Cortical network functional connectivity in the
descent to sleep
Linda J. Larson-Priora,1, John M. Zempelb, Tracy S. Nolana, Fred W. Priora, Abraham Z. Snydera,b,
and Marcus E. Raichlea,b,c,d,1
aDepartment of Radiology, Washington University School of Medicine, 4525 Scott Avenue, St. Louis, MO 63110;bDepartment of Neurology, Washington
University School of Medicine, St. Louis, MO 63110;cDepartment of Neuroscience, Washington University School of Medicine, St. Louis, MO 63110;
anddDepartment of Biomedical Engineering, Washington University School of Engineering, St. Louis, MO 63130
Contributed by Marcus E. Raichle, January 28, 2009 (sent for review June 6, 2008)
Descent into sleep is accompanied by disengagement of the con-
scious brain from the external world. It follows that this process
should be associated with reduced neural activity in regions of the
brain known to mediate interaction with the environment. We
examined blood oxygen dependent (BOLD) signal functional con-
nectivity using conventional seed-based analyses in 3 primary
sensory and 3 association networks as normal young adults tran-
sitioned from wakefulness to light sleep while lying immobile in
the bore of a magnetic resonance imaging scanner. Functional
connectivity was maintained in each network throughout all ex-
amined states of arousal. Indeed, correlations within the dorsal
attention network modestly but significantly increased during
light sleep compared to wakefulness. Moreover, our data suggest
that neuronally mediated BOLD signal variance generally increases
in light sleep. These results do not support the view that ongoing
BOLD fluctuations primarily reflect unconstrained cognition.
Rather, accumulating evidence supports the hypothesis that spon-
taneous BOLD fluctuations reflect processes that maintain the
integrity of functional systems in the brain.
default network ? fMRI ? neuroimaging ? non-rapid eye movement sleep
manifest subjectively as altered awareness and objectively as
reduced responsiveness to environmental stimuli. The electro-
physiological correlates of sleep are sufficiently pronounced and
characteristic as to be defining (1, 2). Thus, natural sleep is
characterized by a sequence of electroencephalographically de-
fined stages that may be broadly divided into nonrapid eye
movement (NREM) and rapid eye movement (REM) that
cyclically alternate throughout the sleep period.
NREM sleep cerebral blood flow and metabolism are reduced
in cortical association areas (3–7), as well as in the brainstem,
thalamus, basal ganglia, and basal forebrain (3, 4, 7). NREM
sleep is accompanied by reduced responsiveness to stimuli in
regions involved in executive function, attention, and perceptual
processing (5, 7, 8). The deepest NREM sleep states are
characterized by low frequency oscillations in the EEG during
which cognition is thought to be greatly reduced (9–13). During
REM, cerebral blood flow and metabolism remain decreased in
prefrontal and parietal regions but are increased in paralimbic
areas, anterior cingulate, and thalamus (3, 7, 14), a pattern
consistent with the emotionality and reduced logicality notable
in during dreaming (7, 15, 16). REM sleep is also marked by
atonia in skeletal muscles, reducing the ability to overtly respond
successively deeper stages of NREM and then REM sleep
progressively disengage the self from the environment.
It is now well-established that slow (?0.1 Hz) spontaneous
fluctuations of the blood oxygen dependent (BOLD) signal show
phase correlation in widely distributed functional networks (for
review see ref. 17). The topography of these networks has proven
here is a physiologically distinct change in the state of the
brain during sleep in comparison to wakefulness that is
to be highly consistent regardless of whether they are computed
by correlation against selected seed regions (17–19) or by blind
source separation methods (20–22). We here generically refer to
all such methods as functional connectivity MRI (fcMRI).
Remarkably, the networks obtained by fcMRI closely match the
topographies of functional responses obtained by task-related
fMRI using typical sensory, motor, and cognitive paradigms.
Thus, fcMRI-defined networks appear to be highly stable.
The assumption that spontaneous BOLD fluctuations repre-
sent uncontrolled cognition follows naturally from the well-
established relation between task-related responses and directed
cognition. By measuring functional signal correlations in human
subjects during the transition from wakefulness to sleep, we
directly tested this idea.
Three higher order functional systems located within associ-
(23), which acts to focus perceptual processes on selected
features of the environment (23, 24); the executive control
system (25), which governs overt responses particularly in cir-
cumstances of complex and potentially conflicting contingen-
cies; and the default system (26), which constitutes a set of
regions in which activity is suppressed relative to quiet wake-
fulness during performance of externally oriented tasks. Current
theoretical accounts of cognitive operations represented in the
default system emphasize social cognition, episodic memory,
and the construction of models of the external worlds (for a
recent comprehensive review see ref. 27). We also examined
fcMRI in 3 primary sensory systems (visual, auditory, and
somatomotor). Should fcMRI measures primarily reflect men-
tation, descent into sleep should be accompanied by significantly
reduced fcMRI in those networks supporting higher order
cognition with no change seen in primary sensory systems.
Ten healthy young adult subjects (22–24, 6 female) participated
course of the scan session. Two subjects returned for a second
night, one of whom again attained light NREM sleep. Thus, our
data set is composed of 6 sleep records and 5 nonsleep records.
Functional connectivity was examined using distributed net-
work seeds (Table 1, see Methods). The purpose of this study was
to evaluate defined functional networks for shifts in their
interregional connectivity in 2 different brain states. Thus, we
average network activity pattern to obtain the strongest measure
Author contributions: L.J.L.-P. and M.E.R. designed research; L.J.L.-P., J.M.Z., and T.S.N.
performed research; F.W.P. and A.Z.S. contributed new reagents/analytic tools ; L.J.L.-P.,
J.M.Z., F.W.P., and A.Z.S. analyzed data; and L.J.L.-P. wrote the paper.
The authors declare no conflict of interest.
1To whom correspondence may be addressed. E-mail: email@example.com or marc@
This article contains supporting information online at www.pnas.org/cgi/content/full/
www.pnas.org?cgi?doi?10.1073?pnas.0900924106 PNAS ?
March 17, 2009 ?
vol. 106 ?
no. 11 ?
of network, rather than regional, connectivity. Single seeded
network connectivity was also evaluated, producing similar
results to those found using the distributed network seed method
(Figs. S1–S3). We found that all 3 higher order systems main-
1 and S1). Further, these networks were consistent across
subjects (Fig. 1).
Primary sensory systems show little reduction in energy me-
tabolism or blood flow during light NREM sleep, and BOLD
responses are similarly unaffected (3, 4, 28). We evaluated 3
primary sensory systems (Table 1) for changes in connectivity
between wakefulness and light NREM sleep. As expected, these
systems maintained their connectivity structure in early sleep,
showing a consistent spatial pattern across subjects (Figs. 2
While all systems examined maintained their interregional
functional connectivity, the level at which this correlated activity
was maintained differed between systems (Fig. S5). Indeed, the
correlation values for the visual, auditory, and somatomotor
regions of interest (ROI) remained statistically unchanged be-
tween wake and sleep (Fig. 2 and Figs. S3 and S5). Contrary to
what one might predict, the dorsal attention network showed a
significant increase in BOLD correlations in light NREM sleep
relative to waking (P ? 0.015, n ? 36 [6 subjects, 6 ROIs]). As
one might expect, the executive control network decreased in
correlation strength in light sleep although this effect did not
reach statistical significance (P ? 0.08, n ? 18 [6 subjects, 3
ROIs]). Interestingly, the default network, like the sensory
systems, showed no change in its BOLD (P ? 0.83, n ? 18 [6
subjects, 3 ROIs]) correlation structure in light NREM sleep.
To verify that the distributed seed ROI technique generated
reliable results, we repeated the computations using as a seed
ROI the most prominent node in each functional system (dorsal
attention ? LIPS; default ? PCC; control ? LOP). This analysis
replicated the main findings in all particulars (Fig. S2). Paired t
tests again showed a significant increase in BOLD (P ? 0.03, n ?
30 [6 subjects, 5 ROIs]) correlation strength in dorsal attention
network ROIs. The executive control network illustrated a
nonsignificant reduction in correlation strength (P ? 0.3, n ? 12)
and the default network was again unchanged in temporal (P ?
0.5, n ? 18 [6 subjects, 3 ROIs]) correlation strength.
It has been reported that significant differences in BOLD
signal variance occur in NREM sleep (29, 30) that may represent
Table 1. Regional network seeds
Network Seed region name
DefaultPosterior parietal/precuneus (PCC) (19)
Left medial prefrontal (LMPFC) (19)
Right medial prefrontal (RMPFC) (19)
Left lateral parietal (LLP) (19)
Right lateral perietal (RLP) (19)
Left intraparietal sulcus (LIPS) (19)
Right intraparietal sulcus (RIPS) (19)
Left frontal eye fields (LFEF) (19)
Right frontal eye fields (RFEF) (19)
Left auditory (LAud) (53)
Right auditory (RAud) (53)
Left visual (LVC) (52)
Right visual (RVC) (52)
Right somatomotor (LSM) (54)
Right somatomotor (RSM) (54)
Dorsal anterior cingulate (dACC) (51)
Left operculum (LOP) (51)
Right operculum (ROP) (51)
ROIs used to create distributed network seeds are noted in bold. Numbers in parentheses are references.
There is wide-spread correspondence in the principal ROIs for each network across subjects (n ? 6), and network connectivity is maintained in sleep for all
Conjunction analysis of cognitive network seed correlations in wake (i) and light NREM sleep (ii). Seed ROIs are indicated by open circles (i) in all cases.
www.pnas.org?cgi?doi?10.1073?pnas.0900924106 Larson-Prior et al.
a BOLD signature of sleepiness (30). These studies focused
specifically on visual areas, which have been shown by others to
increased signal fluctuation was also noted in whole brain. We
evaluated changes in BOLD signal variance for both the whole
brain signal regressed out of our functional connectivity analysis
overall increase in variance in sleep relative to waking was seen
in the whole brain signal (Fig. 3Ai: wake 4.77 ? 2.0, sleep 5.59 ?
2.2, n ? 6, P ? 0.1). Signal variance in each cognitive system
component ROI generally increased in light NREM sleep (Fig.
3Aii), although this did not reach statistical significance. The
BOLD signal is illustrated for one subject in the transition from
wake to light NREM sleep for the default network ROI and is
typical of the level of signal variance in this transition for all
systems analyzed (Fig. 3B). The observed changes in signal
variance were not attributable to subject motion as subjects
exhibited significantly less movement during sleep (0.41 ? 0.06
mm, n ? 7) than in wake (0.69 ? 0.1 mm, P ? 0.0004). It remains
possible that the increase in whole brain BOLD signal variance
reflects altered patterns of respiration (32).
We analyzed the BOLD spectral content to determine
whether, like the EEG, it might exhibit a consistent shift in
spectral content toward slower frequencies during the transition
from waking to light NREM sleep (Fig. 3C). While there was a
tendency for a shift to lower frequency content in BOLD in the
descent to sleep in those systems that exhibited changes in
functional connectivity strength, this was statistically nonsignif-
icant when evaluated at 50 and 90% cumulative power (spectral
edge) by network (parametric; P ? 0.14 SE50, P ? 0.18 SE90;
nonparametric; P ? 0.44 SE50, P ? 0.30 SE90).
The notion that the brain progressively disconnects from the
external world as subjects fall asleep led us to hypothesize that
measures of functional connectivity should similarly decrease in
primary sensory and higher order cognitive systems if these
measures reflect active information processing. However, while
(open circles) were created in the left sensory cortex (see Table 1) and show strong cross-hemispheric connectivity across all subjects (n ? 6) in wake (i) that is
maintained in sleep (ii).
Conjunction analysis of cross-hemispheric connectivity in sensory systems showing that connections are maintained in light NREM sleep. Seed regions
steps (Aii, solid bars). Whole brain signal variance showed a statistically significant increase with descent to sleep. There was a general trend toward increased
in signal variance across state is illustrated. BOLD timecourses for each ROI in the default network are shown overlaid in the transition from wake (red) to sleep
(blue). (C) Analysis of the BOLD spectral content demonstrates a general trend toward lower frequency bins in sleep (dashed lines) that was not statistically
significant using either parametric (T, P ? 0.14, SE50, P ? 0.18, SE90, n ? 6 per network) or nonparametric (Wilcoxon rank sum, P ? 0.44 SE50, P ? 0.30 SE90,
n ? 6 per network) methods.
BOLD signal variance was examined in the regressed whole brain signal (Ai, whole brain) and in each distributed network ROI following all regressions
Larson-Prior et al.PNAS ?
March 17, 2009 ?
vol. 106 ?
no. 11 ?
there were some changes in functional connectivity in early
NREM sleep, most systems exhibited little or no change. Thus,
there was no evidence of reduced functional connectivity in the
sensory (visual, auditory, and somatomotor) or cognitive net-
works (dorsal attention, default and executive control) exam-
ined. In particular the default network, widely associated with
subjective awareness (33, 34), showed no measurable change in
that our small sample size (n ? 6) limits our ability to exclude
quantitatively small changes; thus, to achieve an 80% confidence
limit (? ? 0.05) for the observed effects size in the default
network would require an additional 1000 subjects. The only
network, which (24) would arguably be the most likely to show
reductions in network connectivity in sleep since it is known for
its role in attention to the external environment. While power
remained low (d ? 0.12, 70% confidence limit, ? ? 0.05), this
network actually increased its connectivity. Despite the limita-
tions in sample size, we are confident that if functional connec-
tivity does change in early sleep, the magnitude of such changes
is small. This is broadly consistent with previous reports on the
effects of light sleep on network connectivity in humans (30). We
conclude that the available data do not support the view that
intrinsic BOLD fluctuations primarily reflect conscious menta-
tion (17, 35). Rather, these intrinsic processes appear to exhibit
quasi-structural properties that are preserved across levels of
Like sleep, anesthetic states are ones in which volitional
cognitive processes are essentially abolished. Although repre-
expected to produce reductions in interregional network con-
nectivity. However, in keeping with the view that functional
connectivity is preserved across levels of arousal, a recent study
examining primate seed-based functional connectivity under
isoflurane anesthesia (36) found that functional network con-
nections based on spontaneous BOLD fluctuations were main-
tained under deep anesthesia.
Correlated activity within nodes of functional networks is
necessary for the generation of normal functional organization
during development (37–39). In addition, activity dependent
changes in synaptic weighting are believed to underlie experi-
ential learning throughout the life span, promoting the dynamic
reconfiguration of neural networks to meet changing sensori-
motor and cognitive processing demands (40–42). Sleep, par-
ticularly deep slow wave sleep, has been posited to represent a
mechanism by which changes in synaptic weighting accumulated
in wakefulness are homeostatically normalized (43, 44). Given
the ubiquitous nature of such processes in establishing and
dynamically regulating neural network activity, such correlated
activity may also be necessary for the maintenance of functional
organization throughout the lifespan.
Perhaps most intriguing is the difference in connection
strength between the functional networks examined in this study.
While the minimal change in cross-hemispheric sensory systems
was anticipated, the statistical lack of change in the default
network was surprising. Behaviorally, the default network was
defined on the basis of its disengagement from active cognitive
processing (26, 45), and others have reported reductions in
metabolic activity and blood flow in states of reduced conscious-
specific to a fronto-parietal network largely encompassing those
regions that define the default network. Thus, given the loss of
conscious volitional cognition represented by both sleep and
anesthesia, both the default and attentional networks might
logically be expected to reduce their connectivity in these states.
The fact that this connectivity is maintained, even strengthened
(as in the attentional network), suggests that the maintenance of
connectivity in these networks is fundamental to brain function.
The slight reduction in connectivity seen in the executive control
network may reflect the well-known disengagement of executive
control during sleep (6, 7).
Nonsignificant changes in BOLD signal variance in selected
spectral content toward lower frequencies (Fig. 3). These effects
qualitatively correspond to very well known electrophysiological
effects but they are much smaller in magnitude. Similar shifts in
BOLD spectral power have been described in human subjects
during anesthesia (48), where low frequency spectral power was
reported to coincide with changes in intraregional correlation
strengths. In our studies, these changes did not significantly
correlate with state.
It is possible that larger sample sizes, descent to deeper
NREM sleep, or REM sleep (where electrical activity more
closely resembles that of wake), may result in clear changes in
network connectivity. It is also possible that the abnormally
restricted and noisy environment in which subjects slept affects
connectivity in early sleep (for discussion, see SI Methods).
However, the maintenance of functional networks under general
anesthesia suggests that such connections, while they may
change, will not be completely abolished. That the connectivity
of interregional neural networks known to play a role in waking
state function is maintained across all examined states of con-
sciousness, suggests that maintenance of these network connec-
tions through ongoing spontaneous activity is of fundamental
importance to the living brain.
Subjects. Ten right-handed, healthy human subjects (ages 22–54, 6 females)
were recruited from the campus of Washington University under a protocol
approved by the University’s Human Studies Committee. All subjects gave
returned for a second sleep study.
30 ms, 4 mm3voxels, 2.013 sec/volume, 1 sec pause between frames) was
acquired using an EPI sequence locally modified to enhance the signal/noise
(1 ? 1 ? 1.25 mm) sagittal, T1-weighted magnetization-prepared rapid gra-
dient-echo scan. fMRI runs were 20 min (398 volumes) in duration. Sleep
latency, in most healthy subjects, falls within this time window (49). BOLD
acquisition continued without interruption during interrun intervals (45 sec,
restart recordings. This protocol ensured that subjects did not experience
abrupt (arousing) changes in the auditory environment. Sleep sessions were
conducted at night and included three to four 20 min runs. Sessions were
terminated when subjects indicated that they were either unable to continue
sleeping or were uncomfortable.
fMRI Data Preprocessing. fMRI data preprocessing included compensation
of slice-dependent time shifts and elimination of intensity differences in
even-odd slices resulting from interleaved acquisition, rigid body correction
for interframe head motion, intensity scaling (to whole brain modal value of
1000), and atlas registration by affine transformation (50). Each fMRI run was
transformed to atlas space and resampled to 3 mm3voxels.
Electroencephalography (EEG). Electroencephalography (EEG) data were
acquired simultaneously with fMRI (DC-3500 Hz, 20 KHz sampling rate) using
the MagLink™ (Compumedics Neuroscan) system (modified 10/20, 64 elec-
trodes) and the Synamps/2™ amplifier. Sixty-four EEG leads were placed in an
extended version of the International 10–20 system using the MagLink™ cap
(Compumedics Neuroscan), including an external cardiac lead (In Vivo Re-
CZ. Gradient artifact and ballistocardiogram were reduced using Scan 4.5 and
Curry 6.0 software respectively (Compumedics Neuroscan). Instantaneous
Hz; delta, 1–4 Hz) was computed for a 15 electrode transverse bipolar mon-
(see SI Methods). Data were also visually scored in 30 sec epochs by an
experienced observer (J.M.Z.) Fig. S6 according to standard criteria (1, 2). The
EEG was impacted by recording in the scanner bore and from artifact reduc-
www.pnas.org?cgi?doi?10.1073?pnas.0900924106Larson-Prior et al.
Fig. S7, SI Methods).
Analysis. Functional connectivity was assessed using methods described pre-
viously (19). The seed regions used to produce these maps are noted in Table
1, with those ROIs used to construct distributed network seeds noted in BOLD
type. Briefly, following regression of noise signals (whole brain signal, ven-
tricular signal, and white matter signal) (see ref. 19), the averaged BOLD time
series was extracted from 12 mm diameter spheric volumes centered on foci
defined by Talairach coordinates (Table 1, Fig. S8). The extracted seed time
series was then correlated to all other brain voxels to produce spatial corre-
lation maps. Correlation coefficients from each unique tested pair were used
to construct a correlation matrix used to evaluate correlations within and
between identified networks during quiet waking and sleep (Fig. S3), where
sleep was defined as those periods in which stable, stage 2 sleep was attained
(14.7–37.6 min, see Table S1). Seeds defined for the task positive attention
network (19) were centered on the intraparietal sulcus (IPS) and the frontal
were centered on the medial prefrontal cortex (MPF), the lateral parietal
cortex (LP), and the posterior cingulate/precuneus region (PCC). For the exec-
utive control network (51), seeds were centered on the dorsal anterior cingu-
late/medial superior frontal cortex (dACC) and on the bilateral insula/frontal
opercular region (LOP and ROP). Three additional seeds representing a dis-
(PCC?LLP?LMPF), the attention network (bilateral IPS ?FEF), and the exec-
utive control network (dACC?LOP?ROP). Seed regions of interest were also
constructed for primary visual (VC), auditory (Aud), and somatomotor (SM)
cortices (VC, ref. 52; Aud, ref. 53; SM, ref. 54). Results were calculated as
Fisher-z transformed correlation values and group data were evaluated using
the degree of agreement in connectivity maps for each network.
Statistical Analysis. Random effects analyses were performed on fcMRI group
data (P ? 0.01, multiple comparison corrected) and displayed using in-house
software. CARET brain mapping software (http://brainmap.wustl.edu/caret;
ref. 55) and the PALS human cortical atlas (56) were used to create display
maps based upon these spatial image maps. A repeated measures analysis of
variance (MANOVA) was performed to assess the effect of state on network.
This analysis yielded no significant effect of state F (5, 30)?0.342, P ? 0.6. For
cognitive networks, in which observations were not balanced, an analysis of
variance (ANOVA) was performed with network, state, and the interaction
between networks was found [F (2, 71) ? 61.08, P ? 0.001], but there was no
significant effect of state [F (1, 71)?0.28, P ? 0.6] or the interaction of network
and state [F (2, 71) ? 0.403, P ? 0.67] Planned comparisons of state in each
network are reported as paired t tests. Group data were analyzed using JMP
7.0 (SAS Insititute, Inc.). For analyses of BOLD variance, BOLD time series data
were extracted for seed and distributed seed ROI’s from the default, execu-
using an autocorrelation method and the mean psd was calculated across
subjects for each ROI (n ? 6 per network ROI). To clarify possible shifts in the
power plots for statistical testing of the spectral edge (SE), calculated at 50
(SE50), and 90 (SE90) percent cumulative power. Both parametric (Student’s t
test (T) and nonparametric (Wilcoxon signed rank) methods were used to
assess statistical differences in psd between waking and sleep states for each
network of interest.
ACKNOWLEDGMENTS. The authors thank Dr. C. Hildebolt for statistical assis-
NS006833 to M.E.R.
1. Rechtshaffen A, Kales A (1968) A Manual of Standardized Terminology, Techniques,
and Scoring System for Sleep Stages of Human Subjects (Brain Information Service/
Brain Res Inst., Univ. of California, Los Angeles).
2. Iber C, Ancoli-Israel S, Chesson AL, Quan SF (2007) The AASM Manual for the Scoring
3. Braun AR, et al. (1997) Regional cerebral blood flow throughout the sleep-wake cycle.
tomography. J Sleep Res 9:207–231.
5. Kjaer TW, Law I, Wiltschiotz G, Paulson OB, Madsen PL (2002) Regional cerebral blood
flow during light sleep—A H(2)(15)O-PET study. J Sleep Res 11:201–207.
6. Dang-Vu TT, et al. (2005) Cerebral correlates of delta waves during non-REM sleep
revisited. NeuroImage 28:14–21.
7. Maquet P, et al. (2005) Human cognition during REM sleep and the activity profile
within frontal and parietal cortices: A reappraisal of functional neuroimaging data.
Prog Brain Res 150:219–226.
8. Portas CM, et al. (2000) Auditory processing across the sleep-wake cycle: Simultaneous
EEG and fMRI monitoring in humans. Neuron 28:991–999.
9. Campbell KB, Colrain IM (2002) Event related potential measures of the inhibition of
information processing: II. The sleep onset period. Int J Psychophsyiol 46:197–214.
10. Massimini M, et al. (2005) Breakdown of cortical effective connectivity during sleep.
11. Steriade M, Nunez A, Amzica F (1993) Intracellular analysis of relations between the
slow (?1 Hz) neocortical oscillation and other sleep rhythms of the electroencepha-
logram. J Neurosci 13:33252–33265.
12. Timofeev I, Steriade M (1997) Low frequency rhythms in the thalamus of intact-cortex
and decorticated cats. J Neurophysiol 76:4152–4168.
13. Timofeev I, Grenier F, Bazhenov M, Sejnowski TJ, Steriade M (2000) Origin of slow
cortical oscillations in deafferented cortical slabs. Cereb Cortex 10:1185–1199.
14. Nofzinger EA, et al. (2002) Human regional cerebral glucose metabolism during
non-rapid eye movement sleep in relation to waking. Brain 125:1105–1115.
15. Hobson JA, Pace-Schott EF (2002) The cognitive neuroscience of sleep: Neuronal
systems, consciousness and learning. Nat Rev Neurosci 3:679–693.
dreams. Trends Cogn Sci 6:23–30.
17. Fox MD, Raichle ME (2007) Spontaneous fluctuations in brain activity observed with
functional magnetic resonance imaging. Nat Rev Neurosci 8:700–711.
cortex of resting human brain using echo-planar MRI. Magn Reson Med 34:537–541.
19. Fox MD, et al. (2005) The human brain is intrinsically organized into dynamic, anticor-
related functional networks. Proc Natl Acad Sci USA 102:9673–9678.
20. Beckmann CF, DeLuca M, Devlin JT, Smith SM (2005) Investigations into resting-state
connectivity using independent component analysis. Philos Trans R Soc Lond B
21. Damoiseaux JS, et al. (2006) Consistent resting-state networks across healthy subjects.
Proc Natl Acad Sci USA 103:13848–13853.
22. Mantini D, Perrucci MG, Del Gratta C, Romani GL, Corbetta M (2007) Electrophysio-
in the brain. Nat Rev Neurosci 3:201–215.
24. Husain M, Nachev P (2007) Space and the partietal cortex. Trends Cogn Sci 11:30–36.
architecture of top-down control. Trends Cogn Sci 12:99–105.
26. Raichle ME, et al. (2001) A default mode of brain function. Proc Natl Acad Sci USA
27. Buckner RL, Andrews-Hanna JR, Schacter DL (2008) The brain’s default network:
Anatomy, function, and relevance to disease. Ann NY Acad Sci 1124:1–38.
28. Czisch M, et al. (2002) Altered processing of acoustic stimuli during sleep: Reduced
auditory activation and visual deactivation detected by a combined fMRI/EEG study.
29. Fukunaga M, et al. (2006) Large-amplitude, spatially correlated fluctuations in BOLD
fMRI signals during extended rest and early sleep stages. Magn Reson Imaging
and light sleep: A simultaneous EEG-fMRI study. Hum Brain Mapp 29:671–682.
31. McAvoy M, et al. (2008) Resting states affect spontaneous BOLD oscillations in sensory
and paralimbic cortex. J Neurophysiol 100:922–931.
32. Birn RM, Diamond JB, Smith MA, Bandettini PA (2006) Separating respiratory-
variation-related fluctuations from neuronal-activity-related fluctuations in fMRI.
33. Laureys S (2005) The neural correlate of (un)awareness: Lessons from the vegetative
state. Trends Cogn Sci 9:556–559.
34. Gusnard DA, Akbudak E, Shulman GL, Raichle ME (2001) Medial prefrontal cortex and
Acad Sci USA 98:4259–4264.
35. Vincent JL, et al. (2006) Coherent spontaneous activity identifies a hippocampal-
parietal memory network. J Neurophysiol 96:3517–3531.
36. Vincent JL et al. (2007) Intrinsic functional architecture in the anesthetized monkey
brain. Nature 447:83–86.
37. Maffei A, Nelson SB, Turrigiano GG (2004) Selective reconfiguration of layer 4 visual
cortical circuitry by visual deprivation. Nat Neurosci 12:1353–1359.
38. Lu W, Constantine-Paton M (2004) Eye opening rapidly induces synaptic potentiation
and refinement. Neuron 43:237–249.
39. Butts DA, Kanold PO, Shatz CJ (2007) A burst-based ‘‘Hebbian’’ learning rule at
retinogeniculate synapses links retinal waves to activity-dependent refinement. PLoS
40. Adani Y, Sagi D, Tsodyks M (2002) Context-enabled learning in the human visual
system. Nature 415:790–793.
41. Hihara S, et al. (2006) Extension of corticocortical afferents into the anterior bank of
the intraparietal sulcus by tool-use training in adult monkeys. Neuropsychologia
42. Baeg EH, et al. (2007) Learning-induced enduring changes in functional connectivity
among prefrontal cortical neurons. J Neurosci 27:909–918.
Larson-Prior et al.PNAS ?
March 17, 2009 ?
vol. 106 ?
no. 11 ?
43. Tononi G, Cirelli C (2003) Sleep and synaptic homeostasis : A hypothesis. Brain Res Bull Download full-text
44. Tononi G, Cirelli C (2006) Sleep function and synaptic homeostasis. Sleep Med Rev
in cerebral cortex. J Cognit Neurosci 9:648–663.
47. Qiu M, Ramani R, Swetye M, Constable RT (2007) Spatial nonuniformity of the resting
CBF and BOLD responses to sevoflurane: In vivo study of normal human subjects with
magnetic resonance imaging. Hum Brain Mapp,10/1002/hbm. 20472.
48. Kiviniemi V, et al. (2005) Midazolam sedation increases fluctuation and synchrony of
the resting brain BOLD signal. Magn Reson Imaging 23:53–537.
49. Johns MW (2000) Sensitivity and specificity of the multiple sleep latency test (MSLT);
the maintenance of wakefulness test and the Epworth sleepiness scale: Failure of the
MSLT as a gold standard. J Sleep Res 9:5–11.
50. Ojemann JG, et al. (1997) Anatomic localization and quantitative analysis of gradient
refocused echo-planar fMRI susceptibility artifacts. NeuroImage 6:156–167.
52. Kang HC, Burgund ED, Lugar HM, Petersen SE, Schlaggar BL (2003) Comparison of
functional activation foci in children and adults using a common stereotactic space.
53. Lehmann C, et al. (2007) Dissociated lateralization of transient and sustained blood
oxygen level-dependent signal components in human primary auditory cortex. Neu-
54. Fox MD, Corbetta M, Snyder AZ, Vincent JL, Raichle ME (2006) Spontaneous neuronal
activity distinguishes human dorsal and ventral attention systems. Proc Natl Acad Sci
55. Van Essen D, et al. (2001) An integrated software suite for surface-based analyses of
cerebral cortex. J Am Med Inform Assoc 41:1359–1378.
56. Van Essen DC (2005) A population-average, landmark- and surface-based (PALS) atlas
of human cerebral cortex. NeuroImage 28:635–662.
www.pnas.org?cgi?doi?10.1073?pnas.0900924106 Larson-Prior et al.