The Memory Doctor
She was fascinated. But what excited her most was watching the process unfold in her own hands. She was given a rat and a cage—a "Skinner box"—in which to train it. By selectively administering food during a series of repetitions, she taught the rat to look at, then approach, then press a lever. By the time she was done, the rat had learned to run straight to the lever as soon as it was put in the cage. She had made the animal do her bidding.
In 1966, she entered Stanford's graduate program in mathematical psychology. She might as well have walked into a men's locker room. She was the only woman admitted to the program that year. All her professors were men. Her classmates unanimously voted her least likely to succeed as a psychologist. They placed bets on when she'd quit. The betting pool turned out to be a damning test of mathematical psychology. Its abstract equations, designed to explain and predict behavior, couldn't account for the particulars of this young woman. She had lost her home and her mother. Compared with that, exams were easy.
She soon outgrew the boys' game. The deeper she waded into mathematical psychology, the less she liked its simplifications. People were more complicated than that. So was she. In her second year at Stanford, she was assigned to mentor an incoming student. She married him instead. On June 30, 1968, she became Elizabeth Loftus.
She thought she had found the love of her life. She would serve her husband's career, just as her mother had done for her father. But then she fell in love again. Not with a person, but with a field of study: memory.
Part III: Leading the Witness
In 1968, Elizabeth Loftus discovered what she wanted to do with her life. She wanted to experiment on people.
Loftus was 24. For several years, she had studied psychology and helped her professors with their research. She had been a cog in the academic system. But now, in her third year of graduate school at Stanford, she was finally getting a chance to design, run, and analyze her own experiments. She knew the thrill of training a rat to press a lever. But this was different. Now she was working with human beings.
Her topic was semantic memory. She was trying to find out how people's brains stored and retrieved words. She couldn't see inside their heads, but she could administer inputs and measure outputs, as she had done with her rat. The inputs were questions, and the output was response time. Sometimes she asked her subjects to name a "yellow fruit." Sometimes she asked them to name a "fruit that is yellow." On average, they answered the latter question a quarter of a second faster than the former. From this, she drew an inference: The brain organized such information by the noun, not the adjective.
Loftus loved the whole thing: conceiving the experiment, trying it out on people, measuring their performance, drawing conclusions about the mind. But not everyone was impressed. Shortly after earning her Ph.D., she had lunch in New York with her cousin, a lawyer. When her cousin asked about her work, Loftus proudly told her about the yellow-fruit study. Her cousin dryly asked how much it had cost the government.
The conversation stung Loftus. She was running her own memory experiments, but they were just about words. Why did it matter how people recalled yellow fruit? Wasn't there something more worthwhile to study?
What did she really care about? As an experimental psychologist, she decided to answer the question by studying her own behavior. What did she talk about when the topic was hers to choose? What did she like to bring up at parties? The answer was crime. She loved books, movies, TV shows, and news stories about it. Maybe she could become a crime expert. She could use the science of memory to help the justice system.
The first step was to find a project somebody would pay for. The Department of Transportation was offering money to study car accidents. Accidents weren't crimes, but they involved eyewitnesses, so she started there. She showed people films of collisions and quizzed them about what they had seen. Sometimes she asked how fast the cars had been going when they "hit" or "contacted" each other. Sometimes she asked how fast they had been going when they "smashed" into each other. The "smashed" question produced estimates 7 miles per hour faster than the "hit" question and 9 miles per hour faster than the "contacted" question. The questions were skewing the answers.
In another experiment, she showed her subjects a multi-car collision and asked some of them, "Did you see a broken headlight?" She asked others a leading version of the question: "Did you see the broken headlight?" Of the six questions posed in this dual format, three referred to things that weren't in the film. Compared with subjects who heard the question with an "a," those who heard it with a "the" were twice as likely to say they had seen a bogus item. (To experience one of Loftus' traffic experiments, click on the adjacent
Still, that was just lab work. Loftus wanted to get involved in a real court case. In 1973, after moving to the University of Washington, she called up the Seattle public defender's office and volunteered to help as a memory expert. In exchange, she got to watch the case unfold. It was a murder trial that hinged on conflicting memories over how much time had elapsed for premeditation. It ended in acquittal.
Loftus packaged her memory expertise with her accident studies in an article for Psychology Today. She challenged the reliability of eyewitness testimony, mentioned her work in the Seattle murder case, and noted that the defendant had been acquitted. It was practically an advertisement. Attorneys read the article and picked up their phones. Her career in legal consulting was launched.
She was exactly what defense lawyers needed. The chief threat to their clients was incriminating witness testimony. Loftus could shake the jury's faith in such recollections without attacking the witness personally. Memory errors were natural. The witness, like the defendant, was innocent. Even police, who caused misidentifications by contaminating witnesses' memories with mug shots and lineups, often didn't realize what they were doing.
Over the next 35 years, Loftus testified as a memory expert in more than 250 hearings and trials. She worked dozens of famous cases: Ted Bundy, O.J. Simpson, Rodney King, Oliver North, Martha Stewart, Lewis Libby, Michael Jackson, the Menendez brothers, the Oklahoma City bombing, and many more.
Why did a woman who had endured assault in her own life defend accused predators? Part of it was the structure of criminal law: Her work created doubt, and doubt was an ally of the defense. Part of it was her empathy for the accused: She had always been suspicious of criminal allegations and lenient toward small-time offenders. And where empathy failed, scientific rigor took over. Memory's fallibility was a fact. By testifying to that fact, she believed she was serving justice.
Loftus alters Lesley Stahl's eyewitness memory
Her job was to explain how memory errors could contaminate eyewitness testimony. For example, when eyewitnesses were shown lineups of possible culprits, they sometimes selected a face that was familiar for a different reason. Loftus demonstrated this by showing experimental subjects six photos while they listened to a crime story. One photo depicted the culprit; the others depicted innocent characters in the story. Three days later, she showed the same subjects a photo of one of the innocent characters along with three photos of other people. From these four pictures, she asked them to pick the criminal. Twenty-four percent of the subjects correctly refused to pick a photo. Sixteen percent picked one of the three new photos. Sixty percent picked the photo of the innocent character. They remembered it from the crime story but confused it with the perpetrator.
Police lineups worsened this confusion. In another experiment, after watching a mock crime, subjects were offered a lineup that didn't include the perpetrator. One-third of them picked somebody anyway. But when the cops conveyed extra confidence—"We have the culprit and he's in the lineup"—78 percent of the subjects picked somebody.
Then there were prosecutions based on coached child testimony, such as the McMartin Preschool sex-abuse case. To measure children's suggestibility, Loftus and a colleague showed them several one-minute films, followed by leading questions. "Did you see a boat?" they asked one child. Afterward, the child remembered "some boats in the water." "Did you see some candles start the fire?" they asked another. "The candle made the fire," the child said later. Other kids, after being asked about bees and bears, recalled bees and bears. None of these things—bees, bears, boats, candles—were in the films.
Not even Loftus was immune to suggestion. In 1988, after 13 years of testifying about memory's fallibility, she was told by her uncle that she was the one who had found her dead mother in the swimming pool. The sights and sounds of that awful morning came back to her—the corpse face down, the nightgown, the screaming, the stretcher, the police cars. But within three days, her uncle recanted the story, and other relatives confirmed that her aunt, not Loftus, had found the body. The memories of the memory expert were false.
The incident strengthened Loftus' conviction that such recollections shouldn't be trusted in court. The more cases she saw, the more passionate she became about her work. She saw herself as Oskar Schindler, rescuing as many innocent souls as she could. "The beauty I find in helping the falsely accused is something I like about myself," she wrote in an essay years later. "It's the deeper part of who I am."
The passion and the work took their toll. In court, she endured cross-examination and vilification. And at home, her husband gave up competing for her attention. In 1991, they divorced.
But by then, she had a bigger problem.
Part IV: The Recipe
In the summer of 1990, Elizabeth Loftus got a phone call from an attorney in San Francisco. A man named George Franklin had been charged with murdering a child, based on the recollection of his daughter, Eileen. Loftus, a psychologist, had testified in dozens of cases about the fallibility of eyewitness memory. But this case was different. The murder had happened 21 years earlier. Eileen's purported memory, however, was less than a year old. According to the prosecution, she had repressed it.
Repressed? How could such a crucial memory vanish for 20 years, leaving its owner completely unaware of its existence, and then resurface in full color? Loftus had her own bad childhood memory—being molested when she was 6—but she had never purged it. She searched the psychological literature and found no basis for the repression theory. George Franklin's attorney had a different theory: Eileen Franklin had never seen the crime. In her head, she had blended details of the murder, as it was reported in the press, with an imaginary picture of her father doing it. She had developed a false memory.
At the trial, Loftus explained how memories naturally eroded over time and became susceptible to distortion. She told the jury about an experiment in which she had shown people a video of a robbery and shooting. After the video, the viewers had watched an erroneous television report about the shooting. When they were asked afterward to describe the incident, many of them blended false details from the television report with their recollections of the video. They clung to these altered memories even when the experimenters suggested that they might be mistaken. Something like that must have happened to Eileen Franklin.
In previous cases, such testimony had swayed juries. But not this time. The prosecutor forced Loftus to admit that she had never studied memories like Eileen Franklin's. Loftus had proved that people could misidentify random perpetrators, not that they could mistakenly accuse their own fathers. She had proved that memories could be altered, not that they could be wholly invented. Her work seemed irrelevant. In November 1990, George Franklin was convicted.
The nightmare was just beginning. Repressed memories were surfacing everywhere. In June 1991, Marilyn Van Derbur, a former Miss America, told the world that at age 24, she had discovered her father's sexual abuse of her as a child. Later that year, Roseanne Barr claimed to have recovered 30-year-old memories of both parents molesting her. ("He used to chase me with his excrement and try to put it on my head," she said of her father.) Women were suing their parents for millions of dollars. Hundreds of accused families sought help.
If these memories weren't real, where were they coming from? Eileen Franklin claimed that her memory had surfaced during hypnosis, therapy, a dream, or a flashback. Barr said hers had emerged during individual and group therapy.
Loftus began to read popular books that told women and therapists how to recover memories of sexual abuse. The books urged therapists to ask their clients about childhood incest. They listed symptoms that supposedly indicated abuse even if it wasn't remembered. They invited women to search for memories by imagining the abuse. They encouraged group therapy in which women could hear one another's stories of being victimized.
These ideas sounded fishy. Suggestion, indoctrination, authority, inference, imagination, and immersion were known to alter memories in police interrogations and experiments. But could they create a whole memory? Could the recent surge of incest recollections be the product of recovered-memory therapy?
To find out, Loftus went to a talk by George Ganaway, a respected psychiatrist, at the American Psychological Association's annual meeting in August 1991. Armed with case studies, Ganaway argued that "iatrogenic implantation"—implantation by therapists—was creating false memories of satanic ritual abuse.
Loftus suspected the same phenomenon was creating incest memories more broadly. But how could she expose it? In her book, The Myth of Repressed Memory, she described her next thought:
"While I couldn't prove that a particular memory emerging from therapy was false, perhaps I could step around to the other side of the problem. Through careful experimental design and controlled studies, perhaps I could provide a theoretical framework for the creation of false memories, showing that it is possible to create an entire memory for a traumatic event that never happened."
This was a pivotal decision. Loftus wasn't a detective. She was a designer of experiments. She couldn't start with seemingly recovered memories and demonstrate that they were false. But she could start with false memories and demonstrate how they were seemingly recovered.
Piece by piece, she analyzed and incorporated her adversaries' methods. For example, she noted in her book, "to parallel the therapeutic process, the memory had to be implanted by someone the subject trusted and admired, either a relative, friend, or a respected authority figure." That insight led to what became known as the "lost-in-the-mall experiment." (For more on the genesis of this experiment, see Part 2.) Each subject was given summaries of four incidents from his childhood. Three stories were true; one was false. The false story followed a formula: You got lost in a mall or department store, you cried, you were found by an old person. The summaries were written with the help of older relatives who knew the true incidents and the family. One woman, for example, was falsely told:
You, your mom, Tien, and Tuan all went to the Bremerton K-Mart. You must have been 5 years old at the time. Your mom gave each of you some money to get a blueberry Icee. You ran ahead to get into the line first, and somehow lost your way in the store. Tien found you crying to an elderly Chinese woman. You three then went together to get an Icee.
Loftus speaking at the Beckman Center, March 2007
The subjects were told that their relatives had recalled all four incidents. They were asked to fill in the details of each incident or, if they couldn't remember it, to write, "I do not remember this." In follow-up interviews, they were asked to think more about each incident and to retrieve any additional details they could recall. Of the 24 people subjected to this procedure, six came to remember the fake story as true.
From this experiment, Loftus began to sketch what she called a "recipe" for planting memories. First, you needed the subject's trust. A therapist had that; so did a family member. Then, by suggesting that the incident might have happened, you planted a seed. The subject would think about it, and the idea, if not the scene, would start to become familiar. The people and places mentioned—Tien, blueberry Icees, the Bremerton K-Mart—would evoke real memories, and these would begin to blur with the suggested scenario. By coaxing the subject to imagine the scene, you could accelerate this confabulation. Gradually, she would add details, seizing authorship of the story and securing its authenticity. The fabrication was out of your hands now. The memory was hers.
Loftus speaking at the Beckman Center, March 2007
This recipe was what the incest-survivor books were unwittingly teaching. It was what the recovered-memory therapists, with equal folly, were practicing. They hooked their readers and clients with checklists of supposed symptoms: headaches, guilt, low self-esteem, fear of darkness. Then they induced collaboration. "Let yourself imagine or picture what might have happened to you," said one book. "Occasionally you may need a small verbal push to get started. Your guide may suggest some action that seems to arise naturally from the image you are picturing." The guide, a therapist, supplied personal knowledge to help the process along. Group therapy helped, too. The more incest memories a woman heard, the more plausible her own victimization became. The more images she absorbed, the easier it was to picture the scenes she had repressed.
Will Saletan covers science, technology, and politics for Slate and says a lot of things that get him in trouble.