In 1983 the police in Fauquier County, Virginia, arrested Earl Washington, a twenty-two-year-old, mentally disabled farmhand. He was a suspect in a burglary, but during two days of questioning, detectives asked him about five other crimes. Washington, who had the IQ of a ten-year-old, confessed to all of them. Though four of the cases against him were dismissed, he was convicted of the fifth, a brutal rape and murder, and sentenced to death. Washington spent a total of seventeen years in prison before he was exonerated by DNA evidence in 2000. Five different appellate courts — including the U.S. Supreme Court — had upheld his conviction.

Confessions are seen as the gold standard of evidence in a trial, but cases like Washington’s are more common than people think. Law professor Richard A. Leo has spent several decades trying to bring attention to the problem of false confessions. The public has not always been supportive of his efforts. The average citizen, he says, presumes suspects are guilty and believes they deserve whatever they get. Leo’s work has been cited by the Supreme Court, and he’s been involved in many high-profile cases in which people have given false confessions, including the West Memphis Three and the Central Park Five. In 2010 he was featured in a PBS Frontline documentary about the Norfolk Four, who were the subject of Leo’s book, co-written with Tom Wells, The Wrong Guys.

To understand how the police coerce an innocent suspect into admitting guilt, Leo has examined interrogation techniques, undergoing the appropriate training and sitting in on nearly two hundred interrogations. He says the problem is not necessarily a matter of misconduct by detectives, most of whom are “decent people who follow the rules.” Rather it’s a pattern of errors resulting from misguided methods and a presumption that police have arrested the guilty party.

Born in Italy, Leo grew up in Southern California, where his family moved when he was three. He describes himself as an “accidental lawyer”: while earning his PhD in social psychology at the University of California, Berkeley, in the early nineties, he was given the opportunity to earn a doctor of jurisprudence concurrently almost for free. (He’s proud of having completed both degrees in four years.) He never wanted to practice law in a courtroom, but he’s often been called as an expert witness or hired as a consultant.

Currently the Hamill Family Professor of Law and Psychology at the University of San Francisco, Leo is the author of several books, including Confessions of Guilt and The Miranda Debate, and he’s won awards from the American Society of Criminology, the Society for the Study of Social Problems, and the American Psychological Association. Before coming to the University of San Francisco, he taught at the University of Colorado, Boulder, and the University of California, Irvine. He has eight-year-old twin daughters and in his spare time plays guitar and is an avid boxing fan.

I met with Leo one rainy afternoon this past winter. He’d just changed offices and apologized for the mess; his shelves overflowed with law books. We spoke for two hours, and he often shook his head or laughed uncomfortably at the tragic absurdities of our criminal-justice system.


499 - Richard A. Leo


Leviton: The National Registry of Exonerations currently lists more than two thousand established cases of wrongful criminal conviction for the most serious violent crimes, the earliest from 1989. The Innocence Project at Cardozo School of Law in New York City estimates that false confessions are the third-largest contributor to wrongful convictions. Why do people confess to crimes they didn’t commit?

Leo: To understand that, you have to look at the sequence of errors that often precede the confession.

First the police mistakenly come to view an innocent person as guilty. Ideally an investigator is like a scientist — open to what the data show, letting the facts guide the investigation, testing hypotheses: “I’m going to start with the boyfriend and see where that leads.” Some police detectives do follow this process, but others believe they are endowed with intuition, a sixth sense about who’s guilty.

An FBI agent called me today to discuss one of my studies, and he mentioned how cops keep talking about knowing in their gut when someone is guilty. The agent told me he got hunches, too, but didn’t consider them trustworthy.

Whichever method they use, once the police come to the conclusion that someone committed the crime, they are trained to interrogate. At that point their goal isn’t to gather information; it’s to build a case against the person they’ve already decided is guilty. They want to get a confession.

To do this, they use methods academics like me call accusatory, manipulative, and coercive. If the police interrogate you, and you deny the charge, they will not honor your denials. Their goal is to break you down.

Interviews, which often precede interrogations, do not presume guilt. In interviews officers ask open-ended questions, let the subject do most of the talking, and don’t take an accusatory tone. The police are trained to call every conversation with a suspect an “interview” when they testify in court or talk to the media; they never use the word interrogation. Yet their training manuals make a clear distinction between the two. You don’t subject someone to an interrogation unless you’re reasonably sure that person is guilty.

One way to put it is: in an interview they ask you questions; in an interrogation they tell you the answers.

Leviton: So the police refuse to consider the possibility that a person in their interrogation room is innocent?

Leo: Yes. When I taught at the University of California, Irvine, I would invite an FBI agent to speak to my class. He would admit that false confessions do happen — but never to him. How would he even know that?

False confessions happen all over the country. Take this case in upstate New York: A woman’s baby has sustained an injury, and her boyfriend — not the child’s father — has blood on him. The woman is told by the police that the boyfriend is a natural suspect, but they decide to give her a lie-detector test and tell her she failed it, no matter the outcome.

They interrogate her in a lengthy, high-pressure process. Some police might think of it more as a seduction; others are tougher, telling her she’d better stop bullshitting them, and the only way they can help her is if she admits what she did.

This is the second error: coercing the suspect to confess. When I talk to people who’ve been proven innocent after giving false confessions, they often say they just couldn’t take it anymore. They stopped thinking about long-term consequences and only wanted to get out of the room where they’d been held for hours. They said what the police wanted to hear just to go home, thinking they could prove their innocence later.

Now, in this woman’s case I don’t know if she’s innocent. I do know they didn’t record the interrogation. The so-called lie detector is garbage and inadmissible in court in most states, about as valuable as reading tea leaves. She did plead guilty to a child-endangerment charge and is going to spend two and a half years in prison.

I know of hundreds of people who were proven innocent after they’d confessed — and I don’t just mean juveniles and people with mental impairments or illnesses, all of whom are disproportionately represented when it comes to false confessions. Most innocent people who confess are quite normal. Few of us can withstand such interrogation techniques.

Innocence Project cofounder Peter Neufeld talks about “stone-cold innocence,” which is when the evidence completely clears someone. A rape was committed, a man was falsely imprisoned largely because of a confession, and later a DNA test proves that someone else actually did it. There was no physical evidence linking the first man to the crime, but he confessed. Of the 350 postconviction DNA exonerations documented by the Innocence Project, around 30 percent involved police-extracted confessions.

Leviton: You’ve written about how “confirmation bias” influences the way we evaluate people. Does it come into play in criminal investigations?

Leo: Absolutely. Confirmation bias is the tendency to look for information that supports our beliefs and ignore information that doesn’t. This is a well-documented human tendency. We all have it. It’s why police develop tunnel vision, lock in on one explanation, and discount other possibilities.

Princeton psychologist Daniel Kahneman won the 2002 Nobel Prize in Economic Sciences for showing how, in uncertain circumstances, people use rules of thumb or stereotypes to make decisions instead of factual analysis and reason. Sometimes we have time to evaluate carefully, but when we don’t, we resort to quick, intuitive, emotional thinking. We judge others based on how they dress, whether they make eye contact, and many other factors. It’s not that police are bad people — it’s that they are human, and have biases, and make mistakes, just like everyone else.

Leviton: What other errors do investigators make?

Leo: The next big error is leaking nonpublic details about the case to the suspect, which he or she then parrots back to the interrogator, making the false confession look more convincing: How did he know the victim was in bed, or wearing a particular article of clothing, or assaulted in a certain way? Because he was repeating back information that had been revealed to him, often inadvertently, during the course of the interrogation.

The psychology of interrogation is one of capture and negotiation: “I’ve got you, Mark,” the interrogator will say. “I’ve got a videotape that shows you were at the crime scene. Your buddy told me you did it, and we’ve got your blood at the scene, too.” But none of these things is true. When you say you were never there and didn’t know the victim, the response is “Mark, we know you knew her. She was hit on the head three times with a gun that matches the one you own.” Twelve hours later you’re admitting you hit her three times with the butt of the gun. The police are blind to the fact that they fed you those details.

Leviton: There’s nothing illegal about police lying to a suspect during an interrogation?

Leo: No, it’s totally legal. The courts have given it their blessing year in and year out. In 99.9 percent of cases the police suffer no consequences for lying. Judges look the other way except in the most blatant violations.

There are two sorts of lies: ones about evidence — “We have your bloody fingerprint” — and broader lies like “I’m here to help you.” When the police say they can write reports a certain way or talk to the district attorney or influence the judge, that’s a lie. They want you to believe they are just trying to help you get a more lenient sentence, but cops are separate from prosecutors and have no power over sentencing. You should shut up and get a lawyer. Police are simply trying to get a statement from you that fits their narrative of how you committed the crime.

Oftentimes they exaggerate the potential charges. For instance, if you might reasonably be charged with kidnapping, a cop might suggest you’re going to be charged with kidnapping with intent to commit a sexual assault. The difference in sentences for those two crimes is about twenty years. Police use the prospect of a worse charge to get you to admit to a lesser one.

Leviton: This is also a prosecutorial technique, to inflate charges so that a plea bargain looks more sensible. And then they never have to prove anything in court.

Leo: Trials are rare. Only about 3 percent of cases go to trial. When I was in graduate school decades ago, it was about 10 percent.

Leviton: Would the whole system collapse if suspects stopped cooperating with police and prosecutors and insisted on going to trial?

Leo: I’m not convinced it would. If we had more-efficient courts — they are horribly inefficient — it could be handled.

Leviton: The justice system seems to operate on the assumption that, in order to catch a lot of criminals, we have to tolerate the occasional mistake.

Leo: That’s the working assumption, yes. In 1765 the English jurist William Blackstone said it was “better that ten guilty persons escape than that one innocent suffer.” English law became the basis for American law. It used to be that the idea of convicting one innocent person to catch ten guilty ones was abhorrent. But if we took a poll of Americans today, what do you think the acceptable ratio would be? How many innocent victims would people say should be sacrificed to make us safer?

At Irvine I would ask my undergraduates how many in the class were in favor of the death penalty. About half would raise their hands. I’d then ask whether they’d still be in favor of it if one innocent person in a thousand were executed. Almost all hands would stay up. How about one in a hundred? At one in ten, some hands would still be up, and I would want to cry.

Leviton: Tell me about the case of Eddie Lowery.

Leo: In July 1981 a seventy-four-year-old woman named Arta Kroeplin was raped and assaulted in her home in Ogden, Kansas. Afterward she was unable to describe the rapist, and police had no physical evidence linking anyone to the assault.

Eddie Lowery, an active soldier at Fort Riley, had been involved in a minor automobile accident near the crime scene that night. Two officers interrogated Lowery over two days, without giving him food, and they did not record it. He initially denied all their allegations but eventually broke down and agreed that he’d done it. His first trial ended with a hung jury; a second jury convicted him of rape, aggravated battery, and aggravated burglary. He received a sentence of eleven years to life. After serving ten years, he was paroled and registered as a sex offender. In 2002 DNA testing excluded him as the rapist, and his conviction was vacated a year later.

Lowery sued the police in civil court and received $7.5 million for his wrongful conviction and imprisonment.

Leviton: Would the system change if there were more successful lawsuits against police departments?

Leo: It might lead to better police training — making sure that officers are able to recognize and prevent false confessions. Police could also be better trained on suspects’ constitutional rights and how not to violate them. And perhaps there would be better internal monitoring by police departments of interrogation practices, to make sure they didn’t expose the department to lawsuits.

Once the police come to the conclusion that someone committed the crime, they are trained to interrogate. At that point their goal isn’t to gather information; it’s to build a case against the person they’ve already decided is guilty. They want to get a confession.

Leviton: People who watch police dramas on TV might be surprised to learn that not all investigations uncover conclusive physical evidence of a suspect’s guilt.

Leo: When I was a student at the University of California, Berkeley, in the 1980s, an undergraduate named Bradley Page was convicted of voluntary manslaughter in the killing of his girlfriend, Bibi Lee, who’d disappeared while jogging. He launched a massive search for her, but of course the first thing the police did was arrest him. He confessed under interrogation and later recanted, saying he’d been exhausted and confused by sixteen hours of questioning, six hours of which were not recorded. The police convinced Page that he had repressed the memory of committing the murder. There was no physical evidence linking him to the crime, and his confession contradicted many facts in the case. She had a skull fracture, for example, which supposedly had been caused by Page backhanding her in the heat of an argument. Mike Tyson couldn’t have cracked her skull by backhanding her.

Page was convicted at a second trial (the jury was hung the first time) and spent several years in prison. He was, by all reports, a broken man after that, punished for a crime he almost certainly didn’t commit, although he was never proven innocent. The murder was likely committed by a convicted serial killer.

As I said, once the police fixate on your guilt, they stop investigating and begin building a case against you. Some scholars talk about this as a “suspect-driven” investigation versus an “evidence-driven” investigation. In the Lowery case there were a number of details that didn’t add up, but rather than pursue evidentiary questions, the police decided to go for a confession.

Leviton: In 2000 Corethian Bell found his mother dead and called 911. Under interrogation he admitted that he’d murdered her. Did he?

Leo: No. This case is utterly amazing to me. Bell was mentally ill and had drug problems. His mother was brutally murdered, and there was a lot of physical evidence, none of which connected her son to the crime.

Under police interrogation, after many denials, Bell eventually said that he had killed his mother. In his videotaped confession he referred to many nonpublic details of the crime, all of which he must have gotten from the police during their fifty-hour interrogation. DNA evidence later identified the true killer as a violent sex offender. Before that, Bell spent seventeen months in Cook County Jail. Once the police had Bell’s confession, they just dug in, and anything that didn’t fit — such as his admission that he’d shot her when in fact she’d been stabbed — was “fixed.”

You might think that the more heinous the crime, the less likely someone would be to admit to it, but studies have shown that two-thirds of murder cases are “solved” by confessions, compared to just a quarter of all crimes.

When you let police lie, they can make up anything. At one police-training seminar I attended, a detective bragged that he’d told a suspect he had found a “molecule match” at the crime scene. There’s no such thing as a molecule match, but the suspect believed it and confessed.

Leviton: Is there more pressure to obtain a confession when it’s a murder?

Leo: Absolutely. They don’t interrogate suspected car thieves or burglars for sixteen hours. For my PhD dissertation I spent almost a year observing the Oakland Police force. The homicide detectives there had no limit on the amount of overtime pay they could receive. One of them told me that overtime pay was the reason interrogations sometimes went from 10 PM to 10 AM.

Leviton: And confessions obtained after long interrogations, sometimes without food or water or bathroom trips or cigarette breaks, are nonetheless considered “voluntary”?

Leo: Whenever a confession is entered as evidence in a trial, there’s always a judge who has ruled it voluntary. Suspects, however, often say they were threatened or believed the only way to get out of the room was to say what the police wanted.

Few interrogations are video- or audiotaped, and even when there is a recording, judges almost never question how the confession was obtained or suppress it as evidence. They instinctively support the prosecution.

When the defense tries to get a confession excluded — because it was coerced and the conditions under which it was elicited violated due process — many times the judge will say, “Well, Counselor, this is a close call, but I have confidence in the jury.” The judge kicks the can down the road, because if the jury acquits, the judge won’t be accused of being soft on crime, and prosecutors won’t raise money to oppose the judge in the next election.

Judges should show more critical thinking and skepticism. Instead they lean toward believing the police when there’s a “swearing contest” — the defendant’s word against the police’s.

As early as 1967, in Chapman v. California, the Supreme Court ruled that a violation of a defendant’s constitutional rights was not automatically sufficient to overturn a conviction. They deemed some such violations “harmless.” In 1991, in Arizona v. Fulminante, the court ruled that it can also be “harmless” for juries to hear confessions that should have been ruled inadmissible. But Brian Wallace and Saul Kassin of the John Jay College of Criminal Justice have shown that judges and juries see a confession as such powerful evidence that they do not discount it, even when it would be legally and logically appropriate to do so.

We need a better standard than whether or not a confession is voluntary. Who knows what “voluntary” means anyway? If you stick a gun to my head and say, “Give me your wallet,” to some extent I’m making a voluntary choice to give it to you. I could say no. But legally we would say that’s an involuntary action, because of the threat. Yet I’ve seen a judge examine an interrogation transcript in which a suspect is threatened and then, twenty pages later, confesses; the judge ruled that, because enough time had passed, it wasn’t the threat that had caused the confession but the desire to tell the truth.

Leviton: So let’s connect these dots in false-confession cases: The police go after a confession because they know they’ll have a much harder time proving guilt without it. Then the confession hardly ever gets suppressed as evidence because judges don’t want to appear soft. And the jury ranks the confession as the number-one piece of evidence proving guilt.

Leo: Confessions trump just about everything else. There are rape cases in which the DNA evidence introduced at trial fails to connect the suspect to the crime, but juries think, “Maybe he wore a condom.” In what universe do rapists wear condoms? In the language of the law, confessions are “prejudicial.” Once a confession has been entered into the stream of evidence, even if it’s contradicted by the facts, it’s very hard for the defense to undo the damage. In a sense, the real trial has taken place in the interrogation room.

Leviton: I was shocked to learn from your work about the “error-insertion trick,” in which police deliberately put small mistakes into a written confession before asking the accused to read and sign it. For instance, police might put a wrong birth date in the document.

Leo: Yes, and when the accused corrects the error, it becomes evidence of voluntariness and deliberation. The cop can say, “Sure, I wrote the confession, but look here — he corrected three errors.” Another trick is to coerce the accused into signing an apology to the victim.

Leviton: How can the defense prove a confession was false?

Leo: A colleague and I have come up with four ways. One is to show the crime didn’t occur: the murder victim whose body was never found later shows up alive, or the baby who supposedly died from being shaken is later found to have died of an infection. The second way is through physical impossibility. When I lived in Laguna Beach, California, a man who confessed to burning down houses there had actually been in jail in Mexico at the time of the fires. The third is DNA or other scientific evidence that definitively rules out the accused. And the fourth is when the true perpetrator is identified. But most of the time, if you are falsely convicted of a crime, your ability to prove you didn’t do it comes down to pure luck or something out of your control. Your crime might have taken place near a surveillance camera, or maybe the victim had a cellphone for which call records could be obtained.

My book The Wrong Guys examines the case of the young Navy sailors who became known as the Norfolk Four. There were multiple false confessions to a rape-murder in that case. The DNA evidence pointed to a much more credible suspect, but the police clung to their theory that the victim’s husband and three accomplices had committed the crime. The real murderer was eventually identified, and in 2009, well after the sailors had agreed to plea bargains, some pro bono lawyers got the case before Tim Kaine, who was then governor of Virginia. He held a press conference and said he couldn’t get past the fact that there were so many confessions. He decided to commute the men’s sentences and release them from prison but without pardoning them.

These men qualified for “stone-cold innocent” in my opinion. The confessions were coerced, and DNA evidence showed that the true perpetrator had acted alone. But in the eyes of the law, these men were convicted and discharged, not found innocent.

Leviton: What becomes of people who are proven innocent after years in prison? How does it affect their lives?

Leo: The damage is often horrific. I’ve written about Jerry Townsend, the son of a sharecropper from Greensville, Mississippi, who had an IQ of around 60 and was convicted of rape and murder in 1980. During interrogation he implicated himself in more than twenty homicides in Miami and other cities. No physical evidence linked him to any of these murders, and it was obvious even to the police that he’d falsely confessed to at least some, but he was convicted of several. He spent twenty-two years in prison until a judge vacated his convictions, calling them “an enormous tragedy.”

In prison Townsend lived in constant fear for his life. He was classified as a maximum-security prisoner, and his activities outside his cell were extremely limited. Perhaps more important, because his convictions involved the rape and murder of a child, guards and inmates especially disliked him. As a sex offender, he had only limited visits with family and was not allowed to see his young daughter. After his release on June 15, 2001, he showed many signs of trauma. He walked slowly, with his head down and his shoulders stooped. He constantly looked behind him as if expecting to be ambushed.

Leviton: You’ve written that some people who falsely confess actually believe they committed the crime. This is surprising to me. It’s one thing to decide under interrogation that I’m better off confessing, but to believe I did something I don’t remember doing?

Leo: Here’s what happens: At some point the innocent suspect says, “Look, if I’d done this crime, I’d have some memory of it.” A clever interrogator says, “Not necessarily. People do things all the time that they don’t remember afterward.” The officer claims to have enough evidence to convict the man — a lie. The suspect begins to think, Could I have repressed the memory of the crime? Could I have had a blackout?

Tom Franklin Sawyer of Clearwater, Florida, was interrogated in 1986 for fourteen hours about the death by strangulation of his next-door neighbor. An appeals-court judge later said Sawyer had been subjected to “grossly leading questions” and deprived of food, drink, and sleep. Sawyer was an alcoholic. He hadn’t had a drink in three months, but the interrogator told him that sometimes alcoholics have “dry blackouts” — a fabrication. When you let police lie, they can make up anything. At one police-training seminar I attended, a detective bragged that he’d told a suspect he had found a “molecule match” at the crime scene. There’s no such thing as a molecule match, but the suspect believed it and confessed.

Most suspects defer to authority and believe what the police tell them, but the police can be complete liars when they want to extract a confession. The suspect is thinking, “Why would they lie to me? I must have done it. My daughter’s accusing me of sexual abuse. She would never lie. How could this have happened? How do they have evidence?” The police then start to work on cutting out the suspect’s conditional words: I might have. I could have. Maybe. Probably. The interrogation continues for hours. Eventually the suspect may even be fairly certain he did it.

Leviton: Do people who believe their own false confession go on thinking they did it?

Leo: Cases in which people genuinely internalize the false memories for lengthy periods of time are rare. They might believe they are guilty for several hours or even days after the interrogation, but it’s not like being in a cult, where you are indoctrinated day after day for a long time and your whole belief system changes.

The police need the suspect to believe just long enough to confess. After the statement has been signed, if the suspect recants, the interrogator chalks it up to regret.

Leviton: You mentioned that suspects defer to authority. So false confessions don’t arise only from a desire to get out of the room but also from a desire to cooperate with someone who’s in charge.

Leo: I think so. Interrogations almost always begin with suspects saying they didn’t do it and showing a desire to help the police find out who did. People who are innocent believe their innocence is transparent and that the police will eventually see this and let them go. They believe they are protected and have nothing to hide, but they don’t realize the many ways that facts and memories and statements can be manipulated and massaged.

Leviton: Let’s talk about the protections U.S. citizens have under the Constitution and our laws. The most visible protection is what’s called the Miranda warning.

Leo: The warning is “You have the right to remain silent. Anything you say can and will be used against you in a court of law. You have the right to an attorney. If you cannot afford an attorney, one will be provided for you.” Then you are asked if you understand these rights and later to acknowledge this in writing. The police might then say, “Having these rights in mind, do you wish to speak to me?”

The warning originated from the 1966 Supreme Court case Miranda v. Arizona. The court decided that the Fifth Amendment provision against self-incrimination — which says you may not be compelled to be a witness against yourself — applies not just in court but also in police custody. Many law professors and lawyers were angry about this ruling because it created a new right in the interrogation room that had not previously seemed to exist; they said the Fifth Amendment referred only to trials. But the Warren Court saw interrogations as intrinsically coercive, and it tried to discourage the coercion by ensuring suspects would be aware of their right to refuse interrogation.

At the time many conservatives pilloried the Warren Court. In 1968 Richard Nixon ran in part on an anti–Warren Court platform. Most observers thought the Miranda ruling would hamper law enforcement’s ability to solve important crimes, because suspects would all invoke their right to remain silent and have a lawyer with them during questioning.

It didn’t turn out that way. Around 80 percent of people, if not more, waive their Miranda rights. In subsequent years the Supreme Court has created gaping exceptions to Miranda: half the time the police don’t even need to read suspects their rights before questioning them, because the questioning is considered “noncustodial” — meaning the suspect is not under arrest — even when it’s a closed-door interrogation inside a police station by officers wearing guns.

What the police do is say, “Mark, I have a few questions I’d like to ask you. I’ve got all the paperwork at the station. Can you meet me there?” When you arrive, they say, “I’m going to close the door for privacy, but it’s not locked. Anytime you want a bathroom break, just let me know.” Now they don’t have to read you your rights, even though you’re in a closed interrogation room.

The Miranda warnings have been watered down by the courts and gamed by police officers until they are almost completely ineffective in preventing false or unreliable confessions. It’s like the health warning on a pack of cigarettes: it protects the tobacco company, but it doesn’t really affect consumers’ behavior. It just means they’ve given “informed consent.”

Leviton: So Miranda has turned out to be a tremendous advantage for police?

Leo: Yes. The police have realized that almost everyone waives his or her rights, and after the police have read the Miranda card, anything else they do is OK, because they followed procedure.

Leviton: But when someone asks for a lawyer, the police have to grant the request, right?

Leo: It’s unlikely police will ignore a straight, clear request for a lawyer. It’s more common for the suspect to ask, “Should I get a lawyer?” and the cop will reply, “You can get a lawyer if you want, but then we can’t really help you.” If the suspect keeps talking after that — and a lot of them do — the judge will rule that the suspect didn’t unequivocally say, “I want a lawyer.”

There was a case in which the suspect, after being read his rights, remained completely silent for two hours and forty-five minutes. Then the cop told him that God would forgive him if he just told the truth, and the suspect confessed. His defense attorney argued to suppress the confession because he had never waived his Miranda rights, but the judge decided the suspect had waived his Miranda rights through his silence. In other words it’s now up to the defendant to prove he invoked his rights, instead of the prosecution having to prove he waived them. In order to invoke your right to remain silent, you have to speak.

In one case a suspect kept saying, “I plead the Fifth!” and the cop later argued that he’d never explicitly said he wanted a lawyer, so it was OK to continue the interrogation. Only in the most extreme cases, when the person clearly invoked his or her Miranda rights and was ignored, do judges suppress a confession.

Leviton: Do wealthy people tend to avoid interrogation?

Leo: I certainly think so. When I studied two hundred cases in the Oakland Police Department, the vast majority of suspects interrogated were lower or working class. People of greater means are less likely to speak to police if they are accused of a crime. With white-collar crime there are few confessions. Those people get lawyered up quickly.

Leviton: The wealthy are also more aware that the police are required to prove guilt, and that they are not required to help the police.

Leo: Right. Interrogators not only present themselves as a friend to the suspect; they suggest that the suspect has to show he or she didn’t commit the crime, which is untrue.

Pretending to be the suspect’s friend — a common, well-documented interrogation technique — is fundamentally deceptive, because the interrogator is the opposite of the suspect’s friend. The detective’s goal is to get the suspect to incriminate himself. We could debate whether this kind of deception is justified, but that it occurs is indisputable.

Leviton: In a perverse way the cop is being honest when he says to a person of color or someone living in poverty, “What do you think’s going to happen to a person like you in front of a jury?” The cop has probably seen how the system is loaded against them.

Leo: Yes. I’ve heard cops say to a minority suspect, “What do you think an all-white jury is going to think of you?”

Leviton: As more people have been exonerated and released, and these cases have been publicized, have there been changes in public opinion?

Leo: Public support for the death penalty has dropped from over 80 percent twenty years ago to just over 50 percent today, and polls show a slight majority of Americans would prefer that life without possibility of parole be the most severe punishment allowed. A 2014 Pew Research Center analysis found that support for the death penalty declines during times of low violent-crime rates, such as we are experiencing now. Pew also reported that news of exonerations and botched executions has been eroding support.

People who are innocent believe their innocence is transparent and that the police will eventually see this and let them go. They believe they are protected and have nothing to hide, but they don’t realize the many ways that facts and memories and statements can be manipulated and massaged.

Leviton: In the 1920s when J. Edgar Hoover became director of the Bureau of Investigation — the precursor to the FBI — he was seen as a reformer who wanted to replace brutal interrogations with scientific forensics, centralized fingerprint databases, and so forth.

Leo: Yes, he was trying to professionalize the police. The Supreme Court at the time did not like the “third degree” — which was basically interrogation by torture. They would invalidate confessions if the suspect had been whipped or held in a room for six hours, or if the police had threatened to arrest the suspect’s spouse or take away a child. But there was no clear line; each case was decided as it came up.

Another reason use of the third degree declined was the invention of the polygraph — the so-called lie detector. The polygraph, it was thought, could read your mind. Beating a prisoner came to be seen as messier and less reliable than using a machine. Polygraph results are almost never admissible in court, and failing a lie-detector test simply means you registered more emotional arousal when asked one question than you did when asked another. Your heart beat faster. So what? You might react that way for all sorts of reasons.

President Nixon, on one of the White House tape transcripts, said about polygraphs, “I don’t know how accurate they are, but I know that they’ll scare the hell out of people.”

If I interrogate you for a crime, I might use a polygraph to scare you. I’ll tell you the polygraph proved your guilt and get you to falsely confess. At trial the polygraph test isn’t admissible, but the confession is. Defense lawyers never bring up a failed polygraph test in court, because they think the jury will take it as another reason to convict, but I think the defense should tell the jury that the polygraph is junk science and the results aren’t admissible, and then explain how the polygraph was used to get the defendant to falsely confess. Otherwise the jury has no idea how the confession was obtained.

Police sometimes might believe they can use the polygraph to determine real guilt or innocence, but it’s basically no more accurate than guessing.

The Miranda warnings have been watered down by the courts and gamed by police officers until they are almost completely ineffective in preventing false or unreliable confessions. It’s like the health warning on a pack of cigarettes: it protects the tobacco company, but it doesn’t really affect consumers’ behavior.

Leviton: Now there’s another gizmo, a “stress test,” that supposedly shows when a suspect is lying.

Leo: Another fraud. It’s called the Computer Voice Stress Analyzer and was invented by cops, not scientists. It purports to measure microtremors in the voice. Scientists have found no evidence that such microtremors even exist, and there is no logical reason to believe these nonexistent microtremors in the voice could distinguish between truth and deception if they were real. Again, the results aren’t admissible. It’s just another trick to extract a confession.

Leviton: Have any officers spoken out about the problem of false confessions?

Leo: The police do not talk about this; they usually don’t even recognize it. Because cops do not acknowledge the existence of false confessions or how they cause them, they will never be able to solve the problem. There needs to be pressure put on police departments from the outside.

Leviton: What reforms do you advocate?

Leo: I think all interrogations should be recorded. We should have a reliable record of the entire process, not just partial notes. Too often the defendant says the police threatened to arrest his wife, and the cop denies it, and the judge and jury side with the police. If there were a recording, we’d know.

Leviton: You want all the facts available.

Leo: Yes. Why not use digital cameras in interrogation rooms? They’re inexpensive, easy to install, and can be voice- or motion-activated. The camera isn’t necessarily going to make it easier for the defense. I saw one interrogation videotape in which the suspect said he was blind, and when the police left the room, he started reading a newspaper. In another, while the police were gone, the suspect got a tissue and started wiping blood off his shoe. The police came back in and took his shoes as evidence. The video camera doesn’t have a bias. It’s there to record the truth. Can’t we all agree we want the truth?

Leviton: Would it be legally necessary to reveal that video cameras or tape recorders were present?

Leo: In most states, no. Eleven states legally require that both parties approve the recording of conversations, although there are some exceptions made for law enforcement. In “one-party consent” states, you can be recorded without your knowledge. At a police station there’s no reasonable expectation of privacy, so the police can usually surreptitiously record without telling you.

There should not be selective recording, however, such as when police record a confession but nothing that came before or after it. That’s exactly what happened in the Central Park Five case. A jogger was found brutally beaten and raped in New York City’s Central Park, and the teen suspects’ videotaped confessions included details that turned out to have been revealed to them by the police. Kharey Wise, one of the defendants, first said the jogger’s head injuries were the result of punches. After police prompting, he changed that to a rock. A few moments later the rock turned into a brick. He said he was with his friend Al. A little while later Al disappeared from his confession, replaced by someone named Eddie. Another suspect, Antron McCray, told the police the jogger wore a T-shirt and blue shorts; she wore a long-sleeved jersey and black tights.

Leviton: Donald Trump was calling for the death penalty after the rape happened, and during his presidential campaign he repeated his claim that the defendants were guilty, even though they’d been exonerated.

Leo: It’s not just that those five Latino and black men have been exonerated; they have been proven innocent. The DNA evidence shows that Matias Reyes was the true perpetrator. He confessed to doing it alone; he was seen by the police in the park that day; he provided details that were not known publicly or fed to him by interrogators; and he had a history of violently raping women.

Leviton: What would need to happen for all interrogations to be recorded?

Leo: It would take legislation, in most cases. In 2013 California state senator Ted Lieu read a study I’d done with Steven Drizin about 125 false confessions. Lieu was disturbed to learn that one-third of those cases were juveniles. He sponsored a bill that would mandate full recording of any interrogation of a minor, and it passed. In September 2016 Governor Jerry Brown signed subsequent legislation extending this policy to all homicide suspects. There is still a clause to allow for situations in which the lack of a recording was unintentional or the result of equipment failure, so it’s not ironclad. But every state should have this protection in place.

Chicago attorney Thomas P. Sullivan says that once police departments start videotaping interrogations, they see the benefits. They even use the videotapes to train officers, showing good and bad techniques in the interrogation room and explaining why an attorney was able to suppress a confession later. Sullivan suggests that if police are reluctant to use cameras, they should start by recording all homicide interrogations. Once they do this, he thinks, they will want to extend it to every interrogation.

Leviton: If you are walking down the street, and the police pull up next to you and start asking you questions, what should you do?

Leo: You’re not in custody unless you are told you’re under arrest. Street encounters are considered “temporary detentions,” and you can leave if the officers respond in the negative to the question “Am I under arrest?” If you voluntarily answer questions from the police in this situation, they can use your answers to incriminate you.

I would argue that all police should wear body cameras in the field because it will help, not hurt, them. Studies show that cameras protect police against false accusations; they save time, because officers don’t have to write as much in their reports when there’s a video record; and they facilitate plea bargains.

Leviton: Should jury instructions at trial be revised?

Leo: I think there’s room for improvement. A good jury instruction might be “Research shows that people do make false confessions. Don’t decide a confession is automatically true. You must determine its credibility. Here are the reasons someone might falsely confess to a crime. . . .”

But jury instructions come near the end of the process. If you want to prevent wrongful convictions, the earlier you intervene, the better. If the defense can get a false confession thrown out, the jury doesn’t have to conclude that it was coerced.

Police should be required to have some reasonable basis for believing that an individual committed a crime before interrogating. Officers can’t just go into your house looking for contraband or pull you over on the highway for no reason; they are required to have a reasonable basis or a warrant. But they can put you into an interrogation room without any cause.

Another reform would be to have judges hold pretrial reliability hearings regarding confessions. Some judges do this already. The defense should be able to argue before a trial begins that a confession is weak, that it has the earmarks of a false confession, that there are discrepancies and contradictions, and that it would be prejudicial to put it before a jury.

Leviton: Are there reforms you think won’t help?

Leo: Some academics have proposed making it illegal for police to lie, or argued for better Miranda warnings, or for an adult to be present when a minor is questioned. I’m less likely to advocate for these reforms. For instance, I’ve found that an additional adult around an underage suspect often becomes another interrogator. And if the police weren’t allowed to lie, I think they’d just learn to adapt, as they did with Miranda.

Leviton: Do you think DNA testing should be compulsory?

Leo: I don’t see a major downside to it, but I also don’t think it’ll make a difference in the vast majority of cases, because there is biological evidence in only around 10 percent of criminal cases. Civil libertarians might oppose it, because suspects’ DNA will go into databases even if they are innocent, which violates privacy rights. But I care more about getting the correct verdict. If the cost of reducing wrongful convictions is for all suspects to give DNA samples, I’m in favor of paying it. I feel the same way about security cameras in public places. This isn’t the government snooping on your private life; this is an effort to counter the very real possibility of convicting the wrong person. We cannot afford to keep making mistakes and sending so many people who aren’t guilty to prison.

Human error is just too prevalent. Something like fifty people are involved in every wrongful conviction: the police, the prosecutor, their investigators, the judge, the jurors, the appellate judges. It’s staggering. They all missed the fact that the defendant was innocent.