5. The Sin of Suggestibility.

The Sin of Suggestibility.

SHORTLY AFTER an EI AI cargo plane took off from Schiphol Airport in
Amsterdam on October 4, 1992, two engines failed and the pilots attempted
to return to the airport. They never made it back: the plane crashed into an
eleven-story apartment building in a southern suburb, killing thirty-nine
residents and all four members of the airline crew. Reporters and television
cameras descended on the chaotic scene, and the tragedy dominated news
in the Netherlands for days. People throughout the country saw, read,
heard, and talked about the catastrophe.
Ten months later, a group of Dutch psychologists probed what members of their university communities remembered about the crash. The researchers asked a simple question: "Did you see the television film of the
moment the plane hit the apartment building?" Fifty-five percent of respondents said "yes:' In a follow-up study, two-thirds of the participants
responded affirmatively. They also recalled details concerning the speed
and angle of the plane as it hit the building, whether it was on fire prior to
impact, and what happened to the body of the plane right after the collision. These findings are remarkable because there was no television film of
the moment when the plane actually crashed.
The psychologists had asked a blatantly suggestive question: they implied that television film of the crash had been shown. Respondents may
have viewed television footage of the postcrash scene, and they probably
read, imagined, or talked about what might have happened at the moment
of impact. Spurred on by the suggestive question, participants misattributed information from these or other sources to a film that they never
watched.
In 1997, a PBS television audience watching a Scientific American Frontiers documentary on memory hosted by Alan Aida learned that something
similar can result from a seemingly innocent everyday activity: looking at
photographs. In collaboration with the show's producers, I devised a memory experiment based on my laboratory's recent research; AIda was the
naive but willing subject. We met at a park in Brookline, Massachusetts, on
a sunny fall morning; the camera rolled as we sat on a bench in front of a
young man and woman about to begin a staged picnic. AIda knew they
were actors, and suspected that his memory would be tested. He paid close
attention as the pair enjoyed a drink, put on sunscreen, combed their hair,
ate a sandwich, took a picture, and did various other things that people
would be more or less likely to do while enjoying a picnic on a beautiful day.
Two days later, we met in my Harvard office. I showed photographs of
the bucolic scene to AIda, asking him only to judge the esthetic quality of
each one. He quickly sensed that I was up to something. When he saw a
photo of the actors eating potato chips - and failed to remember any potato chips at the picnic - he identified the experiment's key feature. Some
photos showed events that had actually occurred at the picnic, whereas
others were visual suggestions: they showed actions that might have happened, but did not. He wondered aloud whether we had set out to play
tricks on his memory.
After finishing up with the photos, I read out a series of objects and
actions, and instructed AIda to say "yes" when he remembered that an item
had occurred at the picnic. He had to be careful, I warned, because - just
as he suspected - some of the items had appeared only in the photos he
saw a few minutes ago, and were not part of the actual picnic. Despite
his skepticism and generally accurate memory, AIda soon stumbled: he
claimed to remember seeing the young woman file her nails at the picnic,
but she had done so only in a photo he had seen minutes earlier. He stumbled again moments later when he recalled a bottle of water at the picnic
that had appeared only in a photo. AIda accepted his erroneous recollection
with characteristic good humor, and I assured him that his experience was
common.
Suggestibility in memory refers to an individual's tendency to incorporate misleading information from external sources - other people,
written materials or pictures, even the media - into personal recollections. Suggestibility is closely related to misattribution in the sense that the
conversion of suggestions into inaccurate memories must involve misattribution. However, misattribution often occurs in the absence of overt suggestion, making suggestibility a distinct sin of memory.

Suggested memories can seem as real as genuine ones. On May 31,
a front-page story in the New York Times described the baffling case
2000,
of Edward Daly, a Korean War veteran who made up elaborate - but
imaginary - stories about his battle exploits, including his involvement in
a terrible massacre in which he had not actually participated. While weaving his delusional tale, Daly talked to veterans who had participated in the
massacre and "reminded" them of his heroic deeds. His suggestions infiltrated their memories. "I know that Daly was there," pleaded one veteran.
"I know that. I know that."
Suggestibility is worrying for multiple reasons: leading questions can
contribute to eyewitness misidentifications; suggestive psychotherapeutic
procedures may foster the creation of false memories; and aggressive interviewing of preschool children can result in distorted memories of alleged
abuses by teachers and others. The stakes are high in these cases for affected
individuals, so understanding and countering suggestibility is just as important for addressing social and legal concerns as it is for furthering psychological theory.

INFLUENCING EYEWITNESSES.

In the Dutch study of EI AI crash memories, researchers provided factually
incorrect information regarding the existence of film that captured the
plane colliding with an apartment building. In so doing, they followed a
procedure pioneered by the University of Washington psychologist Elizabeth Loftus which has since been used in numerous subsequent laboratory
studies. People first witness an everyday event on slides or videotape, then
answer a question containing misleading suggestions about the event, and
finally take a memory test that probes their recollections of the original incident. For example, in one recent study by the psychologist Philip Higham
at the University of British Columbia, participants viewed videotape of a
staged robbery at a convenience store. They were then given misleading
suggestions about an item of clothing worn by the store attendant, and
later tried to remember the attendant's clothing and other details of the
scene.
Describing the procedure to a psychology class from his own memory,
a research student who had helped to carry out the project explained that
in the video the attendant wears a white apron. He confidently elaborated
on the details of his recollection to convey to the students exactly what happened. Much to his shock, the student soon discovered that he had unwittingly demonstrated the power of the effect he was investigating. The attendant did not wear a white apron in the video: it had been suggested later by
the experimenter.
Previous experiments had shown that suggestive questions produce
memory distortion by creating source memory problems like those in the
previous chapter: participants misattribute information presented only in
suggestive questions about the original videotape. Higham's results provide
an additional twist. He found that when people took a memory test just
minutes after receiving the misleading question, and thus still correctly recalled that the "white apron" was suggested by the experimenter, they
sometimes insisted nevertheless that the attendant wore a white apron in
the video itself. In fact, they made this mistake just as often as people who
took the memory test two days after receiving misleading suggestions, and
who had more time to forget that the white apron was merely suggested.
The findings testify to the power of misleading suggestions: they can create
false memories of an event even when people recall that the misinformation was suggested.
The results have potentially important implications for police interrogations of eyewitnesses, because they imply that when suggestive questioning is used, memories for an original event may be altered even when people realize that the interrogator mentioned a critical bit of information.
Although few data are available regarding the extent of suggestive questioning of eyewitnesses, a British study using actual interviews indicates
that approximately one of every six questions that police posed to eyewitnesses was in some way suggestive.
In the numerous misinformation studies based on the work of Elizabeth Loftus, memory distortion results from suggestions that provide blatantly inaccurate information, such as a nonexistent white apron. But even
more subtle suggestions that do not contain specific inaccuracies can also
influence eyewitness testimony. Consider this excerpt from a Missouri case: Oh, my God ... I
EYEWITNESS TO A CRIME ON VIEWING A LINEUP:
don't know ... It's one of those two ... but I don't know ... Oh, man
... the guy a little bit taller than number two ... It's one of those two,
but I don't know.
EYEWITNESS THIRTY MINUTES LATER, still viewing the lineup and
having difficulty making a decision: I don't know ... number two?
OFFICER ADMINISTERING LINEUP: Okay.

DEFENSE ATTORNEY, MONTHS LATER AT TRIAL:
You were positive it was number two? It wasn't a maybe?
EYEWITNESS: There was no maybe about it ... I was absolutely positive.
The eyewitness spent about thirty minutes viewing a lineup of four
people while trying to identify her attacker from among them. She expressed considerable doubt while making her choice, but later at trial
disavowed that she had experienced even a hint of uncertainty. The psychologist Gary Wells wondered whether confirming feedback from the
administering officer - simply saying "okay" - served as a suggestive
procedure, increasing the witness's confidence in her memory. If so, the implications for courtroom testimony would be significant: an eyewitness's
level of confidence is the single most important determinant of whether a
jury believes that the witness has identified a suspect correctly. When confronted with a highly confident eyewitness, juries tend to focus more on
that person's believability than on the original witnessing conditions that
may have made it difficult for the witness to perceive or identify the perpetrator. Even though juries believe confident witnesses more than uncertain
ones, eyewitness confidence bears at best a tenuous link to eyewitness accuracy: witnesses who are highly confident are frequently no more accurate
than witnesses who express less confidence. To make matters worse, eyewitness confidence can be inflated when a witness is told that another witness
identified the same suspect, or when witnesses rehearse their testimony repeatedly during trial preparations. Clearly, eyewitness confidence is not
set in stone at the time an event occurs. But is it so malleable that seemingly innocuous confirming feedback - a mere "okay" - can inflate it
significantly?
To find out, Wells and Amy Bradfield showed people a security video
in which a man enters a Target store, and told them that in the moments
following the scene they witnessed, the man murdered a security guard.
The subjects then tried to identify the gunman from a set of photos - even
though the actual gunman did not appear in any of the photos. Some participants then received confirming feedback: "Good, you identified the actual suspect." Others received no feedback, and still others received disconfirming feedback that the suspect was in one of the photos they had not
selected. Finally, all subjects assessed how well they had been able to view
the suspect, and evaluated the certainty, clarity, and other features of their
memories.
Compared with those who received disconfirming or no feedback,
people who received confirming feedback claimed higher confidence and
trust in their memories, a better view and clearer recollection of the gunman, and heightened recall of facial details. There was, of course, no basis
for these claims: subjects in all three conditions had the same opportunity
to perceive and remember the gunman. Despite the fact that the witnesses
were dead wrong, their confident assertions of a good initial view of the
suspect and a clear, detailed recollection would have been extremely convincing to a jury.
These findings are especially important in light of the legal criteria for
evaluating the validity of eyewitness reports. Faced with evidence that suggestive questioning could influence eyewitness testimony, in 1972 the Supreme Court ruled in Neil v. Biggers that such procedures do not necessarily disqualify an eyewitness account if there are grounds to believe that the
report is fundamentally accurate. The Biggers criteria hold that the probable accuracy of an eyewitness account depends on the witness's certainty,
ability to describe the suspect, and initial opportunity to view and pay attention to a crime (as well as on the amount of time between the incident
and attempted identification). As Wells and Bradfield point out, however,
their results show that confirming feedback, for instance, can influence several of the very criteria used to evaluate the credibility of evidence obtained
with suggestive procedures, creating a type of Catch-22:
an argument that the use of feedback is suggestive would not result in a
successful motion to suppress the evidence because the eyewitness is
certain, claims to have had a good view, and so on. Of course, the eye witness is certain, claims to have had a good view, and so on because of
the suggestive procedure, but the Biggers criteria do not allow for such
an analysis ... arguing that a suggestive procedure is not a problem because of the eyewitness's high standing on the Biggers criteria is a bit
like arguing that a forensic DNA procedure that contaminated the suspect's blood with the sample at the crime scene is not a problem be cause the lab results show that the match is virtually perfect.
In light of Wells and Bradford's results, and the court's reliance on the
Biggers criteria, it is difficult to overemphasize the importance of limiting
suggestive procedures during police interviews. But suggestibility is not the
only concern that police confront when interviewing an eyewitness: they
want to elicit as much accurate information as possible. To improve eyewitness recall, some police professionals advocate the use of hypnosis. The
hypnotist uses an induction technique that directs the subject to relax and
concentrate on a specific object or activity: staring at a picture on the wall
while experiencing the sensation that one's eyelids are becoming heavy, or
imagining oneself lying on a warm beach. Once a sufficiently deep hypnotic state has been established, the hypnotist attempts to prod recall by
asking the subject to go back in time and reexperience the original event,
or perhaps imagine a giant television screen that depicts the incident they
witnessed.
Hypnotic procedures sometimes produce spectacular outcomes in actual crimes. One of the most impressive occurred in the 1976 kidnapping of
twenty-six children and their bus driver in Chowchilla, California. Three
masked men hijacked the bus at gunpoint, took the children and the driver
to a quarry, and buried them six feet underground. After the children and
driver miraculously escaped, FBI agents tried unsuccessfully to obtain information from them about the kidnappers. The bus driver then submitted
to a hypnotic interview, and recalled correctly five of six numbers from the
license plate of the kidnappers' van. This crucial information ultimately led
to the arrest and conviction of all three criminals.
Despite this and other impressive successes, the status of testimony
obtained through the use of hypnosis remains controversial. Hypnotic procedures frequently elicit inaccurate reports, and sometimes amplify the
suggestive effects of misleading information. Recent reviews of the scientific literature have turned up little or no evidence that hypnosis reliably
enhances the accuracy of eyewitness memory. But hypnosis can bolster
witness confidence. In view of its potentially powerful impact on juries, the
specter of confident - but inaccurate - testimony from a witness who
had been hypnotized continues to be a source of grave concern.
Advocates of hypnotically aided testimony, such as the forensic psychologist Martin Reiser, highlight dramatic successes and point out that
hypnosis does not invariably lead to heightened suggestibility. And, indeed,
if an investigation is stalled and other procedures have failed, hypnosis
might be useful for obtaining leads that can be subsequently checked by independent evidence. Hypnosis can also serve as a kind of "face-saving" device. Sometimes witnesses are initially reluctant to provide information for
fear of reprisals or personal embarrassment. If they later change their
minds but want to avoid admitting having lied earlier, they can "recover" a
memory through hypnosis. Indeed, such face-saving incidents could account for some of the apparent successes of hypnotic interviews.
Because of problems with hypnotically aided testimony, researchers
have sought to develop other procedures for increasing the retrieval of accurate information from eyewitnesses without also increasing suggestibility. One effective approach is known as the «cognitive interview." Developed initially in the 1980s by the cognitive psychologists Ronald Fisher
and Edward Geiselman, the cognitive interview is based on findings and
ideas established in controlled studies of memory, and specifically avoids
the use of suggestive or leading questions. The original cognitive interview
included four components. The first involves asking a witness to try to report everything about the relevant incident, important because police often ask highly specific questions that do not maximize witness recall, such
as «What color was his shirt?" instead of "Describe your attacker:' To stimulate recall of details that may not be retrieved initially, a second component of the cognitive interview requires the witness to try to reinstate mentally the context or setting in which the incident occurred. Numerous
laboratory studies have shown that such mental reinstatement of context
can enhance memory retrieval. Third, witnesses are asked to try to recall
events in different temporal orders: starting at the beginning and proceeding to the end, and vice versa. This procedure, too, has yielded improved recall in controlled studies. Finally, witnesses are asked to try to take different
perspectives on an event, such as mentally viewing the event from the perspective of the perpetrator or the victim, to help them notice features of the
incident that they otherwise might have overlooked. In the early 1990S,
these four cognitive procedures were supplemented with procedures that
foster social interaction and communication between the interviewer and
the witness.
Many experiments have compared the cognitive interview to standard
police interviewing techniques. Virtually all of them find that the cognitive
interview yields significant - and sometimes extremely large gains in
witness recall. These effects have been observed with various types of interviewers, ranging from novice college students to experienced police officers, and with different types of witnesses, including a range of adults, elderly individuals, and children.
As with hypnosis, however, the cognitive interview can yield greater
reporting of inaccurate information. But the amount of inaccurate information is typically small- many studies find no sign of it at all- and
there is no evidence that the cognitive interview reduces eyewitness accuracy. Because the results so far indicate that the cognitive interview raises
recall without a corresponding increase in suggestibility, many policeincluding all police forces in England and Wales - now receive training in
the cognitive interview and use it routinely when interrogating witnesses.
Further, several features of the cognitive interview are included in the
guidelines for collecting eyewitness evidence developed by Attorney General Janet Reno's working group (see Chapter 4).
Suggestibility is also a concern in relation to a highly disturbing outcome of some police interrogations: false confessions. Some false confessions occur because suspects wish to terminate mental or physical abuse,
even though they know they did not commit a crime; others happen spontaneously, without coercion, and may reflect attention seeking or related
pathology. But in a subset of false confessions - nobody knows exactly
how many - people develop the false belief that they committed a crime.
The Harvard professor Hugo Munsterberg was the first psychologist to call
attention to this latter type of false confession. In his classic 1908 book, On
the Witness Stand, Munsterberg observed that emotional stress, combined
with social pressure and suggestion, could distort memory to the point at
which people falsely believe they had committed a crime.
False confessions by political prisoners were widespread in the Soviet
Union during the heyday of totalitarian rule. "The Communists are skilled
in the extraction of information from prisoners and in making prisoners
do their bidding," observed the authors of a 1956 article on Communist interrogation techniques. "It has appeared that they can force men to confess
to crimes which they have not committed, and then, apparently, to believe
in the truth of their confessions and express sympathy and gratitude toward those who have imprisoned them."
Even in modern Western societies, people continue to make false confessions in which they mistakenly believe in their own guilt. In a British
case reported in the 1970s, Peter Reilly came home to discover the body
of his murdered mother. He immediately notified the police, who identified him as a suspect and administered a polygraph test that Reilly failed.
Though initially he denied the murder, Reilly eventually became convinced
that he must have committed the crime, and signed a written confession.
Two years later, he was exonerated by new evidence showing that he could
not have murdered his mother.
Reilly's experience illustrates what the clinical psychologist Gisli Gudjonsson calls a "memory distrust syndrome." Although Reilly did not develop detailed recollections of having committed murder, in the face of coercive questioning by police he began to distrust his own memory and
eventually disregarded it entirely. To abandon trust in his memory for such
a grisly event - more precisely, in his failure to recollect the murderReilly would have had to abandon the memory monitoring strategy that in
Chapter 4 I called the "distinctiveness heuristic": expecting to remember
distinct details of an experience. Normally, someone who had participated
in a horrific event such as murdering his own mother would surely expect
to remember the incident. The memory distrust syndrome can develop
when it is plausible that one might forget even a violent crime - perhaps
when a person is intoxicated, or believes he could have repressed a horrendous event. When somebody no longer expects to remember an event
clearly, it is easier to distrust memory.
In some instances of false confessions, suspects initially believe in
their own innocence, but in the course of suggestive police questioning,
may eventually develop specific recollections of a crime they did not commit. In a highly publicized case from the mid-1990s, the Washington State
sheriff's deputy Paul Ingram confessed to the sexual abuse of his two
daughters and participation in a bizarre cult that included satanic rituals,
animal sacrifices, and the murder of babies. In response to coercion and
bullying from local police officers, Ingram had reported full-blown "memories" of these ghastly activities - memories that he came to believe he
had previously repressed. Even though no hard evidence for any of the confessed activities was ever produced, and Ingram eventually retracted his
confession, he was sent to prison and remains there today.
Coercive questioning by police is frequently involved in false confessions. Gisli Gudjonsson and associates in London recently described the bizarre conclusion of a case involving a seventeen-year-old man who had
been routinely interviewed by police in connection with the investigation
of a brutal murder. He became preoccupied with "visions" of the victim's
face and began to wonder whether he had committed the crime. The young
man turned himself into police voluntarily, initially stating that "it might
have been me:' but that "I don't know if I killed her or not. I keep seeing
her." During the course of the next twenty-four hours, he developed the belief that "I must have done it because I can see a picture of her," and then
finally asserted with conviction that "I am sure I killed her ... I know I did
it." Although there was no other evidence to support this claim, the man
was imprisoned on the basis of his written confession. He served twentyfive years of his sentence before new evidence led to a reversal of his conviction.
This latter case raises the possibility that some individuals may be
especially prone to false confessions because they are easily suggestible.
Gudjonsson has developed a scale for measuring individual differences in
what he calls "interrogative suggestibility": the tendency to change claims
about the past in response to misleading information and suggestive questions. Gudjonsson found that people who had made a criminal confession
that they later retracted were more influenced by suggestive questions than
"deniers" who steadfastly refused to acknowledge any involvement in a
crime despite forensic evidence against them. The memory performance of
the two groups on standard clinical tests did not differ.
It is still hard to fathom how anyone could admit to carrying out an
act - much less a violent crime - that he or she did not commit. The
memory sins considered earlier in the book - transience, absent-mindedness, blocking, and some types of misattribution - are so familiar from
our daily experience that we can relate to them all too easily. But the kind of
suggestibility involved in false confessions is alien to the realities of everyday remembering and forgetting. Not surprisingly, mock juries are highly
skeptical of the possibility that people would ever confess to crimes they
did not commit.
Experiments by Saul Kassin's group at Williams College show that
false confessions may not be as aberrant as they first seem. College students
seated at a computer were instructed to type in a series of spoken letters one group typed at a hurried pace, another at a more leisurely rate. All students had been instructed not to press the AL T key because it would cause
the program to crash. None of the students actually hit the ALT key, but the
experimenter falsely accused them of doing so. After denying the charge,
half the students in each group heard a confederate "witness" say that she
saw the error; there was no witness for the other half. Nearly 70 percent of
students eventually signed a false confession that they had hit the AL T key.
The effect was particularly striking in the group that responded quickly
and also heard a witness back up the experimenter: all of them signed the
confession, and 35 percent produced a detailed false recollection of how
they made the error.
Kassin's results are troubling, because they suggest that under the
right conditions many of us can be induced to confess to an act that we
never performed. Of course, people may not expect to remember hitting
the AL T key on a computer, whereas they would ordinarily expect to remember committing a violent crime. It may be relatively easier to make
people confess to hitting the AL T key than committing a crime because
they are less likely in the experimental situation to invoke the distinctiveness heuristic - if I had committed this act I surely would have remembered it. This interpretation is supported by the finding that false confession was especially common in people who typed rapidly during the initial
task. These individuals probably expected less from their memories than
those who had more time to respond, perhaps reasoning that because they
responded so quickly they were likely to make a mistake and unlikely initially to remember doing so.
The consequences of suggestibility in eyewitness testimony and police
interrogation can be shattering, but its harmful effects are not confined
to these public arenas. Suggestibility can even shape recollections of the
mostly intensely personal and private aspects of our pasts.

THE RISE AND FALL OF FALSE MEMORY SYNDROME.

In 1992, a group of alarmed middle-aged adults formed the first organization ever dedicated to the topic of memory distortion: False Memory
Syndrome Foundation. Comprised mainly of parents in the midst of disturbing conflicts with their adult daughters, the early members of the foundation told stories that seemed shocking at the time, but became numbingly familiar as the 1990S progressed. Educated and intelligent middleclass women entered psychotherapy for depression or related problems,
only to emerge with recovered memories of previously forgotten childhood
sexual abuse, typically perpetrated by fathers and sometimes by mothers.
The parents who formed the foundation, and many others like them, angrily disputed the validity of the memories their children embraced. Accusers and their supporters castigated the parents for denying a reality they
could not accept.
As I mentioned in Chapter 3, some recovered memories of childhood
abuse that people had not thought about for years have been corroborated
and appear to be accurate. But when the crisis first boiled over in 1992,
many professionals and accused parents were quick to blame an alleged
epidemic of false memories on suggestive techniques used by some psychotherapists hypnosis, guided imagery exercises where people imagine
possible abuse scenarios, and the like - to call up supposedly forgotten
traumas. As the 1990S unfolded, several kinds of evidence indicated that
many recovered memories are inaccurate: implausible recollections of bizarre practices in satanic cults whose existence was never documented; the
lack of scientific support for commonly used memory recovery techniques;
and a steadily increasing number of women who retracted their memories.
Early on, however, memory researchers were roped into the fray as potential arbiters of the reality of recovered memories. Many pressing questions
required solid scientific answers: Is it possible to create false memories of
traumatic autobiographical events? What kinds of techniques are most
likely to promote illusory recollections? Are certain kinds of people especially susceptible to implantation of memories for events that never
happened?
In the early 1990S, memory researchers did not have good answers to
these questions. Psychologists knew in general terms that memory is suggestible, but for the most part they had to rely on evidence from such experimental techniques as those pioneered by Elizabeth Loftus, in which
suggested details of an event seep into the recollections of eyewitnesses.
Critics objected - and rightly so - that these kinds of suggested memories involve little more than the minutiae of an experience; they do not
demonstrate or imply that people can develop full-blown false recollections of a trauma such as sexual abuse. Memory researchers, the critics
contended, would have to do far better before their findings could inform
the ongoing debate. And they have. In an ironic contrast to its devastating
effects on families, and the bitter divisions it has produced within psychology and psychiatry, the recovered memories debate has had a salutatory
effect on memory research by stimulating a new wave of studies on suggestibility.
Appropriately enough, Elizabeth Loftus - a pivotal figure in earlier
suggestibility studies and a lightning rod in the recovered memories debate
reported one of the first attempts to implant experimentally a mildly
traumatic autobiographical incident. In what came to be known as the
"lost in the mall" study, a teenager named Chris was asked by his older
brother Jim to try to remember the time Chris had been lost in a shopping
mall at age five. He initially recalled nothing, but after several days Chris

produced a detailed recollection of the event. The study achieved instant
notoriety because, according to Jim and other family members, Chris never
was lost in a shopping mall. Following up with a larger group of twentyfour participants, Loftus documented that after several probing interviews,
approximately one-quarter of the participants falsely remembered being
lost as a child in a shopping mall or a similar public place.
The psychologist Ira Hyman and his group at Western Washington
University have successfully implanted false memories of other childhood
experiences in a significant minority of participants in their experiments.
Hyman asked college students about various childhood experiences that,
according to their parents, had actually happened, and also asked about a
false event that, their parents confirmed, had never happened. For instance,
students were asked: "When you were five you were at the wedding reception of some friends of the family and you were running around with some
other kids, when you bumped into the table and spilled the punch bowl on
the parents of the bride." Participants accurately remembered almost all of
the true events, but initially reported no memory for the false events. However, approximately 20 to 40 percent of participants in different experimental conditions eventually came to describe some memory of the false event
in later interviews. In one experiment, more than half of the participants
who produced false memories described them as "dear" recollections that
included specific details of the central event, such as remembering exactly
where or how one spilled the punch. Just under half reported "partial" false
memories, which included some details but no specific memory of the central event.
Hyman's results implicate visual imagery as a culprit in suggested
memories. People in his studies who produced false memories of childhood experiences scored higher on scales that measure vividness of visual
imagery than did individuals whose recollections were more accurate. Further, when Hyman and associates specifically instructed people to try to
imagine an event if they could not recall it initially, they found more false
memories compared with when participants were allowed to sit quietly and
think about whether the event had occurred. These results make sense in
light of other evidence that true recollections of actual events are often
characterized by rich and detailed visual imagery. If imagery is a kind of
mental signature of true recollections, then embellishing a false memory
with vivid mental images should make it look and feel like a true memory.
In recent collaborative work with Elizabeth Loftus, the Italian psychologist Giuliana Mazzoni asked whether another type of suggestive procedure can produce false memories: dream interpretation. Some therapists
use their patients' dreams to make inferences about what happened to
them in the past. Could dream interpretation help to create, rather than reveal, past experiences? To find out, Mazzoni and Loftus asked people to rate
their confidence that various kinds of experiences had or had not happened to them. One group then participated in an ostensibly unrelated task
two weeks later in which a clinical psychologist interpreted their dreams.
The psychologist suggested to them that their dreams included repressed
memories of events that had happened to them before the age of three upsetting experiences such as being abandoned by parents, getting lost in a
public place, or being lonely and lost in unfamiliar surroundings. The participants previously indicated that such events had never happened to
them. Nonetheless, when they were again asked about early experiences
two weeks after having their dreams interpreted, the majority now claimed
to remember one or more of the three suggested experiences for which they
had previously denied any memory. Nothing of the sort was found in a
control group that did not receive any suggestions regarding their dreams.
The kinds of events falsely recollected in the studies by Loftus, Mazzoni, and Hyman are sometimes mildly upsetting, but do not involve serious trauma. More recent experiments have obtained similar results with
more disturbing events. Using procedures like those reported by Hyman,
the Canadian psychologist Stephen Porter and coworkers successfully implanted false childhood memories of a serious animal attack, serious outdoor accident, and serious harm perpetrated by another child in approximately one-third of the college students in their experiments. Of course,
there may be limits to the kinds of memories that can be successfully suggested. In one study, for example, 15 percent of participants generated false
recollections of being lost in a shopping mall, but none generated false
memories of a childhood enema.
Still, it's hard not to be impressed by just how many different kinds of
memories can be suggested. Consider, for instance, your very earliest recollection: What is the first thing you can recall from childhood? The psychoanalyst Alfred Adler believed that earliest memories have great psychological significance, providing important insights into the very core of an
individual's personality. For most of us, earliest memories date from three
to five years of age; there is no evidence that people can remember incidents that occurred before they were two years old, most likely because the
brain regions necessary for episodic memory are not yet fully mature until
that age.
In one recent study, people generally reported earliest memories from
when they were three or four years old, as in most previous research. The
experimenters then introduced a suggestive procedure in which they asked
subjects to visualize themselves as toddlers and try to "get in touch" with
even earlier memories. They offered assurances that just about anyone can
remember very early events, such as a second birthday, by "letting go" and
working hard to visualize the event. Following the suggestive procedure,
people reported earliest memories that dated, on average, to approximately
eighteen months - well before the accepted offset of childhood amnesia.
Indeed, one-third of those exposed to the suggestive procedure reported an
earliest recollection from prior to twelve months, whereas nobody did so
without suggestions. Because there is no other evidence that people can recall events from this early in their lives, these newly discovered "memories"
almost certainly do not reflect accurate recall of events. Consistent with
this idea, those individuals who came up with earliest memories from prior
to twenty-four months were more suggestible on the Gudjonsson interrogative suggestibility scale than those who did not.
Visualization is not the only suggestive procedure that can influence
what people claim to remember about early childhood. In a separate study,
hypnotic suggestions yielded earlier autobiographical memory reports
than instructions to relax or to count numbers visually; nearly four of
every ten participants who received hypnotic suggestions claimed to remember events that occurred at or before their first birthday.
If doubts linger about whether memories dating prior to the age of
two years are products of suggestion, rather than accurate recovered memories, results from the laboratory of the late Canadian hypnosis researcher
Nicholas Spanos should end the debate. Consider the following question:
Can you recall whether at the hospital where you were born a colored mobile hung above your crib? Of course you cannot. Spanos and collaborators
told people they wanted to find out whether there were colored mobiles
above their birth cribs. They informed one group that hypnosis allows people to remember events from the first days of life by regressing them to an
earlier time so that they can relive those experiences. These individuals
then received a hypnotic regression treatment, and mentally "returned" to
the day after their birth. A second group listened to the same speech that
hypnosis can unlock early memories, and then heard that they would receive an equally effective nonhypnotic treatment called "guided mnemonic
restructuring." They were encouraged to "reexperience" the day after their
birth, but were not administered a hypnotic regression treatment. A control group was told nothing about hypnosis or memory enhancement, and
simply tried to recall what dangled over their cribs the day after they were
born.
No one in the control group came up with memories of a mobile over
his crib, but about half of the individuals in the other two groups did.
Regardless of whether they actually received hypnotic regression treatment,
some people who had been led to expect that they would be able to recall
experiences from the first day after birth expressed a strong belief that they
had done so.
Though not unprecedented - other evidence shows that people
sometimes "remember" past lives and alien abductions, usually under the
influence of hypnosis - these results are important because they underscore the key role of expectancies in producing false memories. The mere
suggestion that participants should expect to recall something from the
first day of life was sufficient to lead half of an otherwise ordinary sample
of introductory psychology students to believe that they had recovered a
patently preposterous memory.
Based on what I said in Chapter 4 about the distinctiveness heuristic
and what we expect from our memories, it is perhaps not surprising that
people readily come up with false memories from early childhood and infancy. Ordinarily we would not expect to recall the incidents of early life
with vividness or clarity, in the same way that we would expect to do so for
a recent event. It is extremely difficult to implant false memories of salient
personal experiences that allegedly occurred yesterday, such as becoming
lost in a shopping mall, because we expect to remember yesterday's events
with some clarity and detail. For recent events, we can invoke a distinctiveness heuristic: if the suggested event had occurred, we would have remembered it vividly. But we expect little of recollections from early childhood
and thus are more likely to interpret fuzzy images or vague feelings of familiarityas signs of an emerging memory, particularly if we are instructed
to expect that such recollections are possible.
Suggestibility's pernicious effects highlight the idea that remembering
the past is not merely a matter of activating or awakening a dormant trace
or picture in the mind, but instead involves a far more complex interaction between the current environment, what one expects to remember,
and what is retained from the past. Suggestive techniques tilt the balance
among these contributors so that present influences playa much larger role
in determining what is remembered than what actually happened in the
past.
At the same time, results like those reported by Spanos and others
provide a sobering perspective on the recovered memories controversy.
Recollections of early experiences are extremely malleable, more so than
many would have believed less than a decade ago. When suggestive techniques such as hypnosis and guided imagery are used to hunt for memories
from vulnerable periods of childhood, they comprise a potentially dangerous recipe for producing false memories. Surveys of psychotherapists conducted in the early and mid-1990s indicate that many believe that hypnosis
and guided imagery can unlock buried childhood memories, and therefore
use these techniques to stimulate clients' recollections. In view of the data
we've considered, it should not be at all surprising if a subset of those clients recalled events that never occurred.
People with especially vivid imagery and those who score high on interrogative suggestibility scales appear to be at risk for creating some types
of false memories. Ira Hyman has also found that individuals who obtain
high scores on a scale that measures self-reported tendencies toward lapses
in attention and memory are more likely to create false childhood memories than are people who obtain lower scores on that scale. Experiments
with college students reveal that higher scores on that scale are also associated with greater false recognition of semantic associates - words such as
sweet after the study of related words such as candy, sour, sugar, bitter, and
so forth (see Chapter 4). A recent study in my laboratory led by Susan
Clancy documented a similar relationship in adult women. In that study,
we found that women who report recovered memories of childhood sexual abuse showed elevated false recognition of semantic associates on the
sweet test compared with women who were abused as children and always
remembered it, and with nonabused control subjects.
It is conceivable that the women reporting recovered memories were
abused as children, forgot about it, and later recalled the abuse; the early
trauma might be responsible for heightened susceptibility to false recognition. However, this hypothesis does not explain why women who reported
recovered memories showed more false recognition than women who always remembered their abuse. An alternative possibility is that the recovered memories are inaccurate, reflecting a vulnerability to memory distortion which also results in more false recognition of semantic associates. We
cannot be certain about the causal sequence: early trauma produces heightened false memories, or greater susceptibility to false memories produces
inaccurate reports of early trauma. However, Clancy has recently led another study that shows that people who "remember" being abducted and
abused by aliens also show increased false recognition of semantic associates. Because the abduction memories are surely false, these results indicate
that heightened false recognition of semantic associates in the laboratory
may indeed reflect elevated risk for experiencing false memories outside
the lab. At the very least, our results add to the evidence that some people
are more vulnerable to false recognition than others.
As the 1990S concluded, there were clear signs that the recovered
memory crisis had started to ease. Perhaps because of new knowledge
about suggestibility and memory which encourages therapists to adopt a
more conservative approach to memory recovery, and perhaps because of
successful lawsuits brought against them by retractors, the incidence of
new cases involving disputed recovered memories has plummeted. In the
False Memory Syndrome Foundation's winter 1999 newsletter, the director,
Pamela Freyd, reported that the foundation "now receives dramatically
fewer calls and letters from people asking for assistance." And Freyd concluded, "The drop is of such magnitude that we can finally phase out that
part of the FMSF organization that responded to those calls." The rise and
fall of the recovered memories controversy parallels a related crisis that
hinged on the suggestibility of the most vulnerable memories of all.
SUGGESTIBILITY IN PRESCHOOL
On April 19, 1999, the Boston attorney James Sultan sent me a copy of an
amicus brief filed days earlier in the case of Commonwealth of Massachusetts v. Cheryl Amirault LeFave. LeFave, along with her brother Gerald and
mother, Violet, had been convicted over a decade earlier of molesting children at their family-run Fells Acres Daycare Center in Malden, a small suburb just north of Boston. The Amiraults' story resembled other highly pub licized daycare cases that seemed to spread like wildfire in the 1980s and
early 1990S, such as McMartin in Los Angeles and Little Rascals in Edenton,
North Carolina. In these latter cases, preschool children reported that they
had been subject to revolting, even horrific acts. The children's accusations
involved not only sexual abuse, but also bizarre claims of bloody torture,
murder, being forced to eat dead babies, and even taking trips on alien
spaceships. Yet there was a lack of medical evidence that the children had
suffered abuse, and no adult visitors ever noticed anything amiss in the
daycare centers where wrongdoing allegedly occurred. None of the schools
had a prior history of problems: Fells Acres, for example, had been operating for eighteen years without any accusations of impropriety, prior to the
initial charges against Gerald Amirault in 1984. The children who made the
accusations had almost invariably endured suggestive questioning by police or child-care professionals.
There was, however, a critical difference between Fells Acres and the
other preschools, reflected in the amicus brief I received. Prosecutors failed
to obtain a conviction in the McMartin case, and finally gave up trying. The
convicted teachers from the Little Rascals daycare center eventually were
released after new evidence became available. Despite concerted efforts on
the part of her defense attorneys and the urging of leading researchers in
the area of child memory, the Commonwealth of Massachusetts maintained that Cheryl Amirault LeFave belonged in prison.
Amirault LeFave and her mother, Violet, had been offered parole during 1992 in exchange for an admission of guilt, but refused to admit to
crimes that they claimed they did not commit (no such offer was or has
been made to Gerald Amirault). They were granted new trials in 1995, and
released from prison. But prosecutors successfully appealed, and in 1997 the
Supreme Judicial Court of Massachusetts overturned the decision to award
the Amirault women new trials, ordering them to return to prison. Culminating a subsequent frenzy of legal maneuvers, in May 1997 Judge Isaac
Borenstein overturned the convictions of Cheryl and Violet on the technical grounds that they had not been allowed (literally) to face their accusers
- the Fells Acres children - directly in court. Violet Amirault died of
cancer in September 1997. The prosecution was preparing an appeal of the
decision that had freed Cheryl when her attorney, James Sultan, told the
court that he had novel evidence warranting a new trial.
Sultan had enlisted the aid of Dr. Maggie Bruck from McGill University in Montreal, a respected expert on the suggestibility of children's recollections. Dr. Bruck contended that the Commonwealth owed Cheryl another trial because new research on child suggestibility spoke directly to the
possibility that the interviewing techniques used with the Fells Acres children led them to provide inaccurate reports. The amicus brief that James
Sultan sent me in April backed up Dr. Bruck's interpretation of the new
research and its potential relevance to the guilt or innocence of Cheryl
Amirault LeFave. Twenty-nine researchers with established credentials in
the study of memory, including me, had signed the brief.
The bulk of the evidence described in the amicus brief came from
striking demonstrations regarding the nature and extent of suggestibility in
children's recollections of personal experiences. Beginning in the early
1900s, researchers had shown that suggestive questions can distort children's reports about the past - sometimes to a greater extent than seen in
adults. But prior to 1990, almost all of this research had examined children
who were older than the preschoolers whose memories were disputed in
the Fells Acres case and other similar instances. At the time of the Amiraults' conviction, there were only a handful of studies available concerning
the suggestibility of preschoolers like those who testified against them. Further, the early studies had focused on whether small details of an incident
could be suggested to children through misleading questions. If asked
about the hair color of a bald man who had visited them, children who "remembered" that the man had black hair were considered suggestible. But
this type of research fell far short of determining whether suggestive questions could lead children to provide an inaccurate recollection of an entire
event that, in reality, never occurred.
Sultan and Bruck's concerns about the Amirault case centered on interviews with the Fells Acres children conducted by the pediatric nurse Susan Kelley. None of the Fells Acres children had spontaneously reported
abuse to their parents, and they denied being abused when asked about it
initially. Reports of abuse emerged only after questioning by parents, police, Kelley, and others (based on concerns stemming from an incident in
which one child engaged in sex play with a cousin). This observation is key
because new research has shown that children's spontaneous recollections
tend to be accurate, whereas their responses to specific questions are more
likely to include distortions. For instance, in a 1996 study two- to five-yearold children were interviewed about treatment that they had just received
in an emergency room. The researchers found that when children were
asked open-ended questions, such as "What happened?" they provided accurate details of their experiences. But when they were asked more specific
questions, such as "Where did you hurt yourself?" the incidence of inaccurate details grew dramatically: from 9 percent in response to open-ended
questions, to 49 percent in response to specific questions.
Bruck noted that in her interviews of the Fells Acres children, Susan
Kelley never began by asking an open-ended question such as "What happened?" Instead, she proceeded directly to specific questions: queries about
teachers, whether they were nice, and so forth. Kelley also frequently repeated specific questions, seeming to refuse to take no for an answer. For
instance, investigators developed the hypothesis that a down whom the
children had mentioned in connection with the daycare center was in some
way related to the alleged abuse. In the following exchange, Kelley repeatedlyasks the child about the clown's action:
Did the clown touch you?
KELLEY:
CHILD: No....
KELLEY: You said the clown took your clothes off.
CHILD: Yeah.
KELLEY: And then what happened?
CHI L 0: Well, nothing really.
KELLEY: Did the clown touch any ... Will you show me if the clown
touched any part of you?
CHI L 0: No, he didn't touch me any KELLEY: Now, pretend this was you. Did the clown touch you? Where
did the clown touch you?
CHILD: Right there [indicates footJ.
KELLEY: Did he take your underpants off?
CHILD: [No responseJ
KELLEY: Then what did he do?
CHILD: Nothing else.
KELLEY: No? Did he touch you?
CHILD: I want to wear that now.
KELLEY: Oh, but I want you to tell if the clown touched you.
CHILD: Yeah.
With other Fells Acres children, repeated interviews were conducted
on separate occasions when an initial interview failed to yield satisfactory
answers, with results much like those depicted in the quoted transcript:

negative responses were eventually replaced by positive ones. Such repeated
questioning is alarming because studies conducted by Bruck and others
have found that when children are interviewed twice, and produce details
in a second interview that were not mentioned in the first, the new details
,are highly likely to be inaccurate. In related studies, Bruck and the CorI
nell psychologist Stephen Ceci repeatedly asked children questions about
events that their parents indicated had never occurred, such as getting a
finger caught in a mousetrap and going to the hospital. The children were
encouraged to think about and imagine the events. After repeated ques tioning, 58 percent of preschoolers reported detailed recollections of at
least one event that they initially said had never occurred; 25 percent gener ated false memories for a majority of such events.
Some of the deleterious effects of suggestive questioning are attribut able to basic vulnerabilities of young children's memory systems. A grow ing number of laboratory studies indicate that young children have special
difficulties remembering source information - exactly when and where a
particular incident or action occurred. When children are repeatedly asked
about particular events, the incidents may begin to feel familiar simply be cause the examiner has mentioned them numerous times. Lacking detailed
memory for the source of the feeling of familiarity, preschoolers may begin
to mix together bits and pieces of different past episodes, or even intrude
elements of fantasy and imagination. Source memory problems may also
explain why parents can sometimes inadvertently suggest experiences to
children which actually never occurred. In one study, preschoolers visited
"Mr. Science" at a university laboratory and watched him conduct some
experiments. Four months later, the children's parents received written
descriptions of the experiments, others that the child had not witnessed,
and a further incident that had not actually occurred: "Mr. Science wiped
[child's name] hands and face with a wet-wipe. The cloth got close to
[child's name] mouth and tasted really yucky." Parents read the stories to
their children three times. When asked later about what they had seen in
the laboratory, children frequently remembered experiments that had been
mentioned only by their parents. When asked whether Mr. Science put
something yucky in their mouth, more than half the preschoolers said
"yes." Poor source memory is the likely culprit.
Some of the reports generated by children in daycare cases such as
Fells Acres may also be attributable to social pressures that often surround
the interview situation. For example, Maggie Bruck documented a number
of instances in which Susan Kelley held out promises and even bribes in exchange for testimony.
At the time of Cheryl Amirault LeFave's trial, little was known about
the effect of social influence on the accuracy of a child's recollection. Researchers had typically examined the effects of suggestive questioning in
isolation from the social pressures that were often present during interviews in the 1980s daycare cases. And, indeed, some studies found that
when preschoolers received only a single suggestive question, they rarely
produced false reports about the central features of an event, such as
whether strangers had taken off their clothes.
Newer studies have begun to fill in the gap. In 1998, the psychologists
Sena Garven, James Wood, and their coworkers from the University of
Texas at El Paso took advantage of a new resource that had not been available during the Fells Acres trial: transcripts of interviews from the McMartin case. As with the Susan Kelley interviews, investigators in the McMartin
affair applied various kinds of social pressures in an attempt to elicit information from recalcitrant preschoolers. In addition to asking suggestive
questions, interviewers offered praise and rewards for sought-after information, expressed disappointment or disapproval when children failed to
come up with a desired answer, repeated questions that initially yielded no
responses, and invited children to speculate about what might have happened by pretending or imagining.
Garven and associates compared the McMartin techniques to a control condition involving suggestive questioning only. Preschoolers watched
and listened as a graduate student introduced as Manny Morales told them
the story of The Hunchback of Notre Dame. After the story he gave out cupcakes and napkins, said goodbye, and departed. A week later, children in
the control group were asked about a few things that Manny had done,
such as taking off his hat and asking the children to sit quietly and listen.
They were also asked suggestive questions about things that Manny had not
done: tearing a book, putting a sticker on a child's knee, saying a bad word,
throwing a crayon at a child who was talking, and so forth. Children in the
social incentive group were asked the same questions, except that the interviewers also used the influence techniques identified in the McMartin transcripts.

and six-year-olds said "yes" to just over half of the misleading questions,
whereas five- and six-year-olds in the control group said "yes" to fewer
than 10 percent of the misleading questions. Results were similar for fouryear-olds and even worse for three-year-olds: they said "yes" to 81 percent
of misleading questions in the social incentive condition compared to 31
percent in the control condition. These findings leave little room for doubt
that social incentive techniques like those used by investigators in McMartin and Fells Acres have a devastating effect on the accuracy of what preschoolers report about past experiences.
In another highly publicized case, nineteen adults in the small town of
Wenatchee, Washington, were convicted of running a child sex ring. But
the convictions have been questioned because a thirteen-year-old girl who
served as a key witness recanted her testimony, claiming that the chief police investigator forced her to generate allegations of sexual abuse. "I had to
make it all up," she reflected. "First 1 said it didn't happen ... and then he
forced me to make up a lie." Experiments by the psychologists Jennifer
Ackil and Maria Zaragoza have shown that forcing elementary school children to answer a suggestive question about what happened in a video they
saw earlier creates a serious source memory problem: the children confused their own answers with what had happened in the video.
Despite the persuasiveness of the new research on child suggestibility
to the twenty-nine scientists who signed Dr. Bruck's amicus brief, prosecutors continued to insist to the court that grounds were insufficient to warrant a new trial for Cheryl Amirault leFave. In the winter of 1998, Maggie
Bruck described the new research to Judge Isaac Borenstein, highlighting
important differences from the earlier work that was available at the time
of the initial Amirault trial. Despite rebuttals from the prosecution, Judge
Borenstein found Bruck's arguments convincing. He enumerated stinging
criticisms of the prosecution's evidence and ruled in favor of a new trial.
Borenstein did not, however, have the final word. In August 1999, the Supreme Judicial Court of Massachusetts, siding with the prosecution's claim
that evidence concerning child suggestibility was available at the time of
trial and that Bruck had not added anything fundamentally new, overturned Judge Borenstein's decision and reinstated Amirault leFave's conviction. The decision seemed to ensure that she would return to prison. But
days before she was scheduled to do so, in late October 1999, prosecutors
and defense attorneys agreed to a deal: Amirault leFave was set free on
time served, but remains a convicted felon. During a ten-year parole pe riod, she cannot discuss the case on television or profit in any way from her
involvement in it. Gerald Amirault remains in prison and his sister cannot
visit him.
Though tragic for the Amiraults, the Fells Acres parents and children,
and those entangled in related preschool cases, the interviewing errors of
the 1980S are also responsible for the novel research in the 1990S which
should provide benefits for children and the rest of society. Knowing more
about what factors raise suggestibility in young children - leading questions, social incentives, forced responding, and the like - also means
knowing more about how to lower it. Interviewers who rely on simple
open-ended questions, and avoid the risky techniques used in the past,
stand an excellent chance of obtaining accurate information from even
very young witnesses.
Suggestibility remains a worrisome vulnerability of memory, especially in young children. Yet despite its potential to wreak more havoc than
any other of the seven sins, suggestibility is probably the easiest to neutralize. Whereas countering such problems as transience and absent-mindedness, for instance, requires putting forth the effort to perform elaborative
encoding techniques or to construct external memory aids, avoiding suggestibility's harmful consequences mainly involves knowing what not to do.
There is no longer any reason why police and mental health professionals
who interview children or adults in legal or therapeutic contexts should repeat the kinds of errors that were made before psychologists declared a
kind of research war on suggestibility during the 1990S. By revealing just
how permeable to suggestions our recollections can be, the new studies
provide weapons that can allow society to better protect the integrity of
memory from external influences that, if left unchecked, are likely to corrupt it.