By Sara Zhang | WIRED | August 3, 2015
JAMES HOLMES WALKED into a midnight screening of The Dark Knight Rises in 2012 carrying three guns including a semi-automatic rifle and opened fire, killing 12 people and injuring 70 more. Nobody, not even his defense attorneys, denied that. But those attorneys still told a jury and a judge that Holmes was not guilty of those crimes—because he was insane. Last month, that jury rejected that assertion, finding Holmes guilty on all counts.
Holmes’ plea didn’t get him off, but it did get people talking about the insanity defense again. It’s a rare move for defense attorneys these days, even quaint sounding. Psychiatrists no longer call patients “insane.” It’s not a clinical diagnosis. Yet the term persists in the courtroom—along with many other practices unsupported by modern psychology and neuroscience.
Americans inherited a legal system shaped by history, not by science. “The legal system is resistant to change and resistant to paying attention to scientific research,” says Adam Benforado, a law professor at Drexel University and author of the recent book Unfair: The New Science of Criminal Injustice. The system assumes that innocent people don’t confess to crimes they didn’t commit. It presumes that eyewitness testimonies are reliable. It counts on the impartiality of jurors.
None of those things are borne out by evidence.
More than one in four people exonerated by DNA evidence initially confessed to the crime.
But why? Why confess if you’re innocent?
Interrogations are designed to mentally break down the guilty, but those psychological tricks leave the innocent vulnerable as well, says Saul Kassin, a psychologist at Williams College who has studied false confessions. “When people are asked. ‘why did you confess?’ they say, ‘I wanted to go home.’” he says. “That’s a testament to their state of mind.”
The problem begins with pre-interrogation interviews, where police are often trained to look for behaviors—like jittering or avoiding eye contact—that supposedly reveal lying. In fact, peer-reviewed research shows those behaviors are not reliable signs of lying at all.
By the time the police haul someone in for interrogation, they’re already assuming the suspect is guilty. They can then spend hours badgering the suspect before adopting a more sympathetic pose with minimizing remarks like “We think you were provoked” or “You just had too much to drink.” Courts have ruled police are not allowed to promise leniency—precisely because it leads to false confessions. But Kassin’s research shows that in a lab setting, suspects interpret those minimizing remarks as implicit promises of leniency.
A confession, true or false, ripples outward. Knowing that a suspect has confessed can sway the testimony of witnesses and forensic experts alike. (The unreliability of physical forensic evidence is a whole other can of worms.) It’s intuitively easy to believe a confession—less intuitive is understanding the situations that lead people to confess.
In 1987, Jennifer Thompson got on a witness stand and said she was “absolutely sure” Ronald Cotton had raped her three years earlier in Burlington, North Carolina. Eight years later, DNA evidence exonerated Cotton of the crime.
Despite Thompson’s certainty in court, she barely recognized Cotton when she first had to pick him out of photos and a line-up. She would later describe how the detectives’ efforts to reassure her—“You did great” and “It’s the same person you picked from the photos”—made her confident in her decision. When she met the real perpetrator, she didn’t recognize him at all, remarking at the time, “I have never seen him in my life.”
Thompson’s faulty recollection is not surprising given what neurobiologists know about memory. Every act of recall involves a process called reconsolidation, the recreation of the memory. On a molecular level, reconsolidating a memory is much like forming a new one. In other words, memories are just reconstructions. That makes them startlingly easy to change.
It doesn’t even take an especially leading question to implant false details. Elizabeth Loftus, a psychologist who has done groundbreaking research on eyewitness memory, conducted a study where participants watched a filmed car crash. A week later, witnesses asked “How fast were the cars going when they smashed into each other” were more likely to report seeing broken glass than those asked the more neutral, “How fast were the cars going when they hit each other.” There was no glass.
One relatively easy way for the legal system to minimize the problems of reconsolidation would be double-blinding police line-ups. If the police officer present doesn’t know who the suspect is, he or she can’t unknowingly drop hints or make remarks that could influence an identification, like the things police officers said to Thompson.
Here’s a radical idea: Courtrooms should be virtual, with lawyers, witnesses, and defendants replaced by avatars. Benforado makes this suggestion in Unfair to illustrate how deeply biases run.
Race and gender are obvious sources of bias—so obvious that courts have ruled jurors can’t be rejected on the basis of race or gender alone. Lawyers, however, can easily circumvent that by offering up alternative and neutral-sounding excuses for rejecting a potential juror. Over 20 percent of African Americans executed in the US were convicted by all-white juries.
And plenty of biases aren’t covered by law at all—for example, attractiveness. Jurors in a simulated courtroom, specifically men, are more likely to be biased against overweight female defendants. “We are not the people that we wish we were,” says Benforado, “The people we wish we were are kind of like computer robot humans who are able to control their biases.”
One of the best examples of how biases change outcomes is date rape. Legal standards, such as whether mistaken belief of consent is allowed as a defense, vary from jurisdiction to jurisdiction. So it would make sense that outcomes in date rape cases would also vary. But no. Rather, the demographics of the jury are more important. Older, conservative women are more likely to acquit men in these cases than younger, liberal women. Research into cultural cognition shows that a person’s values and thus judgments are inextricably tied to his or her cultural background. Of course, jury consultants know all this, and they select jurors to stack the deck for one side. The legal system might not have metabolized the latest science, but for-pay consultants have.
The Insanity Plea
In the 46 states that accept a version of “insanity” as a legal defense, the word has a very particular legal definition. But fundamentally, to be found not guilty of a crime by reason of insanity, the accused has to be shown to be unable to tell right from wrong.
That puts forensic psychiatrists and juries in the difficult position of reconstructing a defendant’s mental state during the time of the crime. “There’s no reason for us to have strong confidence in our view,” says William Carpenter, a psychiatrist who testified for the defense in the trial of John Hinckley, Jr., found not guilty by reason by insanity for attempting to assassinate President Ronald Reagan. “Reasonable people would and did disagree over whether Hinckley was legally sane,” says Carpenter.
In fact, in the Holmes case, court-appointed psychiatrists and the defense’s experts disagreed about his legal sanity. Now, they did agree that he was schizophrenic. Psychiatric diagnoses are based on criteria laid on out in the Diagnostic and Statistical Manual of Mental Disorders, the so-called bible of psychiatrists. The DSM lays out the diagnostic criteria for schizophrenia: symptoms such as delusions, hallucinations, and disorganized speech have to be present for at least six months. But the word “insane” does not appear. It’s not a medical term, but it is still a legal one.
Insanity pleas are rare—in part because they rarely work. If anything, the insanity plea represents a humane impulse to treat rather than punish the mentally ill. But no brain scan or blood test or even DSM checklist can tell if someone is sane by law. The legal system has generated its own workable definition. But in other instances, it has a lot of catching up to do.
Read the full article originally published by WIRED.