Speaking of that New Yorker article – it’s well worth reading. It’s about US law enforcement’s widespread reliance on “the Reid technique” for eliciting confessions, which was concocted by a retired cop out of…nothing in particular.

A growing number of scientists and legal scholars, though, have raised concerns about Reid-style interrogation. Of the three hundred and eleven people exonerated through post-conviction DNA testing, more than a quarter had given false confessions—including those convicted in such notorious cases as the Central Park Five. The extent of the problem is unknowable, because there’s no national database on wrongful convictions. But false confessions, which often lead to these convictions, are not rare, and experts say that Reid-style interrogations can produce them.

See, the goal with confessions is not quantity but quality. I mean, sure, it would be great for law enforcement and thus for the rest of us if they could get truthful confessions in every case, but the “truthful” part is key. Dragging false confessions out of exhausted subjects is not helpful.

Thirty-five years ago, a postdoctoral fellow in psychology named Saul Kassin began researching the psychological factors that affect jury decisions. He noticed that whenever a confession was involved, every juror voted guilty. Alibis and fingerprints didn’t matter in these cases. Kassin read the U.S. Supreme Court’s 1966 Miranda decision and found that it repeatedly cites the Reid Technique manual as the most authoritative source on American interrogation techniques. When he bought the manual, he says, “my first impression was, my God, this reads like a bad psychology textbook. It was filled with assertions with no empirical proof.”

Today, Kassin has appointments at Williams College, in Massachusetts, and at John Jay College of Criminal Justice, in New York, and is widely regarded as a leading expert on false confessions. He believes that the Reid Technique is inherently coercive. The interrogator’s refusal to listen to a suspect’s denials creates feelings of hopelessness, which are compounded by the fake file and by lies about the evidence. At this point, short-term thinking takes over. Confession opens something of an escape hatch, so it is only natural that some people choose it.

Do you think the Miranda ruling makes this not a problem? More than 80% of suspects waive their Miranda rights.

The Reid interrogation technique is predicated upon an accurate determination, during Behavioral Analysis, of whether the suspect is lying. Here, too, social scientists find reason for concern. Three decades of research have shown that nonverbal signals, so prized by the Reid trainers, bear no relation to deception. In fact, people have little more than coin-flipping odds of guessing if someone is telling the truth, and numerous surveys have shown that police do no better. Aldert Vrij, a professor of psychology at the University of Portsmouth, in England, found that law-enforcement experience does not necessarily improve the ability to detect lies. Among police officers, those who said they paid close attention to nonverbal cues did the worst. Similarly, an experiment by Kassin showed that both students and police officers were better at telling true confessions from false ones when they listened to an audio recording of an interview rather than watch it on video. In the experiment, the police officers performed less well than the students but expressed greater confidence in their ability to tell who was lying. “That’s a bad combination,” Kassin said.

It’s the Dunning-Kruger combination. You really don’t want that in police interrogations.

Such studies suggest that a troubling chain of events can easily take place in the mind of an interrogator. During the Behavioral Analysis Interview, the detective begins to form an impression, based in part on the suspect’s body language. The impression could be wrong, but the detective, sensitized to those responses, notices them more and pays less attention to others—an instance of confirmation bias. Increasingly convinced that he’s dealing with a liar, the detective questions more aggressively, and this, in turn, triggers more nervousness. The behaviors create a feedback loop, ratcheting up the suspicion and anxiety to the point where the detective feels duty-bound to get a confession. Psychologists call this cycle the “Othello error,” for the tragic escalation of accusation and fear that leads Othello to kill Desdemona.

Gregg McCrary, a retired F.B.I. agent, told me that Reid-style training creates a tendency to see lies where they may not exist, with an unhealthy amount of confidence in that judgment. “They just assume they’re interviewing the guilty guy,” he said.

And the wrong people go to prison, and sometimes to the lethal injection room.

There are alternatives.

In 1990, after a flurry of false-confession scandals in Britain, the government appointed a commission of detectives, academics, and legal experts to develop an interview method that would reflect up-to-date psychological research. After two years’ work, the commission unveiled their technique, called PEACE, for Preparation and Planning, Engage and Explain, Account, Closure, Evaluate. Training was provided for police departments throughout England and Wales, starting with major-crimes units. By 2001, every police officer in England and Wales had received a basic level of instruction in the method.

The method differed dramatically from previous practices. Police were instructed not to try to obtain confessions but to use the interview as a way to gather evidence and information, almost as a journalist would. They were to focus on content rather than on nonverbal behavior, and were taught not to pay attention to anxiety, since it does not correlate with lying. Instead, police were trained to ask open-ended questions to elicit the whole story, and then go back over the details in a variety of ways to find inconsistencies. For the suspect, lying creates a cognitive load—it takes energy to juggle the details of a fake story. Part of the process involved thorough preparation: police learned to spend hours drawing diagrams of the route they hoped an interview would take. Bluffing about evidence was prohibited. “We were not allowed to lie, coerce, or minimize,” Andy Griffiths, a detective superintendent with the Sussex Police Department, told me. Their job was simply to get as much information as possible, which, along with corroborating evidence, would either inculpate the suspect or set him free.

There are law enforcement people in the US working on developing approaches like that. Let’s hope they make headway.