Sigma Xi, The Scientific Research Society

Sleep to Remember: The brain needs sleep before and after learning new things, regardless of
the type of memory. Naps can help, but caffeine isn't an effective substitute
Author(s): Matthew P. Walker
Source: American Scientist, Vol. 94, No. 4 (JULY-AUGUST 2006), pp. 326-333
Published by: Sigma Xi, The Scientific Research Society
Stable URL: http://www.jstor.org/stable/27858801 .
Accessed: 13/02/2015 12:50

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Sigma Xi, The Scientific Research Society is collaborating with JSTOR to digitize, preserve and extend access
to American Scientist.

http://www.jstor.org

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions
 to Remember
Sleep

The brainneeds sleep beforeand afterlearningnew things,regardlessof the type

ofmemory.Naps can help, but caffeineisn'tan effective
substitute

Matthew P.Walker

being a student? Faced
with a big test the next day, you
Remember
have to learn something in a hurry?the
General Prologue to Chaucer's Canter
bury Tales, maybe,
or two-and-a-half oc
taves of a difficult scale on the clarinet.
After long hours of practice you notice
it's themiddle of the night. You haven't
quite mastered the task. Should you
forgo sleep formore practice?
Depending on the stakes, most peo
ple would say yes. But based on how
our brains work, the answer is
prob
ably no. In fact,psychologists have sus
pected for some time that sleep relates
somehow to the development ofmem
ory, although the reasons were unclear.
Behavioral tests showed that adequate
sleep before and after a training session
was essential for learning,whether the
taskwas tennis or algebra.
Scientists frommany disciplines have
confirmed and elaborated those suspi
cions over the past decade. In view of
the collective illumination ofmuch con
gruent data, most neuroscientists now
believe that sleep is integral to learning
and memory. However, some subtle
ties lie beneath this blanket statement:
Neither memory nor sleep is simple in
terms of structure and function.What's
more, the intersections between types
ofmemory and phases of sleep are also
governed by sometimes cryptic vari
ables. Despite the inevitable discrepan
P. Walker earned his Ph.D. from the
Matthew
Medical Research Council in London in 1999. Fol
lowing stints as a postdoctoral fellow and instructor
in psychiatry and psychology atHarvard Medical
School, he became an assistant professor ofpsychol
ogy in 2004. He is currently director of the Sleep
and Neuroimaging Laboratory. Address: Sleep and
Neuroimaging Laboratory, Department ofPsychia
try,Beth Israel Deaconess Medical Center, Feldberg,
FD 862, Harvard Medical School, Boston, MA
02115. Internet: mwalker@hms.harvard.edu
cies that arise in complex fields of study,
thepreponderance of behavioral, neuro
anatomical, physiological, cellular and
molecular evidence supports the idea
that periods of the sleep cycle actively
orchestrate changes in certain categories
of memory.
Sleep, Memory
The electrical signature of a sleeping
brain is different from that of an awake
brain, but equally big differences exist
during thevarious phases of sleep. Rapid
eye movement (REM) sleep, associated
with themost vivid dreaming, produces
brain waves (measured with an electro
encephalograph) thatmost resemble the
ones found in awake subjects.Across the
night, REM sleep alternateswith its an
tithesis,non-REM (NREM) sleep, about
every 90minutes inhumans. In primates
(including human beings), NREM sleep
has four substages. Psychologists refer
to the deepest and most electrically dis
tinctiveof these substages, NREM 3 and
NREM 4, as slow-wave sleep (SWS) be
cause of the characteristic low-frequency,
high-amplitude brain waves.
Similar to sleep, memory exists in cord and brainstem.
several forms. The most popular clas
sification scheme is based on the dis
tinction between those memories that memories are rarely so distinct. For
you can declare verbally and those that example, learning to speak a language
you have to show throughperformance.
Psychologists call these categories declar Figure 1. Sleep is critical to cognitive functions,
ative and nondeclarative
memory, respec
tively.The former is fact-based and in
cludes answers to questions like "What
is the value of Planck's constant?" and
"Where did I put my keys?" The an
nals of neuropathology, brain imaging
and modern computer models agree
thatdeclarative memory requires a part
of the brain called the hippocampus,
which lieswithin themedial temporal
lobe. This little structure is a nexus for
filing and retrieving information from
theneocortex, and italso seems to bind
together different perceptual elements
of a single event ("Ah, yes. I put the
keys down tonab the last doughnut").
In contrast, nondeclarative memory
is the "know how" memory, rather
than the "know what" memory, and is
manifest as an action or behavior. The
category is divided further into proce
dural and implicitmemory. The former is
responsible formovements, habits and
skills (such as how to ride a bicycle);
the latterencompasses less familiarphe
nomena such as classical
conditioning,
habituation and priming (also known
as the power of
suggestion). These dif
ferent forms of nondeclarative memory
depend on somewhat differentbrain re
gions. Although a map of these regions
must be overly simplistic, neuroscien
tistsconsider the core structures forpro
cedural memory tobe the striatum,mo
tor cortex and cerebellum; conditioning
engages the cerebellum and, for emo
tional learning, the amygdala; priming
involves theneocortex; and habituation
isbased on reflexpathways in the spinal
Although these categories are con
veniently separate on paper, real-life
particularly memory. Even during highly cho
reographed missions aboard the space shuttle,
astronauts are instructed to sleep eight hours
per night?a difficult feat given the excitement
and weightlessness. Recent studies demon
strate that learning requires physical changes
in the brain, at least some of which occur dur
of the sleep cycle. Here,
ing specific phases
Mission Guion Bluford
Specialist (left) and
Commander Richard "Dick" Truly (right) doze
while floating on themiddeck of the Challeng
er in 1983.
(Photograph courtesy of NASA.)

326 American Scientist, Volume 94

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions
www.americanscientist.org 2006 July-August 327

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions
 stages of sleep brain wave pattern

nonREMI

MM

?
Figure 2. During sleep, a person cycles through periods of electrically distinct brain activity. Rapid eye movement (REM) and non-REM (NREM)
sleep alternate about every 90 minutes, the ratio of NREM to REM sleep shifts as the progresses (as shown here for a person falling
although night
asleep atmidnight). NREM stages 3 and 4,which are characterized by high-amplitude, low-frequency waves, predominate during the firsthalf of the
but stage 2 NREM and REM are more common later.
night, sleep

requires several memory modes, rang on the passage from the firstmental If that encoded memory is destined for
ing from nondeclarative memory for glimmer to a permanent record. The long-term storage, itwill go through
how tomove themouth and tongue, steps occur on a continuum, although successive
stages
to become more sta
tomemory of grammatical rules and the exact timing is variable by task, ble in a process known as consolida
structure (a mix of the conscious and strength of the initialmemory, circum tion. In classical psychology, a memory
unconscious), todeclarative memory for stance and individual. Thus, making is consolidated when, in the absence
vocabulary.
someone's
acquaintance forms an of mental rehearsal, it becomes stur
But regardless of type, all memory ephemeral representation of that per dy enough to resist disruption from
appears to go through similar stages son's name and facewithin the brain. competing new learning, perceptions,
thoughts or actions.
Recent findings show that consolida
declarative nondeclarative
tion goes beyond simply stabilizing or
memory memory
fixatingmemories?it enhances them
as well. The two processes seem to be
distinct: Although stabilization appears
to occur over time regardless of brain
state, enhancement occurs primarily, if
not exclusively, during sleep. This "of
fline" effect can restore previously lost
memories or produce additional learn
ing, both without the need for further
practice. In other words, the enhance
ment phase of memory consolidation
is an active process, not merely one of
simple maintenance; the brain contin
ues to learn even though ithas stopped
practicing.
I've chosen in this article to focus
primarily on the influence of sleep on
encoding and consolidation, but the
later stages of memory processing are
also important. In them, new patterns
Figure 3. Human memory can be classified several ways. Most schemes distinguish between de of information are integratedwith past
clarative and nondeclarative memory. The former is consciously accessible and fact-based (know
experiences and knowledge. At about
ing what), and includes general knowledge (semantic) and autobiographical memory (episodic). the same time,memories can be reor
In the brain, these categories require the diencephalon and medial temporal lobe, including the
Nondeclarative memory to our conscious mind and includes proce
is inaccessible ganized and moved to new anatomical
hippocampus.
sites in a process called translocation.
dural memory how), as well as implicit forms of learning,
for actions, habits and skills (knowing
on various other parts of the brain. For For declarative memories, thismeans
which depend example, learning to swing a golf club
that thememory trace is no longer ex
requires the striatum, motor cortex and cerebellum; nonassociative learning involves reflex path
ways in the spinal cord and elsewhere; priming engages the neocortex; and associative learning
clusive to the hippocampus, but be
requires the amygdala or cerebellum, depending on whether the cued behavior has an emotional comes more distributed through por
or a physical component. (Functional anatomy after Squire and Knowlton 1994.) tions of the cortex.

328 American Scientist, Volume 94

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions
Active Memory Enhancement ing out the difficult portions of the se provement for both motor memories,
Many investigators (including myself quence. In essence, the sleeping brain not just the second. These findings sug
and my colleagues) have explored the appears towork specifically on themost gest several conclusions: First, newly
particularly robust link between sleep problematic parts of a memory, selec learned motor memories are initially
and procedural learning, the category tively solving them for the next day. As unstable and vulnerable to interference
of memory that includes perceptual a consequence, these from competing motor memories. Sec
overnight changes
and motor skills. In 2002, we published make performance more automated and ond, thememory gradually becomes
a study detailing the effects of more stable and resistant to such compe
sleep correspondingly faster.
on a task similar We further teased theprocess of tition after several hours awake. Finally,
finger-tapping (very apart
to learning a piano scale). We told the enhancement by having subjects learn these data demonstrate that follow
subjects to press a specific five-button one sequence, then interfering with that ing such stabilization across the day, a
sequence as fast as possible, and then motor memory by asking them later night of sleep enhances thosememories,
tested them again at 12 and 24 hours to learn a new sequence. In one group,
thereby resulting in improved perfor
after training. The subjects who slept people learned the firstsequence, then mance the following day. Thus, itseems
normally during the first 12-hour in learned the second 10minutes later.We that thedevelopment of thesememories
terval performed the task 20 percent retested on each memory aftera night of consists of at least three separate stages.
faster and 35 percent more accurately, sleep: Only the second memory, learned It is important tonote that the conclu
but an equivalent period awake provid last, showed significantoffline improve sions we drew from experiments with
ed no significant benefit. However, the ments in accuracy. However, when we the motor-memory task are unlikely
group that stayed awake for the first12 allowed six hours between learning of to be universal. Whereas our task ap
hours (and failed to improve during the firstmemory and learning of the peared linked to stage 2 NREM, other
that time) caught up with the other co second, then tested each after a night forms of procedural learning seem to
hort by 24 hours?after they too had a of sleep, we saw significant offline im depend on other sleep phases. In 1994
night of sleep. Furthermore, we noted
that the amount of overnight improve
ment correlated with the amount of
stage 2NREM sleep, particularly late in
the night. During this period, so-called
sleep spindles?short, high-frequency
bursts of electrical activity?reach peak
density. We and other neuroscientists
hypothesize that these spindles trig
ger intracellular events that modify
the connections between neurons and
may lead to overnight improvements
inmemory.
When we analyzed the transition
speed profiles for individual subjects (in
other words, the time between the first
and second button, the second and third
button, and so on), we found that the
speed of individual key-press transitions
within the sequence was unequal. Some
transitions seemed easy (fast)and others
problematic (slow), as if the subjectwas
parsing the entire sequence intomore
manageable sub-sequences during the
initial training, a phenomenon termed
chunking. (In a similar fashion,people of
ten chunk a long telephone number into
a string of
two-or-three-digit numbers
for easier memorization). Remarkably,
after a night of sleep, the slow, problem
atic transitions had improved, whereas
the fast, easy transitions remained the
same. In contrast,
people who were
trained and retested after eight hours Figure 4.Memory goes through several stages, independent of rehearsal, intent or awareness, on
the way from an ephemeral to a more permanent state. After its initial encoding,
awake did not improve at all. We in representation
a memory is stabilized and enhanced during the process of consolidation. Many studies show
terpreted these findings tomean that that the latter stage requires sleep. With time, a memory becomes into the fabric of
integrated
sleep-dependent consolidation unifies themind. At some point, a declarative memory no longer depends on the hippocampus but ex
or "stitches together" these smallermo ists in distributed form in the cortex, linked by a web of associations to other, related memories.
tor-memory units into one long motor Solid red lines represent periods of known processing; dotted red lines indicate hypothesized
memory program by selectively smooth or variable periods of processing. Note the logarithmic time scale.

www.americanscientist.org 2006 July-August 329

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions
 of sleep the firstnight after learning,but
then have two recovery nights of sleep
before being tested, they still show no
improvement on the task. Thus, sleep
dependent memory consolidation in
stead appears to be an all-or-nothing
event: If you don't sleep within the
first24 hours after learning these new
memories, they are lost. The prospect
is particularly frightening in our 24/7,
don't-wait
hurry-up, society.

In Praise ofNaps
One twist in this story is thatdespite the
clear importance of nightly sleep for full
memory enhancement, short daytime
naps yield surprisingly largebenefits.We
used thefinger-tapping task to compare
theperformance gains of two groups of
subjects:One had a 60- to 90-minutemid
Figure 5. Sleep enhances memory. The author and his colleagues trained subjects on a finger
day nap afteramorning training session,
tapping task (a form of procedural memory) in the morning. All participants improved with and the other did not. Later the same
practice. Later that day, the author tested half the subjects and found that their performance day, the group thathad napped was sig
had not changed (a, green bar at center). However, the other half, which the author tested after
nificantly (about 16 percent) better at per
normal sleep, improved (b, purple bar at center). A night of sleep subsequently these sequences than theyhad
significantly
enhanced thememory of the first group (a, purple bar), but the second group did not continue to forming
been thatmorning. In contrast, and as
get better following another 12 hours awake (b, purple bar at right). Asterisks indicate significant
we noted in theearlier study,people who
differences between initial and later testing; error bars represent standard error.
did not nap and remained awake across
Avi Kami, Dov Sagi and coworkers provements the next day. Subsequently, theday did not improve.
at theWeizmann Institute of Science Robert Stickgold,my frequent collabora Interestingly, when we tested both
in Rehovot, Israel, demonstrated that tor atHarvard, showed that the degree of these groups (subjects who did or
subjects who learned todistinguish spe of improvement correlated positively did not nap) again the next day follow
cific details in a patterned image (a so with the time spent in REM sleep and ing a full night of sleep, those subjects
called visual-skill task) also improved SWS. SteffenGais and his colleagues in who had napped showed only a 7 per
their performance after sleep (but not JanBorn's research team at theUniver cent additional increase inperformance
after an equivalent period of wakeful sity of L?beck inGermany suggest that speed, resulting in a summed total im
ness). In that task, the enhancement ap SWS initiates consolidation, but subse provement of 23 percent. However, sub
peared to depend on REM sleep, since quent REM sleep promotes additional jectswho had not napped were nearly
subjects who were deprived of this type enhancement. Stickgold has since dem 24 percent fasterafter sleeping thenight.
of sleep across a night showed no im onstrated that if subjects are deprived Therefore, by the following day, both
groups had improved by approximately
- -
before awake interval after the same total amount. These data sug
gest theremay be a limit to how much
absolute improvement sleep can trigger
across a 24-hour period. Thus, amidday
nap changes the time course of when
that offline motor memory improve
ment occurs, but ultimately not the total
benefit, as the two groups improved by
the same amount after24 hours.
Daytime naps also appear to improve
the learning of a visual skill, although
key-press transition key-press transition the effectsare subtly different.Stickgold,
Sara C. Mednick and their colleagues at
Figure 6. Sleep selectively enhances themost difficult portions of a complex motor task. Subjects Harvard have shown thatwhen people
learned to tap quickly a sequence of five numbers on a keyboard (for example, 4-1-3-2-4), which
rehearse the visual-perceptual task (de
contained four unique key-press transitions: (A) 4 to 1, (B) 1 to 3, (C) 3 to 2, and (D) 2 to 4. Not all
scribed earlier) several times across the
transitions are equally hard: Some are particularly slow, as reflected in greater reaction times, and
therefore difficult (labeled as problem points). After training, some subjects slept while others day, theybegin to getworse rather than
remained awake. This figure shows representative performance data from separate individuals,
better. However, if the subjects take a
either before and after sleep (a) or before and after an equivalent of wakefulness (b). Fol 30- to 60-minute nap among these tests,
period
lowing sleep (a, purple line), subjects performed the task more quickly, but the improvement was
then thedeterioration halts. If they sleep
specific to the slowest, problem-point transition. By contrast, eight hours awake did not improve longer?60 to 90 minutes?then per
performance, and the transition profile remained unaltered (b, yellow line). formance not only stops declining, but

330 American Scientist, Volume 94

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions
becomes enhanced. Furthermore, these task

nap-based enhancements didn't occlude performance

the improvement that normally comes
with sleep (unlike our results with the high
motor skill). These resultsmay indicate
REM sleep
that certain parts of thebrain (those that
after training
perform the visual-perceptual task) can
fatigue locally,but that short periods of
sleep remedy this condition. Further,
low
thedata suggest that longer naps, which REM sleep
likely contain SWS and REM sleep, lead (no training)
to enhancement ofmemory.

Plastic Brain Changes

To encode new information, the brain Figure 7. The pattern of brain activity that accompanies learning reappears during the subject's
must physically change. Neuroscien REM sleep that night. These images show brain activity recorded with positron-emission
tists call the propensity for this kind of tomography during the performance of a visuomotor task (a), during the REM sleep of sub
who the same task earlier that and the REM sleep of untrained
change plasticity, and it can operate at
jects practiced day (b), during
subjects (c). In each case, the investigators subtracted baseline activation data from an awake,
the level of individual synapses and
resting brain. These patterns show that compared to untrained controls, the patterns of brain
cells, between different circuits and
activity in the REM sleep of trained subjects more closely resembled the activation pattern
even across different brain regions. If, evoked the task itself. et al.
by (From Maquet 2000.)
as thebehavioral results show, sleep en
hances new learning, and if learning past few years, several research teams, turing PET snapshots of subjects' brain
requires plasticity, then the consequence including my own, have used these activitywhile they trained on a motor
of sleep forpeople who've just formed neuroimaging techniques to observe task during the day and then perform
new memories should include physical physical, learning-specific changes in ing scans thatnight,Maquet did see the
changes to the structure of thebrain. the sleeping brain. These results rein reemergence of themotor-learning pat
Of course, it's devilishly tricky to ob force the ideas thatmemory enhance tern during REM sleep. This signature
serve the precise changes that accom ment depends on sleep, and that sleep REM replay did not appear inuntrained
pany memory formation in the human reshapes memories within out brains. subjects. Furthermore, those individu
brain, even when we know where to In 2000, PierreMaquet from theUni als who learned more during the day
look. The best noninvasive tools to ex versity of Liege, Belgium, collaborated time exhibited more replay during REM
amine changing patterns of brain activ with scientistsatUniversity College Lon sleep. This last detail suggests that the
ity are positron-emission tomography don to conduct a study that used PET to process of learning itself (rather than
(PET) and functional magnetic reso see if the sleeping brain subsequently simply performing the task) dictates the
nance imaging (fMRI). Both techniques "replays" the same pattern of activity altered physiology during sleep. The
measure themetabolic activity of spe thatoccurred when a taskwas originally more the brain learns, themore it de
cific brain regions in real time. In the learned during the day. Indeed, by cap mands from sleep at night.

W??? daytime
napgroup W??? controlgroup (no nap)

ffiilP
liflli

across day overnight total(across day

and overnight)
time course of change

Figure 8. Daytime naps can fillmuch the same role inmemory enhancement as a full
nights sleep, although adding
a nap to a
night's sleep doesn't
offer any greater benefit than a night of sleep alone. In this experiment, subjects practiced a motor skill task in themorning and either napped (60
to 90 minutes) or remained awake until retested later the same day, subjects who had napped
atmidday evening. When (purple) were significantly
faster (16 percent), but the performance of controls (red) was unchanged. After a full night of sleep, the performance speed of subjects in the nap
group only increased by an additional 7 percent, but the control group sped up by 24 percent. Therefore, within 24 hours, both groups averaged the
same total amount of
delayed learning. Several famously creative thinkers throughout history have been dedicated nappers, including Leonardo
Da Vinci, Salvador Dali, Buckminster Fuller and Thomas Edison (shown here asleep on a laboratory bench in 1911). Asterisks indicate significant

performance differences between training and testing. (Photograph courtesy of the National Park Service, Edison Historical Site.)

www.americanscientist.org 2006 July-August 331

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions
 ments used unrelated words, such as
dog-leaf, Born used related pairs, such
as dog-bone. The former task required
subjects to form and retain completely
novel associations (dog-leaf), but Born's
task called for the strengthening or tag
ging ofwell-formed associations (dog
bone) for subsequent recall. Thus, it's
possible that sleep is not an absolute
requirement for the consolidation of de
' clarativememory, but itcould be neces
+9 -5f|-7
5_g| ] sary forspecific tasks, such as those that
activation strength deactivation strength
play on semantic associations.
Figure 9. Physical changes in the brain accompany sleep-dependent learning. The author used In addition, emotion can affect de
functional magnetic-resonance imaging to compare patterns of brain activity between subjects clarative memory, and my colleagues
who had either slept or remained awake after training on a test of motor-skill memory. Follow
and I recently showed that sleep also
ing sleep, the left cerebellum (a), the right primary motor cortex (b) and the right hippocampus
enhances emotionally charged de
(c) were more active. Several regions also showed less activity post-sleep, including the left and
clarative memories more than neutral
right parietal lobes (d) and other areas responsible for emotion and motivation (not shown).
ones. We firstpresented to subjects a
My laboratory used a different ex these data showed that sleep-dependent mixture of emotionally evocative and
perimental design to examine sleep motor learning coincided with large neutral pictures. Half of the group re
dependent plasticity by comparing pat scale restructuring of thememory rep mained awake for 12 hours, half slept.
ternsof brain activation during memory resentation within the brain overnight. At the end of thisperiod, we tested the
recall, either after a night of sleep or fol We think these changes enable subjects subjects on whether they recalled ever
lowing the same period ofwakefulness. to return the next day and perform the seeing the images. Generally, people
Our hypothesis was that ifmemory task faster,more accurately and more who had slept scored better across the
had improved the next day, then these automatically following sleep. board. However, the breakdown by
performance enhancements must be ac
group and image type was remark
companied by specific changes within Memory Modifiers able: Among the sleepers, recognition
the brain and would be evident from Although not all studies of human of emotional images improved by 42
thepattern of "where" thememory was behavior support the link between percent relative to the awake group.
being recalled after sleep, as shown in declarative learning and sleep, many Without sleep, the subjects' recall of
theMRI activation images. Our quarry: do. For example, some studies that emotional scenes was not significantly
sleep-dependent restructuring of the used a verbal memory task reported different from that of neutral images.
neural representation ofmemory. no change in the architecture of the These data indicate that sleep, rather
We again used the finger-tapping sleeping brain after training, but oth than time per se, selectively helps to
task, inwhich two groups are trained ers found the opposite: significantly consolidate this form of emotional
and then both are tested 12 hours later. more REM sleep among subjects who memory. This type of consolidation
One of them sleeps through the night, had intensive training in a foreign may be related toREM sleep late in the
and the other remains awake across the language earlier that day. In the lat night: The brain structures that lightup
day. Charing the testingphase, we exam ter experiment, themore an individu at that time, together with the neuro
ined brain activitywith fMRI.After con al learned, the greater her increase in chemicals released, are the same ones
trolling for individual differences and REM sleep (a result similar toMaquet's that support emotional memories.
circadian fluctuations inmetabolism, we study of themotor-learning task). Still, Taken together, these and other stud
identified several regions that differed the findings are not in perfect agree ies suggest that sleep plays an important
between groups. The sleepers' brains ment (likemany in science), and some but nuanced role in conscious learning.
showed more intense activity and a larg open questions remain, including the Although the contribution of REM sleep
er active area in the right primary mo degree of sleep alteration after training to simple, emotion-free declarative
tor cortex and leftcerebellum, changes on declarative tasks and the degree of tasks isnot settled, a substantial body of
consistent with improving the accuracy learning impairment that follows selec evidence indicates that SWS and REM
and speed of movement. The medial tive sleep deprivation. sleep contribute (respectively) to the
prefrontal lobe and hippocampus were Recent papers by Born and his col consolidation of complex and emotion
also more active, consistentwith the fact leagues show that subjects who learned ally salient declarative memories?par
that thesebrain regions can help toorder certainword pairs performed better af ticularly those memories that link to
the sequence of movements. ter early sleep?the part of the night networks of preexisting associations.
Some areas of the sleepers' brains largely devoted to SWS?and their
were less active, including the leftand brainwaves showed more frequent Caffeine IsNo Substitute
right parietal cortices and the extended sleep spindles during this period than So far, I've discussed the need of sleep
limbic system.The formercould reflecta did those of controls. However, earlier after learning formemory consolidation.
reduced need for conscious monitoring studies with similar tasks had reported But what about sleep before learning?
as a result of improved task automation; no connection to sleep. The discrepancy Not surprisingly, a sleepless night also
the lattercould indicate a lessened emo may reflect subtle details of how mem hampers the process of encoding novel
tionalburden of the task.Taken together, ory works. Whereas previous experi information. For example, being awake

332 American Scientist, Volume 94

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions
 Two principal questions promise to
dominate thisfield in the future: What is
itabout sleep?brain chemistry,regional
brain activation, electrical oscillations?
that triggers changes in individual syn
apses, cells and circuits? And, what is
the role of sleep in postconsolidation
processes, such as integrating memo
ries, and even erasing them?Neurosci
entists need towork across disciplines
to answer these questions, but with
the current growth of the field, I expect
that important advances will continue
to emerge. Treatments for disorders
of memory (and perhaps even the en
all stimulustypes h positive negative hancement of normal function)may not
Figure 10. Sleep deprivation exerts variable consequences
on declarative memory depending on the be a dream of the distant future, but
emotional significance of thememory. The author found that on average, 36 hours without sleep
reality in our time.
resulted in a 40 percent decline in the ability to form new memories (a). However, when he segre
was
gated the data according to the emotional tenor of each item in the test,he found that the deficit Bibliography
most pronounced for positive and, to a lesser extent, neutral associations; recall of those memories
Gais, S., and J. Born. 2004. Low acetylcholine
with a negative association were impaired the least as a result of sleep deprivation (b). Asterisk
during slow-wave sleep is critical for de
indicates statistical significance (p<0.05); double asterisk indicates high significance (p<0.01). clarative memory consolidation. Proceed
ings of theNational Academy of Sciences of the
for36 hours robs people ofmuch of their finding that agrees with the hypothesis U.S.A. 101:2140-2144.
ability to perform a test of "temporal" thatemotion improves learning. By con Gais, S., M. Molle, K. Helms and J.Born. 2002.
memory (rememberingwhen things oc trast, the sleep-deprived subjects were Learning-dependent increases in sleep

curred) and, interestingly,cripples their
insight intohow well they actually per
formed the task, according to a study by
Yvonne Harrison and JamesA. Home at
Loughborough University in theUnited
severely compromised in remembering spindle density. Journal of Neuroscience
22:6830-6834.
positive and, to a lesser extent, neutral
stimuli (almost a 60 percent decrement Kami, A., D. Tanne, B. S. Rubenstein, J. J.Aske
nasy and D. Sagi. 1994. Dependence on REM
forpositive associations). Furthermore, a
sleep of overnight improvement of percep
negative emotional memories were tual skill. Science 265:679-682.

Kingdom. This bodes ill for the success
of people who need to acquire and ana
lyze new information while sleep de
prived, such as physicians or soldiers.
They may think that they're doing just
fine on little sleep, but because their
ability to judge is compromised, they
are more likely to be wrong. Harrison
and Home also showed that caffeine,
while increasing general alertness, did
notmend theperformance deficits.
However, not all forms of declarative
memory are affected equally by sleep
somewhat more resistant to the effects Maquet, P., et al 2000. Experience-dependent
of sleep deprivation; the groups were changes in cerebral activation during human
REM sleep. Nature Neuroscience 3:831-836.
not significantly different in their recol
S. C., K. Nakayama, J.L. Cantero, M.
lection of negative word pairs. Mednick,
Atienza, A. A. Levin, N. Pathak and R. Stick
The take-home message from these on
gold. 2002. The restorative effect of naps
studies is that sleep before learning is
perceptual deterioration. Nature Neuroscience
critical for the brain to be able to lay 5:677-681.
down new memories. Without it, the Squire, L. R., and B. J.Knowlton. 1994. Memory,
initial coding of information suffersdra hippocampus and brain sysytems. In The
Cognitive Neurosciences, ed. M. S. Gazzaniga.
matically, resulting inmemories thatdo
not persist in the long term. Further Cambridge, Mass.: MIT Press.
Stickgold, R., D. Whidbee, B. Schirmer, V. Patel
more, the impact of inadequate sleep and J.A. Hobson. 2000. Visual discrimina
may be different for different types of tion task improvement: A multi-step process

deprivation. My lab recently looked at information, specifically positive forms occurring during sleep. Journal of Cognitive
the influence of emotion on verbal learn of learning, meaning it's harder to re Neuroscience 12:246-254.

member the happy events of the day Walker, M. P., T. Brakefield, J.A. Hobson and R.
ing in subjects who had slept normally
or not at all for 36 hours prior to learn without a good night's rest. Stickgold. 2003. Dissociable stages of human
memory consolidation and reconsolidation.
ing.After we presented pairs ofwords Nature 425:616-620.
thathad positive (happy, love, sunlight), Dreams Becoming Reality Walker, M. P., T. Brakefield, A. Morgan, J.A.
negative (cancer,grief, jail) or neutral as The publication rate in the field of sleep Hobson and R. Stickgold. 2002. Practice with
sociations, the participants slept as they and memory has doubled in each of the sleep makes perfect: sleep dependent motor
skill learning. Neuron 35:205-211.
wished fortwo nights and then returned; past two decades?a rate that eclipses
we surprised them at thatpoint with an the growth of research in either sleep
unexpected testofword recognition. or
memory alone. These reports, from
The data showed that subjects who cellular and molecular studies in ani
were sleep-deprived before learn mals to behavioral studies in humans, For relevant Web links, consult this
issue of American Scientist Online:
ing remembered 40 percent less than provide converging evidence that pre
controls?a striking impairment?but training sleep prepares the brain for
this deficit was not the same in each learning, and posttraining sleep trig http://www.americanscientist.org/
emotional category. Rested subjects gers memory consolidation through lssueTOC/issue/861
had better recall of positive and nega neural plasticity, leading to enhanced
tive stimuli than of neutral stimuli, a recall thenext day.

www.americanscientist.org 2006 July-August

This content downloaded from 128.235.251.160 on Fri, 13 Feb 2015 12:50:45 PM

All use subject to JSTOR Terms and Conditions