Journal of Electromyography and Kinesiology 17 (2007) 57–73

www.elsevier.com/locate/jelekin

Muscles within muscles: Coordination of 19 muscle segments

within three shoulder muscles during isometric motor tasks
J.M.M. Brown a, J.B. Wickham b,*
, D.J. McAndrew a, X.-F. Huang a

a
Department of Biomedical Science, University of Wollongong, Wollongong 2500, Australia
b
Department of Anatomy and Physiology, School of Human Biosciences, La Trobe University, Vic. 3086, Australia

Received 10 April 2005; received in revised form 24 September 2005; accepted 26 October 2005

Abstract

The aim of the present study was to determine how the intra-muscular segments of three shoulder muscles were coordinated to pro-
duce isometric force impulses around the shoulder joint and how muscle segment coordination was influenced by changes in movement
direction, mechanical line of action and moment arm (ma).
Twenty male subjects (mean age 22 years; range 18–30 years) with no known history of shoulder pathologies, volunteered to partic-
ipate in this experiment. Utilising an electromyographic technique, the timing and intensity of contraction within 19 muscle segments of
three superficial shoulder muscles (Pectoralis Major, Deltoid and Latissimus Dorsi) were studied and compared during the production of
rapid (e.g. approximately 400 ms time to peak) isometric force impulses in four different movement directions of the shoulder joint (flex-
ion, extension, abduction and adduction).
The results of this investigation have suggested that the timing and intensity of each muscle segment’s activation was coordinated
across muscles and influenced by the muscle segment’s moment arm and its mechanical line of action in relation to the intended direction
of shoulder movement (e.g. flexion, extension, abduction or adduction). There was also evidence that motor unit task groups were
formed for individual motor tasks which comprise motor units from both adjacent and distant muscles.
It was also confirmed that for any particular motor task, individual muscle segments can be functionally classified as prime mover,
synergist or antagonist – classifications which are flexible from one movement to the next.
 2005 Elsevier Ltd. All rights reserved.

Keywords: Electromyography; Functional differentiation; Muscle segment; Shoulder muscles; Motor control; Motor unit task group

1. Introduction et al. [5] has suggested that large absolute differences in

It is now well accepted that motor units, within individ-
ual muscle segments of single skeletal muscles, can be inde-
pendently controlled by the central nervous system (CNS)
to produce particular motor outcomes [3,21]. This phe-
nomenon, which may be termed ‘functional differentiation’
[18], has been described within a number of individual skel-
etal muscles, including the Tensor Fascia Latae [18], the
Gluteus Maximus and Medius [16], the Triceps Surae [3],
the Biceps Brachii [2] and the Pectoralis Major [19]. Ettema
*
Corresponding author. Tel.: +61 3 94795725; fax: +61 3 94795784.
E-mail address: j.wickham@latrobe.edu.au (J.B. Wickham).
moment arms between different segments of a single muscle
partially explain this phenomenon.
More recent studies on the Deltoid [24] have shown that
skeletal muscle has a greater potential for functional differ-
entiation than hitherto accepted. Wickham and Brown [24]
have determined that the Deltoid muscle, commonly
described as having only three (anterior, middle and poster-
ior heads) functional muscle segments [10], is composed of
at least seven muscle segments which all have the potential
to be independently coordinated by the CNS. Functionally
independent muscle segments provided the CNS greater
flexibility to ‘‘fine tune’’ the activity of its Deltoid motor
units when controlling movements of the shoulder joint.

1050-6411/$ - see front matter  2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jelekin.2005.10.007
58 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73
Furthermore, Wickham and Brown [24] have suggested
that the seven individual muscle segments of the Deltoid
may be functionally classified as either prime mover, syner-
gist or antagonist muscle segments based upon their
mechanical lines of action (joint movement most likely
from approximation of a segment’s origin and insertion)
and periods of activation. Attempts to assign functional
classifications to individual muscle segments, classifications
hitherto used to describe whole muscle function [1] are use-
ful in characterising the activity of each individual muscle
segment. For example, the prime mover segments of the
Deltoid were found to have the most efficient mechanical
lines of action, the earliest and longest durations of activa-
tion and the highest amplitudes of myoelectric activity.
Deltoid muscle segments with increasingly divergent
mechanical lines of action (synergist segments) were found
to have increasingly later onsets, shorter periods of activa-
tion and lower intensities of myoelectric activity. Finally,
antagonist segments, with mechanical lines of action oppos-
ing the movement, were activated last, had the shortest
periods of activation and variable amplitudes of myoelec-
tric activity [24].
Applying functional classifications to muscle segments
within a single muscle is useful to provide a clearer under-
standing of how the CNS controls motor unit activity
within a muscle during a range of motor tasks. However,
there have been few studies into how the CNS controls
the muscle segments of individual muscles constituting a
muscle group around a particular joint. We do not as yet
understand how the muscle segments of individual muscles,
within a muscle group, are coordinated together to produce
motor tasks or how that coordination is affected by factors
such as movement direction and moment arm.
Therefore, the aim of the present study was to determine
how the muscle segments of individual shoulder muscles
were coordinated together to produce isometric force
impulses around the shoulder joint and how that coordina-
Fig. 1. The experimental setup. Depicted is one of the subjects seated in the experimental chair with the arm cast set at 20 of abduction in the frontal
plane. In this position subjects performed the static tasks of shoulder joint flexion, extension, abduction and adduction.
tion was influenced by movement direction, mechanical
line of action and moment arm. This investigation com-
pared the activation of 19 muscle segments within three
superficial shoulder muscles (Pectoralis Major, Deltoid
and Latissimus Dorsi) during the execution of rapid (e.g.
400 ms time to peak) isometric force impulses in four differ-
ent movement directions (shoulder-flexion, -extension,
-abduction and -adduction). Our intention was to under-
stand how individual muscle segments were coordinated
not only within a single muscle, but across a group of mus-
cles that controlled a single joint.
2. Methods
2.1. Procedures
Twenty male subjects (mean age 22 years; range 18–30
years) with no known history of shoulder pathologies, vol-
unteered to participate in this experiment.
The subjects sat in an adjustable dental chair within a
wire cage with their extended right upper limb positioned
snugly within an arm cast (Fig. 1). The arm cast was immo-
bilised by attachment to the wire cage to permit the sub-
jects to perform a series of isometric shoulder joint tasks.
The arm cast, so fitted, ensured that the right upper limb
was fixed securely in space and did not move, other than
due to unavoidable compression of its soft tissues, during
any of the isometric motor tasks investigated.
Each subject performed four isometric tasks; shoulder-
abduction, -adduction, -flexion and -extension, at 75% of
MVC, a movement time (MT) approximating 400 ms and
with an inter-trial period of at least 30 s. Each task was per-
formed with the right upper limb secured in 20 of abduc-
tion in the coronal plane with subject feedback on
movement accuracy being provided by an oscilloscope.
Ten trials were recorded from each isometric movement
task. Following an auditory trigger signal, each subject was
 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73 59

asked to match an electronically generated force–time
curve on an oscilloscope which provided the subject with
feedback on movement direction, movement speed and
the force of contraction. The order in which the trials were
performed was randomised to minimise muscular fatigue
and order effects. To permit normalisation of the intensity
data, three MVC contractions were recorded for each
movement direction at the beginning of the experiment.
The MVC contractions were recorded with an inter-trail
interval of 3 min.
2.2. Force recordings
Two load cells, connecting the arm cast to the wire cage,
recorded the isometric force produced by the subjects in the
coronal (abduction and adduction) and the sagittal (flexion
and extension) planes. The output of each load cell was
amplified by a DC amplifier.
2.3. Electromyographic recordings
Nineteen miniature gold bipolar surface electrodes
(6.5 mm inter-electrode distance; 1.6 mm active plates)
were located over the muscle segments of the three shoul-
der muscles investigated using anatomical landmarks
derived through cadaveric dissection (Fig. 2). The anatom-
ical criteria used to identify muscle segments within each of
the three muscles may be found in Wickham [23] and Wick-
ham et al. [25]. In short these included, distinctive origins/
insertions, architectural differences (strap vs. pennate

Fig. 2. Muscle segments of Pectoralis Major (P), the Deltoid (D) and the Latissimus Dorsi (L). ‘‘d’’ indicates bipolar surface electrode site. Note that in
this view segment L6 of the Latissimus Dorsi lies anterior to segment L5. Muscle segments were identified by obvious anatomical segmentation (e.g.
Deltoid) or by division of relatively homogenous muscle tissue (e.g. L1 and L2 of Latissimus Dorsi).
60 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73

fibered), intramuscular fascial thickenings and >10 differ-
ence in lines of action between adjacent segments. One or
more of these criteria needed to be present. In all 19 bipolar
surface electrodes were applied to each subject; seven over
the Deltoid and six each over the Pectoralis Major and the
Latissimus Dorsi (Fig. 1).
Electrode gel was injected into the electrode wells by
utilising a syringe coupled to a 1 mm cannula. The surface
electrodes were secured to the subject by double-sided tape
after the skin had been shaved, abraded and washed with
alcohol to reduce skin resistance. Reference surface elec-
trodes (3M paediatric electrodes) were positioned over
the acromion, clavicle and the anterior superior iliac spine.
The 19 bipolar surface electrodes and their reference elec-
trodes were connected to differential preamplifiers and
HUMTEC 100 EMG amplifiers (input impedance of
1 · 1013 X, CMRR of 110 db, signal to noise ratio of
1000:1, 1000k gain). The raw EMG signals were amplified
and filtered (10 Hz high pass and 1 kHz low pass; Butter-
worth filters) prior to storage.
The isometric force impulse and the raw EMG wave-
forms were stored on an IBM compatible microcomputer
after A–D conversion at 1 kHz. A 2000 ms recording per-
iod, commencing at least 200 ms before the beginning of
the each motor task, was recorded for each trial.
2.4. Data analysis
Data analysis of the EMG waveforms utilised a digital
signal processing (DSP) package for determinations of
muscle segment contraction intensity and timing (Fig. 3).
This program permitted the raw EMG signals to be pro-
cessed (rectified, low passed, high passed, etc.) prior to
analysis.
2.5. Timing analysis
Firstly, the time of force onset and the timing of force
peak were determined with the temporal features of muscle
segment activation determined in relation to these values
(Fig. 3).
To determine the timing of each muscle segment, the
raw EMG waveforms were rectified and smoothed with a
20 Hz low pass filter (Fig. 3). Threshold detectors (10%
peak amplitude), combined with visual analysis, were used
to locate the onset (ON), peak (PEAK) and end (OFF) of
activity within each muscle segment (Fig. 3). From this
data the duration of segmental activation could also be
determined. ON and OFF were determined in relation to
the beginning of the isometric force onset while PEAK
was measured in relation to the timing of peak isometric
force (Fig. 3).
2.6. Intensity analysis
Each EMG waveform, representing the activity of a seg-
ment within a particular trial was integrated to determine
the intensity of muscle segment contraction. The EMG
waveforms were full wave rectified and then integrated
(10 ms bins) to determine the muscle segment’s intensity
of activation or discharge rate (DR: mVms/ms). The dis-
charge rate was the average intensity of motor unit con-

Fig. 3. The analysis screen window that was used to locate the temporal and consequently the intensity parameters of the EMG signals. The top window
shows a representative raw EMG signal with its 20 Hz linear envelope displayed in the middle window. The middle window also displays the threshold
detector set at 10% of peak segmental intensity (horizontal line) that was used to locate the on/off positions of the EMG signals. Visual detection of the
signal was used when there was a low signal to noise ratio and the threshold detector could not determine an appropriate on/off location. The force record
from the 100 kg prime load cell is displayed in the bottom window. The duration of each burst was integrated (time reset; 1 ms bins) to give a value of
average V/ms (total V ms divided by duration). Representative data from one subject (segment D1) during a 75% MVC shoulder joint flexion task.
 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73
traction throughout the muscle segments period of activa-
tion. This value was then normalised against the discharge
rate found during the MVC contractions associated with
each particular movement direction.
2.7. Biomechanical analysis
Three-dimensional cartesian coordinates of each of the
19 muscle segments were determined through use of a ver-
tical milling machine. This sort of machine, which is com-
monly used in engineering workshops for shaping metal
very accurately (to within 0.01 mm), was considered more
than adequate for our biomechanical analysis. By securing
a cadaver (bisected at the waist) to a specially designed
frame in front of the milling machine, 3D coordinates of
each muscle segment and the axis of rotation of the shoul-
der joint were determined (Fig. 4). This was made possible
by securing custom made pointers to the mill vice and then
manipulating the milling machine X, Y and Z mill slides to
approximate the pointers at each segmental attachment
point. A detailed description of the utilisation of the verti-
cal milling machine may be found in Wickham et al. [25].
AutoCAD was utilised to graphically represent the biome-
chanical relationships (e.g. mechanical lines of action)
between each of the muscle segments and the axis of shoul-
der joint rotation. In addition, moment arms (ma) were
determined for each individual segment for both shoulder
abduction/adduction and flexion/extension.
2.8. Functional classifications of muscle segments
The functional classifications of each muscle segment
were simplified from those previously proposed by Wick-
ham and Brown [24]. In the present study, functional clas-
sifications of muscle segments were applied as followed:
Fig. 4. A composite representation (AutoCad) of the lines of action to the
humerus for each segment of the Pectoralis Major (P), the Deltoid (D) and
the Latissimus Dorsi (L). Coronal view with the upper limb abducted to
20 in the coronal plane. Representative data from one cadaver.
61
(a) Prime Mover segment: A muscle segment which was
activated first and had an agonist moment arm for
the intended motor task. Prime mover segments had
muscle fibres with mechanical lines of action similar,
or identical to, the direction of the intended joint
motion.
(b) Synergist segment: A muscle segment, with an agonist
muscle arm, which was activated significantly
(p < 0.05) later than the prime mover segment/s.
Often, the synergist segments had mechanical lines
of action divergent away from the prime mover seg-
ments and the intended movement direction.
(c) Antagonist segment: A muscle segment with an antag-
onist moment arm which was activated with, or sig-
nificantly (p < 0.05) after, the prime mover segment/
s. In all cases, antagonist segments had mechanical
lines of action and moment arms that opposed the
direction of intended joint movement.
The functional classification of a muscle segment was
determined after review of all muscles segments’ contribu-
tions, from the three muscles investigated, to the motor
task. Therefore particular functional classifications (e.g.
prime mover segment) for a given motor task may be
applied to the muscle segments of one or more muscles.
2.9. Statistical evaluations
Repeated measures analysis of variance and Student
Neumann Keuls techniques were used to evaluate the data.
Pearson’s correlation coefficients were used to determine
the relationship between the timing of activity in muscle
segments of adjacent muscles. Statistical evaluations were
designed to determine whether the timing and contraction
intensity of muscle segment activity varied significantly:
(a) within each individual muscle and (b) across all three
muscles investigated. The level of significance was set at
(p < 0.05).
3. Results
3.1. Biomechanical data
Table 1 provides a biomechanical analysis of each of the
19 muscle segments derived from one representative cada-
ver. Note that the largest muscle segment cross-sectional
areas were found in segments P1 and P2 (clavicular head)
of the Pectoralis Major, segment D3 (middle head) of the
Deltoid and segment L2 (upper fibres) of Latissimus Dorsi.
Within both the Deltoid and the Latissimus Dorsi, the
‘‘stronger’’ muscle segments were found towards the mid-
dle of the muscle. In contrast, the upper segments (clavicu-
lar head) of the Pectoralis Major had the greatest force
generation potential.
Table 2 illustrates the moment arms for each muscle seg-
ment derived for both flexion/extension and abduction/
adduction of the shoulder joint. All muscle segments of
62 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73

Table 1
Anthropometric data from one cadaver
Volume (cm3) Muscle length (cm) CSA (cm2) % of total muscle mass Adjusted CSA (cm2) Force (N)
P1 36.3 14.9 2.4 20.5 3.1 91.9
P2 50.8 17.1 3.0 25.1 3.8 112.6
P3 27.3 17.5 1.6 13.2 2.0 59.1
P4 33.9 19.4 1.7 14.7 2.2 65.8
P5 35.5 19.2 1.8 15.6 2.3 69.8
P6 26.6 20.3 1.3 11.0 1.7 49.4
L1 23.9 19.2 1.2 13.2 1.2 36.6
L2 42.9 23.1 1.9 19.7 1.8 54.6
L3 41.5 23.9 1.7 18.3 1.7 50.7
L4 44.4 25.6 1.7 18.3 1.7 50.7
L5 47.3 28.3 1.7 17.7 1.6 49.1
L6 26.7 22.1 1.2 12.7 1.2 35.2
D1 55.3 15.6 3.5 15.2 4.2 126.1
D2 47.5 14.6 3.3 14.0 3.9 115.8
D3 62.1 12.4 5.0 21.5 6.0 178.6
D4 59.3 14 4.2 18.2 5.0 151.2
D5 42.6 12.1 3.5 15.1 4.2 125.6
D6 18.9 11 1.7 7.4 2.0 61.4
D7 26.7 13.4 2.0 8.5 2.4 71.0
The volume was calculated by the mass (g) divided by the density (g/cm3). The muscle length did not include any tendinous components. The volume (cm3)
divided by the muscle length (cm) was used to calculate CSA. The percentage of total muscle mass (% of total muscle) column refers to the cross-sectional
area (CSA in cm2) column with the segment now expressed as a percentage of the total muscle mass for that muscle. The adjusted CSA (cm2) column has
been obtained by dividing a representative figure (cm2) from the literature [22], for the particular muscle, by the percentage of total muscle mass obtained
in this study. The above study used 29 deltoid, 16 Pectoralis Major and 18 Latissimus Dorsi specimens. The average of these values was used for each
muscle. The size of the cadaver used in this study approximated values obtained by other authors (see text). The force data was obtained by multiplying the
adjusted CSA (cm2) by a constant of 30 N [4].

Table 2
Segmental moment arms
Segment Moment arms at 20 abduction in a sagittal plane Moment arms at 20 abduction in a frontal plane
Flexor Extensor Abductor Adductor
P1 20.5 (1.8) 38.5 (3.3)
P2 53.3 (4.2) 56.4 (5.6)
P3 49.7 (4.1) 52.6 (5.1)
P4 32.2 (2.4) 55.6 (5.9)
P5 12.1 (1.1) 43.9 (2.1)
P6 6.1 (0.7) 41.6 (1.4)
L1 59.7 (4.8) 57.3 (4.9)
L2 54.4 (4.2) 52.1 (4.8)
L3 48.5 (3.8) 45.9 (3.5)
L4 30.6 (2.5) 35.5 (2.2)
L5 20.9 (1.9) 30.8 (1.6)
L6 15.8 (1.5) 27.8 (1.1)
D1 17.7 (1.6) 2.0 (1.5)
D2 23.7 (2.2) 17.5 (0.3)
D3 8.3 (0.8) 35.4 (1.4)
D4 35.5 (3.1) 30.1 (0.7)
D5 45.9 (3.8) 16.9 (0.3)
D6 50.4 (4.1) 8.2 (0.3)
D7 62.9 (5.9) 10.5 (1.2)
The moment arm data is given in millimetres with segment actions also extrapolated and labeled as either shoulder joint flexors/extensors (anterior or
posterior to the joint centre respectively) and/or abductors/adductors (superior or inferior to the joint centre respectively). The moment arm data has been
determined from both frontal (abductor and adductor) and sagittal (flexor and extensor) plane views. The axis of rotation of the shoulder joint was
considered as a fixed centre of rotation located in the middle of the humeral head with this position being used in similar experiments [8]. The coordinates
for each segmental attachment point gathered from the milling machine was put into AUTOCAD software were moment arms were calculated. Standard
deviations are in brackets. Grand mean data from 20 subjects.
 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73 63

the Pectoralis Major were found to have both shoulder Segment

flexion and shoulder adduction moment arms. Similarly, P6
P5
all muscle segments of the Latissimus Dorsi were shown
P4
to have both shoulder extension and shoulder adduction P3
moment arms. In contrast, the muscle segments of the Del- P2
toid were found to have moment arms for all four move- P1
ment directions; segments D1–D3 (anterior and middle
heads) had shoulder flexion moment arms while segments D1
D4–D7 (posterior head) had shoulder extension moment D2
arms. Abductor moment arms were associated with D3
D4
segments D1–D6 while a single segment, D7, was found
D5
to have a shoulder adductor moment arm. D6
D7
3.2. Shoulder abduction
L1
L2
3.2.1. Timing of muscle segment activation
L3
To produce an isometric shoulder abduction impulse, L4
myoelectric activity was found only within the muscle seg- L5
ments of the Deltoid (Figs. 5 and 6); all muscle segments of L6
Pectoralis Major and Latissimus Dorsi were myoelectri-
cally silent (<10% MVC activity) during this task. The % MVC
mean data (Table 3) indicated that the Deltoid’s middle 75
segments, D3 and D4, were activated first more than
40 ms before the initial rise of the force impulse, with adja- 50
cent muscle segments being activated successively later
(p < 0.05). The postero-medial segment D7 was activated 25
last some 34 ms after the initial rise of the force impulse.
The timing of peak myoelectric activity was identical in
-100 0 100 200 300 400 500 600 700
segments D1–D6 occurring, on average, some 74 ms before
the peak of the force impulse (Fig. 5). In contrast, the peak Time (ms)
activation of the antagonist segment D7 occurred signifi- Fig. 5. Shoulder abduction. Timing of muscle segment activation.
cantly (p < 0.05) later at 26 ms before the peak of the force P = Pectoralis Major; D = Deltoid; L = Latissimus Dorsi. Vertical dotted
impulse (Table 3). lines indicate the beginning (FcON) and the peak (FcPEAK) of the
The duration of muscle segment activation within the isometric force impulse. Note the waves of segment activation spreading
from D3 to more medial segments. m = peak of myoelectric activity in
Deltoid was significantly (p < 0.05) longer in the muscle’s that segment, which contrasted with the myoelectric silence seen in all
middle segments (D2–D5), those anatomically closer to segments of the Pectoralis Major and the Latissimus Dorsi. Group mean
the most efficient mechanical line of action for shoulder data.
abduction, than its more peripheral obliquely orientated
muscle segments (D1, D6 and D7) (Table 3). In all
segments the duration of activation was longer than As seen in Table 4, the adduction force impulse was ini-
movement time (MT = 408 ms) (Fig. 5). tiated by a broad activation of all muscle segments within
Latissimus Dorsi (L1–L6) followed significantly later
(p < 0.05) by the lower muscle segments of the Pectoralis
3.2.2. Intensity of muscle segment activation Major (P4–P6). The mean data indicated that segments
As seen in Fig. 6 and Table 3, the intensity (%MVC) of L1–L6 and P4–P6 were all activated before the initial rise
Deltoid segment activation was significantly (p < 0.05) of the force impulse (FcON) (Table 4). There followed a
higher in the Deltoid’s middle segments (D2–D5) when later (p < 0.05) activation of the upper fibres of the Pecto-
compared to the muscles more peripheral muscle segments ralis Major (P1–P3), along with the D6 segment of the Del-
(D1, D6 and D7). toid, after the initial rise of the force impulse. In contrast,
segment D7 of the Deltoid had a significantly (p < 0.05)
3.3. Shoulder adduction earlier activation than segment D6 just 7 ms before FcON
(Table 4).
3.3.1. Timing of muscle segment activation Peak myoelectric activity occurred, in all muscle seg-
The shoulder adduction task elicited segmental activity ments, before the peak of the force impulse (FcPEAK).
within all three muscles investigated although this activity However, it was found that segments L2, L3, L4 and L6
was confined to only the two postero-medial segments of the Latissimus Dorsi reached their highest myoelectric
(D6 and D7) of the Deltoid (Figs. 7 and 8). activity significantly (p < 0.05) earlier (approximately
64 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73

Fig. 6. Shoulder abduction. Intensity of muscle segment contraction in Latissimus Dorsi (L1–L6), Deltoid (D1–D7) and Pectoralis Major (P1–P7).
Diagram based on group mean data. Note that muscle segment activity is limited to the Deltoid muscle only and more intense in the middle (adductor)
segments of the muscle.

Table 3 3.3.3. The sequence of muscle segment activation

Shoulder abduction; timing (ms) and intensity (%MVC) Correlation of muscle segment activation times during
On (ms) Peak (ms) Duration Intensity the adduction motor task (Fig. 9) indicated that each mus-
(ms) (%MVC) cle segment, within the three muscles investigated, was acti-
x s.d. x s.d. x s.d. x s.d. vated in a sequence that ignored anatomical boundaries
between adjacent muscles. As seen in Fig. 9, the CNS pro-
D1 04* 26 68 43 554* 96 38* 24
D2 35* 19 85 58 617 87 83 33 duced three waves of muscle segment activation (sequential
D3 53 19 82 37 636 76 87 17 activation of muscle segments) during the adduction motor
D4 42 23 81 31 657 96 88 20 task. Two muscle segment activation waves, centred on
D5 24* 24 59 31 624 81 73 23 segment L3, progressively activated the muscle segments
D6 05* 21 66 41 560* 84 58* 30
of Latissimus Dorsi as well as the postero-medial segments
D7 34* 26 26* 66 528* 70 51* 30
of the Deltoid. The other wave of muscle segment activa-
Mean and standard deviations. Significant (p < 0.05) difference: *to seg-
tion began within segments P5 and P6 (lower fibres) and
ment D3. Movement time (FcOn to FcPeak) of 408 ms.
then progressively activated higher muscle segments of
the Pectoralis Major. All three muscle segment activation
waves were associated with high correlation coefficients
117 ms) than all other muscle segments (approximately (R2) of between 0.88 and 0.94 (Fig. 9).
65 ms) investigated (Table 4).
The duration of muscle segment activation was found to 3.4. Shoulder flexion
be longest in segments L1–L6, D7 and P3–P6 (approxi-
mately 584 ms). Segments P1, P2 and D6 were all found 3.4.1. Timing of muscle segment activation
to have significantly (p < 0.05) shorter periods of activation The shoulder flexion task was accomplished by wide-
(approximately 495 ms) during this task (Table 4). All spread activation of the Pectoralis Major (P1–P6), the ante-
active muscle segments had periods of activity longer than rior four segments of the Deltoid (D1–D4) and only one
adduction movement time of 390 ms (Table 4). segment (L1) of the antagonist Latissimus Dorsi
(Fig. 10). The task was initiated by simultaneous activation
3.3.2. Intensity of muscle segment activation of four adjacent muscle segments on the anterior aspect of
The intensities of muscle segment activation (%MVC) the shoulder; P1 and P2 (clavicular head) with D1 and D2
were similar throughout the muscle segments of Latissimus (anterior head) (Fig. 10). A significantly (p < 0.05) later
Dorsi and the lower segments of the Pectoralis Major activation occurred in segments P3–P6, D3, D4 and L1
(Table 4). All muscle segments of Latissimus Dorsi (L1– (Table 5). The remaining segments (D5–D7; L2–L6) of
L6), and the lower segments of Pectoralis Major (P4–P6) the Deltoid and the Latissimus Dorsi respectively remained
had significantly (p < 0.05) higher intensities of activation inactive during this task (Fig. 11).
(approximately 64% MVC) than the other activated muscle All active muscle segments across the three muscle inves-
segments (approximately 34% MVC) (Table 4 and Fig. 8). tigated, except segment P4, reached their maximum inten-
 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73
Fig. 7. Shoulder adduction. Timing of muscle segment activation.
P = Pectoralis Major; D = Deltoid; L = Latissimus Dorsi. Vertical dotted
lines indicate the beginning (FcON) and the peak (FcPEAK) of the
isometric force impulse. m = peak of myoelectric activity in that segment.
Note the sequential onset of segment activation, the similar timing of
segmental peak activity and the myoelectric silence within the more
anterior segments of the Deltoid. Group mean data.
sity (peak) simultaneously some 73 ms before the peak of
the flexion force impulse (FcPK). Segment P4 had a signif-
icantly (p < 0.05) later peak activation approximately
46 ms before FcPK (Table 5).
The duration of muscle segment activation was shown
to be significantly (p < 0.05) shorter in segments D4
(513 ms) and L1 (349 ms) than in the other activated mus-
cle segments (585 ms). All segments, except the antagonist
L1, had durations of activity longer than movement time
(MT = 408 ms) (Table 5).
3.4.2. Intensity of muscle segment activation
As seen in Table 5, two of the four initially activated
muscle segments (D1 and D2) had significantly (p < 0.05)
higher intensities of contraction (approximately 55%
65
MVC) than muscle segments (approximately 29% MVC)
with later activation times.
3.5. Shoulder extension
3.5.1. Timing of muscle segment activation
The results for shoulder extension were almost the mir-
ror image of those for shoulder flexion (Figs. 12 and 13).
Shoulder extension was achieved through activation of
the Latissimus Dorsi (L1–L6), the posterior five segments
of the Deltoid (D3–D7), and the P3 segment of the Pecto-
ralis Major (Fig. 12). The extension force impulse was ini-
tiated by synchronous activation of the upper muscle
segments of Latissimus Dorsi (L1–L4) in conjunction with
the posterior muscle segments of the Deltoid (D5 and D7)
and segment P3 of the antagonist Pectoralis Major muscle.
Other muscle segments were either activated significantly
(p < 0.05) later (D3, D4, D6, L5, L6,) or not activated at
all (P1, P2, P4–P6; D1 and D2) (Table 6).
The timing of peak muscle activation was reached in all
muscle segments before the peak of the force impulse
(FcPEAK) (Table 6). Note that segments L3, L6 and P3
had the earliest (p < 0.05) myoelectric peak, approximately
100 ms prior to FcPK compared to between 60 ms and
70 ms prior to FcPK in the other active segments (L1,
L2, L4, L5, D3–D7) (Fig. 13).
All active muscle segments within the three investigated
muscles, except segment D3, had similar durations of acti-
vation (approximately 630 ms). Segment D3 had a signifi-
cantly (p < 0.05) shorter duration of activation
(approximately 574 ms) although all active muscle seg-
ments were activated longer than movement time
(MT = 430 ms) (Table 6).
3.5.2. Intensity of muscle segment activation
As seen in Table 6, segments L2 and L3 along with seg-
ments D5–D7 had significantly (p < 0.05) higher intensities
of muscle activation (approximately 73% MVC) when
compared with other (L1, L4–L6, D3, D4 and P3) active
muscle segments (approximately 42% MVC).
4. Discussion
4.1. Overview
The aim of the present study was to determine how 19
muscle segments within three adjacent superficial shoulder
muscles, the Pectoralis Major, the Deltoid and the Latissi-
mus Dorsi, were controlled by the CNS during the produc-
tion of four rapid isometric shoulder tasks (abduction,
adduction, flexion and extension).
Of particular interest was to understand how the timing
and intensity of muscle segment activation within and
across the three superficial shoulder muscles was influenced
by the muscle segments mechanical line of action (joint
movement most likely from approximation of a segment’s
origin and insertion) and its moment arm. We wished to
66 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73

Fig. 8. Shoulder adduction. Intensity of muscle segment contraction in Latissimus Dorsi (L1–L6), Deltoid (D1–D7) and Pectoralis Major (P1–P7).
Diagram based on group mean data. Note that muscle segment activity included the postero-medial segments of the Deltoid – a muscle normally described
as an abductor of the shoulder joint.

Table 4
Shoulder adduction; timing (ms) and intensity (%MVC)
On (ms) Peak (ms) Duration (ms) Intensity (%MVC)
x s.d. x s.d. x s.d. x s.d.
P6 24 m 37 61 m 40 587 119 62 15
P5 25 m 26 72 m 51 588 153 66 20
P4 15 m 23 58 m 51 579 112 56 19
P3 11 m* 36 53 m 31 540 107 41 m 18
P2 21 m* 48 56 m 34 511 m* 100 34 m* 18
P1 26 m* 34 50 m 47 498 m* 83 33 m* 26
D1
D2
D3
D4
D5
D6 22 m# 46 62 m 46 477 m 133 28 m 18
D7 7m 28 70 m 35 536 89 47 m 36
L1 43 23 87 m 53 612 109 55 34
L2 53 32 100 52 601 99 56 22
L3 66 30 132 56 600 112 79 39
L4 58 25 128 75 594 123 67 24
L5 48 28 86 m 46 594 130 64 25
L6 46 25 109 35 596 134 70 31
Mean and standard deviations. Within muscle analysis: significant differences (p < 0.05): * to segment P5; # to segment D7; 
to segment L3. Between
muscle analysis: significant difference (p < 0.05) : m to segment L3. Movement time (FcOn to FcPeak) of 390 ms.

determine whether functional classification of muscle seg-
ments as ‘‘prime movers’’, ‘‘synergists’’ and ‘‘antagonists’’,
as described previously within the Deltoid muscle [24], was
more widely applicable to muscle segments within a group
of active muscles.
In general, the results of this investigation indicated that
the timing and intensity of each muscle segment’s activa-
tion was often determined by the segment’s function
through its mechanical line of action, in relation to the
intended movement direction, and its moment arm. The
results support the hypothesis that the function of each
muscle segment, to initiate the motor task (prime mover),
to support the motor task’s goal (synergist) or to act in
opposition to the motor task (antagonist), determined the
muscle segments onset, duration of activity and intensity
of activation. Furthermore, our results clearly showed that
 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73 67

Segment
L6 FcON
L5
R2 = .94
L4

L3

L2

L1 R2 = .94
D7

D6

D5

D4

D3

P1

P2

P3

P4
R2 = .88
P5

P6

-75 -50 -25 0 25 50

Time (ms)
Fig. 9. Onset (ON) of muscle segment activation within the Latissimus
Dorsi (L), Deltoid (D) and the Pectoralis Major (P) during shoulder
adduction. Note the waves of muscle segment activation sweeping out
from segment L3 both superiorly to the Deltoid (L3–D6) and inferiorly
(L3–L6). Within the Pectoralis Major, the excitation wave passes from
inferior (P6) to superior (P1) segments. FcOn = initial rise of the isometric
force impulse. Group mean and SEM data. High correlation coefficients
(R2) characterised the waves of segment activation.
the activity of muscle segments within adjacent, and more
distant muscles, are closely coordinated and, in agreement
with Kuechle et al. [12], that the functional classifications
of muscle segments were fluid and may change from one
motor task to the next. Finally, our results support the con-
tention [2,19,24] that individual skeletal muscles are com-
posed of discrete sub-volumes (segments) of muscle tissue
which may be independently controlled by the CNS to pro-
duce the desired motor outcome.
4.2. Prime mover, synergist and antagonist muscle segments
Initially we have shown that within the three superficial
shoulder muscles investigated, there were muscle segments
which had a range of mechanical lines of action and
moment arms related to the intended movement direction.
Those muscle segments which were activated first and
which tended to have higher contraction intensities and
longer periods of activation were termed the prime mover
Fig. 10. Shoulder flexion. Timing of muscle segment activation. P = Pec-
toralis Major; D = Deltoid; L = Latissimus Dorsi. Vertical dotted lines
indicate the beginning (FcON) and the peak (FcPEAK) of the isometric
force impulse. m = peak of myoelectric activity in that segment. Note the
sequential activation of segments within the Pectoralis major and the
(anterior and middle) Deltoid, and the myoelectric silence throughout
most of the Latissimus Dorsi and the posterior segments of the Deltoid.
Group mean data.
segments [24]. These muscle segments were found to have
large agonist moment arms and favourable mechanical
lines of action for the intended motor task. The early acti-
vation of the prime mover segments to initiate production
of the isometric force impulses is consistent with the find-
ings of Flanders and Soechting [6] who have reported that
the first ‘‘peak’’ of muscle activation was always close to
the mechanical line of action of the muscle.
Muscle segments with more divergent mechanical lines
of actions or smaller agonist moment arms than the prime
mover segments, the synergist segments, were generally
activated next with similar, or smaller, durations and inten-
sities of activation. Most agonist muscles involved in each
motor task had both prime mover and synergist muscle
segments the exception of the Latissimus Dorsi during
shoulder adduction whose six muscle segments (L1–L6)
68 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73

Table 5 In general, antagonist muscle segments tended to have

Shoulder flexion; timing (ms) and intensity (%MVC) delayed onsets, shorter periods of activity and lower con-
On (ms) Peak (ms) Duration Intensity traction intensities than the prime mover segments (e.g.
(ms) (%MVC) D7 in shoulder abduction; D6 during shoulder adduction
x s.d. x s.d. x s.d. x s.d. and L1 during shoulder flexion) (Tables 3–5). An exception
P6 3m* 44 53 29 596 88 27 .* 16 was segment P3 (Pectoralis Major) during shoulder exten-
P5 2 m* 36 59 49 602 86 31 .* 19 sion which was co-activated with the prime mover seg-
P4 4 m* 32 46 c 41 576 82 27 .* 16 ments of the Latissimus Dorsi (Table 6). This was an
P3 11 m* 37 80 52 567 99 24 .* 13
P2 38 32 88 49 576 136 33 .* 17
unexpected result given that numerous reports [11,15] of
P1 47 26 75 34 625 123 42 . 22 ‘‘triphasic’’ muscle activation patterns during rapid isomet-
ric motor tasks have suggested that the antagonist (Ant)
D1 27 23 70 29 586 126 55 24
D2 29 23 75 33 594 125 55 24 muscle burst is always activated after the initial agonist
D3 11 m# 28 82 38 548 130 44 .# 16 (Ag1) period of activation.
D4 28 m# 23 65 25 513 m 96 19 .# 15 However, the involvement of antagonist muscle seg-
D5 ments in these rapid isometric motor tasks is consistent
D6
with the findings of Flanders and Soechting [6] who state
D7
that antagonist muscle activity, in part, is important to
L1 35 m 20 85 20 349 m 43 11 . 24 counteract internal joint forces which may cause tissue
L2
damage.
L3
L4 In general these results, obtained from the superficial
L5 shoulder muscle group, broadly support the previous find-
L6 ings of Wickham and Brown [24] for the Deltoid muscle.
Mean and standard deviations. Within muscle analysis: significant differ- These findings, with notable exceptions to these generalisa-
ences (p < 0.05): * to segment P1; # to segment D2. Between muscle tions, will now be discussed in more detail with reference to
analysis: significant differences (p < 0.05) : m to segment P1; c to segment each movement direction.
P2; . to segment D1. Movement time (FcOn to FcPeak) of 408 ms.

all appeared to have a prime mover function (Fig. 6 and
Table 4). This notable exception will be discussed further
below.
We have also noted in the results that muscle segments
with antagonist moment arms, whether located within an
antagonist muscle (e.g. D1–D6 Deltoid during shoulder
adduction) or as a single segment within an agonist muscle
(e.g. D7 during shoulder abduction), had characteristic
periods of activation (Figs. 4, 6, 9 and 11).
4.3. Influence of movement direction
The most simple motor control strategies were employed
by the CNS during the shoulder abduction task (Table 3).
Here the middle muscle segments of the Deltoid (D3 and
D4) were found to have the most favourable mechanical
lines of action and the greatest moment arms to produce
shoulder abduction in the coronal plane. Shoulder abduc-
tion was initiated by activation of the prime mover seg-

Fig. 11. Shoulder Flexion. Intensity of muscle segment contraction in Latissimus Dorsi (L1–L6), Deltoid (D1–D7) and Pectoralis Major (P1–P7).
Diagram based on group mean data. Note that the highest intensities of muscle segment contraction were seen in adjacent segments of the Deltoid and the
Pectoralis Major. Also note the unique antagonist role played by muscle segment L1.
 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73
Fig. 12. Shoulder extension. Timing of muscle segment activation.
P = Pectoralis Major; D = Deltoid; L = Latissimus Dorsi. Vertical dotted
lines indicate the beginning (FcON) and the peak (FcPEAK) of the
isometric force impulse. m = Peak of myoelectric activity in that segment.
Note that the pattern of muscle segment activity is almost the reverse of
that seen in shoulder flexion (Fig. 9). Group mean data.
ments D3 and D4. These segments, along with the adjacent
segments D2 and D5, had the longest periods of activation
and the highest intensities of contraction within the Deltoid
muscle. The more peripheral and divergent segments of this
muscle, D1, D6 and D7, all had later onsets, shorter peri-
ods of activation and lower contraction intensities reflect-
ing their less favourable mechanical lines of action and
smaller abduction moment arms (Table 2).
Interestingly, segment D7 on the postero-medial
aspect of the Deltoid was found uniquely to have an
antagonist moment arm for shoulder abduction. Accord-
ingly, its period of activation was very late (e.g. 34 ms
after FcON), short and of low intensity. Although D7
was anatomically contingent with segment D6, it is
clearly seen that the CNS controlled the motor units of
D7 independently of those within D6. The difference in
69
moment arms between D7 (antagonist ma = 10.5 mm)
and D6 (agonist ma = +8.2 mm) for the shoulder abduc-
tion task appeared to explain their contrasting periods of
activity.
A more complicated and difficult to explain muscle
segment control strategy was seen during shoulder adduc-
tion (Table 4). This motor task involved all three super-
ficial muscles including segments of the Deltoid muscle
(Fig. 6) – a muscle which Hughes and Kai-Nan [9] had
previously predicted should have had no activity for this
motor task.
Shoulder adduction was initiated by activation of the
entire Latissimus Dorsi (L1–L6), some 52 ms (group
mean) before the initial rise of the force–time curve
(FcON) (Fig. 6). Therefore, all segments of the Latissi-
mus Dorsi could be considered prime mover segments
although their mechanical lines of action and adduction
moment arms were variable (Table 1). In contrast, the
segments of the Pectoralis Major were all activated as
synergists with the lower fibres (P4–P6) and then the
upper fibres (P1–P3) being activated in sequence (Table
4). The utilisation of all segments of the Latissimus
Dorsi as the prime movers for shoulder adduction, in
preference to those of the Pectoralis Major, was an unex-
pected result which was difficult to explain. Both muscles
have large adductor moment arms (Table 2) and muscle
segments with effective mechanical lines of action for the
motor task. Both muscles also insert into the bicipital
groove [7] and have similar cross-sectional areas (Table
1). Possibly the more posterior orientation of the Latiss-
imus Dorsi muscle segments afforded biomechanical
advantages to the CNS in the initiation of this motor
task. However, further work is required before this result
can be fully explained.
During shoulder adduction, all segments of the antag-
onist Deltoid muscle were myoelectrically silent except
for segments D6 (antagonist ma = 8 mm) and D7 (ago-
nist ma = +10.5 mm) (Fig. 6 and Table 4). The onset of
segment D7 activation suggested a synergist function,
with the lower fibres of the Pectoralis Major (P4–P6),
to assist production of the adduction force impulse. In
contrast, segment D6 clearly functioned as an antagonist
to the motor task. D6 was coactivated with the fibres of
the Pectoralis Major although only at a very low (28%
MVC) intensity of activation. Similar findings were made
by Flanders and Soechting [6] who noted, in the poster-
ior (and medial) Deltoid during isometric motor tasks,
that motor units may be directionally tuned in opposite
movement directions within the same muscle segment.
Muscle segment function during shoulder flexion and
shoulder extension appeared as a mirror image. During
shoulder flexion (Table 5), the clavicular head of Pectoralis
Major (P1 and P2), along with the anterior head of the Del-
toid (D1 and D2), acted as the prime mover segments. All
four segments had good mechanical lines of action and
large moment arms for the shoulder flexion task (agonist
ma = +18 mm to +53 mm). The lower fibres of Pectoralis
70 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73

Fig. 13. Shoulder extension. Intensity of muscle segment contraction in Latissimus Dorsi (L1–L6), Deltoid (D1–D7) and Pectoralis Major (P1–P7).
Diagram based on group mean data. Note the unique role of the antagonist muscle segment P3 during the extension task. Segment activity, in shoulder
extension, is almost a mirror image of that seen in shoulder flexion (Fig. 10).

Major (P3–P6) and the middle fibres of the Deltoid (D3– P4) were not necessarily activated as prime mover segments
D4) then acted as synergist segments. These segments all if their mechanical lines of action were too divergent from
had more divergent mechanical lines of action although movement direction.
their flexion moment arms were still impressive (e.g., P3 With an antagonist moment arm of 60 mm, segment
agonist ma = 50 mm; P4 agonist ma = 32 mm) (Table 2). L1 was the only segment of the Latissimus Dorsi to be
It is interesting to note in this motor task that muscle seg- active during the shoulder flexion task. Its late activation,
ments with very large agonist moment arms (e.g. P3 and shorter duration of activity and low intensity of activation
clearly defined this segment’s antagonist role (Table 5).
Table 6
The horizontal orientation of its muscle fibres may have
Shoulder extension; timing (ms) and intensity (%MVC) also produced an optimal ability to oppose the forward
On (ms) Peak (ms) Duration Intensity
translatory motion of the humerus during the flexion
(ms) (%MVC) motor task.
x s.d. x s.d. x s.d. x s.d.
In contrast shoulder extension (Table 6) was initiated by
the upper fibres of Latissimus Dorsi (L1–L4) and two seg-
P6
P5
ments of the Deltoid (D5 and D7) (Fig. 11). Again these
P4 prime mover segments had good mechanical lines of action
P3 34 19 83 57 629 51 11 m 29 for shoulder extension and large extension moment arms
P2 (ma range = +63 mm (D7) to +31 mm (L4)) (Table 2).
P1 The synergist segments (L5, L6, D6 and D4) were then acti-
D1 vated to complete the motor task. Again, these synergist
D2 segments had more divergent mechanical lines of action
D3 43 m# 27 45 m 57 574 c 116 33 m# 10
and generally smaller extension moment arms (ma ran-
D4 5m 29 53 m 49 622 126 50 m# 17
D5 25 27 60 m 36 641 99 63# 14 ge = +50 mm to +21 mm).
D6 14 m 30 79 m 48 595 87 81 25 Two antagonist muscle segments were activated during
D7 37 30 62 m 39 637 100 84 30 shoulder extension. Segment D3 of the Deltoid, with an
L1 45 28 62 m 46 670 135 58 m 22 antagonist moment arm of 8.3 mm, was activated after
L2 39 27 80 m 44 633 127 67 19 the prime mover segments. Its late activation, shorter dura-
L3 49 31 126 71 664 155 71 31 tion of activity and lower intensity of activation (33%
L4 28 25 78 m 61 624 74 59 m 13 MVC), clearly reflected its antagonist role. In contrast seg-
L5 7 m 45 61 m 31 608 96 42 m 20
L6 3 m 39 92 56 612 73 43 m 15
ment P3 of the Pectoralis Major, with an antagonist
moment arm of 50 mm, had an onset similar to the prime
Mean and standard deviations. Within muscle analysis: significant differ-
ences (p < 0.05):  to segment L3; # to segment D7. Between muscle
mover segments. As seen in Table 6, segment P3 was co-
analysis: significant differences (p < 0.05): m to segment L3; c to segment activated with the prime mover segments L1–L4, D5 and
L1. Movement time (FcOn to FcPeak) of 430 ms. D7. Its peak intensity and duration of activity (Table 6)
 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73
were also similar to the prime mover segments although its
intensity of activation was significantly (p < 0.05) lower at
11% of MVC. It is difficult to explain why this particular
antagonist muscle segment, the only segment of the Pecto-
ralis Major to be activated during shoulder extension,
should have such an early activation although its horizon-
tal muscle fibre orientation may be implicated. Activation
of the P3 segment may have been required early to protect
the anterior shoulder from injury during the vigorous acti-
vation (approximately 73% of MVC group mean) of the
prime mover segments; a role commonly given to antago-
nist muscle action [6]. If this hypothesis is true, it is evident
that the CNS has considerable flexibility regarding the tim-
ing of antagonist muscle activity during the production of
motor tasks.
With regard to the peak of muscle segment activation, it
is clear that for all muscle segments involved in the four
motor tasks investigated, maximal motor unit activation
occurred before the peak of the isometric force–time curve.
In general, prime mover segments had earlier peak intensi-
ties when compared to synergist segments although this
was not always the case. There was, however, some limited
evidence to suggest that the peak activation of some antag-
onist muscle segments was significantly (p < 0.05) later
than the agonist muscle segments (Table 3).
4.4. Motor unit task groups
It was evident that to produce each of the four move-
ment tasks, motor unit task groups were formed both
within, and where appropriate, across the three shoulder
muscles, and then activated (timing and intensity) sequen-
tially to produce the movement task. Motor unit task
groups may be defined as a group of motor units, from
within a muscle or across several muscles, which are active
simultaneously for a particular motor task [14]. It is
thought that differences in moment arms, or physiological
factors such as fibre type, may explain the existence of
motor unit task groups, and therefore functionally inde-
pendent muscle segments within muscle [5].
In the present study, each motor task was accomplished
through the formation of at least two, and up to four, dif-
ferent motor unit task groups which often comprised
motor units from adjacent, and more distant, muscles.
For example, during shoulder abduction two motor unit
task groups were formed within the Deltoid to achieve
the motor task. The first motor unit task group formed
included the prime mover segments (D3, D4), a common
feature in all movement tasks in this study. A second motor
unit task group was then formed by the remaining synergist
and antagonist segments (D1, D2, D5, D6 and D7) which
all had smaller abduction moment arms, or in the case of
D7, an adduction moment arm. Contrary to the hypothesis
of Flanders and Soechting [6], there was no evidence to
suggest that the motor units of the middle (D3) and poster-
ior (D4–D7) Deltoid were tightly coupled during this
motor task.
71
Shoulder adduction was accomplished with the forma-
tion of three motor unit task groups. The first contained
all motor units within the Latissimus Dorsi muscle which
acted together as the prime movers for the motor task. A
second group was formed by motor units within the lower
fibres of Pectoralis Major along with the agonist segment
D7 of the Deltoid. A third group was formed by motor
units within the upper fibres of the Pectoralis Major along
with the antagonist D6 segment of the Deltoid. Fig. 8
clearly shows the three motor unit task groups and con-
firms that the activation of segments (motor units) within
each task group was sequential away from muscle segments
with higher agonist moment arms.
Two motor unit task groups were formed during shoul-
der flexion while three were formed during shoulder exten-
sion. In both movement tasks, the first group contained the
prime mover segments while the second comprised syner-
gist segments. All motor unit task groups contained motor
units from a number of muscles. The antagonist segments
either formed their own motor unit task group (shoulder
extension) or were distributed either to the first (shoulder
extension) or the second (shoulder flexion) motor unit task
group.
4.5. Muscle segment function and the CNS
The production of muscle force is a product of
moment arm, muscle size, the muscle’s length tension rela-
tionship as well as the timing and intensity of motor unit
activation; the latter controlled by the CNS [12]. Our
results support the suggestion that the CNS, through its
upper and lower motoneurons, has the flexibility to ‘‘fine
tune’’ and coordinate the timing and intensity of motor
unit activity both within a single muscle and between
adjacent and more distant muscles, to meet the intended
goals of the motor task.
It has been previously demonstrated, in monkey cortex,
that pyramidal tract neurons from layer 5 of M1 (the pri-
mary motor cortex) directly innervate lower motor neurons
in layer IX of the spinal cord [20]. These are the motoneu-
rons which have direct control over the activation of a
muscle’s constituent fibres.
Given that there is a proven relationship between the
activity of upper motoneurons within a cortical column,
and the subsequent EMG activity within an active muscle
[17], the results of this study may be used to gain further
insight into the strategies used by the CNS to control vol-
untary movement.
Specifically, our study supports the contention that spe-
cific motor columns within the human M1 (primary motor
cortex), have the ability to independently control discrete
subpopulations of motor units within a single skeletal mus-
cle. This has previously only been confirmed in animal
studies [13,17]. Therefore, our results give further insight
into the complexity of the task faced by the CNS to control
each and every muscle involved in even the most simple of
motor tasks.
72 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73
4.6. Limitations
Several limitations in the experimental design may have
influenced the results.
Firstly, it should not be concluded that the 19 muscle
segments identified in this study represented the ‘‘ultimate’’
design imposed by the CNS. In some instances (e.g. Latiss-
imus Dorsi), the anatomical borders of each muscle seg-
ment were found by simple division of relatively
homogenous muscle tissue. Even when the muscle segments
were clearly defined anatomically (e.g. Deltoid), it may be
possible that each segment could be further subdivided into
ever smaller functional units. It is not beyond possibility
that the ‘‘ultimate muscle segment’’ would be represented
by a single muscle spindle and its associated muscle fibers!
However, it remains probable that similar findings, to
those found here, would have been found if fewer, or more,
muscle segments had been investigate. The exciting possi-
bility now exists for the application of a multi-electrode
fine-wire EMG technique, to an in vitro analysis of animal
muscle, whereby more muscle segments may be investi-
gated and greater clarity provided to their neuromuscular
control.
A second limitation was the problem of crosstalk
between adjacent bipolar electrode pairs. If detectable lev-
els of crosstalk were present, the ability to differentiate
between the activity of adjacent muscle segments would
be compromised. Considerable effort was made to mini-
mize the effects of electrode crosstalk. The bipolar elec-
trodes were specifically designed to have small active
plates (1.6 mm in diameter) and inter-electrode distances
(6.5 mm). This electrode design, using the crosstalk ‘‘rule
of thumb’’ reported by Basmajian and DeLuca [1], would
suggest meaningful pickup of muscle fiber activity only
within approximately 6.5 mm of each bipolar electrode
pair. Adjacent bipolar electrode pairs were generally
spaced more than 2 cm from each other so as to minimize,
as much as practicable, electrode crosstalk. In addition,
regardless of possible electrode crosstalk, the results pre-
sented here provide strong evidence of the sequential acti-
vation of adjacent muscle segments to control these
isometric tasks (Fig. 8). It would have been expected that
high levels of electrode crosstalk would have negated the
possibility of this result.
5. Conclusion
The aim of this study was to determine how 19 muscle
segments within the Latissimus Dorsi, Deltoid and Pecto-
ralis Major, were controlled by the CNS to produce four
isometric shoulder motor tasks.
The results of this investigation have suggested that the
timing and intensity of each muscle segment’s activation
were coordinated across muscles and influenced by the
muscle segment’s moment arm and its mechanical line of
action in relation to the intended shoulder movement direc-
tion (e.g. flexion, extension, abduction or adduction).
There was also evidence that motor unit task groups were
formed for individual motor tasks which comprised motor
units from both adjacent and distant muscles. In addition,
the results supported the hypothesis that individual muscles
segments may be functionally classified as prime mover,
synergist or antagonist; functional classifications which
were generally determined by the muscles segments
moment arm and/or mechanical line of action.
Acknowledgements
The authors wish to thank the University of Wollon-
gong for technical and financial support to Dr. Wickham
during his doctoral studies and to Dr. Brown during his
sabbatical leave in Germany. Thank you also to Mr. Luke
Bones and Mr. Daniel Wickham for assistance with prepa-
ration of the illustrations and to the reviewers for the
excellent comments and suggestions. The work described
in this study was approved by the Human Ethics Commit-
tee of the University of Wollongong.
References
[1] Basmajian JV, DeLuca C. Muscles alive; their functions revealed by
electromyography. 5th ed. Baltimore: Williams & Wilkins; 1985.
[2] Brown JMM, Solomon C, Paton ME. Further evidence of functional
differentiation within biceps brachii. Electromyograph Clin Neuro-
physiol 1993;33:301–9.
[3] Campbell KM, Norman BA, Biggs L, Blanton PL, Lehr RP.
Electromyographic investigation of the relative activity among
four components of the triceps surae. Am J Phys Med
1973;52(1):30–41.
[4] Enoka RM. The neuromechanical basis of kinesiology, 2nd ed. In:
Human kinetics, 1994.
[5] Ettema GJC, Styles G, Kippers V. The moment arms of 23 muscle
segments of the upper limb with varying elbow and forearm positions:
Implications for motor control. Human Movement Sci
1998;17(2):201–20.
[6] Flanders M, Soechting JF. Arm muscle activation for static forces in
three-dimensional space. J Neurophysiol 1990;64(6):1818–37.
[7] Gray’s Anatomy. The anatomical basis of medicine and surgery.
Sydney: Churchill Livingstone; 1990.
[8] Hogfors C, Sigholm G, Herberts P. Biomechanical model of the
human shoulder – I. Elements. J Biomech 1987;20(2):157–66.
[9] Hughes RE, Kai-Nan A. Force analysis of rotator cuff muscles. Clin
Orthopaed Related Res 1996;1(330):75–83.
[10] Jenkins DB. In: Hollinshead’s functional anatomy of the limbs and
back. Sydney: W.B. Saunders; 1998.
[11] Karst GM, Hassan Z. Antagonist muscle activity during forearm
movements under varying kinematic and loading conditions. Exp
Brain Res 1987;67:391–401.
[12] Kuechle DK, Newman SR, Itoi E, Niebur GL, Morrey BF, An K-N.
The relevance of the moment arm of shoulder muscles with respect to
axial rotation of the glenohumeral joint in four positions. Clin
Biomech 2000;15(5):322–9.
[13] Lemon R. The output map of the primate motor cortex. Trends
Neurosci 1988;11:501–6, 1989, erratum appears in Trends Neurosci
Apr;12(4):158..
[14] Loeb GE. Motorneuron task groups: Coping with kinematic heter-
ogeneity. J Exp Biol 1985;115:137–46.
[15] MacKinnon CD, Rothwell JC. Time-varying changes in corticospinal
excitability accompanying the triphasic EMG pattern. J. Physiol.
1998;506P:114P.
 J.M.M. Brown et al. / Journal of Electromyography and Kinesiology 17 (2007) 57–73
[16] Manueddu Y, Blanc W, Taillard W. Study of the functioning of the
Gluteus Medius and Maximus: An electromyographic analysis. Ann
Kinesither 1989;16:193–201.
[17] Muir RB, Lemon RN. Corticospinal neurons with a special role in
precision grip. Brain Res 1983;61:312–6.
[18] Pare EB, Stern JR, Schwartz JM. Functional differentiation within
the tensor fascia latae. J Bone Joint Surg A 1981;63(9):239–46.
[19] Paton ME, Brown JMM. An electromyographic analysis of func-
tional differentiation in human pectoralis major muscle. J Electro-
myograph Kinesiol 1994;4(3):161–9.
[20] Shinoda Y, Yokota J, Futami T. Divergent projection of individual
corticospinal axons to motoneurons of multiple muscles in the
monkey. Neurosci Lett 1981;23:7–12.
[21] Soderberg GL, Dostal WF. Electromyographic study of three parts of
the gluteus medius muscle during functional activities. Phys Therapy
1978;58(6):691–6.
[22] Veeger HEV, VanDerHelm FCT, VanDerWoude LHV, Pronk GM,
Rozendal RH. Inertia and muscle contraction parameters for
musculoskeletal modelling of the shoulder mechanism. J Biomech
1991;24(7):615–29.
[23] Wickham JB. Muscles within Muscles: The neuromotor activation
patterns of intramuscular segments. PhD thesis, Department of
Biomedical Science, University of Wollongong, 2002.
[24] Wickham JB, Brown JMM. Muscles within muscles: the neuromotor
control of intra-muscular segments. Eur J Appl Physiol 1998;78:219–25.
[25] Wickham JB, Brown JMM, McAndrew DJ. Muscles within muscles:
Anatomical and functional segmentation of selected shoulder joint
musculature. J Musculoskeletal Res 2004;8(1):57–73.
Mark Brown has a Doctoral qualification from
the University of Queensland and is currently a
Senior Lecturer and Assistant Dean at the
University of Wollongong. He was Head of the
Department of Biomedical Science between
1997 and 2001 and is the immediate past Vice
President (exercise science) of Australia’s pro-
fessional Exercise Science association (AAESS).
His research interests include functional seg-
mentation of skeletal muscle and sustainable
transport.
73
James Wickham received a Bachelor of Sports
Science (exercise science) from the University
of New South Wales in 1992 and a Diploma
of Education (physical education) in 1993
from the same institution. In 1995 he received
an Honours Degree from the University of
Wollongong and a PhD in 2002. Since 1999
James has been employed as an Anatomy
Lecturer at La Trobe University in Mel-
bourne. His research interests include using
electromyography to quantify shoulder muscle
activation patterns and functional differentia-
tion within skeletal muscles.
Darryl McAndrew received a Bachelor of Sci-
ence (Human Movement Science) from the
University of Wollongong in 1993 and is cur-
rently an Associate Lecturer within the
Department of Biomedical Science, UoW. He is
concurrently completing a PhD focusing on the
fibre type characteristics of segmental muscle
and a M.Sc. in OH & S, specialising in Occu-
pational Hygeine. His research interests include
identifying muscle fibre type characteristics via
mechanomyography and occupational exposure
to industrial noise.
Xu-Feng Huang received his PhD degree from
the University of New South Wales in 1992. He
is currently an Associate Professor in the
Department of Biomedical Science and the
Director of Neurobiology Research Centre for
Metabolic and Psychiatric Disorders in the
University of Wollongong. His research interest
includes the central regulation of energy
balance.