In 1994, psychologist Elizabeth Loftus was sitting in her office listening to a person who claimed to remember getting lost in a shopping mall at the age of five, crying, and then being found by an elderly man in a plaid shirt. The person spoke confidently. He even remembered that the shirt was plaid.
There was just one problem: none of it had happened.
This was not a patient with psychosis, nor was it a hoax. It was a participant in an experiment whose older brother — at Loftus’s request — had told him a “family story” a few weeks earlier about how little [name] got lost in a shopping mall. The brother told it sincerely, with details, with dates. He himself believed what he was saying, because he had just been given the story in writing, along with instructions to “retell it as your own.”
The person remembered what had been described to him. And the memory stuck.
This episode has a long backstory, and it begins with a class assignment.
Jim Coan, one of Loftus’s students, was writing a term paper in cognitive psychology. He had to come up with something that would test the limits of memory. He asked his mother, sister, and brother to help: tell the “youngest” sibling several real childhood stories, and one fake one, invented by Coan himself. The trick was that the “youngest” sibling was Coan himself, while the storytellers did not know which story was false.
At the debriefing, Coan’s brother could not distinguish the false event from the real ones. More than that — he did not believe it when he was told that one of the events had been invented. He remembered something that had never happened because that was how it had been told to him.
The student term paper became the prototype for a method that, a few years later, appeared in Psychiatric Annals as the article “The Formation of False Memories” (Loftus & Pickrell, 1995). Twenty-four participants. Four events, three of them genuine — confirmed by a relative — and one constructed. Roughly 25% reported remembering the false event.
The figure “25%” sounds modest in the report, but there is a detail that says more: when participants were told at the end of the experiment that one event was false, five out of twenty-four could not identify the fake — and pointed to one of the real events instead. In other words, their ability to distinguish reality from suggestion failed in the opposite direction from the one expected.
Loftus herself called it an “existence proof.” Not “all memories are false,” but “look, a person can remember an event that never happened, and you can make that happen in a laboratory in a few weeks.”
Then comes the most elegant and the most unpleasant part.
In 2002, Kimberly Wade, Maryanne Garry, J. Don Read, and D. Stephen Lindsay published a paper in Psychonomic Bulletin & Review with a title that sells itself: “A Picture Is Worth a Thousand Lies.” Twenty participants. Three interviews, spaced a week to a week and a half apart. And a photograph: the participant — a child aged four to eight — had been digitally inserted into the basket of a hot-air balloon.
The image was made in Photoshop 4.0 on a Macintosh G3. The event was not chosen at random: hot-air balloon rides in New Zealand were accessible enough — festivals, operators — but distinctive enough that relatives would know for certain whether they had happened or not.
The result: 50% of participants had “remembered” the flight by the third interview — fully or partially. Details accumulated from one interview to the next. At the first interview, one participant said he remembered absolutely nothing and “that’s me in the photo, but I have no memory of it,” yet two meetings later he was discussing what year it had been, who was there, and how his mother had stood below with a camera.
The most interesting line in the paper is not the percentages, but the explanation the participants gave themselves. They said: “Well, it’s a photograph, so it must have happened.” Not “I remember,” but “there is evidence, so there has to be a memory.” The photo worked not as a memory, but as permission to remember.
And here lies the elegant twist. The authors broke down the accounts into clauses and counted: less than 30% of the details in the “memories” came from the photograph itself. The rest came from guided imagery, from imagination, from real events that happened to settle alongside it. The photograph did not draw the memory. It planted a seed and gave the participant permission to build the rest on their own.
There is a follow-up that breaks intuition even further. In 2005, Garry and Wade published a paper with the almost mocking title “Actually, a Picture Is Worth Less Than 45 Words” — and showed that narratives produce more false memories than photographs. Forty-five words of text turn out to be stronger than an image.
The study by Porter, Yuille, and Lehman (1999, Law and Human Behavior) approached it from another direction. Seventy-seven participants. Not a shopping mall, not a balloon ride — stressful events: a serious animal attack and the like. Questionnaires were sent to parents; real events were taken from them, and false ones were constructed.
Then came three sessions. Guided imagery, repeated retrieval attempts, gentle pressure: “many people cannot remember right away, think about it some more.” Twenty-six percent “recovered” a complete memory of the false event. Another thirty recalled individual aspects.
This was no longer “a quarter of students.” It was half the sample offering a partial or complete account. And, crucially, the paper does not stop at the number: the authors compared genuine, implanted, and fabricated memories — participants who had already “remembered” the false event were also asked to lie deliberately. Differences emerged in confidence, vividness, number of details, consistency of details, coherence, and level of stress. In other words, the inner texture of a suggested memory is its own thing, distinct from lying — and that is perhaps the most unpleasant part for a courtroom.
There is also a 2000 paper in Psychological Science — about who, exactly, gives in. It turned out that it was not just about the “victim”: participants with memory distortion scored higher on the dissociation scale, but the pairing also mattered. High interviewer extraversion and low participant extraversion — and the likelihood of distortion rises. Suggestion is not a property of a person; it is a property of a conversation. Which, if you think about it, is far more disturbing than “some people are suggestible.”
Here we need to stop and do what popular retellings usually do not: ruin a beautiful story with numbers.
In 2017, Scoboria and colleagues assembled a major review of three paradigms — imagination inflation, false feedback, and memory implantation (Applied Cognitive Psychology, 31(1):2-23). Their conclusion: in implantation studies, some kind of recollective experience arises in an average of 47% of participants, but only around 15% can be considered full-fledged memories. The authors’ wording: susceptibility to false memories of childhood events “appears more limited than previously thought,” and the ratings assigned by researchers do not necessarily match participants’ self-reports.
There is a second qualification as well — from Pezdek and Hodge (Child Development, 1999). They replicated getting lost in a shopping mall, and among children it worked even better. But a false event involving a painful and humiliating enema — almost not at all. Their conclusion was not “you can mold memory into anything,” but “the plausible can be implanted; the implausible cannot.” The difference is fundamental: it means that “a person remembered being lost in a shopping mall” does not automatically entail “a person can remember abuse that never happened.”
And this is where the real war begins, because the question stops being academic.
In the nineties, America was going through an epidemic of “recovered memories”: therapy returned images of childhood abuse to patients, families fell apart, lawsuits followed. Loftus became the leading expert for the other side — and she was called into cases where the stakes were absolute: Ted Bundy, the McMartin cases, the Menendez family, Oklahoma City, later Cosby, Sandusky, Michael Jackson. She joined the scientific advisory board of the False Memory Syndrome Foundation — an organization that critics regarded as created to defend parents accused of abuse.
The criticism targeted both the method and the ethics. Crook and Dean (1999, Ethics & Behavior) accused the research of being unethical and inapplicable to real trauma; Loftus replied that their analysis contained “exaggerations, omissions and errors,” and stressed: the study does not claim that all memories of abuse are false — it shows how easily memory can be manipulated when an older relative “confirms” the event. In 2019, Crook and McEwen added a legal argument, citing Loftus’s 2017 deposition, in which she acknowledged that no safeguard had been built in to rule out the possibility that participants might genuinely have gotten lost without their relatives knowing.
The cruelest episode was Jane Doe. In 1997, psychiatrist David Corwin published in Child Maltreatment a case of a “recovered” memory of abuse that became a textbook example. Loftus and Melvin Guyer had doubts, found the mother, dissected the story piece by piece, and published their analysis in Skeptical Inquirer in 2002. And the woman whose life they had taken apart as a case study filed suit — Taus v. Loftus — for defamation and invasion of privacy.
This is where the point lies that is hard to fit into applause. A method invented to protect innocent people from false accusations once ran roughshod over someone who had not chosen to be an example. Science and its price ended up in the same story.
Memory is not a tape recorder. It is a construction kit that reassembles the past anew each time for the task at hand and — crucially — does not label its blueprints. From this come two equally wrong conclusions.
The first: “if memory is constructed, then no memory can be trusted.” No. The construction kit works with real parts, and most of what it assembles corresponds to what happened.
The second: “I was told, therefore it happened.” Also no. Suggestion requires conditions — plausibility, the authority of the storyteller, repeated attempts to remember, guided imagery, “evidence” such as a photograph. Remove the conditions, and the effect collapses, as with the enema.
The practical conclusion is boring, and therefore reliable: look not at the vividness of a memory, but at its source. Vividness can be cranked up through repetition; a source either exists or it does not. What someone remembers with a tremor in their voice and what someone was told with confidence look identical from the outside. That is exactly why Loftus spent forty years going to court.
Sources: Loftus & Pickrell, Psychiatric Annals 25(12), 1995 · Wade, Garry, Read & Lindsay, Psychonomic Bulletin & Review 9(3), 2002 · Porter, Yuille & Lehman, Law and Human Behavior 23(5), 1999 · Porter, Birt, Yuille & Lehman, Psychological Science, 2000 · Pezdek & Hodge, Child Development 70(4), 1999 · Scoboria et al., Applied Cognitive Psychology 31(1), 2017 · Garry & Wade, Psychonomic Bulletin & Review 12, 2005 · Crook & Dean, Ethics & Behavior, 1999 · Crook & McEwen, 2019 · Blizard & Shaw, Journal of Child Custody, 2019 · Bernstein, Laney, Morris & Loftus, PNAS 102(39), 2005 · Wiki: Lost in the mall technique, Elizabeth Loftus, Jane Doe case.