Jo
ur
na
l o
f P
hy
si
ol
og
y
Most established knowledge of motor unit firing behaviour
is based on experiments performed on extremity muscles
and usually not exceeding 1 min duration. Studies with
longer recording periods have been performed, notably
control and training of individual motor units, with
experimental feedback given by the observed motor unit
(e.g. Basmajian, 1963), and experiments collecting a large
number of firings for statistical evaluation of motor unit
synchrony (Sears & Stagg, 1976). These experimental
protocols are not designed to ascertain the ‘natural’ firing
behaviour of motor units active for long periods of time,
such as low-threshold motor units in postural muscles.
The extensive motor control literature has therefore not
yet addressed those aspects that could be unique to the
control of motor units in postural muscles. In an earlier
publication we documented the phenomenon of motor
unit substitution in the human trapezius muscle (Westgaard
& De Luca, 1999), i.e. the recruitment of higher-threshold
motor units to replace lower-threshold, putative fatigued
motor units that stop firing. Motor unit substitution is an
old concept that has been reported in some scientific papers
and is often mentioned within the clinical community
(Person, 1974; Kato et al. 1981; Fallentin et al. 1993), but
few reports exist that properly document the phenomenon.
In the earlier publication where motor unit firing was
monitored for 10 min, we noted instances of substitution
coinciding with brief periods of reduced excitatory drive to
the motoneuron pool, manifest as a brief period of
reduced activity in the electromyographic signal recorded
by surface electrodes (‘EMG gaps’). When substitution
occurred in a few motor units, the rest remained active
throughout the recording period.
In another study we showed that the firing rates of the
human trapezius motor units were relatively stable and
independent of the overall activity level in the muscle, at
least in the case of slowly augmenting contractions or
contractions maintained at a set contraction level
(Westgaard & De Luca, 2001). An observation of particular
interest was that the firing rate of continuously active
motor units tended to decrease below the level of stable
firing after the first few seconds of a temporary increase in
surface electromyographic (EMG) activity, despite the
root mean square (RMS)-detected amplitude of the surface
Motor unit recruitment and derecruitment induced by brief
increase in contraction amplitude of the human trapezius
muscle
C. Westad *, R. H. Westgaard * and C. J. De Luca †
* Department of Industrial Economics and Technology Management, Norwegian University of Science and Technology, N-7491 Trondheim, Norway
and † NeuroMuscular Research Center, Boston University, Boston, MA 02215, USA
The activity pattern of low-threshold human trapezius motor units was examined in response to
brief, voluntary increases in contraction amplitude (‘EMG pulse’) superimposed on a constant
contraction at 4–7 % of the surface electromyographic (EMG) response at maximal voluntary
contraction (4–7 % EMGmax). EMG pulses at 15–20 % EMGmax were superimposed every minute on
contractions of 5, 10, or 30 min duration. A quadrifilar fine-wire electrode recorded single motor
unit activity and a surface electrode recorded simultaneously the surface EMG signal. Low-
threshold motor units recruited at the start of the contraction were observed to stop firing while
motor units of higher recruitment threshold stayed active. Derecruitment of a motor unit coincided
with the end of an EMG pulse. The lowest-threshold motor units showed only brief silent periods.
Some motor units with recruitment threshold up to 5 % EMGmax higher than the constant
contraction level were recruited during an EMG pulse and kept firing throughout the contraction.
Following an EMG pulse, there was a marked reduction in motor unit firing rates upon return of the
surface EMG signal to the constant contraction level, outlasting the EMG pulse by 4 s on average.
The reduction in firing rates may serve as a trigger to induce derecruitment. We speculate that the
silent periods following derecruitment may be due to deactivation of non-inactivating inward
current (‘plateau potentials’). The firing behaviour of trapezius motor units in these experiments
may thus illustrate a mechanism and a control strategy to reduce fatigue of motor units with
sustained activity patterns.
(Received 11 April 2003; accepted after revision 7 August 2003; first published online 8 August 2003)
Corresponding author R. H. Westgaard: Department of Industrial Economics and Technology Management, Norwegian
University of Science and Technology, N-7491 Trondheim, Norway. Email: rolf.westgaard@iot.ntnu.no
J Physiol (2003), 552.2, pp. 645–656 DOI: 10.1113/jphysiol.2003.044990
© The Physiological Society 2003 www.jphysiol.org

Jo
ur
na
l o
f P
hy
si
ol
og
y
EMG staying equal to or higher than the constant EMG
value. We considered that the temporary depression in
firing could be helpful in inducing substitution in that the
net excitatory input to already active motor units was
reduced and the motor units were thereby closer to
recruitment threshold. The aim of the present study is to
test this hypothesis. A secondary aim is to generate new
knowledge regarding the physiological processes that
promote substitution in the trapezius muscle. A procedure
of contractions at a constant level by the trapezius muscle,
but with short-duration increases in voluntary EMG super-
imposed every minute (‘EMG pulse’) was implemented.
Most experiments were carried out for 10 min, but some
experiments were extended to 30 min to look for evidence
of more frequent or longer-duration substitution in
putative fatigued motor units. The effect of EMG pulse
amplitude on substitution was examined in experiments
of 5 min duration, with EMG pulses of increasing
amplitude imposed every minute.
METHODS
Ten healthy subjects, three males and seven females, volunteered
for the study. The age ranged from 20 to 56 years. The experiments
were performed according to the Declaration of Helsinki. Each
subject read and signed an informed consent form approved by
the local Institutional Review Board prior to participating in the
study. We studied the trapezius muscle, and detected the surface
and intramuscular EMG signals. The indwelling EMG signal was
used to study firing behaviour of trapezius motor units. Force
developed by the trapezius cannot be reliably monitored due to
the complex biomechanics of the muscle synergies controlling
shoulder movement. The root-mean-square (RMS)-detected
trapezius surface EMG signal was therefore used as a proxy
indicator of trapezius force development. The surface EMG signal
was calibrated as a percentage of the RMS-detected EMG activity
at maximal voluntary contraction (%EMGmax).
Experimental procedures
Three experimental procedures were performed. The first
consisted of a contraction of 10 min duration with constant EMG
amplitude, with brief periods (nominally of 2–4 s duration) of
voluntary increase in muscle activity (‘EMG pulses’) super-
imposed every minute. The constant contraction amplitude
ranged from 4 to 7 % EMGmax in all experiments, determined by
the observation of a suitable number of motor units in brief trial
contractions before the start of the experiment proper. The peak
amplitude of the EMG pulses was targeted to 15–20 % EMGmax. In
the second procedure the contraction period was extended to
30 min; the first 10 min was performed in the same manner as in
the first procedure. It continued with a 10-min constant EMG
amplitude contraction without EMG pulses and finally the
procedure of the first 10 min was repeated from 20 to 30 min. The
third procedure consisted of a 5-min constant EMG amplitude
contraction with a series of four EMG pulses at 1 min intervals,
performed at increasing strength from 10 to 25 % EMGmax. At the
end of each procedure the contraction level was briefly reduced
and a ramp contraction was performed to re-examine recruitment
threshold of the motor units. The procedures were carried out in
a fixed order: first the 10-min contraction, then the 5-min
contraction, followed by the 30-min contraction and finally
another 10-min contraction, similar to the first. Between each
procedure, rest periods of at least 2 min were allowed. The
experiment was carried out with the subject seated. Straps placed
over the shoulders provided resistance to the attempted movement
of elevating the shoulders. The shoulder elevations were
performed bilaterally, but with EMG data always collected from
the left trapezius.
The EMG pulses were voluntary contractions controlled by the
subject and initiated by a vocal cue from the experimenter, who
watched a wall-mounted clock also visible to the subject. The
subjects attempted to elevate the shoulders until the trapezius
surface EMG signal, displayed on a visual display monitor in front
of them, reached the required level and then the contraction was
immediately reduced to the constant EMG amplitude. It was
emphasized in the instructions to the subjects that the EMG
response should not drop below the constant EMG amplitude.
The timing in the execution of the contractions was not controlled
except for the instruction that the EMG pulses should be brief. The
need for control in reaching the high point and the return to the
constant contraction baseline nevertheless necessitated feedback
control in the execution of the EMG pulses, which were exercised
in test contractions until an acceptable performance was achieved.
EMG recordings and analyses
The surface EMG signal was detected with an active differential
electrode with two circular areas, 6 mm in diameter and centre-
to-centre distance 20 mm, in skin contact. The electrode was
positioned with the medial recording area 20 mm lateral to the
midpoint of the line between the C7 spinous process and the
acromion (Jensen et al. 1993). The surface EMG signal was band-
pass filtered at 10–1000 Hz. The RMS-detected surface EMG
signal was averaged at a time resolution of 0.2 s for graphical
presentation purposes. The intramuscular EMG signal was
recorded with specialized quadrifilar wire electrodes. These
electrodes were constructed by bonding together four 50-mm
nylon-coated nickel–chrome alloy wires (‘Stablohm 800A’,
California Fine Wire Co, Grover Beach, CA, USA). The wire
bundle was cut transversely, exposing only the cross-section of the
wires. The wire bundle was placed in a 27-gauge needle and a hook
was formed at approximately 1 mm from the exposed end of the
wire. The needle was inserted to a depth of approximately 10 mm
at a location approximately 10 mm medial to the midpoint of a
line between the C7 spinous process and the acromion. The needle
was removed and the wire bundle remained lodged in the muscle.
Three pairs were chosen as the differential input to the amplifiers.
The signals were band-pass filtered from 1 to 10 kHz. All EMG
signals were stored on a digital recorder (DATaRec-A160, Racal-
Heim Systems GmbH, Bergisch Gladbach, Germany). The signals
were subsequently reconverted to an analog form and digitized at
a sampling rate of 50 kHz on a PC.
The intramuscular EMG signals were resolved into the individual
motor unit firing trains using the Precision Decomposition
technique (LeFever et al. 1982; De Luca & Adam, 1999). This
technique uses template matching, template updating, firing
probabilities and superposition resolution to identify the
individual firing times of the motor units with up to 100 %
accuracy (Mambrito & De Luca, 1984). The firing rates of the
motor units were obtained by inverting the time series of the inter-
pulse intervals. The firing rates were subsequently low-pass
filtered at 0.5 Hz. Inspection of the trains of firing rates showed
that the filtering tended to reduce the peak firing rates by
1–2 pulses per second (p.p.s.), but did not affect the estimates of
the post-pulse depression in firing.
C. Westad, R. H. Westgaard and C. J. De Luca646 J Physiol552.2