1. The Sin of Transience

The Sin of Transience.

On OCTOBER 3, 1995, the most sensational criminal trial of our time
reached a stunning conclusion: a jury acquitted o. J. Simpson of murder.
Word of the not-guilty verdict spread quickly, nearly everyone reacted with
either outrage or jubilation, and many people could talk about little else for
days and weeks afterward. The Simpson verdict seemed like just the sort of
momentous event that most of us would always remember vividly: how we
reacted to it, and where we were when we heard the news.

Can you recall how you found out that Simpson had been acquitted?
Chances are that you don't remember, or that what you remember is
wrong. Several days after the verdict, a group of California undergraduates
provided researchers with detailed accounts of how they learned about the
jury's decision. When the researchers probed students' memories again fif-
teen months later, only half recalled accurately how they found out about
the decision. When asked again nearly three years after the verdict, less than
30 percent of students' recollections were accurate; nearly half were dotted
with major errors.

The culprit in this incident is the sin of transience: forgetting that occurs with the passage of time. We are all familiar - sometimes painfully so
- with the everyday consequences of transience. Imagine, for example,
that you are attending an annual meeting of a professional or social group.
A smiling face looms at the other end of the hallway, approaching with an
extended hand, calling out your name, and saying how wonderful it is to
see you again. You smile politely and try to buy some time, but inside you
feel a mounting sense of panic: Who is this person? Why don't I remember
having met him before? He senses your discomfort and reminds you of the
pleasant cup of coffee you enjoyed together at the same meeting last year,
where you discussed, among other things, mutual frustrations with the bad
weather that disrupted your travel plans. If you had seen this person an
hour or a day after you met, you surely would have recognized him.

But a year later, you feel like the befuddled novelist in "Yumiura;' who could not
remember the woman who claimed he proposed marriage, as you struggle
and still can't recall the incident. Muttering weakly something to the effect
that "I sort of remember ... ;' you actually feel as though you are meeting
this person for the first time.

Transience can sometimes leave us feeling rather embarrassed. A female acquaintance of mine attended the wedding of a friend, whose husband she had not met before. Several months later, at a fiftieth birthday
party for her friend, she spotted an unfamiliar man in the corner. She discreetly asked her friend about the stranger - who was the woman's new
husband. My acquaintance says she still cringes when thinking about that
moment.

Perhaps the most pervasive of memory's sins, transience operates silently but continually: the past inexorably recedes with the occurrence of
new experiences. Psychologists and neuroscientists have uncovered reasons
for transience and are developing ways to counter it. The path to the mod-
ern era was set when a young German philosopher, traveling through Europe in the late 1870S, found inspiration that changed his future, and that of
psychology, while browsing in a secondhand Parisian bookstore.

WHEN MEMORY FADES.

The philosopher's name was Hermann Ebbinghaus, and the book that he
encountered, authored by the great German philosopher-scientist Gustav
Fechner, contained experimental methods for studying sensory perception.
When Ebbinghaus began his first academic post in Berlin in 1878, he pursued the flash of insight that had come to him in the Parisian bookstore:
memory, like sensory perception, could be studied using the methods of
science. It would take him seven years to publish his findings, but Ebbinghaus's 1885 monograph shaped the field for decades to come. Probing his
own memory for thousands of meaningless letter strings (psychologists
call them "nonsense syllables") that he had dutifully tried to learn and re-
learn, Ebbinghaus produced the first experimental evidence of transience.
He tested himself at six different times after studying a list of nonsense syl-
lables, ranging from one hour to one month. Ebbinghaus noted a rapid
drop-off in retention during the first few tests; nine hours after he studied a
list of nonsense syllables, he had forgotten approximately 60 percent of the
list. The rate of forgetting then slowed down considerably. After a month's


delay, Ebbinghaus had forgotten just over 75 percent of what he had learned
initially - not that much worse than the amount of forgetting at the nine hour delay.

Ebbinghaus conducted his experiments in the sterile confines of the
laboratory, far removed from the complexities of everyday life; he studied
meaningless strings of letters, not rich and varied personal experiences,
and tested only himself. Despite the evident limitations, these century-old
findings concerning how one man learned and forgot nonsense syllables
have something to say about whether we will recall last week's breakfast
meeting six months from now or remember what we read in yesterday's
newspaper for more than a few hours or days. His conclusion that most
forgetting occurs during early delays, and then slows down at later ones,
has been replicated in countless laboratory experiments. Modern memory
researchers have also extended Ebbinghaus's curve of forgetting outside the
confines of the laboratory, demonstrating that it defines a core feature of
transience.

In the early 1990S, the psychologist Charles Thompson and his col-
leagues at Kansas State University probed the memories of college students
who kept diaries over the course of a semester in which they recorded one
unique event each day. Forgetting was not quite so rapid as in Ebbinghaus's
study, but the shape of the forgetting curve for these everyday events was
generally similar to what others had observed in the laboratory. Thomp·
son's students recorded and tried to remember experiences that varied in
significance. A minority were personally meaningful ("My boyfriend Jake
and I broke up"), but most were rather humdrum ("Watched movies at
Jim's house on the VCR from 8:00 P.M. to 3:45 A.M."; "Mark and I started
to make caramel corn but we soon found out that we were out of baking
soda"). Other evidence concerning an annual happening that most people
value greatly - Thanksgiving dinner - shows clearly that even personally
significant events are not immune from the kind of transience that charac-
terizes the Ebbinghaus forgetting curve.

How well can you recollect the most recent Thanksgiving dinner that
you attended? A study of more than 500 college students suggests that what
you remember very much depends on exactly when you are reading these
words. At regular intervals for six months after Thanksgiving, students
were asked about the overall vividness of their memories of the dinner,
and also about specific details_ Vividness declined rapidly over the first
three months, followed by a more gradual decline for the remaining three
months. The basic form of the Ebbinghaus curve was again observed, but
this time for an event of considerable personal significance.

The drop-off was not, however, quite so steep as in Thompson's diary
studies. This difference may be because some aspects of our most recent
Thanksgiving dinner can be "remembered" on the basis of general knowl-
edge of previous ones. We know that we probably had turkey, even if we
have forgotten the particulars of this year's bird; we also know that we
probably gathered together with family. This type of general knowledge
about what usually happens at Thanksgiving does not fade across just a few
months. Consistent with this suggestion, students' memories for food and
for who attended the dinner dropped off at a relatively slow rate. But memories of details that were specific to the most recent Thanksgiving - such
as what clothes they and others wore, and the contents of conversations were lost much more quickly.

Similar processes operate when people recall a day at work. Try to an-
swer in detail the following three questions: What do you do during a typical day at work? What did you do yesterday? And what did you do on that
day one week earlier? When twelve employees in the engineering division
of a large office-product manufacturer answered these questions, there was
a dramatic difference in what they recalled from yesterday and a week ear-
lier. The employees recalled fewer activities from a week ago than yesterday,
and the ones they did recall from a week earlier tended to be part of a "typical" day. Atypical activities - departures from the daily script - were re-
membered much more frequently after a day than after a week. Memory after a day was close to a verbatim record of specific events; memory after a
week was closer to a generic description of what usually happens. Likewise,
Thompson's diary studies showed that specific details, such as the location
of an event, the people who were there, and the specific date, fade more
rapidly than the general sense of what happened. These observations are
backed up by other laboratory studies indicating that recollections of when
and where an event occurred, or who said what, tend to be especially transient.
At relatively early time points on the forgetting curve - minutes,
hours, and days, sometimes more - memory preserves a relatively de-
tailed record, allowing us to reproduce the past with reasonable if not perfect accuracy. But with the passing of time, the particulars fade and opportunities
multiply for interference - generated by later, similar experiences
to blur our recollections. We thus rely ever more on our memories for
the gist of what happened, or what usually happens, and attempt to reconstruct the details by inference and even sheer guesswork. Transience involves a gradual switch from reproductive and specific recollections to reconstructive and more general descriptions.

When attempting to reconstruct past events based on general knowledge of what usually happens, we become especially vulnerable to the sin of
bias: when present knowledge and beliefs seep into our memories of past
events (see Chapter 6). The combination of transience and bias can get us
into trouble. A management consultant told me about a meeting at which a
partner in a large company made a presentation to an important client in
the presence of his company's CEO and several overseas investors. The
partner related a story relevant to the client's situation about how a particular fast food chain adopted a strategy of raising prices. The story was
based on an incident the partner remembered from a year or two earlier.
But rather than calling on a detailed reproductive memory, the partner had
unknowingly reconstructed the specifics from his present knowledge
the chain had not actually raised prices. Worse yet, a manager who had previously worked at the fast food chain fidgeted uncomfortably. "She started
making faces while he was speaking:' recalls the consultant. "As the partner
was finishing his story, the manager spoke to the associate next to her in
what she thought was a whisper. In a voice that regrettably carried halfway
across the room, she said, 'He doesn't know what he's talking about. They
never raised prices.'" The embarrassed partner had lost specific memory
but was unaware of it.

Transience played the role of troublemaker in another, rather more
public incident, where questions concerning the nature of forgetting assumed national prominence: the 1998 grand jury investigation of William
Jefferson Clinton.


The afternoon of August 17, 1998, was a watershed in the investigation and
eventual impeachment of President Clinton. Testifying before a grand jury
convened by the independent counsel Kenneth Starr, Clinton answered
questions concerning the details of his relationship with Monica Lewinsky,
and about his related testimony in January 1998 in the Paula Jones lawsuit.
Clinton's August 17 remarks will no doubt be remembered by many - and
in the history books - for his verbal jousts with prosecutors regarding the
exact definition of the term "sexual relations:'

But from the perspective of a memory researcher, Clinton's terminological hairsplitting is not nearly so interesting as a second battle he fought
that afternoon: a battle over the characteristics and limits of transience.
Clinton's memory lapses in his grand jury testimony and earlier deposition
in the Jones case were widely viewed as self-serving conveniences designed
to avoid embarrassing admissions. Prosecutors' attempts to establish this
point rested on their intuitions about what is - and is not - reasonable
to forget about an experience at different times after it has occurred.

This debate over transience is vividly illustrated by an exchange be-
tween Clinton and the government counsel Sol Wi sen berg concerning a
meeting between the president and Vernon Jordan on the evening of December 19, 1997. Earlier that day, Jordan had met with an extremely upset
Monica Lewinsky, who had just learned that she had been issued a sub-
poena by the independent counsel's office. Jordan had later told Clinton
about this development. On August 17, nearly eight months later, Wisenberg focused on what Clinton had said back in January 1998 about this
meeting with Jordan. When asked whether anybody other than his attorneys ever told him that Lewinsky had been served with a subpoena by the
independent counsel's office, Clinton told the Jones attorneys, "I don't
think so." But this claim seemed implausible to Wisenberg: "Mr. President,
3.5 weeks before, Mr. Jordan had made a special trip to the White House to
tell you Ms. Lewinsky had been subpoenaed; she was distraught; she had a
fixation over you. And you couldn't remember that 3.5 weeks later?"

Clinton says that his memory is not what it used to be, and offers up
possible explanations of his recent forgetfulness:
... If I could say one thing about my memory - I have been blessed
and advantaged in my life with a good memory. I have been shocked
and so have members of my family and friends of mine at how many
things I have forgotten in the last six years - I think because of the
pressure and the pace and the volume of events in a president's life,
compounded by the pressure of your four-year inquiry, and all the
other things that have happened. I'm amazed - there are lots of times
when I literally can't remember last week.


Wisenberg immediately picks up on Clinton's self-confessed memory
problems. ''Are you saying, sir;' he queries, "that you forgot when you were
asked this question that Vernon Jordan had come on December 19, just 3
weeks before, and said that he's met that day, the day that Monica got the
subpoena?" While not explicitly agreeing, Clinton acknowledges that he
might have forgotten certain aspects of Vernon Jordan's visit. "It's quite
possible that I had gotten mixed up," he proffers. Clinton then asserts
somewhat more emphatically, "All I can tell you is I didn't remember all the
details of all this:'

Given the obsessive pursuit of Clinton by the independent counsel's
office, Wisenberg's questions might be viewed as indiscriminate badgering
by an aggressive attorney. But other parts of the deposition indicate that
Wisenberg did not cast doubt on Clinton's claims about forgetting when
they seemed more plausible. Compare the contentious exchange about forgetting across a three-week interval with an incident that occurs later in the
grand jury deposition. Clinton is asked about a meeting with his aide John
Podesta which occurred seven months earlier. On January 23, two days after
the Lewinsky affair became public, Clinton had allegedly told Podesta that
he had not engaged in any type of sex whatsoever with Lewinsky. When
asked about this exchange, Clinton acknowledges making careful denials to
a variety of people who might have included Podesta, but again appeals to
faulty memory for particulars:

CLINTON: I do not remember the specific meeting about which you
asked or the specific comments to which you referred.
WISENBERG: You don't remember ...
CLINTON: Seven months ago, I'd have no way to remember, no.
In contrast to his pointed probing of Clinton's apparent forgetting of a
three-week-old meeting, Wisenberg allows this assertion to pass unchallenged. He is willing to concede poor memory for a relatively routine ex-
change seven months earlier, but is dubious about any claims of forgetting
across a mere three weeks. The crux of the problem goes all the way back to
Ebbinghaus: How much forgetting is plausible at different times after an
experience has occurred?

Whatever Clinton's motivations when he testified, his self-professed
confusion about the details of what happened is exactly the type of forgetting expected based on both naturalistic and laboratory studies. Nonetheless,
Wisenberg's skepticism that Clinton could have forgotten the entire
meeting with Jordan after a mere three weeks is fully warranted. Clinton,
on the other hand, showed acute awareness of the difference between specific and general memories. Thus, when describing his first encounters
with Lewinsky in early 1996, he acknowledges that he probably met with
her approximately five times, but has specific memory for only two meetings. Clinton draws a sharp distinction between his specific recollections
and more general ones:

I remember specifically - I have a specific recollection of two times. I
don't remember when they were. But I remember twice when, on a
Sunday afternoon, she brought papers down to me, stayed and we were
alone.
And I am frankly quite sure - although I have no specific memory,
I am quite sure - that there were a couple of more times, probably
two more, three times more. That's what I would say. That's what I can
remember. But I do not remember when they were or at what time of
day they were or what the facts were. But I have a general memory that
I would say I certainly saw her more than twice during that period be-
tween January and April 1996 when she worked there.

Was Clinton twisting his testimony to avoid an embarrassing admission? Maybe so, but from the perspective of both naturalistic and labora-
tory memory research, one could hardly find a more apt illustration of how
memory fades over time.

THE BOOMERS' LAMENT.

Whatever the source of the memory complaints made by the fifty-something Clinton, he is certainly not alone among his contemporaries: aging
boomers are grumbling in record numbers about their increasing propen-
sity for forgetting. Laboratory studies show that some of these concerns
may be warranted. Numerous experiments have documented that older
adults (mainly in their sixties or seventies, sometimes fifties) have greater
difficulty than college students remembering information that an experimenter asked them to learn. Further, even when older adults can remember
lists of words or other experimental materials just as well as their younger
counterparts across a delay of a few minutes, their memories deteriorate
more rapidly across days or weeks. These memory deficits are particularly
evident when older adults are required to recollect the particulars of an experience, such as exactly when and where an event occurred. Older adults
lose specific details and tend to rely even more than younger adults on a
general sense of knowing that something happened.

How early does aging begin to affect transience? This question is important to millions of baby boomers entering their forties and fifties (and is
also relevant to the claims of Clinton, who was fifty-two years old in August
1998). Because most investigations of aging memory have compared college students with retirees, relatively less is known about people who occupy the in-between ages. In one recent study, people who had reached
their thirties, forties, fifties, sixties, or seventies took various memory tests
in 1978 and again in 1994. People who were fifty or older at the beginning of
the experiment (in 1978) performed more poorly when learning and recalling word lists and stories in 1994 than they had back in 1978. Those who
were in their thirties in 1978 performed more poorly in 1994 only on the
stories. Among people who were in their thirties in 1978 and those who
were in their fifties that year, the older group performed worse on both
word recall and story recall. So, problems with story recall begin, at the lat-
est, in the early to mid-forties, whereas problems with word recall are not
evident until people reach their fifties. The good news is that none of the
declines were large, with the older groups generally recalling about 10 to 15
percent less than the younger groups.

By the time people reach their sixties and seventies, transience is more
marked and consistent. But even in these older groups, poor recall is not
an inevitable consequence of aging: transience varies considerably among
older individuals. For example, in one study a significant minority of people in their seventies (roughly 20 percent) recalled about as many words
from a recently presented list as college students did.

Why do some older adults continue to show more marked susceptibility to transience than their younger counterparts, whereas others show lit-
tle evidence of decline? Several reports have raised the possibility that educational level plays a role. For example, in a recent Dutch study, elderly
adults aged sixty-five to sixty-nine, seventy to seventy-four, seventy-five to
seventy-nine, and eighty to eighty-five were given a list of words to learn,
then tried to recall them immediately and after a thirty-minute delay. Loss
of information across the delay was faster, and observed at an earlier age, in
less educated people than in more educated people. Whereas sixty-five- to


sixty-nine-year-olds in both groups retained about 65 percent of what they
learned across a delay, eighty- to eighty-five-year-olds with high education
retained about 60 percent of learning across the delay, but those with low
education retained less than 50 percent.

The researchers also noted that their results could reflect a higher
prevalence of Alzheimer's disease or other forms of dementia among those
with lower educational levels, possibly because they have less "mental re-
serve" to calion than more highly educated people. Scientists have long dis-
tinguished between normal declines in memory which accompany aging
(sometimes referred to as "benign senescent forgetfulness") and more pro-
nounced declines that accompany conditions involving actual brain pa-
thology, such as Alzheimer's disease. The brains of Alzheimer's patients are
disfigured by "senile plaques:' deposits of a protein known as "amyloid:'
and by twisted nerve cell fibers called "neurofibrillary tangles:' which inter-
fere with the normal operations of nerve cells. Experiments have shown
that compared with healthy older adults, Alzheimer's patients retain little
of their recent experiences.

An important series of studies by the neurologist Herman Buschke
and colleagues shows that levels of forgetting in a word memory test can
distinguish between healthy older individuals and those with Alzheimer's
disease. In the simplest version of the test, people see a sheet containing
four words belonging to different categories. When the examiner says the
appropriate category name (for example, vegetable), the individual points
to the appropriate word (for example, potato). This procedure ensures that
people pay attention to the words and understand them. After a few minutes, individuals try to remember the words on their own, and are then
given the category names again as prompts for any forgotten items. Failure
to come up with a studied word when given a category cue probably reflects loss of memory across the brief delay. Poor performance on this test
(defined by specific cut-off scores) is almost uniquely associated with the
presence of Alzheimer's disease or some other form of dementia. The test
works because Alzheimer's disease greatly magnifies transience above and
beyond any changes associated with normal aging.

Psychologists and neuroscientists who study memory agree that transience is pervasive and increases as people age. But they have spent dec-
ades struggling with a seemingly straightforward yet maddeningly difficult
question: Why does it happen?

Although there is no simple one-to-one relationship be-
tween an individual brain region and a specific sin of memory, some brain regions
are particularly relevant to specific memory sins. You can start to understand the
regions' locations by recognizing that each hemisphere in the brain is divided into
four major lobes: frontal, temporal, parietal, and occipital.


WITNESSING THE BIRTH OF A MEMORY.

The human brain is perhaps the most complex object in the entire universe, consisting of some one hundred billion nerve cells or neurons and an
even larger number of connections or synapses between them. Neuro-
scientists, who typically study memory in rats, rabbits, monkeys, birds, and
even sea slugs, can record electrical or chemical signals directly from individual neurons, or carefully remove small portions of the brain. This kind
of unfettered access to the brain has always elicited some jealousy on the
part of psychologists such as myself. We haven't had the techniques to
probe the inner workings of human brains with anywhere near the precision available to neuroscientists, and ethical grounds preclude the possibility of making experimental lesions in a person's brain. It is as if the scientific gods had decided to allow neuroscientists into an inner sanctum of the
brain, but restricted psychologists to a remote observation deck.

Straining to catch a glimpse into the inner sanctum, psychologists
have relied, for the most part, on experiments of nature: cases in which
people suffer memory loss as a result of damage to particular parts of the
brain. In the most famous case ever reported, a young man known by the
initials HM was operated on for relief of intractable epilepsy in 1953. The
neurosurgeon, William Beecher Scoville, removed the inner parts of the
temporal lobe on both sides of HM's brain (see Figures 1.1 and 1.2). After the operation, HM seemed normal in most respects; he could perceive
the world around him, carryon a normal conversation, and perform on
IQ tests as well as he had before the operation. But there was something
each of these lobes from the perspective of the surface of the brain's left hemisphere.
FIG U R E 1.2 allows us to peer through the surface and view a number of structures that occupy the inner parts of the brain.

The hippocampus and nearby structures in the inner parts of the temporal lobe
(Figure 1.2) are especially pertinent to the sin of transience. Parts of the frontal
lobe (Figure 1.1) also playa role in transience, are even more centrally involved in
the sins ofabsent-mindedness and misattribution, and may be related to the sin of
suggestibility. The area near the front of the temporal lobe (Figure 1.1, lower left)
appears to playa role in the sin of blocking. The amygdala (Figure 1.2) is closely
related to the sin of persistence. Not much is known about the brain regions involved in the sin of bias, but regions within the left hemisphere may playa significant role. I elaborate on the relationship between brain function and each of the memory sins in Chapters 1 through 7.


terribly wrong: HM seemed to forget his daily experiences as fast as they
occurred. He couldn't remember conversations from minutes earlier. He
failed to recognize doctors who worked with him every day. He forgot
that he had eaten lunch almost as soon as his plate was cleared from the
table. HM has been plagued by this extraordinary form of transience for
nearly fifty years: his memory has never shown even a hint of improvement.
HM revealed a stunning link between transience and the inner parts
of the temporal lobe. Because his amnesia is so profound, the structures
that were removed - including the horseshoe-shaped hippocampus and
part of a region behind it called the parahippocampal gyrus - have fascinated memory researchers ever since HM's case was first reported. These
structures are among the earliest affected and hardest hit by the senile
plaques and neurofibrillary tangles of Alzheimer's disease, which probably
explains why affected patients have such great difficulty remembering recent experiences.

Recently, the scientific gods have grown kinder to psychologists. The
past decade has seen the development of powerful new neuroimaging tools
that allow us to peer into the brain while it learns and remembers. The
technique that researchers are most excited about nowadays is called "functional magnetic resonance imaging;' or fMRI. This technology works by
detecting changes in the brain's blood supply. When a region of the brain
becomes more active, it requires more blood than it does in a less active
condition. But when blood flow rises, something odd happens: there is a
temporary oversupply of oxygenated hemoglobin relative to deoxygenated
hemoglobin, which magnifies the fMRI signal. Using this technique, researchers can determine which parts of the brain "light up" during cogni-
tive activities.

Using fMRI allows us to localize these changes in blood flow quite
precisely, within a few millimeters. Just as the telescope allowed astronomers to see the heavens, and the microscope allowed biologists to peer into
cells of living organisms, fMRI (and a related neuroimaging technique
known as "positron emission tomography," or PET scanning) has opened
up the human brain for psychologists and neuroscientists.

When memory researchers first began to use fMRI and PET scans,
there was great excitement about finally witnessing firsthand what goes on
in the parts of the temporal lobe that were removed from HM - regions
that are clearly central to understanding transience. But promising early reports were followed by a string of failures.

In late 1997, my research team seized on a new way to examine the issue with fMRI. Consider the following questions: If I measure activity in
your brain while you are learning a list of words, can I tell from this activity
which words you will later remember having studied, and which words you
will later forget? Do measurements of brain activity at the moment when a
perception is being transformed into a memory allow scientists to predict
future remembering and forgetting of that particular event? If so, exactly
which regions allow us to do the predicting? Because of technical limita-
tions, early fMRI (and PET) studies could not address this question. But by
1997, fMRI had advanced to the point at which it was possible, at least in
principle, to pose the questions and obtain answers.

In a collaborative effort led by two young stars of fMRI research, An-
thony Wagner and Randy Buckner, our group at the imaging center of
Massachusetts General Hospital came up with an experiment that was un-
doubtedly taxing for the participants. The MRI scanner is not a luxury
suite: a technician gently pushes you, while you lie flat on your back,
headfirst into a narrow tube. You then lie extremely still for an hour or two
(motion disrupts recording of the fMRI signal) while carrying out a task
the experimenter has devised. All the while, the scanner emits loud beeps as
a strong magnetic field is used to detect brain activity.

While holding still in this cacophonous tunnel, participants in our ex-
periment saw several hundred words, one every few seconds, flashed at
them from a computer by specially arranged mirrors. To check that they
paid attention to every word, we asked our volunteers to indicate whether
each word referred to something abstract, such as "thought," or concrete,
such as "garden." Twenty minutes after the scan, we showed subjects the
words they had seen in the scanner, intermixed with an equal number of
words they hadn't seen, and asked them to indicate which ones they did
and did not remember. We knew, based on preliminary work, that people
would remember some words and forget others. Could we tell from the
strength of the fMRI signal which words participants would later remember and which ones they would later forget?

We could. Two regions of the brain showed greater activity when people made abstract/concrete judgments about words they later remembered
compared to those they later forgot. Critically, one area was in the inner
part of the temporal lobe: the parahippocampal gyrus in the left cerebral
hemisphere - one of the regions that HM's surgeon had removed.

The other region whose activity predicted subsequent memory was
located farther forward, in the lower left part of the vast territory known as
the frontal lobes. This finding was not entirely unexpected, because previous neuroimaging studies indicated that the lower left part of the frontal
lobe works especially hard when people elaborate on incoming information by associating it with what they already know. Cognitive psychologists
had known for years that transience is influenced by what happens as peo-
ple register or encode incoming information: more elaboration during encoding generally produces less transient memories. For instance, suppose I
show you a list of words to remember, including lion, CAR, table, and
TREE. For half of the words, I ask you to judge whether they refer to living
or nonliving things; for the other half, I ask you to judge whether they are
in uppercase or lowercase letters. All other factors being equal, you will
later remember many more of the words for which you had made living!
nonliving judgments compared to words for which you had made upper case/lowercase judgments. Thinking about whether the word refers to a living or nonliving thing allows you to elaborate on the word in terms of what
you already know about it; making the uppercase/lowercase judgment does
little to link the word with what you already know. Other experiments have
shown that subsequent memory improves when people generate sentences
or stories that tie together to-be-learned information with familiar facts
and associations.

We thought that something similar might be going on in our fMRI experiment. When the left frontal lobe was strongly activated, people may
have been more successful in elaborating on a study list word in terms of
what they already know about it - dredging up associations or images -
than when the left frontal lobe was more weakly activated. The left para-
hippocampal region, we hypothesized, would then help to "save" this elaboration in memory. Working together, these two parts of the brain helped
to transform perception of a word into an enduring memory of its presentation.

At about the same time that we carried out our fMRI study, a group at
Stanford University completed a related project. During scanning, people
studied pictures of everyday scenes (instead of words) and then tried to re-
member the pictures a few minutes later. Their results were virtually identical to ours, except that the right cerebral hemisphere was prominently involved.
Levels of activity in the lower part of the right frontal lobe, and in
both the right and left parahippocampal gyrus, predicted subsequent re-
membering and forgetting of pictures that volunteers had studied in the
scanner. These findings made good sense, because earlier studies suggested
that the right hemisphere is primarily responsible for coding pictures,
whereas the left hemisphere is responsible for processing words.

The results from these two studies were exciting in part because there
is something fascinating, almost like science fiction, about peering into a
person's brain in the present and telling what she will likely remember and
forget in the future. And beyond an exercise in scientific fortune-telling,
these studies managed to trace some of the roots of transience to the split-
second encoding operations that take place during the birth of a memory.
What happens in frontal and parahippocampal regions during those critical moments determines, at least in part, whether an experience will be re-
membered for a lifetime, or follow the curve described by Ebbinghaus en
route to the oblivion of the forgotten.

THE FIRST SECONDS AFTER PERCEPTION.

In the late 1950S, two articles appeared in psychological journals which
astounded the few scientists who then specialized in memory research.
Trained in the tradition of Ebbinghaus, they were accustomed to observing
the trajectory of forgetting across hours, days, and weeks. The new studies
showed that when people were given the seemingly simple task of remembering three nonsense syllables, they forgot them almost completely in less
than twenty seconds. Nothing like it had ever been reported.

The key to understanding the apparent anomaly lies in a crucial transition that takes place in the moments when a memory is born: from temporary or short-term memory to more permanent long-term memory. Re-
taining information across days, weeks, and years depends on two major
forms of long-term memory. Episodic memory supports remembering of
personal experiences that occurred in a particular time and place: recollections of the surprise birthday party you attended last week, or of the Broad-
way play you saw on your first visit to New York as a child. Semantic memory allows the acquisition and retrieval of general knowledge and facts:
knowing that John Adams and Thomas Jefferson were principal architects
of the Declaration of Independence, or that Yankee Stadium is the House
That Ruth Built.

But a third type of memory intervenes between the moment of
perception and the eventual establishment of long-lasting episodic or semantic memories. Referred to as "working memory:' it holds on to small
amounts of information for short periods of time - usually a few seconds
- while people engage in such ongoing cognitive activities as reading, listening, problem solving, reasoning, or thinking. You need working mem-
ory to understand each and everyone of the sentences I have written thus
far. If you did not have a way to hold on to the beginning of the sentence as
the rest of it unfolds, you would not know what I meant by the time you
reached the sentence's end. Consider, for example, the following two sen-
tences:

The long and demanding course was so difficult that he never shot
below 90.
The long and demanding course was so difficult that he never passed
an exam.
You cannot tell whether the course refers to golf or school unless you hold
on to this word until the end of the sentence. Working memory allows you
to do so, but the system must constantly discard what is no longer needed
at the moment, and devote its resources to the temporary storage of in-
coming information. Unless a special effort is made - such as repeating a
sentence over and over again - information is lost from the system almost
immediately after it enters.

The stunning demonstrations of rapid forgetting in the late 1950S exploited this property of working memory. Immediately after presentation
of a nonsense syllable for study, people were required to count backward
from one hundred by threes. Unable to rehearse the nonsense syllables and
thus keep them in mind, the participants were victimized by rapid loss of
information from working memory.

We've all experienced this kind of transience. After calling directory
assistance to obtain a phone number, you are faced with the choice of pay-
ing an extra few cents for automatic dialing or doing it yourself. If you take
the time to decide which option to pursue, the number will vanish because
you aren't mentally repeating it. Perhaps the phone company understands
the consequences of rapid transience: having forgotten the number while
considering the options, people may be more likely to pay the extra cost for
automatic dialing rather than !ooking up the number again. And we've all
been frustrated by rapid transience in the course of casual conversations.
While listening to a friend, you are reminded of something important to
tell him. But when he unexpectedly changes the topic and starts spilling the
latest gossip about a mutual acquaintance, you suddenly realize that you've
forgotten that critical tidbit you wanted to pass on. It can take considerable
effort to recapture your train of thought and regenerate what you wanted
to tell him.

The main culprit in rapid transience is a part of the working memory
system called the «phonological loop." First postulated by the British psychologist Alan Baddeley, the phonological loop allows us to temporarily
hold on to a small amount of linguistic information. Baddeley conceived
the loop as a «slave" subsystem that assists the «central executive" system of
working memory. This system orchestrates the flow of information into
and out of long-term memory, but because of the continual bombardment
of inputs, the executive frequently needs assistance. The phonological loop
helps out by providing extra temporary storage of words, digits, and other
bits of speech.

This slave subsystem's existence was initially revealed by studies of
brain-damaged patients whose memory problems are virtually a mirror reversal of those seen in the amnesic patient HM. Even though his long-term
memory for daily experiences is nonexistent, HM has no difficulty when
presented with a string of digits and asked to repeat them immediately. He
can easily reproduce sequences of six or seven digits - the same number
that can be recalled by healthy people. In the early 1970s, the neuropsychologists Tim Shall ice and Elizabeth Warrington described an intriguing
patient, known by the initials KF, who had no difficulty remembering daily
experiences from long-term memory, but was unable to remember immediately more than a single digit! KF (and other patients like him) had suf-
fered a stroke that destroyed the back part of his parietal lobe on the surface of the left cerebral hemisphere but did not affect the inner parts of the
temporal lobes that were removed from HM's brain.

The mirror-image strengths and weaknesses in HM and KF showed
that the phonological loop can function independently oflong-term memory. But the results also raised questions about the function of the phonological loop. If people with a dysfunctional loop have no difficulty estab-
lishing new long-term memories, then why do they need it in the first
place? Surely this system did not evolve solely to help us remember telephone numbers for a few seconds. By the 1980s, the function of the phonological
loop seemed so obscure that one cynic derided it as "a pimple on the
face of cognition:'

We now know that the kind of rapid transience associated with a
damaged phonological loop has significant, even grave, consequences. The
early dues came from studies of another brain-injured patient with a dam-
aged phonological loop. The patient could learn word pairs in her native
language, Italian, as quickly as healthy controls. But in contrast to healthy
native Italian speakers, the patient could not learn Italian words paired
with unfamiliar Russian words. Subsequent studies showed similar results:
patients with damage to the phonological loop were almost totally unable
to learn foreign vocabulary.

The phonological loop turns out to be a gateway to acquiring new vocabulary. The loop helps us put together the sounds of novel words. When
it is not functioning properly, we cannot hold on to those sounds long
enough to have a chance of converting our perceptions into enduring long-
term memories. Rapid transience of this kind has consequences that extend
beyond adults with brain damage. Studies of young children show that the
ability to repeat nonsense words provides a sensitive measure of the phonologicalloop's functioning. Children who perform at a high level on this
test have an easier time acquiring new vocabulary than do children who
perform at a low level; the number of nonsense words a child can repeat
immediately is an excellent predictor of vocabulary acquisition. Baddeley
and the psychologist Susan Gathercole have found that children with language deficits perform especially poorly on tests of the phonological loop.
In contrast, other studies have revealed that gifted language learners-
polyglots who have mastered several languages - do especially well on the
same kinds of tests. Far from being a mere "pimple on the face of cognition," the phonological loop is a key player in one of the most fundamental
human abilities: learning a new language.

Neuroimaging studies using fMRI and PET scans have begun to illuminate some of the neural subsystems that are relevant to short-term transience. For instance, neuroimaging studies have isolated the storage compartment of the phonological loop to the back part of the parietal lobe an important finding because, as we have already seen, this part of the system is damaged in brain-injured patients who are plagued by short-term
transience. Another part of the phonological loop, critical for actively repeating information held in short-term storage, depends on lower portions
of the left prefrontal cortex - in the general vicinity of the region discussed earlier which contributes to elaborative encoding. This same region
plays an important role in language output. When a healthy person suffers
the kinds of short-term transience considered so far - such as forgetting
what he or she is about to say, or forgetting a phone number looked up
seconds ago - it's probably because the person fails to activate this part of
the left frontal cortex. The information is then lost from working memory
and unavailable for further elaborative encoding into long-term memory.
Healthy people can circumvent short-term transience by making a concerted
effort to rehearse information, which stimulates the lower left frontal cortex. But brain-damaged individuals, like the patient KF, are perpetually doomed to endless bouts of rapid transience because they lack the
necessary brain structures.

AFTER THE FIRST FEW SECONDS.

Working memory and encoding processes are keys to understanding transience, but they are not the entire story. Whether an experience is quickly
forgotten or remembered for years also depends on what happens after
those first few seconds when a memory is born. Human beings are story-
tellers, and we tend to tell stories about ourselves. Thinking and talking
about experiences not only helps to make sense of the past, but also
changes the likelihood of subsequent remembering. Those episodes and
incidents we discuss and rehearse are protected, at least partially, from
transience; those that we don't ponder or mention tend to fade more
quickly. Of course, the experiences that cause us to ponder and discuss
them repeatedly might simply be more memorable in the first place. After the Loma Prieta earthquake struck the Bay Area in 1989, those who
experienced it firsthand were so eager to relate their memories of this
distinctive and disturbing event that others quickly became saturated
by endless tales of "where I was when the earthquake hit:' Soon a popular
T-shirt appeared admonishing people to refrain from sharing their earthquake stories.

In the diary study conducted by Charles Thompson and associates,
experiences that students reported talking and thinking about most often
were remembered in the richest detail. Numerous laboratory studies have
demonstrated clearly that, even when possible differences in initial memorability are controlled, thinking or talking about a past event enhances


memory for that event compared to experiences that are not rehearsed.
These findings have direct implications for countering transience in everyday life: thinking and talking about everyday experiences is one of the best
ways to retain them later.

Transience can also be exacerbated by what happens after an experience is initially encoded. Consider the study discussed earlier that probed
what people remember from a typical day at work. The next day, their
memories were rich and detailed; a week later, they were little more than
generic descriptions of what usually happens. Imagine, however, that after
leaving work on Monday, some people left for a week's vacation instead of
working for the remainder of the week. Chances are excellent that upon returning, they would possess a richer and more detailed recollection of what
happened at work on the previous Monday than those who worked for the
entire week. Experiences that are similar to those we wish to remember create interference that impairs memory. People who don't leave for vacation
carry out many activities on Tuesday, Wednesday, Thursday, and Friday
which are highly similar to those they carried out on Monday, and thus create substantial interference. People who take off on vacation engage in entirely different activities that create little or no interference.

But long-term transience is probably not entirely attributable to interference from similar experiences: loss of information over time occurs even
when there is little opportunity for interference to playa role. For example,
the psychologist Harry Bahrick tested retention of Spanish vocabulary in
those who had studied Spanish during high school or college. He con-
ducted tests at various time points after people stopped taking Spanish,
ranging from immediately to fifty years later. Bahrick reported a rapid
drop-off in memory for Spanish vocabulary during the first three years after classes had stopped, followed by tiny losses in later years. The drop-off
during the initial years is probably attributable to spontaneous decay or
loss of information.

What happens to those experiences that we can remember after a day
but not after a year? Do they disappear entirely? Or are they lurking in the
background, requiring only the right trigger - a distinctive voice or a
pungent smell- to call them to mind? Memory researchers have debated
these questions for decades. The answer - or at least my answer - in-
volves a partial "yes" to both scenarios. Neurobiological studies of non-
human animals provide mounting evidence that forgetting sometimes


involves literal loss of information. Memories, according to most neurobiologists, are encoded by modifications in the strengths of connections
among neurons. When we experience an event or acquire a new fact, complex chemical changes occur at the junctions - synapses - that connect
neurons with one another. Experiments indicate that with the passage of
time, these modifications can dissipate. The neural connections that encode memories thus may weaken as time passes, perhaps mirroring the
shape of the decay curve first reported by Ebbinghaus. Unless strengthened
by subsequent retrieval and recounting, the connections become so weak
that recall is eventually precluded.

At the same time, however, countless studies have also shown that
seemingly lost information can be recovered by cues or hints that remind
us of how we initially encoded an experience. As time passes and interference mounts, information may be gradually lost to the point that only a
powerful reminder can breach the seemingly inexorable effects of transience by dredging up the remaining fragments of an experience from everweakening neural connections.

This latter point is nicely illustrated by the psychologist Willem
Wagenaar's diary study of his personal memories. Every day for four years,
Wagenaar recorded various aspects of a distinctive event: who was involved, what happened, when and where the event occurred, and a further
distinguishing detail of the event. He did not review the memory diary at
any point during the four years when he was recording entries. Wagenaar
commenced testing himself the day after the recording phase concluded,
probing his memory with different combinations of cues (for example,
who, what, where, when).

Wagenaar found that the more cues he provided, the more likely he
was to remember key details of the event. There were, however, many
events in which no amount of cueing elicited any form of recollection. Intrigued by the question of whether these experiences had disappeared from
his memory altogether, Wagenaar interviewed people who were involved in
ten of the events he had scored as "completely forgotten." In all cases, they
were able to provide additional details that allowed Wagenaar to remember
the event.

Wagenaar's study reveals a common result of transience over months
and years: incomplete rather than total forgetting that leaves in its wake
scattered shards of experience. Vague impressions of familiarity, general


knowledge of what happened, or fragmentary details of experiences are the
most common legacies of transience.


We'd all like to remember more than what remains in the wake of transience. Any attempt to reduce transience should try to seize control of what
happens in the early moments of memory formation, when encoding pro-
cesses powerfully influence the fate of a new memory. All popularly available memory improvement packages recognize and build on this fundamental insight by trying to teach people how to elaborate on incoming
information; a number of available books and articles provide helpful reviews of specific techniques. The most frequently prescribed technique involves some form of visual imagery mnemonics: people are encouraged to
elaborate on information they wish to remember by converting it into vivid
and even bizarre visual imagery. So, for instance, if you want to remember
that my name is Daniel Schacter, you might imagine me surrounded by a
group of lions (Daniel in the lion's den), eyeing a shack into which I hope
to flee for protection.

Visual imagery mnemonics were first discovered by the Greeks more
than two thousand years ago, and are used by most professional mnemonists to perform the spectacular feats of their trade - memorizing a tele-
phone book, or the names of hundreds of people based on just a few sec-
onds of exposure. Controlled studies in the laboratory also clearly show
that ordinary people can use imagery mnemonics to boost memory for
lists of words, names, and other materials. There is a problem, however.
Many of the imagery techniques are complex, require considerable cognitive resources to implement, and are therefore difficult to use spontaneously. The first few times you generate bizarre mental pictures and stories
to encode new information, the process may be challenging and fun. But
the task of repeatedly generating memorable images can eventually become
burdensome enough so that people stop engaging in it. In one study, for example, older adults were able to use mnemonics when instructed to do so
in the lab, but barely one-third of them reported using the techniques in
their everyday lives.

Widely advertised memory improvement programs, such as Mega
Memory, rely heavily on the use of visual imagery and related techniques.
Promotional materials for Mega Memory hold out the enticing prospect


that training can result in a "photographic memory" that will enable you to
remember names and faces, recall lists or appointments without writing
anything down, and even impress your friends and family with demonstrations of mental gymnastics. Glowing testimonials suggest spectacular gains
by individuals who have tried the program.

There is little reason to doubt that these programs will be helpful to
those who make the effort to use the techniques on an ongoing basis. But I
suspect that some people do not realize that to be successful, they must use
the techniques each and every time they want to remember a particular
event or fact. When I took questions during a radio interview, one caller
asked me whether completing the Mega Memory course would "train my
brain" to "take pictures" that ensure subsequent memory. She expected
or at least hoped for a method that would boost her memory in much
the same way that glasses help you to see better: just put them on, and without any effort you notice an immediate improvement in your vision. Un-
fortunately, I explained, mnemonic techniques are not the memory equivalents of eyeglasses: improvements are possible, but they require effortful
use of the technique to encode each individual face, name, event, or fact.

Although there have been few controlled investigations of commercially available memory programs, one recent study examined the effects of
training using Mega Memory and similar Memory Power audiotapes with
groups of older adults. After completing a variety of memory tasks, participants attempted to complete one of the two training programs, or were
assigned to a waiting list. Most of the participants managed to complete
the audiotaped courses, reported generally high levels of satisfaction with
them, and had the subjective sense that their memories had improved as a
result of training. Disappointingly, there was no evidence of memory improvement in those adults who successfully completed either the Mega
Memory or Memory Power programs as compared with the other participants. The researchers concluded that the benefits of these programs for
older adults had been "grossly exaggerated."

To profit from mnemonics, or any technique purported to improve elaborative encoding, the method must be simple enough to use regularly.
One approach that meets this criterion has been documented in numerous laboratory studies: generating elaborations that relate information you
wish to remember to what you already know. A simple way to achieve this
goal is to ask questions about what you wish to remember which force you


to elaborate: What are the distinctive features in the face of the woman
whom 1 just met? What acquaintance does she remind me of, and what are
their similarities and differences?

Promising results have been reported using a variant of this approach
derived from encoding studies in an unlikely population: professional ac-
tors. During the early 1990S, the psychologists Helga and Tony Noice made
an intriguing discovery while studying how professional actors learn and
remember their lines. Rather than using a verbatim memorization strategy,
actors learned a script by asking questions about how the specific words
that a character uses provide insights into the nature of the character and
his goals. The precise grammar, punctuation, and other linguistic elements
served as clues to the character's plans, motivations, and intentions. For instance, when one actor encountered a brief response by his character-
"Yes, I did" - he noted when analyzing the script, "I don't say anything
more than 1 need to say. Short answers." Another actor, considering the line
"Er ... thanks. Thanks;' thought that it suggested "trying to be cool and
man-of-the-worldish, but I stutter a little."

More recently, Noice and Noice have examined whether college students and older adults can benefit from instruction in the kinds of "active
experiencing" used by actors. Results so far have been encouraging. Several
studies have shown that brief training in this strategy enhanced verbatim
recall of a script in psychology students and senior citizens compared to
participants who simply tried to memorize the script. As with imagery
mnemonics, active experiencing demands considerable effort, so it remains
to be seen whether people will use the technique on a regular basis. But the
promising early results remind us that a major principle for countering
transience - enhance elaborative encoding - along with some tools for
realizing it, have been established experimentally. The main stumbling
block involves effective implementation of encoding techniques in every-
day life.

Because elaborative encoding, imagery mnemonics, and related approaches all require cognitive effort, there is undeniable appeal to the
prospect of finding an easy and permanent antidote to transience - the
mnemonic equivalent of corrective lenses. To judge by the amount of advertising it has received, you might guess that the magical mnemonic lenses
have been found in the extracts of leaves from one of the oldest deciduous
trees, the Ginkgo biloba. We've all seen the ads claiming that ingestion of


gingko will produce heightened mental sharpness and enhanced memory
function. And, indeed, many studies have shown that gingko does have a
salutary effect on cerebral circulation. In the few well-controlled studies of
memory that compare ginkgo to placebo controls, modest improvements
have been reported in people who reported serious memory problems
prior to taking ginkgo, but little or no improvement has been observed in
people who reported slight or no memory problems beforehand. Other
studies have revealed that patients with Alzheimer's disease show small improvements in a variety of symptoms after taking ginkgo, probably because
of general improvements in alertness. But there is no evidence that ginkgo
exerts specific effects that reduce transience. Given a choice between taking
gingko or investing some time and effort in developing elaborative encoding strategies, healthy people would be well advised to focus on the latter
approach.

Various other herbs and vitamins have also been touted as memory
aids, but for the most part the evidence supporting their efficacy is slim or
nonexistent. Some suggestive positive results have been reported in several
studies of a nutritional supplement called phosphatidylserine or "PS:' Like
gingko, PS appears to exert generally beneficial effects on a wide variety of
tasks, including some memory tests. Some have gone so far as to tout it as a
"memory cure" for all manner of memory problems associated with ad-
vancing age. However, the apparent ubiquity of PS effects - modest gains
in attention, concentration, speed of responding, and so forth - suggests
that it may operate mainly to augment arousal and alertness, much like a
stiff cup of coffee. Indeed, the authors of a six-step memory-cure program
that advocates regular doses of PS also endorse the kinds of elaborative en-
coding techniques discussed earlier. It is a safe bet that this part of the program is responsible for some of the clinical successes they describe.

Other approaches have focused on hormones that seem related to
transience. For instance, researchers have been examining the possible benefits of estrogen hormone replacement in postmenopausal women. After
menopause, women often complain about memory problems, and laboratory evidence from studies of older women suggests that low levels of estrogen are associated with poor retention of verbal information, such as lists
of words or word pairs, across a time delay. Recent results indicate that estrogen replacement can improve delayed retention of verbal and pictorial
information.


Treatments that effectively combat transience are likely to operate directly on the physiological processes responsible for preserving memories.
A group of neurobiologists led by Joseph Tsien recently took a dramatic
step forward in this direction by identifying a gene that significantly improves retention in experimental mice. The gene makes a protein for a neu-
ral gateway which plays a key role in memory known as the NMDA (N-
methyl-D-aspartate) receptor. The NMDA receptor helps to orchestrate the
flow of information from one neuron to another across the gap known as
the synapse. Several decades ago, the Canadian psychologist Donald Hebb
proposed that memories form when the strength of synaptic connections
rises among neurons that are active at the same time
summed up by the slogan "'Neurons that fire together wire together:'

The NMDA receptor opens when it receives two different signals
at roughly the same time, triggering facilitated neural processing called
"long-term potentiation:' which is believed to help increase synaptic connections and thus promote memory formation. At a relatively young age,
the receptor stays open for longer than at an older age, boosting long-term
potentiation and making it easier for youthful organisms to form new connections. Tsien's group overexpressed the critical gene in experimental
mice, leading to more activity in NMDA receptors. Mice with extra copies
of the gene performed several different kinds of tasks -learning a spatial
layout, recognizing familiar objects, and recalling a fear-inducing shock.
The mutant mice showed enhanced long-term potentiation during learning, and also showed better performance on each of the three memory
tasks than did normal mice. The benefits persisted into adulthood, in effect
allowing older mice to learn like younger ones.

The Tsien group concluded their article with the tantalizing suggestion that the beneficial gene expression effect in mice "reveals a promis-
ing strategy for the creation of other genetically modified mammals with
enhanced intelligence and memory." As exciting as their results may be,
however, nobody yet knows when this kind of approach will lead to the development of treatments that counteract transience in patients with memory disorders or even in people with normal memories. The possibilities
are both enticing and troubling. Tim Tully, a neurobiologist who has carried out pioneering research on the genetic basis of memory, wonders
whether memory-enhancing drugs might eventually find applications that
he would deem personally distasteful. "Think about the pressure on a gen

eral who has thirty minutes to communicate a data-rich conversation of
specifics of bombing missions to a group of pilots before going off to drop
bombs;' suggests Tully. "Do you think he'd cram, then take a memory
enhancer? They'd be chomping at the bit for drugs that could modulate
memory in that fashion." Tully is a pacifist who views this type of application as a perversion of the intent of his and others' research. "I would hate
to see this understanding perfected for the art of war, for all the covert and
overt atrocities that humans push over on each other."

The potential educational implications of memory-enhancing drugs
are also both auspicious and troubling. "What would it be like if a child
popped a memory enhancement pill every day before school?" wonders
Tully. "What would that child's head be like after twelve years of education?
What would the child accomplish with that store of information?" The
possibility of producing a generation of superlearners, free from the limita-
tions of transience, seems highly desirable. But could the brain handle such
an onslaught of information? What happens to those children who don't
have access to the latest memory enhancer? Are they left behind in school
and in later life? "We don't know:' concedes Tully. Some of the same questions apply to adults in the workplace. Imagine that your likelihood of obtaining a promotion grows if you learn and retain more job-related information, and you can do so by taking a memory enhancer. Failing to
take the drug could put you at a competitive disadvantage. Would you take
it, even if the drug produced worrisome side effects, or if the side effects
were unknown? These are the kinds of questions that we will have to face
eventually, given the pace of progress in research on the neurobiology of
memory.

Public excitement over the prospects of genes and drugs that would
reduce or eliminate forgetting perhaps reflects an underlying fear of the
catastrophic consequences of Alzheimer's disease, or even of the milder effects of normal age-related memory decline. In her disturbing short story,
''Almost No Memory," Lydia Davis describes a woman whose memories for
all past experiences - even those that happened a day ago or an hour ago
- are characterized by the hazy quality of incomplete forgetting that normally develops over longer periods of time for most of us. Davis's semi-
amnesic protagonist records her thoughts and ideas in notebooks that she
consults to obtain a sense of her self and her past. But the result of this exercise is more perplexing than it is enlightening:


And so she knew by this that these notebooks truly had a great deal to
do with her, though it was hard for her to understand, and troubled her
to try to understand, just how they had to do with her, how much they
were of her and how much they were outside her and not of her, as they
sat there on the shelf, being what she knew but did not know, being
what she had read but did not remember reading, being what she had
thought but did not now think, or remember thinking, or if she
remembered, then she did not know whether she was thinking it now or
whether she had only once thought it, or understand why she had a
thought once and then years later the same thought, or a thought once
and then never that same thought again.
This swirling confusion cuts to the heart of why transience is perhaps
the most terrifying of the seven sins: it undermines memory's role in connecting us to past thoughts and deeds that define who we are. The great
British poet William Wordsworth recognized this connection. In his Ode:
Intimations of Immortality from Recollections of Early Childhood, Wordsworth meditated on the qualities of faded childhood memories, acknowledging with some regret, "The things which I have seen I can now see no
more." He celebrated the importance of the faint echoes that remained
from his ever-receding past:

But for those first affections,
Those shadowy recollections,
Which, be they what they may,
Are yet the fountain-light of all of our day,
Are yet a master-light of all of our seeing.