Functional Neuroimaging of Normal Human Sleep by Positron Emission Tomography

J. Sleep Res.
(2000) 9, 207±231
Functional neuroimaging of normal human sleep

by positron emission tomography
P. MAQUET
Cyclotron Research Centre, University of LieÁge, LieÁge, Belgium
Accepted in revised form 23 April 2000; received 28 September 1999
SUMMARY Functional neuroimaging using positron emission tomography has recently yielded
original data on the functional neuroanatomy of human sleep. This paper attempts to
describe the possibilities and limitations of the technique and clarify its usefulness in
sleep research. A short overview of the methods of acquisition and statistical analysis
(statistical parametric mapping, SPM) is presented before the results of PET sleep
studies are reviewed. The discussion attempts to integrate the functional neuroimaging
data into the body of knowledge already acquired on sleep in animals and humans
using various other techniques (intracellular recordings, in situ neurophysiology,
lesional and pharmacological trials, scalp EEG recordings, behavioural or psycholo-
gical description). The published PET data describe a very reproducible functional
neuroanatomy in sleep. The core characteristics of this `canonical' sleep may be
summarized as follows. In slow-wave sleep, most deactivated areas are located in the
dorsal pons and mesencephalon, cerebellum, thalami, basal ganglia, basal forebrain/
hypothalamus, prefrontal cortex, anterior cingulate cortex, precuneus and in the mesial
aspect of the temporal lobe. During rapid-eye movement sleep, signi®cant activations
were found in the pontine tegmentum, thalamic nuclei, limbic areas (amygdaloid
complexes, hippocampal formation, anterior cingulate cortex) and in the posterior
cortices (temporo-occipital areas). In contrast, the dorso-lateral prefrontal cortex,
parietal cortex, as well as the posterior cingulate cortex and precuneus, were the least
active brain regions. These preliminary studies open up a whole ®eld in sleep research.
More detailed explorations of sleep in humans are now accessible to experimental
challenges using PET and other neuroimaging techniques. These new methods will
contribute to a better understanding of sleep functions.
Several neuroimaging techniques are now available to

INTRODUCTION
explore brain dynamics in humans: electroencephalography
Brain function may be investigated at various levels of (EEG), magnetoencephalography (MEG), near infrared spec-
description, from molecular interactions to the behaviour of troscopy (NIRS), single photon emission tomography
organisms (Churchland and Sejnowski 1992). Functional (SPECT), positron emission tomography (PET) and function-
neuroimaging studies describe cerebral function at the systems al magnetic resonance imaging (fMRI). Each of these
level, showing macroscopic interactions of brain areas engaged techniques is based on a particular signal and has its own
in sensory, motor or cognitive tasks (Frackowiak et al. 1997). advantages and disadvantages, in terms of spatial and
In the description of the cerebral processes, they act as an temporal resolution, safety and cost (Toga and Mazziotta
interesting link between the cellular level and psychological 1996). Most of them have now been successfully applied to
observation. As such, they are thus likely to provide an sleep studies and have provided valuable contributions to this
original description of brain function. ®eld. Coverage of all these data is beyond the scope of this
review. This paper is devoted mainly to recent advances made
by PET in the understanding of human sleep. Indeed, a wealth
Correspondence: P. Maquet, Cyclotron Research Centre, University of of data has recently been published in this ®eld. PET has
LieÁge, Department of Neurology, CHU Sart Tilman, LieÁge, Belgium. bene®ted from two technological advances: the iterative
Ó 2000 European Sleep Research Society 207

208 P. Maquet
H215O scan technique and statistical parametric mapping

Cerebral blood ¯ow as a marker of neuronal activity during sleep
(SPM). These methods have allowed the exploration of
regional cerebral haemodynamics with a good localizing Activation studies based on CBF measurements rely on the
power. It appeared timely to present an overview of the physiological observation that `¼the brain possesses an
methodological aspects of these techniques and the results intrinsic mechanism by which its vascular supply can be varied
obtained in sleep. locally in correspondence with local variations of functional
This paper is written for sleep researchers not involved in activity' (Roy and Sherrington 1890). In other words, CBF is
functional neuroimaging and consists of three main sections. used as a marker of neuronal activity.
In the methodology section, we present a short description of Here, we show that:
neuroimaging with PET, H215O scans and SPM. We also 1 The activity of local neuronal populations can be traced by
examine whether cerebral blood ¯ow (CBF) can still be regional CBF changes. This assumption is valid although
considered a marker of neuronal activity during sleep. These there is a transient uncoupling between oxygen metabolism
technical discussions expose the validity of PET neuroimaging and regional cerebral blood ¯ow (rCBF) during cerebral
in sleep as well as several technical points discussed recur- activation.
rently in the published PET sleep data. The second section 2 During sleep, this assumption remains also true.
reviews the results obtained in wakefulness, slow-wave sleep 3 The in¯uence of a potential neurogenic tone on rCBF,
(SWS) and rapid eye movement (REM) sleep. A short note although possible, is probably negligible. This is also true in
on dreaming processes is also proposed. The last section sleep.
presents future development of human sleep neuroimaging, in The precise mechanisms of the haemodynamic response to
particular with the advent of fMRI and multimodal coreg- local neuronal activity are not yet not known in every detail.
istrations. Post-synaptic potentials and action potentials cause ion move-
ment across neuronal membranes which, in turn, induce an
increase in metabolism of local neuronal and glial populations
METHODOLOGICAL ASPECTS (Magistretti and Pellerin 1996a,b). During this metabolic
response, several compounds, which mediate the vasodilatory
The H215O paradigm
haemodynamic response in the surrounding microcirculation,
Activation studies with PET usually imply that multiple CBF are produced. The principal vectors of this vascular response
measurements are made in the same subject under various are K+ ions (Paulson and Newmann 1987; Kuchinsky 1997),
conditions, and that several subjects are studied in the same H+ ions (Kuchinsky 1997), nitric oxide (NO; Northington
protocol. It is comparison of the blood ¯ow distributions et al. 1992; Dirnagl et al. 1993) and adenosine (Ko et al. 1990;
observed in the various conditions or their correlation with a Dirnagl et al. 1993), although peptides (Edvinsson 1985),
physiological parameter, that provides the localization of prostaglandins (PG; Leer et al. 1993) and kinines (Itakura
activated cerebral areas, suspected to be involved in the and Okuno 1993) have also been implicated in these processes.
sensory, motor or cognitive process explored by the experi- These events occur near the synapses and a number of
mental protocol. arguments suggest that regional CBF primarily re¯ects local
CBF is estimated by administering oxygen-15 labelled synaptic activity (Jueptner and Weiller 1995). On top of these
compounds (water or butanol). The very short half-life of local processes, CBF is also regulated by general factors such as
oxygen-15 (123 s) has several advantages. First, as radioactiv- perfusion pressure (autoregulation; Paulson et al. 1990), PCO2
ity becomes negligible after four or ®ve half-lives, the scans can and PO2 and possibly a neurogenic tone (Lou et al. 1987).
be repeated at 8±10-min intervals and multiple CBF measure- A local static factor such as capillary density is also function-
ments can be obtained within the same experimental session. ally related to rCBF: the higher the capillary density, the higher
Routine activation studies consist of four to 12 injections per the rCBF (Klein et al. 1986; Borowsky and Collins 1989;
subject. Second, H215O is distributed quickly in the body, in Kuchinsky 1997).
proportion to local blood ¯ow. CBF measurements can thus During cerebral activation, there is a transient uncoupling
be made on a timescale ranging from 40 s to 2 min. between the local metabolic rates for oxygen (CMRO2) and
If an arterial line is put into the radial artery, arterial blood rCBF. Although the time courses of these events are still being
samples can be obtained during scanning and quantitative debated, they can be summarized as follows. When a local
values of CBF can be computed, otherwise statistical analysis neuronal population is ®ring, the local need for oxygen
is based on `tomographic counts', i.e. the estimated number of increases abruptly. The content in deoxy-haemoglobin in the
disintegrations computed at each voxel. local vasculature increases quickly, during the ®rst hundreds of
Some activation studies are based on glucose metabolism. milliseconds after the activation. This was observed using both
This requires the use of [18F]¯uorodeoxyglucose. Fluorine-18 optical spectroscopy (Malonek and Grinvald 1996) and
has a longer half life (110 min). Consequently, this compound magnetic resonance imaging (MRI; Ernst and Hennig 1994).
is less suitable for functional neuroimaging studies, the After a very short time, the local CBF increases in response to
dierent scans being necessarily obtained in the same subject the neuronal activation. The rCBF increases out of proportion
on dierent days. with the local metabolic needs and the local content of
Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

PET neuroimaging of normal sleep 209
deoxy-haemoglobin actually decreases. As deoxy-haemoglobin forward to suggest regulation of CBF by direct in¯uences from
is paramagnetic, its decrease can be detected in MRI and this is neural structures. For some authors, the local vasodilatation
actually at the origin of the blood oxygen level dependent occurring in an activated area was too quick to be accounted
(BOLD) signal, largely used in functional MRI. Despite this for by metabolic factors and this suggested a direct neural
uncoupling between rCBF and oxygen metabolic rates, it in¯uence on cerebral vessels (Lou et al. 1987). However, since
should be borne in mind that the vascular changes still localize the hypothesis was proposed, several humoral compounds
precisely the cerebral regions at work. They thus can be used have been found which could explain this rapid vascular
reliably in functional neuroimaging. response: NO, PG, K+. Another argument, which was
Little data exist about the metabolic CBF regulation during thought to suggest neural control of brain vessels, is the
sleep but it is likely that in this condition, rCBF also remains a uncoupling between the increase in CBF and oxygen con-
marker of synaptic activity. The basic mechanisms of CBF sumption in an activated cerebral area (Lou et al. 1987).
regulation investigated to date persist during sleep. The However, although not completely understood, this phenom-
microcirculation does not seem to be modi®ed by sleep. As enon might, at least in part, rely on metabolic factors
during wakefulness (GoÈbel et al. 1990; Kuchinsky and Paulson (Magistretti and Pellerin 1996a, b).
1992; Villringer et al. 1994), all capillaries remain open during The most intriguing ®nding suggesting neurogenic control of
sleep in rats (Zoccoli et al. 1996). Likewise, it was recently local CBF is the presence of nerve ®bres in the vicinity of
shown in lambs that autoregulation persists during SWS and cerebral vessels (Iwayama et al. 1970; Edvinsson et al. 1972).
REM sleep (Grant et al. 1998). In sleep, as during wakeful- This innervation is both of extrinsic (ortho- and para-sympa-
ness, acute decreases in cerebral perfusion (by in¯ating a cu thetic) and intrinsic (intracerebral) origin (Pinard 1991).
around the brachiocephalic artery) cause a decrease in cerebral Several brain structures have been shown to have remote
vascular resistance in such a way that CBF returns to normal vasomotor actions: locus coeruleus (LC), raphe nucleus (RN),
control values after a few tens of seconds. Although the basal forebrain (BF), intralaminar thalamic nuclei, parabra-
response and its time course is quantitatively dierent in active chial nucleus, bulbar reticular formation, nucleus of tractus
sleep than in quiet sleep or in wakefulness, these results solitarius and fastigial nucleus of the cerebellum (Busija 1993).
provide clear evidence that the vasoactive mechanisms under- In theory, one could imagine a situation in which, because of
lying cerebral autoregulation persist in sleep. Finally, the the neurogenic tone, an increase in rCBF would occur without
available data seem to show the persistence of a close any change in local metabolism. Such a situation would lead to
relationship between local metabolic needs and rCBF. In the detection of spurious cerebral activation by functional
goats, global levels of CBF parallel cerebral oxygen consump- neuroimaging. However, it should be stressed that neurogenic
tion, although during REM sleep, the increase in oxygen CBF control was observed under very unnatural conditions,
consumption alone cannot account for the increase in CBF using stimulations or lesions formed by electrical or chemical
(Santiago et al. 1984). In humans, cerebral metabolic rates for means in anaesthetized animal preparations. Their role under
oxygen (CMRO2) and CBF also vary in parallel (Madsen et al. physiological circumstances is suspected but remains unclear.
1 1991b, c; Hoshi et al. 1994). The only discrepant data were Furthermore, possible neurogenic CBF control would alter
obtained by NIRS and suggest that at sleep onset and only in PET data under only three conditions. First, CBF modi®ca-
certain subjects, CBF might decrease while CMRO2 increases tions due to neural mechanisms should be systematic; they
in frontal areas (Hoshi et al. 1994). These results must be should occur in every subject under particular conditions (as it
treated with caution given the great interindividual, intra- is, stage of sleep), or at least in a vast majority. If this was not
individual and spatial variability of metabolic patterns during the case, i.e. if the neurogenic control is not systematic, group
sleep as well as possible artefacts arising from skin blood analysis would average them out. Second, CBF modi®cations
volume (Ogata et al. 1998; Yamamoto et al. 1998). due to neural control should be prolonged. In PET, the
At the regional level, at this point, it seems reasonable to cerebral activity is integrated over 40±120 s. Transient vaso-
suggest that, qualitatively, CBF regulation during sleep relies motor reactions are unlikely to be detected by PET. They
on the same mechanisms as during wakefulness. There might might, however, aect fMRI data. Third, CBF variations due
be quantitative dierences with waking CBF regulation. For to neural control should not be accompanied by modi®cations
example, in lambs, NO has a major in¯uence on CBF during in regional metabolism. Indeed, if local metabolism changed
sleep as well as during wakefulness and its eects seem simultaneously, the rCBF signal would still re¯ect modi®ca-
quantitatively dierent depending on sleep stage (Zoccoli et al. tions of neuronal activity. For these reasons, the eects of
1998). Likewise, local levels of adenosine are lower during neurogenic tone are usually considered negligible during
sleep than during wakefulness (Porkka-Heiskanen et al. 1997) wakefulness and are not a general concern for functional
and might modulate local blood ¯ow dierently in sleep. neuroimaging with PET.
Despite these possible quantitative changes, it remains that As far as sleep studies are concerned, some of the structures
modi®cations in local CBF re¯ect variations in neural activity. mentioned above are involved in both sleep and rCBF
Although metabolic needs are the major parameter modu- regulation. They both modify their activity throughout the
lating regional vascular supply, the role of neurogenic control sleep/waking cycle and, under the unnatural experimental
has been evoked recurrently. Various arguments have been put conditions described above, have some vasoactive actions. It is

210 P. Maquet
thus interesting to examine whether this can explain some thalamic nuclei. At present, there remains the possibility that
variations of regional cerebral blood ¯ow during sleep. In the frontal and cingulate activations may, in part, be explained by
following, we concentrate on four nuclei: locus coeruleus (LC), these vascular eects, at least in wakefulness and REM sleep.
rapheÂ nuclei (RN), basal forebrain (BF) and intralaminar Consequently, the available evidence does not lend much
thalamic nuclei. support to a signi®cant in¯uence of LC and RN on functional
The role of LC in CBF regulation under physiological neuroimaging data. There remains the possibility that BF and
conditions remains to be established. LC stimulation was intralaminar thalamic activity could modify cortical blood
shown to decrease CBF, together with cerebral metabolism ¯ow, especially in frontal regions. The relative importance of
(Raichle et al. 1975; Abraham et al. 1979; Goadsby et al. 1982; these eects, with regard to the CBF regulation by local
Buchweitz et al. 1985), whereas LC lesions by electrolysis or neurono-glial metabolism remains unknown.
6-hydroxydopamine have provided controversial results: an Finally, it should be stressed that, in the speci®c context of
increase in CBF (Bates et al. 1977) or the absence of any CBF sleep, neurogenic control of hypothalamic CBF was suspected
2 change (Dahlgren et al. 1981b; Reddy et al. 1986). in cats. At the beginning of REM sleep episodes, caudate
Results concerning the vasomotor in¯uence of RN are also temperature was found to increase while local CBF decreased.
controversial. Lesions of RN do not dramatically modify CBF After a lesion in the posterior hypothalamus, the decrease in
(Dahlgren et al. 1981; McBean et al. 1990; Underwood et al. CBF was suppressed, while temperature did not increase
1992; Kelly et al. 1995) or cerebral glucose metabolism further or even decreased. These data were interpreted as
(Cudennec et al. 1988; Pappius et al. 1988; McBean et al. re¯ecting the eects of neurogenic CBF regulation by the
1990; Kelly et al. 1995). Electrical or chemical stimulation of posterior hypothalamus (Denoyer et al. 1991).
RN does not have univocal eects on CBF: some authors
reported an increase in CBF (Goadsby et al. 1985a, b; Statistical parametric mapping
Bonvento et al. 1989; Cudennec et al. 1993), others observed
a decrease in CBF (Underwood et al. 1992). The stimulation Statistical parametric mapping is a statistical process which
site seems to be of critical importance: stimulation of ventro- tests hypotheses about regionally speci®c eects. The results
lateral RN decreases blood ¯ow while a CBF increase is are presented as statistical maps where voxel values re¯ect the
observed with more caudal stimulations (Underwood et al. probability of change in rCBF. Under the null hypothesis,
1992). In any case, the RN stimulation also causes glucose these voxel values are distributed according to a known
metabolism increases in various parts of the brain (Cohen et al. probability density function (e.g. Gaussian). Ùnlikely' voxel
1996): manipulations of the 5-HT system actually seem to values are interpreted as due to a signi®cant, regionally speci®c
modify the regression slope between CBF and CMRGlu. Under eect attributed to the sensory, motor or cognitive aspects of
these conditions, CBF remains a marker of neuronal activity. the experimental protocol.
Stimulations of BF increase the cortical acetylcholine level The most widely used software for statistical parametric
(Casamenti et al. 1986; Kurosawa et al. 1989). Both electrical mapping is itself known as SPM (statistical parametric
and chemical BF stimulations cause an ipsilateral increase in mapping). While SPM is now the standard method in
cortical blood ¯ow (Biesold et al. 1989; Lacombe et al. 1989; functional neuroimaging, it should be emphasized that other
Adachi et al. 1990). The eects predominate on frontal and methods do exist. They are usually based on similar premises
parietal cortices and are attenuated by muscarinic and and are not discussed here. A detailed account of statistical
nicotinic antagonists. The increase in CBF would not be parametric mapping can be found in Frackowiak et al. (1997).
paralleled by similar changes in cerebral metabolism (Hall- In this section, we summarize the main steps of the analysis of
stroÈm et al. 1990; Kimura et al. 1990). The eects of BF PET activation studies. We stress some points that have been
lesions are controversial. Some authors report a decrease in source of discussion in recent sleep studies.
both CBF and glucose metabolism (Iadecola et al. 1983; Orzi The characterization of brain mapping data essentially relies
et al. 1988; Kiyosawa et al. 1989; Gomi et al. 1991), others do on two basic principles of brain organization: segregation and
not observe any signi®cant change (Lamarca and Fibiger 1984; integration. Functional segregation pertains to the specializa-
Namba et al. 1991; Scremin et al. 1991). We cannot rule out tion of certain brain areas for some functions. For instance,
that, under conditions of high BF activity (wakefulness, REM Broca's and Wernicke's areas are known to participate in
sleep), an increase in frontal or parietal CBF would be partly language. In contrast, functional integration re¯ects how
due to the intervention of a BF vasomotor eect. dierent regions interact to mediate a function. A compre-
The electrical stimulation of centromedian and parafascicu- hensive understanding of brain function needs the exploration
lar nuclei of the intralaminar thalamus causes an increase in of both cerebral specialization and connectivity. Examples of
CBF but not in glucose metabolism (Mraovitch et al. 1986; each type of contribution in sleep are provided below.
Mraovitch and Seylaz 1987). The changes predominate over the
Methods
frontal and cingulate cortex, as well as the subthalamic area.
Further studies should complement these ®ndings: chemical In SPM, the data analysis proceeds in three steps: spatial
rather than electrical stimulation as well as lesion studies would transformation, statistical modelling and assessment of signi®-
con®rm or in®rm the vasomotor role of these intralaminar cance. In the following, each step is described brie¯y.

Spatial transformations. Before embarking on a statistical be used in hypothesis-led analysis but should be avoided in
comparison of a particular voxel under the dierent experi- exploratory investigations.
mental conditions, spatial transformations are necessary
3 (Friston et al. 1995a). Intuitively, one has to ensure that a Assessment of signi®cance. The ®nal step of statistical inferences
particular voxel represents the same brain area in all the scans takes into account the multiple comparisons that have been
of the same subjects and in the dierent experimental subjects. 4 made (Friston et al. 1991a, 1994). In a typical SPM, several
Three types of spatial transformation are usually considered: hundred thousand voxels are tested. Consequently, even at a
realignment, spatial normalization and coregistration. threshold of P 0.001, there should be several hundred voxels
Realignment of the various scans of the same subject corrects exceeding the threshold by chance only. A Bonferroni correction
for the subject's head movement during the experimental would obviously be too conservative and no signi®cant results
session (during night in sleep studies). These intramodal intra- could survive such a correction. Furthermore, a Bonferroni
individual spatial transformations are based on a `rigid body' correction would consider that voxels are independent while in
modi®cation of the images. These simple translations and fact, they are not; the image reconstruction process, and the
rotations bring all the scans from one subject into register. The smoothing performed during the analysis, render adjacent
second transformation consists of normalizing the individual voxels highly correlated. So a procedure was developed, based
data into a standard `stereotaxic' space. The most widely used on the theory of Gaussian random ®elds, which takes into
space is that described by Talairach and Tournoux (1988). This account the smoothing of the image and corrects the multiple
step is mandatory in the analysis of a group of subjects, comparisons on the basis of independent elements of the images,
because of individual dierences in brain size and shape. This the RESELs (resolution elements). Use of the random ®elds
interindividual intramodal spatial transformation implies not theory puts some constraints on the SPM structure, which
only translations and rotations, but also other linear (scaling, should be a good representation of a continuous stationary
sheering) and nonlinear deformations. These image transfor- Gaussian random ®eld. These results are usually reported as
mations are not intended to obtain a point-to-point matching signi®cant at `corrected' P-levels (P < 0.05).
of individual brains to a standard brain but allow group
analysis and facilitate the communication between laborat-
Application of SPM in sleep
ories. However, this standard space has several limitations.
Talairach's atlas is based on a single hemisphere and does not The use of SPM in sleep studies highlights three issues: head
take into account the interhemispheric asymmetries, or the movements during sleep, quanti®cation of CBF values and the
known interindividual variability of brain anatomy. Further- choice of contrasts.
more, the localization proposed by Talairach's atlas in certain
regions (vermis, medial aspects of occipital and temporal Head movements. Head movements are not a greater problem
lobes, etc.) may not match exactly the individual anatomy (as in sleep studies than in studies conducted during wakefulness.
estimated by the subject's MRI data). Another spatial trans- However, the experimental session in sleep studies is inevitably
formation consists in coregistering the subject's PET data and longer than in usual daytime protocols, since the subjects have
his/her structural MRI scan. This (intra-individual intermodal) to stay »8 h in the scanner. The longer the session, the higher
step increases the localizing power of the analysis. Finally, the probability of head movements in the scanner. Although
images are smoothed by a Gaussian kernel in order to increase head movements are reduced by sleep per se, some movements
the signal-to-noise ratio. may be observed at sleep onset or awakening. If these
movements are large, attenuation correction is no longer
correct. [Attenuation of photons emitted by the object (i.e. the
Statistical modelling. The second step of the SPM analysis subject's head) causes an underestimation of the true local
models the data at each and every voxel, using the general isotope concentration, especially in the areas situated deep in
linear model (Friston et al. 1995b), in terms of the estimates of the object. Correction for this attenuation is usually based on a
the model parameters. It partitions the measured responses measure of the density of the head structures by a `transmis-
into various components, including the eects of interest and sion' scan. In this scan, rod sources ®lled with a positron
those of confounding factors. One of these confounding emitter (usually germanium) turn around the object.] Of
factors is the global ¯ow (discussed later). The dierence in course, the correction of the attenuation is valid only if
parameter estimates (e.g. the treatment eects) are assessed transmission and functional (èmission') scans are in register.
using linear contrasts, leading to a t or a Z statistic. In the Some authors correct for movement between transmission and
most simple case, a contrast is a vector as [)1 1], which simply emission scans (Andersson et al. 1998). Furthermore, head
subtracts the mean observed during the ®rst condition from movements make heavy corrections necessary during realign-
the mean during the second one. This step provides a ®rst set ement. As a rule of thumb, we discard data from subjects who
of signi®cant voxels, before correction for multiple compar- moved more than the original voxel size (i.e. »4 mm).
isons is applied. The probability level usually used with this
®rst step has been empirically chosen at P £ 0.001 (uncorrected), CBF quanti®cation. A debated question is whether CBF
which corresponds to Z ³ 3.09. This ®rst level of inference may should be quanti®ed during PET sleep studies. There is no

212 P. Maquet
univocal answer to this question and the methodology should with statistical parametric mapping. First, raw images seem
be adapted to each experimental situation. PET studies may particularly suited to SPM. They are created by convolutions
answer two main questions: (i) Àre there global CBF of Poisson distributions and by the central limit theorem, voxel
modi®cations across states of vigilance (W, SWS, REMS)?'; values are distributed according to a normal distribution.
and (ii) `Within a speci®c state, what is the regional distribu- Second, it is often argued that variations in PCO2 occur during
tion of brain activity (i.e. CBF)?'. sleep (Sullivan 1980) and that this may change CBF and SPM
There is no doubt that to answer the ®rst question requires results. However this possibility seems unlikely, PCO2 changes
CBF quanti®cation. Global CBF modi®cations are reported do not seem to modify the regional CBF distribution (Ramsay
by many authors, using various techniques and in dierent et al. 1993). Consequently, it is unlikely that PCO2 modi®ca-
species. Detailed reviews cover these data and summarize their tions during sleep could bias the SPM results. Third, raw data
results (Madsen and Vorstrup 1991; Franzini 1992). Brie¯y, images are less prone to condition-dependent global activity
these data show that CBF in REMS is systematically higher changes. Global ¯ow is a confounding factor that masks the
than in SWS (Goas et al. 1969; Lecas and Bloch 1969; regional CBF variations for which SPM is looking. SPM
Tachibana 1969; Seylaz et al. 1975; Dufour and Court 1977; adjusts the regional CBF for possible modi®cations in global
Milligan 1979; Mukhtar et al. 1982; Richardson et al. 1985, CBF. This can be done either by proportional scaling or by an
1994; Lenzi et al. 1987; Meyer et al. 1987; Parisi et al. 1988; ANCOVA (global CBF being the confounding covariate;
Abrams et al. 1990; Zoccoli et al. 1993, 1994; Gerashchenko Friston et al. 1990). The adjustments may cause some
and Matsumura 1996), similar or slightly elevated as compared trouble if global CBF changes do actually depend on
to W (Kanzow et al. 1962; Reivich et al. 1968; Townsend et al. experimental conditions.
1973; Shapiro and Rosendor 1975; Sakai et al. 1980; Santi- Indeed, if the adjustment is made outside the range of
ago et al. 1984; Lenzi et al. 1986, 1987; Parisi et al. 1988; Cote observed values (Fig. 1), one can falsely detect a signi®cant
and Haddad 1990; Madsen et al. 1991a). Various results are activation (deactivation) in the case of decreased (increased)
observed during SWS: compared with W, decreases (Town- global ¯ow. Under the same conditions, one may miss a true
send et al. 1973; Sakai et al. 1980; Greisen et al. 1985; Meyer activation (deactivation) if global CBF increases (decreases).
et al. 1987; Madsen et al. 1991a, b; Gerashchenko and In general, one should be cautious in discussing signi®cant
Matsumura 1996), increases (Reivich et al. 1968; Goas et al. activation when there may be a global ¯ow decrease (e.g. SWS)
1969; Seylaz et al. 1975; Shapiro and Rosendor 1975) or no or signi®cant deactivation in conditions with high global CBF
change in CBF (Mangold et al. 1955; Santiago et al. 1984; (e.g. REM sleep). Because raw data voxel values depend not
Lenzi et al. 1986, 1987; Parisi et al. 1988; Cote and Haddad only on CBF, but also on PCO2 and on the injected dose, their
1990; Zoccoli et al. 1994) are reported. The reasons for these variance is usually greater but they are less dependent on the
discrepancies remain unclear. The diversity of species investi- vigilance state. In this sense, they are more appropriate than
gated and the techniques used are certainly part of the answer. quantitative CBF images in which global ¯ows are likely to
However, other factors are less understood, such as the depend on the vigilance state. As a rule of thumb, we check
intriguing possibility of sleep-independent across-night CBF scan total counts to ensure the absence of condition-dependent
decreases (Hajak et al. 1994) or the possible confounding eect global ¯ow changes before proceeding to the SPM analysis. In
of circadian rhythms, which may also be involved. A better sum, to describe global changes in CBF, quantitative CBF
knowledge of CBF regulation during sleep is certainly needed measurements are mandatory. To determine rCBF distribu-
at this stage. The measures of global CBF would certainly help tion, both quantitative and qualitative CBF measurements
in this endeavour. may be included in the SPM. In this context, quantitative CBF
In contrast, when the regional distribution of cerebral blood values are more dicult to obtain and, in theory, more delicate
¯ow is considered, there seem to be several advantages to the to use in the SPM.
use of raw images (in `tomographic counts') rather than
quantitative CBF images. The main drawback of quantitative The contrasts. Statistical maps cannot be interpreted without
CBF studies is that arterial sampling is needed. Usually, taking into account the contrasts used to create them. It should
subjects are catheterized in one arm for venous infusions and be clear that no one contrast is right or wrong. Each answers a
in the other with the arterial line. This position is particularly slightly dierent question. Thus, it is necessary bear in mind
uncomfortable, especially after several hours of immobility. the contrast used in order to interpret each statistical map. In
This severely limits the possibility of observations during a the discussion of the sleep results (see below), the reader should
whole night. In contrast, when limiting the analysis to raw refer to the tables in which the contrasts are speci®ed for each
images, no arterial line is needed and sleep stability is greatly study.
improved: up to 12 scans may be regularly obtained per subject In the case in which one wants to compare the CBF
and per night. This increases the statistical power, by impro- distribution in various states of vigilance (SWS, REM sleep
ving the estimation of variance and the assessment of and wakefulness, for instance), the analysis relies on categor-
signi®cance [see above and (Frackowiak et al. 1997)]. ical comparisons. The most direct contrast that comes to mind
There are also some theoretical considerations that make is to compare one particular sleep state with wakefulness.
quantitative CBF data more delicate than raw images for use However, it should be borne in mind that wakefulness has its

Figure 1. Eects of condition-dependent

global ¯ow changes on the estimation of
regional activity by ANCOVA. Cerebral
blood ¯ow (CBF) observed during the
control condition (e.g. rest) is given in
green; the experimental condition, CBF
observed in a brain region, the activity
of which is not altered by the task is
given in blue; the experimental condi-
tion, CBF observed in a brain region,
the activity of which is modi®ed by the
task is given in red. (a) A simple situ-
ation in which there is no modi®cation
of global CBF. The regional increase in
rCBF is measured on the axis of the
ordinates (red arrow) (Friston et al.
1990). (b) Eects of increased global
CBF in the experimental condition.
In addition to a genuine local increase
in rCBF (red regression), there is an
increase in global ¯ow (rightward shift).
This causes an underestimation of the
activation (red arrow). In contrast, a
region that is not particularly activated
but in which the rCBF follows the
increase in global CBF (blue regression)
will appear as signi®cantly (and errone-
ously) deactivated (blue arrow).
(c) Eects of decreased global CBF in
the experimenatl condition. In addition
to a genuine local decrease in rCBF (red
regression), there is a decrease in global
¯ow (leftward shift). This causes an
underestimation of the deactivation (red
arrow). In contrast, a region that is not
particularly deactivated but where the
rCBF follows the decrease in global
CBF (blue regression) will appear as
signi®cantly (and erroneously) activated
(blue arrow).

214 P. Maquet
own characteristic CBF distribution (see the following section their behaviour to the experimental situation. These processes
on wakefulness), which will be subtracted from the sleep are known to rely on polymodal associative frontal and
haemodynamic pattern. parietal cortices, which are signi®cantly activated in this
Another possibility is to try to extract the characteristic CBF condition. Accordingly, as shown in Fig. 2, the most active
distribution of each state, building on the fact that brain brain areas in this condition are located in the prefrontal,
organizations in SWS, REM sleep and W are so dierent that anterior cingulate, parietal cortices and in the precuneus.
these states may be considered as orthogonal, as exempli®ed Similar results concerning the resting condition have been
by Jones (1991) (Fig. 2). In this perspective, each state is obtained in very dierent experimental situations. Andreasen
compared with the rest. For instance, REMS ± (W + SWS)/2 and co-workers observed that cerebral regions more active in
provides the brain areas which are speci®cally activated in the resting state than in memory tasks are located in the
REM sleep. This type of contrast leads to more restricted prefrontal and parietal cortex, in good agreement with our
rCBF changes but these are more speci®c, i.e. characteristic, of waking CBF distribution (Andreasen et al. 1995). Re-
each state. analysing PET studies of visual information processing,
Finally, a more elegant way to analyse PET data is to look Shulman et al. (1997) showed that the most active areas during
for signi®cant regressions between CBF distribution and a the resting state were in the posterior cingulate/precuneus,
physiological parameter in a speci®c condition such as delta inferior parietal cortices, left dorso- and ventrolateral prefron-
power density in SWS. In this case, regional CBF modi®ca- tal cortex and medial frontal areas. These brain areas are in
tions are speci®c to the speci®ed parameter. very good agreement with those we observed. They report,
however, an important activity in the right amygdala, which we
did not observe, possibly because this structure is even more
`CANONICAL' SLEEP AND WAKING
active in REM sleep (see below). Although a redistribution of
BRAIN MAPS
CBF by task performance cannot be ruled out, this CBF
In this section, we describe the brain areas in which local distribution was proposed to be due to monitoring by the
activity is maximally and speci®cally modi®ed during each subjects of the environment and internal state. As mentioned
state of consciousness: wakefulness, SWS and REM sleep. previously, we must also consider the possible vasomotor
Some results are interpreted easily in view of the mechanisms action of basal forebrain on these frontal and parietal areas.
of sleep generation described at the cellular level in animals. These results in wakefulness should be kept in mind in the
For other results, cellular correlates of CBF modi®cations are interpretation of sleep data. When the waking state serves as
not yet known completely. In this case, we try to ®nd some the control for a sleep condition, parieto-frontal hyperactivity
support for a link between sleep processes and a brain area at a is subtracted from the sleep distribution.
higher level of description (EEG, psychological eects of sleep
or sleep deprivation). The latter situation inevitably leads to a
Slow-wave sleep
more controversial discussion but may also induce new
working hypotheses. During SWS, large neuronal populations oscillate synchro-
In the following paragraphs, we consider published results nously within certain frequency bands (Steriade et al. 1990a,
that were at least signi®cant at the level P < 0.001 uncorrected 5 1993c, d; Steriade and Amzica 1998). As sleep deepens,
(corresponding to a Z score of at least 3.09). spindles then delta rhythms appear, both sculptured by a slow
(<1 Hz) oscillation (Steriade et al. 1993b, d). These rhythms
have been described in animals at the cellular level but EEG
Wakefulness
data suggest their existence in humans (Achermann and
During wakefulness, the rCBF distribution is of course highly Borbely 1997; Amzica and Steriade 1997). At the cellular
variable and depends on the cognitive tasks in which the level, these rhythms alternate short bursting periods with long
subject is engaged. This is the foundation of functional silent lapses of hyperpolarization. To understand PET results,
neuroimaging in cognitive neuroscience. In the particular case it should be borne in mind that PET averages these cellular
of sleep studies, subjects were scanned while lying still in the processes over time and space.
dark, with their eyes closed. This condition has long been used Over time, brain haemodynamics are integrated for 40±
as the `control' resting condition. It is now admitted that this 120 s. During this time, neurones discharge in a bursting
situation is particularly uncontrolled: the subject may be in all mode. It turns out that the in¯uence on CBF of the
sorts of mental conditions, including boredom, anxiety, mental hyperpolarization periods outweighs that of the short bursts;
imagery, memory retrieval, etc. These activities may vary from the hallmark of SWS is a decrease in regional CBF. This
one subject to the other. However, the resting condition is
characterized by at least two characteristics: (i) the constant
recording of a rhythm on posterior leads by electroencepha- Figure 2. Surface rendering of brain areas more active in wakefulness
than in sleep (slow-wave or rapid eye movement sleep). The data
lography; and (ii) a set of cognitive processes common to all concern the 11 subjects reported in Maquet et al. (1996, 1997) and
subjects, which include attending and interpreting external were analysed with SPM 95. Displayed voxels are signi®cant at
stimuli, holding the material in memory as well as adapting P < 0.05 after correction for multiple comparisons.


216 P. Maquet
37 Table 1 Decreases in rCBF during SWS
38 Maquet et al. Braun et al. Andersson et al. Kajimura et al. Ho¯e et al.
(1997) (1997) (1998) (1999) (1997)
Quanti®cation ) + + + )
Global activity + + + ) + +
Adjustment
Brainstem ¯ ¯ ¯ ¯ ¯
Thalami ¯ ¯ ¯ ¯ ¯ ¯
Basal forebrain ¯ ¯ ¯
Basal ganglia ¯ ¯ ¯ ¯
Prefrontal cortex Orbital ¯ DLPFC ¯ DLPFC ¯ ¯ Orbital ¯
Orbital ¯ Orbital ¯
Parietal cortex ¯ ¯ ¯ ¯ ¯
Precunes/posterior ¯ ¯ ¯ ¯ ¯
cingulate cortex
Anterior cingulate ¯ ¯ ¯ ¯ ¯
cortex
Mesio-temporal ¯ ¯ ¯
cortex
Insula-temporal cortex ¯ ¯
Cerebellum ¯ ¯ ¯ ¯
Remarks CBF CBF CBF CBF CBF CBF
[(W + REMS)/2] W ) SWS W ) SWS W ) SWS* W ) SWS
) SWS
+, CBF quanti®cation and global activity adjustment have been performed; ), CBF quanti®cation and global activity adjustment have not been
performed; DLPFC, dorso-lateral prefrontal cortex.
*Actually, all neorortical areas decrease except the periolandic and occipital cortex.

Regression with delta power density.
reasoning applies for modi®cations in local CBF. Global CBF tively (Steriade et al. 1993b, d). Similar proportions (82 and
variations in SWS also depend on general factors, as described 100%, in reticular and thalamocortical cells, respectively) have
earlier. In consequence, these periods of hyperpolarization been reported for delta rhythm (Curro Dossi et al. 1992).
represent the main cellular event that explains the neuroimag- These ®gures are usually lower in natural sleep. It seems
ing data during SWS. reasonable to suggest that the higher the proportion of local
The degree of deactivation of a particular brain area (in neurones engaged in sleep oscillations, the lower the local
terms of rCBF) is dependent on the local mechanisms of CBF.
generation of the sleep rhythm. In the thalamus, active In sum, taking into account regionally speci®c cellular
inhibition by GABA, as well as intrinsic membrane properties mechanisms, brain areas with a high proportion of neurones
of thalamic neurones, plays an important role in the mainten- committed in abundant synchronous sleep oscillations are
ance and modulation of sleep rhythms (Steriade et al. 1993a). likely to have the lowest regional CBF. These areas are
In contrast, in the cortex, sleep rhythm is related to disfacili- described in the following paragraphs as showing a maximal
tation of pyramidal neurones and interneurones (Contreras CBF decrease, having a negative association with SWS or
et al. 1996). These dierent mechanisms possibly require being deactivated. This does not imply that these regions `need'
dierent metabolic demand and may explain why thalamic more sleep. However, it seems tempting to suggest that, if
nuclei are the among the most deactivated areas during SWS. synchronous oscillations of SWS have a favourable eect on
Likewise, within the cortex itself, some brain areas might be neuronal function (Steriade et al. 1993c), the deactivated areas
more deactivated than others (see below), depending on the in particular bene®t from these sleep processes. Likewise, the
local modulation of cellular processes underlying sleep deactivated areas are not necessarily actively involved in sleep
rhythms. At present, the cellular bases of such a regional generation or maintenance, although it turns out that the
intracortical modulation are not completely understood. cellular processes observed in many of these regions show that
Moreover, the proportion of the local neuronal population this is actually the case.
engaged in the sleep oscillations may also explain the Five studies (Braun et al. 1997; Ho¯e et al. 1997; Maquet
distribution of rCBF. Because of its limited spatial resolution, et al. 1997; Andersson et al. 1998; Kajimura et al. 1999)
PET also averages the local activity over a whole population reported on regional CBF distribution in SWS (Table 1 and
of neurones. Only a fraction of this population adopts a Fig. 3). The functional neuroanatomy of SWS is characterized
bursting pattern. For example, in anaesthetized cats, 88, 65 by a low CBF in central core structures (brainstem, thalamus,
and 44% of neurones were engaged in slow oscillations in the basal forebrain), in basal ganglia and cerebellum, as well as in
cortex, the reticular thalamus or the dorsal thalamus, respec- some cortical areas (frontal, parietal, mesiotemporal).

Brainstem deactivations were described in the pontine and stimulation elicits a shift of spike discharge pattern of cortical
mesencephalic tegmentum. This brain area encompasses many neurones from phasic to tonic and, consequently leads to EEG
nuclear groups among which cholinergic and aminergic activation (Metherate et al. 1992). By the same token, the unit
reticular nuclei and the mesencephalic reticular formation. In activity of cortically projecting basal forebrain neurones (most
the currently held theory of sleep generation, the activity of the of which are cholinergic), increase their discharge rate during
mesencephalic reticular formation and cholinergic reticular EEG activation (Detari and Vanderwolf 1987). On the other
nuclei decreases at sleep onset, leading to a disfacilitation of hand, BF/AH is also important for sleep generation. BF/AH
thalamic nuclei (Steriade 1993a). The activity of these brain- lesions with excitatory aminoacids induce a prolonged insom-
stem structures must remain low during sleep because resump- 7 nia (Szymusiak and McGinty 1986a) while its electrical
tion of their tonic activity would immediately cause an arousal stimulation produces sleep (Siegel and Wang 1974). This
reaction (Steriade 1993a; McCormick and Bal 1997). Aminer- region contains cells that increase their ®ring at sleep onset
gic (serotonergic and catecholaminergic) nuclei also decrease (Szymusiak and McGinty 1986b; Sherin et al. 1996). Despite
their ®ring rate during slow sleep (Steriade and McCarley the functional importance of the latter cellular group, these
1990b). Consequently, the low CBF in the brainstem re¯ects cells represent only a minority of BF/AH neurones (37%;
their permissive role on the generation of sleep rhythms. Nunez 1996). Most of BF/AH neurones increase their ®ring
Thalamic deactivation is the most reproducible pattern rate during wakefulness and REM sleep and adopt a bursting
observed by neuroimaging techniques in human sleep (stage 2 mode of ®ring during SWS (Khateb et al. 1992). This bursting
sleep and SWS). It has already been suggested by early glucose pattern would seem to be related to intrinsic membrane
metabolism studies (Buchsbaum et al. 1989; Maquet et al. properties, notably a low threshold calcium current (Khateb
1990, 1992) and was con®rmed by recent CBF studies. In stage et al. 1992). That these ®ndings, obtained in rats may be
2 sleep, Ho¯e et al. (1997) showed that the medial thalamic transposed to humans is suggested only by functional neuro-
blood ¯ow was inversely correlated with the power density imaging. The low BF/AH blood ¯ow might re¯ect that in
within the spindle frequency band (12±15 Hz). In deep SWS, humans also, the majority of BF/AH neurones discharge in
all H152 O studies reported important CBF decreases in the bursts during SWS.
thalamic nuclei (Braun et al. 1997; Ho¯e et al. 1997; Maquet Basal ganglia, especially the striatum, were not expected to
et al. 1997; Andersson et al. 1998). Although the bulk of be among the maximally deactivated brain areas during SWS
thalamic nuclei seem deactivated when categorical compar- (Braun et al. 1997; Maquet et al. 1997). Indeed, they are not
isons are used, CBF is negatively correlated with delta power usually considered to participate in the networks of SWS
density (1.5±4 Hz) essentially in the medial thalamus (Ho¯e generation in animals. For these reasons, we cautiously
et al. 1997). suggested that these deactivations might re¯ect the eects of
In animals, thalamic neurones, hyperpolarized because of frontal and thalamic activity on the basal ganglia (Maquet
the disfacilitation described above, change their ®ring mode et al. 1997). Frontal cortex (Selemon and Goldman-Rakic
6 from tonic to phasic (Steriade et al. 1990b, c). First, GAB- 1985) and intralaminar thalamic nuclei (Macchi et al. 1984;
Aergic reticular thalamic nucleus bursts in the spindle Sadikot et al. 1990) represent the major sources of aerents to
frequency range and entrains thalamo-cortical neurones in basal ganglia. They are also among the most deactivated brain
spindle oscillation. As sleep deepens, thalamic neurones areas (see related discussions). They might entrain the basal
become more hyperpolarized and a clock-like delta rhythm ganglia neuronal population in highly synchronized oscilla-
appears in thalamocortical cells, due to their intrinsic mem- tions. At the cellular level also, the activity of striatal neurones
brane properties. This delta rhythm arises from the interplay has been shown to depend directly on frontal cortex activity.
between a low threshold calcium current (It) and a hyperpo- During sleep or anaesthesia, the striatal membrane potential
larization-activated K+ current (Ih). The thalamic delta alternates long phases of hyperpolarization with bursts of
rhythm is conveyed to the cortex, which further reorganizes discharges (Wilson 1994). Cortical stimulation can elicit
it and incorporates it into a slow, cortically induced, slow membrane potential shifts similar to the spontaneous ones
rhythm (Steriade and Amzica 1998). Neuroimaging data (Wilson 1994).
support the hypothesis that the same mechanisms are at work Braun et al. (1997) proposed a more active role for basal
in humans in stage 2 and SWS. ganglia in the generation of SWS. Their proposal is based
Basal forebrain/anterior hypothalamus was found to be essentially on the anatomical connectivity of basal ganglia and
deactivated during SWS in two studies (Braun et al. 1997; may be summarized as follows. Basal ganglia are in a position
Maquet et al. 1997). Given the spatial resolution of PET data, to modulate cortical arousal because they receive information
the area encompasses the anterior part of the hypothalamus as from the reticular core via the intralaminar thalamus and
well as the basal forebrain. These brain areas are structurally might transfer it to the cortex via the ventral anterior and
and functionally heterogeneous (Szymusiak 1995). On the one ventromedial thalamic nuclei. Furthermore, basal ganglia
hand, BF/AH is a critical site for cortical arousal. Ibotenic acid might also regulate the activity of the reticular core via their
lesions of basal forebrain reduces acetlylcholinesterase cortical projections to the pontine and mesencephalic reticular forma-
staining and results in an ipsilateral increase in slow waves tion. In favour of the role of basal ganglia in SWS regulation,
(Buzsaki et al. 1988). In contrast, electrical and chemical BF it should be remembered that the mechanical removal of the

218 P. Maquet
striatum (caudate nucleus) causes long-lasting insomnia in cats cause an EEG arousal (Nishino and Mignot 1998). In the same
(Villablanca et al. 1976). In contrast, low-frequency electrical line, moda®nil binds with low anity to the dopamine
stimulation of the caudate nucleus induces EEG synchroniza- transporter and this might play a part in its arousing eects
tion and behavioural inhibition (Siegel and Wang 1974). (Mignot et al. 1994). In addition, dopamine release within the
Furthermore, dopaminergic uptake inhibitors (which, however, striatum decreases during SWS although the ®ring rate of VTA
may or may not be speci®cally acting on the basal ganglia) neurones is not modi®ed during sleep (Trulson 1985). Finally,

SWS is reduced in Huntington's patients, who suer from a on aminoacidergic neurotransmission (Steriade et al. 1993d).
8 predominant striatal lesion (Wiegand and Lauer 1991). The participation of intracortical inhibitory neurones (Steriade
Whether active or passive, the role of the basal ganglia in et al. 1993a), as well as cortico-cortical connection ®bres
the regulation of sleep remains unclear (Nishino and Mignot (Amzica and Steriade 1995a, b), has also been emphasized.
1998) and deserves further investigation. Although neurophysiological studies in animals recorded
The cerebellar hemispheres are also signi®cantly depressed cortical activity from many sites (sensory, motor, visual) of
during SWS. This has been attributed to the reduction of primary and associative cortices, no regional modulation of
sensorimotor function (decreased muscle tone and proprio- sleep oscillations was observed at the cellular level. However,
ception) characterizing the sleeping subject (Braun et al. 1997; regions such as the orbito-frontal cortex have not been
Ho¯e et al. 1997). speci®cally investigated. In this respect, it is important to note
The cortical deactivations are certainly the most intriguing that the oscillatory frequency of cortical areas was dierent on
®ndings from neuroimaging studies. SWS profoundly aects either side of a lidocaine-inactivated brain area (Amzica and
the activity of the whole cortex, not only in the areas in which 9 Steriade 1995a, Fig. 9). This ®nding suggests that dierent
CBF is lowest. However, in these regions, a larger proportion local oscillators may exist in the cortical mantle, which are
of the local neuronal population is probably engaged in the normally synchronized by intracortical or thalamo-cortical
highly synchronous oscillation of sleep. Recent neurophysio- networks. However, it clearly appears that the cellular
logical studies explored the role of the cortex in the generation substrate of the regional cortical CBF depression is not yet
of sleep rhythms. The cortical mantle has been involved in the known. Further arguments suggesting a functional link
generation of a slow rhythm (<1 Hz) which organizes the between sleep processes and the most deactivated areas is thus
thalamically generated rhythms (spindles and delta) and to be found at higher levels of description.
synchronizes oscillations within the cortical and cortico- Braun et al. (1997) observed that CBF in fronto-parietal
thalamo-cortical networks (Steriade et al. 1993b, d; Steriade association cortices decreased markedly in SWS while the
and Amzica 1998). The cortex would also seem to be activity levels of unimodal sensory cortices were preserved.
implicated in the generation of a delta rhythm, less regular This ®nding was recently con®rmed by Kajimura et al. (1999).
than the clock-like thalamic delta, although the cellular They found that absolute levels of rCBF were lower in all
mechanism of this oscillation has not yet been described in cortical areas except the perirolandic and occipital cortex
detail (Steriade and Amzica 1998). during SWS (Kajimura et al. 1999). This observation sug-
The slow rhythm relies on the properties of the cortical gested that the ®rst cortical relay areas for exteroceptive
networks and the alternate silent periods of hyperpolarization stimuli remain active during SWS. Although attractive, this
with activity periods (Steriade et al. 1993b). The long-lasting hypothesis is challenged by another way of interpreting the
silent periods would appear to be due to a cascade of data. It should again be emphasized that polymodal associa-
disfacilitation (Contreras et al. 1996). Activity periods depend tion cortices are the most active areas during wakefulness,
whereas primary sensory and motor cortices do not reach such
high levels of activity during resting conditions. The dieren-
Figure 3. Decreases in rCBF during slow-wave sleep (SWS). (a) Data tial behaviour of primary and associative cortices could thus
from Maquet et al. (1997). Contrast: [(W + REMS)/2] ± SWS. Mid- be due more to processes occurring during wakefulness than to
sagittal and transverse sections showing deactivation of the central genuine SWS processes. In keeping with this interpretation,
core structures (brainstem and thalamic nuclei), the basal forebrain, eigenimage analysis of our sleep data shows that 30% of the
the basal ganglia, the orbito-frontal and anterior cingulate cortices and
the precuneus. Section numbers on the sagittal section refer to the
variance could be explained by a fronto-parietal network
respective transverse sections. Reproduced with permission from the which is more active in wakefulness that in sleep (Fig. 4). [The
Journal of Neuroscience. (b) Data from Braun et al. (1997). Contrast: ®rst PC (50% of variance) essentially dierentiates REM sleep
presleep W ± SWS. Transverse sections showing deactivation of the from both SWS and wakefulness, in relation with negative
brainstem, the thalamus, the basal forebrain, the basal ganglia, the loadings in the lateral prefrontal cortex and parietal cortex.]
orbito-frontal cortex, the frontal and parietal cortices on the convexity.
Also note the deactivation of the cerebellum and the insular cortex.
Furthermore, neurophysiological data in animals have
Reproduced with permission from the authors and Brain. (c) Data shown that exteroceptive aerent volleys are blocked mainly
from Andersson et al. (1998). Contrast: W ± SWS. Transverse sections at the thalamic levels during SWS (Steriade et al. 1990b, c).
showing deactivation of the thalamus as well as of the frontal and This ®nding minimizes the role of cortical areas in aerent
parietal cortices on the convexity. Reproduced with permission from processing during SWS. A third hypothesis, which reconciles
the authors and the Journal of Cerebral Blood Flow and Metabolism.
(d) Data from Kajimura et al. (1999). Contrast: W ± SWS. Surface
the previous suggestions, would be that polymodal associative
rendering of brain areas deactivated in SWS. For the four images on cortices are both more active during wakefulness and more
the left, data were not adjusted for global blood ¯ow changes in SWS. profoundly in¯uenced by SWS rhythms than primary cortices.
All cortical areas are deactivated with the exception of primary Current intracellular recordings indeed show that cellular
cortices. For the four images on the right, the data were adjusted for correlates of SWS rhythms are more pronounced in associ-
global ¯ow changes in SWS. The regional deactivations are localized in
brainstem, the thalamus, the cerebellum, the posterior cingulate cortex
ation cortical areas (Steriade, personal communication). Like-
and left parietal cortex. Reproduced with permission from the authors wise, in cats involved for some time in a active visual task,
and the Journal of Neuroscience. neurones in the associative visual cortex may adopt a bursting

220 P. Maquet
Figure 4. Analysis of the sleep data pub-

lished in Maquet et al. (1996, 1997) by
singular value decomposition (Frac-
kowiak et al. 1997). This analysis is a
simple way of measuring a pattern of
correlated cerebral activity. Singular value
decomposition is an operation that
decomposes the original time series (the
series of scans) into the patterns in space
and time which account for the most
variance±covariance of the original data.
The identi®ed spatial modes (also called
eigenimages) can be interpreted on the
basis of their temporal mode, i.e. their
expression over dierent brain states. The
®gure shows the second eigenimage, which
accounts for 30.6% of the variance (a).
The spatial mode with negative loadings is
characterized by a fronto-parietal net-
work, which also includes the precuneus
(b). The corresponding temporal mode (c)
indicates that this component is due to the
contrast between waking and sleep scans.
The negative scores of waking scans echo
the negative loadings shown in (b) and
suggest that the fronto-parietal network is
most expressed in wakefulness.

pattern typical for the sleeping cortex and become less At this point, it is fair to note that other signs observed after
responsive to visual stimulation, while the primary visual area short-term sleep deprivation do not depend on ventro-medial
maintains a normal response to visual inputs (Pigarev et al. but on lateral frontal areas. For instance, verbal ¯uency
1997). (Friston et al. 1991b) is known to rely on the dorsolateral
Below we discuss in more detail three brain areas: the prefrontal cortex and is rapidly impaired by sleep deprivation
ventro-mesial frontal regions, the precuneus and the mesio- (Horne 1988). Thus, it remains open as to whether the eects
temporal areas. of SWS and regional deactivations are restricted to orbito-
Ventro-mesial areas of the frontal lobe essentially encom- frontal regions or also involve frontal convexity.
pass both the anterior cingulate cortex and the orbito-frontal The precuneus, on the medial aspect of the parietal lobe, is
cortex. This terminology is proposed by Damasio (1994) and signi®cantly deactivated during SWS (Braun et al. 1997;
is very convenient here because these regions have signi®- Maquet et al. 1997; Andersson et al. 1998). The interpretation
cantly low CBF during SWS (Braun et al. 1997; Ho¯e et al. of this deactivation is uncertain. As can be seen in Fig. 2, this
1997; Maquet et al. 1997). A functional link between these association cortex is particularly active during wakefulness. Its
areas and SWS processes is suggested by three lines of activity contrasts the waking state from both SWS and REM
evidence. First, early studies showed that electrical stimula- sleep (Fig. 4). As for the association cortices of the convexity,
tion of the orbito-frontal cortex causes an outbreak of slow the activity of the precuneus would thus appear to re¯ect more
waves on EEG recordings and induces behavioural sleep brain operations taking place in conscious wakefulness than in
(Sterman and Clemente 1962; Lineberry and Siegel 1971). At sleep. Curiously, the precuneus and posterior cingulate cortex
present, these results are dicult to reconcile with the were shown to be hypoactive in mental states of decreased or
prominent theory of sleep generation which invokes mainly abolished consciousness, e.g. pharmacological sedation (Fiset
a disfacitilitation of thalamo-cortical networks under the et al. 1999), hypnotic state (Maquet et al. 1999), vegetative
permissive inactivity of the brainstem structures (see above). state (Laureys et al. 1999). The role of this area in conscious
However, a detailed description of the functional interactions processes certainly deserves further investigation.
occurring in sleep between ventro-mesial frontal areas and Mesio-temporal areas were found to be deactivated during
diencephalic or brainstem structures probably deserves further SWS (Maquet et al. 1997; Andersson et al. 1998). This would
investigation. suggest that highly synchronous oscillations also occur in these
Second, it is noteworthy that the topography of slow wave areas. At the cellular level, such activities are indeed observed
activity in humans predominates over frontal regions (Zeit- during SWS in the hippocampal formation of rats. Two main
lhofer et al. 1993; Werth et al. 1996). Although the highest types of activity are observed in the hippocampus in rats: theta
delta power densities are observed over the frontal convexity, rhythm and sharp waves, respectively, associated with
the limited spatial resolution and spatial sampling of EEG c oscillations and ripples (Chrobak and Buzsaki 1998). Sharp
recordings make it dicult to determine the precise localiza- waves and ripples are observed during consumatory beha-
tion of frontal areas most intimately related to this higher delta viours and SWS. They originate from a synchronous discharge
power density. of CA3 populations which impinge on CA1 neurones. Sharp
Third, total sleep deprivation, the eects of which can be waves are transmitted to the super®cial layers of the entorhinal
attribued primarily to SWS deprivation (Horne 1993), rapidly cortex where they can also be recorded. The presence of similar
induces signs and symptoms similar to ventro-mesial frontal activities in primate hippocampal formation is uncertain. Early
lesions. Patients with orbito-frontal or anterior cingulate observations in chimpanzees reported slow waves and K
lesions are usually described as impulsive, disinhibited and complexes in the hippocampus during non-REM sleep but in
distractible. Interestingly, related signs, such as childish the human hippocampal complex, stage 4 sleep was essentially
humour (Kollar et al. 1966), disinhibition and irritability characterized by low voltage fast activity (Freemon and Walter
(Bliss et al. 1959), distractibility (Norton 1970), and persever- 1970). However, some high voltage M-shaped transients were
ations (Horne 1988) are likewise observed after short-term also recorded (Freemon and Walter 1970). More recently,
total sleep deprivation in man. As orbito-frontal and anterior using single unit recordings in epileptic patients, it was shown
cingulate cortices are closely related to the regulation of that ®ring rates in the entorhinal cortex were lowest during
emotions and behaviour, it is also noteworthy that a recent SWS, while hippocampal neurone activity was less prone to
meta-analysis shows that one of the main eects of sleep state-related changes (Staba et al. 1998).
deprivation in humans is modi®cation of aect (Pilcher and From the psychological standpoint, the hippocampal for-
Hucutt 1996). Likewise, ventromesial frontal areas have been mation is intimately involved in explicit memory systems
implicated in decision-making (Bechara et al. 1994; Bechara (Squire and Zola 1996). Functional relationships between sleep
et al. 1995). This process aims to integrate large numbers of and memory remain subject to debate (Stickgold 1999). Some
facts, past experience and estimations of future consequences recent data show that SWS deprivation has detrimental eects
in order to adapt the behaviour in the best interest of the on episodic memory trace consolidation (Plihal and Born
individual (Adolphs et al. 1996). Once again, sleep deprivation 1997).
rapidly leads to de®cits in decision-making (Harrison and In consequence, all these arguments suggest that the
Horne 1998, 1999). deactivation of mesio-temporal areas in SWS re¯ect local slow

222 P. Maquet
synchronous oscillations, possibly related to memory traces Table 3 Decreases in rCBF during REM sleep
processing. At present, further evidence is needed to support Maquet et al. (1996) Braun et al. (1997)
this hypothesis.
Quanti®cation ) +
Global activity + +
REM sleep adjustment
Preformal cortex ¯ ¯
In contrast to SWS, REM sleep is characterized by sustained Parietal cortex ¯ ¯
neuronal activity (Jones 1990; Steriade et al. 1990) and, Precunes/posterior ¯ ¯
correspondingly, by high cerebral energy requirements cingulate cortex
Remarks CBF CBF
(Maquet et al. 1990) and blood ¯ow (Madsen et al. 1991a;
REMS ) [(W + SWS)/2] REMS ) SWS
Madsen and Vorstrup 1991; Franzini 1992). In this working
brain, some areas are more active than others; in contrast, +, CBF or glucose metabolism quanti®cation and global activity
other regions have lower than average CBF. This section deals adjustment have been performed; ±, CBF or glucose metabolism
quanti®cation and global activity adjustment have not been
with these two types of structure.
performed.
Three studies described the functional neuroanatomy of
REM sleep using PET; two used H215O CBF measurements
synaptic tracking in this region during REM sleep. Some
(Maquet et al. 1996; Braun et al. 1997), the third measured the
of the local neuronal populations monosynaptically project to
glucose metabolism with [18F]¯uorodeoxyglucose (Nofzinger
the thalamic nuclei, which was also found to be activated
et al. 1997). The core functional characteristics of REM sleep
bilaterally in the three published studies.
are: (i) activation of the mesopontine tegmentum, thalamic
It was usually thought that thalamic nuclei, and in particular
nuclei, limbic/paralimbic structures and posterior cortices; and
the intralaminar thalamus, dispatched this activation diusely
(ii) relative deactivation of frontal and parietal cortices
and homogeneously to the cortex. Early studies in rats and cats
(Tables 2 and 3, Fig. 5).
had suggested the heterogeneity of cerebral metabolism during
Activation of the dorsal aspect of the mesopontine tegmen-
sleep, in particular REM sleep (Ramm and Frost 1983, 1986;
tum is in good agreement with the role played by this area
Lydic et al. 1991). Functional neuroimaging in humans
in REM sleep generation in animals. It should be emphasized
de®nitely showed that the distribution of telencephalic activity
that the activation encompasses many local nuclear structures
is heterogeneous and speci®c to REM sleep. The functional
that cannot be sorted by PET (systems level description)
signi®cance of the cortical pattern of activity is still open to
and that the increased regional CBF re¯ects an intense
discussion.
We were intrigued by the bilateral activation of amygdaloid
Table 2 Increases in rCBF during REM sleep complexes during REM sleep. Furthermore, neuroanatomical
evidence in the Macaque showed that cortical areas, which
Maquet et al. Braun et al. Nofzinger et al.
were activated during REM sleep (anterior cingulate cortex,
(1996) (1997) (1997)
parietal lobule), receive numerous amygdalar projections,
Quanti®cation ) + + whereas deactivated cortices (frontal and parietal) are practi-
Global activity + + + cally devoid of amygdalar aerents (Amaral and Price 1984;
adjustment
Brainstem
Amaral et al. 1992; Aggleton 1993). This anatomical argument
Thalami suggested that amygdaloid complexes could modulate cortical
Basal forebrain activity during REM sleep. Our hypothesis was even more
Amygdala likely when we were able to show that functional interactions
Hippocampal
formation
Orbito-frontal
cortex Figure 5. (a) Brain areas with signi®cant rCBF increase during REM
Parietal lobule sleep. (i) Data from Maquet et al. (1996). Contrast: REMS ±
Extrastriate cortex [(W + SWS)/2]. Transverse sections showing activation of pontine
Anterior cingulate tegmentum, the thalami, the anterior cingulate cortices and the right
cortex parietal operculum. Reproudced with permission from Nature. (ii)
Insula-temporal Data from Braun et al. (1997). Contrast: REMS ± SWS. Transverse
cortex sections showing activation of the brainstem and the thalamus, as well
Cerebellum as the orbito-frontal, anterior cingulate and insular cortices. Note the
Remarks CBF CBF Glucose activation of parahippocampal gyrus on the mesial aspect of the
REMS ) REMS ) SWS metabolism temporal lobe and of lateral occipital areas. Reproduced with
[(W + SWS)/2] REMS ) W permission from the authors and Brain. (b) Brain areas with signi®cant
rCBF decrease during REM sleep. Note the deactivation of frontal and
+, CBF or glucose metabolism quanti®cation and global activity parietal cortices. (i) Data from Maquet et al. (1996). Contrast:
adjustment have been performed; ), CBF or glucose metabolism [(W + SWS)/2] ± REMS. Reproduced with permission from Nature.
quanti®cation and global activity adjustment have not been (ii) Data from Braun et al. (1997). Contrast: W ± REMS. Reproduced
performed. with permission from the authors and Brain.


224 P. Maquet
between the amygdala and occipito-temporal cortices were In humans, REM sleep deprivation is more detrimental to
dierent in the context of REM sleep than in SWS or in W procedural than explicit memory tasks (Smith 1995). For
(Maquet and Phillips 1998). example, REM sleep deprivation alters a perceptual implicit
Furthermore, animal data suggested a role for amygdala in learning of a visual discrimination task (Karni et al. 1994).
REM sleep modulation. Early experiments showed that a Furthermore, improvement in the same task is correlated with
unilateral daily 1-min stimulation of rat amygdala increased the amount of REM sleep in the second part of the night
the amount of REM sleep (Smith and Young 1980). Likewise, (Stickgold 1999). Likewise, modi®cations of the visual ®eld in
electrical stimulation of the central nucleus of amygdaloid normal subjects using goggles (which forces them to learn
complexes increases PGO activity (Calvo et al. 1987), while implicitly to move in their environment) causes an increase in
carbachol injections in the same nucleus increase both REM REM sleep in the following days (De Koninck and PreÂvost
sleep and PGO waves (Calvo et al. 1996). Intra-amygdalar 1991). In contrast, explicit memory, which depends on the
injections of serotonin produce a shift from SWS to REM function of the mesio-temporal areas, has not been success-
sleep in rats (Sanford et al. 1995). fully related to any REM sleep (Smith 1995). Further research
However, other neuroimaging studies found activation in is needed to specify the role of REM sleep in memory
the hippocampal formation with (Nofzinger et al. 1997) or processes and especially those involving the mesio-temporal
without (Braun et al. 1997) simultaneous amygdalar activa- areas.
tion. This suggested that REM sleep would appear to be Neocortical activations were observed inconstantly during
characterized more by an activation of the limbic system than REM sleep. Early studies mentioned signi®cant increases in
by an amygdalar hyperactivity alone. As the limbic system is a glucose metabolism (Maquet et al. 1990) or CBF (Madsen et al.
phylogenetically old structure and is deeply involved in the 1991a) in occipito-temporal areas. More recently, Braun
regulation of aective and social behaviours (MacLean 1990), et al. (1997) observed such occipital activation while Nofzinger
these results are compatible with, but not decisively supportive et al. (1997) and Maquet et al. (1996) did not. The cause of these
of, the hypothesis of a genetic re-programming during REM discrepancies is possibly that posterior cortex activation in REM
sleep, a process which would rehearse the repertoire of innate sleep is variable across subjects and the group analysis averages
behaviours and maintain the psychological individuation out the individual activations, unless they occur in the same
(Jouvet 1998). regions in all (or at least the majority of) individuals.
The reasons why the hippocampal formation is activated in It is interesting to note that the functional relationships
some studies while the amygdala shows up in others remain between posterior cortical areas is dierent in REM sleep than
unclear. One reason might be that the previous waking in wakefulness (Braun et al. 1998). During REM sleep, a
experience was dierent in the experimental populations. negative correlation is observed between the activity of
Indeed, both the amygdala and hippocampal formation are extrastriate cortex and primary visual areas, in contrast to
critical structures for memory systems (Bechara et al. 1995). that usually observed during wakefulness. These results, like
The relationship between sleep and memory processes is the amygdalo-temporal interactions we described (Maquet and
beyond the scope of this paper and would deserve a review in Phillips 1998), strongly suggest that not only is the functional
itself. In the following, we summarize only data which can neuroanatomy of REM sleep dierent from wakefulness, but,
provide some explanation for the PET data. in addition, functional interactions between neuronal popula-
At the cellular level, theta rhythms, together with gamma tions are also dierent in REM sleep than in wakefulness.
oscillations, are recorded in hippocampal formation during Further research should explore the characteristics of func-
REM sleep and exploratory behaviour in rats. In a multistep tional and eective connectivity in REM sleep.
hypothesis, during sleep there would be an interplay between
sharp waves/ripples of SWS and theta/gamma oscillations
A note on dreaming
during REM sleep. This interplay would be involved in the
consolidation of memory traces (Buzsaki et al. 1994). Functional neuroimaging of normal subjects during REM
At the behavioural level, more emphasis has been put on sleep also bears on the neural correlates of dreaming activity.
tasks that do not depend on the hippocampal formation. In In our own study, subjects were systematically awakened after
rats, REM sleep deprivation disturbs the learning of various each sleep scan (Maquet et al. 1996). All subjects awakened
tasks (Hennevin et al. 1995; Smith 1995). These eects are during REM sleep had a dream recall (Maquet et al. 1996),
observed only if the deprivation occurs during certain periods, whereas only one dream was reported after SWS awakening.
the PS windows, after the learning session (Smith 1985). As a consequence, at present, the functional maps concern
Furthermore, the amount of REM sleep increases during the REM sleep with dreaming: we lack any observation of REM
learning period (Hennevin et al. 1995). This increase in REM sleep without dreaming and a sucient number of SWS with
sleep is speci®cally linked to memory processes because: (i) it dreaming. The latter two groups would be interesting to have a
does not occur in nonlearners; (ii) it is only transient, with comprehensive view of the neural correlates of dreaming in
REM sleep levels decreasing to baseline levels as soon as the humans.
animal masters the task; and (iii) it occurs at speci®c times Consequently, all that can be said is that the distribution of
after learning sessions. cerebral activity in REM sleep is consistent with some

characteristics of dream recalls (Maquet et al. 1996). Their

CONCLUSIONS AND PERSPECTIVES
perceptual character, essentially visual and auditory, might
rely on the activity of temporo-occipital areas, which were Despite a relentless research eort and an increasing number
shown to be activated during REM sleep in some glucose of investigative tools available, sleep is one of the last
metabolism (Maquet et al. 1990) and CBF (Madsen et al. behaviours, whose function remains unknown. Among the
1991a; Braun et al. 1997) studies. The emotional load in many possible approaches to explore sleep processes, func-
dreams might be related to high limbic activity, especially to tional neuroimaging enables the study of cerebral specializa-
amygdalar activation. The peculiar functional connectivity tion and integration during sleep in man at the systems level.
between amygdaloid complexes and temporo-occipital cortices The published work is only at a preliminary stage and new
would also contribute to the aective aspect of dreaming data should soon increase our understanding of sleep processes
(Maquet and Phillips 1998). The relative hypoactivity of the in humans. In this endeavour, progress will come from two
prefrontal cortex would explain the characteristic lack of types of advance: better experimental protocols, dealing with
insight, the alteration of time perception or even the delusional more speci®c aspects of sleep and new neuroimaging tech-
belief of being awake, as recently discussed by Hobson et al. niques, including fMRI; and multimodal imaging.
(1998). The cerebral organization described above proposes a
Any further comment on the functional neuroanatomy of stereotyped, ®xed picture of sleep processes, a picture of
dreaming remains speculative. At this point, it is useful to `canonical' human sleep. This description may give the
recall some basic characteristics of dreaming activity. First, impression that sleep is a stereotypical process. In contrast,
dreams, like other cognitive processes (attention, memory, we know that sleep may adapt dynamically in response to
language, mental imagery), result from interactions of neuro- modi®cations in the outer or inner environment.
nal populations, especially in the cortex. Second, dreams are Several speci®c aspects of sleep deserve further investigation.
not observable directly. We have access only to dream recall, The correlates of sleep rhythms, EEG transients and phasic
i.e. the verbal translation of a whole world of perceptions and events should be speci®ed. Up to now, the results have
feelings felt by the dreamer during his sleep. Third, the essentially dealt with SWS and spindle activity (Ho¯e et al.
dreamer is not aware of his dreams until he is awakened and 1997), and many other activities remain to be explored.
recovers the full consciousness of the waking state: while he is Concerning the ocular saccades of REM sleep, preliminary
dreaming during sleep, the dreamer is not conscious he is results have certainly shown activation of prefrontal and
dreaming. We do not consider here the particular case of lucid parietal areas (Hong et al. 1995), but this deserves con®rma-
dreamers, who, to our knowledge, have not yet been explored tion with a more powerful and localizing statistical analysis.
by functional neuroimaging. Fourth, most dream recall occurs The internal organization of sleep during the night, i.e. the
immediately after awakening and their memory rapidly van- temporal course of sleep processes, is also worth describing it
ishes in a few tens of seconds. in detail. How does brain organization change in relation to
One possible scenario would thus be the following. During the exponential decrease in slow-wave activity in SWS, or with
REM sleep, in a cholinergic context, brain function is the modi®cation of spindling activity during the night? Are late
characterized by predominant interactions between mesio- REM sleep periods comparable with early ones? How is the
temporal areas and posterior sensory cortices, while supra- brain activity organized in non sleep-deprived subjects or
modal associative cortices in frontal and parietal cortices are during a siesta?
relatively quiescent. This mode of brain function carries a low Another line of research concerns the role of sleep in the
level of consciousness but is favourable to a hallucinatory, interaction with the environment. How does the brain process
oneiric mentation. On arousal, brain activity is reorganized, external stimuli during sleep? In particular during REM sleep,
with a preponderance of frontal and parietal cortex function how is it that your alarm clock ringing in the morning is
and the arising of full consciousness. During the transient incorporated into your dream and does not wake you up? Are
state of awakening, the subject would become abruptly aware stimuli in REM sleep processed dierently than in wakefulness
of neuronal interactions going on in REM sleep. As these at the cortical level? Recent fMRI studies suggest that primary
interactions are replaced by those prevailing in wakefulness, sensory areas continue to process auditory inputs during sleep
the dream vanishes and gives way to self and environmental (Portas et al. 1999).
awareness. The amnesia of dreams at awakening would An even more intriguing question is the possible in¯uence of
appear to be related to the ephemeral character of brain previous experience on brain activity in sleep. This question is
function reorganization. The relative hypoactivity in REM related to possible functional links between sleep, memory
sleep of the prefrontal and parietal cortices, which are processes and, more generally, brain plasticity. Slow-wave
involved in working memory and episodic memory processes activity in SWS increases in the central area controlateral to the
during wakefulness, might also contribute to the amnesia at hand submitted to a vibratory stimulus in the preceding evening
awakening. In this view, it clearly appears that a better (Kattler et al. 1994). Likewise, preliminary results suggest
understanding of the REM sleep neuronal interactions is the experience-related changes in brain CBF distribution during
core problem on which most research eort should be sleep after a training period (Maquet and Phillips 1998, Maquet
focused. et al. 2000).

226 P. Maquet
Interindividual variability should also be investigated sys-

REFERENCES
tematically, more probably thanks to fMRI acquisitions. Does
every individual organize their his/her brain function during Abraham, W. C., Delanoy, R. L., Dunn, A. J. and Zornetzer, S. F.
sleep in the same way? Does the functional neuroanatomy of Locus coeruleus stimulation decreases deoxyglucose uptake in
ipsilateral mouse cerebral cortex. Brain Res., 1979, 172: 387±392.
sleep in the elderly dier from young subjects? Along the same Abrams, R. M., Post, J. C., Burch®eld, D. J., Gomez, K. J., Hutchison,
lines, pathological populations are worth exploring: how is A. A. and Conlon, M. Local cerebral blood ¯ow is increased in
brain activity organized in depressed, narcoleptic or apnoeic rapid-eye-movement sleep in fetal sheep. Am. J. Obstet. Gynecol.,
patients in sleep? Do treatments normalize their functional 1990, 162: 278±281.
neuroanatomy during sleep? Preliminary data suggest that Achermann, P. and Borbely, A. A. Low-frequency (<1 Hz) oscilla-
tions in the human sleep electroencephalogram. Neuroscience, 1997,
depressed patients do not activate the anterior cingulate cortex 81: 213±222.
during REM sleep (Nofzinger et al. 1998). Adachi, T., Biesold, D., Inanami, O. and Sato, A. Stimulation of the
Pharmacological trials should explore the eects of hypnotic nucleus basalis of Meynert and substantia innominata produces
drugs. Is the regional activity in drug-induced sleep compar- widespread increases in cerebral blood ¯ow in the frontal, parietal,
able with physiological sleep? Preliminary observations on and occipital cortices. Brain Res., 1990, 514: 163±166.
Adolphs, R., Tranel, D., Bechara, A., Damasio, H. and Damasio, A. R.
CBF distribution during sleep induced by zolpidem have been
Neuropsychological approaches to reasoning and decision-making.
reported (Finelli et al. 1998, Finelli et al. 2000). 10 In: Damasio, A. R., Damasio, A., Christen, Y. (Eds) Neurobiology of
It appears that the PET studies presented in this paper were Decision-Making. Springer-Verlag, Berlin, 1996: 157±179.
very crude in their description of sleep functional neuroanat- Aggleton, J. P. The contribution of the amygdala to normal and
omy but they opened up the way for a whole new ®eld of abnormal emotional states. Trends Neurosci., 1993, 16: 328±333.
Amaral, D. G. and Price, J. L. Amygdalo-cortical projections in the
research. No doubt future studies will provide a picture of
monkey (Macaca fascicularis). J. Comp. Neurol., 1984, 230: 465±
more versatile sleep processes, which adjust individual beha- 496.
viour to various external and internal adaptive constraints. Amaral, D. G., Price, J. L., PitkaÈnen, A. and Carmichael, S. T.
New neuroimaging techniques now challenge PET as 11 Anatomical organization of the primate amygdaloid complex. In:
valuable methods for sleep investigations. At present, sleep Aggleton, J. P. (Ed) The Amygdala: Neurobiological Aspects of
Emotion, Memory, and Mental Dysfunction. Wiley-Liss, New York,
research in fRMI remains a challenge given its sensibility to
1992: 1±66.
movement artefacts, the subject's discomfort and the diculty Amzica, F. and Steriade, M. Cellular substrates and laminar pro®le of
of EEG recordings. Some of these problems have already been sleep K-complex. Neuroscience, 1997, 82: 671±686.
solved, however (Huang-Hellinger et al. 1995), and functional Amzica, F. and Steriade, M. Disconnection of intracortical synaptic
MRI should soon provide a description of sleep processes with linkages disrupts synchronization of a slow oscillation. J. Neurosci.,
1995a, 15: 4658±4677.
better spatial and temporal resolutions. Likewise, a multimo-
Amzica, F. and Steriade, M. Short- and long-range neuronal
dal approach should certainly also be considered. Combining synchronization of the slow (<1 Hz) cortical oscillation. J. Neuro-
two (or more) neuroimaging techniques would allow a more 12 physiol., 1995b, 73: 20±38.
precise description of the spatio-temporal brain activities and Andersson, J. L. R., Onoe, H., Hetta, J., Lidstrom, K., Valind, S.,
regional interactions during sleep (Maquet and Phillips 1998). Lilja, A., Sundin, A., Fasth, K. J., Westerberg, G., Broman, J. E.,
Watanabe, Y. and Langstrom, B. Brain networks aected by
As should appear clear, a new era is on the horizon which
synchronized sleep visualized by positron emission tomography.
should lead to a better understanding of macroscopic sleep J. Cereb. Blood Flow Metab., 1998, 18: 701±715.
processes in humans. This body of data will have to be integrated Andreasen, N. C., O'Leary, D. S., Cizadlo, T., Arndt, S., Rezai, K.,
with the information provided by other techniques and describ- Watkins, G. L., Boles Ponto, L. L. and Hichwa, R. D. Remember-
ing similar mechanisms at other levels of observation. This ®nal, ing the past: two facets of episodic memory explored with positron
emission tomography. Am. J. Psychiatr., 1995, 152: 1576±1585.
multidisciplinary step in the interpretation of experimental data
Bates, D., Weinshilboum, R. M., Campbell, R. J. and Sundt, T. M.
seems essential to unravel the functions of sleep. The eect of lesions in the locus coeruleus on the physiological
responses of the cerebral blood vessels in cats. Brain Res., 1977, 136:
431±443.
ACKNOWLEDGEMENTS Bechara, A., Damasio, A. R., Damasio, H. and Anderson, S. W.
Insensitivity to future consequences following damage to human
The author is particularly indebted to Professor Mircea prefrontal cortex. Cognition, 1994, 50: 7±15.
Steriade, Professor Karl Friston and the anonymous reviewers Bechara, A., Tranel, D., Damasio, H., Adolphs, R., Rockland, C. and
for their comments on a previous version of the paper. The Damasio, A. R. Double dissociation of conditioning and declarative
knowledge relative to the amygdala and hippocampus in humans.
author would like to thank Allan Braun, Jesper Andersson and Science, 1995, 269: 1115±1118.
Naofumi Kajimura for their sleep PET images. The author is Biesold, D., Inanami, O., Sato, A. and Sato, Y. Stimulation of the
Senior Research Associate and Research Associate at the nucleus basalis of Meynert increases cerebral cortical blood ¯ow in
Fonds National de la Recherche Scienti®que de Belgique rats. Neurosci. Lett., 1989, 98: 39±44.
(FNRS). The personal research reported in this review was Bliss, E. L., Clarck, L. D. and West, C. D. Studies of sleep deprivation.
Arch. Neurol., 1959, 81: 348±359.
supported by FNRS grants, by Special Funds for Scienti®c Bonvento, G., Lacombe, P. and Seylaz, J. Eects of electrical
Research of the University of LieÁge and by the Queen stimulation of the dorsal raphe nucleus on local cerebral blood
Elisabeth Medical Foundation. ¯ow in the rat. J. Cereb. Blood Flow Metab., 1989, 9: 251±255.

Borowsky, I. W. and Collins, R. C. Metabolic anatomy of brain: a Dahlgren, N., Lindvall, O., Sakabe, T., Stenevi, U. and SiesjoÈ, B. K.
comparison of regional capillary density, glucose metabolism, and Cerebral blood ¯ow and oxygen consumption in the rat brain after
enzyme activities. J. Comp. Neurol., 1989, 288: 401±413. lesions of the noradrenergic locus coeruleus system. Brain Res.,
Braun, A. R., Balkin, T. J., Wesensten, N. J., Carson, R. E., Varga, 15 1981b, 209: 11±23.
M., Baldwin, P., Selbie, S., Belenky, G. and Herscovitch, P. Damasio, A. Descartes's Error. Emotion, Reason, and the Human
Regional cerebral blood ¯ow throughout the sleep±wake cycle ± An 16 Brain. Putnam Books, 1994.
(H2O)-O-15 PET study. Brain, 1997, 120: 1173±1197. De Koninck, J. and PreÂvost, F. Le sommeil paradoxal et le traitement
Braun, A. R., Balkin, T. J., Wesensten, N. J., Gwadry, F., Carson, R. de l'information: une exploration par inversion du champ visuel.
E., Varga, M., Baldwin, P., Belenky, G. and Herscovitch, P. Revue Can. Psychologie, 1991, 45: 125±139.
Dissociated pattern of activity in visual cortices and their projections Denoyer, M., Sallanon, M., Buda, C., Delhomme, G., Dittmar, A. and
during human rapid eye movement sleep. Science, 1998, 279: 91±95. Jouvet, M. The posterior hypothalamus is responsible for the
Buchsbaum, M. S., Gillin, J. C., Wu, J., Hazlett, E., Sicotte, N., increase of brain temperature during paradoxical sleep. Exp. Brain
Dupont, R. M. and Bunney, W. E. Regional cerebral glucose 17 Res., 1991, 84: 326±334.
metabolic rate in human sleep assessed by positron emission Detari, L. and Vanderwolf, C. H. Activity of identi®ed cortically
tomography. Life Sci., 1989, 45: 1349±1356. projecting and other basal forebrain neurones during large slow
Buchweitz, E., Edelman, N. H. and Weiss, H. R. Eect of locus waves and cortical activation in anaesthetized rats. Brain Res., 1987,
coeruleus stimulation on regional cerebral oxygen consumption in 437: 1±8.
the cat. Brain Res., 1985, 325: 107±114. Dirnagl, U., Lindauer, U. and Villringer, U. Role of nitric oxide in the
Busija, D. W. The role of central neural pathways in the regulation of coupling of cerebral blood ¯ow to neuronal activation in rats.
13 cerebral blood ¯ow. In: Phillis, J. W. (Ed) The Regulation of Neurosci. Lett., 1993, 149: 43±46.
Cerebral Blood Flow CRC Press, Boca Raton, FL, 1993: 65±77. Dufour, R. and Court, L. Le deÂbit ceÂreÂbral sanguin au cours du
Buzsaki, G., Bickford, R. G., Ponomare, G., Thal, L. J., Mandel, R. sommeil paradoxal du lapin. Arch. Italiennes Biologie, 1977, 115:
and Gage, F. H. Nucleus basalis and thalamic control of neocortical 57±76.
activity in the freely moving rat. J. Neurosci., 1988, 8: 4007±4026. Edvinsson, L. Functional role of perivascular peptides in the control of
Buzsaki, G., Bragin, A., Chrobak, J. J., Nadady, Z., Sik, A., Hsu, M. 18 cerebral circulation. Trends Neurosci., 1985, 126±131.
and Ylinen, A. Oscillatory and intermittent synchrony in the Edvinsson, L., Nielsen, K. C., Owman, C. and Sporrong, B.
hippocampus: relevance to memory trace formation. In: Buzsaki G. Cholinergic mechanisms in pial vessels. Histochemistry, electron
14 et al. (Eds) Temporal Coding in the Brain. Springer-Verlag, Berlin, microscopy and pharmacology. Zeitung Zellforschung, 1972, 134:
1994: 145±172. 311±325.
Calvo, J. M., Badillo, S., Morales-Ramirez, M. and Palacios-Salas, P. Ernst, T. and Hennig, J. Observation of a fast response in functional
The role of the temporal lobe amygdala in ponto-geniculo-occipital MR. Magnet. Reson. Med., 1994, 32: 146±149.
activity and sleep organization in cats. Brain Res., 1987, 403: 22±30. Finelli, L., Landoldt, H., Buck, A., Roth, C., Berthold, T., Acher-
Calvo, J. M., Simon-Arceo, K. and Fernandez-Mas, R. Prolonged mann, P. and Borbely, A. Functional neuroanatomy of human
enhancement of REM sleep produced by carbachol microinjection sleep after zolpidem: a H215O-PET study. J. Sleep Res., 1998,
into the amygdala. Neuroreport, 1996, 7: 577±580. 7 (Suppl. 2): 88.
Casamenti, F., Deenu, G., Abbamondi, A. L. and Pepeu, G. Changes Finelli, L. A., Landolt, H. P., Buck, A., Roth, C., Berthold, T.,
in cortical acetylcholine output induced by modulation of the Borbely, A. A., Achermann, P. Functional neuroanatomy of human
nucleus basalis. Brain Res. Bull., 1986, 16: 689±695. sleep states after zolpidem and placebo: A H2150-PET study [In
Chrobak, J. J. and Buzsaki, G. Operational dynamics in the hippo- Process Citation]. J. Sleep Res., 2000, 9: 161±173.
campal±entorhinal axis. Neurosci. Biobehav. Rev., 1998, 22: 303±310. Fiset, P., Paus, T., Daloze, T., Glourde, G., Meuret, P., Bonhomme,
Churchland, P. S. and Sejnowski, T. J. The Computational Brain. MIT V., Hajj-Ali, N., Backman, S. B. and Evans, A. C. Brain
Press, Cambridge, MA, 1992. mechanisms of propofol-induced loss of consciousness in humans:
Cohen, Z., Bonvento, G., Lacombe, J. and Hamel, E. Serotonin in the a positron emission tomography study. J. Neurosci., 1999, 19: 5506±
regulation of brain microcirculation. Prog. Neurobiol., 1996, 50: 5513.
335±362. Frackowiak, R. S. J., Friston, K. J., Frith, C. D., Dolan, R. J. and
Contreras, D., Timofeev, I. and Steriade, M. Mechanisms of long Mazziotta, J. C. Human Brain Function. Academic Press, San Diego,
lasting hyperpolarizations underlying slow sleep oscillations in cat CA, 1997.
corticothalamic networks. J. Physiol. (Lond.), 1996, 494: 251±264. Franzini, C. Brain metabolism and blood ¯ow during sleep. J. Sleep
Cote, A. and Haddad, G. G. Eects of sleep on regional blood ¯ow Res., 1992, 1: 3±16.
distribution in piglets. Pediat. Res., 1990, 28: 218±222. Freemon, F. R. and Walter, R. D. Electrical activity of human limbic
Cudennec, A., Bonvento, G., Duverger, D., Lacombe, P., Seylaz, J. system during sleep. Comp. Psychiatr., 1970, 11: 544±551.
and McKenzie, E. T. Eects of dorsal raphe nucleus stimulation on Friston, K., Ashburner, J., Frith, C., Poline, J. B., Heather, J. and
cerebral blood ¯ow and ¯ow-metabolism coupling in the conscious Frackowiak, R. S. J. Spatial realignment and normalization of
rat. Neuroscience, 1993, 55: 395±401. images. Human Brain Mapping, 1995a, 2: 165±189.
Cudennec, A., Duverger, D., Nishikawa, T., McRae-Degueurce, A., Friston, K. J., Frith, C. D., Liddle, P. F., Dolan, R. J., Lammertsma,
McKenzie, E. T. and Scatton, B. In¯uence of ascending serotonergic A. A. and Frackowiak, R. S. J. The relationship between global and
pathways on glucose use in the conscious rat brain. I. Eects of local changes in PET scans. J. Cereb. Blood Flow Metab., 1990, 10:
electrolytic or neurotoxic lesions of the dorsal and/or median rapheÂ 458±466.
nucleus. Brain Res., 1988, 444: 214±226. Friston, K. J., Frith, C. D., Liddle, P. F. and Frackowiak, R. S. J.
Curro Dossi, R., Nunez, A. and Steriade, M. Electrophysiology of a Comparing functional (PET) images: the assessment of signi®cant
slow (0.5±4 Hz) intrinsic oscillation of cat thalamocortical neurones change. J. Cereb. Blood Flow Metab., 1991a, 11: 690±699.
in vivo. J. Physiol., 1992, 447: 215±234. Friston, K. J., Frith, C. D., Liddle, P. F. and Frackowiak, R. S. J.
Dahlgren, N., Lindvall, O., Nobin, A. and Stenevi, U. Cerebral Investigating a network model of word generation with positron
circulatory response to hypercapnia: eects of lesions of central 19 emission tomography. Proc. R. Soc. Lond., 1991b, 244: 101±106.
dopaminergic and serotoninergic neuron systems. Brain Res., 1981a, Friston, K., Holmes, A. P., Worsley, K. J., Poline, J. B., Frith, C. D.
230: 221±233. and Frackowiak, R. S. J. Statistical parametric maps in functional

228 P. Maquet
imaging: a general approach. Human Brain Mapping, 1995b, 2: 189± Hoshi, Y., Mizukami, S. and Tamura, M. Dynamic features of
20 210. hemodynamics and metabolic changes in the human brain during
Friston, K. J., Worsley, K. J., Frackowiak, R. S. J., Mazziotta, J. C. all-night sleep as revealed by near-infrared spectroscopy. Brain Res.,
and Evans, A. C. Assessing the signi®cance of focal activations 1994, 652: 257±262.
using their spatial extent. Human Brain Mapping, 1994, 1: 214±220. Huang-Hellinger, F. R., Breiter, H. C., McCormack, G., Cohen, M. S.,
Gerashchenko, D. and Matsumura, H. Continuous recordings of brain Kwong, K. K., Baker, J. R., Belliveau, J. W. and Rosen, B. R.
regional circulation during sleep/wake state transitions in rats. Am. Simultaneous functional magnetic resonance imaging and electro-
J. Physiol., 1996, 270: R855±R863. physiological recording. Human Brain Mapping, 1995, 3: 13±23.
Goadsby, P. J., Lambert, G. A. and Lance, J. W. Dierential eects on Iadecola, C., Mraovitch, S., Meeley, M. P. and Reis, D. J. Lesions of
the internal and external carotid circulation of the monkey evoked the basal forebrain in rat selectively impair the cortical vasodilation
by locus coeruleus stimulation. Brain Res., 1982, 249: 247±254. elicited from cerebellar fastigial nucleus. Brain Res., 1983, 279:
Goadsby, P. J., Piper, R. D., Lambert, G. A. and Lance, J. W. Eect 41±52.
of stimulation of nucleus raphe dorsalis on carotid blood ¯ow. I. 22 Itakura, T. and Okuno, T. Tachykinins and cerebral blood ¯ow. In:
The cat. Am. J. Physiol., 1985a, 248: R263±R269. Phillis, J. W. (Ed) The Regulation of Cerebral Blood Flow CRC
Goadsby, P. J., Piper, R. D., Lambert, G. A. and Lance, J. W. Eect Press, Boca Raton, FL, 1993: 195±211.
of stimulation of nucleus raphe dorsalis on carotid blood ¯ow. II. Iwayama, T., Furness, J. B. and Burnstock, G. Dual adrenergic and
21 The monkey. Am. J. Physiology, 1985b, 248: R257±R262. cholinergic innervation of the cerebral arteries of the rat. Circ. Res.,
Goas, J. Y., Seylaz, J., Mamo, H., McLeod, P., Caron, J. P. and 1970, 26: 635±646.
Houdart, R. Variations du deÂbit sanguin du cortex de l'homme au Jones, B. E. Paradoxical sleep and its chemical/structural substrates in
cours du sommeil. Revue Neurologique (Paris), 1969, 120: 159±176. the brain. Neuroscience, 1990, 40: 637±656.
GoÈbel, U., Theilen, H. and Kuchinsky, W. Congruence of total and Jones, B. E. Paradoxical sleep and its chemical/structural substrates in
perfused capillary network in rat brains. Circulation Res., 1990, 66: the brain. Neuroscience, 1991, 40: 637±656.
271±281. Jouvet, M. Paradoxical sleep as a programming system. J. Sleep Res.,
Gomi, S., Gotoh, F., Ishihara, N., Tanaka, K., Ishikawa, Y., 1998, 7 (Suppl. 1): 1±5.
Takashima, S. and Mihara, B. Eects of lesioning the substantia Jueptner, M. and Weiller, C. Review: does measurement of regional
innominata on autoregulation of local cerebral blood ¯ow in rats. cerebral blood ¯ow re¯ect synaptic activity? Implications for PET
J. Cereb. Blood Flow Metab., 1991, 11: 66±71. and fMRI. Neuroimage, 1995, 2: 148±156.
Grant, D. A., Franzini, C., Wild, J. and Walker, A. M. Cerebral Kajimura, N., Uchiyama, M., Takayama, Y., Uchida, S., Uema, T.,
circulation in sleep: vasodilatory response to cerebral hypotension. Kato, M., Sekimoto, M., Watanabe, T., Nakajima, T., Horikoshi,
J. Cereb. Blood Flow Metab., 1998, 18: 639±645. S., Ogawa, K., Nishikawa, M., Hiroki, M., Kudo, Y., Matsuda,
Greisen, G., HellstroÈm-Vestas, L., Lou, H., Rosen, I. and Svenningsen, H., Okawa, M. and Takahashi, K. Activity of midbrain reticular
N. Sleep±waking shifts and cerebral blood ¯ow in stable preterm formation and neocortex during the progression of human
infants. Pediatric Res., 1985, 19: 1156±1159. non-rapid eye movement sleep. J. Neurosci., 1999, 19: 10065±
Hajak, G., KlingelhoÈfer, J., Schulz-Varszegi, M., Matzander, G., 10073.
Sander, D., Conrad, B. and RuÈther, E. Relationship between Kanzow, E., Krause, D. and KuÈhnel, H. Die Vasomotorik der
cerebral blood ¯ow velocities and cerebral electrical activity in sleep. Hirnrinde in den Phasen desynchronisierter EEG-AktivitaÈt im
Sleep, 1994, 17: 11±19. natuÈrlichen Schlaf der Katze. P¯uÈger Arch., 1962, 274: 593±607.
HallstroÈm, A., Sato, A., Sato, Y. and Ungerstedt, U. Eect of Karni, A., Tanne, D., Rubenstein, S., Askenasy, J. J. M. and Sagi, D.
stimulation of the nucleus basalis of Meynert on blood ¯ow and Dependence on REM sleep of overnight improvement of a
extracellular lactate in the cerebral cortex with special reference to perceptual skill. Science, 1994, 265: 679±682.
the eect of noxious stimulation of skin and hypoxia. Neurosci. Kattler, H., Dijk, D. J. and BorbeÂly, A. A. Eect of unilateral
Lett., 1990, 116: 227±232. somatosensory stimulation prior to sleep on the sleep EEG in
Harrison, Y. and Horne, J. A. One night of sleep loss impairs humans. J. Sleep Res., 1994, 3: 159±164.
innovative thinking and ¯exible decision making. Org. Behav. Kelly, P. A. T., Ritchie, I. M., McBean, D. E., Sharkey, J. and
Human Decision Processes, 1999, 78: 128±145. Olverman, H. J. Enhanced cerebrovascular responsiveness to
Harrison, Y. and Horne, J. A. Sleep loss aects risk-taking. J. Sleep hypercapnia following depletion of central serotonergic terminals.
Res., 1998, 7 (Suppl. 2): 113. J. Cereb. Blood Flow Metab., 1995, 15: 706±713.
Hennevin, E., Hars, B., Maho, C. and Bloch, V. Processing of learned Khateb, A., Muhlethaler, M., Alonso, A., Sera®n, M., Mainville, M.
information in paradoxical sleep: relevance for memory. Behav. and Jones, B. E. Cholinergic nucleus basalis neurons display the
Brain Res., 1995, 69: 125±135. capacity for rhythmic bursting activity mediated by low-threshold
Hobson, J. A., Pace-Schott, E. F., Stickgold, R. and Kahn, D. To dream calcium spikes. Neuroscience, 1992, 51: 489±494.
or not to dream? Relevant data from new neuroimaging and Kimura, A., Sato, A. and Takano, Y. Stimulation of the nucleus
electrophysiological studies. Curr. Opin. Neurobiol., 1998, 8: 239±244. basalis of Meynert does not in¯uence glucose utilization of the
Ho¯e, N., Paus, T., Reutens, D., Fiset, P., Gotman, J., Evans, A. C. cerebral cortex in anesthetized rats. Neurosci. Lett., 1990, 119: 101±
and Jones, B. E. Regional cerebral blood ¯ow changes as a function 104.
of delta and spindle activity during slow wave sleep in humans. Kiyosawa, M., Baron, J. C., Hamel, E., Pappata, S., Duverger, D.,
J. Neurosci., 1997, 17: 4800±4808. Riche, D., Mazoyer, B., Naquet, R. and McKenzie, E. T. Time
Hong, C. C., Gillin, J. C., Dow, B. M., Wu, J. and Buchsbaum, M. S. course of eects of unilateral lesions of the nucleus basalis of
Localized and lateralized cerebral glucose metabolism associated Meynert on glucose utilization in the cerebral cortex. Positron
with eye movements during REM sleep and wakefulness: a positron tomography in baboons. Brain, 1989, 112: 435±455.
emission tomography (PET) study. Sleep, 1995, 18: 570±580. Klein, B., Kuchinsky, W., SchrõÈ ock, H. and Vetterlein, F. Interdepen-
Horne, J. A. Human sleep, sleep loss and behaviour. Implications for dency of local capillary density, blood ¯ow, and metabolism in rat
the prefrontal cortex and psychiatric disorder. Br. J. Psychiatr., brain. Am. J. Physiol., 1986, 251: H1333±H1340.
1993, 162: 413±419. Ko, K. R., Ngai, A. C. and Winn, H. R. Role of adenosine in
Horne, J. A. Sleep loss and `divergent' thinking ability. Sleep, 1988, 11: regulation of regional cerebral blood ¯ow in sensory cortex. Am. J.
528±536. Physiol., 1990, 259: H1703±H1708.

Kollar, E. J., Slater, G. R., Palmer, J. O., Docter, R. F. and Mandell, Magistretti, P. J. and Pellerin, L. Cellular bases of brain metabolism
A. J. Stress in subjects undergoing sleep deprivation. Psychosom. and their relevance to functional imaging: evidence for a prominent
Med., 1966, 28: 101±113. role of astrocytes. Cerebral Cortex, 1996a, 6: 50±61.
Kuchinsky, W. Neuronal-vascular coupling. In: Villringer, A. and Magistretti, P. J. and Pellerin, L. Cellular mechanisms of brain energy
23 Dirnagl, U. (Eds) Optical Imaging of Brain Function, and Metab- metabolism. Relevance to functional brain imaging and neurode-
olism. 2. Physiological Basis and Comparison to Other Functional 27 generative disorders. Ann. NY Acad. Sci., 1996b, 777: 380±387.
Neuroimaging Methods. Advances in Experimental Medicine and Malonek, D. and Grinvald, A. Interactions between electrical
Biology. Plenum Press, New York, 1997: 167±175. activity and cortical microcirculation revealed by imaging spec-
Kuchinsky, W. and Paulson, O. B. Capillary circulation in the brain. troscopy: implications for functional brain mapping. Science,
Cerebrovasc. Brain Metab. Rev., 1992, 4: 261±286. 28 1996, 272: 551±554.
Kurosawa, M., Sato, A. and Sato, Y. Stimulation of the nucleus Mangold, R., Sokolo, L., Conner, E., Kleinerman, J., Therman, P. O.
basalis of Meynert increases acetylcholine release in the cerebral and Kety, S. S. The eects of sleep and lack of sleep on the cerebral
cortex in rats. Neurosci. Lett., 1989, 98: 45±50. circulation and metabolism of normal young men. J. Clin. Invest.,
Lacombe, P., Sercombe, R., Verrechia, C., Philipson, V., McKenzie, E. 1955, 34: 1092±1100.
T. and Seylaz, J. Cortical blood ¯ow increases induced by Maquet, P., Degueldre, C., Del®ore, G., Aerts, J., PeÂters, J. M., Luxen,
stimulation of the substantia innominata in the unaesthetized rat. A. and Franck, G. Functional neuroanatomy of human slow wave
Brain Res., 1989, 491: 1±14. sleep. J. Neurosci., 1997, 17: 2807±2812.
Lamarca, M. V. and Fibiger, H. C. Deoxyglucose uptake and choline Maquet, P., Dive, D., Salmon, E., Sadzot, B., Franco, G., Poirrier, R.
acetyltransferase activity in cerebral cortex following lesions of the and Franck, G. Cerebral glucose utilization during stage 2 sleep in
nucleus basalis magnocellularis. Brain Res., 1984, 307: 366±369. man. Brain Res., 1992, 571: 149±153.
Laureys, S., Goldman, S., Phillips, C., Van Bogaert, P., Aerts, J., Maquet, P., Dive, D., Salmon, E., Sadzot, B., Franco, G., Poirrier, R.,
Luxen, A., Franck, G. and Maquet, P. Impaired eective cortical von Frenckell, R. and Franck, G. Cerebral glucose utilization
connectivity in vegetative state preliminary investigation using PET. during sleep±wake cycle in man determined by positron emission
Neuroimage, 1999, 9: 377±382. tomography and [18F]2-¯uoro-2-deoxy-D-glucose method. Brain
Lecas, J. C. and Bloch, V. DonneÂes fournies par la rheÂoenceÂphalog- Res., 1990, 513: 136±143.
raphie continue dans le sommeil paradoxal du chat. J. Physiologie Maquet, P., Faymonville, M. E., Degueldre, C., Del®ore, G., Franck,
(Paris), 1969, 61 (Suppl. 1): 146. G., Luxen, A. and Lamy, M. Functional neuroanatomy of hypnotic
Leer, C. W., Armstead, W. M. and Shibata, M. Role of eicosanoids state. Biol. Psychiatr., 1999, 45: 327±333.
24 in cerebral hemodynamics. In: Phillis, J. W. (Ed) The Regulation of Maquet, P., Laureys, S., Peigneux, P., Fuchs, S., Petiau, C., Phillips,
Cerebral Blood Flow CRC Press, Boca Raton, FL, 1993: 297±313. C., Aerts, J., Del Fiore, G., Meulemans, T., Luxen, A., Franck, G.,
Lenzi, P., Cianci, T., Guidalotti, P. L. and Franzini, C. Regional spinal Van der Linden, M., Smith, C., Cleeremans, A. Experience-
cord blood ¯ow during sleep±waking cycle in rabbit. Am. J. dependent changes in cerebral activation during human REM sleep.
Physiology, 1986, 251: 957±960. Nat Neurosci., 2000, 3.
Lenzi, P., Cianci, T., Guidalotti, P. L., Leonardi, G. S. and Franzini, Maquet, P., PeÂters, J. M., Aerts, J., Del®ore, G., Degueldre, C., Luxen,
C. Brain circulation during sleep and its relation to extracerebral A. and Franck, G. Functional neuroanatomy of human rapid eye
hemodynamics. Brain Res., 1987, 415: 14±20. movement sleep and dreaming. Nature, 1996, 383: 163±166.
Lineberry, C. G. and Siegel, J. E. E. G. Synchronization, behavioral Maquet, P. and Phillips, C. Functional brain imaging of human sleep.
inhibition, and mesencephalic unit eects produced by stimulation J. Sleep Res., 1998, 7 (Suppl.1): 42±47.
of orbital cortex, basal forebrain and caudate nucleus. Brain Res., McBean, D. E., Sharkey, J., Ritchie, I. M. and Kelly, P. A. T. Evidence
1971, 34: 143±161. for a possible role for serotoninergic systems in the control of
Lou, H. C., Edvinsson, L. and McKenzie, E. T. The concept of cerebral blood ¯ow. Brain Res., 1990, 537: 307±310.
coupling blood ¯ow to brain function: revision required? Ann. McCormick, D. A. and Bal, T. Sleep and arousal: thalamocortical
Neurol., 1987, 22: 289±297. mechanisms. Annu. Rev. Neurosci., 1997, 20: 185±215.
Lydic, R., Baghdoyan, H. A., Hibbard, L., Bonyak, E. V., DeJoseph, Metherate, R., Cox, C. L. and Ashe, J. H. Cellular bases of neocortical
M. R. and Hawkins, R. A. Regional brain glucose metabolism is activation: modulation of neuronal oscillations by the nucleus
altered during rapid-eye movement sleep in the cat: a preliminary basalis and endogenous acetylcholine. J. Neurosci., 1992, 12: 4701±
study. J. Comp. Neurol., 1991, 304: 517±529. 4711.
Macchi, G., Bentivoglio, M. and Molinari, D. The thalamo-caudate Meyer, J. S., Ishikawa, Y., Hata, T. and Karacan, I. Cerebral blood
versus thalmo-cortical projections as studied in the cat with ¯ow in normal and abnormal sleep and dreaming. Brain Cognition,
¯uorescnet retrograde double labeling. Exp. Brain Res., 1984, 54: 1987, 6: 266±294.
225±239. Mignot, E., Nishino, S., Guilleminault, C. and Dement, W. C.
MacLean, P. D. The Triune Brain in Evolution: Role of Paleocerebral Moda®nil binds to the dopamine uptake carrier site with low
Functions. Plenum Press, New York, 1990. anity. Sleep, 1994, 17: 436±437.
Madsen, P. L., Holm, S., Vorstrup, S., Friberg, L., Lassen, N. L. and Milligan, D. W. A. Cerebral blood ¯ow and sleep state in the normal
Wildschiodtz, G. Human regional cerebral blood ¯ow during newborn infant. Early Human Dev., 1979, 3/4: 321±328.
rapid-eye-movement sleep. J. Cereb. Blood Flow Metab., 1991a, Mraovitch, S., Lasbennes, F., Calando, Y. and Seylaz, J. Cerebrovas-
11: 502±507. cular changes by electrical stimulation of the centromedian-para-
Madsen, P. L., Schmidt, J. F., Holm, S., Vorstrup, S., Lassen, N. A. fascicular complex in rat. Brain Res., 1986, 380: 42±53.
and Wildschiodtz, G. Cerebral oxygen metabolism and cerebral Mraovitch, S. and Seylaz, J. Metabolism-independent cerebral
blood ¯ow in man during light sleep (stage 2). Brain Res., 1991b, vasodilatation elicited by electrical stimulation of the centro-
25 557: 217±220. median±parafascicular complex in rat. Neurosci. Lett., 1987, 83:
Madsen, P. L., Schmidt, J. F., Wildschiodtz, G., Friberg, L., Holm, S., 269±271.
Vorstrup, S. and Lassen, N. L. Cerebral O2 metabolism and cerebral Mukhtar, A. I., Cowan, F. M. and Stothers, J. K. Cranial blood ¯ow
blood ¯ow in humans during deep sleep and rapid-eye-movement and blood pressure changes during sleep in the human neonate.
26 sleep. J. Appl. Physiol., 1991c, 70: 2597±2601. Early Human Dev., 1982, 6: 59±64.
Madsen, P. L. and Vorstrup, S. Cerebral blood ¯ow and metabolism Namba, H., Irie, T., Fukushi, K., Yamasaki, T., Tateno, Y. and
during sleep. Cerebrovasc. Brain Metab. Rev., 1991, 3: 281±296. Hasegawa, S. Lesion of the nucleus basalis magnocellularis does not

230 P. Maquet
aect cerebral cortical blood ¯ow in rats. Neurosci. Res., 1991, 12: Ramsay, S. C., Murphy, K., Shea, S. A., Friston, K. J., Lammertsma,
463±467. A. A., Clark, J. C., Adams, L., Guz, A. and Frackowiak, R. S. J.
Nishino, S. and Mignot, E. Neurobiology of narcolepsy. Neuroscien- Changes in global cerebral blood ¯ow in humans: eect on regional
tist, 1998, 4: 133±143. cerebral blood ¯ow during a neural activation task. J. Physiol.,
Nofzinger, E. A., Mintun, M. A., Wiseman, M., Kupfer, D. J. and 1993, 471: 521±534.
Moore, R. Y. Forebrain activation in REM sleep: an FDG PET Reddy, S. V. R., Yaksh, T. L., Anderson, R. E. and Sundt, T. M.
study. Brain Res., 1997, 770: 192±201. Eect in cat of locus coeruleus lesions on the response of cerebral
Nofzinger, E. A., Nichols, T. E., Kupfer, D. J. and Moore, R. Y. blood ¯ow and cardiac output to altered PaCO2. Brain Res., 1986,
Forebrain function during REM sleep in depression: [18F]-FDG 365: 278±288.
PET studies. Soc. Neurosci. Abstr., 1998, 24: 1519. Reivich, M., Isaacs, G., Evarts, E. and Kety, S. The eect of slow wave
Northington, F. J., Matherne, G. P. and Berne, R. M. Competitive sleep and REM sleep on regional cerebral blood ¯ow in cats.
inhibition of nitric oxide synthase prevents the cortical hyperemia J. Neurochem., 1968, 15: 301±306.
associated with peripheral nerve stimulation. Proc. Natl Acad. Sci. Richardson, B. S., Caetano, H., Homan, J. and Carmichael, L.
USA, 1992, 89: 6649±6652. Regional brain blood ¯ow in the ovine fetus during transition to
Norton, R. The eects of acute sleep deprivation on selective attention. the low-voltage electrocortical state. Dev. Brain Res., 1994, 81:
Br. J. Psychol., 1970, 61: 157±161. 10±16.
Nunez, A. Unit activity of rat basal forebrain neurons: relationship to Richardson, B. S., Patrick, J. E. and Abduljabbar, H. Cerebral
cortical activity. Neuroscience, 1996, 72: 757±766. oxidative metabolism in the fetal lamb: relationship to electrocor-
Ogata, S., Shiotsuka, S., Yamamoto, R., Takahashi, K., Hirasawa, H., tical state. Am. J. Obstet. Gynecol., 1985, 153: 426±431.
Igawa, M., Koyama, K., Kitamura, N., Nakabayashi, T., Saijyo, T., Roy, C. S. and Sherrington, C. S. On the regulation of the blood-
Maki, A., Yamashita, Y., Yamamoto, T., Koizumi, H., Atsumi, Y. supply of the brain. J. Physiol., 1890, 11: 85±108.
and Toru, M. Near-infrared measurement in sleep ± relation to Sadikot, A. F., Parent, A. and FrancËois, C. The centre meÂdian and
NREM sleep stages. Neuroimage, 1998, 7: S129. parafascicular thalamic nuclei project respectively to the sensori-
Orzi, F., Diana, G., Casamenti, F., Palombo, E. and Fieschi, C. Local motor and associative-limbic striatal territories in the squirrel
cerebral glucose utilization following unilateral and bilateral lesions monkey. Brain Res., 1990, 510: 161±165.
of the nucleus basalis magnocellularis in the rat. Brain Res., 1988, Sakai, F., Stirling Meyer, J., Karacan, I., Dermap, S. and Yamamoto,
462: 99±103. M. Normal human sleep: regional cerebral hemodynamics. Ann.
Pappius, H. M., Dadoun, R. and High, M. The eect of Neurol., 1980, 7: 471±478.
p-chlorophenylalanine on cerebral metabolism and biogenic amine Sanford, L. D., Tejani-Butt, S. M., Ross, R. J. and Morrison, A. R.
content of traumatized brain. J. Cereb. Blood Flow Metab., 1988, 8: Amygdaloid control of alerting and behavioral arousal in rats:
324±334. involvement of serotonergic mechanisms. Arch. Italiennes Biologie,
Parisi, R. A., Neubauer, J. A., Frank, M. M., Santiago, T. V. and 1995, 134: 81±99.
Edelman, N. H. Linkage between brain blood ¯ow and respiratory Santiago, T. V., Guerra, E., Neubauer, J. A. and Edelman, N. H.
drive during rapid-eye-movement sleep. J. Appl. Physiol., 1988, 64: Correlation between ventilation and brain blood ¯ow during sleep.
1457±1465. J. Clin. Invest., 1984, 73: 497±506.
Paulson, O. B. and Newmann, E. A. Does the release of potassium Scremin, O. U., Torres, C., Scremin, A. M. E., O'Neal, M., Heuser, D.
from astrocyte endfeet regulate blood ¯ow? Science, 1987, 237: 896± and Blisard, K. S. Role of nucleus basalis in cholinergic control of
898. cortical blood ¯ow. J. Neurosci. Res., 1991, 28: 382±390.
Paulson, O. B., Strandgaard, S. and Evinsson, L. Cerebral autoregu- Selemon, L. D. and Goldman-Rakic, P. S. Longitudinal topography
lation. Cerebrovasc. Brain Metab. Rev., 1990, 2: 161±192. and interdigitation of corticostriatal projections in the rhesus
Pigarev, I. N., Nothdurft, H. C. and Kastner, S. Evidence for monkey. J. Neurosci., 1985, 5: 776±794.
asynchronous development of sleep in cortical areas. Neuroreport, Seylaz, J., Pinard, E., Mamo, H., Goas, J. Y., Luft, A. and Correze, J.
1997, 8: 2557±2560. L. Human cerebral blood ¯ow during sleep. In: Ingvar D. H. and
Pilcher, J. S. and Hucutt, A. I. Eects of sleep deprivation on 30 Lassen N. A. (Eds) Brain Work: the Coupling of Function,
performance: a meta-analysis. Sleep, 1996, 19: 318±326. Metabolism, and Blood Flow in the Brain. Alfred Benzon Symposium
Pinard, E. Notion de couplage deÂbit-meÂtabolisme du cerveau: validiteÂ VIII. Munksgaard, Copenhagen, 1975: 235±252.
et limites. Agressologie, 1991, 32: 359±365. Shapiro, C. M. and Rosendor, C. Local hypothalamic blood ¯ow
Plihal, W. and Born, J. Eects of early and late nocturnal sleep on during sleep. Electroencephalogr. Clin. Neurophysiol., 1975, 39: 365±
declarative and procedural memory. J. Cogn. Neurosci., 1997, 9: 369.
534±547. Sherin, J. E., Shiromani, P. J., McCarley, R. W. and Saper, C. B.
Porkka-Heiskanen, T., Strecker, R. E., Thakkar, M., Bjokum, A. A., Activation of ventrolateral preoptic neurons during sleep. Science,
Greene, R. W. and McCarley, R. W. Adenosine: a mediator of the 1996, 271: 216±219.
sleep-inducing eects of prolonged wakefulness. Science, 1997, 276: Shulman, G., Fiez, J., Corbetta, M., Buckner, R. L., Miezin, F. M.,
1265±1268. Raichle, M. E. and Petersen, S. E. Common blood ¯ow changes
Portas, C. M., Krakow, K., Allan, P., Josephs, O. and Frith, C. across visual tasks: II. Decreases in cerebral cortex. J. Cogn.
Sensory processing across the sleep±wake cycle: a functional MRI Neurosci., 1997, 9: 648±663.
29 study in humans. Soc. Neurosci. Abstr., 1999, 25: 2065. Siegel, J. and Wang, R. Y. Electroencephalographic, behavioral, and
Raichle, M. E., Hartman, B. K., Eichling, J. O. and Sharpe, L. G. single-unit eects produced by stimulation of forebrain inhibitory
Central noradrenergic regulation of cerebral blood ¯ow and structures in cats. Exp. Neurol., 1974, 42: 28±50.
vascular permeability. Proc. Natl Acad. Sci. USA, 1975, 72: 3726± Smith, C. Sleep states and learning: a review of the animal literature.
3730. Neurosci. Biobehav. Rev., 1985, 9: 157±168.
Ramm, P. and Frost, B. J. Cerebral and local cerebral metabolism in Smith, C. Sleep states and memory processes. Behav. Brain Res., 1995,
the cat during slow wave and REM sleep. Brain Res., 1986, 365: 69: 137±145.
112±124. Smith, C. and Young, J. Reversal of paradoxical sleep deprivation by
Ramm, P. and Frost, B. J. Regional metabolic activity in the rat brain amygdaloid stimulation during learning. Physiol. Behav., 1980, 24:
during sleep±wake activity. Sleep, 1983, 6: 196±216. 1035±1039.

Squire, L. R. and Zola, S. M. Structure and function of declarative and Talairach, J. and Tournoux, P. Co-Planar Stereotaxic Atlas of the
nondeclarative memory systems. Proc. Natl Acad. Sci. USA, 1996, Human Brain. George Thieme-Verlag, Stuttgart, 1988.
93: 13515±13522. Toga, A. W. and Mazziotta, J. C. Brain Mapping. The Methods.
Staba, R. J., Fox, I. J., Behnke, E. J., Fields, T. A., MacDonald, K. Academic Press, San Diego, CA, 1996.
A., Bragin, A., Fried, I., Simmons, J., Mann, C., Wilson, C. L. Townsend, R. E., Prinz, P. N. and Obrist, W. D. Human cerebral blood
and Engel, J. Single unit recordings in human hippocampus and ¯ow during sleep and waking. J. Appl. Physiol., 1973, 35: 620±625.
entorhinal cortex during sleep. Abstr. Soc. Neurosci., 1998, 24: Trulson, M. E. Simultaneous recording of substantia nigra neurons
920. and voltammetric release of dopamine in the caudate of behaving
Steriade, M. and Amzica, F. Coalescence of sleep rhythms and their cats. Brain Res. Bull., 1985, 15: 221±223.
chronology in corticothalamic networks. Sleep Res. Online, 1998, 1: Underwood, M. D., Bakalian, M. J., Arango, V., Smith, R. W. and
1±10. Mann, J. J. Regulation of cortical blood ¯ow by the dorsal raphe
Steriade, M. Central core modulation of spontaneous oscillations and nucleus: topographical organization of cerebrovasucular regulatory
sensory transmission in thalamocortical systems. Curr. Opin. regions. J. Cereb. Blood Flow Metab., 1992, 12: 664±673.
Neurobiol., 1993, 3: 619±625. Villablanca, J. R., Marcus, R. J. and Olmstead, C. E. Eects of
Steriade, M., Contreras, D., Curro Dossi, R. and Nunez, A. The caudate nucleus or frontal cortex ablations in cats, II. Sleep±
slow (<1 Hz) oscillation in reticular thalamic and thalamocor- wakefulness, EEG, and motor activity. Exp. Neurol., 1976, 53:
tical neurons: scenario of sleep rhythm generation in interacting 31±50.
thalamic and cortical networks. J. Neurosci., 1993, 13: 3284± Villringer, A., Them, A., Lindauer, U., EinhauÈpl, K. and Dirnagl, U.
3299. Capillary perfusion in the rat brain cortex. An in vivo confocal
Steriade, M., Gloor, P., Llinas, R. R., Lopes de Silva, F. H. and microscopy study. Circ. Res., 1994, 75: 55±62.
Mesulam, M. M. Basic mechanisms of cerebral rhythmic activities. Werth, E., Achermann, P. and Borbely, A. A. Brain topography of the
Electroencephalogr. Clin. Neurophysiol., 1990a, 76: 481±508. human sleep EEG antero-posterior shifts of spectral power.
Steriade, M., Jones, E. G. and Llinas, R. R. Thalamic Oscillations and Neuroreport, 1996, 8: 123±127.
31 Signalling. John Wiley & Sons, New York, 1990b. Wiegand, M. A. A., M. and Lauer, C. J. Nocturnal sleep in
Steriade, M. and McCarley, R. W. Brainstem Control of Wakefulness Huntington's disease. J. Neurol., 1991, 238: 203±208.
and Sleep. Plenum Press, New York, 1990. Wilson, C. J. Generation of natural ®ring patterns in neostriatal
Steriade, M., McCormick, D. A. and Sejnowski, T. J. Thalamocortical 36 neurons. In: Arbuth Nott, G. W. and Emsson, P. C. (Eds) Progress
oscillations in the sleeping and aroused brain. Science, 1993a, 262: in Brain Research (Vol. 99). Chemical Signalling in the Basal
32 679±685. Ganglia. Elsevier, Amsterdam, 1994: 277±297.
Steriade, M., Nunez, A. and Amzica, F. A novel slow (<1 Hz) Yamamoto, R., Shiotsuka, S., Ogata, S., Takahashi, K., Hirasawa, H.,
oscillation of neocortical neurons in vivo: depolarizing and hyper- Igawa, M., Koyama, K., Kitamura, N., Nakabayashi, T., Saijyo, T.,
33 polarizing components. J. Neurosci., 1993b, 13: 3252±3265. Maki, A., Yamashita, Y., Yamamoto, T., Koizumi, H., Atsumi, Y.
Sterman, M. B. and Clemente, C. D. Forebrain inhibitory mech- and Toru, M. Hemodynamic changes during and around REM sleep
anisms: sleep patterns induced by basal forebrain stimulation in the measured by near-infrared spectroscopy. Neuroimage, 1998, 7: S128.
behaving cat. Exp. Neurol., 1962, 6: 103±117. Zeitlhofer, J., Anderer, P., Obergottsberger, S., Schimicek, P., Lurger,
Stickgold, R. Sleep: o-line memory processing. Trends Cogn. Sci., S., Marschnigg, E., Saletu, B. and Deecke, L. Topographic mapping
1999, 2: 484±492. of EEG during sleep. Brain Topogr., 1993, 6: 123±129.
Sullivan, C. E. Breathing in sleep. In: Orem, J. and Barnes C. D. (Eds) Zoccoli, G., Bach, V., Cianci, T., Lenzi, P. and Franzini, C. Brain
Physiology in Sleep. Academic Press, New York, 1980: 214±272. blood ¯ow and extracerebral carotid circulation during sleep in rat.
Szymusiak, R. Magnocellular nuclei of the basal forebrain: substrates Brain Res., 1994, 641: 46±50.
of sleep and arousal regulation. Sleep, 1995, 18: 478±500. Zoccoli, G., Bach, V., Nardo, B., Cianci, T., Lenzi, P. and Franzini, C.
Szymusiak, R. and McGinty, D. Sleep suppression following kainic Spinal cord blood ¯ow changes during sleep±wake cycle in rat.
acid-induced lesions of the basal forebrain. Exp. Neurol., 1986a, 94: Neurosci. Lett., 1993, 163: 173±176.
598±614. Zoccoli, G., Grant, D., Wild, J., Smolich, J. and Walker, A. Nitric
Szymusiak, R. and McGinty, D. Sleep-related neuronal discharge in oxide inhibition abolishes sleep±wake dierences in cerebro-vascu-
35 the basal forebrain of cats. Brain Res., 1986b, 370: 82±92. lar resistance. J. Sleep Res., 1998, 7 (Suppl. 2): 312.
Tachibana, S. Relation between hypothalamic heat production and Zoccoli, G., Lucchi, M. L., Andreoli, E., Bach, V., Cianci, T., Lenzi, P.
intra- and extracranial circulation factors. Brain Res., 1969, 16: 405± and Franzini, C. Brain capillary perfusion during sleep. J. Cereb.
416. Blood Flow Metab., 1996, 16: 1312±1318.

Functional Neuroimaging of Normal Human Sleep by Positron Emission Tomography

Uploaded by

Copyright:

Available Formats

Functional Neuroimaging of Normal Human Sleep by Positron Emission Tomography

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Functional Neuroimaging of Normal Human Sleep by Positron Emission Tomography

Uploaded by

Copyright:

Available Formats

J. Sleep Res.

Functional neuroimaging of normal human sleep

Accepted in revised form 23 April 2000; received 28 September 1999

Several neuroimaging techniques are now available to

Ó 2000 European Sleep Research Society 207

H215O scan technique and statistical parametric mapping

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Figure 1. Eects of condition-dependent

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

37 Table 1 Decreases in rCBF during SWS

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Figure 4. Analysis of the sleep data pub-

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

characteristics of dream recalls (Maquet et al. 1996). Their

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Interindividual variability should also be investigated sys-

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

Ó 2000 European Sleep Research Society, J. Sleep Res., 9, 207±231

You might also like

Figure 1. Eects of condition-dependent