SHORT-TERM UNILATERAL RESISTANCE TRAINING AFFECTS THE

AGONIST–ANTAGONIST BUT NOT THE FORCE–AGONIST ACTIVATION

RELATIONSHIP
NEALE A. TILLIN, MSc, MATTHEW T.G. PAIN, PhD, and JONATHAN P. FOLLAND, PhD
School of Sport, Exercise and Health Sciences, Loughborough University, Leicestershire, Ashby Road, Loughborough, Leicestershire
LE11 3TU, UK
Accepted 10 August 2010

ABSTRACT: In this study we investigated the contribution of such as electrode relocation and between-session
neural adaptations to strength changes after 4 weeks of unilat- variability in fascia, subcutaneous fat, and skin im-
eral isometric resistance training. Maximal and submaximal iso-
metric knee extension contractions were assessed before and pedance may explain the inconsistent results of
after training. Surface electromyography (EMG) data were col- past research.18 Normalizing EMG amplitude to a
lected from the agonist and antagonist muscles and normalized supramaximal compound muscle action potential
to evoked maximal M-wave and maximal knee flexor EMG,
respectively. The interpolated twitch technique (ITT) was also (M-wave) may remove some of these confounding
used to determine activation at maximum voluntary force factors,19 but few resistance training stud-
(MVF). MVF increased in the trained (þ20%) and untrained ies14,15,17,20,21 have employed this technique. The
(þ8%) legs. Agonist EMG at MVF increased in the trained leg
(þ26%), although activation determined via the ITT was force–agonist EMG relationship through the range
unchanged. In both legs the position of the force–agonist EMG of contraction intensities may also provide insight
relationship was unchanged, but antagonist coactivation was into the link between changes in activation and
lower for all levels of agonist activation. Strength gains in the
trained leg were due to enhanced agonist activation, whereas force production. Previous studies have reported a
decreased coactivation may have affected strength changes in rightward shift in the force–agonist EMG relation-
both legs. ship (i.e., greater force for the same level of activa-
Muscle Nerve 43: 375–384, 2011
tion) after 2–6 months of resistance train-
Marked increases in muscle function, specifically ing,7,13,16,22 suggesting that peripheral adaptations
strength and rate of force development (RFD), were the primary determinants of long-term
have been found during the early phase (2–4 strength gains. In contrast, it may be expected that
weeks) of a training program.1–4 Despite recent the force–agonist EMG relationship will remain
evidence of hypertrophy after just 20 days of resist- constant, although extended, over a short-term (<2
ance training,5 early strength and RFD gains are months) resistance training intervention, when the
typically attributed to neural adaptations. Specific primary adaptations may be neural. However, evi-
mechanisms put forward include increased activa- dence for this effect is inconsistent23 and limited.
tion of the agonist muscles and decreased activa- The ITT has also produced highly equivocal evi-
tion of the antagonists, but evidence for these dence of enhanced agonist activation during maxi-
mechanisms is equivocal. Typically, agonist neural mal voluntary efforts after resistance training. Some
activation has been assessed using either surface studies have reported an increase,2,14,24–26 and
electromyography (EMG) or the interpolated others have reported no change.17,27,28 The afore-
twitch technique (ITT; see Shield and Zhou6). An- mentioned studies, however, defined activation as
tagonist activation has also been measured with the ratio of superimposed twitch force to twitch
EMG, but its relationship to agonist activation force evoked at rest. This method assumes that the
through the range of contraction intensities has superimposed twitch is elicited at maximum volun-
not been considered in the context of training. tary force (MVF), and that the relationship between
Studies that have used EMG to assess the effect superimposed twitch force and voluntary force is
of resistance training on agonist activation during linear,29 which are both unlikely scenarios.30,31 The
maximal contractions have reported an difference between MVF and theoretical maximum
increase,2,3,7–10 whereas others have not.11–17 Issues force (TMF; force at maximal activation) may pro-
vide a more valid measure of activation, where TMF
has been extrapolated from an appropriate model
Abbreviations: ANOVA, analysis of variance; BF, biceps femoris; EMG,
electromyography; ITT, interpolated twitch technique; Mmax, maximal M- of the superimposed twitch–voluntary force
wave; MVC, maximal voluntary contraction; MVF, maximal voluntary force; relationship.29
M-wave, compound muscle action potential; RF, rectus femoris; RFD, rate
of force development; RMS, root mean square; ST, superimposed twitch Agonist and antagonist activation contribute
force; TMF, theoretical maximal force; VF, voluntary force; VL, vastus later- simultaneously to net force production, and thus
alis; VM, vastus medialis
Key words: coactivation, interpolated twitch technique, isometric strength assessing their coactivation (via a statistical analysis
training, knee extensors, rate of force development of their relationship) will provide greater insight
Correspondence to: N. A. Tillin; e-mail: [email protected]
into the neuromuscular adaptations to resistance
V
C 2011 Wiley Periodicals, Inc.

Published online 15 February 2011 in Wiley Online Library

training. Decreased coactivation would theoreti-
(wileyonlinelibrary.com). DOI 10.1002/mus.21885 cally increase net force production but may
Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011 375
compromise joint integrity. These competing program. Measurements were taken from both legs
demands likely influence the nature of the adapta- in the 10 days prior to the start of training (pre-
tion to any specific training stimulus and may training) and 2–6 days after the last training ses-
partly explain why previous studies have reported sion (posttraining). Measurement trials were com-
antagonist activation to increase,32,33 decrease,8,11 pleted at a consistent time of day and involved a
or remain unchanged8,14 after resistance training. standard 60–90-min protocol. Specifically, knee
Although these studies compared antagonist and extension/flexion force and surface EMG of the
agonist activation qualitatively, they did not quan- superficial knee extensors and biceps femoris were
tify coactivation. Consequently, their results pro- recorded in response to electrically stimulated
vide limited insight into the contribution of antag- twitch, explosive voluntary, and maximal voluntary
onist activation to force output before and after isometric contractions. All contractions, including
training. The complete agonist–antagonist EMG measurements and angles, were completed in an
relationship should provide a comprehensive isometric strength testing chair,39,40 with a con-
assessment of coactivation and its association with stant knee and hip angle of 85 and 100 , respec-
force changes posttraining. tively (180 representing full extension).
Alongside strength, RFD has been identified as
Training. Participants completed four training ses-
an important descriptor of functional perform-
sions per week for 4 weeks. Each session lasted 10–
ance.34–36 RFD over periods >100 ms appears
15 min and consisted of four sets of 10 unilateral
closely related to strength,37 but RFD over shorter
isometric knee extensions using the same appara-
periods (
50 ms) appears to be more related to
tus as the measurement trials. In each contraction,
agonist activation.34,38 A short-term resistance
participants were instructed to ramp up to 75% of
training program designed to increase strength
maximal voluntary force (MVF) (determined in
through neural adaptations might produce differ-
the measurement trials and then re-determined at
ential effects on RFD according to the time period
the start of the first session of each week, as
of measurement, but this possibility has not been
detailed in what follows) over 1 s, hold it for 3 s,
explored. Investigating this issue may offer further
and relax. Two seconds separated each contrac-
insight into the mechanisms that determine
tion, and 2 min separated each set. Training was
strength and RFD and could have implications for
completed on one leg (trained leg) chosen at ran-
resistance training.
dom, and participants were instructed to avoid
The purpose of this study was to investigate the
contracting the contralateral leg (untrained leg)
contribution of agonist and antagonist activation
during each effort. Typically, participants struggled
to changes in strength and RFD after 4 weeks of
to achieve 75% of MVF by the fourth set, in which
unilateral isometric resistance training. To provide
case they were instructed to extend the knee as
a more comprehensive assessment, maximal neural
hard as possible. Real-time biofeedback of the
activation of the agonists was assessed with EMG
force response, with a target marker at 75% MVF,
normalized to Mmax, and using the ITT with appro-
was provided on a computer monitor, and verbal
priate extrapolation. Furthermore, the force–ago-
encouragement was given throughout.
nist EMG and agonist EMG–antagonist EMG rela-
tionships were examined. Measurements. Participants were firmly secured in
the strength testing chair with a waist belt and
METHODS
shoulder straps. Force was measured with a cali-
Participants. Nine male participants (age 21 6 1
brated U-shaped aluminum strain gauge (linear
years, height 1.82 6 0.05 m, mass 81 6 7 kg) com-
response up to 1000 N41), which was in series with
pleted the study. Participants were recreationally
an ankle strap placed 2 cm proximal to the medial
active (moderate exercise
3 times per week), but
malleolus, and positioned perpendicular to tibial
were not involved in any form of strength or power
movement during knee extension/flexion. The
training. All the participants were healthy, injury-
force signal was amplified (500), interfaced with
free, and provided written informed consent prior
an analog-to-digital converter (CED Micro 1401;
to their involvement in this study, which was
CED, Cambridge, UK), and sampled at 2000 Hz
approved by the Loughborough University ethics
with a personal computer using Spike2 software
advisory committee.
(CED, Cambridge, UK). The force signal was
notch filtered at 100 and 200 Hz in both directions
Overview. Participants completed three trials with an infinite impulse response digital filter (q-
before (one familiarization and two measurement factor of 100) to remove harmonics of the main
trials, each trial was 2–3 days apart) and two mea- frequency.
surement trials (2–3 days apart) after a 4-week uni- Surface EMG was recorded mains from the rec-
lateral isometric knee extensor strength training tus femoris (RF), vastus lateralis (VL), vastus
376 Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011
medialis (VM), and long head of the biceps femo- dow), and peak RFD normalized to twitch peak
ris (BF), using a Bagnoli-4 EMG system (Delsys, force (RFD/peak force) were also averaged across
Boston, Massachusetts). After preparation of the the three supramaximal twitch contractions.
skin (shaving, lightly abrading, and cleansing with Explosive Voluntary Contractions. Participants
70% ethanol), a single differential surface elec- then completed a warm-up of submaximal voluntary
trode configuration (1-cm intersensor distance; contractions, followed by 10 explosive voluntary
Model DE-2.1; Delsys, Boston, Massachusetts) was contractions (each separated by 20 s). In each con-
attached over each muscle using adhesive interfa- traction, participants attempted to extend the knee
ces. To normalize electrode placement across par- as ‘‘fast’’ and as hard as possible for 1 s from a
ticipants and trials, the same experienced investiga- relaxed state. Explosive voluntary contractions were
tor placed the electrodes over the belly of each completed separately from maximal voluntary con-
muscle, parallel to the presumed orientation of tractions (MVCs; detailed in what follows) because
the muscle fibers, at 50%, 55%, 90%, and 45% previous work has highlighted the importance of
of the distance between the greater trochanter and instruction with regard to performance outcome,42
lateral femoral condyle for the RF, VL, VM, and whereby explosive performance (RFD) was greatest
BF, respectively. The reference electrode was when participants received the ‘‘fast-and-hard’’
placed on the patella of the same limb. EMG sig- instruction. Furthermore, due to the technicalities
nals were amplified 100 (differential amplifier of the explosive contractions, 10 efforts were com-
20–450 Hz) and sampled at 2000 Hz with the same pleted to ensure that a valid measure of the partici-
analog-to-digital converter and personal computer pant’s capacity for voluntary RFD was recorded.
as the force signal prior to being bandpass filtered During each explosive contraction, participants
in both directions between 6 and 500 Hz using a were instructed to avoid any countermovement
second-order Butterworth digital filter. (knee flexion prior to knee extension). A com-
The femoral nerve was electrically stimulated via puter monitor displayed both force (on a sensitive
a constant-current, variable-voltage stimulator scale around resting values) and the slope of the
(DS7AH, Digitimer, Ltd., UK) with square-wave force–time curve. The latter was used to provide
pulses (0.1-ms duration) to elicit twitch contrac- immediate biofeedback of performance, specifi-
tions and compound muscle action potentials (M- cally peak slope of each contraction, and the for-
waves). The anode, a 7  10 cm carbon rubber mer highlighted any countermovement. The three
electrode (Electro-Medical Supplies, Greenham, contractions with the largest peak slope and no
UK), was taped to the skin over the greater tro- discernible countermovement or pre-tension
chanter. The cathode, a custom-adapted stimula- (change of baseline force of <0.5 N during the
tion probe 1 cm in diameter (Electro-Medical Sup- 100 ms prior to contraction onset) were used for
plies, Wantage, UK), which protruded 2 cm analysis, and all RFD and EMG variables were aver-
perpendicular from the center of a plastic base (4 aged across these three explosive contractions.
 5 cm), was taped to the skin over the femoral Analysis consisted of measurement of force–time
nerve in the femoral triangle. The precise location and EMG–time curves in three 50-ms time windows
of the cathode was determined as the position that (0–50, 50–100, and 100–150 ms) after their respec-
evoked the greatest twitch response for a particular tive onsets. Impulse (the area underneath the
submaximal electrical current (typically 30–50 mA). force–time curve) and RFD (slope of the force–
time curve) were measured for each time window,
Protocol. The following protocol was first com- and the latter was reported in absolute terms and
pleted on one leg and then repeated on the other. normalized to MVF (detailed in what follows), spe-
The order in which legs were tested was random- cifically RFD/MVF. The root mean square (RMS)
ized between participants but remained consistent of the EMG signal during each time window was
for each individual. Once in the chair a series of calculated for each muscle. Agonist (RF, VL, and
twitch contractions were elicited at incremental VM) EMG RMS values were normalized to Mmax
current intensities until a simultaneous plateau in (i.e., EMG RMS/Mmax) and averaged across the
force and M-wave response of each muscle was three muscles to give a mean quadriceps value. An-
achieved (typically between 100 and 160 mA). tagonist (BF) EMG RMS values were normalized to
Thereafter, the electrical current was increased by maximal BF RMS EMG determined during the
20%, and three supramaximal twitches were eli- knee flexor MVCs (detailed in what follows).
cited at 12-s intervals. The average peak–peak M- Maximal Voluntary Contractions. After the ex-
wave response of these three supramaximal plosive contractions, participants completed four
impulses was defined as the Mmax for each muscle. knee extensor isometric MVCs (separated by 30
Peak force, peak RFD (peak slope of the force– s), in which they were instructed to push as hard as
time curve determined by a 1-ms moving time win- possible for 3 s. Biofeedback and verbal
Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011 377
encouragement were provided during and between
each MVC. Knee extensor MVF was the greatest vol-
untary force achieved by a participant in any of the
knee extensor MVCs or explosive contractions dur-
ing that laboratory visit. Peak voluntary force in
each MVC was typically between 95% and 100% of
MVF. After the MVCs participants completed one
submaximal contraction at 20%, 40%, 60%, and
80% of MVF (separated by 20 s), during which they
were requested to achieve the required force (rep-
resented with a target line on the computer moni-
tor) and hold it steady for up to 5 s.
Neural activation was assessed by measuring the
RMS EMG of each muscle during a 200-ms epoch,
at MVF (100 ms either side of MVF) and during a FIGURE 1. The relationship between voluntary force (VF;
stable segment of each submaximal contraction expressed as a percentage of maximum VF) and superimposed
that was not influenced by the interference pattern twitch force (ST; expressed as a percentage of control twitch
of a superimposed twitch (detailed in what follows). peak force). This relationship is represented with a model first
proposed by Scaglioni et al.26 that contains both a linear and
Agonist and antagonist RMS EMG values were nor-
an exponential component.
malized as described for the explosive contractions,
and the former were averaged to give a mean quad-
riceps value. To illustrate the effects of training on Furthermore, the model was required to meet the
the force–agonist EMG relationship and agonist– same criteria when fitted separately to the data
antagonist EMG relationship, agonist (mean quad- from each of the four conditions (pretrained, post-
riceps)-normalized EMG at the five contraction trained, pre-untrained, post-untrained). We found
intensities (20%, 40%, 60%, 80%, and 100% the best fit (of various linear and curvilinear mod-
MVF) was plotted against absolute force, and antag- els) was generated by a model previously proposed
onist-normalized EMG, respectively, for each partic- by Scaglioni and colleagues,26,44 which contains
ipant pre- and posttraining. An appropriate func- both linear and exponential components:
tion for these relationships was then generated.   
ST
Maximum voluntary neural activation was also VF ¼ a ð1  bÞe cST þ b 1  (1)
d
determined via the ITT. Supramaximal twitch con-
tractions (two superimposed twitches, separated by The coefficient a corresponds to the value of
1 s) were evoked during a stable segment of the VF when ST ¼ 0% and represents TMF as a per-
second and fourth MVCs and each submaximal centage of MVF; b weights the linear and exponen-
contraction. Single twitch contractions were cho- tial components; c is the constant of the exponen-
sen to assess activation rather than a train of tial portion; and d corresponds to the value of ST
impulses, as previous work has reported no differ- when VF ¼ 0%.
ence in activation calculated with single or multi- When applied to the data for all participants
ple impulses.29,43,44 Futhermore, during extensive and conditions, this model generated a function
pilot work we found neural inhibition45 to be for the ST–VF relationship shown in eq. (2) that
greater when participants were anticipating a had an adjusted R2 value of 0.97:
larger superimposed stimulus. For each voluntary   
0:036ST ST
contraction the average force increment of the two VF ¼ 101:5 ð1  0:44Þe þ 0:44 1 
100
superimposed twitches (ST) was expressed as a
(2)
percentage of average peak force of two control
twitches evoked at rest 1–2 s immediately after that The model shown in eq. (1) was fitted to the
contraction (average coefficient of variation of individual data of each leg pre- and post-training.
peak force for the two control twitches was 1.8%). The TMF (x-axis intercept) generated by each
The relationship between voluntary force (VF, as a function was converted to absolute force, and the
percentage of MVF) and normalized ST was plot- percentage of TMF achieved at MVF was defined
ted (Fig. 1) using the data from all participants, as the level of maximum neural activation.
measurement trials (two before and two after), To complete the trial, participants performed
and conditions (trained and untrained legs). To three isometric MVCs of the knee flexors (sepa-
determine a model that could best describe this rated by 30 s), at a knee joint angle of 105 . This
relationship, criteria were set whereby the ST at VF knee joint angle was required, as participants were
¼ 0% was 100% and VF at ST ¼ 0% was 100%. not comfortable performing the knee flexor MVCs
378 Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011
Table 1. Coefficient of variation (CV) of key dependent variables
agonist activation and TMF; explosive voluntary
between the two pre-training measurement trials. RFD (absolute and normalized); impulse; mean
quadriceps EMG activation at 0–50, 50–100, and
Dependent variable CV (%)
100–150 ms from onset; and twitch peak force and
Maximal voluntary force (MVF) 3.0 twitch peak RFD (absolute and normalized).
Agonist-normalized EMG at MVF 10.9
Submaximal agonist normalized EMG 12.7 RESULTS
Antagonist-normalized EMG at MVF 19.7 Reliability. There was no difference in any of the
Submaximal antagonist-normalized EMG 29.7
Explosive force at 50 ms 12.8
dependent variables between the two pretraining
Explosive force at 100 ms 5.3 measurement trials (paired t-tests, P > 0.05).
Explosive force at 150 ms 4.5 Therefore, changes between pre- and posttraining
Twitch peak force 6.0 measurement trials were considered to be an effect
Twitch peak RFD 10.6 of training. Coefficients of variation of key depend-
CV for submaximal agonist- and antagonist-normalized EMG are means ent variables between the two pretraining measure-
of the CVs recorded at 20%, 40%, 60%, and 80% of maximal voluntary ment trials are presented in Table 1.
force.
Maximal Voluntary Contractions. MVF increased in
at an 85 knee joint angle. RMS EMG during a both the trained (þ20%, paired t-test, P ¼ 0.001)
200-ms epoch at the greatest knee flexor force and untrained (þ8%, paired t-test, P ¼ 0.007) legs.
(100 ms either side) was measured to give maximal This increase was greater in the trained legs, result-
BF RMS EMG. ing in a time-by-leg interaction effect (ANOVA, P
¼ 0.002; Fig. 2).
Data Analysis and Statistics. For the explosive vol- There was a 26% increase of agonist-normal-
untary and involuntary contractions identification ized EMG at MVF in the trained leg (pre, 0.094 6
of force and EMG onsets was made manually. 0.021; vs. post, 0.119 6 0.022 RMS/Mmax; paired t-
Details of this method have been published previ- test, P ¼ 0.046), and this increase was similar for
ously.38 Briefly, the same investigator identified sig- submaximal contractions at 20%, 40%, 60%, and
nal onsets with a constant y-axis scale of 1 N and 80% of MVF (þ26–31%; all P < 0.02; Fig. 3A).
10 mV, for force and EMG, respectively, and an x- Whereas agonist-normalized EMG in the untrained
axis scale of 500 ms. Manual identification is con- leg was unchanged at MVF (pre, 0.102 6 0.011 vs.
sidered the ‘‘gold standard’’ method for detecting post, 0.107 6 0.025 RMS/Mmax; paired t-test, P ¼
signal onsets.46–49 0.53), and at 40–80% of MVF, it did increase at
To assess the reliability of each dependent vari- 20% of MVF (þ15.7%; P ¼ 0.023; Fig. 3B) post-
able, the values recorded in the first and second training. The positions of the force–agonist EMG
pretraining measurement trial were compared with relationship, specifically the quadratic coefficients
paired t-tests. The coefficient of variation of each of the relationship, were unchanged posttraining
dependent variable (averaged across the trained for both the trained (paired t-tests, 0.44 < P <
and untrained legs) between the two pretraining 0.80; Fig. 3A) and untrained (paired t-tests, 0.65 <
measurement trials was also recorded. P < 0.98; Fig. 3B) legs.
Before group values (mean 6 SD) for the
trained and untrained legs were generated, de-
pendent variables for each participant were aver-
aged across the two pretraining measurement trials
and across the two posttraining measurement tri-
als. For each dependent variable the influence of
time (pretraining vs. posttraining) and leg (trained
vs. untrained) was analyzed with a two-way
repeated-measures analysis of variance (ANOVA).
Paired t-tests were then used to determine within-
leg differences between pre- and posttraining
measures, and stepwise Bonferroni correction was
employed for all time-series data. Statistical analysis
was completed using SPSS (version 16), and the
significance level was set at P < 0.05. Dependent
FIGURE 2. Maximal voluntary force (MVF) during isometric
variables included: MVF; normalized agonist and
knee extensions of the trained (filled squares) and untrained leg
antagonist EMG activation at MVF; the gradient (open circles) before and after resistance training. **Time  leg
and coefficients of the agonist EMG–force relation- interaction effect (P < 0.01). Values are presented as mean 6
ship and agonist–antagonist EMG relationship; ITT SD (n ¼ 9).

Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011 379
 paired t-test, P ¼ 0.42). There was a downward
shift in the position of the agonist-antagonist EMG
relationship, with a decrease in the y-intercept
(Fig. 4) in both the trained (51.7%; paired t-test,
P ¼ 0.014; Table 2) and untrained (48.6%;
paired t-test, P ¼ 0.097; Table 1) legs, but no
change in the gradient of the relationship of ei-
ther leg (paired t-tests, 0.34 < P < 0.59). This indi-
cates reduced coactivation for a given level of ago-
nist activation.

Explosive Voluntary Contractions. There was no

change in the absolute force achieved at 50, 100,
or 150 ms during explosive isometric knee exten-
sions, in either the trained (Fig. 5A) or untrained
(Fig. 5C) legs. Accordingly, there was no change in

FIGURE 3. The relationship between force and agonist (three

superficial quadriceps) normalized EMG during isometric knee
extensions at 20%, 40%, 60%, 80%, and 100% MVF, performed
before (squares, solid line) and after (circles, dotted line)
resistance training, by the trained (A) and untrained (B)
legs. Relationships are represented as quadratic functions
(pretrained, y ¼ 1  107x2 þ 8  105x þ 0.0007;
posttrained, y ¼ 1  107x2 þ 8  105x þ 0.0009; pre-
untrained, y ¼ 1  107x2 þ 0.0001x  0.0026; post-untrained,
y ¼ 9  108x2 þ 0.0001x  0.0051). Values are expressed as
mean 6 SD (n ¼ 9). Agonist-normalized EMG of the trained
leg increased at all force levels (P < 0.05).

In contrast, there was no change in maximum

neural activation determined via ITT in either the
trained (pre, 98.6 6 0.8 vs. post, 98.7 6 0.9%;
paired t-test, P ¼ 0.64) or untrained (pre, 98.5 6
0.8 vs. post, 98.5 6 0.8%; paired t-test, P ¼ 0.97)
legs. Consequently, changes in TMF were propor-
tional to changes in MVF in both the trained
(þ20%; pre, 599.2 6 84.8 vs. post, 718.2 6 118.2
N; paired t-test, P ¼ 0.001) and untrained (þ8%;
pre, 592.3 6 52.0 vs. post, 637.8 6 63.9 N; paired FIGURE 4. The relationship between agonist and antagonist
t-test, P ¼ 0.006) legs. normalized EMG during isometric knee extensions at 20%,
There was a tendency for the absolute level of 40%, 60%, 80%, and 100% MVF, before (squares, solid line)
antagonist-normalized EMG at MVF to increase by and after (circles, dotted line) resistance training, in the trained
(A) and untrained (B) leg. Relationships are represented as lin-
8% in the trained leg (pre, 0.115 6 0.047 vs. post,
ear functions. Values are mean 6 SD (n ¼ 9). There was a
0.124 6 0.044 RMSmax; paired t-test, P ¼ 0.053), downward shift in the position of the agonist- and antagonist-
but no change occurred in the untrained leg (pre, normalized EMG relationship in the trained (P < 0.05) and
0.125 6 0.076 vs. post, 0.107 6 0.048 RMSmax; untrained (P < 0.1) legs.

380 Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011
 Table 2. Linear regression parameters for the relationship between agonist- and antagonist-normalized EMG during isometric knee
extensions at different levels of force before and after strength training in the trained and untrained leg.
Leg Linear parameter Before After P-value
Trained Gradient 0.97 6 0.41 1.04 6 0.43 0.59
y-intercept (RMS/RMSmax) 0.02 6 0.01 0.01 6 0.01 0.01
Untrained Gradient 0.95 6 0.66 0.77 6 0.33 0.34
y-intercept (RMS/RMSmax) 0.03 6 0.02 0.02 6 0.01 0.10

Values are expressed as mean 6 SD (n ¼ 9).

absolute RFD or impulse during any of the three Twitch Contractions. Pairwise comparisons re-
50-ms time windows. When force was normalized to vealed no significant change in twitch peak force
MVF, there was a decrease in the percentage of or absolute peak RFD in either the trained or
MVF achieved at 100 and 150 ms by both the untrained leg after the intervention (Table 3).
trained (100 ms, 15%; 150 ms, 12%; paired t- However, normalized peak RFD (to twitch peak
tests, P < 0.01; Fig. 5B) and untrained (100 ms, force) decreased in the trained leg by 7%, but
8%; 150 ms, 6%; paired t-tests, P < 0.01; Fig. not the untrained leg (Table 3). There was no
5D) legs after the intervention. This appears to change in M-wave amplitude of the RF, VL, or
have been due to a decrease in normalized RFD VM in the trained or untrained leg after training
during the 50–100-ms time window in the trained (Table 3).
leg (pre, 7.3 6 1.3 vs. post, 6.0 6 1.3 MVF s1;
paired t-test, P ¼ 0.028), and a tendency for the DISCUSSION
same effect in the untrained leg (pre, 7.1 6 0.69 In this study we have investigated the effects of a 4-
vs. post, 6.1 6 1.0 MVF s1; paired t-test, P ¼ week unilateral isometric resistance training inter-
0.074). There was also no training effect on agonist vention on the strength, RFD, agonist, and antago-
or antagonist normalized EMG during the initial nist activation of trained and untrained limbs.
150 ms of the explosive contractions of either leg. Strength gains in the trained leg (20%) were

FIGURE 5. Absolute (A, C) and normalized (B, D) force (normalized to maximal voluntary force; MVF) during explosive isometric knee
extensions, before (filled squares) and after (open circles) strength training, in the trained (A, B) and untrained leg (C, D). Data are
mean 6 SD of all participants (n ¼ 9). **Difference between pre- and posttraining values (P < 0.01).

Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011 381
 Table 3. Maximal evoked twitch contractions of the knee extensors before and after strength training in the trained and untrained leg.
Trained Untrained

Pre Post P Pre Post P

Peak force (N) 141 6 39 144 6 33 0.62 142 6 37 135 6 30 0.33
Peak RFD (N s1) 3977 6 970 3753 6 649 0.30 3951 6 978 3809 6 736 0.52
Normalized peak RFD (peak force s1) 28.8 6 5.1 26.7 6 4.7 <0.01 28.3 6 4.3 28.7 6 3.9 0.68
Rectus femoris Mmax (mV) 1.4 6 0.7 1.4 6 0.4 0.79 1.6 6 1.0 1.5 6 0.8 0.69
Vastus lateralis Mmax (mV) 2.6 6 0.8 2.4 6 0.6 0.67 3.4 6 1.5 2.5 6 1.3 0.06
Vastus medialis Mmax (mV) 8.2 6 3.1 7.1 6 3.0 0.25 7.2 6 3.3 9.3 6 3.7 0.09

Values are expressed as mean 6 SD of all participants (n ¼ 9).

similar to increases in agonist normalized EMG at controversial.30,31 There are suggestions that the
MVF (26%). The position of the force–agonist ITT is insensitive at high voluntary forces50 and
EMG relationship and the magnitude of the rest- that the ST–VF relationship may become increas-
ing twitch remained unchanged, indicating negligi- ingly asymptotic and potentially confound extrapo-
ble muscular adaptation. Despite an increase in an- lation up to TMF.51 Considering the similar
tagonist-normalized EMG at MVF in the trained increases in force and agonist-normalized EMG
leg, which appears to be a consequence of greater and a consistent force–agonist EMG relationship, it
joint loading, coactivation was lower for any given seems likely that the ITT was insensitive to changes
level of agonist activation (i.e., a downward shift of in agonist activation after strength training.
the agonist–antagonist activation relationship Evidence for changes in maximum agonist EMG
occurred). In contrast to this accumulated evi- after resistance training is controversial.3,7–9,11–13,16
dence for neural changes, agonist activation as However, none of these earlier studies involved nor-
determined by the ITT was unaffected by training. malization to Mmax, so their results may have been
Strength gains of 8% in the untrained leg may be confounded by EMG reliability issues such as relocat-
due to a downward shift of the agonist–antagonist ing electrodes and variable impedance of fascia, sub-
EMG relationship, and there was no evidence of cutaneous fat, and skin.18 Two more recent studies of
increased agonist activation (normalized EMG or short duration (nine training sessions), which nor-
ITT). maliz agonist EMG to Mmax, reported no change in
Strength changes in the trained leg, þ20% in neural drive after training.14,17 However, after 16 train-
16 training sessions, were comparable to other ing sessions in this study and 18 training sessions in a
studies that used a similar training interven- previous report,17 agonist-normalized EMG did
tion.2,11,12,14 It appears that strength gains in our increase. Therefore, although early strength gains to
study were largely due to enhanced agonist activa- resistance training appear to be primarily due to
tion, as shown by the increase in maximum agonist enhanced agonist activation, a sufficient training vol-
EMG normalized to Mmax and the consistent posi- ume may be required to detect this adaptation. It is
tion of the force–agonist EMG relationship (i.e., important to note that, although there may be limita-
no change in force for the same neural drive). On tions to the information that surface EMG can pro-
the contrary, maximum activation determined by vide,52 the results of this study support the use and
the ITT was unaffected by training. Previous stud- sensitivity of EMG normalized to Mmax as a global
ies have reported no change17,27,28 or typically measure of neural activation.
modest increases (
5%) in activation with ITT af- To fully understand the coordinated changes
ter resistance training.2,14,24–26 These investiga- in agonist and antagonist activation after training
tions, however, used the ratio of superimposed and their combined influence on force produc-
twitch force to control twitch force, irrespective of tion, we assessed the complete agonist–antagonist
MVF, so their results do not necessarily reflect the EMG relationship. There was a downward shift in
maximal level of voluntary activation, which is only the entire agonist–antagonist EMG relationship,
achieved at MVF. Care was taken in this study to representing reduced coactivation posttraining for
define an appropriate ST–VF relationship29 for any given level of agonist activation. Despite this
each individual31 in order to extrapolate up to shift in the agonist–antagonist EMG relationship
TMF. Furthermore, conditions of the study with training, antagonist-normalized EMG at MVF
included a low-compliance strength chair, prior increased in the trained leg. Previous research has
potentiation of the superimposed twitch,29 and reported inconsistent effects of resistance training
nerve stimulation.44 Nevertheless, the validity of on antagonist activation,11,14,32,33,53 but these stud-
quantifying neural activation with the ITT remains ies did not quantify coactivation. It is thus unclear
382 Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011
how their observed changes in antagonist activa- the trained leg suggests that there were possible
tion influenced force production. The increase in adaptations to the muscle’s contractile properties.
antagonist activation in this study appears to be a The time-course of the twitch response, and there-
consequence of increased at MVF agonist activa- fore normalized twitch RFD, is expected to be
tion and force and is likely to be a protective influenced by cross-bridge cycling rate,37 which is
mechanism required to maintain joint integrity.54 greatest in muscle fibers that express type IIX myo-
Nevertheless, the shift in the agonist–antagonist sin heavy chain isoforms.59 The training-induced
EMG relationship indicates a less than proportion- transition of MHC expression from type IIX to IIA
ate increase in antagonist activation at MVF after (fast-to-slow) is well documented60,61 and has been
training that, if replicated throughout all of the observed after just 4 weeks of resistance train-
knee flexor muscles, may have contributed to the ing.62,63 If this effect occurred in our study it could
observed gains in strength. The high coefficient of have contributed to the drop in evoked and volun-
variation of antagonist-normalized EMG (20–30%) tary normalized RFD, although this possibility can-
was not unexpected.55 This variation is an artifact not be confirmed without collection of tissue sam-
of the low recorded antagonist-normalized EMG ples. It is important to note that this training
values (0.12 of RMSmax at MVF), and reflects intervention may also have induced hypertrophy,
only a 2–3% change in the level of antagonist acti- as this adaptation has been observed previously
vation. Nevertheless, pre- and posttraining meas- over a similar time period.5 However, although the
urements were averaged across two trials to cross-sectional area of the quadriceps muscles was
improve their reliability. not recorded in this study, the consistent force–
Changes in force production were similar, but agonist EMG relationship and twitch amplitude of
of a smaller magnitude, in the untrained leg com- the trained leg suggests that peripheral adaptations
pared with the trained leg. Previous studies have were minimal.
also indicated enhanced strength in the untrained In conclusion, for the trained leg, strength
limb after unilateral resistance training.11,25,56–58 gains appear to be due to greater agonist activa-
Although there was no change in agonist- or antag- tion during the maximum voluntary effort with no
onist-normalized EMG at MVF of the knee exten- change in the position of the force–agonist activa-
sors, there was a tendency for the agonist–antago- tion relationship. The increase in force and ago-
nist EMG relationship of the untrained leg to shift nist activation appears to have caused an increase
downward (i.e., lower coactivation). It seems likely in antagonist activation at MVF; however, there
that neural adaptations account for the strength was an overall decrease in coactivation for any
gains in the untrained limb, but they are too given level of agonist activation and force post-
subtle to provide significant changes and may only training. The ITT was insensitive to changes in
become evident when assessing the entire agonist– agonist activation. Adaptations were specific to the
antagonist EMG relationship. high-force, non-explosive contractions performed
Despite increased strength there was no change in training.
in RFD or impulse during the explosive voluntary
The authors thank Josh Bakker-Dyos, Christopher Davison, and
contractions in either the trained or untrained Matt Cross for their help and hard work during data collection. We
legs. Consequently, normalized RFD decreased also thank the participants for their efforts in the trials and com-
during the 50–100-ms time window in both legs af- mitment to the training. This work was funded by Loughborough
ter resistance training. Although strength and RFD University.
are generally thought to be related,37 the strength
gains in this study appear to be due to neural REFERENCES
1. Barry BK, Warman GE, Carson RG. Age-related differences in rapid
adaptations specific to maximum force production muscle activation after rate of force development training of the
that did not transfer to RFD. Previous studies fo- elbow flexors. Exp Brain Res 2005;162:122–132.
2. Del Balso C, Cafarelli E. Adaptations in the activation of human skel-
cusing on training explosive force, rather than etal muscle induced by short-term isometric resistance training. J
maximum force with no explosive component (as Appl Physiol 2007;103:402–411.
3. Narici MV, Roi GS, Landoni L, Minetti AE, Cerretelli P. Changes in
in the present study), showed increased RFD and force, cross-sectional area and neural activation during strength
agonist activation during explosive voluntary con- training and detraining of the human quadriceps. Eur J Appl Physiol
Occup Physiol 1989;59:310–319.
tractions.1,2,8,10,35 The results of our study there- 4. Abe T, DeHoyos DV, Pollock ML, Garzarella L. Time course for
fore highlight the importance of training specific- strength and muscle thickness changes following upper and lower
body resistance training in men and women. Eur J Appl Physiol
ity for enhancing the neural mechanisms that 2000;81:174–180.
underlie the performance of explosive and maxi- 5. Seynnes OR, de Boer M, Narici MV. Early skeletal muscle hypertro-
phy and architectural changes in response to high-intensity resist-
mum strength tasks. ance training. J Appl Physiol 2007;102:368–373.
Although neural adaptations appear to explain 6. Shield A, Zhou S. Assessing voluntary muscle activation with the
twitch interpolation technique. Sports Med 2004;34:253–267.
the majority of changes observed in this study, the 7. Hakkinen K, Komi PV. Electromyographic changes during strength
surprising decrease in normalized twitch RFD in training and detraining. Med Sci Sports Exerc 1983;15:455–460.

Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011 383
 8. Hakkinen K, Kallinen M, Izquierdo M, Jokelainen K, Lassila H, Mal- 37. Andersen LL, Aagaard P. Influence of maximal muscle strength and
kia E, et al. Changes in agonist–antagonist EMG, muscle CSA, and intrinsic muscle contractile properties on contractile rate of force de-
force during strength training in middle-aged and older people. velopment. Eur J Appl Physiol 2006;96:46–52.
J Appl Physiol 1998;84:1341–1349. 38. Tillin NA, Jimenez-Reyes P, Pain MTG, Folland JP. Neuromuscular
9. Kubo K, Ohgo K, Takeishi R, Yoshinaga K, Tsunoda N, Kanehisa H, performance of explosive power athletes versus untrained individu-
et al. Effects of isometric training at different knee angles on the mus- als. Med Sci Sports Exerc 2010;42:781–790.
cle–tendon complex in vivo. Scand J Med Sci Sports 2006;16:159–167. 39. Bojsen-Moller J, Magnusson SP, Rasmussen LR, Kjaer M, Aagaard P.
10. Van Cutsem M, Duchateau J, Hainaut K. Changes in single motor Muscle performance during maximal isometric and dynamic contrac-
unit behaviour contribute to the increase in contraction speed after tions is influenced by the stiffness of the tendinous structures.
dynamic training in humans. J Physiol 1998;513:295–305. J Appl Physiol 2005;99:986–994.
11. Carolan B, Cafarelli E. Adaptations in coactivation after isometric re- 40. Parker DF, Round JM, Sacco P, Jones DA. A cross-sectional survey of
sistance training. J Appl Physiol 1992;73:911–917. upper and lower limb strength in boys and girls during childhood
12. Kubo K, Kanehisa H, Ito M, Fukunaga T. Effects of isometric train- and adolescence. Ann Hum Biol 1990;17:199–211.
ing on the elasticity of human tendon structures in vivo. J Appl Phys- 41. Jones DA, Parker DF. Development of a portable strain gauge
iol 2001;91:26–32. to measure human muscle isometric strength. J Physiol 1989;145:
13. Narici MV, Hoppeler H, Kayser B, Landoni L, Claassen H, Gavardi 11P.
C, et al. Human quadriceps cross-sectional area, torque and neural 42. Sahaly R, Vandewalle H, Driss T, Monod H. Maximal voluntary force
activation during 6 months strength training. Acta Physiol Scand and rate of force development in humans—importance of instruc-
1996;157:175–186. tion. Eur J Appl Physiol 2001;85:345–350.
14. Pucci AR, Griffin L, Cafarelli E. Maximal motor unit firing rates during 43. Behm DG, St-Pierre DM, Perez D. Muscle inactivation: assessment of
isometric resistance training in men. Exp Physiol 2006;91:171–178. interpolated twitch technique. J Appl Physiol 1996;81:2267–2273.
15. Rich C, Cafarelli E. Submaximal motor unit firing rates after 8 wk 44. Scaglioni G, Martin A. Assessment of plantar flexors activation
of isometric resistance training. Med Sci Sports Exerc 2000;32:190–196. capacity: nerve versus muscle stimulation by single versus double
16. Garfinkel S, Cafarelli E. Relative changes in maximal force, EMG, pulse. Eur J Appl Physiol 2009;106:563–572.
and muscle cross-sectional area after isometric training. Med Sci 45. Button DC, Behm DG. The effect of stimulus anticipation on the
Sports Exerc 1992;24:1220–1227. interpolated twitch technique. J Sports Sci Med 2008;7:520–524.
17. Cannon J, Kay D, Tarpenning KM, Marino FE. Comparative effects 46. Allison GT. Trunk muscle onset detection technique for EMG sig-
of resistance training on peak isometric torque, muscle hypertrophy, nals with ECG artefact. J Electromyogr Kinesiol 2003;13:209–216.
voluntary activation and surface EMG between young and elderly 47. Pain MT, Hibbs A. Sprint starts and the minimum auditory reaction
women. Clin Physiol Funct Imaging 2007;27:91–100. time. J Sports Sci 2007;25:79–86.
18. Folland JP, Williams AG. The adaptations to strength training: mor- 48. Pulkovski N, Schenk P, Maffiuletti NA, Mannion AF. Tissue Doppler
phological and neurological contributions to increased strength. imaging for detecting onset of muscle activity. Muscle Nerve 2008;
Sports Med 2007;37:145–168. 37:638–649.
19. Gandevia SC. Spinal and supraspinal factors in human muscle fa- 49. Moretti DV, Babiloni F, Carducci F, Cincotti F, Remondini E, Rossini
tigue. Physiol Rev 2001;81:1725–1789. PM, et al. Computerized processing of EEG–EOG–EMG artifacts for
20. Pensini M, Martin A, Maffiuletti NA. Central versus peripheral adap- multi-centric studies in EEG oscillations and event-related potentials.
tations following eccentric resistance training. Int J Sports Med 2002; Int J Psychophysiol 2003;47:199–216.
23:567–574. 50. Herbert RD, Gandevia SC. Twitch interpolation in human muscles:
21. Colson SS, Martin A, Van Hoecke J. Effects of electromyostimulation mechanisms and implications for measurement of voluntary activa-
versus voluntary isometric training on elbow flexor muscle strength. tion. J Neurophysiol 1999;82:2271–2283.
J Electromyogr Kinesiol 2009;19:e311–9. 51. Kooistra RD, de Ruiter CJ, de Haan A. Conventionally assessed vol-
22. Moritani T, deVries HA. Neural factors versus hypertrophy in the time untary activation does not represent relative voluntary torque pro-
course of muscle strength gain. Am J Phys Med 1979;58:115–130. duction. Eur J Appl Physiol 2007;100:309–320.
23. Rabita G, Perot C, Lensel-Corbeil G. Differential effect of knee 52. Keenan KG, Farina D, Maluf KS, Merletti R, Enoka RM. Influence of
extension isometric training on the different muscles of the quadri- amplitude cancellation on the simulated surface electromyogram.
ceps femoris in humans. Eur J Appl Physiol 2000;83:531–538. J Appl Physiol 2005;98:120–131.
24. Reeves ND, Narici MV, Maganaris CN. Effect of resistance training 53. Hakkinen K, Newton RU, Gordon SE, McCormick M, Volek JS,
on skeletal muscle-specific force in elderly humans. J Appl Physiol Nindl BC, et al. Changes in muscle morphology, electromyographic
2004;96:885–892. activity, and force production characteristics during progressive
25. Shima N, Ishida K, Katayama K, Morotome Y, Sato Y, Miyamura M. strength training in young and older men. J Gerontol A Biol Sci
Cross education of muscular strength during unilateral resistance Med Sci 1998;53:B415–423.
training and detraining. Eur J Appl Physiol 2002;86:287–294. 54. Cochrane J, Lloyd D, Ackland T, Besier T, Elliott B. The effect of
26. Scaglioni G, Ferri A, Minetti AE, Martin A, Van Hoecke J, Capoda- lower limb training on muscular support of the knee and risk of an-
glio P, et al. Plantar flexor activation capacity and H reflex in older terior cruciate ligament injury. Proceedings of the XXIVth Interna-
adults: adaptations to strength training. J Appl Physiol 2002;92: tional Symposium on Biomechanics in Sports. Vol. 1.Salzburg,
2292–2302. Austria: University Press; 2006. p 1–4.
27. Brown AB, McCartney N, Sale DG. Positive adaptations to weight-lift- 55. Krishnan C, Williams GN. Variability in antagonist muscle activity
ing training in the elderly. J Appl Physiol 1990;69:1725–1733. and peak torque during isometric knee strength testing. Iowa
28. Jones DA, Rutherford OM. Human muscle strength training: the Orthop J 2009;29:149–158.
effects of three different regimens and the nature of the resultant 56. Adamson M, Macquaide N, Helgerud J, Hoff J, Kemi OJ. Unilateral
changes. J Physiol 1987;391:1–11. arm strength training improves contralateral peak force and rate of
29. Folland JP, Williams AG. Methodological issues with the interpolated force development. Eur J Appl Physiol 2008;103:553–559.
twitch technique. J Electromyogr Kinesiol 2007;17:317–327. 57. Lee M, Gandevia SC, Carroll TJ. Unilateral strength training
30. Taylor JL. Point: the interpolated twitch does/does not provide a increases voluntary activation of the opposite untrained limb. Clin
valid measure of the voluntary activation of muscle. J Appl Physiol Neurophysiol 2009;120:802–808.
2009;107:354–355. 58. Fimland MS, Helgerud J, Solstad GM, Iversen VM, Leivseth G, Hoff
31. de Haan A, Gerrits KH, de Ruiter CJ. Counterpoint: the interpolated J. Neural adaptations underlying cross-education after unilateral
twitch does not provide a valid measure of the voluntary activation strength training. Eur J Appl Physiol 2009;107:723–730.
of muscle. J Appl Physiol 2009;107:355–357. 59. Larsson L, Moss RL. Maximum velocity of shortening in relation to
32. Simoneau E, Martin A, Porter MM, Van Hoecke J. Strength training myosin isoform composition in single fibres from human skeletal
in old age: adaptation of antagonist muscles at the ankle joint. Mus- muscles. J Physiol 1993;472:595–614.
cle Nerve 2006;33:546–555. 60. Baumann H, Jaggi M, Soland F, Howald H, Schaub MC. Exercise
33. de Boer MD, Morse CI, Thom JM, de Haan A, Narici MV. Changes training induces transitions of myosin isoform subunits within histo-
in antagonist muscles’ coactivation in response to strength training chemically typed human muscle fibres. Pflugers Arch 1987;409:
in older women. J Gerontol A Biol Sci Med Sci 2007;62:1022–1027. 349–360.
34. de Ruiter CJ, Kooistra RD, Paalman MI, de Haan A. Initial phase of 61. Pette D, Staron RS. Mammalian skeletal muscle fiber type transitions.
maximal voluntary and electrically stimulated knee extension torque de- Int Rev Cytol 1997;170:143–223.
velopment at different knee angles. J Appl Physiol 2004;97:1693–1701. 62. Jurimae J, Abernethy PJ, Blake K, McEniery MT. Changes in the my-
35. Aagaard P, Simonsen EB, Andersen JL, Magnusson P, Dyhre-Poulsen osin heavy chain isoform profile of the triceps brachii muscle follow-
P. Increased rate of force development and neural drive of human ing 12 weeks of resistance training. Eur J Appl Physiol Occup Physiol
skeletal muscle following resistance training. J Appl Physiol 2002;93: 1996;74:287–292.
1318–1326. 63. Staron RS, Karapondo DL, Kraemer WJ, Fry AC, Gordon SE, Falkel
36. Newton RU, Kraemer WJ. Developing explosive muscular power: JE, et al. Skeletal muscle adaptations during early phase of heavy-re-
implications for a mixed methods training strategy. Strength Cond J sistance training in men and women. J Appl Physiol 1994;76:
1994;16:20–29. 1247–1255.

384 Neural Adaptations to Short-Term Resistance Training MUSCLE & NERVE March 2011