4. The Sin of Misattribution.

The Sin of Misattribution.

I have been here before,
But when or how I cannot tell:
I know the grass beyond the door,
The sweet keen smell,
The sighing sound, the lights around the shore ...

- "Sudden Light;' Dante Gabriel Rossetti, 1854

On February 24, 1896, members of the Societe Medico-Psychologique in
Paris learned about a bizarre case of memory disturbance. Accounts of patients suffering from amnesia were not uncommon in the 1890S, but the
thirty-four-year-old man described that day had a different problem: he remembered events that never occurred. After suffering from malaria several
years earlier, Louis was regularly overwhelmed by feelings of familiarity
when he encountered situations that were, in reality, entirely novel. In the
midst of his brother's marriage ceremony, he felt certain he had attended
the same event a year earlier. Admitted to a new hospital because of emotional problems, Louis was sure he had been there before. When he first
met Dr. Arnaud, the French psychiatrist who reported his case to the
Societe Medico-Psychologique, he insisted, "You know me, doctor! You
also welcomed me last year, at the same time of the day, and in this same
room. You asked me the same questions, and I gave you the same answers."

Louis was of great interest to the psychologists and psychiatrists who
had assembled in Paris to hear Arnaud's presentation. The late nineteenth
century has been called a golden age for the study of memory, and French
psychology had played a large part in it. Although Ebbinghaus's groundbreaking 1885 experiments are best known today, four years earlier the
French psychologist Theodule Ribot had written a classic book, Diseases of
Memory, which described how brain damage or psychological disturbance
could produce amnesia for the recent or remote past. Ribot also described
cases where memory is present, but wrong. Called "paramnesias" or "false
memories;' these distortions had sparked a spirited and sometimes heated
debate: How widespread are such false memories in the general population? Do they signal the presence of clinical pathology? Is there just one
type of paramnesia, or are there many? Proponents of opposing views debated in a special 1893 issue of the Revue Philosophique.

When Arnaud presented Louis to the Paris meeting of the society in
1896, he placed his problems squarely in the context of the ongoing debate
by rejecting the terms most commonly used to describe the aberrant remembering that Louis displayed. "I believe that it would be better to abandon the words false memory and paramnesia;' he boldly asserted, going on
to argue - seemingly paradoxically - that "the phenomenon in question
may not be associated with memory at all." Arnaud suggested a new expression to describe the inappropriate familiarity that plagued patients like
Louis: the illusion of deja vu. He insisted that deja vu is a special experience, distinct from other kinds of memory distortions, because of its intensity, the conviction that a present experience is identical to a past one, and
the feeling that one knows precisely what is going to happen next.

Arnaud helped to propel the phrase "deja vu" into common usage, but
he was not the first to describe the experience. Dante Gabriel Rossetti captured the feeling of deja vu in his 1854 poem "Sudden Light;' and as early as
1849, Charles Dickens wrote of a similar experience in David Copperfield.
"He seemed to swell and grow before my eyes;' David relates upon encountering Uriah Heap. "The room seemed full of the echoes of his voice; and
the strange feeling (to which no one is quite a stranger) that all this had occurred before, at some indefinite time, and that I knew what he was going
to say next, took possession of me."

But what did Arnaud mean when he claimed that Louis's deja vu experiences "may not be associated with memory at all"? Many prior interpretations of deja vu tended toward the mystical, conjecturing that it reflects memory for a past life and thus provides evidence for reincarnation,
or perhaps involves telepathic eavesdropping into someone else's memory.
Other, less exotic explanations held that people have feelings of deja vu
when a present experience stirs up a similar - though not identicalpast experience. For Arnaud, however, deja vu had nothing to do with either the paranormal or partial memories of a similar past episode. Instead,
he characterized it as a kind of bad judgment: a misattribution of current
sensations and experiences onto the past.

To understand better what Arnaud had in mind in 1896, fast forward a
century and consider a 1993 experiment by the Canadian cognitive psychologist Bruce Whittlesea. Participants first studied a list of common words.
On a later memory test, some words from the study list and some new
words appeared in capital letters at the end of a sentence; participants
judged whether the capitalized words had appeared earlier. In some of the
sentences, the word at the end of the sentence was highly predictable from
the preceding words: "The stormy seas tossed the BOAT:' In other sentences, the final word was less predictable: "She saved her money and
bought a LAMP:'

When the capitalized word had not appeared previously on the study
list, and participants should have responded "new," they sometimes incorrectly responded "old." Most important, people were more likely to claim
incorrectly that they had previously seen the highly predictable new words
than that they had seen the less predictable new words. Participants also
named highly predictable words faster than they named less predictable
words. Whittlesea suggested that participants misattributed their fast responses for the highly predictable words to an earlier encounter that, in
fact, had never occurred: speedy, fluent responding - a consequence of
the word's predictability - was misinterpreted as familiarity.

In this experiment, then, people claimed that they had a prior experience - seeing a word on a study list - for reasons having nothing to do
with memory, just as Arnaud had asserted about deja vu a century earlier.
Deja vu, he thought, might occur because features of a present situation
trigger responses, perhaps analogous to the fluent processing of predictable
words in Whittiesea's experiment, that are mistakenly attributed to a past
experience.

Deja vu occurs relatively infrequently, and there is still no convincing
explanation of precisely what features of a present experience would produce the kinds of mistaken judgments that Arnaud theorized about to his
Parisian audience. Yet misattributions in remembering are surprisingly
common. Sometimes we remember events that never happened, misattributing speedy processing of incoming information, or vivid images that
spring to mind, to memories of past events that did not occur. Sometimes
we recall correctly what happened, but misattribute it to the wrong time or
place. And at other times misattribution operates in a different direction:
we mistakenly credit a spontaneous image or thought to our own imagination, when in reality we are recalling it - without awareness - from
something we read or heard. Though we know little more about deja vu today than we did back in the days of Arnaud over a century ago, we have
learned a great deal about other forms of misattribution. This is hard-won
knowledge with potentially vital consequences for society: misattribution
can alter our lives in strange and unexpected ways.

EYEWITNESS MISATTRIBUTIONS AND SOURCE MEMORY.

Anybody who remembers the Oklahoma City bombing in 1995 probably
also recalls the failed search for John Doe 2. John Doe 1 - soon identified
as Timothy McVeigh - was apprehended shortly after the bombing in
April of that year. At the same time, the FBI mounted a nationwide manhunt for a second suspect they believed had accompanied McVeigh when
he rented a van from Elliott's Body Shop in Junction City, Kansas, two days
before the bombing. An artist's sketch of John Doe 2, depicting a young
square-faced man with dark hair and a stocky build wearing a blue and
white cap, appeared continually on television and was featured in newspapers around the country. Despite massive efforts that resulted in the successful prosecution of McVeigh and his friend Terry Nichols, and polls indicating that seven in ten Americans believed that another accomplice had
eluded the law, John Doe 2 was never found. What happened?

After tracing McVeigh's rental van, the FBI interviewed employees at
Elliott's Body Shop. A secretary and the shop owner recalled that only one
man, matching McVeigh's description, had rented a van on April 17, 1995,
two days before the bombing; he had made his reservation under the alias
"Robert Kling." The mechanic Tom Kessinger, who watched the transaction, recalled seeing two men. One fit McVeigh's description: tall and fair
with short blond hair. The other was shorter and stockier, dark-haired,
wore a blue and white cap, and had a tattoo beneath his left sleeve. Based on
Kessinger's recollection, the search for John Doe 2 began.

The source of Kessinger's memory, however, appears to lie in an unrelated visit to Elliott's Body Shop a day later, when Army Sergeant Michael Hertig and his friend, Private Todd Bunting, also rented a van in
Kessinger's presence. Hertig, like McVeigh, was tall and fair. Bunting was
shorter and stockier, dark-haired, wore a blue and white cap, and had a tattoo beneath his left sleeve - a match to the description of John Doe 2.

After initiating the unsuccessful manhunt for the elusive second suspect, FBI agents reviewed records of Hertig and Bunting's visit to Elliott's
Body Shop. They reluctantly concluded that John Doe 2 was Private Todd
Bunting, an innocent man with no connection to the bombing. Kessinger
had correctly recalled Bunting's features, depicted in the infamous picture
of John Doe 2 that circulated nationwide, but had misattributed them to
the wrong episode a day earlier.

This kind of mistaken identification is not unprecedented. In a famous case from the mid-1950s, a British ticket agent, robbed at gunpoint,
later identified an innocent sailor as the gunman. The sailor had previously
purchased tickets from the same agent, who had misattributed the familiarity of the sailor's face to the robbery. In a later incident, the psychologist
Donald Thomson was accused of rape based on a victim's detailed memory
of his face. Thomson was cleared because he had an impeccable alibi: he
was in the midst of a live television interview (ironically, on the fallibility of
memory) at the moment the rape occurred. The victim had been watching
the show and misattributed her memory of Thomson's face to the rapist.

Both Thomson and the British sailor were fortunate to escape wrongful imprisonment. But how many other times have similar misattributions
produced inaccurate eyewitness testimony leading to the conviction of an
innocent person? Nobody knows for certain, but consider two facts. First,
according to estimates made in the late 1980s, each year in the United States
more than seventy-five thousand criminal trials were decided on the basis
of eyewitness testimony. Second, a recent analysis of forty cases in which
DNA evidence established the innocence of wrongfully imprisoned individuals revealed that thirty-six of them (90 percent) involved mistaken eyewitness identification. There are no doubt other such mistakes that have
not yet been rectified.

These chilling numbers create a sense of urgency regarding the need
to understand better the nature of eyewitness misattributions, and to take
steps to minimize them. The specific type of misattribution in the John
Doe 2 incident is sometimes referred to as "unconscious transference:' The
idea is that a witness such as Kessinger incorrectly attributes a face's familiarity to the wrong source because he unconsciously transfers memory of
the individual from one context to another. Recent laboratory studies indicate that when individuals make an eyewitness misidentification, they are
not necessarily unaware of having previously encountered a person in multiple contexts. For example, participants who watched a film of a robbery
that included an innocent bystander in a separate scene sometimes later
mistakenly identified the bystander as the robber. But the process that led
to the misidentification was not entirely unconscious: many participants
believed - incorrectly - that the bystander and the robber were the same
person.

Whether unconscious or not, the kinds of eyewitness misidentifications observed in cases such as John Doe 2 and others fit well with research
showing that people often have sketchy recollections of the precise details
of previous experiences - when and where they encountered a person or
object. This vagueness creates fertile soil for the occurrence of "source
misattributions," in which people recall correctly a fact they learned earlier,
or recognize accurately a person or object they have seen before, but misattribute the source of their knowledge. Experiments have shown, for instance, that people may remember perfectly well that they saw a previously
presented face but misremember the time or place that they saw it, much as
happened to Tom Kessinger in Elliott's Body Shop.

Consider what's involved in remembering the details of what a person
looks like and where you saw him. You attend a business meeting at a palatial downtown office on Tuesday morning and meet two executives with
whom you will be negotiating: Thomas Wilson, a silver-haired vice president wearing horn-rimmed glasses and a conservative blue suit, and Frank
Albert, a thirty-something financial analyst wearing a bow tie and colorful
suspenders. Later that afternoon, you head out to the suburbs to meet with
two prospective clients who have just started a new company in rather
cramped quarters. The computer programmer, Eric Merton, is a recent college graduate wearing jeans and a silver earring; the company president,
Elaine Green, is a slightly older woman dressed in a more traditional business suit.

If I ask you a week later about your meetings last Tuesday, to provide
an accurate report you will need to remember the individual features of the
people and places you visited. But it is not sufficient to recall a vice president, financial analyst, computer programmer, and company president;
horn-rimmed glasses, colorful suspenders, a bow tie, silver earring, jeans,
and traditional business suits; Mr. Wilson, Mr. Albert, Mr. Merton, and Ms.
Green; a large downtown office and a smaller one in the suburbs. You also
need to remember which person was wearing what and which face goes
with which name, who worked out in the suburbs and who worked downtown, and what position each person holds. In addition to recording and
retrieving the individual features, you need to link them together in memory so that you can recall the correct conjunctions of people, attire, positions, and places.

Psychologists refer to this linking process as the problem of "memory
binding": gluing together the various components of an experience into a
unitary whole. When individual parts of an experience are retained but
memory binding fails, the stage is set for the kinds of source misattributions seen in the John Doe 2 incident and in other episodes of mistaken
eyewitness recollection.

Source confusions are sometimes attributable to a binding failure: at
the time an event occurs, an action or object is not properly bound to a
particular time and place. Binding failures may also contribute to memory
confusions between events we actually experience and those we only think
about or imagine. About to leave your house, you think about closing and
locking the basement door. An hour later in the car, you are suddenly
gripped by a sense of panic: Did I actually close the door or did I only
imagine doing it?

After he retired, the psychology professor Lew Lieberman became increasingly vexed by such confusions. "It is as if before you do something;'
he reflected, "you kind of picture yourself doing it and then later, you do
not remember whether it was just the picture or the reality:' He wondered
whether others experience something similar. In fact, numerous experiments have shown that when people imagine seeing an object, or carrying
out an action, they sometimes later claim that they actually perceived the
object or performed the action.

In one particularly neat experiment, younger and older adults saw
an object such as a magnifying glass and researchers later asked them to
imagine a lollipop (similar object), or saw a hanger and later imagined a
screwdriver (unrelated object). Older adults were more likely than younger
adults to insist that they had actually seen the imagined lollipop, but were
no more likely than younger adults to claim they had seen the imagined
screwdriver. Older participants seemed to have special difficulty binding
the appearance of perceived objects (for example, a "round shape") to the
context of presentation. So, after seeing one "round shape" - the magnifying glass - and imagining another similar shape - the lollipop - elderly
participants could not recall details associated with the actual perception of
the magnifying glass, and were thus prone to source misattribution.

If associated details are bound together with an object or action, it becomes easier to recall whether an incident actually occurred. Fretting in the
car about whether you've left your basement door wide open, you carry out
a frantic mental search, trying to recall some specific object or action that
proclaims that you indeed carried out what you had thought about doing.
Your mind eases as you remember seeing a cat running away when you
closed the door. But if you hadn't bound together the perception of the
frightened animal with the act of closing the door, you might still be trying
to sort out imagination from reality.

Binding failures can also result in a striking illusion known as a
"memory conjunction error." Having met Mr. Wilson and Mr. Albert during your business meeting, you reply confidently the next day when an associate asks you the name of the company vice president: "Mr. Wtlbert."
You remembered correctly pieces of the two surnames, but mistakenly
combined them into a new one. Cognitive psychologists have developed
experimental procedures in which people exhibit precisely these kinds of
erroneous conjunctions between features of different words, pictures, sentences, or even faces. Thus, having studied spaniel and varnish, people
sometimes claim to remember Spanish. Or having seen the drawings of the
two faces shown in Figure 4.1, people often claim to remember having seen
a new face (also shown in the figure) that combines features of the two that
they had actually seen. If individual features of the words or faces are retained, but are not bound together adequately when people initially study
them, memory conjunction errors can result.

FIGURE 4.1 After seeing each of the faces in the left and center of the figure, people later claim erroneously to remember seeing the face on the right, which contains conjunctions of elements from the other faces. This type of misattribution is
known as a "memory conjunction error."

Recent studies of brain-injured patients suggest that the hippocampus
plays an important role in binding processes that, when disrupted, contribute to memory conjunction errors. Patients with damage restricted to the
hippocampus are even more likely than healthy controls to make memory
conjunction errors for recently studied words and faces. The patients perceive faces and words as wholes, but when tested only a few seconds or
minutes later, they are likely to miscombine features of separate faces or
syllables of different words. The damaged hippocampus no longer provides
the mnemonic glue needed to hold together parts of a face or word in
memory. This idea receives further support from recent brain imaging
studies using PET scans. The hippocampus became especially active when
people learned pairs of unrelated words (for example, "level/need"), which
places heavy demands on binding processes.

Source misattributions and memory conjunction errors can also occur because of faulty memory retrieval processes. When a face seems familiar, people need to reflect on, or "monitor," the outputs of memory to determine why. Patients whose frontal lobes have been damaged by stroke, or
partially removed during surgery, have difficulty engaging in such retrieval
monitoring processes. They tend to make snap judgments about the source
of a feeling of familiarity, and hence make more source misattribution errors than do healthy controls.

Healthy elderly adults who perform poorly on tests that are sensitive
to frontal lobe abnormalities also tend be especially prone to source misattributions. Recall that older adults often claimed that they had previously
seen a lollipop they had only imagined, when in reality they had actually
seen a similarly shaped magnifying glass. When participants took this test
two days after seeing and imagining objects, older adults who scored most
poorly on other tests that are sensitive to frontal lobe damage also made the
most confusions between perceived and imagined objects. But when memory testing was carried out only fifteen minutes after seeing and imagining
objects, there was no relationship between these source misattributions
and scores on frontal lobe tests. Monitoring processes that depend on the
frontal lobes are probably most heavily taxed after a two-day delay, when
trying to recall whether you perceived or imagined an object is very difficult and requires considerable reflection. If only fifteen minutes have
passed, then the memory task is easier and does not draw on the frontal
lobes as much.

Related retrieval failures also contribute to memory conjunction errors. Susan Rubin and her collaborators found that in a sample of older
adults, those who performed most poorly on tests of frontal lobe function
also tended to make many memory conjunction errors, such as claiming to
remember barley from an earlier list when they actually saw barter and valley. These elderly adults failed to scrutinize their memories sufficiently, relying instead on the strong sense of familiarity engendered by seeing a conjunction word like barley.

A strong sense of general familiarity, together with an absence of specific recollections, adds up to a lethal recipe for misattribution. Understanding this point may be a key to reducing the egregious consequences of
rnisattribution in eyewitness testimony. Gary Wells and his group at Iowa
State University have shown that common lineup identification practices
may often promote misattribution because people are encouraged to rely
on familiarity. In standard lineup procedures, witnesses are shown a number of suspects; after seeing all of them, they attempt to identify the culprit.
Wells finds that under such conditions, witnesses tend to rely on relative
judgments: they choose the person who, relative to the others in the lineup,
looks most like the suspect. The problem is that even when the suspect is
not in the lineup, witnesses will still tend to choose the person who looks
most like him. Witnesses rely on general similarities between a face in the
lineup and the actual culprit, even when they lack specific recollections.
Happily, however, Wells has also shown how to minimize reliance on such
relative judgments: ask witnesses to make a "thumbs-up or thumbs-down"
decision about each suspect right after seeing the face, instead of waiting
until all suspects' faces have been displayed. This procedure encourages
people to scrutinize their memories carefully and examine whether the pictured suspect matches the details of their recollections. Thankfully, law
enforcement officials are learning more about this and related methods
to increase eyewitness accuracy. In early 1998, Attorney General Janet
Reno formed a working group consisting of psychologists (including Gary
Wells), police, and attorneys to develop guidelines for collecting eyewitness
evidence. The working group published a widely available set of guidelines
based on rigorous scientific studies.

Wells's studies on reducing eyewitness false alarms involve separating
the mnemonic wheat from the chaff - creating conditions that induce
people to rely on accurate recollections of what really happened rather than
being misled into errors based on general resemblance. To me, they also
lead to a fundamental question with far-reaching implications: Is it possible to tell true memories from false?

LOOKING FOR A TRUTH MACHINE.

In the summer of 1996, I helped to run a cognitive neuroscience institute at
Dartmouth College. My family and I stayed at a lovely country inn located
in the nearby Vermont countryside. Returning there after a day of lectures
by distinguished speakers, I was stunned by an unexpected, even surreal
sight: the door to our room was virtually covered with slips of paper, each
containing a phone message from a newspaper, television news show, or radio program. Media from allover the world wanted to talk to me - right
away.

That morning, an article in the Tuesday science section of the New
York Times described new PET scan studies I had conducted with several
colleagues which examined brain activity while people experienced true
and false memories. Though other studies had used PET and fMRI to peer
inside the brain as participants recalled true memories of previous experiences, none had looked at brain activity when they called up false memories of incidents that had never occurred. The possibility that brain imaging might function as kind of a high-tech lie detector, flawlessly sorting out
true from false memories, is undeniably intriguing.

It is easy to have people recall true memories during a scan: ask them
about words or pictures that you showed them before scanning, or inquire
about past experiences outside the laboratory. But how does one induce
false memories in a PET scanner? A year before our PET study, the psychologists Henry 1. Roediger and Kathleen McDermott rediscovered a procedure developed by James Deese in the 1950S which reliably leads people to
insist that they experienced an event - the occurrence of a word in a list
- that never actually happened (the procedure is referred to as the "DRM"
or Deese/Roediger-McDermott procedure). The experimenter first reads
out lists of associated words. One list, for instance, would contain thread,
pin, eye, sewing, sharp, point, prick, thimble, haystack, thorn, hurt, injection,
syringe, cloth, knitting. Another list might contain bed, rest, awake, tired,
dream, wake, snooze, blanket, doze, slumber, snore, nap, peace, yawn, drowsy.
On a later memory test, subjects decide whether each of several words had
been read aloud earlier: sewing, door, needle, sleep, candy, awake. Most of
the time, people correctly remember that they had earlier heard sewing and
awake, and correctly state that they had not heard door and candy. More interestingly, people frequently claim - confidently but incorrectly - that
they heard needle and sleep. You might even have made this error yourself
as you looked over the test words.

This false memory effect occurs because all the words in the first set
are associated with needle and all the words in the second set are associated
with sleep. Hearing each word in the study list excites or activates related
words. Because needle and sleep are related to all the associates, they become more activated than other words - so highly activated that only
minutes later, people swear that the experimenter said the word. Could
PET scans distinguish between these true and false memories, even though
the experimental participants themselves do not?

A few minutes before they entered the scanner, participants in our experiment heard a series of associate lists. Then, during one scan they made
recognition judgments about previously presented words, such as sewing or
awake, and during another scan, they made judgments about associated
words that had not been presented, like needle and sleep. As expected, people claimed to remember the nonpresented words almost as often as they
claimed to remember the presented ones. Brain activity was, overall, remarkably similar during true and false recognition: a network of regions
showed heightened activity regardless of whether people were claiming to
remember words they had heard previously or associates they only thought
they had heard. The frontal lobes responded very strongly, and there were
also signs of activity in the inner parts of the temporal lobe, near the hippocampus, during both true and false recognition. Because the hippocampus and surrounding areas play such an important role in true memories,
we thought that activation in this area during retrieval of a false memory
might mislead people into feeling confident of hearing a word that was
never actually presented.

But despite the striking similarities between the regions activated during true and false recognition, there were also tantalizing hints of differences. A part of the frontal lobe thought to be involved in scrutinizing or
monitoring memories showed greater activity during false than true recognition. It was as if people sensed something odd about words like sleep and needle, and were scrutinizing them especially carefully before capitulating
to the powerful memory illusion. There was also more activity during true
than false recognition in a part of the temporal lobe on the surface of the
left hemisphere - a region that stores the sounds of words. Was the PET
scan picking up a faint echo of hearing a word that had actually been presented?

The possibility of using brain imaging to separate truth from fiction
- perhaps in a therapist's office or court of law - has a surreal, futuristic
appeal. In James Halperin's imaginative novel The Truth Machine, brainscan technology has advanced to the point at which it unfailingly separates
truth from deceit. Politicians must now make promises under the watchful eye of the scanner, which will immediately reveal devious intentions.
Though telling truth from intentional deceit is different from distinguishing between true and false memories - the liar intends to be deceptive,
whereas the faulty rememberer tries to be truthful- the prospect of a
"memory truth machine" captured the imagination of the reporters whose
phone numbers were plastered across my door. Could we now use PET
scans to settle disputes over recovered memories of childhood abuse, where
one person vividly remembered horrific abuse and another denied it just as
firmly? Could they help to decide whether the memory of an eyewitness is
accurate?

These kinds of questions are fascinating, and their potential implications for society are enormous. But our results forced me to throw cold water on speculations about these far-reaching consequences. The similarities
between true and false recognition were striking and widespread, the differences small and no more than suggestive. We had used an experimental
task whose relation to everyday life is unknown, and had looked at only one
type of testing condition. We didn't know whether we'd get the same results
if we changed any aspect of our procedure. Based on our experiment, brain
imaging would not be used soon to decide between true and false memories in the courtroom or anywhere else.

We soon followed up our initial findings, and the results justified my
caution. Differences in brain activity during true and false recognition
turned out to depend on the details of the testing procedure. Because of
limits on PET technology, we had to test all previously studied words during one scan, all non presented associates of studied words in a second, and
unrelated new words in a third scan. This feature of testing encouraged
subjects to scrutinize their memories carefully before responding "old" or
"new," because all the words within a particular scan would tend to seem
equally familiar (or unfamiliar) to them. We reasoned that such careful
scrutiny contributed to the different patterns of brain activity during true
and false recognition.

To test this idea, we recorded electrical activity in the brain from
various scalp locations using "event-related potentials;' which reflect the
brain's electrical response to specific sensory stimuli. Event-related potentials offer the ability to track brain activity over just a few thousandths of a
second. Unlike PET - which provides a picture of brain activity averaged
across a minute or so - the event-related potential technique allows us to
mix together studied words, non presented associates, and unrelated new
words during a single memory test. Under these conditions, previously
studied words or associate lures tend to jump out as familiar in comparison
to unrelated new words, so people are more likely to make snap judgments
about them. With this type of test we saw no reliable differences in electrical activity during true and false recognition.

These results contain an important positive lesson. Just as in Gary
Wells's studies showing that certain test conditions reduce the incidence of
eyewitness false alarms, the physiological data suggest that test conditions
that encourage people to carefully examine their memories broaden differences between true and false memories. Further studies using various procedures for inducing false recognition support this conclusion.

Other researchers have used electrical recordings to examine brain
activity during memory conjunction errors (for example, remembering
Spanish after seeing spaniel and varnish). Memory conjunction errors occur because people misattribute the strong familiarity elicited by two previously exposed syllables to having seen them together as a single integrated
unit. Results suggest that it is possible to tell apart memory conjunction errors from true recollections. Electrical responses differed when college students correctly remembered words that were shown earlier, compared to
when they incorrectly claimed to remember totally new words, or "syllable
lures" (new words that shared a syllable with a previously studied word,
such as Spanish after seeing only varnish). When students made memory
conjunction errors, electrical responses were clearly distinguishable from
those associated with accurate remembering, but were indistinguishable
from electrical responses that accompanied false alarms to new words or
syllable lures.

False alarms were produced by a general sense of familiarity that was
strongest for conjunction words and weakest for entirely new words. Conjunction words were, to some extent, just as familiar as the words that were
actually presented: in each case, participants had seen both of the word's
syllables. But correct "old" responses involved specific recollections of having seen the two syllables together. These detailed memories were associated with very different patterns of electrical activity than were the general
feelings of familiarity which misled people to say they had seen the new
conjunction words.

Several related studies using electrical recordings or fMRI have also
shown that brain activity differs when people call up specific recollections
of past experiences compared with when they respond on the basis of general familiarity. Further, people who are highly susceptible to the Deese/
Roediger- McDermott memory illusion - they remember falsely as often
as they remember accurately - show identical brain electrical activity during true and false memories. But people who are less susceptible to the illusion - they remember accurately more often than they remember falsely
- show different patterns of brain activity during accurate and inaccurate
remembering.

These results suggest that misattribution can be mitigated by encouraging people to base their memory decisions on specific recollections,
rather than relying on overall familiarity. In the Deese/Roediger-McDermott procedure, for example, having heard numerous associated words,
people may be lulled into basing their memory decisions on whether a test
word is strongly associated with previous studied ones, and thus seems
highly familiar, rather than demanding specific recollections.

Lana Israel and I tested this idea by showing people pictures at the
same time they heard lists of semantic associates. So, for example, when
they heard a list including butter, flour, milk, and dough, each word was accompanied by a picture - a stick of butter, pile of flour, carton of milk,
ball of dough, and so forth. Later, we asked participants whether they remembered studied words like butter and associates that had not been studied, such as bread. We thought that the pictures would be so distinctive and memorable that people would later claim to remember having heard a
word only when they could also recollect having seen a picture. That is exactly what we found.

Based on several experiments, we hypothesized that studying pictures
along with words helped experimental participants to invoke a "distinctiveness heuristic": a rule of thumb that leads people to demand recollections
of distinctive details of an experience before they are willing to say that they
remember it. Consider the following question: Do you recall that one page
earlier in this book I confessed that I suffer from a multiple personality disorder, and that I actually have nineteen separate personalities, each with a
different name? You can confidently assert that I never said any such thing
because you invoke a distinctiveness heuristic: if I had made such a confession one page earlier, you would have been startled; surely you would possess a detailed recollection of what I wrote and how you reacted to it. We
can invoke a distinctiveness heuristic whenever we expect that our memories will contain rich and detailed information about an experience. In
experiments that use the Deese/Roediger-McDermott word associates procedure, however, people typically do not expect to retrieve distinctive recollections of specific words, and so are misled into falsely recognizing associates that they had never studied. But after studying pictures along with the
words, participants expect more from their memories. They easily reject
items that do not contain the distinctive pictorial information they are
seeking - much as you easily rejected my assertion about multiple personalities.
The distinctiveness heuristic can help older adults to avoid false recognition. Elderly adults are sometimes especially prone to false recognition.
They have a harder time than younger adults calling up specific recollections, and tend to rely more on general familiarity - a potent combination
for producing misattribution. Yet when provided with highly memorable
information to study, older adults can invoke a distinctiveness heuristic as
effectively as younger adults to reduce false memories.

However, older adults often do not expect to recollect specific details
of past experiences; in fact, they may expect to recall little or nothing. Unfortunately, expecting little from their memories can create serious problems for the elderly. As the cognitive psychologist Larry Jacoby has pointed
out, con artists know precisely how to exploit this feature of aging memory.
The Cleveland Better Business Bureau warns of a scam called "Where's the
Check?" Con artists collect personal information from older adults during
a telephone conversation. When they call back the next day, the crooks
determine whether the senior has forgotten the conversation, and hence
would be likely to forget other events. If so, the con artist makes a false
claim about an incident that never occurred, such as, "We received your
check for $1200, but it should only have been for $950. Send us another
check for $950 and we'll simply return the first check to you," In another
variant, the scam artist asserts, "Our records indicate that you paid $2400,
leaving a balance of only $600. Let's make out a check today to clear the
balance:' Not remembering the conversation - and not expecting to
many embarrassed seniors send along the check to avoid further problems.

This sad, expensive outcome results from a failure to invoke a distinctiveness heuristic: if I had sent a check for $1200, or for $2400, I surely
would have remembered it. Because many seniors ordinarily tend to recall
relatively little distinctive information about past experiences, some don't
expect to remember writing the check and are therefore not surprised
when it seems that they've forgotten. Our studies using the Deese/Roediger-McDermott procedure and related tasks show that when armed witll
specific recollections, older adults can use the distinctiveness heuristic effectively. With the aging of the baby boom generation, growing numbers of
people will no doubt become targets of similar frauds that prey on fuzzy
memories and low expectations for specific recall. To reduce the vulnerability of an aging population to such scams, it would be worthwhile to try
to alter older adults' expectations of their own memories, perhaps by incorporating training into memory courses for senior citizens which explains
the distinctiveness heuristic and how to use it effectively.
The good news from our research is that with a little guidance, older
adults can guard against false memories by learning to scrutinize their
memories carefully in order to avoid distortions and errors.

SEEING FILM STARS EVERYWHERE.

When defenses against misattribution are seriously damaged, however,
people make bewildering, even bizarre claims about the past that sever the
connection between memory and reality. In 1991, a British photographer in
his mid-forties, known in the medical literature by the initials MR, started
to have problems with vision and then memory. He had difficulty recalling
events from the recent and remote past. More disturbingly, MR experienced intense feelings of familiarity about people he did not know. He began asking his wife regularly whether a passing stranger was "somebody"
- a screen actor, television newsperson, or local celebrity. MR became so
convinced that his feelings were real that he often could not resist approaching befuddled strangers and asking whether they were indeed famous celebrities. Vexed by the sensation that he was "seeing film stars
everywhere;' MR sought help from a psychiatrist who concluded that the
false familiarity did not originate from psychological problems.

When given formal tests, MR recognized the faces of actual celebrities
as accurately as healthy volunteers. But MR also "recognized" more than
three-quarters of unfamiliar faces, whereas healthy controls hardly ever
did. Neurological exams revealed that MR had been afflicted by multiple
sclerosis. The disease, which attacks the myelin sheath that protects nerve
cells, had caused damage in the vicinity of the frontal lobes.

The fact that MR's frontal lobes were compromised by multiple sclerosis provides an important clue to the origin of his unusual disorder
(most multiple sclerosis patients are not afflicted with this type of recognition disorder). A similar clue comes from work by the University of Arizona neurologist Steven Rapcsak, who has described patients who falsely
recognize new faces after damage to the lower and inner parts of the right
frontal lobe.

The damaged frontal regions normally play an important role in assessing or monitoring signals provided by other neural systems. In cases of
false facial recognition, brain damage may have resulted in faulty connections between frontal systems and those elsewhere, which seem to be involved in face recognition. The British neuropsychologist Andrew Young
has proposed that encountering a familiar face excites a "face recognition
unit:' containing a description of what the person's face looks like. When
activated, this unit sends out signals that we take as a sign that the face is familiar to us. These signals, however, do not provide any details concerning
the person's identity. Recall of such information requires activation of a
separate "person identity node" (see Chapter 3) that contains details of a
person's occupation, interests, background, and related information.

Rapcsak suggests that patients with frontal lobe damage do not sufficiently monitor or scrutinize signals generated by weakly activated face
recognition units located elsewhere. Several lines of research indicate that
regions near the back of the brain, in the lower parts of the temporal lobe
and nearby areas in the occipital lobe, record and retrieve visual descriptions of faces. Single-cell recordings from monkeys, for instance, have revealed "face cells" that respond more strongly to faces than to other objects.

And recent fMRI studies in people have shown something similar. The
fusiform gyrus, a key part of the visual regions in the rear of the brain,
shows exceptionally strong activity when people look at faces compared to
many other kinds of visual objects. Damage to the fusiform gyrus typically
results in loss of the ability to recognize well-known faces as familiar.

According to Rapcsak and others, when we encounter a face, the
fusiform region becomes highly activated, which excites face recognition
units. But because these units contain only visual information, the source
of familiarity is left unspecified - we don't know whether the face seems
familiar because we have encountered it before or because it resembles
other faces we know. For a familiar face, a face recognition unit should excite a related person identity node, allowing us to recall details about the
person. Problems arise when a novel face excites a face recognition unit
producing a weak feeling of familiarity - but does not call forth detailed
information about the person from a person identity node. Frontal monitoring systems must now intercede and demand recall of person-specific
information. The patients with frontal lobe damage studied by Rapcsak
and coworkers fail to engage spontaneously in such monitoring operations,
instead blithely accepting signals from an activated face recognition unit as
indicators of familiarity. Importantly, Rapcsak was able to reduce false recognition of faces in his patients by requiring them to respond "familiar"
only when they could also produce specific information about a person.
Because patients cannot produce specific information about unfamiliar
faces, they manage to resist the impulse to call the face familiar.

In addition to "seeing film stars everywhere;' MR also frequently
claimed to recognize made-up names that were constructed to sound like
pop stars (Sharon Sugar) or historical figures (Horatio Felles). When asked
about the identity of the nonfamous people who seemed so familiar to
him, MR could do no better than generic labels: singer, politician, or sports
star. In striking contrast, however, MR did not falsely recognize made-up
place names: he knew that Jakarta is an actual city and that Wabera is not.
Likewise, MR did not falsely recognize such made-up words as legify or
florrical. MR's problems are restricted to recognizing people, suggesting
that frontal systems can fail miserably in a specific domain while performing normal monitoring operations in other areas.

We don't precisely understand this phenomenon. However, the finding could illuminate one of the weirdest of all misattributions: the Fregoli
delusion. In 1927, the French psychiatrists Courbon and Fail described a
schizophrenic woman who believed she was a "victim of enemies." The patient felt certain that two French actresses were attempting to persecute her.
Courbon and Fail named the delusion after the Italian actor Leopoldo
Fregoli, who delighted Parisian audiences of the time with his ability to
mimic other people. The hallmark of the Fregoli delusion is an unshakable
belief that a stranger is "inhabited" by a friend, relative, or famous person.
Whereas patients such as MR experience general feelings of false familiarity, Fregoli patients are victimized by specific false memories.

The Fregoli delusion usually occurs in psychiatric patients, but neurologists and neuropsychologists have recently reported cases in which the
delusion occurs after brain injury even when patients do not have a prior
psychiatric history. In one case, a twenty-seven-year-old woman from Madeira, studying English in London, suffered a severe head injury when she
fell to the road from a London bus that unexpectedly moved forward as she
was about to exit. Patient IR suffered severe damage to the lower and inner
parts of the right frontal lobe - regions implicated previously in abnormal false recognition - as well as injury to other parts of the frontal cortex. As she was recovering in the hospital, IR became convinced that a female patient in a nearby bed was her mother. The subjective conviction was
so powerful that IR attempted to get into bed with the bewildered patient
several times, and followed her to other parts of the hospital after she was
moved. The Fregoli delusion finally subsided after a month, when IR's father confirmed that her mother was in a hospital back home in Madeira. IR
had twisted one bit of accurate knowledge - that her mother was in a hospital- into a compelling delusion.

Formal testing showed that IR was plagued by memory problems,
from time to time came up with stories or confabulations about events that
had never occurred, and even developed a delusion that a young nephew
was being cared for elsewhere in the hospital. The researchers who studied
IR concluded that part of her problem involved a breakdown in frontal
lobe monitoring systems that normally scrutinize memories for plausibility
and coherence. IR's difficulties seemed to involve misinterpretation of signals from particular person identity nodes. IR did not "see film stars everywhere" but instead became confused about the identity of a specific individual. We still don't know exactly why different patients develop different
forms of misattribution, but I suspect that brain imaging techniques will
soon help to unravel that.

WHAT A GREAT IDEA I HAD: THE PERILS OF CRYPTOMNESIA.

William Wallace is a legendary figure in Scottish history. Popularized by
Mel Gibson's portrayal in the 1995 film Braveheart, Wallace was also the
subject of an acclaimed biography authored in the same year by the Scots man James Mackay. But Mackay's world soon fell apart amid allegations
that he had plagiarized large sections of his book from a 1938 biography of
Wallace written by the late Scottish historian Sir James Ferguson.

"I'm just totally unaware of it, this is purely unconscious, I do assure
you," pleaded Mackay. "I have tried always to find new material on the people I've written about." Is it possible to reproduce significant parts of another person's work without being aware of the material's origin? Mackay's
plea of unconscious influences should be viewed with some skepticism: he
was accused of blatant plagiarism in other books, and the Scottish historian
Geoffrey Barrow called his biography of Wallace "the worst case of plagiarism I've seen for quite a while, and maybe even the worst ever:' But there
is other evidence that people can, in good faith, produce from memory
another individual's writings or ideas while unknowingly misattributing these creations to themselves - a type of misattribution known as
"cryptomnesia." Cryptomnesia constitutes a mirror image of some of the
misattributions considered earlier in the chapter. In false recognition, for
example, people misattribute a feeling of familiarity to a novel event,
whereas in cryptomnesia, people misattribute novelty to something that
should be familiar.

During the early 1900S, the psychoanalyst Carl Jung discovered that
Friedrich Nietzsche had taken parts of Also sprach Zarathustra from a story
he had read in his youth. Nietzsche wrote:
Now about the time that Zarathustra sojourned on the Happy Isles, it
happened that a ship anchored at the isle on which the smoking moun tain stands, and the crew went ashore to shoot rabbits. About the
noontide hour, however, when the captain and his men were together
again, they suddenly saw a man coming towards them through the air,
and a voice said distinctly: "It is time! It is high time." But when the fig ure drew close to them, flying past quickly like a shadow in the direc tion of the volcano, they recognized with the greatest dismay that it
was Zarathustra.

Jung noted the resemblance to an old ghost story written by the German physician and poet Kerner:
The four captains and a merchant, Mr. Bell, went ashore on the island
of Mount Stromboli to shoot rabbits. At three o'clock they mustered
the crew to go aboard, when, to their inexpressible astonishment, they
saw two men flying rapidly towards them through the air ... They
came past them very closely, in the greatest haste, and to their utmost
dismay descended amid the burning flames into the crater of the terri ble volcano, Mount Stromboli. They recognized the pair as acquain tances from London.

The similarities between the two passages are unmistakable, but Jung
concluded that Nietschze had not intentionally copied Kerner's work; he
had simply forgotten the source of his ideas. A remarkable example of unintentional plagiarism came to the fore after George H. Daniels's 1971 book,
Science in American Society, received a warm review in the journal Science.
In a letter to Science, Daniels wrote that soon after the review appeared, he
became aware that parts of the book incorporated quotations from other
sources that he had only acknowledged in a general way. "To first cite as a
major source the author of a still current book:' explained Daniels, "who,
in many cases, would be a likely reviewer of my book, and then to deliberately steal from him, would require a degree of naivete much greater than
mine:' What had happened? Reconstructing his efforts, Daniels wrote, he
realized he had memorized and unconsciously reproduced the content of
several books; when he thought he was describing them generally he was in
fact quoting from them. "I have certainly been aware that I had an extraordinary ability to remember material when I wanted to:' Daniels reflected
ruefully, "but I have never before realized that I did it unconsciously:'

We are all potentially susceptible to cryptomnesia, and in some instances, we can catch ourselves in the act. The psychologist Graham Reed
describes a time when he woke up in the middle of the night with a catchy
tune running through his mind. He excitedly developed the tune the next
morning and worked feverishly on it throughout the day. When he thought
about a title for his wonderful new creation, Reed realized it already had
one - The Blue Danube Waltz! People can even unintentionally "plagiarize" their own ideas. The late psychologist B. F. Skinner relates that "one of
the most disheartening experiences of old age is discovering that a point
you have just made - so significant, so beautifully expressed - was made
by you in something you published a long time ago."

On the face of it, cryptomnesia is difficult to study under controlled
conditions: how does an experimenter induce people to unintentionally
plagiarize the ideas of others? In 1989, Alan Brown and Dana Murphy at
Southern Methodist University came up with a procedure for doing so.
They asked groups of four people to produce examples from a specified
category, one person after another. For example, if the experimenter said
"fruit;' the group members might say, in turn, "apple," "pear;' "orange;' and
"peach." On a later test, people were required to generate new examples
from the same categories which no one in the group had previously mentioned. Despite explicit instructions not to produce what others had said,
participants sometimes "plagiarized" answers, such as "apple" or "pear;'
even though others had previously given them.

Cryptomnesia in this experiment is probably attributable to an unconscious influence of memory known as "priming." When people hear
other group members generate such words as apple or pear, the words become activated, or primed, in memory. The priming persists over time, so
that when participants attempt to generate new category members later,
the activated words spring to mind easily. Failing to recollect that they previously heard a primed word, people believe that they are producing it for
the first time themselves.

Recent research indicates that cryptomnesia can be reduced by requiring people to pay careful attention to the source of their ideas. The psychologist Richard Marsh at the University of Georgia asked groups of college
students to generate novel solutions to each of two problems: 1) What are
some ways in which the university might be improved? and 2) How can the
number of traffic accidents in the United States be reduced? Similar to the
research above, some participants returned a week later and were asked to
generate novel solutions that no one in the group had mentioned a week
earlier. Sometimes these students produced ideas that others had suggested
the previous week. But with a second group of students who returned a
week later and tried to generate novel solutions, the researchers explicitly
encouraged the participants to consider carefully whether the idea was new
or related in some way to ideas that other group members had proposed
the previous week. These students plagiarized less frequently than those in
the other group. People sometimes do not spontaneously scrutinize the origin of their ideas, leaving themselves open to the undue influence of priming. Instructions to consider possible origins of an idea can, at least to some
extent, override the influence of priming and allow people to take advantage of the information that they possess about the source of an idea.

Misattributions in cryptomnesia are produced by some of the same
factors that create false recognition: a failure to consider or use specific recollections about the source of recalled information. Such a combination
can wreak havoc in everyday life: witness the misguided search for John
Doe 2, con artists who exploit the elderly, and the bizarre manifestations of
the Fregoli delusion.

Larry Jacoby has noted the similarity between attributions in remembering and those invoked in social situations. In famous experiments conducted by the social psychologist Stanley Schachter, people were injected
with adrenaline in a pleasant or a frustrating situation. Subjects in the former group felt happy; those in the latter group became angry. The adrenaline created an ambiguous sense of arousal that people attributed to positive or negative features of the situation. The adrenaline-induced arousal
resembles the fluent or speedy mental activity that people sometimes attribute - rightly or wrongly - to a feeling of familiarity rooted in a past
experience. Perhaps this sensation is what the French psychiatrist Arnaud
had in mind when he tried to explain the peculiar illusion of deja vu which
so often gripped his patient. It was as if Louis received jolts of adrenaline
that he struggled to interpret, ultimately attributing them to past experiences that had never occurred.

The odd experiences of a patient like Louis, and the annoying misattributions that happen frequently in everyday life, teach an important lesson about the nature of memory. We often need to sort out ambiguous signals, such as feelings of familiarity or fleeting images, that may originate in
specific past experiences, or arise from subtle influences in the present.
Relying on judgment and reasoning to come up with plausible attributions, we sometimes go astray. When misattribution combines with another of memory's sins - suggestibility - people can develop detailed
and strongly held recollections of complex events that never occurred.
During the past decade, such recollections have been linked with deeply
troubling events in the therapist's office, the courtroom, and the preschool.
The resulting shock waves have fractured families and shattered lives.