J Neurophysiol 100: 922–931, 2008.
First published May 28, 2008; doi:10.1152/jn.90426.2008.
Resting States Affect Spontaneous BOLD Oscillations in Sensory
and Paralimbic Cortex
Mark McAvoy,1 Linda Larson-Prior,1 Tracy S. Nolan,1 S. Neil Vaishnavi,1 Marcus E. Raichle,1,2,3
and Giovanni d’Avossa2,4
Departments of 1Radiology, 2Neurology, and 3Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri;
and 4Department of Psychology, Bangor University, Wales, United Kingdom
Submitted 31 March 2008; accepted in final form 25 May 2008
INTRODUCTION
Neural activity is continuously modulated even during rest.
For example, scalp recorded EEGs provided initial evidence
that during wakeful rest the brain exhibits a posterior predominant, continuously modulated, 8- to 12-Hz activity, i.e., the
alpha rhythm (Adrian and Matthews 1934; Berger 1929, 1930).
Although the EEG correlates are well characterized, no comparable description exists for blood oxygenation level-dependent (BOLD) correlates of resting states. On the one hand,
studies of simultaneously recorded EEG and BOLD signals
have yielded somewhat contradictory results regarding the
temporal and spatial relation between BOLD amplitude and
alpha power. Although some have reported that BOLD activity
in the occipital cortices is positively correlated with ongoing
modulations of the EEG alpha power (Laufs et al. 2003), others
have reported a negative correlation (Feige et al. 2005; Goldman et al. 2002; Moosmann et al. 2003). Recent observations
have indicated that increases in alpha power are indeed assoAddress for reprint requests and other correspondence: M. McAvoy, Dept.
of Radiology, Washington Univ. School of Medicine, 4525 Scott Ave.,
Campus Box 8225, Rm. 2110, St. Louis, MO 63110 (E-mail: mcavoy
@npg.wustl.edu).
922
ciated with decreases in the BOLD signal in occipital and
parietal regions (de Munck et al. 2007), although this relation
shows very significant individual differences (Goncalves et al.
2006). Furthermore, BOLD correlates of spontaneous variations of alpha rhythm power have been found both in frontal
and parietal regions (Goncalves et al. 2006; Laufs et al. 2006),
despite the predominance of alpha rhythms over the posterior
occipital-parietal lobe.
On the other hand, functional connectivity studies (Biswal
et al. 1995; Cordes et al. 2001; Fransson 2005; Greicius et al.
2003; Hampson et al. 2002; Lowe et al. 1998) have provided
strong support for the notion that spontaneous BOLD oscillations reflect, at least in part, spontaneous oscillations in neural
activity. However, functional connectivity methods are also
sensitive to non-neural signals and therefore do not provide an
unbiased estimate of the value of neural correlations.
In this paper we took a different approach. Rather than pursuing
the BOLD correlates of EEG modulations or the pattern of
BOLD–BOLD correlations, we simply asked whether different
resting states are associated with changes in the power of the
ongoing BOLD signal. We characterized the BOLD signal in
terms of its frequency spectrum to detect regions where the
spectral density of the BOLD signal was affected by the resting
state type. Analysis in the frequency domain is ideally suited to
quantify the properties of the BOLD signal under steady-state
conditions and in the absence of externally paced events
(Damoiseaux et al. 2006). Finally, we present estimates of the
correlation between components of the BOLD signal that were
specifically modulated by resting state type and therefore were
likely to reflect purely neural effects. These data were used to
establish the temporal coherence of resting state effects across
regions.
METHODS
Subjects and paradigm
Ten healthy right-handed subjects (6 females; average age, 23.6 yr)
participated in a simultaneous BOLD functional MRI (fMRI)/EEG
experiment. The BOLD data were obtained from a public database
available at brainscape.org (study BS003). The experiment comprised
nine scans, each 5.5 min long. Before each scan, subjects were
instructed to either keep their eyes open, eyes closed, or maintain
fixation on a foveal crosshair for the duration of the scan. The order
in which the three conditions were run was randomized.
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked “advertisement”
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
0022-3077/08 $8.00 Copyright © 2008 The American Physiological Society
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
McAvoy M, Larson-Prior L, Nolan TS, Vaishnavi SN, Raichle
ME, d’Avossa G. Resting states affect spontaneous BOLD oscillations in sensory and paralimbic cortex. J Neurophysiol 100: 922–931,
2008. First published May 28, 2008; doi:10.1152/jn.90426.2008. The
brain exhibits spontaneous neural activity that depends on the behavioral state of the organism. We asked whether the blood oxygenation
level-dependent (BOLD) signal reflects these modulations. BOLD
was measured under three steady-state conditions: while subjects kept
their eyes closed, kept their eyes open, or while fixating. The BOLD
spectral density was calculated across brain voxels and subjects.
Visual, sensory-motor, auditory, and retrosplenial cortex showed
modulations of the BOLD spectral density by resting state type. All
modulated regions showed greater spontaneous BOLD oscillations in
the eyes closed than the eyes open or fixation conditions, suggesting
that the differences were endogenously driven. Next, we examined the
pattern of correlations between regions whose ongoing BOLD signal
was modulated by resting state type. Regional neuronal correlations
were estimated using an analytic procedure from the comparison of
BOLD–BOLD covariances in the fixation and eyes closed conditions.
Most regions were highly correlated with one another, with the
exception of the primary visual cortices, which showed low correlations with the other regions. In conclusion, changes in resting state
were associated with synchronous modulations of spontaneous BOLD
oscillations in cortical sensory areas driven by two spatially overlapping, but temporally uncorrelated signals.
RESTING STATES AFFECT SPONTANEOUS BOLD OSCILLATIONS
EEGs
923
冘
42
Imaging methods and preprocessing
BOLD data were downloaded from an electronic database publicly
available at brainscape.org (study BS003). The imaging data were
available in a preprocessed format. The acquisition parameters and
preprocessing steps are briefly summarized here. fMRI scans were
acquired with a Siemens 3 Tesla Allegra system (Erlagen, Germany).
An asymmetric spin-echo echo-planar-imaging (EPI) sequence was
used [repetition time (TR) ⫽ 3.013 s, which included a 1-s gap, echo
time (TE) ⫽ 25 ms, flip angle ⫽ 90°]. Each run was comprised of 110
volumes of 32 contiguous 4-mm axial slices (4 mm isotropic in plane)
with whole brain coverage. Structural images included one sagittal
MP-RAGE T1-weighted image (TR ⫽ 2.1 s, TE ⫽ 3.93 ms, flip
angle ⫽ 7°, 1 ⫻ 1 ⫻ 1.25 mm) and a T2-weighted fast spin-echo
image.
Image preprocessing included the following steps: 1) compensation
for slice-dependent time shifts, 2) elimination of odd/even slice
intensity differences caused by interpolated acquisition, 3) realignment of all data acquired in each subject within and across runs to
compensate for rigid body motion, and 4) normalization to a whole
brain mode value of 1,000 (Ojemann et al. 1997). The functional data
were transformed into atlas space (Talairach and Tournoux 1988) by
computing a sequence of affine transformations (1st frame EPI sequence to T2-weighted fast spin-echo to MP-RAGE to atlas representative target), which were combined by matrix multiplication.
Reslicing the functional data to 3-mm isotropic voxels in conformity
with the atlas than involved only one interpolation. For cross-modal
(i.e., functional to structural) image registration, a locally developed
algorithm was used (Rowland et al. 2005). Whole brain images after
preprocessing included 48 ⫻ 64 ⫻ 48 3-mm isotropic voxels.
Analysis: power spectral density and statistical maps
The analysis comprised three steps: detrending, estimation of spectral density, and statistical inference at the group level. Subject
specific general linear models (GLMs) (Friston et al. 1995) were used
to estimate the amplitude of the power spectrum (i.e., spectral density)
from each scan’s detrended data. To remove the linear drift from the
BOLD time series, each scan was detrended by fitting a constant and
linear trend to the acquired time series by ordinary least squares using
the following regression model
BOLDj共t兲 ⫽ c0j ⫹ c1jt
(1)
where c0j is the constant and c1jt is the linear trend for the jth scan.
The estimated linear trend c1jt was subtracted from the BOLD time
series. To estimate the spectral density, subject-specific GLMs were
fit to the detrended BOLD time series, obtained in the jth scan, by
ordinary least squares using the following model
J Neurophysiol • VOL
关ajk cos共2␲fkt兲 ⫹ bjk sin共2␲fkt兲兴
(2)
k⫽1
where c0j is the constant term and ajk and bjk are the coefficients of the
42 sine/cosine pairs (i.e., full Fourier basis set) modeling frequencies
from one full cycle (f1 ⫽ 0.0031 Hz) to 42 full cycles (f42 ⫽ 0.132 Hz)
determined by a practical Nyquist limit [i.e., 1/(2.5 ⫻ TR)]. Spectral
density, the amplitude of modulation at each frequency, was computed from the Fourier coefficients ajk and bjk using the following
formula
冑
冘
#scans
A共 f k兲 ⫽
共ajk2 ⫹ bjk2兲
j⫽1
#scans
,
k ⫽ 1, . . ., 42
(3)
The spectral density of the BOLD signal was computed separately for
scans in which subjects kept their eyes closed, eyes open, or maintained fixation. The spectral density was normalized by the mean of
the nine constant terms c0j (1 for each scan) and multiplied by 100 to
express it in percent BOLD signal change.
The spectral density maps were smoothed with a 6-mm full-widthat-half-maximum three-dimensional Gaussian kernel to blur interindividual differences in brain anatomy. The aim of the statistical
analysis was to detect regions in which the spectral density of ongoing
BOLD oscillations was modulated by the resting state type. For this
purpose, we used a group level statistic: a two-factor, repeatedmeasures, voxel-wise ANOVA. The random factor was subjects, and
the two fixed factors were resting state type, i.e., eyes open, eyes
closed, and fixation, and frequency, which contained 42 levels. The
ANOVA yielded three statistical maps: 1) the main effect of resting
state type (F2,18), which reflects the effect of behavioral condition on
the overall variance of the ongoing BOLD signal; 2) the resting state
type by frequency interaction map (F41,738), which highlights those
regions in which the behavioral state affects the amplitude of
the ongoing BOLD signal in a frequency dependent fashion; and 3) the
main effect of frequency (F41,369), which reflects the deviation of
the spectral density of the ongoing BOLD signal from that of white
noise whose spectral density is constant across the frequency spectrum. The voxel-wise statistical maps (main effect of resting state
type, interaction of resting state type by frequency, and the main effect
of frequency) were gaussianized and corrected for multiple comparisons (z ⱖ 3.0, minimum 45 face connected voxels, P ⬍ 0.05
corrected) with a Monte Carlo based method (Forman et al. 1995;
McAvoy et al. 2001).
Regional analyses
Fifteen regions of interest (ROIs) were defined on the basis of the
multiple comparisons– corrected resting state type by frequency interaction map (Fig. 1; Table 1). ROIs were drawn by hand. All regional
statistics including variances and covariances were based on the BOLD
time series averaged across all voxels within a region. Repeated-measures ANOVAs and Student’s t-test were used to determine the
significance of differences in pairwise regional BOLD–BOLD correlations among resting state types. For this purpose, the correlation
coefficients were first z-transformed using a standard procedure
(Fisher 1921).
Principal component analysis
The estimated neural correlations were entered into a 15 ⫻ 15
matrix, i.e., one element for each possible regional pairing. To
examine how many components accounted for the neural effects, the
eigenvectors and eigenvalues of the correlation matrix were computed. This is simply a principal components analysis. The decom-
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
In 9 of 10 subjects, EEG data were acquired simultaneously with
fMRI (DC-3,500 Hz, 20-kHz sampling rate) using the MagLink
system and the synamps/2 amplifier (Compumedics Neuroscan) with
32 electrodes (modified 10/20), including bipolar, vertical eye, and
cardiac leads. The reference electrode was located between Pz and Cz
and the ground electrode between Fz and FPz. Gradient artifact was
reduced using Scan 4.3 software (Compumedics Neuroscan) with
ballistocardiogram artifacts reduced using in-house software (Vincent
et al. 2007). These procedures also allowed the detection of the EKG
epochs and the estimation of the duration of the interbeat interval.
EEG was inspected both visually and using spectral information in the
alpha, theta, and delta bands to ensure that subjects were not asleep.
The effects of resting state on the mean and SD of the interbeat
interval were assessed statistically using repeated-measures ANOVAs
that included subjects as the random factor and resting state type as
the fixed factor.
detrended BOLDj共t兲 ⫽ c0j ⫹
924
MCAVOY ET AL.
position of the correlation matrix yields a set of eigenvalues that
represent the proportion of the overall variance accounted by the
corresponding eigenvector and a set of eigenvectors whose scalar
elements represent the weighted contribution of each regional BOLD
response to that component. The product of the scalar elements of the
eigenvector and its eigenvalue gives the proportion of the BOLD
variance accounted by the eigenvector for each region.
Measurement of end-tidal CO2
Three male subjects performed 5.5 min of eyes closed and eyes
open rest (randomized order between subjects) while lying supine in
a mock MR scanner. Continuous end-tidal CO2 was collected via
nasal cannula and measured using the InVivo Magnitude (InVivo,
Orlando, FL) patient monitoring system. Data were recorded at 100
Hz using PowerLab Chart Software (ADInstruments, Colorado
Springs, CO). The protocol was approved by the Washington University Institutional Review Board.
To facilitate comparison with the BOLD signal, the computation of
spectral density was carried similarly to the analysis of the BOLD
data. First, the end-tidal CO2 measurements were interpolated by
means of cubic splines (interp1, Matlab v5.2, The MathWorks,
Natick, MA) to the BOLD data TR of 3.013 s. The interpolated time
series was detrended and fit with the full Fourier basis set, which
exactly replicated the model used to analyze the BOLD data (see Eqs.
1 and 2). To statistically assess changes in end-tidal CO2 across
resting state types, a repeated-measures ANOVA was performed on
J Neurophysiol • VOL
the spectral density (see Eq. 3). Subjects were the random factor and
resting state type and frequency were the fixed factors.
RESULTS
Spontaneous BOLD activity is affected by resting state type
The main effect of resting state type highlighted a subset of
regions also found within the interaction of resting state type
by frequency map; hence we present only the latter. Figure 1A
shows representative axial slices obtained from the resting state
type by frequency interaction map. Displayed values are ztransformed F statistics. Several regions in the occipital lobe,
including primary visual cortex, the lateral occipital and MT
complexes, fusiform gyri, and dorsal occipital-parietal regions
were highlighted. Furthermore, regions outside visual striate
and extrastriate cortex were found, whose ongoing BOLD
activity modulated as a function of resting state type. These
included left and right retrosplenial cortex, left auditory cortex,
and bilateral sensory-motor cortices. Peak coordinates, center
of mass, and region sizes are listed in Table 1. A notable
feature of these data are that the effect of resting state type in
visual areas seems to replicate the anatomical pattern of activation evoked by a visual stimulus, with clear boundaries
between separate visual regions, such as primary visual cortex,
lateral occipital, and MT as shown in Fig. 1B. However, the
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
FIG. 1. Statistical map of the resting state
type by frequency interaction. A: regions
whose spectral density depended on whether
the subject rested with eyes closed, eyes
open, or while maintaining fixation included
sensory-motor, auditory, and primary visual
cortex, as well as secondary and tertiary
visual cortices and retrosplenial cortex.
Therefore the effects of resting state type on
spontaneous blood oxygenation level-dependent (BOLD) oscillations are focal and engage exclusively a network of sensory and
paralimbic regions. B: the same data are
presented overlaid on the flattened cortical
surface (Van Essen 2005; http://brainmap.
wustl.edu/caret). Displayed statistics are
gaussianized F statistics.
RESTING STATES AFFECT SPONTANEOUS BOLD OSCILLATIONS
TABLE
925
1. Peak coordinates of regions identified in Fig. 1
Peak Coordinates
Center of Mass
Abbreviation
x
y
z
x
y
Left sensory-motor
Right sensory-motor
Left ventral intraparietal sulcas/visual area 3a
Right ventral intraparietal sulcas/visual area 3a
Left visual cortex
Right visual cortex
Left lateral occipital
Right lateral occipital
Left medial temporal
Right medial temporal
Left retrosplenial cortex
Right retrosplenial cortex
Left fusiform gyrus
Right fusiform gyrus
Left auditory cortex
LSM
RSM
LVipV3a
RVipV3a
LV1
RV1
LLO
RLO
LMT
RMT
LrSplen
RrSpen
LFus
RFus
LAud
⫺35
35
⫺23
26
⫺5
11
⫺20
44
⫺47
38
⫺14
20
⫺14
20
⫺50
⫺42
⫺39
⫺81
⫺93
⫺90
⫺84
⫺90
⫺90
⫺78
⫺69
⫺63
⫺51
⫺63
⫺51
⫺24
54
63
24
27
6
9
12
0
9
6
⫺6
⫺3
⫺9
⫺9
18
⫺37
37
⫺18
23
⫺5
11
⫺18
36
⫺43
44
⫺13
17
⫺15
21
⫺49
⫺35
⫺37
⫺87
⫺91
⫺89
⫺85
⫺91
⫺87
⫺77
⫺70
⫺61
⫺56
⫺60
⫺56
⫺24
z
49
60
27
27
6
10
12
6
9
8
⫺5
⫺5
⫺11
⫺12
18
Volume, mm3
2,295
1,620
2,106
1,269
3,591
1,539
1,242
1,404
1,242
1,782
2,241
3,456
2,241
3,672
2,376
Peak coordinates and center of mass are given in mm according to the atlas of Talairach and Toumoux (1988).
effects of behavioral state cannot be trivially attributed to
visually evoked responses being added to the ongoing activity
for two reasons. First, our data clearly indicate that areas
beyond primary and extrastriate visual cortex (e.g., sensorymotor cortex and auditory cortex) modulated their ongoing,
spontaneous BOLD signal according to resting state type.
These areas are not known to respond to passive visual stimulation. Second, although intermittent visual stimulation and
hence phasic BOLD responses could have been introduced by
spontaneous eye movements and eye blinks, this should result
in greater modulation of spontaneous BOLD activity during the
eyes open condition than the eyes closed condition. This was
not the case. Figure 2C shows the spectral density of the BOLD
signal in the left occipital pole, which probably corresponds to
the foveal representation of primary visual cortex. In this
region, the comparison of the spectral densities indicates that
the amplitude of spontaneous modulation of the BOLD signal
was greater when subjects kept their eyes closed than during
eyes open or fixation. Table 2 reports the values of the
temporal variance of the BOLD signal in right visual cortex in
the three different resting state types for all 10 subjects.
Variance is a simpler measure of variability than the spectral
FIG. 2. Spectral density of regional ongoing BOLD signals for 3 different resting state
types averaged over 10 subjects. Regions
include right sensory-motor (A), left auditory
(B), left visual (C), and right retrosplenial
(D) cortex, selected from the map shown in
Fig. 1. Across these and all the other regions,
the eyes closed condition was associated
with greater modulation of the ongoing
BOLD signal than the eyes open and fixation
conditions. These effects were mostly limited to 3 frequency bands centered at 0.01,
0.03, and 0.04 Hz. Error bars are standard
error computed across the 10 subjects.
J Neurophysiol • VOL
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
Region
926
TABLE
MCAVOY ET AL.
2. Subject variance for RV1 and LrSplen
RV1
LrSplen
Subject
Fixation
Eyes Open
Eyes Closed
Fixation
Eyes Open
Eyes Closed
1
2
3
4
5
6
7
8
9
10
Mean
SE
0.3205
0.3984
0.3479
0.5516
0.2601
0.6739
0.2056
0.2814
0.3427
0.2022
0.3584
0.0476
0.3203
0.8841
0.3032
0.5547
0.2638
0.6970
0.1691
0.4816
0.3148
0.2845
0.4273
0.0710
0.4805
1.3643
0.3256
1.1225
0.3177
1.4911
0.3549
0.2873
0.3227
0.5487
0.6615
0.1498
0.1551
0.2289
0.5943
0.2013
0.1623
0.1222
0.3210
0.1800
0.8862
0.1429
0.2994
0.0786
0.1324
0.4749
0.5322
0.1909
0.2293
0.1968
0.3584
0.2304
0.5356
0.2361
0.3117
0.0479
0.2839
0.4765
0.6945
0.4274
0.3834
0.2880
0.9740
0.1709
1.1844
0.2093
0.5092
0.1072
density, which breaks the signal in frequency-dependent components, and it shows that, in this particular region, the BOLD
signal power almost doubled in the eyes closed condition
compared with the fixation condition. Moreover, greater amplitude of spontaneous BOLD oscillations in the eyes closed
condition was not limited to primary visual cortex but generalized to all areas that showed a significant interaction of
resting state type by frequency, without exceptions. To illustrate this point, spectral densities from regions in right sensorymotor cortex, left auditory cortex, and right retrosplenial cortex
are shown in Fig. 2, A, B, and D. We conclude that visual
stimulation introduced by eye movements and blinks could
hardly account for the spectral differences, because one would
expect greater BOLD modulations during eyes open than eyes
closed, at least in visual areas. On the contrary, our data
showed consistent decreases in BOLD power across a set of
sensory areas for several modalities in the fixation and eyes
open conditions compared with the eyes closed condition.
Therefore BOLD modulations by resting state type are most
likely driven by endogenous signals rather than slavishly reflecting the sensory input.
Interestingly, the spectral density plots shown in Fig. 2,
A–D, also illustrate two other points. First, the spectral density
of the BOLD signal is significantly nonuniform, in agreement
with previous observations (Purdon and Weisskoff 1998;
Weisskoff 1996; Zarahn et al. 1997). We also found that the
power distribution of the BOLD signal had a 1/f distribution
not only in gray matter but across the entire brain (data not
shown). Second, the modulation of the ongoing BOLD activity
by resting state type has a complex, band-pass behavior; the
effects were mainly restricted to at least three bands with center
frequencies at 0.01, 0.03, and 0.04 Hz, respectively.
Effects of resting state type are by two spatially overlapping,
but temporally uncorrelated, oscillatory modes
To further characterize the effects of resting state on the
ongoing BOLD signal, we developed a procedure to estimate
the correlation between neural components of the BOLD response in regions showing an interaction of resting state type
by frequency. The raw correlation coefficients measure the
similarity between simultaneous BOLD signals recorded from
separate brain regions. A difficulty in interpreting these
BOLD–BOLD correlations is that the BOLD time series not
J Neurophysiol • VOL
only reflects local neural activity but also other physiological
and nonphysiological signal sources. Hence correlations based
on the BOLD time series provide a biased estimate of neural
correlations. To overcome this limitation, we used a simple
model of the regional BOLD variances and pairwise covariances
to remove the effects of non-neural signals (see APPENDIX). This
model assumes that 1) there exist both neural and non-neural
sources of BOLD variability, 2) sources of variability unrelated
to local neural activity are not affected by resting state type,
and 3) non-neural sources of variability are uncorrelated with
neural sources of BOLD variability. To show the effect of
resting state on raw BOLD–BOLD correlations, we present the
correlations between left retrosplenial cortex and right occipital
cortex with the remaining regions, because these two regions
show somewhat different results.
The values of the pairwise regional BOLD–BOLD correlations between the left retrosplenial region and all the other
regions in the eyes closed and fixation conditions as well as the
estimated neural correlations are shown in Fig. 3A. BOLD–
BOLD correlations with left retrosplenial cortex were found to
vary significantly over regions (F13,117 ⫽ 20.26, P ⫽ 3.79e⫺24)
and to differ significantly between resting state types (F1,9 ⫽
6.61, P ⫽ 0.030). A significant interaction of region by resting
state type (F13,117 ⫽ 6.06, P ⫽ 1.58e⫺8) also indicated that the
effects of resting state type on the correlation with the left
retrosplenial region strongly depended on the region. For each
regional pairing, BOLD–BOLD correlations in the eyes closed
and fixation conditions were statistically compared using a
two-way, paired Student’s t-test. A single asterisk indicates a
significant difference at the 0.05 level and a double asterisk at
the 0.01 level. The black bars report the correlation between
the components of the BOLD signal specifically modulated by
resting state type.
Figure 3B shows the pairwise regional BOLD–BOLD correlations between right primary visual cortex region and all
other regions in the eyes closed and fixation conditions. The
estimated correlation between signal components present in the
eyes closed but not in the fixation condition, i.e., the neural
component, is also shown. Although the BOLD–BOLD correlations were found to vary regionally (F13,117 ⫽ 25.22, P ⫽
5.079e⫺28), the interaction of region by resting state type was
weak (F13,117 ⫽ 1.82, P ⫽ 0.047), and no significant difference
was found between correlations in the eyes closed and fixation
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
All values are calculated from the BOLD time series in units of percent change of baseline. RV1, right visual cortex; LrSplen, left retrosplenial cortex.
RESTING STATES AFFECT SPONTANEOUS BOLD OSCILLATIONS
927
procedure assessed the regional coherence of simultaneous
changes in BOLD signal. Figure 4 shows the proportion of
regional variance accounted for by each of the 15 eigenvectors
for each of the 15 regions. For all but two of the regions, the
first eigenvector accounted for 55% or more of the overall
neural modulation. The first and second eigenvectors accounted for a comparable proportion of the overall variance
(30 –50% each) in left and right visual cortex regions. We
conclude that, in the eyes closed condition, most of the increase
in activity was caused by two oscillatory modes: one that was
synchronous across all regions and a second that was synchronized between left and right occipital cortex but did not
generalize across the entire network.
FIG. 3. Pairwise correlations between (A) the left retrosplenial region and
all other regions and (B) right striate cortex and all other regions. Gray and
empty bars are the group averaged correlations for the fixation and eyes closed
conditions, respectively. The black bars represent the estimated correlation for
the neural signal component modulated by resting state type. The significance
of the difference between the fixation and eyes closed condition was evaluated
with a 2-way, paired Student’s t-test (*P ⬍ 0.05, **P ⬍ 0.01).
conditions (F1,9 ⫽ 0.023, P ⫽ 0.88). We conclude that the
effect of resting state type on the correlation between right
visual cortex and other regions was small and of uncertain
significance.
The median value and its 95% CI for the regional correlations of the component of the BOLD signal that was modulated
by resting state type were 0.67 and 0.64 – 0.70, respectively.
These values are higher than those reported in previous studies
of regional BOLD correlations, based on the same data (Fox
et al. 2005), suggesting that the procedure successfully isolated
signal components related to changes in neural activity across
resting states. The finding that spontaneous local field potentials also exhibit equally high temporal coherence in nonhuman
primates (Leopold et al. 2003) is consistent with this viewpoint.
To determine whether there is a dominant component driving neural activity in the eyes closed condition relative to the
fixation condition, the eigenvectors and eigenvalues of the
matrix of estimated neural correlations were computed. This
J Neurophysiol • VOL
A potential confound is the effect of physiological variables
on cerebral blood flow and therefore BOLD. If physiological
noise changes across resting state types, contrary to our assumption, the observed BOLD effects may not be ascribed to
changes in neural activity, contrary to the viewpoint expressed
here. To address this point, we assessed whether heart rate, a
source of physiological noise, was affected by resting state
type, by comparing the mean interbeat interval and variability of
the interbeat interval across resting state types. The ballistocardiogram was recovered from the scanner record using a set
of procedures detailed in a previous publication (Vincent et al.
2007). The mean of the duration of the interbeat interval,
averaged across subjects, was 0.960, 0.938, and 0.938 s in the
eyes closed, eyes open, and fixation conditions, respectively.
Likewise, the SD of the duration of the interbeat interval
averaged across subjects was 0.0858, 0.0896, and 0.0797 s in
the eyes closed, eyes open, and fixation conditions, respectively. Repeated-measures ANOVAs showed that neither the
mean interbeat interval or the SD of the interbeat interval was
affected by resting state type (F2,16 ⫽ 2.68, P ⫽ 0.0993 and
FIG. 4. Proportion of each region’s BOLD variance accounted for by each
eigenvector. The filled diamond and square is the proportion of regional BOLD
variance in left and right visual cortex, respectively, accounted by each of the
eigenvectors. Unfilled circles represent the data for all the other regions. The
rank of the eigenvectors increases along the horizontal axis. The first eigenvector accounted for ⬎55% of the variance in all but 2 regions, which were left
and right striate visual cortex. The second eigenvector accounted for 40 –50%
of the variance in the 2 occipital regions and ⬍15% in the remaining regions.
None of the other components accounted for ⬎25% of the variance in any of
the regions, suggesting that the first and second eigenvectors carried most
information relevant for describing the instantaneous pattern of BOLD oscillations across the cortical network.
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
Non-neuronal sources of noise are not affected by resting
state type
928
MCAVOY ET AL.
DISCUSSION
There is a large body of literature on resting state BOLD (for
reviews, see Fox and Raichle 2007; Gusnard and Raichle
2001). Much work has concentrated on a network of regions
that are active during wakeful rest in absence of a task
(Shulman et al. 1997), known as the default mode regions
(Raichle et al. 2001). In this study, we identified the brain
regions whose spontaneous BOLD activity is modulated by
resting state type, that is, whether subjects kept their eyes open,
closed, or maintained fixation, rather than by the comparison of
task versus rest condition. Overall, our results highlighted a
network of sensory and paralimbic regions where the amplitude of the ongoing, spontaneous BOLD oscillation depends on
resting state type. This network did not overlap with the default
network except for possibly retrosplenial cortex (Raichle et al.
2001).
FIG. 5. Spectral density of the group averaged end-tidal CO2. End-tidal
CO2 data were obtained in 3 subjects lying in a mock scanner while resting
with their eyes open or closed. The ordinate is the amplitude of CO2 oscillations, the abscissa the frequency. No consistent difference between the amplitude of end-tidal CO2 oscillations were observed in the frequency range ⬍0.04
Hz, where most of the effects of resting state type on the BOLD signal were
found.
J Neurophysiol • VOL
Effects of resting state type are endogenously driven
The first noteworthy result is that resting state type affects
the ongoing BOLD activity in several sensory and paralimbic
regions in the posterior portion of the brain. Prior studies of
resting state effects on spontaneous BOLD oscillations have
not reported any difference between eyes open and eyes closed
wakeful rest. Recently, it has been noted that light sleep is
associated with increased BOLD oscillations in visual cortex
and across the entire brain compared with resting with eyes
open (Fukunaga et al. 2006). Horovitz et al. (2008) have found
a more focal anatomical pattern involving visual cortex, cingulate gyrus, superior temporal gyrus, lateral and medial sensory-motor regions, and inferior parietal and frontal cortex.
Our data suggest a similar, but even more restricted, anatomical pattern for the effects of resting state type during wakefulness. In fact, we found that only sensory-motor, auditory,
primary, and extra-striate visual cortices, as well as retrosplenial cortex (see Fig. 1), showed significant increases of their
ongoing BOLD activity during the eyes closed condition compared with the eyes open or fixation conditions. The latter two
conditions differed little, with fixation showing somewhat
lower amplitude modulation of the ongoing BOLD signal than
eyes open (see Fig. 2, A–D). These results clearly rule out the
possibility that the effects of resting state type may arise
spuriously as a consequence of intermittent visually evoked
responses caused by eye movements or blinks, since this
should have resulted in greater BOLD modulation during eyes
open and fixation than during eyes closed, contrary to what we
found. Could changes in the amplitude of neural and BOLD
modulation simply result from changes in the mean level of
activity? It is known that visually responsive neurons exhibit a
positive relation between average spike counts and spike count
variance (reviewed in Koch 1999). Therefore the finding that
modulations of the BOLD signal are increased in the eyes
closed condition compared with the eyes open and fixation
conditions may simply reflect that neural activity and its
modulation is greater in the former than the latter conditions.
This hypothesis is not supported by a previous fMRI study in
which resting state conditions were alternated within scans
rather than between scans. In fact, Feige et al. (2005) have
shown that the average BOLD signal increases during eyes
open compared with eyes closed in visually responsive areas.
Hence it is quite unlikely that differences in the baseline level
of neural activity accounted for differences in the amplitude of
modulation around the baseline. Finally, a potential concern is
that the effects of resting state type on the BOLD signal could
be caused by changes in physiological noise rather than neural
activity. Previous work has shown that the effects of physiological sources of noise on the spontaneous BOLD signal are
widespread and have been suggested to closely match the
anatomy of the default system (Birn et al. 2006). However,
estimates of the contribution of physiological noise to the
spontaneous BOLD oscillations in gray matter have indicated
that it accounts for a small proportion of the overall signal. For
example, Wise et al. (2004) measured the effect of variation of
end-tidal CO2 on BOLD signal variability. They found that
end-tidal CO2 accounted for 6.5% of the overall voxel-wise
BOLD variance in gray matter voxels and 25% of the overall
BOLD variance for the signal computed averaging over the
entire brain. Similar estimates have been obtained by others
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
F2,16 ⫽ 1.39, P ⫽ 0.277, respectively). This result suggests that
resting state type had no major effects on the breathing cycle,
since if it did, this should have had, in turn, consequences on
the cardiac rhythm (Anrep et al. 1936).
To characterize the possible contribution of CO2 variations
to the effects of resting state on the BOLD signal, we measured
end-tidal CO2 during the eyes closed and eyes open conditions
in three subjects lying in a mock scanner. The group average
power spectra of the end-tidal CO2 time series are presented in
Fig. 5. A repeated-measures ANOVA showed no effect of
resting state type (F1,2 ⫽ 0.155, P ⫽ 0.732) but a significant
interaction of resting state type by frequency (F41,82 ⫽ 3.024,
P ⫽ 1.021e⫺5). The average spectral density plot shows that, in
the eyes closed condition, the variations in end-tidal CO2 were
larger for frequencies ⬍0.02 Hz, but for frequencies ⬎0.02 Hz,
the variations were larger in the eyes open condition. Overall,
the effects of resting state on BOLD and end-tidal CO2, shown,
respectively, in Figs. 2 and 5, match poorly across the frequency range considered. We conclude that the effects of
resting state on end-tidal CO2 variability cannot adequately
explain the effects of resting state on spontaneous oscillations
of the BOLD signal.
RESTING STATES AFFECT SPONTANEOUS BOLD OSCILLATIONS
Relation between ongoing BOLD and EEG alpha power
We cannot conclusively attribute the BOLD effects described here to a specific electrophysiological phenomenon.
However, we were struck by several observations indicating
that modulations of spontaneous BOLD oscillations and alpha
rhythms by resting state type share a number of commonalities.
First, the effects were found in sensory but not regions of the
default system (Gusnard and Raichle 2001), with the possible
exception of retrosplenial cortex. This suggests that they may
not be mediated by top down mechanisms but rather by bottom up
mechanisms, possibly involving subcortical structures (Hughes
and Crunelli 2005). Second, the cortical distribution of the
BOLD effects of resting state type match quite closely the
distribution of alpha rhythms, which arise not only from visual
regions, but also from other sensory regions, including sensorymotor cortex (Pineda 2005) and auditory cortex (Niedermeyer
1990). The modulation of the spectral density of the ongoing
BOLD signal resembles those of the alpha power frequency
spectra obtained under the same behavioral conditions (see Fig.
4, A and B, in Linkenkaer-Hansen et al. 2001). Recent data
have confirmed this, having shown spontaneous BOLD oscillations in primary visual cortex were most coherent with power
modulations in the alpha band (Mantini et al. 2007).
Two partially independent cortical networks
support vigilance
resting state type on end-tidal CO2 variability, nevertheless,
these effects did not match the effects of resting state type on
spontaneous BOLD oscillations. The estimated correlation coefficients varied between 0.29 and 1.0, which is in the same
range or higher than those reported for correlations between
local field potentials recorded from adjacent regions of visual
cortex in nonhuman primates (Leopold et al. 2003). Interestingly, we found that BOLD fluctuations were mostly synchronous within the set of regions highlighted in Fig. 1, suggesting
a common, shared origin for these signals. There were nevertheless departures from complete synchronization. BOLD oscillations in left and right primary visual cortex were partly
independent from those in other regions, including regions in
extrastriate visual cortex. While surprising, the finding that
resting state signals may differentiate between visual striate
and extrastriate regions is not entirely novel. For example,
Damoiseaux et al. (2006) reported a decomposition of resting
state data into independent networks. These included two
visual networks, one centered on primary visual cortex and the
medial occipital-parietal lobe and the second on extrastriate
visual cortex along the lateral aspect of the posterior occipitaltemporal and parietal lobe. Similarly, resting state networks,
obtained during light sleep in infants, have highlighted a
network centered on the medial aspect of the occipital-parietal
lobe but failed to show an equivalent network in lateral regions
(Fransson et al. 2007). Therefore medial occipital structures
may hold a specific status developmentally, which is maintained in the adult. More generally, our correlation analysis
indicates that the effects of resting state type, rather than
representing amplitude modulations of local, independent neuronal activities, instead represent the activity of two oscillatory
networks that are temporally asynchronous but partly overlapping.
Functional significance of modulation of spontaneous
BOLD activity
The focality of the modulations of the ongoing BOLD signal
also raises the question of their behavioral significance. Patients with strokes involving the medial portion of the occipital
and temporal lobe, where the strongest effects of resting state
type on the BOLD signal were found, show remarkable disturbances of their vigilance (Caplan 1980; Medina et al. 1974,
1977). This clinical picture is specific for strokes affecting this
particular location or the right temporal parietal junction (Mesulam and Weintraub 1992), the latter being thought to mediate
reorienting to novel sensory stimuli (Corbetta and Shulman
2002). These clinical observations could be used to speculatively advance the hypothesis that the cerebral system, here
highlighted by the comparison of resting states, supports basic
aspects of vigilance through the gating of sensory processing.
APPENDIX
Regions affected by resting state type show prominent pairwise correlations. An unbiased estimate of the correlations
between regional neural activities was obtained from the differences between BOLD data in the eyes closed and fixation
conditions, based on the assumption that the covariance structure of non-neural signals did not change with resting state
type. Although this assumption is not completely supported by
our physiological recordings, which indicated some effects of
J Neurophysiol • VOL
Model of regional variance and pairwise
regional covariance
The model was formulated to obtain an estimate of pairwise
regional correlations that reduced the effects of non-neural sources of
variability on the BOLD signals. The observed BOLD signal in the
fixation condition F can be modeled as the sum of the neural activity
while fixating NF and all other non-neural sources NO
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
(Birn et al. 2006; Shmueli et al. 2007). To account for the
difference in BOLD variance between eyes closed and fixation
in right visual cortex and left retrosplenial cortex, for example,
which are in the order of 70% (see Table 2), the contribution of
physiological noise to BOLD variance should, accordingly,
vary by an order of magnitude, if one considers the voxel-wise
effects, or triple, if one considers the whole brain effects. This
seems quite implausible. Also, physiological measures obtained during the scans and outside the scanner do not support
the idea that resting state type caused major changes in physiological variables. Heart beat duration and beat to beat variations in heart beat duration estimated from electrical recordings obtained during the imaging procedure did not show a
significant effect of the resting state type on the cardiac cycle.
Moreover, end-tidal CO2 recorded in three subjects while they
lay in a mock scanner, with their eyes either open or closed, did
show resting state effects (see Fig. 5). However, these effects
matched poorly the effects of resting state type on spontaneous
BOLD oscillations (see Fig. 2). In conclusion, although we
cannot entirely rule out some contribution from physiological
noise, we think that a more plausible explanation of the effects
of resting state type on spontaneous BOLD oscillations is that
local spontaneous neural activity was affected by resting state
type.
929
930
MCAVOY ET AL.
F ⫽ N F ⫹ NO
(A1)
2
Since the increased power measured in the eyes closed condition must
correspond to an increase in variance, which must coincide with an
incremental increase in neural activity, the observed BOLD signal in
the eyes closed condition C is equal to the increase in neural activity
caused by maintaining the state of eyes closed NC added to that
already present during fixation NF and all other non-neural sources NO
(A2)
Simulation methods
Our goal is to compare correlations in neural activity between pairs
of regions. We will have six observed variances/covariances computed from the BOLD data for each region pair:
Var共F 1兲 ⫽ ␴ N2 O ⫹ ␴N2 F ⫹ 2␳NO NF ␴NO ␴NF
Var共F 2兲 ⫽ ␴
Var共C 1兲 ⫽ ␴
2
N O1
⫹␴
2
NF1
1
1
⫹␴
2
NF2
⫹␴
2
NC1
1
1
⫹ 2␳NO NF ␴NO ␴NF
2
2
2
⫹ 2␳NO NF ␴NO ␴NF
1
1
1
1
2
1
⫹ 2␳NO NC ␴NO ␴NC ⫹ 2␳NF NC ␴NF ␴NC
1
Var共C 2兲 ⫽ ␴
2
N O2
⫹␴
2
NF2
⫹␴
2
NC2
1
1
1
⫹ 2␳NO NF ␴NO ␴NF
2
2
2
1
1
1
2
2
Cov共F 1,F 2兲 ⫽ ␳ NO NO ␴NO ␴NO ⫹ ␳NO NF ␴NO ␴NF
1
2
1
2
1
2
1
1
2
⫹ 2␳NO NC ␴NO ␴NC ⫹ 2␳NF NC ␴NF ␴NC
2
2
2
2
2
2
2
⫹ ␳NO NF ␴NO ␴NF ⫹ ␳NF NF ␴NF ␴NF
2
1
2
1
1
2
Cov共C 1,C 2兲 ⫽ ␳ NO NO ␴NO ␴NO ⫹ ␳NO NF ␴NO ␴NF ⫹ ␳NO NC ␴NO ␴NC
1
2
1
2
1
2
1
2
1
2
1
1
1
2
1
1
2
1
2
1
2
1
2
2
1
2
1
⫹ ␳NF NC ␴NF ␴NC ⫹ ␳NC NC ␴NC ␴NC
2
1
2
1
1
2
2
2
⫹ ␳NO NF ␴NO ␴NF ⫹ ␳NF NF ␴NF ␴NF ⫹ ␳NF NC ␴NF ␴NC ⫹ ␳NO NC ␴NO ␴NC
2
1
2
Assuming the neural activity is independent and therefore uncorrelated with non-neural sources of variance, with a bit of algebra, the
correlation between the neural activity of regions 1 and 2 in the eyes
closed state ␳ NC NC can be shown to be
1
␳ NC NC ⫽
1
2
Cov共C1 ,C2 兲 ⫺ Cov共F1 ,F2 兲 ⫺ ␳NF NC ␴NF ␴NC ⫺ ␳NF NC ␴NF ␴NC
1
␴NC ␴NC
2
1
2
1
2
2
1
2
2
where
␴ N2 F ⫽Var共F 1 兲⫺ ␴ N2 O
1
␴
2
N F2
1
⫽ Var共F2 兲 ⫺ ␴
2
NO2
冑
␴ NC ⫽ ⫺␳NF NC ␴NF ⫾ ␳N2 F NC ␴N2 F ⫹ Var共C1 兲 ⫺ Var共F1 兲
1
1
1
1
1
1
2
2
2
2
2
The expected value of ␳ NC NC was estimated by Monte Carlo integration over
the remaining parameters 共i.e., ␴N2 O , ␴N2 O , ␳NF NC , ␳NF NC , ␳NF NC , and ␳NF NC 兲
which were assumed to have a uniform distribution over their domains.
1
2
1
2
1
1
1
2
2
1
2
2
1
J Neurophysiol • VOL
1
We thank T. Conturo, E. Akbudak, and A. Snyder for development of MRI
procedures. M. D. Fox, D. S. Marcus, T. R. Olsen, M. Ramaratnam, and K. A.
Archie are the developers of brainscape.org. L. E. Couture assisted with the
collection of the end-tidal CO2 data.
GRANTS
This work was supported by National Institute of Neurological Disorders
and Stroke Grant NS-006833.
REFERENCES
Adrian ED, Matthews BHC. Berger rhythm: potential changes from the
occipital lobes in man. Brain 57: 355–385, 1934.
Anrep GV, Pascual W, Rossler R. Respiratory variations of the heart rate. I.
The reflex mechanism of the respiratory arrhythmia. Proc R Soc Lond B Biol
Sci 119: 191–217, 1936.
Berger H. Über das elektrenkephalogramm des menschen. Arch Psychiatr
Nervenkr 87: 527–570, 1929.
Berger H. Über das elektrenkephalogramm des menschen II. J Psychol Neurol
40: 160 –179, 1930.
Birn RM, Diamond JB, Smith MA, Bandettini PA. Separating respiratoryvariation-related fluctuations from neuronal-activity-related fluctuations in
fMRI. Neuroimage 31: 1536 –1548, 2006.
Biswal BM, Yetkin FZ, Haughton VM, Hyde JS. Functional connectivity in
the motor cortex of resting human brain using echo-planar MRI. Magn
Reson Med 34: 537–541, 1995.
Caplan LR. “Top of the basilar” syndrome. Neurology 30: 72–97, 1980.
Corbetta M, Shulman GL. Control of goal-directed and stimulus-driven
attention in the brain. Nat Rev Neurosci 3: 201–215, 2002.
Cordes D, Haughton VM, Arfanakis K, Carew JD, Turski PA, Moritz CH,
Quigley MA, Meyerand ME. Frequencies contributing to functional connectivity in the cerebral cortex in “resting-state” data. AJNR 22: 1326 –1333,
2001.
Damoiseaux JS, Rombouts SA, Barkhof F, Scheltens P, Stam CJ, Smith
SM, Beckmann CF. Consistent resting-state networks across healthy subjects. Proc Natl Acad Sci USA 103: 13848 –13853, 2006.
de Munck JC, Goncalves SI, Huijboom L, Kuijer JPA, Pouwels PJW,
Heethaar RM, Lopes da Silva FH. The hemodynamic response of the
alpha rhythm: an EEG/fMRI study. Neuroimage 35: 1142–1151, 2007.
Feige B, Scheffler K, Esposito F, Di Salle F, Hennig J, Seifritz E. Cortical
and subcortical correlates of electroencephalographic alpha rhythm modulation. J Neurophysiol 93: 2864 –2872, 2005.
Fisher RA. On the ‘probable error’ of a coefficient of correlation deduced
from a small sample. Metron 1: 1–32, 1921.
Forman SD, Cohen JD, Fitzgerald M, Eddy WF, Mintun MA, Noll DC.
Improved assessment of significant activation in functional magnetic resonance imaging (fMRI): use of a cluster-size threshold. Magn Reson Med 33:
636 – 647, 1995.
Fox MD, Raichle ME. Spontaneous fluctuations in brain activity observed
with functional magnetic resonance imaging. Nat Rev Neurosci 8: 700 –711,
2007.
Fox MD, Snyder AZ, Vincent JL, Corbetta M, Van Essen DC, Raichle
ME. The human brain is intrinsically organized into dynamic, anticorrelated
functional networks. Proc Natl Acad Sci USA 102: 9673–9678, 2005.
Fransson P. Spontaneous low-frequency BOLD signal fluctuations: an fMRI
investigation of the resting-state default mode of brain function hypothesis.
Hum Brain Map 26: 15–29, 2005.
Fransson P, Skiold B, Horsch S, Nordell A, Blennow M, Lagercrantz H,
Aden U. Resting-state networks in the infant brain. Proc Natl Acad Sci USA
104: 15531–15536, 2007.
Friston KJ, Holmes AP, Poline JB, Grasby PJ, Williams SC, Frackowiak
RS, Turner R. Analysis of fMRI time-series revisited. Neuroimage 2:
45–53, 1995.
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
Var(F1) observed variance in the fixation state of region 1
Var(F2) observed variance in the fixation state of region 2
Var(C1) observed variance in the eyes closed state of region 1
Var(C2) observed variance in the eyes closed state of region 2
Cov(F1,F2) observed covariance between regions 1 and 2 in the
fixation state
Cov(C1,C2) observed covariance between regions 1 and 2 in the
eyes closed state
Note Var(x) ⫽ Cov(x,x). Substituting Eqs. A1 and A2 into the six
observed values and solving the algebra yields the following six
relations
1
2
ACKNOWLEDGMENTS
C ⫽ N C ⫹ NF ⫹ NO
2
N O2
冑
␴ NC ⫽ ⫺␳NF NC ␴NF ⫾ ␳N2 F NC ␴N2 F ⫹ Var共C2 兲 ⫺ Var共F2 兲
RESTING STATES AFFECT SPONTANEOUS BOLD OSCILLATIONS
J Neurophysiol • VOL
Medina JL, Rubino FA, Ross E. Agitated delirium caused by infarctions of
the hippocampal formation and fusiform and lingual gyri: a case report.
Neurology 24: 1181–1183, 1974.
Mesulam MM, Weintraub S. Spectrum of primary progressive aphasia.
Baillieres Clin Neurol 1: 583– 609, 1992.
Moosmann M, Ritter P, Krastal I, Brink A, Thees S, Blankenburg F,
Taskin B, Obrig H, Villringer A. Correlates of alpha rhythm in functional
magnetic resonance imaging and near infrared spectroscopy. Neuroimage
20: 145–158, 2003.
Niedermeyer E. Alpha-like rhythmical activity of the temporal lobe. Clin
Electroencephalogr 21: 210 –224, 1990.
Ojemann JG, Akbudak E, Snyder AZ, McKinstry RC, Raichle ME,
Conturo TE. Anatomic localization and quantitative analysis of gradient
refocused echo-planar fMRI susceptibility artifacts. Neuroimage 6: 156 –
167, 1997.
Pineda JA. The functional significance of mu rhythms: translating “seeing“
and “hearing” into “doing”. Br Res Rev 50: 57– 68, 2005.
Purdon PL, Weisskoff RM. Effect of temporal autocorrelation due to physiological noise and stimulus paradigm on voxel-level false-positive rates in
fMRI. Hum Brain Map 6: 239 –249, 1998.
Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. A default mode of brain function. Proc Natl Acad Sci USA 98:
676 – 682, 2001.
Rowland DJ, Garbow JR, Laforest R, Snyder AZ. Registration of
[18F]FDG microPET and small-animal MRI. Nucl Med Biol 6: 567–572,
2005.
Shmueli K, van Gelderen P, de Zwart JA, Horovitz SG, Fukunaga M,
Jansma JM, Duyn JH. Low-frequency fluctuations in the cardiac rate as a
source of variance in the resting-state fMRI BOLD signal. Neuroimage 38:
306 –320, 2007.
Shulman GL, Fiez JA, Corbetta M, Buckner RL, Miezin FM, Raichle ME,
Petersen SE. Common blood flow changes across visual tasks. II. Decreases
in cerebral cortex. J Cogn Neurosci 9: 648 – 663, 1997.
Talairach J, Tournoux P. Co-planar Stereotaxic Atlas of the Human Brain.
New York: Thieme, 1988.
Van Essen DC. A population-average, landmark- and surface-based (PALS)
atlas of human cerebral cortex. Neuroimage 15: 635– 662, 2005.
Vincent JL, Larson-Prior LJ, Zempel JM, Snyder AZ. Moving GLM
ballistocardiogram artifact reduction for EEG acquired simultaneously with
fMRI. Clin Neurophsyiol 118: 981–998, 2007.
Weisskoff RM. Simple measurement of scanner stability for functional
NMR imaging of activation in the brain. Magn Reson Med 36: 643– 645,
1996.
Wise RG, Koijiro I, Poulin MJ, Tracey I. Resting fluctuations in arterial
carbon dioxide induce significant low frequency variations in the BOLD
signal. Neuroimage 21: 1652–1664, 2004.
Zarahn E, Aguirre GK, D’Esposito M. Empirical analyses of BOLD fMRI
statistics. I. Spatially unsmoothed data collected under null-hypothesis
conditions. Neuroimage 5: 179 –197, 1997.
100 • AUGUST 2008 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 14, 2017
Fukunaga M, Horovitz SG, van Gelderen P, de Zwart JA, Jansma JM,
Ikonomidou VN, Chu R, Deckers RH, Leopold DA, Duyn JH. Largeamplitude, spatially correlated fluctuations in BOLD fMRI signals during
extended rest and early sleep stages. Magn Reson Imaging 24: 979 –992,
2006.
Goldman RI, Stern JM, Engel J Jr, Cohen MS. Simultaneous EEG and
fMRI of the alpha rhythm. Neuroreport 18: 2487–2492, 2002.
Goncalves SI, de Munck JC, Pouwels PJW, Schoonhoven R, Kuijer JPA,
Maurits NM, Hoogduin JM, Van Someren EJW, Heethaar RM, Lopes
da Silva FH. Correlating the alpha rhythm to BOLD using simultaneous
EEG/fMRI: intersubject variability. Neuroimage 30: 203–213, 2006.
Greicius MD, Krasnow B, Reiss AL, Menon V. Functional connectivity in
the resting brain: a network analysis of the default mode hypothesis. Proc
Natl Acad Sci USA 100: 253–258, 2003.
Gusnard DA, Raichle ME. Searching for a baseline: functional imaging and
the resting human brain. Nat Rev Neurosci 2: 685– 694, 2001.
Hampson M, Peterson BS, Skudlarski P, Gatenby JC, Gore JC. Detection
of functional connectivity using temporal correlations in MR images. Hum
Brain Map 15: 247–262, 2002.
Horovitz SG, Fukunaga M, de Zwart JA, van Gelderen P, Fulton SC,
Balkin TJ, Duyn JH. Low frequency BOLD fluctuations during resting
wakefulness and light sleep: a simultaneous EEG-fMRI study. Hum Brain
Map 29: 671– 682, 2008.
Hughes SW, Crunelli V. Thalamic mechanisms of EEG alpha rhythms and
their pathological implications. Neuroscientist 11: 357–372, 2005.
Koch C. Biophysics of Computation: Information Processing in Single Neurons. New York: Oxford UP, 1999, p. 350 –373.
Laufs H, Holt JL, Elfont R, Krams M, Paul JS, Krakow K, Kleinschmidt
A. Where the BOLD signal goes when alpha EEG leaves. Neuroimage 31:
1408 –1418, 2006.
Laufs H, Krakow K, Sterzer P, Eger E, Beyerle A, Salek-Haddadi A,
Kleinschmidt A. Electroencephalographic signatures of attentional and
cognitive default modes in spontaneous brain activity fluctuations at rest.
Proc Natl Acad Sci USA 100: 11053–11058, 2003.
Leopold DA, Murayama Y, Logothetis NK. Very slow activity fluctuations
in monkey visual cortex: implications for fuctional neuroimaging. Cereb
Cortex 13: 422– 433, 2003.
Linkenkaer-Hansen K, Nikouline VV, Palva JM, Ilmoniemi RJ. Longrange temporal correlations and scaling behavior in human brain oscillations. J Neurosci 21: 1370 –1377, 2001.
Lowe MJ, Mock BJ, Sorenson JA. Functional connectivity in single and
multislice echoplanar imaging using resting-state fluctuations. Neuroimage
7: 119 –132, 1998.
Mantini D, Perrucci MG, Del Gratta C, Romani GL, Corbetta M. Electrophysiological signatures of resting state networks in the human brain.
Proc Natl Acad Sci USA 104: 13170 –13175, 2007.
McAvoy MP, Ollinger JM, Buckner RL. Cluster size thresholds for assessment of significant activation in fMRI. Neuroimage 13: S198, 2001.
Medina JL, Chokroverty S, Rubino FA. Syndrome of agitated delirium and
visual impairment: a manifestation of medial temporo-occipital infarction.
J Neurol Neurosurg Psychiatry 40: 861– 864, 1977.
931