J Neurophysiol 110: 1869 –1881, 2013.
First published July 31, 2013; doi:10.1152/jn.00196.2013.
Frequency response of vestibular reflexes in neck, back, and lower limb
muscles
Patrick A. Forbes,1 Christopher J. Dakin,3 Alistair N. Vardy,1 Riender Happee,1
Gunter P. Siegmund,3,6 Alfred C. Schouten,1,2 and Jean-Sébastien Blouin3,4,5
1
Department of Biomechanical Engineering, Faculty of Mechanical, Maritime and Materials Engineering, Delft University of
Technology, Delft, The Netherlands; 2Laboratory of Biomechanical Engineering, Institute for Biomedical Technology
and Technical Medicine (MIRA), University of Twente, Enschede, The Netherlands; 3School of Kinesiology, University of British
Columbia, Vancouver, British Columbia, Canada; 4Brain Research Center, University of British Columbia, Vancouver,
British Columbia, Canada; 5Institute for Computing, Information and Cognitive Systems, University of British Columbia,
Vancouver, British Columbia, Canada; and 6MEA Forensic Engineers & Scientists, Richmond, British Columbia, Canada
Forbes PA, Dakin CJ, Vardy AN, Happee R, Siegmund GP,
Schouten AC, Blouin JS. Frequency response of vestibular reflexes in
neck, back, and lower limb muscles. J Neurophysiol 110: 1869 –1881,
2013. First published July 31, 2013; doi:10.1152/jn.00196.2013.—Vestibular pathways form short-latency disynaptic connections with neck
motoneurons, whereas they form longer-latency disynaptic and polysynaptic connections with lower limb motoneurons. We quantified
frequency responses of vestibular reflexes in neck, back, and lower
limb muscles to explain between-muscle differences. Two hypotheses
were evaluated: 1) that muscle-specific motor-unit properties influence the bandwidth of vestibular reflexes; and 2) that frequency
responses of vestibular reflexes differ between neck, back, and lower
limb muscles because of neural filtering. Subjects were exposed to
electrical vestibular stimuli over bandwidths of 0 –25 and 0 –75 Hz
while recording activity in sternocleidomastoid, splenius capitis, erector spinae, soleus, and medial gastrocnemius muscles. Coherence
between stimulus and muscle activity revealed markedly larger vestibular reflex bandwidths in neck muscles (0 –70 Hz) than back (0 –15
Hz) or lower limb muscles (0 –20 Hz). In addition, vestibular reflexes
in back and lower limb muscles undergo low-pass filtering compared
with neck-muscle responses, which span a broader dynamic range.
These results suggest that the wider bandwidth of head-neck biomechanics requires a vestibular influence on neck-muscle activation
across a larger dynamic range than lower limb muscles. A computational model of vestibular afferents and a motoneuron pool indicates
that motor-unit properties are not primary contributors to the bandwidth filtering of vestibular reflexes in different muscles. Instead, our
experimental findings suggest that pathway-dependent neural filtering,
not captured in our model, contributes to these muscle-specific responses. Furthermore, gain-phase discontinuities in the neck-muscle
vestibular reflexes provide evidence of destructive interaction between different reflex components, likely via indirect vestibular-motor
pathways.
vestibular reflexes; neck muscles; bandwidth; electrical vestibular
stimulation; neural filtering
of the oscillatory activity of a muscle
may provide insight regarding the neural source driving that
muscle. For example, corticomuscular oscillations have generally been observed in the 15- to 30-Hz bandwidth (Baker et al.
1997; Murthy and Fetz 1992; Salenius et al. 1997; Salmelin
and Hari 1994), whereas muscular oscillations arising from the
THE FREQUENCY CONTENT
Address for reprint requests and other correspondence: J.-S. Blouin, Univ. of
British Columbia, 210-6081 University Blvd., Vancouver, BC, Canada V6T
1Z1 (e-mail: [email protected]).
www.jn.org
reticular structures appear to occur in the 10- to 20-Hz range
(Blouin et al. 2007; Grosse and Brown 2003). Recently, vestibular reflexes elicited via electrical stimuli were found to
influence lower limb muscles during standing balance over a
bandwidth of 0 –25 Hz (Dakin et al. 2007, 2011). This bandwidth corresponds with the known dynamic range of the
vestibular system (Armand and Minor 2001; Huterer and
Cullen 2002), suggesting that the vestibular system influences
motor activity mainly over a 0- to 25-Hz bandwidth. The
short-latency, high-frequency neck-muscle responses to vestibular stimuli (Colebatch et al. 1994; Lee et al. 2008; Murofushi
et al. 2002; Watson and Colebatch 1998; Wu et al. 1999),
however, suggest that the frequency response of vestibular
reflexes operates at higher frequencies for the neck muscles.
These differences between body regions likely result in muscle-dependent variation in the expression of vestibular reflexes.
The present study aims to characterize the frequency response
of vestibular reflexes across neck, back, and lower limb muscles and to determine the mechanisms underlying potential
between-muscle differences.
Differences in the frequency response between the neck,
back, and lower limb vestibular reflexes may be due to differences in 1) the underlying firing rates of these muscles, or
2) the neural pathways mediating these reflexes. Motor unit
firing rate is known to affect the amplitude (Kilner et al. 2000;
Ushiyama et al. 2010) and frequency bandwidth of brainmuscle correlations (Ushiyama et al. 2012). The firing rates of
neck muscles motor units are generally higher (10 –15 Hz;
Falla and Farina 2008; Falla et al. 2010; Schomacher et al.
2012) than those of lower limb muscles (6 – 8 Hz; Dalton et al.
2009; Mochizuki et al. 2006, 2007) and therefore may be
responsive to vestibular stimuli over a wider bandwidth. Based
on these firing rate data, we hypothesized that the bandwidth of
vestibular reflexes will be larger for the neck muscles than for
the lower limb muscles. We first evaluated this hypothesis
experimentally by estimating the coherence between the input
electrical stimulus and muscle activity to determine the bandwidth of vestibular reflexes in each muscle. We then used a
simple computational model composed of a population of
vestibular afferents and a motoneuron pool to determine
whether changes in firing rate alone could explain these differences.
0022-3077/13 Copyright © 2013 the American Physiological Society
1869
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
Submitted 18 March 2013; accepted in final form 30 July 2013
1870
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
METHODS
Subjects
Eleven healthy subjects (10 males, 1 female; age 20 – 48 yr; mass
73 ⫾ 10 kg; height 1.73 ⫾ 0.10 m; mean ⫾ SD) with no self-reported
history of neurological disorders or injuries took part in 2 separate
experiments (8 in experiment 1; 4 in experiment 2; 1 subject performed both experiments). The experimental protocol was explained
before each experiment, and all subjects gave written, informed
consent. The experiment conformed to the Declaration of Helsinki and
was approved by the University of British Columbia’s Clinical Research Ethics Board.
Vestibular Stimuli
Vestibular stimulation was delivered to subjects using carbon
rubber electrodes (⬃9 cm2) in a binaural bipolar arrangement. The
electrodes were coated with Spectra 360 electrode gel (Parker Laboratories, Fairfield, NJ) and secured over the subject’s mastoid processes with an elastic headband. The stimuli were delivered as analog
signals via a data acquisition board (PXI-6289; National Instruments,
Austin, TX) to an isolated constant-current stimulator (STMISOL;
Biopac, Goleta, CA). The signals were generated offline using MATLAB
(The MathWorks, Natick, MA), and identical signals were presented
to all subjects.
Subjects were exposed to 90-s filtered white noise stochastic
vestibular stimuli (SVS) of two bandwidths (0 –25 and 0 –75 Hz). The
signals were normalized to provide equal amplitude per frequency,
resulting in different root-mean-square (RMS) amplitudes for each
bandwidth: 0.83 mA (0 –25 Hz) and 1.38 mA (0 –75 Hz). The 0- to
25-Hz stimulus was considered our baseline condition since it had
been used previously to characterize the frequency response of the
vestibular reflexes in the lower limb muscles (Dakin et al. 2007,
2010). The 0- to 75-Hz stimulus was chosen to examine the shortlatency (approximately 13–17 ms) and higher-frequency vestibular
reflexes previously observed in neck muscles following 500-Hz shorttone acoustic stimuli (Wu et al. 1999) or transient electrical vestibular
stimulation (see Rosengren et al. 2010 for review).
Protocol
The 1st experiment (8 participants) was performed to evaluate the
bandwidth and frequency response of vestibular reflexes across the
different muscles (i.e., our main hypotheses). Subjects stood on a
force plate (Bertec 4060-80; Bertec, Columbus, OH) with their feet
parallel and medial malleoli 2–3 cm apart. For each trial, subjects
were instructed to stand relaxed, lean forward slightly, close their
eyes, hold their arms by their sides, and rotate their torso and head
axially to the left (i.e., leftward yaw). Torso yaw was maintained at
30° using a torso-mounted laser (level with T2), and head yaw was
maintained at 90° relative to the feet (60° relative to the torso) using
a head-mounted laser. The head was also rotated in extension such
that the Reid plane was tilted up by 18° horizontally. This head
position maximizes the postural response to binaural bipolar electrical
vestibular stimulation in the anterior-posterior direction (Cathers et al.
2005; Fitzpatrick and Day 2004) along the line of action of the right
gastrocnemius and soleus muscles. Orienting the head 60° to the left
(relative to the torso) ensured activation of the right sternocleidomastoid (r-SCM) and the left splenius capitis (l-SPL) muscles. Each 90-s
stimulus (0 –25 and 0 –75 Hz) was applied twice in a random order.
Two additional trials were performed without vestibular stimulation,
one before and one after all stimulation trials, to estimate the frequency characteristics of the muscles in the tested posture and to
evaluate the effects of the digital filters (see Signal Analysis).
The first experiment applied an ideal axial torsion of the neck (60°)
and torso (30°) to ensure activity in the measured neck and back
muscles. However, responses across muscles may be affected by the
variation in descending commands and/or propriospinal reflex contributions specific to maintaining this posture. Therefore, the 2nd experiment (4 participants) was performed to evaluate and eliminate any
confounding effects of the amount of muscle length on the different
responses across muscles. We investigated this effect in the same 2
neck muscles (r-SCM and l-SPL) using a custom-built axial-torsion
device. Subjects wore a helmet that was restrained in yaw but could
move in all other degrees of freedom. This was accomplished using
two universal joints and a telescoping shaft, allowing subjects to sway
freely while applying torque through the neck at any orientation
(Dakin 2012). Subjects performed three tasks: 1) head free with the
head (60° relative to the torso) and torso (30° relative to the feet)
rotated to the left (HF-60), thereby replicating experiment 1; 2) head
restrained with the neck (60° relative to the torso) and torso (30°
relative to the feet) rotated to the left (HR-60) and applying a torque
generating equivalent muscle activity of r-SCM to that of task 1; and
3) head restrained with the head and torso facing forward (0°; HR-0)
and applying a torque generating equivalent muscle activity of r-SCM
to that of task 1. These tasks were chosen to evaluate, first, the effects
of restraining the head in yaw using the torsion device (task 2 vs.
1) and, second, the variation in neck-muscle length between the
two orientations (i.e., 60 vs. 0°, task 3 vs. 2). Only the 90-s, 0- to
75-Hz stimulus was applied in this experiment and repeated three
times to improve the frequency response estimates of vestibular
reflexes (coherence, gain, and phase; see Signal Analysis). The
torques required to match muscle activity across conditions were
established in preliminary experiments performed without stimulation where the RMS of the electromyogram (EMG) was matched
to the HF-60 condition (task 1).
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
Vestibulospinal pathways and time delays between their
activation and subsequent excitation/inhibition of motoneurons
differ across muscles. Vestibular signals form short-latency
disynaptic excitatory and inhibitory connections with neck
motoneurons (Shinoda et al. 1997) via different vestibular
nuclei (medial vs. lateral) and neural pathways (medial vs.
lateral vestibulospinal) compared with lower limb muscles
(Shinoda et al. 2006; Wilson and Peterson 1981; Wilson and
Schor 1999). The dynamic behavior of these pathways varies;
compared with medial vestibular neurons, lateral vestibular
neurons have a lower sensitivity over a wide range of frequencies and demonstrate a “cutoff” behavior by synchronizing
their firing with the depolarizing phase of high-frequency
sinusoidal stimuli (Uno et al. 2003). As such, these differences
may filter the descending vestibular signals to varying degrees.
In addition, the potential for multiple vestibular pathways
projecting on neck muscles (see reviews by Wilson and Schor
1999 and Goldberg and Cullen 2011) may result in their
destructive interaction due to a fixed time delay between them
(Matthews 1994). This manifests itself as discontinuities in the
phase relationship between an input stimulus and the muscles
response (Matthews 1994). To determine whether the frequency response of vestibular reflexes varies between muscles,
we characterized the gain and phase response of the vestibular
reflexes across muscles. We hypothesized that the gain and
phase of vestibular reflexes will differ between muscles because of different neural pathways and the resulting neural
filtering between the vestibular system and the motoneurons of
the neck, back, and lower limbs. This hypothesis was not
explored with our simple computational model due to the
limited information regarding the properties of various vestibular pathways.
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
Data Collection
Signal Analysis
Raw neck EMG data were first high-pass filtered using a phaseless
eighth-order Butterworth digital filter (⫺3 dB at 100 Hz). The high
cutoff was necessary to eliminate the stimulation artifact given the
close proximity of the neck intramuscular electrodes to the vestibular
stimulation site. To ensure that digital filtering did not substantially
alter the spectral characteristics of the rectified signals, the effect of
increasing cutoff frequencies (40, 60, 80, and 100 Hz) was evaluated
using data from the 0- to 25-Hz stimulation condition. The frequency
response of vestibular reflexes (coherence, gain, and phase; see
Coherence [-]
below) for all muscles was unaffected with exception of a slight
drop in gain at all frequency points with increasing cutoff frequency, thus ensuring that responses were unaffected by a highpass cutoff of 100 Hz.
Repeated trials within each subject were concatenated to create
either 180- (experiment 1) or 270-s (experiment 2) data records. Data
from all subjects (EMG and force) were then concatenated again to
create a single pooled data set for each condition. Before concatenation, the filtered EMG of each trial was normalized by its vector norm
to ensure equal contributions to the averaged response from all
subjects. Both the individual and pooled EMG data were full-wave
rectified, and the autospectra for the SVS and the activity of each
muscle as well as the cross-spectra between SVS and activity of each
muscle were calculated. Data were sectioned into segments of 1 s
before calculating the auto- and cross-spectra, and then the spectra
were averaged in the frequency domain. This yielded a frequency
resolution of 1 Hz for all spectra. A repeated-measures ANOVA was
performed on the frequencies of the peak values extracted from the
EMG autospectra of experiment 1 to evaluate the effects of muscle
type and stimulation (5 muscles ⫻ 3 stimuli: 0 –25 Hz, 0 –75 Hz, and
no stimulation). Pairwise comparisons with a Bonferroni correction
were made to evaluate differences between combinations of two
muscles. For the l-SPL, an increased frequency resolution was necessary to evaluate the frequency response of the vestibular reflexes at
low frequencies. For these purposes, data were reprocessed with a
segment length of 4 s, yielding a frequency resolution of 0.25 Hz.
Coherence was calculated to explore the bandwidth of the frequency response of the vestibular reflexes across muscles. Given that
the input signals were filtered white noise, coherence is a measure of
the linear relationship between the electrical vestibular (input) and
muscle activity (output) signals across the frequencies considered. At
each frequency point, coherence ranges from 0 for systems with no
linear relation to 1 for linear systems without noise (Pintelon and
Schoukens 2001). The 95% confidence limit for coherence spectra
was derived from the number of disjoint segments (individual subjects: 180 and 270, pooled subjects: 1,440 and 1,080; for experiments
1 and 2, respectively) to indicate frequencies where coherence was
significantly different from 0 (Halliday et al. 1995). Significant
changes in SVS-EMG coherence between different muscles were
identified using a difference of coherence (DoC) test (Amjad et al.
1997). The DoC test was applied on the Fisher transform (tanh⫺1) of
the coherency (square root of the coherence) values, evaluated for the
0- to 75-Hz stimulus condition and compared with a ␹2-distribution
with k-1 degrees of freedom (k is the number of muscles included in
the comparison). Since we were interested in the coherence difference
across the entire stimulus bandwidth (0 –75 Hz), pairwise comparisons were made to highlight differences between neck-neck, neckback, and neck-lower limb muscle combinations. Two additional DoC
tests were performed separately for r-SCM and l-SPL coherences
Gain [-]
Phase [rad]
0.2
0
r-mGAS
Intramuscular
10
0
Surface
−2π
0.1
−4π
0
10
0
20
40
60
Frequency [Hz]
80
−6π
−1
0
20
40
60
80
Frequency [Hz]
0
20
40
60
80
Frequency [Hz]
Fig. 1. Coherence, gain, and phase frequency estimates for the right medial gastrocnemius (r-mGAS) muscle measured using surface and intramuscular electrodes
during 0- to 75-Hz stimulus during experiment 2. Pooled responses include all subject data (n ⫽ 4). The horizontal lines in the coherence plot represent the level
above which the coherence is significant.
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
In the first experiment, EMG was collected from neck, back, and
lower limb muscles. Intramuscular EMG was recorded in the r-SCM
and l-SPL using pairs of 0.05-mm wire (Stablohm 800A; California
Fine Wire) inserted under ultrasound guidance (MicroMaxx; SonoSite; Blouin et al. 2007). One of the two wires of each electrode had
2–3 mm of exposed wire. This electrode design allowed the recording
of multiunit EMG potentials. All wires were placed near the center of
the horizontal cross-section of a muscle. In the r-SCM, the wire
always remained superficial to the readily identifiable cleidomastoid
subvolume (Kamibayashi and Richmond 1998). Both wires were
inserted at about the C4 level. Surface EMG was recorded for the right
erector spinae (r-ESP) muscle 2–3 cm laterally from midline at the
level of L3. Surface EMG was recorded for the right soleus (r-SOL)
and medial gastrocnemius (r-mGAS) muscles; r-mGAS was recorded
for comparison with previous studies. All EMG signals were amplified (⫻200 –2,000; NeuroLog; Digitimer, Hertfordshire, United Kingdom) and band-pass filtered (intramuscular 0.05–1,000 Hz; surface
10 –1,000 Hz). EMG, vestibular stimuli, and force plate data were
then digitized and recorded at 2,000 Hz via a digital data acquisition
board (PXI-4495; National Instruments) using a custom LabVIEW
software program (National Instruments).
In the second experiment, EMG was collected from the r-SCM,
l-SPL, and r-mGAS muscles only. Intramuscular EMG was recorded
in the r-SCM and l-SPL while both surface and intramuscular EMG
was recorded in the r-mGAS. We recorded both electrode types in
r-mGAS to eliminate the possibility that differences observed between
neck and lower limb muscles in experiment 1 were an artifact of using
surface and intramuscular electrodes. After all, the spectral properties
of surface EMG differ from those of intramuscular EMG (reviewed in
De Luca 1997 and Farina et al. 2004). From these experiments, we
observed equivalent frequency responses of vestibular reflexes (coherence, gain, and phase; see Signal Analysis) for both electrode types
of the r-mGAS during the 0- to 75-Hz stimulus for all conditions (see
head-free results in Fig. 1). This indicated that it was possible to
provide a direct comparison of responses from the two electrode types
across different muscles in experiment 1.
1871
1872
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
Vestibular Afferent-Motoneuron Pool Model
To assess whether differences in motoneuron properties explained
the different frequency responses we observed in neck and lower limb
muscles, we built a computational model of the vestibular reflexes
(Fig. 2A) and then varied the motoneuron firing rates and action
potential durations to encompass the behavior of the neck and lower
limb motoneurons. Coherence, gain, and phase estimates were then
used to evaluate our hypothesis, where it was expected that the lower
motoneuron firing rates and longer action potential durations would
limit the coherent bandwidth of SVS-EMG coupling. It is noted here
that we did not attempt to fit the model to the experimental data but
simply evaluated whether the simulated results matched any of the
muscle-dependent variations.
The model consisted of a pool of 300 motoneurons that received
input from 2 sources: tonic supraspinal input and vestibular afferent
input. Development details of the model (i.e., tonic supraspinal input,
vestibular afferent input, and motoneuron pool) can be found in the
APPENDIX, and information regarding the model simulation and parameter variation are provided here.
Model simulation. The tonic input (RT) was varied to reflect the
physiological range of motor unit firing rates in neck (10 –15 spikes
per second; Farina and Falla 2008; Falla et al. 2010; Schomacher et al.
2012), back (7– 8 spikes per second; Barkhaus et al. 1997), and lower
limb (6 – 8 spikes per second; Dalton et al. 2009; Mochizuki et al.
2006, 2007) muscles at the contraction levels expected in these
conditions (i.e., 15–30% maximal voluntary contraction). To examine
the effects of motoneuron firing rate on the frequency response of the
vestibular reflexes, we simulated firing frequencies ranging between 6
and 18 spikes per second. The required RT input was established in
preliminary simulations with natural vestibular afferent output (i.e.,
without electrical stimulation).
To simulate EMG signals, the motoneuron spike signals were
replaced by motor unit action potential (MUAP) waveforms. We used
the same wave shape as implemented by Stegeman et al. (2010) where
the half-pulse wave shape was defined by:
冉 冊关共
h(t) ⫽ 5sin
␲t
d⁄2
e
1
t
⫺1
␶ d⁄2
兲兴
for 0 ⬍ t ⱕ d ⁄ 2
which was mirrored in time and amplitude for d/2 ⬍ t ⬍ d and aligned
with the motoneuron spike signals at t ⫽ d/2. The duration (d) of the
MUAP was varied between 8 and 24 ms, reflecting the variation in
A
Motoneuron pool
Stimuli
MN1
Vestibular
afferents
(regular/
irregular)
Fig. 2. A: schematic of the vestibular-afferent
motoneuron (MN)-pool model that simulates
the electrical vestibular stimulation experiments. The motoneuron pool is innervated by
vestibular afferents (irregular and regular) as
well as a tonic input. The output is the summation of motor unit action potentials
(MUAPs) that correspond to the motoneuron
spiking. B: motor unit action potential profiles used for 1 of the stimulation parameter
variations. EMG, electromyography.
Rate to
spike train
MN2
Convolve
with
MUAP
Tonic Input
MN...
Rate to
spike train
MN300
B
MUAP profile
0 ms
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
24 ms
Model
EMG
Delay
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
obtained from experiment 2 to evaluate the effects of head restraint
(i.e., HF-60 vs. HR-60) and head orientation (i.e., HR-60 vs. HR-0).
A significance level of ␣ ⫽ 0.05 was used for all analyses.
Gain- and phase-frequency estimates were calculated from the
pooled data to describe the frequency response of the vestibular
reflexes at frequencies with significant coherence. Gain and phase
indicate the magnitude and timing of the output EMG relative to the
input SVS and were assessed with two primary objectives: 1) to
identify muscle-dependent filtering behavior as shown by gains and
phases that decrease with increasing frequency; and 2) to identify the
destructive interaction of multiple reflex pathways as shown by
phase-frequency discontinuities. Both would support the hypothesis
that pathway-dependent neural filtering plays a role in the variation of
vestibular reflexes across muscles. Gains were normalized within each
muscle to the mean gain at the lowest frequency point (1 Hz) across
the three stimulus conditions allowing comparison across muscles. As
a result, the gain is plotted without units. Normalized gains that
remained close to one up to higher frequencies indicate vestibular
reflexes that can respond over a wider bandwidth. Finally, coherence-,
gain-, and phase-frequency estimates of SVS-to-horizontal forces
were also calculated, and they verified that the postural responses
were similar to previous studies (Dakin et al. 2010; Mian and Day
2009). As a result, the force data were not presented in this manuscript.
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
the frequency of the peak gain and phase discontinuities, which
occurred at the same frequency point, as well as the magnitude of the
peak gain during the 0- to 75-Hz condition to assess the effects of
motor unit firing rate and MUAP duration (5 rates ⫻ 5 durations).
These particular features of gain and phase were chosen because they
were similar to responses observed in the experimental data.
spectral properties (Stegeman et al. 2010) and durations (Boonstra and
Breakspear 2012) expected in small (i.e., neck) and large (i.e., lower
limb) muscles, respectively (Fig. 2B). The damping factor (␶) smooths
the onset of the wave shape and was set at 0.18 ms as used in previous
studies (Boonstra and Breakspear 2012; Stegeman et al. 2010). The
amplitude of each MUAP was randomly distributed between 0 and 1
to account for variation in the number and position of the muscle
fibers within each motor unit with respect to measurement electrodes
(Boonstra and Breakspear 2012). In addition, the polarity of the
MUAPs was equally divided between being initially positive or
negative. The model output was the summation of all MUAPs delayed
by 15 ms to approximate neck-muscle vestibular reflex latencies
reported in the literature (Rosengren et al. 2010). Additional delays
were simulated to represent back and lower limb muscle latencies;
however, these data were not presented, as their effect was limited to
changes in the phase slope only.
The vestibular afferent-motoneuron pool model was implemented
in MATLAB (The MathWorks) and simulated using the MATLAB
Distributed Computing Toolbox and Engine for parallel processing on
a 12-core computer. The effect of 3 variables on the coherence, gain,
and phase estimates was evaluated: stimulus bandwidth (0 –25 and
0 –75 Hz), motor unit firing rate (6, 8, 12, 16, and 18 spikes per
second), and MUAP duration (8, 12, 16, 20, and 24 ms). Each
stimulus bandwidth and motor unit firing rate combination was
simulated 10 times each for a total of 100 simulations. The MUAP
convolution was then performed on each simulation for the 5 MUAP
durations, generating a total of 500 simulation results. The vestibular
afferent model was simulated for 90 s using the experimentally
applied vestibular signals with a discrete time step of 0.1 ms. The
resultant spike train was downsampled by 10, whereby the spike trains
were summed over bins of 1 ms. The motoneuron pool model was
then run with a discrete time step of 1 ms. To avoid potential start-up
effects, the 1st 1 s was removed from the data.
Coherence, gain, and phase estimates were calculated in the same
manner as described for the experimental data, although only 89
segments of 1 s were used. In the frequency domain, this produced a
resolution of 1 Hz. A repeated-measures ANOVA was performed on
Electrical Vestibular Stimuli and Filtered EMG
0-25 Hz
0-75 Hz
RESULTS
Experimental
Filtered, unrectified EMG signals from all subjects reflected
typical multiunit recordings and showed no obvious pattern of
covariance between the stimuli and muscle EMG data (see
sample data from experiment 1 in Fig. 3A). The matching
shapes of the autospectra of the rectified EMG for the stimulation and nonstimulation data indicated successful filtering of
the stimulation artifact in neck muscles (Fig. 3B; note the
autospectra of the nonstimulation trials were offset from the
stimulation trials only for better visualization). The repeatedmeasures ANOVA showed a significant effect of muscle type
(F4,28 ⫽ 46.6, P ⬍ 0.001) and no significant effect of stimulus
(F2,14 ⫽ 2.7, P ⫽ 0.103) on the frequency of autospectra peaks.
Within these individual data, the autospectra of neck muscles
showed peak values at higher frequencies (r-SCM: 16.4 ⫾ 2.0
Hz, l-SPL: 13.3 ⫾ 1.5 Hz) relative to the back (r-ESP: 7.5 ⫾
2.4 Hz) and lower limb muscles (r-mGAS: 11.0 ⫾ 1.9 Hz,
r-SOL: 8.4 ⫾ 0.9 Hz) when averaged across conditions (0 –25
Hz, 0 –75 Hz, and no stimulation). A pairwise comparison
revealed no significant difference between the no-stimulation
and either SVS condition for all muscles.
Frequency bandwidth of vestibular reflexes. Significant
SVS-EMG coherence was observed for all muscles in experiment 1 (Fig. 4A). Overall, the bandwidth of coherent frequencies was larger for neck muscles compared with lower limb and
B
Power Spectrum
5 mA
10
0
−2
10
Stimuli
−2
10
500 mV
−3
10
r-SCM
−2
10
500 mV
−3
10
l-SPL
−3
10
100 mV
−4
10
r-ESP
−1
500 mV
10
−2
r-mGAS
10
−2
200 mV
10
r-SOL
−3
10
0
10
0.5 s
1
10
2
10
Frequency [Hz]
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Fig. 3. Sample stimuli, EMG data, and power
spectra of experimental data. A: 0.5 s of the
0- to 25- and 0- to 75-Hz stimuli and corresponding filtered, unrectified EMG signals of
a typical subject from each measured muscle.
B: power spectra of the 0- to 25- and 0- to
75-Hz stimuli and corresponding pooled, filtered, and rectified EMG from all measured
muscles. The gray, dashed lines in EMG
results represent the nonstimulation trials,
offset from the stimulation conditions for
visualization purposes. r-SCM, right sternocleidomastoid; l-SPL, left splenius capitis;
r-ESP, right erector spinae; r-SOL, right soleus.
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
A
1873
1874
A
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
Coherence [-]
0.06
Gain [-]
Phase [rad]
r-SCM
0
10
0
B
0
−1π
0.03
10
−2π
−1
0.1
0
0.05
10
0
−1π
0.05
−2π
0
0
10
20
10
10
0
C
0
−1
0
0
0
10
20
10
20
−4π
0.04
r-ESP
0
10
−2π
10
0
D
0
0.02
−4π
−1
0.2
r-mGAS
0
10
−2π
−4π
10
0
E
0
0.1
−1
−6π
0.1
0
r-SOL
SVS−25
10
0
SVS−75
−2π
0.05
−4π
10
0
0
20
40
60
Frequency [Hz]
80
−1
−6π
0
20
40
60
80
Frequency [Hz]
0
20
40
60
80
Frequency [Hz]
Fig. 4. Coherence, gain, and phase frequency estimates for all muscles elicited by the 2 stimuli: 0 –25 and 0 –75 Hz. Pooled responses include all subject data
(n ⫽ 8). A: r-SCM. B: l-SPL. C: r-ESP. D: r-mGAS. E: r-SOL. Coherence, gain, and phase frequency estimates for l-SPL (B) using a segment length of 4 s are
included as insets in each plot to detail the low-frequency responses. The horizontal, segmented lines in the coherence plots represent the level above which the
coherence is significant. Solid lines in the phase plots were fitted onto the different regions of linearly decreasing phase for visualization using a linear regression
algorithm. SVS, stochastic vestibular stimuli.
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
l-SPL
0.1
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
A
bandwidth for either comparison (i.e., head restraint and head
orientation) in both muscles.
Gain and phase responses of the vestibular reflexes. Gain
and phase estimates of the vestibular stimuli-to-EMG relationship for all muscles in experiment 1 were plotted at frequency
points with significant coherence (Fig. 4). Back and lower limb
muscle gain and phase indicate a low-pass behavior, where
phase decayed monotonically over one or two regions and gain
was flat up to a cutoff frequency of approximately 16, 14, and
3 Hz for r-SOL, r-mGAS, and r-ESP, respectively. In r-SCM,
a discontinuity in phase was observed over the region where
coherence was not significant. This discontinuity was characterized by a positive shift in phase lag by ⬃90° at 20 Hz.
Within each region (the low, 1–15 Hz, and high, 22– 62 Hz,
frequencies), phase decreased linearly and monotonically, although with different slopes (Fig. 4). These were accompanied
with gains that were flat and close to 1 in the low-frequency
region and that increased by a factor of ⬃2 in the highfrequency region. In l-SPL, three separate linear regions of
monotonically decreasing phase were defined by inflection
points at 19 and 36 Hz (Fig. 4). An additional phase discontinuity (positive shift of ⬃90°) was observed at ⬃4 Hz, which
was more clearly revealed when analyzing the data with a
frequency resolution of 0.25 Hz (Fig. 4B, insets in l-SPL data).
l-SPL gains peaked at ⬃10 Hz to a maximum of 2–3 over a
range of 6 –18 Hz but otherwise remained close to 1 for all
other frequency points. The gain and phase of both neck
muscles were comparable across all subjects with the exception
of the r-SCM phase discontinuities, which were observed in 10
of 12 subjects.
We then examined whether the gain and phase characteristics of the r-SCM and l-SPL muscles observed in experiment 1
were associated with the length of these muscles. Gain and
phase estimates of these two neck muscles in the head-free
condition (HF-60) of experiment 2 were similar to the results of
experiment 1 (Fig. 6), with the exception of a less-defined
low-frequency (⬃10-Hz) peak in the l-SPL. This was attributed to the same subject with coherence that fell below the
significance threshold at low frequencies, and therefore no gain
peak was found. The most important result of this experiment
was that gain and phase responses were unaffected by either
the restraint in yaw (HR-60) or head orientation (HR-0).
B
200
40
All muscles
r-SCM vs l-SPL
Fig. 5. Difference of coherence test of measured muscles during the 0- to 75-Hz stimulus. A: comparison of all 5 muscles. Horizontal, segmented lines represent the significance
level for the ␹2-distribution (␣ ⫽ 0.05).
B: pairwise comparison of different muscle
combinations. Horizontal, segmented lines
represent the significance level for the ␹2distribution (P ⫽ 0.05). Comparison of the
neck (r-SCM and l-SPL) and lower limb
(r-mGAS) muscles showed the largest difference in coherence at low and high frequencies.
0
r-SCM vs r-ESP
l-SPL vs r-ESP
80
100
0
120
0
0
20
40
Frequency [Hz]
60
80
0
0
r-SCM vs r-mGAS
l-SPL vs r-mGAS
20
40
60
80
Frequency [Hz]
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
back muscles. In both lower limb muscles, well-defined peaks
in coherence spectra were observed at 4 – 8 and 11–16 Hz with
no significant coherence ⬎30 Hz. In the r-ESP muscle, significant coherence was limited to a range of about 1–15 Hz. A
distinct peak was seen at 2 Hz in r-ESP for all conditions, and
sporadic significant coherence was found around 7– 8 Hz.
Coherence was significant up to ⬃70 Hz in both r-SCM and
l-SPL. Both muscles exhibited coherence peaks between 5 and
8 Hz. In the r-SCM, coherence decreased below significance
between 18 and 25 Hz and thereafter remained significant up to
75 Hz with a broad peak at about 40 –50 Hz. In the l-SPL, an
additional low-frequency peak was observed at ⬃2 Hz, which
was more clearly revealed when analyzing the data with a
frequency resolution of 0.25 Hz (Fig. 4B, inset in l-SPL data).
These observations were confirmed by the DoC tests, which
revealed significant differences in the coherence estimates
between the five muscles at almost all stimulated frequencies
(Fig. 5A). Pairwise DoC revealed coherence in l-SPL to be
larger than r-SCM at low frequencies (⬍20 Hz) and r-SCM to
be larger than l-SPL at high frequencies (30 –70 Hz). Coherence in both neck muscles was larger than in r-ESP at all
frequencies where significant differences were detected. The
magnitude of the DoC test was largest when comparing neck
and lower limb muscles, with larger coherences in lower limb
muscles at low frequencies (⬍20 Hz) and presence of coherence in neck muscles at high frequencies (30 –70 Hz).
To determine whether these differences in coherence across
muscles were due to differences in muscle length, we examined the potential effects of muscle length on the bandwidth of
the neck vestibular reflexes. The SVS-EMG coherence of the
neck muscles in experiment 2 was similar to the results of
experiment 1, with the exception of a decrease in l-SPL
coherence at low frequencies (⬍20 Hz). This was attributed to
a single subject where coherence was below significance at
almost all frequencies below 20 Hz; all other subjects demonstrated the characteristic low-frequency peak. More importantly, the bandwidth of coherence was unaffected by either the
restriction of head yaw (HR-60) or head orientation (HR-0). A
slight reduction in coherence was observed in restrained conditions relative to the head-free condition; however, the DoC
test revealed significant differences at only sporadic frequency
points (no more than 6 nonadjacent points) across the entire
1875
1876
r-SCM
A
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
Coherence [-]
Gain [-]
0
10
−1π
10
−2π
−1
0.1
0
10
HF-60
HR-60
HR-0
0
0.05
−2π
10
0
0
20
40
60
Frequency [Hz]
80
−4π
−1
0
20
40
60
80
Frequency [Hz]
0
20
40
60
80
Frequency [Hz]
Fig. 6. Coherence, gain, and phase frequency estimates for neck muscles elicited by 0 –75 Hz with the head free at 60° (HF-60), head restrained at 60° (HR-60),
and head restrained at 0° (HR-0). Pooled responses include all subject data (n ⫽ 4). A: r-SCM. B: l-SPL. The horizontal, segmented lines in the coherence plots
represent the level above which the coherence is significant.
Together with the coherence results, this indicates that differences in neck-muscle length were not responsible for the
variation in bandwidth or frequency response of reflexes observed across the different muscles.
Simulation
The model yielded coherence and gain/phase relationships
(Fig. 7) that captured some but not all of the behavior observed
in the experimental data. The simulated stochastic stimuli (Fig.
7A) elicited coherence profiles that closely matched the stimulus-dependent experimental findings in the r-SCM. At the
largest stimulus bandwidth (0 –75 Hz), two distinct peaks were
observed in the coherence at ⬃14 and 40 Hz, separated by an
intermediate dip ⬃20 Hz. A low-frequency peak in the gain
and a phase discontinuity (in the form of an inflection point),
which occurred at the same frequency between 10 and 20 Hz,
were similar but not identical to the experimental neck data
(Fig. 7A). Above 20 Hz, the monotonically decreasing phase
captured the simulated delay of 15 ms.
Increasing the motor unit firing rates (Fig. 7B) decreased the
coherence across the entire bandwidth. The frequency of the
1st coherence peak increased with firing rate and at the highest
rate (18 spikes per second) the peak aligned with the firing
frequency of the motor units. Since the firing rate was simulated as an increase in the descending input, the relative
contribution from the vestibular afferent activity at all other
frequencies decreased. The frequency of the gain peak and
phase discontinuity (P ⬍ 0.001, F4,36 ⫽ 64.6) and the magnitude of the gain peak (P ⬍ 0.001, F4,36 ⫽ 34.4) increased (Fig.
7B) with motor unit firing rate. Increasing the MUAP duration
resulted in further reductions in coherence across the entire
bandwidth, although in particular at the highest frequencies
(Fig. 7C). The magnitude of the gain peak (P ⬍ 0.001, F4,36 ⫽
507.0) decreased with increasing MUAP duration (Fig. 7C) but
did not affect the frequency of the gain peak and phase
discontinuity (P ⫽ 0.115, F4,36 ⫽ 2.0). These results indicate
that although both input parameters affected the gain and phase
responses, neither was able to mimic the variation in frequency
responses observed experimentally across the measured muscles.
DISCUSSION
The aim of this study was to characterize the frequency
response of vestibular reflexes in different muscles. We specifically evaluated two hypotheses: 1) that variation in the
bandwidth of vestibular reflexes between muscles are due to
the muscle-specific motor unit properties such as firing rate and
action potential durations; and 2) that the gain and phase of
vestibular reflexes will differ between muscles because of
pathway-dependent neural filtering. The frequency response
of the neck-muscle vestibular reflexes exhibited coherence
across higher bandwidths compared with back and lower limb
muscles. Our mathematical model, however, did not support
our first hypothesis: variations in motor unit properties do not
appear to be the primary determinants of coherence bandwidth.
The discontinuities we observed in the stimulus-EMG phase
relationships of both neck muscles confirmed our second hypothesis and suggest that neural filtering may be responsible
for the muscle-specific responses to vestibular reflexes.
Muscle-Dependent Bandwidth of Vestibular Reflexes: Neural
Mechanisms?
Based on our numerical model, motor unit firing rates and
action potential duration do not solely explain the observed
differences in bandwidth between neck and lower limb vestibular reflexes. Although both motor unit properties had an effect
on the coherence, variations in firing rate and motor unit
waveform were unable to replicate the apparent low-pass
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
l-SPL
0
0.03
0
B
Phase [rad]
0.06
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
A
Coherence [-]
Gain [-]
1
1877
Phase [rad]
10
0.6
0
0-25 Hz
0-75 Hz
0.4
0
10
−1π
0.2
B
−2π
−1
10
0
1
10
0.6
6 sp/s
8 sp/s
12 sp/s
16 sp/s
18 sp/s
0
0.4
0
10
−1π
C
−2π
−1
10
0
1
10
0.6
8 ms
12 ms
16 ms
20 ms
24 ms
0
0.4
0
10
−1π
0.2
−2π
−1
10
0
0
20
40
60
Frequency [Hz]
80
0
20
40
60
80
Frequency [Hz]
0
20
40
60
80
Frequency [Hz]
Fig. 7. Simulated coherence, gain, and phase obtained from the vestibular afferent-motoneuron pool model using single-simulation results with different input
parameters. A: variation of input stimuli (0 –25 and 0 –75 Hz) with motor unit firing rate at 8 spikes per second (sp/s) and MUAP duration at 16 ms. B: variation
of motor unit firing rate (6, 8, 12, 16, and 18 sp/s) with 0- to 75-Hz stimulus and MUAP duration at 8 ms. C: variation of MUAP duration (8, 12, 16, 20, and
24 ms) with 0- to 75-Hz stimulus and motor unit firing rate at 16 sp/s. Vertical lines in the gain plot in B indicate the frequency at which gains peaked for each
simulated condition and that the effect was significant (P ⬍ 0.001). Horizontal lines in the gain plots in B and C indicate the magnitude of the gain peak for
each simulated condition and that the effect was significant (P ⬍ 0.001). The horizontal, segmented lines in the coherence plots represent the level above which
the coherence is significant. Gain and phase were not plotted at frequencies where the coherence was not significant.
behavior, limiting the response bandwidth to ⬍25 Hz, observed in back and lower limb muscles. Nonetheless, the model
captured the high-bandwidth behavior of neck muscles along
with some aspects of the gain and phase characteristics. These
results lead us to suggest that additional filtering in the neural
pathways plays a role in the back and lower limb muscles.
Although neural filtering was not included in our model, our
experimental data support this hypothesis. The frequency response of vestibular reflexes in the back and lower limb
muscles behave as a low-pass filter with a flat gain up to 14 –16
Hz in r-mGAS and r-SOL and 3 Hz in r-ESP. In contrast,
neck-muscle frequency responses were characterized by gains
near or above 1 up to 70 Hz with discrete changes in phase.
One possible neurophysiological mechanism responsible for
the differential filtering of the vestibular reflexes is the transmissibility of head movement information provided by the
different vestibular nuclei to either neck or lower limb motoneurons. During in vitro current stimulation in the guinea pig,
the response dynamics of lateral vestibular nucleus neurons
(LVNn) are fairly flat across a wide bandwidth (approximately
0 –50 Hz) compared with medial vestibular nucleus neurons
(MVNn), which drop off in firing rate sensitivity after the peak
frequency of resonance (⬃10 Hz; Uno et al. 2003). However,
LVNn possess lower firing rate sensitivities to input currents at
all frequencies and often demonstrate a cutoff behavior where
neurons synchronize their firing with the depolarizing phase of
the current input and no response is measured. It is conceivable
that similar phenomena play a role in our experiments whereby
the LVNn connecting to lower limb motoneurons are limited or
unable to transfer the high-frequency vestibular signals induced by the electrical stimulation.
An alternative mechanism is that the spinal circuitry prohibits high-frequency vestibular information from reaching the
back and lower limb motoneurons. In neck muscles, disynaptic
connections form both short-latency excitatory and inhibitory
pathways between the labyrinth and motoneurons (Wilson and
Schor 1999; Wilson and Yoshida 1969). Single vestibulospinal
neurons branch to multiple neck motoneurons whereby muscles activate in synergy to respond to individual canal stimulation (Perlmutter et al. 1998; Shinoda et al. 1992, 2006). On
the other hand, vestibulospinal and reticulospinal tracts connect to lower limb motoneurons via direct connections, which
are exclusively excitatory and indirect connections via interneurons (Davies and Edgley 1994; Grillner et al. 1970; Lund
and Pompeiano 1968; Pompeiano 1972; Shinoda et al. 1986;
Wilson and Yoshida 1969). In humans, vestibular input mod-
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
0.2
1878
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
Neck-Muscle Responses: Evidence of Multiple Reflex
Pathways
Our results suggest that multiple reflex pathways connect the
vestibular system to the neck muscles and that these pathways
may differ between different neck muscles. In l-SPL, gain
peaked at ⬃10 Hz and phase shifted by 90° at ⬃4 Hz, whereas
in r-SCM gain remained more or less flat and phase shifted by
90° at ⬃20 Hz. This phase shift in the neck muscles differed
considerably from the phase response observed in the lower
limb muscles and did not change when reorienting the head
forward. Furthermore, the variation across neck and lower limb
muscles was shown not to be due to electrode type because
vestibular reflex responses in a lower limb muscle (r-mGAS)
were equivalent when using the two recording techniques.
Similar behavior has been observed in the stretch reflex of
the human wrist during mechanical perturbations and attributed
to the interaction of two reflex mechanisms separated by a
fixed delay (Matthews 1993). At frequencies where the two
components are out of phase, destructive interference occurs
without invoking inhibition. For our results, this would imply
a delay of ⬃100 ms in l-SPL and ⬃25 ms in r-SCM and leads
to a question of what two (or more) reflex components could
induce this response in neck-muscle control. In the lower limb
of humans, the short- and medium-latency vestibulomyogenic
components have been hypothesized to originate from, respectively, the otoliths via reticulospinal pathways and the semicircular canals via vestibulospinal pathways (Britton et al.
1993; Cathers et al. 2005). Although canal/otolith interaction
cannot be excluded in neck muscles, this is unlikely because
both sensory end organs generate excitatory and inhibitory
postsynaptic potentials in neck motoneurons with response
latencies as low as 1 ms during electrical stimulation of hair
cells (Shinoda et al. 1997; Uchino et al. 1997; Wilson and
Maeda 1974).
Another possibility is that multiple pathways transmitting
similar information are involved, yielding destructive interaction caused by a time delay between them. Neuroanatomic and
neurophysiological evidence from animal studies regarding the
neck-muscle vestibular reflexes support this proposal. Although neck-muscle motoneurons connect to the labyrinth via
strong disynaptic pathways, more indirect circuits make dominant contributions to vestibular neck reflexes (reviewed in
Goldberg and Cullen 2011; Wilson and Schor 1999). For one,
neck muscles receive many different patterns of excitatory and
inhibitory input from the separate canals and maculae, and
these inputs arrive via multiple vestibulospinal tracts (Shinoda
et al. 1997; Sugiuchi et al. 2004). In addition, transection of the
medial vestibulospinal tract in decerebrate cats results in almost no change in the dynamic properties of either the canaldriven vertical vestibulocollic reflex (Thomson et al. 1995) or
the otolith spinal reflexes (Miller et al. 1982) for many muscles. Such
responses point toward additional pathways contributing to the
reflex responses, and several experimentally supported hypotheses have been put forward, including interneuron connections
from additional vestibulospinal, i.e., bulbospinal, pathways
(Bankoul et al. 1995; Isu et al. 1996), input from the interstitial
nucleus of Cajal (Fukushima et al. 1994), and input via reticulospinal pathways (Peterson et al. 1980). Although the vestibular reflexes in neck muscles observed in this study point
toward the contribution of multiple pathways, additional animal experiments would be needed to test this hypothesis during
head-neck balancing actions.
Study Limitations
A limitation of this study is the inability of our model to
replicate exactly the features of the vestibular reflex responses
of any of the experimentally measured muscles. This may be
due to several physiological features not implemented in this
study. First, the lack of full bandwidth dynamics of vestibular
afferent firing during electrical stimulation were not included
because no literature currently exists to describe this behavior
up to 75 Hz. However, because these properties would most
likely affect both neck and lower limb muscle responses, this
exclusion would not change our results. Second, the effects of
task dependence were not included in the model where ongoing
vestibular input contributing to motoneuron drive is required to
evoke vestibular reflexes, at least in the lower limb muscles
(Fitzpatrick et al. 1994; Luu et al. 2012). As a result, even if the
input provided by the natural firing of vestibular afferents did
not contribute to the motoneuron pool, a vestibular reflex
would still be evoked by the electrical stimulation. Finally, the
motoneuron pool model lacks the detail of more complex
models (Williams and Baker 2009) such as variable motoneuron recruitment thresholds and persistent inward currents.
Nonetheless, despite the above limitations, because the model
was unable to replicate the primary experimentally observed
muscle-dependent variation in responses [i.e., coherence bandwidth, phase (dis)continuity], it was considered sufficient to
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
ulates the transmission of spinal reflex pathways, facilitating
reciprocal Ia inhibition, group I nonreciprocal inhibition, and Ia
presynaptic inhibition (Iles and Pisini 1992; Kennedy and
Inglis 2002; Rossi et al. 1988), suggesting convergence of
vestibular and peripheral signals on interneurons. Spinal interneurons transferring vestibular information may therefore
limit signal transmission to below ⬃25 Hz as observed in our
experiments.
From a functional standpoint, the flat gain observed in neck
muscles indicates a system that responds effectively at all
frequencies. However, since the majority of these frequencies
lie above the frequency response of the mechanical system the
muscle is controlling, it might at first seem difficult to assign
functional value to these responses (Matthews 1994). From a
biomechanical control aspect, it would seem logical that neck
muscles can respond to sensory input across a wider bandwidth; after all, the dynamic range of the sensor must be
greater than the actuator, which must be greater than the
underlying mechanical system to ensure effective control
(Franklin et al. 2009). In lower limb muscles, vestibular reflexes function over a bandwidth of 25 Hz in upright stance,
whereas the system mechanics (i.e., SVS-to-mechanical sway)
extend only as high as ⬃5 Hz (Dakin et al. 2010; Fitzpatrick et
al. 1996). The bandwidth of head motion (i.e., position) during
both voluntary whole body movements and imposed force
perturbations to the head contains frequency components up to
20 Hz (Grossman et al. 1988; Pozzo et al. 1990; Viviani and
Berthoz 1975). Therefore, to maintain head-neck stabilization,
vestibular influence over neck-muscle activation must function
across a much higher bandwidth than it does for lower limb
muscles.
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
evaluate and disprove our first hypothesis. Future efforts to
simulate the physiological mechanisms thought to be responsible for the neural filtering and/or multiple reflex pathways
may help to identify the exact mechanisms responsible for the
muscle-dependent variation of vestibular reflexes.
Regarding the interpretation of our experimental results, we
cannot rule out the possibility that part of the observed responses to electrical vestibular signals are of a nonvestibular
origin (Mian et al. 2010; Spiegel 1942). Although nonvestibular afferents may contribute to responses from the current
study, the short-duration and therefore high-frequency characteristics of vestibular-evoked myogenic potentials using shortduration (2-ms) electrical and nonelectrical stimuli (see Rosengren et al. 2010 for review) suggest that our responses are of a
vestibular origin.
The frequency response of vestibular reflexes in neck muscles showed both a larger bandwidth and discontinuities in
phase that were not present in vestibular reflexes in back or
lower limb muscles. Moreover, differences between the two
neck muscles tested here indicate that vestibular reflexes are
different in anatomically related muscles. Our numerical model
showed that variations in motor unit firing rates between
muscles do not substantially influence the coherence bandwidth. Instead, the extended dynamic range of neck-muscle
vestibular reflexes indicates the presence of additional neural
filtering in lower limb muscles. Discontinuities in the gain and
phase responses of vestibular reflexes in neck muscles provide
evidence of reflex component interaction and additional support for alternative vestibular-motor pathways to the wellestablished direct disynaptic connections.
APPENDIX
Tonic Supraspinal Input
The motoneuron pool received neural input from 100 tonically
firing descending fibers to produce baseline postural muscle activity.
Tonic supraspinal input was converted into spike trains for each fiber
via a stochastic Poisson process, such that each neuron received input
with the same spike rate statistics but with a different random
realization (Schuurmans et al. 2009).
Vestibular Afferent Input
␮A in the perilymphatic space explored the entire dynamic range
(0 –300 spikes per second; Goldberg et al. 1984). Equivalent afferent
output can be achieved with kinetic stimuli on the order of ⫾150°/s2
(Goldberg and Fernandez 1971). Fitzpatrick and Day (2004) combined this information with human perception experiments (Fitzpatrick et al. 2002) to estimate the afferent spike rate during constant
current electrical vestibular stimulation and suggested a 1-mA current
was equivalent to 4 spikes per second.
The vestibular afferent model used a voltage input signal (Vp) to
represent electrical stimulation, whereas our experiments applied
current stimuli (in milliamperes). Depending on the regularity of the
simulated afferents, an input Vp signal as low as 10 mV (irregular) and
as high as 120 mV (regular) can generate afferent spike rates at the
upper limit of the dynamic range (300 spikes per second; see Smith
and Goldberg 1986; Fig. 4B). Averaged across the entire population of
100 afferents in our model, a voltage input of 0.55 mV will generate
a modulation in afferent firing rate of 4 spikes per second, the
estimated equivalent of 1 mA. Therefore, all input stimuli were scaled
by 0.55 to capture the effects of externally applied vestibular stimulation.
Several studies have shown that the dynamic behavior of canal
afferents due to electrical stimulation is composed of a static phase
advance at all frequencies and a modest increase in gain with increasing frequencies (Ezure et al. 1983; Goldberg et al. 1982; Kim et al.
2011). Since these dynamics have only been described up to 20 Hz,
and considering that our stimuli extended up to 75 Hz, they were not
included in the model.
Motoneuron Pool
The motoneuron pool model was based on a discrete time integrate
and fire model (MacGregor and Oliver 1974) and included sodium and
potassium conductances. Arrival of descending and vestibular afferent
input increases the synaptic conductances by an amount (0.03 and
0.01, respectively) representing the amplitude of the postsynaptic
potential (Bashor 1998). Spike generation was modeled as a discrete
event (i.e., having no shape) and occurred when the membrane
potential exceeded the threshold. The threshold of each neuron was
variable with first-order dynamics and was dependent on the membrane potential, thereby capturing cell accommodation. The model
was parameterized according to previous efforts (Bashor 1998;
Schuurmans et al. 2009).
ACKNOWLEDGMENTS
We thank Anoek Geers for assistance with data collection and development
of the experimental setup.
GRANTS
A vestibular afferent model was used to estimate the modulated
output of a population of 100 vestibular afferents during electrical
stimulation (Goldberg et al. 1984; Smith and Goldberg 1986). The
afferent model captured the variability observed in the interspike
interval (ISI) among regular, intermediate, and irregular firing afferents. Discharge regularity was defined by the normalized coefficient
of variation (cv*), which was calculated by dividing the standard
deviation of the ISI by the average ISI and normalized to a standard
ISI (15 ms) to account for the variability of cv within each afferent
(see Goldberg et al. 1984 for more details). Five afferent types were
defined and distributed evenly among the 100 afferents and having
cv*s distributed in roughly octave steps ranging from 0.026 to 0.51
(Smith and Goldberg 1986). This captured the range of afferents
recorded in squirrel monkey otoliths (Fernandez and Goldberg 1976)
and semicircular canals (Goldberg and Fernandez 1971).
No literature was found that quantified afferent firing rates due to
externally applied electrical vestibular stimulation. Stimulation of
monkey vestibular afferents using anodal and cathodal currents of 70
This research was supported by 1) the Dutch Technology Foundation
(STW), which is part of the Netherlands Organization for Scientific Research
(NWO) and partly funded by the Ministry of Economic Affairs, Agriculture
and Innovation (P. A. Forbes, A. N. Vardy, R. Happee, and A. C. Schouten),
and 2) the Natural Sciences and Engineering Research Council of Canada
(C. J. Dakin, G. P. Siegmund, and J.-S. Blouin).
DISCLOSURES
G. P. Siegmund owns shares in a consulting company, and both he and the
company may derive benefit from being associated with this work.
AUTHOR CONTRIBUTIONS
P.A.F., C.J.D., A.N.V., G.P.S., A.C.S., and J.-S.B. conception and design of
research; P.A.F., C.J.D., G.P.S., and J.-S.B. performed experiments; P.A.F.
and A.C.S. analyzed data; P.A.F., C.J.D., A.N.V., R.H., G.P.S., A.C.S., and
J.-S.B. interpreted results of experiments; P.A.F. prepared figures; P.A.F.
drafted manuscript; P.A.F., C.J.D., A.N.V., R.H., G.P.S., A.C.S., and J.-S.B.
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
Conclusions
1879
1880
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
edited and revised manuscript; P.A.F., C.J.D., A.N.V., R.H., G.P.S., A.C.S.,
and J.-S.B. approved final version of manuscript.
REFERENCES
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
Amjad AM, Halliday DM, Rosenberg JR, Conway BA. An extended
difference of coherence test for comparing and combining several independent coherence estimates: theory and application to the study of motor units
and physiological tremor. J Neurosci Methods 73: 69 –79, 1997.
Armand M, Minor LB. Relationship between time- and frequency-domain
analyses of angular head movements in the squirrel monkey. J Comput
Neurosci 11: 217–239, 2001.
Baker SN, Olivier E, Lemon RN. Coherent oscillations in monkey motor
cortex and hand muscle EMG show task-dependent modulation. J Physiol
501: 225–241, 1997.
Bankoul S, Goto T, Yates B, Wilson VJ. Cervical primary afferent input to
vestibulospinal neurons projecting to the cervical dorsal horn: an anterograde and retrograde tracing study in the cat. J Comp Neurol 353: 529 –538,
1995.
Barkhaus PE, Periquet MI, Nandedkar SD. Quantitative motor unit action
potential analysis in paraspinal muscles. Muscle Nerve 20: 373–375, 1997.
Bashor DP. A large-scale model of some spinal reflex circuits. Biol Cybern
78: 147–157, 1998.
Blouin JS, Siegmund GP, Carpenter MG, Inglis JT. Neural control of
superficial and deep neck muscles in humans. J Neurophysiol 98: 920 –928,
2007.
Boonstra TW, Breakspear M. Neural mechanisms of intermuscular coherence: implications for the rectification of surface electromyography. J
Neurophysiol 107: 796 – 807, 2012.
Britton TC, Day BL, Brown P, Rothwell JC, Thompson PD, Marsden CD.
Postural electromyographic responses in the arm and leg following galvanic
vestibular stimulation in man. Exp Brain Res 94: 143–151, 1993.
Cathers I, Day BL, Fitzpatrick RC. Otolith and canal reflexes in human
standing. J Physiol 563: 229 –234, 2005.
Colebatch JG, Halmagyi GM, Skuse NF. Myogenic potentials generated by
a click-evoked vestibulocollic reflex. J Neurol Neurosurg Psychiatry 57:
190 –197, 1994.
Dakin CJ. The Development of Stochastic Vestibular Stimulation and its
Application to Dynamic Vestibular Evoked Responses (PhD thesis). Vancouver, Canada: Univ. of British Columbia, 2012.
Dakin CJ, Inglis JT, Blouin JS. Short and medium latency muscle responses
evoked by electrical vestibular stimulation are a composite of all stimulus
frequencies. Exp Brain Res 209: 345–354, 2011.
Dakin CJ, Luu BL, van den Doel K, Inglis JT, Blouin JS. Frequencyspecific modulation of vestibular-evoked sway responses in humans. J
Neurophysiol 103: 1048 –1056, 2010.
Dakin CJ, Son GM, Inglis JT, Blouin JS. Frequency response of human
vestibular reflexes characterized by stochastic stimuli. J Physiol 583: 1117–
1127, 2007.
Dalton BH, Harwood B, Davidson AW, Rice CL. Triceps surae contractile
properties and firing rates in the soleus of young and old men. J Appl Physiol
107: 1781–1788, 2009.
Davies HE, Edgley SA. Inputs to group II-activated midlumbar interneurons
from descending motor pathways in the cat. J Physiol 479: 463– 473, 1994.
De Luca CJ. The use of surface electromyography in biomechanics. J Appl
Biomech 13: 135–163, 1997.
Ezure K, Cohen MS, Wilson VJ. Response of cat semicircular canal afferents
to sinusoidal polarizing currents: implications for input-output properties of
second-order neurons. J Neurophysiol 49: 639 – 648, 1983.
Falla D, Farina D. Neuromuscular adaptation in experimental and clinical
neck pain. J Electromyogr Kinesiol 18: 255–261, 2008.
Falla D, Lindstrom R, Rechter L, Farina D. Effect of pain on the modulation
in discharge rate of sternocleidomastoid motor units with force direction.
Clin Neurophysiol 121: 744 –753, 2010.
Farina D, Merletti R, Enoka RM. The extraction of neural strategies from the
surface EMG. J Appl Physiol 96: 1486 –1495, 2004.
Fernandez C, Goldberg JM. Physiology of peripheral neurons innervating
otolith organs of squirrel-monkey I. Response to static tilts and to longduration centrifugal force. J Neurophysiol 39: 970 –984, 1976.
Fitzpatrick R, Burke D, Gandevia SC. Loop gain of reflexes controlling
human standing measured with the use of postural and vestibular disturbances. J Neurophysiol 76: 3994 – 4008, 1996.
Fitzpatrick R, Burke D, Gandevia SC. Task-dependent reflex responses and
movement illusions evoked by galvanic vestibular stimulation in standing
humans. J Physiol 478: 363–372, 1994.
Fitzpatrick RC, Day BL. Probing the human vestibular system with galvanic
stimulation. J Appl Physiol 96: 2301–2316, 2004.
Fitzpatrick RC, Marsden J, Lord SR, Day BL. Galvanic vestibular stimulation evokes sensations of body rotation. Neuroreport 13: 2379 –2383,
2002.
Franklin GF, Powell JD, Emami-Naeini A. Feedback Control of Dynamic
Systems. Upper Saddle River, NJ: Prentice Hall, 2009.
Fukushima K, Ohashi T, Fukushima J. Effects of chemical deactivation of
the interstitial nucleus of Cajal on the vertical vestibulo-collic reflex induced
by pitch rotation in alert cats. Neurosci Res 20: 281–286, 1994.
Goldberg JM, Cullen KE. Vestibular control of the head: possible functions
of the vestibulocollic reflex. Exp Brain Res 210: 331–345, 2011.
Goldberg JM, Fernandez C. Physiology of peripheral neurons innervating
semicircular canals of squirrel monkey. I. Resting discharge and response to
constant angular accelerations. J Neurophysiol 34: 635– 660, 1971.
Goldberg JM, Fernandez C, Smith CE. Responses of vestibular afferents in
the squirrel-monkey to externally applied galvanic currents. Brain Res 252:
156 –160, 1982.
Goldberg JM, Smith CE, Fernandez C. Relation between discharge regularity and responses to externally applied galvanic currents in vestibular
nerve afferents of the squirrel-monkey. J Neurophysiol 51: 1236 –1256,
1984.
Grillner S, Hongo T, Lund S. The vestibulospinal tract. Effects on alphamotoneurones in the lumbrosacral spinal cord. Exp Brain Res 10: 94 –120,
1970.
Grosse P, Brown P. Acoustic startle evokes bilaterally synchronous oscillatory EMG activity in the healthy human. J Neurophysiol 90: 1654 –1661,
2003.
Grossman GE, Leigh RJ, Abel LA, Lanska DJ, Thurston SE. Frequency
and velocity of rotational head perturbations during locomotion. Exp Brain
Res 70: 470 – 476, 1988.
Halliday DM, Rosenberg JR, Amjad AM, Breeze P, Conway BA, Farmer
SF. A framework for the analysis of mixed time series/point process
data–theory and application to the study of physiological tremor, single
motor unit discharges and electromyograms. Prog Biophys Mol Biol 64:
237–278, 1995.
Huterer M, Cullen KE. Vestibuloocular reflex dynamics during high-frequency and high-acceleration rotations of the head on body in rhesus
monkey. J Neurophysiol 88: 13–28, 2002.
Iles JF, Pisini JV. Vestibular-evoked postural reactions in man and modulation of transmission in spinal reflex pathways. J Physiol 455: 407– 424,
1992.
Isu N, Thomson DB, Wilson VJ. Vestibulospinal effects on neurons in
different regions of the gray matter of the cat upper cervical cord. J
Neurophysiol 76: 2439 –2446, 1996.
Kamibayashi LK, Richmond FJ. Morphometry of human neck muscles.
Spine (Phila Pa 1976) 23: 1314 –1323, 1998.
Kennedy PM, Inglis JT. Interaction effects of galvanic vestibular stimulation
and head position on the soleus H reflex in humans. Clin Neurophysiol 113:
1709 –1714, 2002.
Kilner JM, Baker SN, Salenius S, Hari R, Lemon RN. Human cortical
muscle coherence is directly related to specific motor parameters. J Neurosci
20: 8838 – 8845, 2000.
Kim KS, Minor LB, Della Santina CC, Lasker DM. Variation in response
dynamics of regular and irregular vestibular-nerve afferents during sinusoidal head rotations and currents in the chinchilla. Exp Brain Res 210:
643– 649, 2011.
Lee KJ, Kim MS, Son EJ, Lim HJ, Bang JH, Kang JG. The usefulness of
rectified VEMP. Clin Exp Otorhinolaryngol 1: 143–147, 2008.
Lund S, Pompeiano O. Monosynaptic excitation of alpha motoneurones from
supraspinal structures in cat. Acta Physiol Scand 73: 1–21, 1968.
Luu BL, Inglis JT, Huryn TP, Van der Loos HF, Croft EA, Blouin JS.
Human standing is modified by an unconscious integration of congruent
sensory and motor signals. J Physiol 590: 5783–5794, 2012.
MacGregor RJ, Oliver RM. A model for repetitive firing in neurons.
Kybernetik 16: 53– 64, 1974.
Matthews PB. Interactions between short-latency and long-latency components of the human stretch reflex during sinusoidal stretching. J Physiol 462:
503–527, 1993.
MUSCLE-DEPENDENT FREQUENCY RESPONSES OF VESTIBULAR REFLEXES
Shinoda Y, Ohgaki T, Sugiuchi Y, Futami T. Morphology of single medial
vestibulospinal tract axons in the upper cervical spinal-cord of the cat. J
Comp Neurol 316: 151–172, 1992.
Shinoda Y, Sugiuchi Y, Futami T, Ando N, Yagi J. Input patterns and
pathways from the six semicircular canals to motoneurons of neck muscles.
II. The longissimus and semispinalis muscle groups. J Neurophysiol 77:
1234 –1248, 1997.
Shinoda Y, Sugiuchi Y, Izawa Y, Hata Y. Long descending motor tract
axons and their control of neck and axial muscles. In: Neuroanatomy of the
Oculomotor System, edited by Büttner-Ennever JA. Amsterdam: Elsevier,
2006, p. 527–563.
Smith CE, Goldberg JM. A stochastic afterhyperpolarization model of
repetitive activity in vestibular afferents. Biol Cybern 54: 41–51, 1986.
Spiegel EA. Cephalogyric reactions of non-labyrinthine origin. Am J Physiol
135: 628 – 632, 1942.
Stegeman DF, van de Ven WJ, van Elswijk GA, Oostenveld R, Kleine BU.
The alpha-motoneuron pool as transmitter of rhythmicities in cortical motor
drive. Clin Neurophysiol 121: 1633–1642, 2010.
Sugiuchi Y, Kakei S, Izawa Y, Shinoda Y. Functional synergies among neck
muscles revealed by branching patterns of single long descending motortract axons. In: Brain Mechanisms for the Integration of Posture and
Movement, edited by Mori S, Stuart DG, and Wiesendanger M. Amsterdam:
Elsevier, 2004, p. 411– 421.
Thomson DB, Ikegami H, Wilson VJ. Effect of MLF transection on the
vertical vestibulocollic reflex in decerebrate cats. J Neurophysiol 74: 1815–
1818, 1995.
Uchino Y, Sato H, Suwa H. Excitatory and inhibitory inputs from saccular
afferents to single vestibular neurons in the cat. J Neurophysiol 78: 2186 –
2192, 1997.
Uno A, Idoux E, Beraneck M, Vidal PP, Moore LE, Wilson VJ, Vibert N.
Static and dynamic membrane properties of lateral vestibular nucleus
neurons in guinea pig brain stem slices. J Neurophysiol 90: 1689 –1703,
2003.
Ushiyama J, Masakado Y, Fujiwara T, Tsuji T, Hase K, Kimura A, Liu M,
Ushiba J. Contraction level-related modulation of corticomuscular coherence differs between the tibialis anterior and soleus muscles in humans. J
Appl Physiol 112: 1258 –1267, 2012.
Ushiyama J, Takahashi Y, Ushiba J. Muscle dependency of corticomuscular
coherence in upper and lower limb muscles and training-related alterations
in ballet dancers and weightlifters. J Appl Physiol 109: 1086 –1095, 2010.
Viviani P, Berthoz A. Dynamics of the head-neck system in response to small
perturbations: analysis and modeling in the frequency domain. Biol Cybern
19: 19 –37, 1975.
Watson SR, Colebatch JG. Vestibulocollic reflexes evoked by short-duration
galvanic stimulation in man. J Physiol 513: 587–597, 1998.
Williams ER, Baker SN. Circuits generating corticomuscular coherence
investigated using a biophysically based computational model. I. Descending systems. J Neurophysiol 101: 31– 41, 2009.
Wilson VJ, Maeda M. Connections between semicircular canals and neck
motoneurons in cat. J Neurophysiol 37: 346 –357, 1974.
Wilson VJ, Peterson BW. Vestibulospinal and reticulospinal systems. In:
Handbook of Physiology. Bethesda, MD: Am. Physiol. Soc., 1981, p.
667–702.
Wilson VJ, Schor RH. The neural substrate of the vestibulocollic reflex. What
needs to be learned. Exp Brain Res 129: 483– 493, 1999.
Wilson VJ, Yoshida M. Comparison of effects of stimulation of Deiters’
nucleus and medial longitudinal fasciculus on neck, forelimb, and hindlimb
motoneurons. J Neurophysiol 32: 743–758, 1969.
Wu CH, Young YH, Murofushi T. Tone burst-evoked myogenic potentials in
human neck flexor and extensor. Acta Otolaryngol 119: 741–744, 1999.
J Neurophysiol • doi:10.1152/jn.00196.2013 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
Matthews PB. The simple frequency-response of human stretch reflexes in
which either short-latency or long-latency components predominate. J
Physiol 481: 777–798, 1994.
Mian OS, Dakin CJ, Blouin JS, Fitzpatrick RC, Day BL. Lack of otolith
involvement in balance responses evoked by mastoid electrical stimulation.
J Physiol 588: 4441– 4451, 2010.
Mian OS, Day BL. Determining the direction of vestibular-evoked balance
responses using stochastic vestibular stimulation. J Physiol 587: 2869 –
2873, 2009.
Miller AD, Roossin PS, Schor RH. Roll tilt reflexes after vestibulospinal tract
lesions. Exp Brain Res 48: 107–112, 1982.
Mochizuki G, Ivanova TD, Garland SJ. Factors affecting the common
modulation of bilateral motor unit discharge in human soleus muscles. J
Neurophysiol 97: 3917–3925, 2007.
Mochizuki G, Semmler JG, Ivanova TD, Garland SJ. Low-frequency
common modulation of soleus motor unit discharge is enhanced during
postural control in humans. Exp Brain Res 175: 584 –595, 2006.
Murofushi T, Takegoshi H, Ohki M, Ozeki H. Galvanic-evoked myogenic
responses in patients with an absence of click-evoked vestibulo-collic
reflexes. Clin Neurophysiol 113: 305–309, 2002.
Murthy VN, Fetz EE. Coherent 25- to 35-Hz oscillations in the sensorimotor
cortex of awake behaving monkeys. Proc Natl Acad Sci USA 89: 5670 –
5674, 1992.
Perlmutter SI, Iwamoto Y, Barke LF, Baker JF, Peterson BW. Relation
between axon morphology in C-1 spinal cord and spatial properties of
medial vestibulospinal tract neurons in the cat. J Neurophysiol 79: 285–303,
1998.
Peterson BW, Fukushima K, Hirai N, Schor RH, Wilson VJ. Responses of
vestibulospinal and reticulospinal neurons to sinusoidal vestibular stimulation. J Neurophysiol 43: 1236 –1250, 1980.
Pintelon R, Schoukens J. System Identification: A Frequency Domain Approach. New York: IEEE Press, 2001.
Pompeiano O. Spinovestibular relations: anatomical and physiological aspects. Prog Brain Res 37: 263–296, 1972.
Pozzo T, Berthoz A, Lefort L. Head stabilization during various locomotor
tasks in humans. I. Normal subjects. Exp Brain Res 82: 97–106, 1990.
Rosengren SM, Welgampola MS, Colebatch JG. Vestibular evoked myogenic potentials: past, present and future. Clin Neurophysiol 121: 636 – 651,
2010.
Rossi A, Mazzocchio R, Scarpini C. Changes in Ia reciprocal inhibition from
the peroneal nerve to the soleus alpha-motoneurons with different static
body positions in man. Neurosci Lett 84: 283–286, 1988.
Salenius S, Portin K, Kajola M, Salmelin R, Hari R. Cortical control of
human motoneuron firing during isometric contraction. J Neurophysiol 77:
3401–3405, 1997.
Salmelin R, Hari R. Spatiotemporal characteristics of sensorimotor neuromagnetic rhythms related to thumb movement. Neuroscience 60: 537–550,
1994.
Schomacher J, Dideriksen JL, Farina D, Falla D. Recruitment of motor
units in two fascicles of the semispinalis cervicis muscle. J Neurophysiol
107: 3078 –3085, 2012.
Schuurmans J, de Vlugt E, Schouten AC, Meskers CG, de Groot JH, van
der Helm FC. The monosynaptic Ia afferent pathway can largely explain
the stretch duration effect of the long latency M2 response. Exp Brain Res
193: 491–500, 2009.
Shinoda Y, Ohgaki T, Futami T. The morphology of single lateral vestibulospinal tract axons in the lower cervical spinal-cord of the cat. J Comp
Neurol 249: 226 –241, 1986.
1881