Investigations and research
Enhancing sleep slow waves with
G.Tononi
B.A. Riedner
B.K. Hulse
F. Ferrarelli
S. Sarasso
E Several lines of evidence
underscore the critical
importance of sleep.
E The ability to ameliorate
the detrimental effects
of sleep loss could have
an immediate practical
impact.
82 MEDICAMUNDI 54/2 2010
natural stimuli
David P. White Chair in Sleep Medicine, Department of Psychiatry,
University of Wisconsin, Madison, WI, USA.
Department of Psychiatry, University of Wisconsin, Madison, WI, USA.
The importance of sleep
Several lines of evidence underscore the critical
importance of sleep [1]. First, sleep is universal.
Despite the danger of being relatively disconnected
from the outside world while asleep, sleep has
been identified in every animal species studied
to date. Second, sleep is a tightly regulated
homeostatic process. The longer one stays awake,
the more sleep pressure increases, and such
pressure can only be relieved by sleep. Last, lack
of sleep has negative consequences. Even modest
sleep restriction leads to cognitive impairment
and intrusion of sleep into waking, which are
responsible for decreased work productivity,
mortality related to automobile crashes, and other
adverse events [2]. Recent surveys suggest that
a large percentage of the population is sleep
restricted, whether by necessity or by choice.
Clearly, the ability to ameliorate the detrimental
effects of sleep loss could have an immediate
practical impact on a society for which the time
for sleep is becoming increasingly curtailed.
Slow waves are critical to maximizing the
power of sleep
The best known marker of the homeostatic
regulation of sleep are electroencephalographic
(EEG) slow waves. Sleep slow waves are high
amplitude EEG waves (>50 microvolt) that occur
once every second or so during deep non-rapid
eye movement (NREM) sleep. The longer one
has been awake, the more frequent and larger
are the slow waves during sleep. Conversely, slow
waves become fewer and smaller the more time
one spends asleep. Until recently, slow waves
have been viewed simply as a useful biomarker
of sleep pressure, with not much thought given
as to why they reflect the need for sleep. However,
a recent proposal - the synaptic homeostasis
hypothesis – suggests that slow waves index sleep
need because they track a fundamental biological
process - the regulation of synaptic strength [3].
During wakefulness, neural circuits in the brain
undergo plastic changes, mainly in terms of
modifications in the strength (or number) of
synapses, which underlie learning and memory.
The synaptic homeostasis hypothesis suggests
that learning-related plastic changes result in a
net increase in synaptic strength by the end of a
waking day. However, stronger synapses need
more energy, space, and cellular supplies, and
continued strengthening may eventually
saturate the ability to learn. Thus, a progressive
increase in the strength of synapses day after day
is unsustainable in the long run. Sleep would
thus be crucial to re-normalize synaptic strength,
restoring the brain to optimal conditions in terms
of energy, space, supplies, and learning ability.
The hypothesis then suggests that the strength
of synapses is reflected in the amplitude of sleep
slow waves, because stronger synapses lead to
better synchronization among cortical neurons,
and the more neurons are synchronized, the
larger the waves that can be recorded by the EEG.
Subsequently, decreasing strength over the course
of a night’s sleep leads to a progressive decline in
neuronal synchronization and thus in the number
and size of slow waves.
Finally, the synaptic homeostasis hypothesis
proposes that slow waves do not just reflect the
regulation of synaptic strength, but that they
have a direct causal role in mediating synaptic
renormalization during sleep. Such a role in
reducing net synaptic strength is supported by
several considerations. The neuronal slow
oscillations underlying EEG slow waves are of
low frequency (around 1 Hz), consist of highly
synchronized burst firing followed by silence,
include an alternation between depolarization
and hyperpolarization of the membrane potential
of neurons [4, 5], and are thus ideally suited to
inducing synaptic depression. In addition, slow
waves occur at a time when the
neuromodulatory milieu of the brain (low
acethylcholine and norepinephrine) is well-suited
to induce depression [6].
Altogether, then, the synaptic homeostasis
hypothesis predicts that slow waves should
mediate at least some of the beneficial functions
of sleep on the brain. Indeed, some initial
evidence supports this prediction. In two recent
studies in humans, for instance, slow waves were
reduced by delivering sounds whenever they
appeared in the EEG, yet without awakening the
subjects. This slow wave deprivation prevented
the improvement in performance that is observed
after a night of sleep in a variety of tasks [7,8].
If curtailing slow waves impairs performance,
could enhancing slow waves improve
performance instead? This might be especially
useful under conditions of sleep restrictions,
when one could try to induce a larger amount
of slow waves over shorter periods of sleep.
Maximizing the restorative value of sleep by
increasing slow waves could be attempted by
both pharmacological and non-pharmacological
means. Recently, the effects of pharmacologically
enhancing slow waves with two separate drugs
have been investigated [9,10]. Both gaboxadol
(a gamma-aminobutyric acid agonist) and
tiagabine (a gamma-aminobutyric acid reuptake
inhibitor) were effective in increasing slow waves
under sleep restriction. Gaboxadol subjects were
less sleepy compared to controls, while tiagabine
subjects demonstrated improved performance on
a sustained attention and a working memory
task. Although these initial results are promising,
pharmacological approaches are hindered by
issues of dependence and tolerance and are often
associated with residual daytime side effects.
Moreover, pharmacological agents cannot be
completely washed out of the metabolic cycle in
cases where an individual, like a doctor on-call
or a soldier at war, is unexpectedly awoken from
sleep and needs to function.
In order to overcome these limitations, it would
be preferable to develop an alternative, non-
pharmacological approach for sleep enhancement.
As early as the 1940’s, investigators considered
peripheral stimulation as a means to improve
sleep. However, early attempts concentrated
exclusively on reducing sleep latency and not so
much on enhancing sleep (for instance see [11,
12]). More recent investigations have focused on
the purported benefits of playing music during
sleep and some have even recognized the
importance of entraining the slow wave rhythm,
designing specific “delta” music with this aim
[13]. Unfortunately, the impact of music on sleep
is typically assessed with subjective reports, and
when sleep scoring and EEG quantification were
performed in a controlled manner, it was found
to have no effect [14].
Another potential method for enhancing slow
waves is transcranial direct current stimulation E The restorative value of
(tDCS), which can produce a widespread shift sleep could be enhanced
in the membrane potential of superficial neurons by both pharmacological
in either a depolarizing direction (anodal and non-pharmalogical
stimulation) or in a hyperpolarizing direction means.
(cathodal stimulation). The human brain
undergoes an endogenous negative DC potential
shift, particularly over frontal cortical areas, at
the onset of slow wave sleep, suggesting that
inducing such DC potential shifts may enhance
or induce slow oscillatory activity [15].
Preliminary work done in our laboratory and a
recent study by Marshall et al. have shown that
anodal tDCS applied intermittently (15 sec on,
15 sec off) does result in an acute increase in
slow wave activity (SWA, EEG power 0.5 - 4 Hz)
at the beginning of the off period [16]. However,
the actual impact on physiological sleep is hard
to evaluate because EEG cannot be recorded
during the on periods because of artifacts. Also,
a recent PET study showed that tDCS produces
a complex pattern of activated and deactivated
brain areas [17] suggesting that the impact on
slow waves may be mixed and difficult to predict.
Perhaps the most direct demonstration that it is
possible to reliably and non-invasively induce
sleep slow waves in humans was obtained by
perturbing the cortex at a rate of approximately
1 Hz using transcranial magnetic stimulation
(TMS) [18]. Unlike tDCS, the efficacy of TMS
could be objectively verified by concurrently
recording evoked EEG slow waves. Much like a
pacemaker, each and every TMS pulse delivered
during NREM sleep with the appropriate
parameters, triggered a full-fledged slow wave
that started under the stimulator and spread to
the rest of the brain. Importantly, TMS evoked
slow waves resemble spontaneous slow waves in
all aspects and the repetitive triggering of these
waves produces an EEG that resembles the
deepest periods of sleep in terms of SWA.
While the TMS study indicated that it is indeed
possible to entrain the cortex specifically at the E Among the various
rate of the slow oscillation, TMS is costly, bulky, natural stimulation
and relatively complex to operate. Thus, we modalities, auditory
wondered whether the same effect could be stimulation is an
achieved using more natural stimuli. Among the obvious first choice.
various natural stimulation modalities, auditory
stimulation is an obvious first choice because it
is safe, easily controllable, and can be
administered non-obtrusively during sleep. MEDICAMUNDI 54/2 2010 83
  1

Figure 1. Alice Sleepware recording

for three different subjects
demonstrating the clear effect of
acoustic tones on EEG sleep slow
waves. Each plot includes a 4 hour
trend plot on the upper portion and
approximately two minutes of
electro-oculogram, EEG, and
electromyogram data on the bottom.
The lowest data line indicates the
timing of the tone stimulation. The
red arrow on the first plot indicates
slow waves on a left frontal lead
(F3-A2) as an example.

E Auditory tones have the

ability to affect the EEG
by producing K-complex
waves during lighter stages
of NREM sleep.

84 MEDICAMUNDI 54/2 2010
Moreover, it has been known since the very first
sleep recordings that auditory tones have the
ability to affect the EEG by producing K-complex
waves during lighter stages of NREM sleep -
waves that are very much, if not the same, as the
spontaneous slow waves of deep sleep [19].
Therefore, we wondered whether by changing the
tone parameters we could obtain an enhancement
of slow waves, similar to what was demonstrated
with TMS.
Our preliminary research with auditory tone
stimulation, as well as some limited results from
olfactory and somatosensory stimuli will be
discussed below. IRB approval was obtained for
all the experiments described and written informed
consent was obtained for each subject after the
relevant procedures had been described in detail
(IRB protocol #2009-0052).
Peripheral stimulation can increase slow
waves
Figure 1 shows examples from three subjects
demonstrating how tones can enhance slow waves.
In each case, brief tones (50 millisecond duration)
were played via headphones or speakers between
  2a

Figure 2. Histograms of all tone ON block versus tone

OFF block comparisons pooled across three subjects
for each sleep cycle. The skew to the right indicates that
in most cases stimulation blocks resulted in an increase
in slow waves (red bars). Note that the effect persists
even in later sleep cycles.

0.8 and 2 Hz, a rate that approximated the

natural cellular oscillation of cortical neurons
during sleep. Sleep was recorded using the
Respironics Alice 5 System with a six-channel
EEG montage that included chin
electromyography and two electro-oculographic
channels. Given that sleep depth can vary from
moment to moment, blocks of 15-20 tones were
played (ON blocks) during NREM in order to
make valid comparisons between stimulation
periods (ON) and no stimulation periods (OFF).
ON blocks were then contrasted with the
2b
preceding and subsequent OFF blocks of equal
length. Importantly, tone intensity was manually
adjusted so as to be above the individual subject’s
auditory threshold during waking, but still quiet
enough as not to awaken the subject from sleep.

Figure 2 shows histograms of the percentage

changes in slow wave power between ON periods
and flanking OFF periods [(ON-OFF)*200/
(ON+OFF), rectified, 0.5-4 Hz filtered EEG
signal downsampled to 1 second resolution,
F3-A2 derivation]. The effects of tones were
evaluated throughout the night, subdivided by
sleep cycle, and pooled across subjects. The
histograms show that ON periods led to higher
slow wave power than the flanking OFF periods,
and that the increase in slow wave power was
present for all 3 sleep cycles. Thus, tones were
effective in increasing slow waves not only at the
beginning of the night, when slow waves are 2c

more abundant (median 12%, mean 14% increase

for the first cycle), but even more effective later
in the night when slow waves are typically fewer
and smaller (median 44%, mean 47% increase
for the third or later sleep cycle), as long as tone
intensity is modified appropriately. Similar effects
were observed for the contralateral derivation
(F4-A2) as well as the central leads (C3-A2 and
C4-A1). A less prominent effect was observed
for the occipital leads.

Is the enhancement of sleep slow waves unique

to acoustic stimulation? Based on theoretical
considerations and results from previous work in
other species, at least two other forms of
stimulation deserve attention: smell and touch.
The olfactory system is especially intriguing and
unique in that it has direct cortical projections
that do not go through the thalamus. Research
in rats suggests that slow waves can actually be MEDICAMUNDI 54/2 2010 85
 E 3

Figure 3. High-density EEG

topographic representation of the
effect of stimulation on SWA.
Normal sleep SWA topography is
characterized by a frontal
predominance which persists during
stimulation nights (top and middle
rows). The frontal increase seen
with acoustic stimulation resembles
normal sleep SWA topography
(ON/OFF bottom row, left). The
peak of the median nerve
stimulation increase is instead
localized over an area near
somatosensory and motor cortices
and is smaller in magnitude (ON/
OFF bottom row, right). Note the
difference in scales between the
ON/OFF ratios for each stimulation
modality.

E Acoustic stimulation
appears to be a promising
means of inducing slow
waves.
86 MEDICAMUNDI 54/2 2010
driven by the olfactory bulb [20]. Thus, olfactory
stimulation may provide a unique opportunity
to bypass the thalamic gate that hinders other
peripheral stimuli during sleep from reaching the
cortex, where slow waves are thought to be
generated. However, our preliminary data in
humans have revealed surprisingly little effect of
olfactory stimulation on sleep slow waves. Air
puffs delivered to the nostrils during NREM sleep
with stimulation frequencies ranging from
0.5 – 2 Hz and either with or without scent have
produced no consistent change in number or
intensity of sleep slow waves.
Regarding touch, cutaneous low-intensity nerve
stimulation was shown to have a synchronizing
effect on the sleep EEG in cats almost 50 years
ago [21]. Therefore, we examined the ability of
median nerve stimulation to enhance slow
waves in humans. Similar to the acoustic
stimulation, we delivered bilateral stimulation
to the median nerves of the sleeping subject.
The stimulation intensities employed in these
experiments did not awaken the subjects, who
reported no conscious perception of the
stimulation upon awakening. At the highest
intensities, some effect of the stimulation on
the ongoing EEG could be observed, but it was
not always consistent. Thus, at least with the
current stimulation paradigm, acoustic
stimulation appears to be a promising means of
inducing slow waves, although other modalities
of stimulation deserve consideration (e.g. tDCS
and vestibular stimulation).
The topography of slow wave enhancement
Sleep is a global cortical phenomenon, and
virtually all cortical neurons exhibit the slow
oscillatory activity underlying EEG sleep slow
waves [22]. Still, topographic analysis of sleep
SWA in humans has revealed that the need for
sleep may not be the same for all cortical areas
[23]. Therefore, in qualifying the enhancement
of sleep slow waves by peripheral stimuli, it is
important to know the topography of slow
wave enhancement, which can be done using
high-density EEG recordings (>= 60 channels).
As shown in Figure 3, acoustic stimulation
produced a diffuse increase in SWA across the
scalp, between 10% and 250% (ratio of SWA
during the ON and OFF blocks). Furthermore,
the topography of the increase in SWA is very
similar to the topographic profile of SWA during
natural sleep, which is characterized by a frontal
predominance [23]. By contrast, high-density
EEG analysis of the effect of median nerve
stimulation during sleep revealed a 30% increase
in SWA that was restricted to somatosensory
and motor cortex contralateral to the dominant
hand, which is probably why no consistent effect
was obtained when only a few channels were
used for the analysis.
Further development and evaluation
The above results identify acoustic stimulation
as a promising non-pharmacological means for
increasing slow waves. The effect is clear - blocks
of tones at the appropriate intensity during NREM
sleep increase the number and size of slow waves
relative to blocks where there are no tones.
Importantly, the impact of tones is lasting,
increasing slow waves even more effectively
during later sleep cycles. Finally, the increase in
slow waves shows topography consistent with
the frontal predominance of normal sleep. Yet
several questions remain to be addressed. Does
the stimulation actually increase overall SWA
over the course of the night relative to a night
where there is no stimulation, or does it merely
enforce a redistribution of SWA in ON and
OFF periods? Can acoustic stimulation
increase the density of slow waves such that a
full-night complement can be obtained over a
shorter period of sleep? If so, would this
increase in slow wave density be sufficient to
compensate for sleep loss by reversing the
negative effects of sleep restriction on cognitive
performance, sleepiness, attention and mood?
Answers to these questions will obviously require
controlled studies with appropriately chosen sham
stimulations. Even if these questions are
successfully answered, translating the results to
an appropriate tool for sleep improvement will
require algorithm optimization in order to deliver
ongoing stimulation of variable intensity titrated
to the ever-changing landscape of sleep
architecture. Basic work will be needed to elucidate
why acoustic stimuli appear to be especially
effective in inducing slow waves, and why the
precise pattern of stimulation (tone duration and E Basic work will be
frequency, tone intervals, and block duration) needed to elucidate why
appears to be so important. Also, despite the acoustic stimuli appear
promise of acoustic stimulation, other modalities to be especially effective
of stimulation may turn out to be effective in inducing slow waves.
provided appropriate stimulation parameters
are found.
Acknowledgements
The authors would like to thank Philips
Respironics for their support during the pilot
phase of these experiments L

References

[1] Cirelli C, Tononi G. Is Sleep Essential? PLoS Biol. 2008
Aug;6(8):e216.
[2] Banks S, Dinges DF. Behavioral and Physiological Consequences of
Sleep Restriction. J Clin Sleep Med. 2007 Aug 15;3(5):519-28.
[3] Tononi G, Cirelli C. Sleep Function and Synaptic Homeostasis.
Sleep Med Rev. 2006;10(1):49-62.
[6] Seol G, Ziburkus J, Huang S, Song L, Kim I, Takamiya K,
Huganir R, Lee H, Kirkwood A. Neuromodulators Control the
Polarity of Spike-Timing-Dependent Synaptic Plasticity. Neuron.
2007 Sep;55(6):919-29.
[7] Aeschbach D, Cutler A, Ronda J. A Role for Non-Rapid-Eye-
Movement Sleep Homeostasis in Perceptual Learning. J Neurosci.
2008 Mar;28(11):2766-72.

[4] Czarnecki A, Birtoli B, Ulrich D. Cellular Mechanisms of Burst
Firing-Mediated Long-Term Depression in Rat Neocortical
Pyramidal Cells. J Physiol. 2007 Jan;578(Pt 2):471-9.
[5] Lubenov E, Siapas A. Decoupling through Synchrony in Neuronal
Circuits with Propagation Delays. Neuron. 2008 Apr;58(1):118-31.
[8] Landsness E, Crupi D, Hulse B, Peterson M, Huber R, Ansari H,
Coen M, Cirelli C, Benca R, Ghilardi M, Tononi G. Sleep-
Dependent Improvement in Visuomotor Learning: A Causal Role for
Slow Waves. Sleep. 2009 Oct;32(10):1273-84.
MEDICAMUNDI 54/2 2010 87
[9] Walsh J, Perlis M, Rosenthal M, Krystal A, Jiang J, Roth T. [23] R iedner BA, Vyazovskiy VV, Huber R, Massimini M, Esser S,
Tiagabine Increases Slow-Wave Sleep in a Dose-Dependent Fashion Murphy M, Tononi G. Sleep Homeostasis and Cortical
without Affecting Traditional Efficacy Measures in Adults with Synchronization: III. A High-Density EEG Study of Sleep Slow
Primary Insomnia. J Clin Sleep Med. 2006 Jan;2(1):35-41. Waves in Humans. Sleep. 2007 Dec 1;30(12):1643-57

[10] Walsh J, Snyder E, Hall J, Randazzo A, Griffin K, Groeger J,

Eisenstein R, Feren S, Dickey P, Schweitzer P. Slow Wave Sleep
Enhancement with Gaboxadol Reduces Daytime Sleepiness During
Sleep Restriction. Sleep. 2008 May;31(5):659-72.

[11] Oswald I. Falling Asleep Open-Eyed During Intense Rhythmic

Stimulation. Br Med J. 1960 May 14;1(5184):1450-5.

[12] Webb WB, Agnew HW, Jr. Sleep Onset Facilitation by Tones.
Sleep. 1979 Spring;1(3):281-6.

[13] Thompson J. Delta Sleep System [CD]: The Relaxation Company,

New York; 1999.

[14] L azic S, Ogilvie R. Lack of Efficacy of Music to Improve Sleep: A

Polysomnographic and Quantitative EEG Analysis. Int J
Psychophysiol. 2007 Mar;63(3):232-9.

[15] Marshall L, Mölle M, Born J. Spindle And Slow Wave Rhythms at

Slow Wave Sleep Transitions are Linked to Strong Shifts in the
Cortical Direct Current Potential. Neuroscience.
2003;121(4):1047-53.

[16] Marshall L, Helgadóttir H, Mölle M, Born J. Boosting Slow

Oscillations During Sleep Potentiates Memory. Nature. 2006
Nov;444(7119):610-3.

[17] L ang N, Siebner H, Ward N, Lee L, Nitsche M, Paulus W,

Rothwell J, Lemon R, Frackowiak R. How does Transcranial DC
Stimulation of the Primary Motor Cortex Alter Regional Neuronal
Activity in the Human Brain? Eur J Neurosci. 2005 Jul;22(2):495-
504.

[18] Massimini M, Ferrarelli F, Esser SK, Riedner BA, Huber R,

Murphy M, Peterson MJ, Tononi G. Triggering Sleep Slow Waves
by Transcranial Magnetic Stimulation. PNAS. 2007 May
15;104(20):8496-501.

[19] L oomis AL, Harvey EN, Hobart GA 3rd. Distribution of

Disturbance Patterns in the Human Electroencephalogram, with
Special Reference to Sleep. J Neurophysiol. 1938;1:413-30.

[20] Fontanini A, Bower JM. Slow-waves in the olfactory system: an

olfactory perspective on cortical rhythms. Trends Neurosci. 2006
Aug;29(8):429-37.

[21] Pompeiano O, Swett JE. EEG and Behavioral Manifestations of

Sleep Induced by Cutaneous Nerve Stimulation in Normal Cats.
Arch Ital Biol. 1962;100:311-42.

[22] Steriade M, Nunez A, Amzica F. A Novel Slow (< 1 Hz) Oscillation

of Neocortical Neurons In Vivo: Depolarizing and Hyperpolarizing
Components. J Neurosci. 1993 Aug;13(8):3252-65.

88 MEDICAMUNDI 54/2 2010