Article

1. NEUROSYSTEMS
Fatigue-induced increase in intracortical communication
between mid ⁄anterior insular and motor cortex during
cycling exercise
Lea Hilty,1,2,3
Nicolas Langer,3
Roberto Pascual-Marqui,4,5
Urs Boutellier1,2
and Kai Lutz3
1
Exercise Physiology, Institute of Human Movement Sciences, ETH Zurich, Zurich, Switzerland
2
Institute of Physiology, University of Zurich, Zurich, Switzerland
3
Institute of Psychology, Department of Neuropsychology, University of Zurich, Switzerland
4
The KEY Institute for Brain-Mind Research, University Hospital of Psychiatry, Zurich, Switzerland
5
Department of Neuropsychiatry, Kansai Medical University Hospital, Osaka, Japan
Keywords: EEG, exercise, homeostatic functions, lagged phase synchronization, supraspinal fatigue
Abstract
In the present study, intracortical communication between mid ⁄ anterior insular and motor cortex was investigated during a fatiguing
cycling exercise. From 16 healthy male subjects performing a constant-load test at 60% peak oxygen consumption (VO2peak) until
volitional exhaustion, electroencephalography data were analysed during repetitive, artefact-free periods of 1-min duration. To
quantify fatigue-induced intracortical communication, mean intra-hemispheric lagged phase synchronization between mid ⁄ anterior
insular and motor cortex was calculated: (i) at the beginning of cycling; (ii) at the end of cycling; and (iii) during recovery cycling.
Results revealed significantly increased lagged phase synchronization at the end of cycling, which returned to baseline during
recovery cycling after subjects’ cessation of exercise. Following previous imaging studies reporting the mid ⁄ anterior insular cortex as
an essential instance processing a variety of sensory stimuli and signalling forthcoming physiological threat, our results provide
further evidence that during a fatiguing exercise this structure might not only integrate and evaluate sensory information from the
periphery, but also act in communication with the motor cortex. To the best of our knowledge, this is the first study to empirically
demonstrate that muscle fatigue leads to changes in interaction between structures of a brain’s neural network.
Introduction
Exercise-induced muscle fatigue is defined as a reversible reduction in
the neuromuscular system’s capacity to generate force or power (Fitts
& Holloszy, 1976; Bigland-Ritchie et al., 1983). Underlying causes of
muscle fatigue are manifold, and have been addressed not only to
peripheral changes within the working muscle but also to central
processes residing in the sensorimotor pathway of the CNS. As a
component of central fatigue, supraspinal fatigue was identified and
defined as a loss of force due to an insufficient output from the motor
cortex (Gandevia et al., 1996; Taylor et al., 2000).
Muscle afferents type III ⁄ IV sensitive to mechanical, chemical or
noxious stimuli of other origin have been assumed to inhibit motor
cortical output cells (Martin et al., 2008) and ⁄ or sites driving the
motor cortex (Gandevia et al., 1996; Taylor et al., 1996, 2000), thus
exerting potential influence on supraspinal fatigue. Indeed, a causal
relationship between spinal l-opioid receptor-sensitive muscle affer-
ents and motor cortical inhibition was demonstrated by a temporary
afferent blockade attenuating a fatigue-related increase in intracortical
inhibition (Hilty et al., 2011). However, a comprehensive under-
standing of a brain’s fatigue-related neural network comprising
assumed relay stations communicating with motor cortex cells
remains elusive.
In a functional magnetic resonance imaging (fMRI) study investi-
gating neural activity during a maximal 2-min handgrip contraction,
brain structures such as the insular and cingulate cortex have been
reported to play an important role in integrating inhibitory influence
arising from group III and IV muscle afferents (Liu et al., 2002,
2003). In studies of experimentally induced muscle pain, an increase
in neural activity within widespread regions of both the insular and
cingulate cortex was proven, besides other regions found to be
activated also by other forms of pain (Kupers et al., 2004; Henderson
et al., 2006). Furthermore, the anterior insular cortex has consistently
been shown to be involved not only in pain processing but also in the
evaluation of other homeostatic processes, such as air hunger and
hunger for food (Tataranni et al., 1999; Brannan et al., 2001; Liotti
et al., 2001; Evans et al., 2002; Craig, 2003b). In accordance,
prominent mid ⁄ anterior insular activation was revealed directly before
task failure in an isometric muscle fatiguing handgrip contraction
potentially alerting the organism to urgent homeostatic imbalance
(Hilty et al., 2010).
Correspondence: Dr K. Lutz, as above.
E-mail: Kai.Lutz@uzh.ch
Received 13 May 2011, accepted 16 September 2011
European Journal of Neuroscience, Vol. 34, pp. 2035–2042, 2011 doi:10.1111/j.1460-9568.2011.07909.x
ª 2011 The Authors. European Journal of Neuroscience ª 2011 Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience

2. In consideration of these findings, we assume that during fatiguing
exercise the mid ⁄ anterior insular cortex is of essential relevance not
only by signalling forthcoming physiological threat (Reiman, 1997),
but also by acting in communication with the motor cortex. On the
basis of surface electroencephalography (EEG) signals transformed
into 3D spatial current density distributions, neurophysiological inter-
and intrahemispheric communication between time series correspond-
ing to different spatial locations can be assessed by computing lagged
phase synchronization of oscillatory neural activity (Pascual-Marqui,
2007a). We hypothesized that at the end of an exhaustive cycling
exercise signalling between mid ⁄ anterior insular and motor cortex is
increased, and thus lagged phase synchronization is higher than at the
beginning.
Materials and methods
Subjects
Seventeen male subjects [age: 25.9 ± 3.5 years (mean ± SD); height:
184.4 ± 5.8 cm; mass: 80.7 ± 8.0 kg; peak oxygen consumption,
VO2peak: 4.7 ± 0.5 L ⁄ min] participated in this study. Due to technical
difficulties, data from one subject had to be excluded, resulting in a
final sample of 16 subjects considered for data analyses. All subjects
were regularly performing a minimum of 3 h weekly endurance
training. They did not suffer from any neurological ⁄ respiratory
disorders, and did not take any medication ⁄ nutritional supplements.
Subjects were instructed to refrain from stressful activities and to
adhere to usual caffeine consumption, 24 and 12 h, respectively, prior
to exercise participation. Each subject was precisely informed about
any risks and inconveniences associated with the experiment before
written informed consent was obtained. All procedures were con-
ducted in accordance to the standards set by the Declaration of
Helsinki.
Procedure
In a pre-test, participants had to perform an all-out incremental
exercise test on a cycling ergometer (Ergoline 800; Ergonometriesys-
teme, Bitz, Germany) in order to assess VO2peak using a calibrated
metabolic cart (OxyconPro, Jaeger, Hoechberg, Germany). This
exercise test, throughout which a freely chosen pedalling frequency
of > 70 revolutions ⁄ min had to be kept constant (Kohler & Boutellier,
2005), consisted of a 4-min warm-up cycling at 100 W followed by
subsequent, iterative workload increases of 50 W every 2 min.
Separated by at least 4 days from this pre-test, an experimental
exercise test had to be performed. Subjects exercised on a racing
bicycle attached to an electromagnetic cycle trainer (Tacx cycleforce
basic T1601, Wassenaar, the Netherlands) by providing the same pre-
assessed pedalling frequency at a workload of 60% VO2peak
(245.6 ± 28.6 W) until volitional exhaustion. Instead of using the
electromagnetic power control of the cycle trainer, its resistance was
mechanically adjusted until appropriate constant workload was
generated at the individual’s chosen pedalling frequency. Both power
output and pedalling frequency were continuously assessed by means
of an SRM Training System (PowerMeter IV; SRM, Jülich,
Germany). However, only pedalling frequency was visualized to the
subjects throughout exercise. An EEG cap was mounted on the
subjects’ scalp and two electrodes were placed on the subjects’ neck
muscles bilaterally (M. trapezius, pars descendens) in order to check
for muscle artefacts. Cables of the cap were stuck on the subjects’
back in order to minimize shifting during exercise. After a 4-min
warm-up cycling inside an air-conditioned Faraday cage (temperature:
19.6 ± 0.3 C, humidity: 31.2 ± 5.6%), subjects rested on the bicycle
and EEG electrodes were checked for required impedances ( 30 kX;
see EEG recordings). In addition, three 30-s sessions of eyes closed
(EC) alternating with three 30-s sessions of eyes open (EO) were
recorded. During EO, subjects had to fixate a cross marked on a white
sheet, which was stuck on a wall in front of them. Throughout the
following constant-load exercise test, EEG analysis periods of 1 min
duration were given, during which the subjects were instructed to keep
their upper body as still as possible while keeping up cycling. These
periods were repeated every 5 min, with the first period assessed
1 min after exercise beginning. Onsets and time progression of each 1-
min period were verbally announced in order to sensitize and remind
the subjects to fixate the cross, relax their jaw muscles and move their
upper body as little as possible. Directly after each period, the subjects
had to verbally rate their perceived exertion (RPE) on the commonly
used Borg scale ranging from 6 (‘no exertion at all’) to 20 (‘maximal
exertion’; Borg, 1975). Heart rate was continuously recorded with a
Polar Sports Tester (Polar Electro, Kempele, Finland). When subjects
were exhausted and gave up cycling, they were instructed to rest
calmly on the bicycle. During this resting period, the same alternating
sessions of EO and EC were carried out again. Thereafter, the
resistance of the cycle trainer was released and subjects cycled at the
same cadence as before providing 50–60 W for another 15 min.
During the last minute of recovery cycling, a final EEG analysis period
was assessed with the same instructions mentioned above.
EEG recordings
Dense array EEG System from Geodesics (128-channel HydroCel
Geodesics Sensor Net, Net Station software 4.3.1; Electrical Geode-
sics, Eugene, OR, USA) was used for EEG data acquisition. An
appropriate elastic tension 128-Ag ⁄ AgCl-Electrode Sensor Net cap
was individually selected from three different sizes, and impedances
were improved by soaking synthetic sponges with KCl-solution for
10 min prior to the experiment. Data were sampled at 500 Hz, filtered
using an infinite impulse filter (0.3–100 Hz), digitized with a 16-bit
A ⁄ D converter and re-referenced to the vertex (Cz).
Data analysis
Brain Vision Analyzer 2.0 software (Brain Products, München,
Germany) was used for off-line processing of the EEG raw data. A
rather narrow band-pass filter from 3.75 to 30 Hz (time constant:
0.0424, 24 db ⁄ octave) was used in order to remove movement and
muscle artefacts, which were to be expected at about 1–1.5 Hz
[pedalling frequency (movement artefacts)] and 30 Hz (muscle
artefacts). Independent Component Analysis was subsequently calcu-
lated in order to effectively detect, separate and remove activity in
EEG records arising from various artefactual sources such as eye
blinks (Jung et al., 1998). In each participant, 20 channels located
marginally on the EEG cap were excluded from further analysis as
they frequently suffered from bad skin contact. From the remaining
108 electrodes intervals of EEG signals were skipped if any of the
following criteria defined in a semi-automatic raw data inspection
were not met: low activity, 0.3 lV; amplitude, )200 lV to 200 lV;
difference (Max)Min), 200 lV; gradient, 50 lV ⁄ ms. Moreover,
if a channel’s signal comprised 10% artefact contaminated intervals
(occurring on average in 2.9 ± 1.7 electrodes per subject), its signal
was interpolated with data of its four nearest neighbours. Then, EEG
artefact-free data were fragmented into data sets of 1-s duration
providing on average 55 ± 6 and 54 ± 5 segments per subject from
2036 L. Hilty et al.
ª 2011 The Authors. European Journal of Neuroscience ª 2011 Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 34, 2035–2042

3. the first and last period of the fatiguing cycling exercise, as well as
56 ± 2 from the last period of recovering cycling. Finally, 1-s
segments of the four conditions were exported into standardized low-
resolution brain electromagnetic tomography (sLORETA) software
(http://www.unizh.ch/keyinst/loreta.htm; Fuchs et al., 2002; Pascual-
Marqui, 2002; Jurcak et al., 2007).
In order to eliminate potential remaining risks of electromyographic
activity contaminating EEG data in the b (12.5–35 Hz) band
(O’Donnell et al., 1974; Davidson, 1988), sLORETA analysis of
EEG data during exercise cycling was restricted to investigations of
EEG signals in the a ⁄ l (7.5–12.5 Hz) band. This frequency band has
been shown to be particularly sensitive to various aspects of motor
activity (Pfurtscheller et al., 2000). In order to judge possible
remaining artefacts, we inspected spectral plots at the scalp locations
C3, C4, T9, T10, O1, O2, AF3, AF4. These plots are provided in
Supporting Information Fig. S1.
Two anatomical regions of interest (ROI) were constructed on each
hemisphere by defining all voxels located within a radius of 5 mm
around designated seed points. For the first ROI on each hemisphere,
seed points were located at Montreal Neurological Institute coordi-
nates (Evans et al., 1993) [)10, )40, 65 (x y z) and 10, )40, 65 (x, y,
z), respectively], as they cover left and right representations of the legs
within the motor cortex. In terms of the second ROI covering left and
right mid ⁄ anterior insular region, Montreal Neurological Institute
coordinates [)40, 5, 10 (x, y, z) and 40, 5, 10 (x, y, z), respectively]
were determined with respect to a previously conducted study,
showing that this part of the insular cortex is involved in limiting
physical performance in a muscle-fatiguing exercise (Hilty et al.,
2010).
Dynamic changes of fatigue-related functional coupling in the a ⁄ l
band between the defined ROI have been assessed by calculating mean
lagged phase synchronization (Pascual-Marqui, 2007a) of intra-
hemispheric connections (left insular cortex with left motor cortex
and right insular cortex with right motor cortex) from both the first and
the last period of fatiguing cycling as well as during the last minute of
recovery cycling. Because lateralization was not the focus of this
study, lagged phase synchronization values within each hemisphere
were averaged over hemispheres. Lagged phase synchronization
computation is based on a collection of single epoch data (cross-
spectra calculated for each condition and subject) and a corresponding
discrete Fourier transform for signal within the ROI. Unlike other
coherence measures, phase synchronization is insensitive to amplitude
information. Because we merely test for increasing communication but
would like to include activating as well as inhibiting signalling, we do
not necessarily assume a positive correlation of amplitudes and thus
expect phase synchronization to be the most suitable measure. In fact,
inhibitory influences from mid ⁄ anterior insular to motor cortex would
be most likely, for which phase synchronization is likely to be most
sensitive. We computed lagged measures rather than regarding
instantaneous synchronization for several reasons. First, this approach
has the advantage over non-lagged calculations that it is insensitive
towards volume conduction (Nolte et al., 2004; Stam et al., 2007).
Second, sLORETA searches for the smoothest possible current density
distribution explaining the measured EEG topography and thereby
maximizes similarity of the assumed signal over a spatially distributed
number of voxels. This artificially inflates instantaneous coherence
measures, which therefore should be avoided. And third, neuronal
signal conduction requires time, which means that only delayed
(lagged) coherence between brain regions can result from neuronal
communication.
Calculations of lagged phase synchronization used in this study are
performed using sLORETA, and are more explicitly described by
Pascual-Marqui (2007b). Briefly, phase synchronization PS is based
on the absolute value of the covariance s between the normalized
Fourier transforms of two signal time series X and Y at frequency x.
The classical phase synchronization, consisting of instantaneous
plus lagged components, can be expressed by the following equation:
PS ¼ jS
x
yxj ¼
1
NR
X
NR
j¼1
xjx
y
jx
with
xjx;
yjx denoting the normalized discrete Fourier transform at
frequency x of the time series X and Y segmented into NR segments of
equal duration. The value of this measure is determined to a large
degree by its instantaneous component, which receives its largest
contribution from non-physiological artefacts such as volume con-
duction (Nolte et al., 2004) and by spatial smoothing. This problem
was addressed early on by Nolte et al. (2004) for the classical
coherence measure, where they proposed the imaginary part as a
measure of physiological connectivity that is not influenced by volume
conduction. Related methods were proposed by Stam et al. (2007) for
phase synchronization. Partly motivated by these previous results, an
improved measure of physiological connectivity was developed
(Pascual-Marqui, 2007b; Pascual-Marqui et al., 2011) that explicitly
models and appropriately separates the instantaneous (non-lagged)
connectivity. Instead of the imaginary part of the phase synchroniza-
tion:
Im sx
^
y
^
x
that would follow from the Nolte et al. (2004) approach, the corrected
measure of lagged phase synchronization (Pascual-Marqui, 2007b) is:
PSlagged ¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
Im sx
^
y
^
x
h i2
1 Re sx
^
y
^
x
h i2
v
u
u
u
u
t
It is shown in Pascual-Marqui et al. (2011) that when the
instantaneous connectivity component is large, the Nolte et al.
(2004) imaginary part tends to zero, whereas the lagged measure
retains non-zero physiologically relevant information on connectivity.
In order to ensure that modulations in lagged phase synchronization
did not arise due to changes in EEG power (Florian et al., 1998), mean
a ⁄ l power was assessed at the same pre-defined ROI (left and right
insular cortex, left and right motor cortex) from the first and the last
period of fatiguing cycling as well as during the last minute of
recovery cycling. Additionally, l power serves as an indicator of
overall neuronal activity in the motor cortex (Laufs et al., 2003;
Pfurtscheller et al., 2006).
Another parameter having been presumed to confound EEG
activity is hyperthermia (Nielsen et al., 2001; Rasmussen et al.,
2004; Ftaiti et al., 2010). Therefore, body temperature was assessed
by using an electronic thermometer (Welch Allyn 690 SureTemp
Plus, Welch Allyn, NSW, Australia; ± 0.1 C accuracy) and by
averaging values from three repetitive oral measurements just prior
to exercise, and from eight measurements performed within 4 min
after exhaustion, respectively. During temperature measurements,
subjects were instructed to keep their mouth permanently closed with
the probe maintained in the right posterior sublingual pocket at the
base of their tongue. Subjects were allowed to drink from a water-
filled CamelBak
as much as preferred, though, only between
Muscle fatigue-induced intracortical communication 2037
ª 2011 The Authors. European Journal of Neuroscience ª 2011 Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 34, 2035–2042

4. periods of EEG measurements and until a RPE of 18 was reached.
Furthermore, after subjects’ cessation of exercise they were
instructed to close their mouth and breathe through their nose.
These restrictions were met, firstly, in order to avoid movement
artefacts due to drinking and, secondly, to ensure proper post-
exercise measurements of oral temperature unaffected by water
intake or increased respiration.
With particular regard to a pioneering study of Schneider et al.
(2009), who recorded brain cortical activity in the a and b band with
EC before and after different kinds of fatiguing exercises, in the
present study both the a and b band were analysed during resting
conditions with EC before and after fatiguing cycling.
Statistics
Electrophysiological data were subjected to repeated-measures ano-
vas with the factor Time with three levels: beginning of cycling; end
of cycling; and recovery cycling. To guard against effects of
heteroscedasticity, Greenhouse–Geisser correction was used. Where
anova revealed significant effects, post hoc t-tests were applied.
Results are expressed as mean ± SD.
To assess differences in a and b band power at resting conditions
during EC between measurements before and after exercise, non-
parametric randomization tests were used. Reported are voxels that
survived significance thresholding at P = 0.05, defined by 5000
randomizations, corrected for multiple comparisons.
Results
Behavioural data, heart rate and body temperature
The mean exercise time of the experimental cycling exercise was
35.5 ± 9.9 min (mean ± SD), with 6.4 ± 1.7 periods of EEG analysis
included. After subjects’ access to water was quit (RPE 18), mean
exercise time until exhaustion (RPE 20) was 10.3 ± 4.7 min. At the
end of the fatiguing protocol, mean heat rate was significantly higher
than at the beginning of cycling (150.5 ± 10.1 vs. 181.7 ± 8.6 beats ⁄
min, respectively, t15 = )18.5, P 0.001). A significant increase was
also found in body temperature after fatiguing cycling compared with
pre-exercise values (36.9 ± 0.2 C vs. 37.0 ± 0.2 C, respectively,
t15 = )2.375, P = 0.031).
Electrophysiological data
anova of lagged phase synchronization between intra-hemispheric
mid ⁄ anterior insular and motor cortex connection revealed a signif-
icant effect of Time within the a ⁄ l band (F1.965,15 = 10.049,
P = 0.001). Subsequent post hoc t-tests showed a significantly higher
lagged phase synchronization at the end of cycling than at the
beginning of cycling (t15 = )3.383, P = 0.004; Fig. 1). A significantly
lower lagged phase synchronization was found during recovery
cycling than at the end of cycling (t15 = 4.158, P = 0.001). No
significant change in lagged phase synchronization was observed from
the beginning of cycling to recovery cycling (t15 = 0.571, P = 0.576).
Analysis of mean a ⁄ l power within the mid ⁄ anterior insular cortex
revealed a significant effect of Time (F1.101,15 = 5.470, P = 0.03;
Fig. 2A). During recovery cycling, a ⁄ l power was significantly lower
than at both the beginning (t15 = 2.787, P = 0.014) and the end of
cycling (t15 = 6.480, P 0.001). No significant change in a ⁄ l power
could be detected from the beginning to the end of cycling
(t15 = 0.995, P = 0.336).
Only a trend was found for Time to influence a ⁄ l power within the
motor cortex (F1.039,15 = 3.838, P = 0.067; Fig. 2B).
Whole-brain analysis during resting conditions of EC revealed a
widespread increase within both the a and b power, which reached
significance (t15 = 4.229, Max t = 9.04, Extreme P 0.001 for a, and
t15 = 4.08, Max t = 4.95, Extreme P = 0.005 for b) only in Brodmann
area 11 after exhaustive cycling exercise (Fig. 3).
Discussion
In the present EEG study, fatigue-related neural communication
between mid ⁄ anterior insular and motor cortex was investigated
during a constant-load cycling exercise by calculating lagged phase
synchronization within a ⁄ l rhythm at the beginning vs. the end of
exercise as well as vs. recovery cycling. As hypothesized, data
revealed a significant increase in lagged phase synchronization at the
end of cycling, which levelled off and returned to baseline during
recovery cycling after subjects’ cessation of exercise.
Lagged phase synchronization unaffected by power
and temperature
This increase in lagged phase synchronization cannot be explained by
changes in power of the respective frequency band (Jung et al., 1998),
because regarding mean a power in the mid ⁄ anterior insular and the
motor cortex at the end of cycling, no significant increase was found.
Rather, a power tended to decrease over time, which – considering
negative correlation of a power with cortical activity (Laufs et al.,
2003) – confirms previous findings from fMRI studies showing an
increased cortical activity within the mid ⁄ anterior insular (Hilty et al.,
2010) and the sensorimotor cortex (Liu et al., 2003) towards the end
of a fatiguing exercise.
As a consequence of prolonged exercise in the heat, high body
temperature has been shown to increase a ⁄ b power index, which is
defined as the area of a power spectrum divided by the area of b power
spectrum (Nielsen et al., 2001; Nybo Nielsen, 2001; Rasmussen
et al., 2004; Ftaiti et al., 2010). In order not to misinterpret EEG data
potentially contaminated by movement and muscle artefacts, b power
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
Lagged
phase
synchronization
** **
Beginning
of cycling
Recovery
cycling
End
of cycling
Pre-exercise
resting
Post-exercise
resting
n.s.
n.s.
Fig. 1. Mean lagged phase synchronization between the mid ⁄ anterior insular
and the motor cortex within a ⁄ l frequency band during resting (pre- ⁄ post-
exercise) and cycling conditions (beginning ⁄ end ⁄ recovery) of EO. Error bars
represent inter-individual SD. Within the cycling conditions a main effect of
Time was revealed. Significant changes from post hoc t-tests are depicted by
**P 0.01; n.s., not significant.
2038 L. Hilty et al.
ª 2011 The Authors. European Journal of Neuroscience ª 2011 Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 34, 2035–2042

5. was not assessed during cycling exercise and, thus, a ⁄ b power index
could not be calculated. However, controlled ambient temperature of
19.6 ± 0.3 C and an observed increase in body temperature of 1 C
at the end of exercise do not allow the assumption of an influence on
EEG, as similar conditions in control trials of hyperthermia-related
studies did not lead to changes in a ⁄ b power index (Nielsen et al.,
2001; Nybo Nielsen, 2001). Moreover, by positioning an electrode
1 cm in front of Cz chosen to represent motor areas of the legs,
hyperthermic-induced increase in a ⁄ b power index was demonstrated
to arise from decreased b power (Nybo Nielsen, 2001) rather than
an increase in a power. Thus, we conclude that our result of decreased
a ⁄ l power at the end of fatiguing exercise is not influenced by thermic
effects.
Role of mid ⁄ anterior insular cortex in homeostasis considering
anatomic connections
Over the past few years, a concept of a central governor has been
elaborated, which posits a brain’s regulative network both integrating
afferent information from the periphery and determining work rate as
well as exercise termination, all to ensure that homeostasis is
maintained (St Clair Gibson Noakes, 2004; Noakes et al., 2005).
Using neuroimaging methods, the anterior insula has been elucidated
as a structure being involved in the evaluation of various distressing
stimuli threatening homeostasis (Cannon, 1935), such as visceral and
cutaneous pain (Binkofski et al., 1998; Baciu et al., 1999; Casey,
1999), air hunger (Brannan et al., 2001; Liotti et al., 2001; Evans
et al., 2002) and hunger for food (Tataranni et al., 1999). Further-
more, prominent mid ⁄ anterior insular activity was also demonstrated
just before cessation of a fatiguing handgrip exercise, suggesting this
structure may be of essential relevance for mediating the termination
of a muscle-fatiguing exercise in order to protect the integrity of the
organism (Hilty et al., 2010). With the current data, previous findings
of a mid ⁄ anterior insular cortex signalling potential physiological
threat (Reiman, 1997; Craig, 2003a) are extended by demonstration of
a fatigue-related increase in communication between the insular and
the motor cortex in the context of an exhaustive bodily exercise.
However, it remains speculative whether this increased intracortical
communication arises from processes driven bottom up, for example
by muscle afferents type III ⁄ IV and ⁄ or from influences interacting
L A R L S R A S P L A R L S R A S P
R A L R S L P S A R A L R S L P S A
top back left top back left
bottom front right bottom front right
t-values
–9.038 –6.779 –4.519 –2.260 0.000 2.260 4.519 6.779 9.038
t-values
–4.955 –3.716 –2.477 –1.239 0.000 1.239 2.477 3.716 4.955
α β
Fig. 3. Statistical parametric maps of sLORETA showing regional changes within the a and b frequency bands after fatiguing cycling exercise compared with
prefatigue baseline during EC. Areas coloured in yellow ⁄ red and blue represent power increases and decreases, respectively, within the specific frequency band.
Orthogonal views from different perspectives are displayed. A, anterior; L, left hemisphere; P, posterior; R, right hemisphere. For interpretation of color references in
figure legend, please refer to the Web version of this article.
0
50
100
150
200
250
300
350
400
450
A
B
*
0
2000
4000
6000
8000
10000
12000
14000
16000
*
n.s. **
Power
[(µA/mm
2
)
2
/Hz]
Power
[(µA/mm
2
)
2
/Hz]
Beginning
of cycling
Recovery
cycling
End
of cycling
Pre-exercise
resting
Post-exercise
resting
Beginning
of cycling
Recovery
cycling
End
of cycling
Pre-exercise
resting
Post-exercise
resting
Fig. 2. Mean a ⁄ l power in the mid ⁄ anterior insular (A) and the motor cortex (B) during resting (pre- ⁄ post-exercise) and cycling conditions
(beginning ⁄ end ⁄ recovery) of EO. Error bars represent inter-individual SD. For the mid ⁄ anterior insular cortex but not for the motor cortex, a main effect of
Time was revealed within the cycling conditions. Significant changes from post hoc t-tests are depicted by **P 0.01; *P 0.05; n.s., not significant.
Muscle fatigue-induced intracortical communication 2039
ª 2011 The Authors. European Journal of Neuroscience ª 2011 Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 34, 2035–2042

6. from top down. From anatomical observations, evidence exists for an
afferent processing via lamina I spino-thalamic pathway involved in
homeostatic regulation in humans: small-diameter afferents relating
information on the physiological status of various tissues – for
example pain (nociceptors), heat ⁄ cold (thermoreceptors), acidic pH
(metaboreceptors) and hypo-osmolarity (osmoreceptors), monosynap-
tically activate neurons, which originate in the lamina I of the
superficial spinal dorsal horn (Craig et al., 2000). In primates, lamina I
neurons project not only to various brainstem homeostatic integration
sites, but also to two sides of the contralateral thalamus, namely, the
posterior part of the ventral medial nucleus and the ventral caudal part
of the medial dorsal nucleus, with the former region being assumed to
further project topographically to the mid ⁄ anterior insular cortex
(Craig et al., 1994). Lamina I neurons have been shown to provide an
ascending pathway for automatic cardiorespiratory adjustments to
muscular work (Wilson et al., 2002), and may be of essential
relevance also during prolonged cycling (Rauch et al., 2005).
From the mid ⁄ anterior insular cortex efferents were found to project
to the primary somatosensory cortex (Flynn et al., 1999), which in
turn has been shown to be directly connected with the motor cortex in
cats (Porter Sakamoto, 1988). Alternatively, intracortical commu-
nication between the mid ⁄ anterior insular and the motor cortex may
occur via relay stations such as motor control areas, for example the
supplementary motor area, which has been shown both to receive
input from mid insula (Flynn et al., 1999) and to project directly to the
motor cortex (Picard Strick, 1996). Thus, anatomical knowledge
gives reason to assume that afferent information about the physiolog-
ical status of the periphery might reach the motor cortex by motor
and ⁄ or sensory efferents from the mid ⁄ anterior insular cortex, which
in turn receives input from the lamina I pathway via thalamic relay.
Fatigue-induced changes in brain cortical activity
Indications that other regions are also involved in the cortical
phenomenon related to muscle fatigue come from a and b power
analyses before and after cessation of exhaustive exercise (Schneider
et al., 2009). These authors found significant increases in a and b
power within parietal as well as limbic regions after an incremental
cycling exercise. In part, we can replicate their findings revealing an
increased b power over parietal regions in our data but, moreover,
widespread increases in a and b power reaching significance were
found in Brodmann area 11 (medial orbitofrontal cortex, mOFC).
Termination of an unpleasant or painful event can be rewarding
(Seymour et al., 2005) – an emotional experience the mOFC has
consistently been shown to play a prominent role in (McClure et al.,
2004; Xu et al., 2009; Sellitto et al., 2010). Moreover, the mOFC has
been demonstrated to be involved in processing mental fatigue
(Tajima et al., 2010), which has been associated with physical fatigue
not only in various disorders such as Parkinson’s disease (Lou et al.,
2001) or multiple sclerosis (Schreurs et al., 2002), but also in
exercising healthy humans (Marcora et al., 2009). Furthermore, the
mOFC is associated with a ‘quasi-hyperbolic’ time-discounting
function (Laibson, 1997), reflecting a trade-off between short-term
and long-term reward-related decision outcomes, and has been shown
to be preferentially activated for choices involving an immediate
reward (McClure et al., 2004). In the context of an exhaustive
exercise, reward has to be postponed, but instead an increasing
perception of discomfort has to be endured. A prominent increase in
power within the mOFC immediately after termination of cycling
might be interpreted as a consequence of a prolonged deactivation of
this structure.
Limitations
Fatigue-induced lagged phase synchrony does not give evidence about
any directional properties of functional interconnectivity, i.e. whether
mid ⁄ anterior insular activity influences motor cortex activity or vice
versa. Based on previous anatomical findings showing efferent
pathways from the mid ⁄ anterior insular cortex projecting onto the
motor cortex, an increasing influence from the mid ⁄ anterior insular
cortex to the motor cortex seems more likely than vice versa.
Moreover, we assume that active inhibitory influence of the
mid ⁄ anterior insular cortex has been exerted on motor cortex activity
for the following reason: a ⁄ l power, which has been shown to
inversely relate to cortical activity, significantly decreased within the
mid ⁄ anterior insular cortex, whereas only a trend of decrease within
the motor cortex was detected at the end of fatiguing exercise with
respect to its beginning. Thus, neuronal activity in the insular cortex
may have increased, but no proportional increase could be observed in
M1.
Conclusion
Data from the present study revealed enhanced lagged phase
synchronization between the mid ⁄ anterior insular and the motor
cortex at the end of a fatiguing cycling exercise, demonstrating a
fatigue-induced increase in communication between those regions.
These new insights provide a valuable basis for further studies
investigating cortical mechanisms of supraspinal fatigue. Applying
causal analysis methods, the motor cortex could be tested on
inhibitory ⁄ excitatory influences exerted from the mid ⁄ anterior insular
cortex. Moreover, it remains to be established whether the motor
cortex is further influenced directly or indirectly by other regions, for
example the prefrontal cortex involved in motivational processes.
Supporting Information
Additional supporting information can be found in the online version
of this article:
Fig. S1. EEG power spectra calculated at eight different scalp
locations are depicted. The color filled curve indicates power of the
processed data at the beginning of cycling (first measurement period)
in the theta (orange), delta (yellow), alpha (green), beta (blue), and
gamma (black) band. In red, the power spectrum measured during the
last minute of cycling before exhaustion is shown.
Please note: As a service to our authors and readers, this journal
provides supporting information supplied by the authors. Such
materials are peerreviewed and may be re-organized for online
delivery, but are not copyedited or typeset by Wiley-Blackwell.
Technical support issues arising from supporting information (other
than missing files) should be addressed to the authors.
Acknowledgement
The present work was in part supported by the Swiss National Science
Foundation (SNF, Grant Nr. 320030-777111).
Abbreviations
EC, eyes closed; EEG, electroencephalography; EO, eyes open; fMRI,
functional magnetic resonance imaging; mOFC, medial orbitofrontal cortex;
ROI, region of interest; RPE, ratings of perceived exertion; sLORETA,
standardized low-resolution brain electromagnetic tomography; VO2peak, peak
oxygen consumption.
2040 L. Hilty et al.
ª 2011 The Authors. European Journal of Neuroscience ª 2011 Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 34, 2035–2042

7. References
Baciu, M.V., Bonaz, B.L., Papillon, E., Bost, R.A., Le Bas, J.F., Fournet, J.
Segebarth, C.M. (1999) Central processing of rectal pain: a functional MR
imaging study. Am. J. Neuroradiol., 20, 1920–1924.
Bigland-Ritchie, B., Johansson, R., Lippold, O.C. Woods, J.J. (1983)
Contractile speed and EMG changes during fatigue of sustained maximal
voluntary contractions. J. Neurophysiol., 50, 313–324.
Binkofski, F., Schnitzler, A., Enck, P., Frieling, T., Posse, S., Seitz, R.J.
Freund, H.J. (1998) Somatic and limbic cortex activation in esophageal
distention: a functional magnetic resonance imaging study. Ann. Neurol., 44,
811–815.
Borg, G. (1975). Simple rating for estimation of perceived exertion. In Borg, G.
(Ed.), Physical Work and Effort. Pergamon, New York, pp. 39–46.
Brannan, S., Liotti, M., Egan, G., Shade, R., Madden, L., Robillard, R.,
Abplanalp, B., Stofer, K., Denton, D. Fox, P.T. (2001) Neuroimaging of
cerebral activations and deactivations associated with hypercapnia and
hunger for air. Proc. Natl. Acad. Sci. USA, 98, 2029–2034.
Cannon, W.B. (1935) Stresses and strains of homeostasis. Am. J. Med. Sci.,
189, 13–14.
Casey, K.L. (1999) Forebrain mechanisms of nociception and pain: analysis
through imaging. Proc. Natl. Acad. Sci. USA, 96, 7668–7674.
Craig, A.D. (2003a) Interoception: the sense of the physiological condition of
the body. Curr. Opin. Neurobiol., 13, 500–505.
Craig, A.D. (2003b) A new view of pain as a homeostatic emotion. Trends
Neurosci., 26, 303–307.
Craig, A.D., Bushnell, M.C., Zhang, E.T. Blomqvist, A. (1994) A thalamic
nucleus specific for pain and temperature sensation. Nature, 372, 770–773.
Craig, A.D., Chen, K., Bandy, D. Reiman, E.M. (2000) Thermosensory
activation of insular cortex. Nat. Neurosci., 3, 184–190.
Davidson, R.J. (1988) EEG measures of cerebral asymmetry: conceptual and
methodological issues. Int. J. Neurosci., 39, 71–89.
Evans, A.C., Collins, D.L., Mills, S.R., Brown, E.D., Kelly, R.L. Peters,
T.M. (1993) 3D statistical neuroanatomical models from 305 MRI volumes.
Nuclear Science Symposium and Medical Imaging Conference, 1993 IEEE
Conference Record., City. pp. 1813–1817 vol.1813.
Evans, K.C., Banzett, R.B., Adams, L., McKay, L., Frackowiak, R.S.
Corfield, D.R. (2002) BOLD fMRI identifies limbic, paralimbic,
and cerebellar activation during air hunger. J. Neurophysiol., 88, 1500–1511.
Fitts, R.H. Holloszy, J.O. (1976) Lactate and contractile force in frog muscle
during development of fatigue and recovery. Am. J. Physiol., 231, 430–433.
Florian, G., Andrew, C. Pfurtscheller, G. (1998) Do changes in coherence
always reflect changes in functional coupling? Electroencephalogr. Clin.
Neurophysiol., 106, 87–91.
Flynn, F.G., Benson, D.F. Ardila, A. (1999) Anatomy of the insula –
functional and clinical correlates. Aphasiology, 13, 55–78.
Ftaiti, F., Kacem, A., Jaidane, N., Tabka, Z. Dogui, M. (2010) Changes in
EEG activity before and after exhaustive exercise in sedentary women in
neutral and hot environments. Appl. Ergon., 41, 806–811.
Fuchs, M., Kastner, J., Wagner, M., Hawes, S. Ebersole, J.S. (2002) A
standardized boundary element method volume conductor model. Clin.
Neurophysiol., 113, 702–712.
Gandevia, S.C., Allen, G.M., Butler, J.E. Taylor, J.L. (1996) Supraspinal
factors in human muscle fatigue: evidence for suboptimal output from the
motor cortex. J. Physiol., 490, 529–536.
Henderson, L.A., Bandler, R., Gandevia, S.C. Macefield, V.G. (2006)
Distinct forebrain activity patterns during deep versus superficial pain. Pain,
120, 286–296.
Hilty, L., Jancke, L., Luechinger, R., Boutellier, U. Lutz, K. (2010)
Limitation of physical performance in a muscle fatiguing handgrip exercise
is mediated by thalamo-insular activity. Hum. Brain Mapp., doi: 10.1002/
hbm.21177 [Epub ahead of print].
Hilty, L., Lutz, K., Maurer, K., Rodenkirch, T., Spengler, C.M., Boutellier, U.,
Jancke, L. Amann, M. (2011) Spinal opioid receptor-sensitive muscle
afferents contribute to the fatigue-induced increase in intracortical inhibition
in healthy humans. Exp. Physiol., 96, 505–517.
Jung, T.-P., Makeig, S., Bell, A.J. Sejnowski, T.J. (1998). Independent
component analysis of electroencephalographic and eventrelated potential
data. In Poon, P. Brugge, J. (Eds), Auditory Processing and Neural
Modeling. Plenum press, New York, pp. 189–197.
Jurcak, V., Tsuzuki, D. Dan, I. (2007) 10 ⁄ 20, 10 ⁄ 10, and 10 ⁄ 5 systems
revisited: their validity as relative head-surface-based positioning systems.
Neuroimage, 34, 1600–1611.
Kohler, G. Boutellier, U. (2005) The generalized force-velocity relationship
explains why the preferred pedaling rate of cyclists exceeds the most efficient
one. Eur. J. Appl. Physiol., 94, 188–195.
Kupers, R.C., Svensson, P. Jensen, T.S. (2004) Central representation of
muscle pain and mechanical hyperesthesia in the orofacial region: a positron
emission tomography study. Pain, 108, 284–293.
Laibson, D. (1997) Golden eggs and hyperbolic discounting. Quart. J. Econ.,
112, 443–478.
Laufs, H., Kleinschmidt, A., Beyerle, A., Eger, E., Salek-Haddadi, A.,
Preibisch, C. Krakow, K. (2003) EEG-correlated fMRI of human alpha
activity. Neuroimage, 19, 1463–1476.
Liotti, M., Brannan, S., Egan, G., Shade, R., Madden, L., Abplanalp, B.,
Robillard, R., Lancaster, J., Zamarripa, F.E., Fox, P.T. Denton, D. (2001)
Brain responses associated with consciousness of breathlessness (air hunger).
Proc. Natl. Acad. Sci. USA, 98, 2035–2040.
Liu, J.Z., Dai, T.H., Sahgal, V., Brown, R.W. Yue, G.H. (2002) Nonlinear
cortical modulation of muscle fatigue: a functional MRI study. Brain Res.,
957, 320–329.
Liu, J.Z., Shan, Z.Y., Zhang, L.D., Sahgal, V., Brown, R.W. Yue, G.H.
(2003) Human brain activation during sustained and intermittent submax-
imal fatigue muscle contractions: an FMRI study. J. Neurophysiol., 90,
300–312.
Lou, J.S., Kearns, G., Oken, B., Sexton, G. Nutt, J. (2001) Exacerbated
physical fatigue and mental fatigue in Parkinson’s disease. Mov. Disord., 16,
190–196.
Marcora, S.M., Staiano, W. Manning, V. (2009) Mental fatigue impairs
physical performance in humans. J. Appl. Physiol., 106, 857–864.
Martin, P.G., Weerakkody, N., Gandevia, S.C. Taylor, J.L. (2008) Group III
and IV muscle afferents differentially affect the motor cortex and motoneu-
rones in humans. J. Physiol., 586, 1277–1289.
McClure, S.M., Laibson, D.I., Loewenstein, G. Cohen, J.D. (2004) Separate
neural systems value immediate and delayed monetary rewards. Science,
306, 503–507.
Nielsen, B., Hyldig, T., Bidstrup, F., Gonzalez-Alonso, J. Christoffersen,
G.R. (2001) Brain activity and fatigue during prolonged exercise in the heat.
Pflugers Arch., 442, 41–48.
Noakes, T.D., St Clair Gibson, A. Lambert, E.V. (2005) From catastrophe to
complexity: a novel model of integrative central neural regulation of effort
and fatigue during exercise in humans: summary and conclusions. Br. J.
Sports Med., 39, 120–124.
Nolte, G., Bai, O., Wheaton, L., Mari, Z., Vorbach, S. Hallett, M. (2004)
Identifying true brain interaction from EEG data using the imaginary part of
coherency. Clin. Neurophysiol., 115, 2292–2307.
Nybo, L. Nielsen, B. (2001) Perceived exertion is associated with an altered
brain activity during exercise with progressive hyperthermia. J. Appl.
Physiol., 91, 2017–2023.
O’Donnell, R.D., Berkhout, J. Adey, W.R. (1974) Contamination of scalp
EEG spectrum during contraction of cranio-facial muscles. Electroencepha-
logr. Clin. Neurophysiol., 37, 145–151.
Pascual-Marqui, R.D. (2002) Standardized low-resolution brain electromag-
netic tomography (sLORETA): technical details. Methods Find. Exp. Clin.
Pharmacol., 24(Suppl D), 5–12.
Pascual-Marqui, R.D. (2007a) Coherence and phase synchronization: general-
ization to pairs of multivariate time series, and removal of zero-lag
contributions. arXiv:0706.1776v3 [stat.ME] July 12, 2007. Available http://
arxiv.org/pdf/0706.1776.
Pascual-Marqui, R.D. (2007b) Instantaneous and lagged measurements of
linear and nonlinear dependence between groups of multivariate time series:
frequency decomposition. arXiv:0711.1455 [stat.ME], November 9, 2007.
Available http://arxiv.org/abs/0711.1455.
Pascual-Marqui, R.D., Lehmann, D., Koukkou, M., Kochi, K., Anderer, P.,
Saletu, B., Tanaka, H., Hirata, K., John, E.R., Prichep, L., Biscay-Lirio, R.
Kinoshita, T. (2011) Assessing interactions in the brain with exact low
resolution electromagnetic tomography (eLORETA). Philos. Trans. R. Soc.
Lond. A, 369, 3768–3784.
Pfurtscheller, G., Neuper, C. Krausz, G. (2000) Functional dissociation of
lower and upper frequency mu rhythms in relation to voluntary limb
movement. Clin. Neurophysiol., 111, 1873–1879.
Pfurtscheller, G., Brunner, C., Schlogl, A. Lopes da Silva, F.H. (2006) Mu
rhythm (de)synchronization and EEG single-trial classification of different
motor imagery tasks. Neuroimage, 31, 153–159.
Picard, N. Strick, P.L. (1996) Motor areas of the medial wall: a review of
their location and functional activation. Cereb. Cortex, 6, 342–353.
Porter, L.L. Sakamoto, K. (1988) Organization and synaptic relationships of
the projection from the primary sensory to the primary motor cortex in the
cat. J. Comp. Neurol., 271, 387–396.
Rasmussen, P., Stie, H., Nybo, L. Nielsen, B. (2004) Heat induced
fatigue and changes of the EEG is not related to reduced perfusion of
Muscle fatigue-induced intracortical communication 2041
ª 2011 The Authors. European Journal of Neuroscience ª 2011 Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 34, 2035–2042

8. the brain during prolonged exercise in humans. J. Therm. Biol., 29, 731–
737.
Rauch, H.G., St Clair Gibson, A., Lambert, E.V. Noakes, T.D. (2005) A
signalling role for muscle glycogen in the regulation of pace during
prolonged exercise. Br. J. Sports Med., 39, 34–38.
Reiman, E.M. (1997) The application of positron emission tomography to the
study of normal and pathologic emotions. J. Clin. Psychiatry, 58(Suppl 16),
4–12.
Schneider, S., Brummer, V., Abel, T., Askew, C.D. Struder, H.K. (2009)
Changes in brain cortical activity measured by EEG are related to individual
exercise preferences. Physiol. Behav., 98, 447–452.
Schreurs, K.M., de Ridder, D.T. Bensing, J.M. (2002) Fatigue in multiple
sclerosis: reciprocal relationships with physical disabilities and depression. J.
Psychosom. Res., 53, 775–781.
Sellitto, M., Ciaramelli, E. di Pellegrino, G. (2010) Myopic discounting of
future rewards after medial orbitofrontal damage in humans. J. Neurosci., 30,
16429–16436.
Seymour, B., O’Doherty, J.P., Koltzenburg, M., Wiech, K., Frackowiak, R.,
Friston, K. Dolan, R. (2005) Opponent appetitive-aversive neural
processes underlie predictive learning of pain relief. Nat. Neurosci., 8,
1234–1240.
St Clair Gibson, A. Noakes, T.D. (2004) Evidence for complex system
integration and dynamic neural regulation of skeletal muscle recruitment
during exercise in humans. Br. J. Sports Med., 38, 797–806.
Stam, C.J., Nolte, G. Daffertshofer, A. (2007) Phase lag index: assessment of
functional connectivity from multi channel EEG and MEG with diminished
bias from common sources. Hum. Brain Mapp., 28, 1178–1193.
Tajima, S., Yamamoto, S., Tanaka, M., Kataoka, Y., Iwase, M., Yoshikawa,
E., Okada, H., Onoe, H., Tsukada, H., Kuratsune, H., Ouchi, Y.
Watanabe, Y. (2010) Medial orbitofrontal cortex is associated with fatigue
sensation. Neurol. Res. Int., doi: 10.1155/2010/671421 [Epub ahead of
print].
Tataranni, P.A., Gautier, J.F., Chen, K., Uecker, A., Bandy, D., Salbe, A.D.,
Pratley, R.E., Lawson, M., Reiman, E.M. Ravussin, E. (1999)
Neuroanatomical correlates of hunger and satiation in humans using
positron emission tomography. Proc. Natl. Acad. Sci. USA, 96,
4569–4574.
Taylor, J.L., Butler, J.E., Allen, G.M. Gandevia, S.C. (1996) Changes in
motor cortical excitability during human muscle fatigue. J. Physiol., 490,
519–528.
Taylor, J.L., Petersen, N., Butler, J.E. Gandevia, S.C. (2000) Ischaemia after
exercise does not reduce responses of human motoneurones to cortical or
corticospinal tract stimulation. J. Physiol., 525, 793–801.
Wilson, L.B., Andrew, D. Craig, A.D. (2002) Activation of spinobulbar lamina
I neurons by static muscle contraction. J. Neurophysiol., 87, 1641–1645.
Xu, L., Liang, Z.Y., Wang, K., Li, S. Jiang, T. (2009) Neural mechanism of
intertemporal choice: from discounting future gains to future losses. Brain
Res., 1261, 65–74.
2042 L. Hilty et al.
ª 2011 The Authors. European Journal of Neuroscience ª 2011 Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 34, 2035–2042