The Science of False Confessions

If you haven’t watched John Oliver’s take-down of American police interrogation techniques, please click on the video above. You will be horrified.

To a large extent, this is an only-in-America problem. If you have watched a lot of British police dramas you will recognize the scene. An investigator sits down with a suspect, turns on her tape recorder, gives the date, time and place of the interview, gives her own name and the name of the suspect. Only then does the interview proceed. A similar scene plays out in recent Scandinavian, Australian or Canadian crime procedural dramas. In America, police interrogations are typically not recorded. And that’s a problem.

Juries love to hear defendants confess. It is commonly regarded as proof of guilt beyond a reasonable doubt. Why, we asks, would an innocent person admit to a crime? It makes not sense. Either you did it or you didn’t. And if you didn’t do the deed, by definition, there will be no memory of the crime for the mind to draw on. With sufficient reflection, we assume, essentially accurate memories can be uploaded from the brain’s memory banks.

No one expects witness memory to be perfect, of course. Memory’s video may be a bit grainy, but it’s good enough. Right?

Wrong!  Elizabeth Loftus, the world’s leading authority on the human memory, believes that memory has more in common with imagination than photography. “Memory is constructed and reconstructed,” she says.  “It’s more like a Wikipedia page — you can go change it, but so can other people.”  (Her terrific TED talk on the subject that will quickly bring you up to speed.)

“When we recall our own memories,” Christopher Chabris and Daniel Simons wrote in a 2014 NYT op-ed, “we are not extracting a perfect record of our experiences and playing it back verbatim . . . We are effectively whispering a message from our past to our present, reconstructing it on the fly each time.”

In The Memory Illusion: Remembering, Forgetting and the Science of False Memory, Julia Shaw describes a 2015 study conducted as part of her doctoral dissertation at the University of British Columbia, Okanagan.  First, Shaw and her associates called up the parents of several dozen ordinary university students and asked them to recall a real event from their child’s adolescent years.  Subjects were asked if they remembered the event and almost all of them did.  Then things got interesting.

Subjects were told (falsely) that their parents had also described an upsetting incident involving behavior so troubling that the police were involved.  Initially, none of the study participants could recall anything of that sort, but they were assured that painful memories could be retrieved with sufficient reflection.  Subjects were asked to go home and give the matter more thought.

In the second interview, participants were told that if they imagine what the painful incident from our past might have been like, genuine memories may be triggered.

Shaw based her experiment on the Reid technique (for more on this watch John Oliver’s video clip). The Reid technique, a method for obtaining eye witness testimony and confessions, is employed almost universally in the United States and Canada (the difference being that, in Canada, interrogations are usually recorded).  This new approach was developed in the late 1950s and early 1960s by John Reid, a former Chicago cop. It was introduced as a humane alternative to old “third degree”, a polite synonym for torture. Prior to Reid, suspects were routinely beaten, scalded and electrocuted into compliance. At the time, the law enforcement community was convinced they couldn’t do their jobs properly if deprived of the third degree.  You couldn’t expect the criminal element to confess of their own good will.

Reid thought he could do better.  During an initial interview with a suspect, Reid looked for non-verbal indications of guilt.  Nervous suspects were considered guilty, especially if they picked at invisible lint on their clothes, looked down or shifted their eyes to the left.  There was no scientific basis for these supposed guilt-indicators, it was all based on Reid’s “professional experience”.

If this initial interview showed evidence of guilt, the gloves came off.  The investigator re-entered the room, standing over the suspect to project dominance and waving a case file supposedly brimming with air tight proof of guilt.  Protestations of innocence, attempted explanations, and requests to look at the file are batted away as the interrogator lays out his theory of the crime as revealed truth.  “You’re time to talk will come later,” suspects are told, “now you’re going to listen.”

Once the suspect stops asserting innocence, the investigator’s voice softens and the “minimization” process begins.  The legal consequences of the alleged crime are set to one side while the moral implications are minimized.  The 1962 edition of Reid’s textbook featured thousands of proposed minimizations, including this gem: “Joe, no woman should be on the street alone at night looking as sexy as she did. Even here today, she’s got on a low-cut dress that makes visible damn near all of her breasts. That’s wrong! It’s too much temptation for any normal man. If she hadn’t gone around dressed like that you wouldn’t be in this room now.”

None of this stands up to scientific scrutiny.

Saul Kassin, Distinguished Professor of Psychology at John Jay College of Criminal Justice in New York City, conducted a series of experiments in which randomly selected college students and trained police investigators were shown videos of actual inmates telling stories–half true, half false.  Participants were asked to detect who was lying.  Kassin found the officers, who focused on non-verbal clues, did less well than the students, but were far more confident in their conclusions.

There may be no scientific support for the Reid technique, but whether used on suspects or potential witnesses, it gets results.

And that’s the problem.

Julia Shaw’s 2015 experiment employed procedures suggested by the Reid technique.   Stephen Porter, Shaw’s Ph.D. supervisor, says they were expecting a modest (though statistically significant) results from their manipulations and were appalled by the actual results.  By the conclusion of the third session, 70% of subjects were recalling hideous, self-incriminating events complete with smells, emotional reactions and detailed physical descriptions.  Once these false memories had taken hold, moreover, it was often difficult to convince test subjects that they had been duped.

False memories feel just as real as true memories.  That’s why the confidence of the witness is a meaningless measure of veracity.  Here’s Julia Shaw’s takeaway:

It is crucial that the justice system becomes aware of overconfidence factors, memory illusions and the problems we have with identification, since these can lead to atrocious situations where we sometimes build cases based on nothing but air.

The Innocence Project reported in 2015 that, of the 325 cases in which defendants were exonerated by incontrovertible DNA evidence, 235 involved mistaken eyewitness testimony and a full 25% involved false confessions.

In his book Searching for Memory, Daniel Schacter recounts a well-known story about Ronald Reagan:

In the 1980 presidential campaign, Ronald Reagan repeatedly told a heartbreaking story of a World War II bomber pilot who ordered his crew to bail out after his plane had been seriously damaged by an enemy hit. His young belly gunner was wounded so seriously that he was unable to evacuate the bomber. Reagan could barely hold back his tears as he uttered the pilot’s heroic response: “Never mind. We’ll ride it down together.” The press soon realized that this story was an almost exact duplicate of a scene in the 1944 film A Wing and a Prayer. Reagan had apparently retained the facts but forgotten their source.

When Hillary Clinton or Brian Williams “remember” being under enemy fire, they aren’t necessarily fibbing to make themselves look good; in all likelihood they remember the bullets whirring past their ears.

Although we are constantly remembering things that didn’t happen, the consequences are usually harmless. But when a low-status individual is hauled down to the police station and subjected to a variation on the Reid technique, memory can easily be contaminated.  Witnesses who initially have nothing to offer eventually “remember” exactly what investigators are looking for.

In their groundbreaking The Science of False Memory, Charles Brainerd and Valerie Reyna show that when we process a past event we store and retrieve ‘verbatim’ memory (what the scene looked like) and ‘gist’ memory (the meaning we assign to the event) separately.  For instance, we can sometimes remember a person’s name without remembering what they look like, or vice versa.  Gist memory (what the event means to us) is usually stronger than verbatim memory (what it looked like).

This is why we often remember what was said–because it is significance to us–but attach it erroneously to a separate verbatim trace.  You might remember a conversation over a few beers (gist memory) but recall that it happened at the wrong bar (verbatim memory).  False memories are created when gist memory is associated with the wrong verbatim memory.  This is why memory is so easily distorted.  We are always confusing gist and verbatim memory traces without realizing we are doing it.

Forgetting is an important brain function.  If we didn’t slough off the inconsequential events of the average day we wouldn’t be able to concentrate on the task at hand.  

As Breynard and Reyna remind us in The Science of False Memory,

When a substantial period of time elapses between the occurrence of a crime and suspect identifications, there are opportunities for witnesses to be exposed to additional information that points to specific suspects but that is not part of their original experience.  Examples include conversations with other witnesses, media reports of the crime, and conversations with investigators.

Brainerd and Reyna show that verbatim face memory (the ability to identify a specific suspect) falls to near zero after a mere eight days while gist facial memory (it was a black guy) often remains relatively strong.

When an eye witness is confronted with an identification test a few days after observing activity, innocent suspects who share salient categories of appearance with culprits will seem familiar (which supports false identification), but verbatim traces of incidental features that differentiate culprits from such suspects are no longer readily accessible.

A month after an event, the authors say, verbatim memory is normally gone and gist memory is getting weak.  

And this is why innocent people, after hours, or even days, of non-stop interrogation, sometimes “remember” doing things they never did, feeling emotions they never felt, or seeing things they never saw. If juries could witness the entire process (which typically shows suspects vacillating between confession and denial) they might grasp the dangers inherent with the Reid method. But they don’t. They just see a signed statement or, a brief clip of audio or video in which the suspect confesses to the crime.

So, whether we’re talking about suspects admitting to a crime, or eye witnesses “remembering” what they saw a month or two after the event, the same psychological manipulation is at work. And that’s why so many innocent people confess to crimes they never committed.

One thought on “The Science of False Confessions

  1. In a rational world it would be recognized that prolonged interrogation in a small room with a bullying interrogator is in fact torture. Prolong the “interview” for as long as it taesk to get a confession and you will get a confession.

    In Japan supescts can be held in custody for 24 days and99 % of crimes in Japan are “solved” by confessions.

Comments are closed.