Sleep deprivation is known to lead to impairments in
the visceral domain; complete lack of sleep leads to death
in experimental animals [12, 21, 41, 45]. Many visceral dis-
orders, especially of the gastrointestinal tract, are seen in
sleep disorders in humans [26, 29, 39]. However, it is like-
ly that because of clear changes in brain electrical activity
accompanying the transition from waking to sleep, and the
fantastic patterns of dreams, most basic studies and
hypotheses on the functional importance of sleep have been
focused on different aspects of brain functions. Reviews of
these studies have been presented in a number of publica-
tions [2, 16, 18]. It is widely recognized that during slow-
wave sleep, cortical neurons show characteristic changes in
activity, transferring to a regime of rare, periodic bursts with
pauses, reflected in slow EEG waves. Animals choose quiet,
dark places for sleep, with soft litter, to decrease the level of
activation of exteroceptors. Neuronal mechanisms increas-
ing thresholds on the sensory information conduction path-
ways to the brain [28, 41] are also activated, while conduc-
tion of signals from the cortical motor zones to the body
muscles are inhibited, i.e., so-called sleep atonia .
However, the following question remains: how could
all the changes in brain activity listed above, which accom-
pany the transition from waking to sleep, be associated with
visceral health? We might hope that an understanding of the
nature of this connection would finally allow sleep to be
converted from a semi-mysterious state to a number of
understood physiological phenomena.
With the aim of combining the cerebral effects accom-
panying the transition from waking to sleep and the viscer-
al consequences of sleep deprivation into a single system,
we proposed a simple but at first glance fantastical hypoth-
esis. The essence of this hypothesis is that during sleep,
the same cortical neurons which during waking analyze
exteroceptive information of different modalities switch to
analyzing interoceptive information arriving from a variety
of visceral systems.
Theoretical analysis of this suggestion showed that the
ambit of this approach includes explanations for many phe-
nomena which appear to have escaped explanation to date.
It becoems clear why cortical neuron spike activity does not
decrease during sleep, but can even increase, despite active
Neuroscience and Behavioral Physiology, Vol. 44, No. 4, May, 2014
The Visceral Theory of Sleep
I. N. Pigarev UDC 612.821.7
0097-0549/14/4404-0421 ©2014 Springer Science+Business Media New York
Translated from Zhurnal Vysshei Nervnoi Deyatel’nosti imeni I. P. Pavlova, Vol. 63, No. 1, pp. 86–104,
January–February, 2013. Original article submitted May 24, 2012. Accepted October 19, 2012.
This review addresses studies seeking experimental confirmation of the author’s visceral theory of sleep,
along with independent investigations whose results are consistent with this hypothesis. The visceral
hypothesis suggests that during sleep, the central nervous system, particularly the cerebral cortex, switch-
es from analyzing exteroceptive information to analyzing signals arriving from interoceptors distributed
throughout all the systems of the body. Substitution of cortical afferentation during sleep implies a simul-
taneous substitution in the targeting of efferent cortical information streams. In waking, these streams are
directed to structures supporting behavior in the environment. During sleep, they switch to structures sup-
porting the efficient operation of all the visceral systems. Analysis of the visceral hypothesis of sleep
shows that many pathological states associated with the sleep–waking cycle can be explained in terms of
impairments to the synchronicity of the switching of information streams in the cerebral cortex going from
waking from sleep and vice versa.
Keywords: sleep, sleep disorders, waking, visceral systems, behavior, cortical activity.
A. A. Kharkevich Institute of Problems in Information
Transmission, Russian Academy of Sciences, Moscow;
blockade of the conduction of exteroceptive signals to the
cortex. In sleep, this activity is maintained by the visceral
afferent stream. Slow-wave cortical activity recorded at
defined stages of sleep may result from interference
between the periodic activities of the various visceral sys-
tems (gastrointestinal tract peristalsis, respiration, signals
from the heart). The reason why the conduction of signals
from the motor cortex to spinal cord motoneurons is
blocked is understandable: this activity during sleep is asso-
ciated with analysis of visceral information and must be
directed not to the system organizing body movements but
to the corresponding visceral system.
However, while this is an attractive approach, there is
a major obstacle which is difficult to overcome – the com-
plexity of the organization of the brain, as has become
increasingly apparent over the last century. It is now wide-
ly recognized that different cortical zones are specialized
processors. Thus, there are few who doubt that the visual
zones process signals arriving from the retina, and many
investigators find it difficult to accept that these neurons
could be involved in analyzing signals arriving, for exam-
ple, from the gastrointestinal tract. However, we should
note that computers are based on the operation of universal,
rather than highly specialized, processors. There is no doubt
that the construction of manmade computers is much sim-
pler than the organization of one of the most powerful
known information processing devices – the animal brain.
It is difficult to accept that the “universal processors” prin-
ciple was not used in creating living brains. This led to the
desire to seek experimental confirmation of the nontrivial
predictions of the visceral hypothesis of sleep.
Most studies performed with the aim of confirming
this hypothesis have been published [6, 30, 32, 33, 35],
while some are conference presentations [7, 8, 31, 34, 38],
and articles with detailed descriptions will appear soon. The
present report provides a brief presentation and systemati-
zation of experimental results obtained with the aim of con-
firming this hypothesis. Most attention has been focused on
new concepts of the organization of information streams in
the brain during sleep and waking on the basis of these
experimental data. This structure of information streams is
expected to lead to an understanding of the mechanisms of
normal and pathological states linked to sleep.
1. EXPERIMENTAL CONFIRMATION OF THE
VISCERAL HYPOTHESIS OF SLEEP
The visceral hypothesis of sleep is based on the sugges-
tion that those neurons which analyze signals from exterocep-
tors during waking switch to analyzing interoceptive informa-
tion on the state of sleep. A natural approach to experimental
confirmation of this hypothesis is to compare the responses of
individual areas of the cerebral cortex to extero- and intero-
ceptive stimulation during sleep and waking.
1.1. Responses of the Sensory Zones of the Cerebral
Cortex to Electrical and Magnetic Stimulation of the
Digestive Organs during Sleep and Waking. The first
series of chronic experiments recorded activity from indi-
vidual neurons in the primary visual cortex in cats in condi-
tions of painless fixation of the head. In waking, these neu-
rons responded to visual stimulation and had classical
receptive fields. When the animals went to sleep, which was
monitored in terms of changes in the EEG pattern and eye
movements, they underwent transient intraperitoneal elec-
trical stimulation of the stomach or small intestine areas.
Stimulation parameters were selected such that stimuli
would not wake the sleeping animals. Our results showed
that electrical stimulation did not wake the cat, but, con-
versely, converted sleep to a deeper phase. Most simple and
complex cells in the primary visual cortex were found to
start responding to intraperitoneal stimulation in the state of
slow-wave sleep. These neurons ceased to respond to elec-
trical stimulation immediately after waking the cats and
returned to the production of visual responses . Similar
experiments were performed with neurons in somatosenso-
ry field 5 of the cat cerebral cortex using active movements
of the forelimbs. In slow-wave sleep, these animals showed
responses to electrical stimulation of the intestine and stom-
ach zones. Figure 1 shows a comparison of the responses of
neurons in the primary visual (A) and somatosensory (B)
cortex to visceral stimulation. During sleep, neurons in the
visual and somatosensory areas of the cortex can be seen to
generate clear responses to electrical stimulation. The
shapes of these responses were different and their latent
periods in the somatosensory cortex were significantly
shorter. Responses to visceral stimulation were absent dur-
ing waking and REM sleep.
Similar experiments were performed on monkeys with
chronic EEG recording of activity in the occipital area
above the visual cortex. Intraperitoneal stimulation in mon-
keys also evoked clear EEG responses during the slow-
wave sleep phase, which disappeared in REM sleep and on
Cortical evoked responses to magnetic stimulation of
the abdomen in monkeys, recorded over the occipital cor-
tex, behaved similarly . The reactions of neurons in
extrastriate visual zone V4 in response to the same stimula-
tion were somewhat different, and in most cases were
inhibitory. Short-latency phasic inhibition was followed
with delays of 10–20 sec by tonic excitation, clearly appar-
ent in some neurons, and still evident in the averaged popu-
lation responses of the neurons studied (Fig. 2). Similar
stimulation during waking never evoked visible changes in
the mean baseline activity of these neurons.
Intraperitoneal electrical stimulation was also used in
chronic experiments in rabbits  using recording of
evoked responses over the visual and somatosensory areas
of the cortex. All three rabbits showed responses to stimu-
lation applied during the slow-wave sleep phase. Visceral
stimulation-evoked responses over the somatosensory and
visual cortex had different patterns and numbers of compo-
nents (Fig. 3). The latent periods of responses, as in cats,
were significantly shorter over the somatosensory cortex.
During waking and REM sleep, there were no responses to
visceral stimulation in either area of the cortex in rabbits.
1.2. Link between Neuron Activity in the Visual
Areas of the Cortex and the Myoelectrical Activity of the
Stomach and Duodenum during Sleep. It is abundantly
clear that the experimental results described above have one
weak point. The electrical stimulation used is not a natural
element of physiological activity. Critics of the hypothesis
have noted that these effects may also not reflect real natu-
ral mechanisms but are some kind of “nonspecific” effects.
The Visceral Theory of Sleep 423
Fig. 1. Responses of visual zone V1 neurons (A) and somatosensory zone 5 neurons (5) in the cat cerebral cortex to electrical
intraperitoneal stimulation during slow-wave sleep and waking. Vertical lines show the moment of stimulation. Raster plots for
individual trials are shown above and averaged histograms are shown below; nis the number of trials.
We cannot completely accept this criticism, as a significant
proportion of our knowledge of the functional organization
of the nervous system has been obtained using electrical
stimulation methods. However, demonstration of a link
between the natural activity of the visceral system and the
sensory zones of the cerebral cortex during sleep has
become a significantly stronger argument in favor of the
visceral hypothesis of sleep. These experiments were per-
formed in collaboration with the late Professor V. A. Bagaev,
director of the Laboratory for Corticovisceral Physiology,
Pavlov Institute of Physiology and his colleague I. I. Busy-
gina. Recording electrodes were implanted into the smooth
muscle walls of the stomach and duodenum of cats, for
recording of the natural myoelectrical activity of these
organs in chronic experimental conditions. The natural
activity of the gastrointestinal tract organs could be com-
pared with the activity of cortical neurons and the overall
EEG during waking and slow-wave and REM sleep. During
uniform slow-wave sleep, the EEG was found to show
episodes of transient desynchronization coinciding with so-
called vagal myoelectrical complexes in the activity record-
ed from the stomach wall. These complexes, separated by
intervals of 50–140 sec, reflect contractions of the stomach
incompletely filled with food (Fig. 4).
Fig. 2. Mean population responses of 61 neurons in visual zone V4 of the monkey cerebral cortex in the state of slow-wave
sleep to magnetic stimulation of the abdomen. The vertical gray line shows the moment of stimulation. The horizontal dotted
line shows the baseline activity level before stimulation, taken as 100%; nis the number of averaged stimuli.
Fig. 3. Averaged evoked responses to intraperitoneal electrical stimulation during slow-wave sleep recorded over the
somatosensory (A) and visual (B) areas of the cerebral cortex in a rabbit. The vertical line shows the moment of stimu-
lation; nis the number of averaged trials.
Analysis of the spike activity of 202 visual cortex neu-
rons showed that 30% of neurons showed changes in spike
frequency at particular periods of slow-wave sleep, which
correlated with the rhythm of the myoelectrical activity of
the duodenum. Furthermore, individual cortical neurons
showed selectivity for rhythms of particular types (pure
rhythms or rhythms with spike potentials). The spike activ-
ity of one third of cortical neurons showed a significant
relationship with low-amplitude periodic changes in the
recorded gastric activity. In waking, none of the study neu-
rons displayed activity linked with gastric or intestinal myo-
electrical activity [7, 34].
1.3. Relationship between the Structure of the
Baseline Activity of Visual Cortex Neurons during Sleep
and Changes in the Composition of the Intragastric
Medium. These results led to the view that during sleep
periods, the composition of the intragastric medium might
be reflected both in the overall EEG and in the spike activ-
ity of individual cortical neurons. Confirmation of this
hypothesis was sought by creating fistulas in cats for deliv-
ery of different substances directly into the gastric lumen
during sleep. Warm water was given in the first experi-
ments. This affected mechanoreceptors in the stomach walls
and altered the acidity of the of the medium; this could alter
the afferentation arriving from chemoreceptors. However, it
would be surprising for a neutral substance – water – given
into the stomach to have direct effects on the activity of
neurons in the cortex. These experiments showed that
administration of water into the stomach during the slow-
wave sleep phase led to an immediate increase in the depth
of the animal’s sleep. The increase in the depth of sleep was
primarily reflected in relaxation of the sleeping animal’s
posture – the eyes were tightly closed, and the animal was
not woken by random sounds. Continuous slow-wave activ-
ity was recorded throughout periods with durations unusual
for daytime sleep in cats. With time, the behavioral pattern
deep sleep showed no change, though the overall EEG pat-
tern did alter significantly, with appearance of a so-called
intermediate sleep phase [22, 23]. Short periods of slow
waves alternated with desynchronization reminiscent of the
inclusions of REM sleep. This state could last up to 40 min.
Analysis of the baseline activity of cortical neurons during
periods of stable slow-wave activity after administration of
water, as compared with analogous periods before adminis-
tration, also showed significant rearrangements of spike
activity structure in most neurons studied, while mean spike
frequency was virtually unchanged .
The experiments described in sections 1.2 and 1.3
required special statistical analysis and large datasets,
which are difficult to obtain during the relatively short peri-
The Visceral Theory of Sleep 425
Fig. 4. Migrating myoelectrical complexes in the cat stomach (A) recorded simultaneously with the spectrogram of the cortical EEG during slow-wave sleep (B).
Light parts on the spectrogram correspond to greater spectral power.
ods of REM sleep. Thus, analysis is restricted to slow-wave
sleep and waking.
We note that in the 1950s, cortical responses to stimu-
lation of different visceral nerves were studied in the labo-
ratory of Bykov and Chernigovskii; the results of these
experiments and references to non-Russian studies have
been reviewed in [1, 4, 10]. These data were obtained in
acute experiments on anesthetized animals. However, when
chronic experiments without anesthesia became possible,
cortical responses to visceral stimulation could not be
reproduced. Neurons in these cortical zones in conscious
animals responded only to visual or somatosensory stimu-
lation; they are now regarded as associative zones for the
corresponding sensory modalities. Cortical responses to
visceral stimulation came to be regarded as probably an
artifact of anesthesia.
In natural conditions, without anesthesia, the cortical
zones have been shown to establish connections with the
visceral organs, though these connections are active only
Our experiments have demonstrated conduction of sig-
nals from the digestive system organs to the cortex during
slow-wave sleep. In REM sleep, conduction of these signals
was again terminated. On the other hand, periods of REM
sleep can be regarded as the deepest sleep, with the greatest
threshold of arousal and the greatest level of muscular ato-
nia. Within the framework of the visceral theory, we do not
see these two states as being fundamentally different, but
suggest that the cortex is occupied processing visceral infor-
mation during both slow-wave sleep and REM sleep. So
what defines the specific features of REM sleep, particular-
ly in the disappearance of slow-wave activity from the EEG?
It can be suggested that during a complete sleep cycle,
including slow-wave and REM phases, the brain performs a
sequential scan of all the body’s life systems. This process
starts with the digestive and respiratory organs and the heart,
which have clearly rhythmic functions. Interference with the
rhythmic activity of these systems also determines slow
waves in the cortical EEG. Scanning then progresses to the
organs without obvious rhythmicity, such as the liver, kid-
neys, vascular system, reproductive organs, muscles, and
tendons. Finally, the brain itself is a body organ, whose
physicochemical state also requires monitoring. The non-
rhythmic stream of afferentation from these organs also
leads to desynchronization of the EEG. Thus, sleep can be
regarded as a single process associated with the involvement
of the brain in analyzing all the body’s visceral systems. The
final section of this review will only address waking and
sleep, without dividing these states into individual phases.
Periodic substitution of cortical afferentation during
the sleep–waking cycle implies the need for modification of
existing views as to the organization of the major informa-
tion streams in the nervous system. The various aspects of
this theme will be evaluated in the second part of the pre-
2. SWITCHING OF INFORMATION STREAMS
IN THE NERVOUS SYSTEM DURING THE
The main information streams in the central nervous
system are shown schematically in Figs. 5–7. It is natural
that regardless of functional state, all conducting pathways
on the schemes remain in their places. However, these path-
ways are in some cases “open” for conducting signals.
These active pathways are indicated in black. Blocked path-
ways, where signal conduction is terminated in certain situ-
ations, are indicated with light gray arrows. Structures
blocking signal transmission are shown as circles, in which
two parallel lines show the state of the gating elements. If
the lines run parallel to the conducting pathways, then con-
duction is open; if they are perpendicular, then conduction
is closed. Each gating element also has a second input,
shown by the inclined thin line with a circle at the end. This
is the input for the control signal which commands a given
apparatus to transfer from the open state to the closed. In
other words, this control input regulates the threshold for
signal conduction. In the brain, the transition from the open
state to the closed can occur gradually, extended over time.
Structures with the properties of such gating elements are
well known in neurophysiology. These include, for exam-
ple, triadic synapses or synapses supporting presynaptic
inhibition. The details of this are of no great interest here.
The important point is that these mechanisms do in fact
exist and that their use in the scheme is not just imagination.
This is of course an extremely simplified scheme. Its
individual blocks are not associated with real brain struc-
tures. The single exception to this is the “cerebral cortex”
block. This is to a considerable extent arbitrary. The cortex
of the real brain is actually within this functional block. It is
very likely that other brain structures may also be in this sit-
uation. We will now ask the reader not to focus on the sim-
plification of the scheme, but on the main idea presented,
which reflects the results of the experiments as presented.
Thus, Fig. 5, Ashows the state of the conducting path-
ways of the nervous system in waking. Information on the
environment or the state of the animal’s body itself (the lower
left corner of the scheme), transformed into nerve spikes by
extero- and interoceptors, passes through open gating ele-
ments and is sent to the central “cerebral cortex” block for
analysis. The results of this analysis are sent to the block
driving behavior and motor activity. Output signals are
transmitted in parallel, to the “Consciousness” block, acti-
vation of whose neurons leads to suppression of perception
of the self and the environment. Many may feel justifiably
confused. Traditionally, consciousness is linked with activ-
ity in the cerebral cortex. However, results from sleep stud-
ies have definitively demonstrated that this is not so.
Consciousness is known to be active in waking and almost
completely absent in sleep and under anesthesia. At the
same time, the mean levels of activity in cortical neurons
are not significantly different in these states. It appears to
follow either that consciousness is not associated with the
activity of cerebral neurons or that the structures associated
with consciousness are in a different location. In the first of
these, the question of the locations of structures associated
with the processes of consciousness is entirely meaningless,
at least at the current level of knowledge of the brain. The
second possibility means that we can look for these struc-
tures. There is a simple experimental approach to answering
this question. If consciousness is active during waking and
inactive during sleep, the structures whose neurons behave
in concert with this need to be found. However, this is not
all. Structures which are candidates for the role of “sub-
strates for higher brain functions” should have a wide circle
of associative connections with the cortical zones leading
the analysis of all types of exteroceptive and proprioceptive
information. Such structures do in fact exist in the brain.
The structures of the basal ganglia have associative connec-
tions with all cortical zones . Furthermore, the excitato-
ry cortical projections to neurons in the main nucleus of the
basal ganglia, i.e., the caudate nucleus, are inactivated in the
state of sleep. As a result, caudate nucleus neurons decrease
their baseline activity during sleep, often to complete ces-
The Visceral Theory of Sleep 427
Surrounding and body environment
Autonomic nervous system
Surrounding and body environment
Autonomic nervous system
Fig. 5. Scheme showing information streams in animals’ bodies in waking (A) and during sleep (B). Black and gray lines
show active and blocked signal transmission channels, respectively. See text for explanation.
sation [27, 37]. Large reductions in the level of baseline
activity during sleep have also been demonstrated for
another structure of the basal ganglia – the globus pallidus
. Positron emission tomography  in humans has
shown that mean cortical activity does not change on the
transition from waking to sleep, while the level of activation
Ganglia of the autonomic
Autonomic nervous system
Autonomic nervous system
Fig. 6. Scheme showing proposed impairments of activation of information channels on the transition from waking
to sleep leading to hypnagogic hallucinations (A), restless legs syndrome (B), and dreams (C).
of the basal ganglia decreases sharply. This is only a small
proportion of the data casting doubt on the traditional views
of the position of the cerebral cortex in the hierarchy of
cerebral structures. However, detailed assessment of this
question is the subject of a separate review. At this point we
will simply note this structure for its connections and con-
sider the extent to which it provides an explanation for a
number of phenomena.
The right-hand side of the schemes presented here
show the animal’s body. During waking, information on the
state of the internal organs, converted into nerve spikes by
interoceptors, arrive in autonomic nervous system struc-
tures which at this time control the internal organs . The
segmental structure of the innervation of the internal organs
does not support the transmission of information about
some visceral systems to others. During waking, these sys-
tems effectively work under the local control of autonomic
nervous system structures.
However, with time, local autonomic nervous system
ganglia become unable to solve problems arising in the vis-
ceral domain independently. Interoceptors assessing the
operating parameters of visceral organs start to send signals
reflecting deviations in ongoing parameters from the genet-
ically specified norms. These signals are evidence perceived
by the animal as the feeling of tiredness, such that they start
to look for a safe and comfortable place to sleep. The details
of the process of the transition to sleep will be discussed
below. We will now address the state of sleep.
During sleep, there is a radical change in information
streams in the nervous system (Fig. 5, B). In the ideal case,
after becoming able to sleep, all gating elements switch
simultaneously to the opposite positions. During sleep, sig-
nal transmission from extero- and proprioceptors to the
cerebral cortex is blocked. However, the same input chan-
nels to the cortex start to carry information on the state of
the body’s visceral systems.
In all probability, the main structure in which
afferentation from extero- and proprioceptors is switched to
interoception is the thalamus. This may also be the main
functional load on the thalamus. Recent studies have shown
that at the level of the lateral geniculate body, the main tha-
lamic nucleus of the visual system, transmission of visual
information to the cortex is blocked [28, 41]. The synapses
of fibers running from the retina are known to make up one
third of all synapses on relay nuclei in the lateral geniculate
body. A second third consists of the synapses of collateral
projections from the visual cortex, whose functions remain
unknown. One third of the synapses is formed by fibers run-
ning from non-visual brain structures, particularly the pon-
tine nuclei . Activation of the pontine nuclei is known
to induce characteristic bursts of neuron activity in the visu-
al cortex during sleep. These are known as ponto-geniculo-
occipital spikes . The maximum numbers of these
spikes are seen during REM sleep. The pontine nuclei lie on
the paths of visceral information streams from the spinal
cord. However, what activation of these spikes is associated
with and what they reflect remain unclear.
In parallel with the shift in cortical afferentation from
extero- to interoceptive, it is natural to expect changes in the
efferent cortical projections. Output information streams
from the cortex during sleep reflect the results of the cortical
processing of visceral information and should not be
addressed to structures associated with motor activity, behav-
ior, or consciousness. On the scheme, the gating blocks on
these pathways are closed during sleep. The fact that con-
sciousness during sleep is detached from the surrounding
world is well known to people on the basis of personal expe-
rience. In addition, we know that at the anatomical-physio-
logical level, signal transmission from the cortex to spinal
cord motoneurons is known to be blocked during sleep. This
leads to relaxation of the body’s muscles, which reaches a
maximum during REM sleep . It has already been noted
above that the transmission of information from the cerebral
cortex to structures of the basal ganglia – an important asso-
ciative center of the brain – ceases during sleep.
At the same time, conduction via some new pathway
from the cerebral cortex to structures associated with pro-
cessing information from all the visceral systems must
open up during sleep. The hypothalamus can be identified
as a candidate for such an associative visceral regulatory
system. However, the efficiency of cortical-hypothalamic
connections during the sleep–waking cycle has not been
When all visceral parameters are normalized as a result
of the involvement of the cortex in visceral integration pro-
cesses, the imperative to sleep is lifted, all gating elements
switch to the opposite position, and the animal wakes.
We can thus see an ideal picture of the switching of
information streams in the sleep–waking cycle. However, it
should be noted that the main elements of this system are
devices blocking conduction via one pathway or another –
not electric relays, but chemical synapses. Their operating
efficiency may depend on large numbers of conditions
external to the synapse. Furthermore, the spike response of
the postsynaptic neuron is determined not only by the effi-
ciency of the control synapse, but also by the current thresh-
old of the neuron itself. Neuron thresholds are known to
depend on their previous activity. If a neuron has recently
been highly active, its excitation threshold will decrease
and, conversely, neurons which have been silent in the
recent past will have increased thresholds. When analyzing
this scheme, it must be borne in mind that real switching of
gating elements in some, generally pathological, conditions
may be extended over a significant period of time. The
resulting nonsynchronicity of the switching of information
streams may involve displacement of exteroceptive and
interoceptive information at the input to the cerebral cortex.
The output signals of the cerebral cortex may also be
addressed to incorrect targets. The next section will consid-
er some likely consequences of such addressing errors.
The Visceral Theory of Sleep 429
3. NONSYNCHRONIZATION OF THE SWITCHING
OF INFORMATION STREAMS IN THE BRAIN
AS A LIKELY CAUSE OF PATHOLOGICAL
PHENOMENA ASSOCIATED WITH
THE SLEEP–WAKING CYCLE
3.1. Hypnagogic Hallucinations. Hypnagogic halluci-
nations constitute a common but quite harmless pathology of
the process of transition from waking to sleep , These
hallucinations appear before falling asleep, at low illumina-
tion levels, and appear as various generally moving beings
(large “beetles”). This may occur when the pressure of sleep
opens the pathways for visceral afferentation to the cortex,
though the conduction of visual information to the cortex is
not blocked and the connection of the cortex with the behav-
ior and consciousness blocks are retained (Fig. 6, A). An
important condition is a low level of illumination. People
continue to perceive the context around them. However, the
intensity of visual afferentation is decreased in these condi-
tions. The intensity of the opened visceral signals stream is
comparable with that of the visual stream. Arriving in the
consciousness block, bursts of spike activity from the vis-
ceral inputs can induce excitation of visual gestalts superim-
posed on the real visual scene which is still being perceived.
It is also easy to explain the movement which is usually
reported in these hallucinations. The onset of sleep has been
shown to start in the visual cortex in the extrastriate zones
. This is how a movement analyzed in visual zone V5
appears. On going to sleep, visceral information will pri-
marily be directed here, while the output signals of this zone,
passing through the still open gating to the consciousness
block, will naturally evoke the feeling of movement. This
may also explain the hallucinatory movements not associat-
ed with the appearance of additional objects, which occurs
in conditions of increasing pressure to sleep. This produces
the sensation, for example, of wavelike movements of the
floor or deformation of particular objects.
Turning on the light also eliminates hypnagogic hallu-
cinations because of the increasing intensity of signals from
the real visual scene on whose background “visceral” com-
ponents become subthreshold, and also because of switch-
ing of the visceral inputs to the cortex due to the arousing
action of bright light.
3.2. Restless Legs Syndrome. The transition from
waking to sleep also produces another, more unpleasant, sit-
uation, reflected in Fig. 6, B. In this situation, the cortical
input to the consciousness block switches normally, while
switching of the output to structures associated with motor
activity, is delayed. When sleep develops, activation in the
motor areas of the cortex induced by what are now visceral
inputs is not blocked but is switched to spinal cord motoneu-
rons. Excitation of motoneurons leads to sudden limb move-
ments, which interrupts the developing sleep. This provides
a model of restless legs syndrome, which occurs quite com-
monly . There are now medications which are quite
effective in preventing the occurrence of such movements
during sleep. When sleep deepens, conduction of visceral
signals to the spinal cord is switched, such that there are no
further problems sleeping. “Extraneous” support is all that is
needed during the transition period.
Eye movements during sleep constitute a special case
of this situation. Eye movements in the REM phase of sleep
are well known. However, the eyes show smooth drift over
very large angles from the central position during slow-
wave sleep too, in some cases exceeding the normal ampli-
tude of movements in waking . As these movements are
slow, they are often not seen on the electrooculogram.
However, as information on eyeball position in the orbits
does not reach consciousness even during waking and the
eyes have an essentially perfectly spherical shape, their
movements beneath the closed lids do not interfere with the
onset of sleep. There are therefore no special mechanisms
for the active suppression of these movements during sleep.
3.3. Dreams. Dreams can be regarded as the common-
est and from time to time even comical “pathology” of sleep.
A possible cause of dreams is outlined in Fig. 6, C. The gat-
ing on the pathway from the cortex to the consciousness
block is not completely switched during sleep. The strongest
output signals reflecting the results of the processing of vis-
ceral information also erroneously enter the consciousness
block. However, the visceral systems are not represented in
our consciousness. Thus, for consciousness, these signals
are generally merely noise. Noise can excite those neurons
which have the lowest response thresholds, i.e., those neu-
rons which were operating the most actively during the pre-
ceding waking period. This phenomenon is reminiscent of
the stochastic resonance principle, when addition of noise to
a subthreshold signal has the result that it starts to be per-
ceived in a threshold system. This mechanism provides a
physiological basis for psychoanalytical approaches based
on analysis of reports of dreams. In fact, the subject of
dreams will in the first instance be items occupying the
greatest proportion of consciousness during waking.
It can also be suggested that at particular time points,
the spatial distribution of excitation running from the cortex
and reflecting the results of analysis of visceral information
is similar to the spatial profile of the gestalts of real objects
formed during waking. As a result, the image formed by
associative connections in the consciousness systems trig-
gers the development of fantastic plots.
Dreams are probably are manifestations of the transi-
tional period from waking to sleep or vice versa, when the
gating on the input to the consciousness block is not yet
completely closed or remains slightly ajar. Full closure
probably occurs on transition to the stationary sleep state.
Thus, this situation, although pathological, is harmless.
However, less pleasant cases do occur, when the gating to
the input to the consciousness block does not switch even
after long periods of time. This produces the situation of
persistent nocturnal nightmares which interrupt normal
sleep. There are also cases of real and profound pathology.
It should be noted that the old legend that dreams are exclu-
sively associated with the REM phase of sleep has not
received support. People also describe dreams on waking
from slow-wave sleep [15, 44]. As dreams can only be eval-
uated in terms of post-waking descriptions, it is very likely
that they also appear during these transition periods.
The Visceral Theory of Sleep 431
Ganglia of the autonomic
Autonomic nervous system
Autonomic nervous system
Fig. 7. Scheme showing d impairments of activation of information channels on the transition from sleep to waking
leading to sleep paralysis (A),somnambulism (B), and visceral hallucinations (C).
However, the question of the time of occurrence of dreams
cannot in principle be solved experimentally, and is thus not
a question for science. This position is given a grounding in
strict logic in Malcolm’s book The State of Sleep .
Previous sections have assessed the possible conse-
quences of impairments to the synchronicity of the switch-
ing of information streams during the transition from wak-
ing to sleep. We will now consider pathological situations
occurring during the transition from deep sleep to waking.
3.4. Sleep Paralysis. A not uncommon phenomenon
occurring on the transition from sleep to waking is so-called
sleep paralysis. People normally wake up, have appropriate
perception of the environment and their own body, but for
some period of time (from several seconds to several min-
utes) are unable to perform any kind of voluntary movement.
The likely cause of this phenomenon is simple, and is shown
in the scheme in Fig. 7, A. There is a delay in the release of
the blockade of the transmission line sending motor com-
mands from the cortex to the behavior and motor activity
block. This is the point at which consciousness has already
been restored but the motor control system remains asleep.
3.5. Somnambulism. Directly opposite cases are also
encountered, in which the pathway from the cortex to the
behavior and motor activity block opens and the pathway
to the consciousness block remains in the blocked state
(Fig. 7, B). This is the phenomenon of somnambulism –
which is quite common, especially in children. People get up
from bed at night and walk trajectories of different lengths.
The eyes are open while sleepwalking, people do not collide
with obstacles, and their movements are well coordinated.
At the end of the trajectory, the person often falls asleep
again, and on waking has no memory of getting to the new
location . Somnambulism is a further argument support-
ing the conclusion that the consciousness block is separate
from the cerebral cortex and structures programming body
movements. It is difficult to say anything definite about con-
nections between the visceral systems and the cerebral cor-
tex during sleepwalking. It is entirely probable that the trig-
ger for these episodes may consist of as yet undiscovered
visceral information streams, inducing dreams en route to
waking, which in turn provoke subsequent sleepwalking.
3.6. Visceral Hallucinations. The term “visceral hal-
lucinations” has not, as far as we know, been used previ-
ously, and the existence of this phenomenon can be regard-
ed as a prediction of the visceral theory of sleep. In the sec-
tion on hypnagogic hallucinations we discussed the effects
of perception evoked by mixed extero- and interoceptive
information at the input to the cortex, a situation which can
occur in the transitional periods between sleep and waking
when some conducting pathways have yet to close com-
pletely and others have not completely opened. As a result,
interoceptive signals are projected into real life situations
and are perceived as hallucinations. In theory, these phe-
nomena could also occur in the opposite direction. By anal-
ogy, we term these visceral hallucinations (Fig. 7, C). It can
be suggested that under the pressure to sleep, the pathways
connecting the visceral systems with the cerebral cortex
start to open. However, rapid transition to sleep may not be
possible under the pressure of the circumstances. The con-
nection between the cortex and the exteroceptive inputs and
outputs for consciousness and motor activity remain open.
The results of analysis of exteroceptive signals may reach
the visceral integration block and be assessed there as cor-
tical signals arising from analysis of the visceral inputs.
Control of the visceral systems will now occur by means of
these “hallucinatory” signals, which cannot produce any
kind of positive effect in the visceral domain.
The pernicious visceral consequences of acute stress
probably operate by this mechanism. The same mechanism
may also explain the occurrence of the visceral components
(nausea, vomiting) of motion sickness. Rhythmic rocking
and constantly reduced gravity, as in space flight, produce
sharp changes in the afferent spike stream arriving from
mechanoreceptors in the walls of the gastrointestinal tract
organs. The urgent need for analysis of the causes of these
unusual changes in afferentation generates pressure to go to
sleep. The pressure of sleep exposes the cortical output to
structures for visceral analysis and, possibly the input of
visceral information to the cortex. However, the high levels
of behavioral activity during this time prevent the transition
to sleep and the cortex remains connected with the power-
ful exteroceptive input stream. This powerful exteroceptive
signal stream arrives at the visceral information analysis
blocks, where it exacerbates the unusual nature of the
afferentation from the organs of the gastrointestinal tract.
The natural response to this will be urgent clearing of the
stomach to remove the sources of the “alarm” signals, i.e.,
As far as we know, the mechanism of these “visceral
hallucinations” has not previously been addressed. It seems
very likely that in a sleep-deprived society, the conse-
quences of visceral hallucinations, which are pernicious for
the visceral systems, may be large. Interference with extero-
and interoceptive information on the pathway to the cere-
bral cortex and at the outputs from the cortex may be the
cause of many psychosomatic illnesses.
4. THE SLEEP-TRIGGERING MECHANISM
The schemes presented in this report provide an under-
standing that the basic elements supporting the transition
from waking to sleep and vice versa are gating elements
opening or closing the conduction of extero- and interocep-
tive information in different directions. Changes in the
states of these gates occur in response to control commands
arriving from the sleep–waking centers. As these com-
mands are simple and uniform in type for all the neurons to
which they are addressed, it would appear that this involves
a limited number of neurons with branched projections over
the whole surface of the cerebral cortex. These are the prop-
erties of neurons in a number of identified “centers” regu-
lating the transition from waking to sleep and vice versa
. It is now important to establish how the commands for
these transitions are formed.
The first stage in this pathway must be performed in all
visceral systems. The task of this level is to make a simple
comparison between current values of parameters determin-
ing the ability of this system to function with their geneti-
cally determined standard values. This process is shown
schematically in Fig. 8, A. The block determining the need
for (or pressure to) sleep simply subtracts the standard val-
ues from the current visceral parameter values. The absolute
size of this difference will also determine the level of tired-
ness or pressure to sleep. All visceral systems must have
these elements for comparison. At the second stage of tak-
ing the decision to go to sleep (Fig. 8, B), tiredness signals
from different visceral systems are summed in a threshold
block designated on the schemes as “Neuron – threshold
element.” The output signal of the threshold element will
also be the “Command to sleep,” which converts the gating
elements from the “state for waking” to the “state for sleep.”
This scheme shows that the final output signal from the
threshold element is entirely determined by the summed
“tiredness” signals. These signals determine only what we
have repeatedly called the pressure to sleep. The second
important element determining the command to trigger
sleep is the threshold command signal. This signal accumu-
lates information on the current state of the body as an ele-
ment of the environment and on the state of the environment
itself, reflecting the current assessment of the body’s poten-
tial to go to sleep. In a safe environment with no other com-
peting needs, the threshold will be low and the transition to
sleep can occur with the first signs of tiredness. In condi-
tions of sleep deprivation, there can be a significant increase
in the threshold, such that even strong pressure to sleep may
be insufficient to overcome it.
However, this does not restrict the sleep control sys-
tem. There is at least one more important element deter-
mining the transition between sleep and waking – the circa-
dian rhythm. It can be suggested that in some situations, the
circadian rhythm operates as a signal directly controlled by
some of the gating elements described above.
Over recent decades, sleep physiologists have dis-
cussed the model of sleep regulation proposed by Borbély
. This model is presented in Borbély’s book The Secrets
of Sleep, which has been translated into Russian and is
available free of charge at www.sleep.ru. In this model, the
transition from waking to sleep is determined by the inter-
action between two processes: the homeostatic and the cir-
cadian. The homeostatic process maintains a particular
amount of sleep per day. The visceral theory of sleep gives
the “homeostatic process” physiological content. In fact,
here is a need to maintain homeostasis – not the abstract
state of sleep, but the functional state of all the body’s vis-
ceral systems, which occurs during sleep as a result of
switching of the cerebral cortex to processing information
regarding the state of these systems.
In summary, we can say that over the last 15 years, the
visceral theory of sleep has received much direct support in
experimental studies, which is unlikely to have happened
without this theory to drive experimental work. In addition,
the theory has proposed real physiological mechanisms for
basic phenomena associated with the sleep–waking cycle,
as well as novel mechanisms which may underlie psycho-
somatic illnesses. Numerous studies demonstrating rela-
tionships between sleep disorders and the occurrence of
pathological deviations in all visceral systems in the human
body  may represent an indirect argument supporting the
visceral theory of sleep.
The author would like to thank M. L. Pigareva,V. B. Do-
rokhov, and E. M. Rutskova for critical comments.
The Visceral Theory of Sleep 433
Current values of
visceral parameters (A) Reference values for
visceral parameters (B)
The intensity of pressure to (need for) sleep is
proportional to the absolute difference between
input signals A and B, ⎪A – B⎪
Threshold control signals
Neuron – threshold element
Fig. 8. Schemes explaining the first stage of the triggering of sleep, occurring independently in each visceral system and leading to pressure to sleep (A),
and the second stage, permitting sleep (B). See text for explanation.
This study was supported by the Russian Foundation
for Basic Research (Grant No. 10-04-00844).
1. K. M. Bykov, Selected Works, Vol. 2, The Cerebral Cortex and the
Internal Organs [in Russian], Gos. Izd. Med. Lit., Moscow (1954).
2. V. M. Koval’zon, Basic Somnology [in Russian], Binom. Lab. Znanii,
3. N. Malcolm, The State of Sleep [Russian translation], IG Progress-
Kultura, Moscow (1993).
4. S. S. Musyashchikova and V. N. Chernigovskii, Cortical and Sub-
cortical Representations of the Visceral Systems [in Russian],
Nauka, Leningrad (1973).
5. A. D. Nozdrachev, Autonomic Nervous System Physiology [in Rus-
sian], Meditsina, Leningrad (1983).
6. I. N. Pigarev, Brain and Sleep. Science in Russia [in Russian]
(2005), Vol. 1, pp. 61–65.
7. I. N. Pigarev, V. A. Bagaev, I. I. Busygina, and N. G. Bibikov, “The
state of sleep – the period of cortical control of visceral functions.
Mechanisms of regulation of the body’s physiological systems dur-
ing adaptation to the environment,” in: Proc. All-Russ. Conf. Celebr.
85th Anniv. Found. Pavlov Institute of Physiological, Russian Academy
of Sciences, St. Petersburg, Dec. 7–9, 2010, Institute of Physiology,
Russian Academy of Sciences (2010).
8. I. N. Pigarev, A. V. Bagaev, I. I. Busygina, et al., “Gastric and duo-
denal activity is reflected in cortical neuron discharges during sleep,”
Ros. Fiziol. Zh.,90, No. 8, 196 (2004).
9. I. N. Pigarev and M. L. Pigareva, “Sleep and the control of visceral
functions,” Ros. Fiziol. Zh.,97, No. 4, 374–387 (2011).
10. V. N. Chernigovskii, Interoceptors [in Russian], Medgiz, Moscow
11. T. J. Balkin, A. R. Braun, N. J. Wesensten, et al., “The process of
awakening: a PET study of regional brain activity patterns mediating
the re-establishment of alertness and consciousness,” Brain,125,
12. R. P. Barf, P. Meerlo, and A. J. W. Scheurink, “Chronic sleep distur-
bance impairs glucose homeostasis in rats,” Int. J. Endocrinol.
(2010), doi: 10.1155/2010/819414.
13. A. A. Borbély, “A two-process model of sleep regulation,” Hum.
Neurobiol.,1, 195–204 (1982).
14. D. C. Brooks and E. Bizzi, “Brain stem electrical activity during
deep sleep,” Arch. Ital. Biol.,101, 648–665 (1963).
15. C. Cavallero, P. Cicogna, V. Natale, et al., “Slow wave sleep dream-
ing,” Sleep,15, No. 6, 562–566 (1992).
16. C. Cirelli and G. Tononi, “Is sleep essential?” PLoS Biol.,6, No. 8,
17. M. H. Chase, “Confirmation of the consensus that glycinergic post-
synaptic inhibition is responsible for the atonia of REM sleep,”
Sleep,31, No. 11, 1487–1491 (2008).
18. S. Datta and R. R. MacLean, “Neurobiological mechanisms for the
regulation of mammalian sleep-wake behavior. Reinterpretation of
historical evidence and inclusion of contemporary cellular and molec-
ular evidence,” Neurosci. Biobehav. Res.,31, No. 5, 775–824 (2007).
19. L. Detary, G. Juhasz, and T. Kukorelly, “Neuronal firing in the pall-
idal region: firing pattern during a sleep-wakefulness cycle,” EEG
Clin. Neurophysiol.,67, No. 2, 159–167 (1987).
20. C. J. Farley, “Restless legs syndrome,” New Eng. J. Med.,348, No. 21,
21. C. A. Everson and A. Szabo, “Recurrent restriction of sleep and
inadequate recuperation induce both adaptive changes and patholog-
ical outcomes,” Am. J. Physiol. Regul. Integr. Comp. Physiol.,297,
22. C. Gottesman, “The transition from slow-wave sleep to paradoxical
sleep: evolving facts and concepts of the neurophysiological pro-
cesses underlying the intermediate stage of sleep,” Neurosci.
Biobehav. Rev.,20, No. 3, 367–387 (1996).
23. C. Gottesman, G. Gandolfo, C. Arnaud, and P. Gauthier, “The inter-
mediate stage and paradoxical sleep in the rat: influence of three
generations of hypnotics,” Eur. J. Neurosci.,10, 409–414 (1998).
24. H. C. Hughes and W. H. Mullikin, “Brainstem afferents to the lateral
geniculate nucleus of the cat,” Exp. Brain Res.,54, No. 2, 253–258
25. N. B. Kavey, J. Whyte, S. R. Resor, Jr., and S. Gidro-Frank,
“Somnambulism in adults,” Neurology,40, 749–752 (1990).
26. L. Keefer, E. J. Stepanski, Z. Ranjbaran, et al., “An initial report of
sleep disturbance in inactive inflammatory bowel disease,” J. Clin.
Sleep Med.,2, 409–416 (2006).
27. A. A. Loshkarev, E. I. Rodionova, E. V. Levichkina, and I. N. Pi-
garev, “Inactivation of the excitatory cortical projections to caudate
nucleus during sleep,” J. Sleep Res.,13, Supplement 1, 614 (2004).
28. L. M. Mukhametov and G. Rizzolatti, “The responses of lateral
geniculate neurons to flashes of light during the sleep–waking cycle,”
Arch. Ital. Biol.,108, No. 2, 325–347(1970).
29. W. C. Orr, “Esophageal function during sleep:Another danger in the
night,” Sleep Med.,8, 105–106 (2007).
30. I. N. Pigarev, “Neurons of visual cortex respond to visceral stimulation
during slow wave sleep,” Neuroscience,62, No. 4, 1237–1243 (1994).
31. I. N. Pigarev, I. Almirall, J. Marimon, and M. L. Pigareva, “Dynamic
pattern of the viscero-cortical projections during sleep. Study in
New Zealand rabbits,” J. Sleep Res.,13, Suppl. 1, 574 (2004).
32. I. N. Pigarev, I. Almirall, and M. L. Pigareva, “Cortical evoked
responses to magnetic stimulation of macaque’s abdominal wall in
sleep-wake cycle,” Acta Neurobiol. Exp.,68, No. 1, 91–96 (2008).
33. I. N. Pigarev, I. Almirall, M. L. Pigareva, et al., “Visceral signals
reach visual cortex during slow wave sleep. Study in monkeys,” Acta
Neurobiol. Exp.,66, No. 1, 69–73 (2006).
34. I. N. Pigarev, V. A. Bagaev, I. I. Busygina, et al., “Causal link
between myoelectrical intestinal activity and cortical neuronal firing
in sleep,” J. Sleep Res.,13, Supplement 1, 575 (2004).
35. I. N. Pigarev, G. O. Fedorov, E. V. Levichkina, et al., “Visually trig-
gered K-complexes. A study in New Zealand rabbits,” Exp. Brain
Res.,210, 131–142 (2011).
36. I. N. Pigarev, H.-C. Nothdurft, and S. Kastner, “Evidence for asyn-
chronous development of sleep in cortical areas,” Neuroreport,8,
No. 11, 2557–2560 (1997).
37. I. N. Pigarev and E. I. Rodionova, “Neuronal activity in caudate
nucleus in the sleep–waking cycle,” J. Sleep Res.,24A, 54 (1995).
38. I. N. Pigarev, M. L. Pigareva, L. P. Filaretova, and E. M. Rutzkova,
“Disturbances of normal sleep pattern as a possible reason of stress-
induced visceral dysfunction,” J. Sleep Res.,17, S1, 262 (2008).
39. S. Ranjbaran, L. Keefer, E. Stepanski, et al., “The relevance of sleep
abnormalities to chronic inflammatory conditions,” Inflamm. Res.,
40. A. Rechtshaffen and B. M. Bergmann, “Sleep deprivation in the rat:
An update of the 1989 paper,” Sleep,25, No. 1, 18–24 (2002).
41. W. Singer, “Control of thalamic transmission by corticofugal and
ascending reticular pathways in the visual system,” Physiol. Rev.,57,
No. 3, 386–416(1977).
42. A. Stocco, C. Lebiere, and J. R. Anderson, “Conditional routing of
information to the cortex: A model of the basal ganglia’s role in cog-
nitive coordination,” Psychol. Rev.,117, No. 2, 541–574 (2010).
43. C. Sturzenegger and C. L. Bassetti, “The clinical spectrum of nar-
colepsy with cataplexy: a reappraisal,” J. Sleep Res.,13, 395–406
44. H. Suzuki, M. Uchiyama, H. Tagaya, et al., “Dreaming during non-
rapid eye movement sleep in the absence of prior rapid eye move-
ment sleep,” Sleep,27, No. 8, 1486–1490 (2004).
45. Y. Tang, F. Preuss, F. W. Turek, et al., “Sleep deprivation worsens
inflammation and delays recovery in a mouse model of colitis,”
Sleep Med.,10, No. 6, 597–603 (2009).