How Police Interrogations Actually Work (And Why Innocent People Confess)

Brendan Dassey was 16 years old. He had an IQ that multiple experts placed in the low-to-mid 70s — borderline intellectual disability. On March 1, 2006, two Manitowoc County detectives brought him into a small room at the Two Rivers Police Department in Wisconsin. No lawyer. No parent. Just Brendan and two investigators who had already decided what happened.

Over the next several hours, those detectives extracted a confession to murder and sexual assault. The transcript is publicly available. It is one of the most uncomfortable things you will ever read. Dassey didn't volunteer details. He guessed. He got things wrong. The detectives corrected him. When he couldn't produce the answer they wanted, they fed it to him. At one point, investigator Tom Fassbender said: "We already know what happened, OK? We just need you to be honest with us." They didn't know what happened. They needed Brendan to say what they believed happened.

The most damning moment came when detectives asked Dassey what he did to the victim's head. He guessed "cut her hair." Wrong answer. They pushed. He guessed again. Finally, Fassbender asked, "Who shot her in the head?" Dassey replied, "He did." The detail — a gunshot to the head — came from the detectives, not from Brendan. But it ended up in his confession as if he'd known it all along.

Making a Murderer put Dassey's interrogation tape in front of millions of viewers in 2015. And something remarkable happened. Millions of people watched the exact same footage and came to opposite conclusions. Some saw a guilty kid who knew too much. Others saw a terrified teenager being manipulated into saying whatever the adults in the room wanted to hear.

That's the problem with police interrogations. They're engineered to produce confessions. Not truth. Confessions. And those are not always the same thing.

The Reid Technique: how American police get confessions

The dominant interrogation method used by law enforcement in the United States is called the Reid Technique. It was developed in the 1960s by John E. Reid, a former Chicago police officer turned polygraph expert. Reid & Associates has trained hundreds of thousands of police officers over the past six decades. If you've ever watched a detective show where the investigator leans across the table, acts sympathetic, and says something like "I know you're not a bad person — just tell me what happened" — that's Reid.

The method is built on a 9-step escalation model. It begins before the suspect even knows they're being interrogated.

Step 1: The Behavioral Analysis Interview

Before the formal interrogation starts, the detective conducts a "non-accusatory" interview. The stated goal is to observe the suspect's baseline behavior and look for signs of deception — breaking eye contact, fidgeting, crossing arms, giving evasive answers. Based on this assessment, the detective decides whether the suspect is lying. The problem? Decades of research have shown that people — including trained police officers — are barely better than a coin flip at detecting deception from body language. A meta-analysis by Bond and DePaulo found that human accuracy at lie detection averages around 54%. Essentially chance.

But the detective proceeds as if they know.

Step 2: Positive confrontation

The detective tells the suspect directly: "We know you did this." This isn't a question. It's a statement. The investigator presents confidence — often manufactured — that the suspect's guilt is already established. They may reference evidence, real or exaggerated. The goal is to make the suspect feel that denial is futile.

Step 3: Theme development

Now the detective offers the suspect a psychological lifeline. "I understand why this happened. You're not a monster. Anyone in your situation might have done the same thing." The detective constructs a narrative that minimizes the suspect's moral responsibility. It was an accident. You were provoked. You were drunk. The victim did something to deserve it. This isn't sympathy — it's strategy. The detective is building a story that makes confessing feel like the reasonable thing to do.

Steps 4-5: Shutting down denials

When the suspect tries to say "I didn't do it," the detective interrupts. Physically — leaning forward, raising a hand — and verbally. "We're past that. I know you were there." Denials are not allowed to gain any momentum. If the suspect withdraws — goes quiet, stares at the floor, stops engaging — the detective moves closer, re-establishes eye contact, and personalizes the conversation. "I'm trying to help you here. But I can only help if you work with me."

Steps 6-7: Minimization and the alternative question

This is the kill shot. The detective presents two versions of events. One sounds bad: "Was this something you planned for weeks? Did you go there intending to kill her?" The other sounds almost forgivable: "Or was it a moment of anger? Something you never meant to happen?" Both options assume guilt. The suspect isn't being asked if they did it. They're being asked why. The "soft" option feels like an escape hatch. Psychologically, after hours of pressure, isolation, and exhaustion, choosing the lesser of two evils can feel like relief.

Steps 8-9: The confession

Once the suspect chooses an alternative — "Yeah, it just happened, I didn't plan it" — the detective locks it down. They get details. They get a written or recorded statement. They bring in witnesses. The confession is now evidence, and in an American courtroom, there is almost nothing more powerful.

Why the Reid Technique breaks people

It works. That's the uncomfortable truth. The Reid Technique produces confessions at a very high rate. The problem is that it produces confessions from guilty people and innocent people, and it has no reliable mechanism for telling the difference.

The psychology isn't complicated. It's just brutal.

Isolation. You're alone in a small, windowless room. No phone. No friend. No lawyer, unless you explicitly ask — and most people don't, because they think asking for a lawyer makes them look guilty. The room is designed to be claustrophobic and disorienting.

Duration. Interrogations can last 4, 8, 12, even 20+ hours. The average person has never experienced sustained psychological pressure from authority figures for that length of time. Sleep deprivation kicks in. Blood sugar drops. The capacity for rational decision-making degrades with every hour.

Authority. A detective is telling you — with complete confidence — that they have evidence proving your guilt. Most people trust authority figures. Most people believe that the system wouldn't target an innocent person. So when a detective says "we know you did this," a part of you starts to wonder if maybe you did.

Minimization. The detective makes confessing sound like the only path that doesn't end in catastrophe. "If this was an accident, we can work with that. If you make me prove it was deliberate, I can't help you." The implicit message: confess now and things will be better. Don't confess and things will be much, much worse. This is functionally a threat wrapped in sympathy, and courts have generally allowed it.

Maximization. Simultaneously, the detective inflates the consequences of not confessing. "The evidence is going to come out. The jury is going to see everything. Without your side of the story, they'll assume the worst." The walls close in.

The human brain has limits. Everybody's brain has limits. Young people break faster. People with intellectual disabilities break faster. People with anxiety, depression, or prior trauma break faster. But given enough time and enough pressure, anyone can be broken. That's not speculation. It's demonstrated by decades of research and hundreds of documented cases.

The false confession problem

According to the Innocence Project, roughly 29% of wrongful convictions overturned by DNA evidence involved false confessions or incriminating statements.

Read that number again. Nearly one in three.

These aren't people who were borderline guilty or probably involved. These are people whose innocence was proven by DNA — the gold standard of forensic evidence — after they confessed to crimes they did not commit. They sat in a room with detectives and said "I did it" when they didn't. And juries convicted them because why would someone confess to something they didn't do?

That question — why would anyone confess to something they didn't do? — is the single biggest obstacle to reforming interrogation practices. People genuinely cannot imagine themselves confessing falsely. So they assume it doesn't happen, or that the people it happens to must be unusual in some way. They're not.

Central Park Five (1989)

Five Black and Latino teenagers — ages 14 to 16 — were accused of beating and raping a white jogger in Central Park. All five confessed after interrogations lasting 14 to 30 hours. Their confessions contradicted each other. Their confessions contradicted the physical evidence. None of their DNA matched. It didn't matter. They were convicted and served between 6 and 13 years in prison. In 2002, Matias Reyes — a convicted serial rapist already in prison — confessed to the crime. His DNA matched. He'd acted alone. All five convictions were vacated.

Think about that for a second. Five confessions. Five convictions. Zero guilty people.

Henry Lee McCollum (1983)

McCollum was 19 years old and intellectually disabled. He was interrogated for five hours without a lawyer. He confessed to the rape and murder of an 11-year-old girl in Red Springs, North Carolina. He was sentenced to death. He spent 30 years on death row — the longest-serving death row inmate in North Carolina history — before DNA evidence pointed to another man, Roscoe Artis, who had been convicted of a similar crime committed a month later just a block away. McCollum was exonerated in 2014.

Thirty years. On death row. For a false confession extracted in five hours.

The Norfolk Four (1997)

Four U.S. Navy sailors — Daniel Williams, Joseph Dick Jr., Derek Tice, and Eric Wilson — confessed to the rape and murder of Michelle Moore-Bosko in Norfolk, Virginia. Their confessions were riddled with inconsistencies and contradicted the physical evidence. Meanwhile, a man named Omar Ballard — who had a history of sexual violence — confessed unprompted and told police he acted alone. His DNA was the only match at the scene. It took over a decade of legal battles before all four sailors were pardoned. Ballard had told the truth from the beginning. The system ignored him because it already had confessions.

These are not edge cases. The National Registry of Exonerations has documented hundreds of false confession cases. They happen in every state, in every decade, and they disproportionately affect young people, people of color, and people with intellectual disabilities.

The alternative: the PEACE model

In 1992, the United Kingdom did something radical. After a series of high-profile wrongful convictions caused by coerced confessions — the Guildford Four, the Birmingham Six — the British government commissioned a complete overhaul of how police conducted interviews. The result was the PEACE model.

PEACE stands for: Preparation and Planning, Engage and Explain, Account, Closure, and Evaluate.

The difference from Reid isn't just procedural. It's philosophical. PEACE is an investigative interview, not an interrogation. The goal is not to extract a confession. The goal is to gather accurate information. The interviewer doesn't decide beforehand whether the suspect is guilty. They ask open-ended questions. They let the suspect talk. They probe inconsistencies with evidence, not with psychological coercion. And critically — deception by the interviewer is not permitted. Police in the UK cannot lie to a suspect about evidence. They cannot say "your fingerprints were at the scene" if they weren't.

In the Reid Technique, detectives are explicitly trained to lie. They can tell a suspect they have DNA evidence when they don't. They can say a witness identified them when no witness exists. They can fabricate evidence entirely. In most U.S. jurisdictions, this is perfectly legal. Courts have upheld it repeatedly. The landmark case Frazier v. Cupp (1969) established that police deception, by itself, does not render a confession involuntary.

PEACE has been adopted as the standard in the United Kingdom, Canada, Australia, New Zealand, and several European countries. Research by Professor Ray Bull and others has found that PEACE produces more detailed and more reliable information than accusatory methods. False confession rates in PEACE-model countries are significantly lower.

Adoption in the United States has been glacial. A few agencies — notably the High-Value Detainee Interrogation Group (HIG), a federal task force — have moved toward evidence-based interviewing. Some states and police departments have started training officers in PEACE or PEACE-influenced methods. But the Reid Technique remains dominant, backed by 60 years of institutional inertia and an industry that profits from selling Reid training courses.

How to spot a coerced confession

If you're reading this and wondering how anyone could watch an interrogation tape and not see the coercion — you're not alone. But spotting it requires knowing what to look for. Defense attorneys and wrongful conviction experts typically flag several indicators.

Information contamination. This is the biggest one. A genuine confession contains details that only the perpetrator could know — where the weapon was hidden, what the victim was wearing, something not released to the public. A coerced confession contains details that the detectives knew and the suspect parroted back. Review the transcript carefully: did the suspect provide the detail, or did the detective provide it and the suspect agree?

Suspect echoing detective language. When someone describes an event they experienced, they use their own words. When someone is repeating what they've been told, they often use the exact phrasing of the person who told them. If a suspect's confession sounds like it was written by a detective, it may have been.

Confession contradicting physical evidence. If the suspect says they stabbed the victim but the victim was shot. If they say they entered through a window that was painted shut. If the timeline in the confession is physically impossible. Genuine confessions align with the evidence. Coerced confessions are often constructed from the detective's theory of the case, which may be wrong.

Duration. Any interrogation exceeding 6 hours should raise eyebrows. Beyond 12 hours, the risk of a false confession increases dramatically. The Central Park Five interrogations lasted up to 30 hours.

No lawyer present. Suspects have a constitutional right to an attorney under the Sixth Amendment and a right to be informed of that right under Miranda. But exercising that right requires affirmatively saying "I want a lawyer." Many people — especially young people, people unfamiliar with the legal system, and people who are scared — don't invoke it. Detectives are trained not to remind them.

Suspect asking to go home. This one is heartbreaking. In multiple false confession cases, the suspect asked some version of "If I say what you want, can I go home?" They don't understand that confessing to murder means they're never going home. They think the interrogation is the punishment. They think agreement is escape. Brendan Dassey asked if he'd be back in time for his sixth-period class. He'd just confessed to murder.

Mandatory recording has changed everything. For decades, interrogations happened behind closed doors with no record of what was said or done. The detective's notes were the only account. Today, more than half of U.S. states require electronic recording of custodial interrogations, and the trend is accelerating. Recording doesn't prevent coercion, but it allows judges, juries, and appeals courts to see exactly what happened. Many of the most famous false confession cases were only exposed because a recording existed.

The tension that makes this real

DetectiveOS has an interrogation room. You sit across from a suspect, review what you know, present contradicting evidence, and try to crack their story. The suspects push back. They deflect. They lie. Your job is to figure out who's telling the truth and who isn't — using evidence, not pressure. It's a simplified version of the real thing, but that core tension is authentic: you're certain someone is hiding something, and the only way through is finding the proof that makes the truth undeniable.

A confession without evidence is just words. Evidence without a confession is still evidence. The best investigators know which one to trust.