In 1974 Elizabeth Loftus, then at the University of Washington, reported the results of experiments that showed how visual memory could be interfered with when the experimenter presented certain kinds of false information to research subjects. In one such experiment, subjects first watched a short film that included a brief segment showing a collision between two cars. Subjects were then immediately asked to write a description of what they had seen, and then were asked to estimate how fast the cars were going when they collided. The key part of the experiment was that different subjects were asked this question in slightly different ways. For some of them the question was, “About how fast were the cars going when they hit each other?” For other subjects the question was, “About how fast were the cars going when they smashed into each other?” For other subjects (the control group), no such question was asked. One week later, all subjects were asked if they had seen broken glass at the accident scene. Only 6 percent of the control subjects said they had, and

only 7 percent of the hit subjects said they had—but 16 percent of the smashed into subjects stated they remembered seeing broken glass. In other experiments Loftus and her co-workers found that the verb chosen to describe the accident could increase their subjects’ estimates of vehicle speed at the time of the accident.

Loftus proposed that these research results could be understood as resulting from the incorporation of information presented during the questioning period into the memory of the original event. This should remind you of the process of retroactive interference described in Chapter 4: Memory of an event occurring at one time can be interfered with by the memory processing of events occurring at a later time. Of course, retroactive interference was a well-known phenomenon long before Loftus’s research. What she did was devise creative new ways of applying this concept to memory testing procedures that were quite different from conventional ones.

Loftus was aware of the possible relevance of her research to legal issues. The way misinformation was presented in her experiments seemed similar to leading questions—questions whose form or content implies or assumes something about an event. However, in 1974 Loftus could not have anticipated how, in just a few years, her research was going to become enmeshed with an incredible and appalling series of events involving child sexual abuse, criminal conviction, and imprisonment of the innocent. These events came about in part because of unfounded and inaccurate beliefs about human memory.

You have probably heard cases involving allegations of childhood sexual abuse and “recovered memory” of abuse in adults. But you may not know the full story. It has been described this way by the psychologists Maggie Bruck and Stephen Ceci:

The full details of some of these cases are sometimes difficult to believe. In 1994 in Wenatchee, Washington, the 11-year-old foster daughter of a police detective started to relate stories to her father and child care workers of how she and others had been sexually molested. This led to 2,300 charges of child molestation, the arrest of more than 40 people, the conviction of several of them, the separation of children from their parents, and an inestimable amount of human suffering—all on the basis of uncorroborated and unsubstantiated testimony from an 11-year-old and a few other children.

What did the children say they remembered happening to them? Here is a sample, as described by Dorothy Rabinowitz, a writer for the Wall Street Journal:

Equally astounding are cases of recovered memory in adults with multiple personality disorder (MPD). MPD is supposed to result from attempts by a person to cope with early traumatic experience, childhood sexual abuse being the most important form of this. The person adopts a new personality (or “dissociates” from his or her real or current personality) and this somehow prevents or reduces stress, conflict, and suffering. This is said to be accompanied by a functional amnesia caused by the complete repression (unconsciously motivated forgetting) of memories of the abuse.

An important history of MPD and recovered memory can be found in a New Yorker article by Joan Acocella. The article’s title is “Hysteria,” a term that was in common use in the nineteenth century to describe disordered and abnormal behavior, almost always in women. One of the most widely known cases is described in the book and movie The Three Faces of Eve, which deals with a woman who had three distinct personalities. Another book, Sybil, which appeared in 1973, raised the stakes in describing a woman who, over the course of an 11-year period of psychotherapy, displayed 16 different personalities: “One could play the piano; another could install sheetrock; two had English accents; two were boys.” (Acocella goes on to observe that “Sybil was more like a club than a person.”) Sybil’s emergent memories, “recovered” with the help of guided imagery, hypnosis, drugs, and relentless questioning by a psychiatrist, read like a movie treatment of a horror-fantasy story. It can’t be proved that the popularization of multiple-personality case histories contributed to events such as those in Wenatchee, but the timing is certainly right.

Why many claims of abuse and recovered memory have been accepted without corroboration and without physical evidence are

questions for students of law and society to ponder. Our concern is with beliefs about memory, what is actually known about memory accuracy, and how this knowledge could or should have influenced the acceptance of such claims. Loftus and others argue that a set of false beliefs about memory and other psychological processes has contributed to the recovered memory issue.

One of these beliefs is that memory is a high-resolution recording device that creates a continuous, detailed copy of experience from which memories can be “played back” with perfect fidelity. This belief is constantly strengthened by movies and television shows that depict people not just remembering something about the past but reexperiencing it in all of its original detail. It is true we have vast visual and auditory long-term memory stores that enable us to recognize objects, faces, scenes, voices, and music, but this does not require belief in a “videorecorder” model of memory.

A second mistaken belief is that there is a memory process known as repression (unconsciously motivated forgetting) that is capable of removing memories, even memories of prolonged or repeated experiences, from deliberate or conscious remembering. In Freud’s theories, repression was the fundamental defense mechanism, a way of keeping disturbing and anxiety-generating thoughts and memories out of consciousness. In classical Freudian theory, repression was also an unconscious process that reduced the accessibility of information in memory and as such is an example of forgetting as a retrieval failure rather than a trace failure. Note that repression is quite different from suppression, which is more like the familiar process of consciously and purposely directing attention away from troubling thoughts or recollections, of not mulling over things (and thereby not strengthening the memories), and of letting time and the ordinary curve of forgetting do their work.

The idea that repressed memory of early abuse is an important cause of psychological and behavioral abnormalities in adulthood is familiar to all of us. We come across it in films, plays, and books such as The Two Faces of Eve and Sybil. It has been with us

since the heyday of psychoanalysis. What is the status of repression as a scientific concept? The brief answer is that trauma-induced repression has not been validated as a fundamental cause of forgetting. A much more forceful statement appears in Richard McNally’s book Remembering Trauma, a recent and detailed evaluation of the available scientific evidence:

This is not to say that memory is unaffected by conflict or anxiety, in conscious and possibly unconscious forms. (One of the present authors still winces at the recollection of being unable to remember his prom date’s name when introducing her to his mother.) However, repression of the kind assumed to result from childhood sexual abuse and to result in multiple personality disorder has never attained the status of an experimental fact. As many memory researchers have pointed out, the memories of severe abuse and punishment are distinguished by the difficulty people have in forgetting the events, not the difficulty in remembering them. There have also been numerous experimental studies of emotion, memory, and the brain (see Chapter 7) but to date none of them have led to identification of the kinds of biologically plausible brain mechanisms that would be required for repression to operate.

Loftus and others suggest that a third contributor to the acceptance of abuse and recovered memory testimony is the failure to understand just how malleable memories might be. According to Bruck and Ceci:

The good news in all of this is that since recovered memory cases started appearing about 25 years ago, a rich experimental science of memory errors and a greatly improved understanding of human memory have come about, and much of the “direct scientific evidence” that Bruck and Ceci say was missing is now available.

The experiment by Loftus described at the beginning of this chapter has been followed by dozens of others that have extended the known set of conditions that can produce false memories. Some of this research is based on standard procedures in human memory labs, testing subjects’ memories for lists of words, pictures, or events. This research has led to the identification of some basic characteristics of false memories. One such finding is that the effects of misinformation can be long-lasting; they aren’t limited to a test at the end of a 15-minute session. Another important finding is that subjects can be very confident about the accuracy of their false memories.

The classic Loftus experiment consists of presenting information to a research subject, following this with some misinformation, and then testing the subject’s memory of the original event or information. But false memories can also be created without any false information being presented or experienced at all, as shown by experiments reported over 40 years ago by James Deese at the University of Virginia. Try it yourself: Read the following list of 15 words at a rate of about one word per second. Then, without looking at the list again, try to recall as many of the words as you can, in any order they come to mind. Keep trying for at least one minute.

Insect

Bug

Fright

Fly

Arachnid

Crawl

Tarantula

Poison

Bite

Creepy

Animal

Ugly

Feelers

Small

If you are like Deese’s subjects, the odds are good that you will have recalled the word spider. But this word does not occur in the list! What the word list consists of are the 15 most common free associations to the word spider, and what Deese was doing in his experiments was exploring the influences of association among words on their recallability.

Many years later Henry Roediger and Kathleen McDermott repeated and extended Deese’s experiments and showed how, under certain conditions, people would falsely recall such “critical words” more than 60 percent of the time. One reason for the high rate of false recall in these experiments is fairly obvious: The 15 words in the list are all associates of the critical word spider, and by the time the list of words has been read and processed into memory, it is likely that the mental representation (or trace) of the critical word has been activated or primed several times. This makes it difficult for the subject attempting to recall the list to distinguish between presented and nonpresented words.

There are some interesting but not well-understood findings with this simple experimental procedure. One is that false recall of critical words is greater—sometimes much greater—when the list of words is heard rather than seen. When research subjects see Deese’s word lists, presented one at a time for 1 second each, the rate of false recall is no more than 30 percent. In contrast, rates of

over 60 percent have been reported when the word lists are heard. And just as in Loftus’s experiments, subjects can be very confident in the correctness of their false recalls and recognitions, even when they are warned not to recall words that were not in the list.

Are the results of these simple experiments relevant to the child sexual abuse and recovered memory issue? What does remembering what you think you saw in a slide show or remembering word lists have to with everyday memory? Do these experiments tell us anything about memory in the real world? There is strong evidence that they do. The basic Loftus experiment has been elaborated and extended in many respects. A particularly important one is that false autobiographical (personal) memories can be created under certain conditions. This goes far beyond modification of the memory of small details of an impersonal event. In one such experiment Loftus and colleagues put together a set of descriptions of actual childhood events for their college student research subjects. Most of these events were actual experiences and were provided to the experimenters by parents and siblings. Some of the events were entirely fictitious but involved plausible and personally meaningful or emotional events, such as having been lost in a shopping mall, spilling champagne at a wedding reception, or being hospitalized. The subjects read the narratives, indicated whether they remembered the events or not, and wrote recollections of the events if they did remember them.

Several important outcomes occurred in such experiments. One was that subjects could recognize a high percentage (75 percent or more) of the true narratives, and this probably represents recall of material after something like a 15-year-long retention interval. The second important result was that about 25 percent of the subjects in these experiments also accepted the fictitious narrative and continued to do so in follow-up testing sessions.

Are there limits to the kind of false memory that can be im-

planted? In some studies it has been shown that the plausibility of the false event can have a strong influence on how readily a fictitious event will be accepted as a real one. Kathy Pezdek demonstrated this in an experiment in which Catholic and Jewish adolescents read narratives which included events that could plausibly have occurred during religious ceremonies as well as implausible events (for example, an event that would be plausible for a Catholic ceremony but implausible for an event during a Jewish ceremony, and vice versa). Both groups of subjects could be induced to accept some false but plausible events but were much less likely to do so for implausible events.

However, other studies along these lines suggest that we don’t yet know the limits of false memory formation. In a study by K. Wade and associates (entitled “One Picture Is Worth a Thousand Lies”), subjects were shown a genuine photograph of themselves as children accompanied by a relative, but with the photograph “pasted” digitally into a photograph of a hot-air balloon ride. This was an unlikely event that family members of the subject were sure never happened. At the end of a series of memory-probing interviews, up to one-half of the subjects recalled some or all of this nonevent as having actually happened.

The special power of pictures to induce false recollection has also been demonstrated in an amusing way. Subjects were first shown pictures of advertisements for Disneyland that included the usual cast of characters plus one who would never have appeared there—Bugs Bunny. (He is a Warner Brothers character and property.) By now you won’t be surprised to hear that Bugs started showing up in the subjects’ recollections of their trips to Disneyland.

The original Loftus experiment and variations on it make a strong case that false personal memories can be created relatively easily and that this set of experimental facts bears directly on evaluation of claims of repression and recovered memory. The same

thing can be said about the Deese experimental procedure, which may at first seem to have little to do with the analysis of real-life false memories. But it does because several studies by Richard McNally at Harvard have shown that there are correlations between personality characteristics and the rate of false remembering in the Deese experiment. In one study, three different groups of subjects were tested: people who reported having been abducted by space aliens but were unable to recall the specific events; people who reported having been abducted by aliens and who had recovered memories of it; and a control group of people who said they had never been abducted by space aliens. (We’re not making this up!) The results were most interesting. First, the two groups of “alien abductees” made more recall and recognition errors than the control group. Second, all three groups recalled about the same number of actually presented list words, so that the alien groups were not just worse at learning and remembering word lists than the control group. Third, compared to the control group, the space alien victims were more hypnotically susceptible, had more symptoms of depression, and had more schizotypal characteristics. (Schizotypal refers generally to having odd beliefs and perceptions.) These results are completely consistent with Acocella’s description of the many people diagnosed with multiple personality disorder who have a history of other kinds of psychological disorders as well.

Other studies of personality differences in relation to false memory have shown that IQ scores are negatively correlated with the propensity for false memory: the lower the IQ score, the more the false recollections in memory experiments.

McNally and associates have performed several laboratory-based assessments of claims about repressed and recovered memory. One experiment was intended to test the hypothesis that people claiming recovered memory are prone to forgetting of emotionally negative experiences. He studied three groups of women: those reporting abuse who also had signs of post-traumatic stress disorder (PTSD), women reporting abuse with no PTSD, and women reporting no history of abuse. One of the experimental

tasks was called directed forgetting, in which subjects study a list of words and are told to either remember or forget each word right after studying it. Some of the words were neutral in emotional content (such as desktop) and some were emotionally negative (such as assault). One of the main results was that abuse-PTSD subjects recalled as many negative as neutral words, whether they were “remember” or “forget” words. The other subjects, in contrast, recalled more “remember” than “forget” words, neutral or unpleasant. In other words, subjects who were thought to be prone to forget unpleasant experiences in fact had good memories for such experiences.

McNally’s findings could be dismissed by arguing that artificial and simplified laboratory tests are not going to show the true effects of abuse and PTSD. But what do you think recovered-memory theorists would say if the research had found the expected differences?

Studies of memory in children have contributed greatly to improved understanding of memory accuracy and memory fallibility. Many of these studies were conducted as experiments in which experiences of different kinds were arranged and observed by the investigator in realistic or naturalistic settings. One reason why these studies are so important for the false memory controversy is that an objective record of the actual events is available. Another reason is that these studies have consistently shown that the manner in which a child’s memory is tested strongly affects the accuracy of the child’s recollection.

The episodic memory system of preschool children is sufficiently developed to allow them to accurately recall experiences over relatively long retention intervals when they are tested in ways that do not bias their responses. This has been shown in numerous studies conducted in real-world settings (for example, a child’s memory for events that occurred during a visit to a pediatrician). This fact should not come as a surprise, given what we

saw in Chapter 3 about memory formation in newborns and infants, who clearly have the ability to form and retain memories. These abilities continue to develop into childhood and beyond. But preschool children sometimes seem more susceptible than older children, adolescents, and adults to false memory formation and retrieval.

Research by Bruck and Ceci and many others has shown that children readily form and express false memories about peripheral details of experienced events, just as in Loftus’s original experiments. But what about more salient, personally meaningful, or emotional events? In one experiment by Bruck and Ceci, children visiting a pediatrician’s office for inoculations were exposed to different kinds of statements made by the adults present about how painful their injections had been (“hurt a lot,” “didn’t hurt at all,” or no statement). A week later, they were questioned about how painful the inoculation had been and about how much they had cried. There was no apparent effect of the adults’ statements because the children told that their inoculations were very painful and those told that the inoculations did not hurt at all did not differ from the control group in terms of their memories of the level of pain or how much they had cried. As Bruck and Ceci observed, “5-year-old children are not sponges soaking up misinformation from the environment and incorporating it into their reports.” The single exposure to biasing information was not enough to influence memory for a salient and emotional event.

However, the creation of false memories started to appear when Bruck and Ceci conducted a series of follow-up suggestive interviews about a year later. During these interviews, some of the children were given additional false information about their original reaction to the inoculations as well as misleading suggestions about events that had not occurred at that time. For example, some children were told that the research assistant had given them their shots. These children were more likely to recall this fictitious event than were children not given this (mis)information. Preschool children’s recollections of salient

events can be influenced by biased questioning and interviewing techniques.

What kinds of questioning and interviewing procedures seem likely to promote false recollections in children? The main ones seem to be these:

• Prior beliefs and biases of the interviewers, especially when interviewers are seeking to confirm these beliefs rather than gather information.

• Repeated questioning: A child may respond accurately when first questioned but will start making erroneous responses when questions are asked repeatedly.

• Specific questions: If a child is simply asked “What happened?,” false recollection is less likely than when the question posed is specific and leading.

All of these seem to apply equally well to situations in which adults are being treated for childhood sexual abuse and recovered memory. Elizabeth Loftus and Katherine Ketcham argue in their book The Myth of Repressed Memory that some therapists see it as their job to “probe” and “dig” for repressed memories over a long series of therapy sessions. They encourage patients to visualize (“guided imagery”) and to free associate; they sometimes hypnotize patients; and many therapists are not at all reluctant to make repeated suggestions to patients.

Anyone who knows the psychological science of false memory creation would immediately recognize this as a prescription for the creation of false memories, and anyone who knows something of the early history of psychoanalysis will recognize something else too—history repeating itself. Some scholars argue that Sigmund Freud has a lot to answer for here. Frederick Crewes has studied the early career and writings of Freud most thoroughly. A brief history of it all starts with Freud’s announcement

in 1895 that certain adult psychological disorders had their origin in sexual molestation (or “seduction” as he put it) during infancy and young childhood and that he had discovered this through lengthy questioning and probing of his patients’ memories. The next development was Freud’s announcement a few years later that he had discovered that events related to him by these patients had probably never occurred, and were instead the product of their own repressed fantasies. As Crewes observes, Freud failed to take the next step and admit that he may have been responsible for creating the fantasies during the original treatment sessions.

Memory researchers are devoting a great deal of effort toward identifying the conditions that produce false memories and to integrate this information into memory theory in general. Here are two of the basic questions they’re asking.

1. How does false information affect memory? We have seen how misinformation following an experience can interfere with attempts to recall some aspects of the experience. But does this mean that memory of the original event itself has been “erased”? Isn’t it possible that the misinformation instead blocks access to the original (and still-intact) memory? These questions are basically no different many from of the basic questions we considered in Chapter 4 regarding ordinary forgetting.

Some research seems to indicate that the memory trace itself is not compromised by misinformation. Michael McCloskey showed this in experiments that used the basic Loftus procedures in addition to an important new procedure. In the standard part of the experiment (see Table 6-1), subjects watched a video that showed, among other things, a man entering a room and stealing an object. The man was shown carrying a hammer. In the next stage, misinformation was presented to one group while a control group experienced no misleading information. The misleading information was a statement such as “What color were the handles of the pliers the man was carrying?” The subjects were then given

the same kind of memory test in which they had to choose which of two objects (screwdriver or hammer) they had seen earlier. Subjects exposed to the misinformation were correct 55 percent of the time while the control group was correct 70 percent of the time.

So far, so good; this is just Loftus’s basic misinformation effect. The critical part of McCloskey’s experiment was what happened with a modified procedure (see Table 6-2) in which subjects were treated exactly as the subjects had been in the standard procedure, except that they were given a recognition test that involved selecting between hammer and screwdriver—that is, a choice between the originally seen object and a new object, one that had not been mentioned in the misinformation stage. The rationale for this experiment was that if the misinformation actually degraded the memory trace, there should have been an effect of the misinformation on memory accuracy in this group of subjects as well. The important result was that subjects in this condition of the experiment showed no effect at all of the misinformation on their memory for the original event, recognizing correctly 70 percent of the time in the control group and 70 percent in the

experimental (misinformation) group. McCloskey concluded that the impairment of memory observed in the standard misinformation experiment was due to some kind of interference process that occurred during the memory test itself, rather than any degradation or replacing of the memory trace. Loftus herself has replicated McCloskey’s experiment and its main results.

McCloskey’s experiments have led to intensive attempts to further analyze the cause of the misinformation effect. If misinformation does not alter the original memory trace, why is performance on tests using the Loftus procedure so strongly and reliably affected by misinformation? These research efforts have led to the kind of conclusion that is so common when any moderately complex psychological process is analyzed carefully: The process of misinformation interference with memory is complex. Under some conditions, there is no evidence of an impaired memory trace, just as McCloskey found; under other conditions, there is some evidence of trace impairment or at least alteration of the memory trace. False memory may in some cases be due to source confusions (remembering an event as occurring in a context or emanating from a source other than the original one); it may be due to the negative effects of similarity on memory; it may be due to a blending of true and false information.

2. Can you tell true from false memories? Researchers have compared many characteristics of true and false memories in order to answer this question. For example, depending on the details of the experiment involved, they have compared the confidence in the accuracy of true and false memories, the clarity of the memories, and the speed at which memory probes are answered to. Some studies have in fact found that true and false memories differ along these lines, at least on average. In the Loftus and Pezdek studies mentioned earlier, recollections of events that had never occurred contained fewer words than recollections of actual events. Subjects also rated their own false recollections as being somewhat less clear and had less confidence in the accuracy of these memories than they had in recollections of actual events. However, these results are true on average. In any experiment

some subjects will sometimes be very confident that their false memory is actually true and report the memory as being quite vivid. At present, there is no generally reliable way of distinguishing between true and false memories.

One new approach to the problem is to measure brain activity during false memory experiments. In an experiment by Brian Gonsalves and Ken Paller, subjects were shown words and asked to “visualize” the object referred to by the word. Some of the words were immediately followed by display of a picture of the object and some were not (Figure 6-1). In a later test subjects had to decide if test words had been shown previously as pictures or not. Event-related potentials (ERPs: recordings of electrical activity in cortical regions of the brain) were measured during presentation of the words and pictures as well as during the test. One important finding was that the ERPs for correct test responses (for example, remembering that “apple” had been presented as a picture) were different from those for incorrect responses (a false memory that “hat” had been presented as a picture). Other studies have used magnetic resonance imaging to determine if true and false memories produce different levels of activity in specific brain regions, and they too have yielded evidence that true and false memories can be distinguished to some extent by these kinds of measurements of brain activity, although with considerably less than perfect accuracy.

FIGURE 6-1 A stream of object names, pictures of objects, and imagined objects.

You will be glad to hear that in the Wenatchee case, all those imprisoned have been released and the children returned to their families. The children (now young adults) who were most responsible for the false claims have retracted all the charges. There have been similar outcomes in other cases, but some of those accused, tried, and sentenced are still in prison.

Two more things need to be said about this issue. First, none of the research considered here and none of the conclusions reached are arguments against the existence of real childhood abuse. We also hope that our description of the psychological characteristics of people claiming to have recovered memories is not taken as a particularly cruel form of “blaming the victim.” The important point is that there are reliable empirical facts about false memory that bear in important ways on genuine understanding of what recovered memory and false memory really represent.

Second, a great deal of suffering and wasted time and resources could have been prevented if those people conducting abuse and recovered-memory investigations had adopted the set of principles that the psychologist Robin Dawes argues should be expected of anyone applying psychology to important problems:

• I have studied the problem extensively.

• I have studied alternative hypotheses about the nature, causes, and possible amelioration of the problem.

• I have evaluated these hypotheses in light of the existing evidence.

• I have tentatively concluded that some of these alternative hypotheses are better supported than are others.

• Therefore, I understand something about the problem and how to address it, although new findings may always prove me wrong.

Finally, we want to stress the point that the creation of false memories is not a pathological process. The apparent ease with

which existing memories can be modified and false memories created has to be put into the context of the overall function and purpose of the human learning and memory system. The formation of false memories seems to take place in much the same way as the formation of any kind of new memory.