Content uploaded by Israel Halperin
Author content
All content in this area was uploaded by Israel Halperin on Nov 30, 2016
Content may be subject to copyright.
RESEARCH ARTICLE
The Effects of Either a Mirror, Internal or
External Focus Instructions on Single and
Multi-Joint Tasks
Israel Halperin
1,2
*, Steven Hughes
1,2
, Derek Panchuk
3,4
, Chris Abbiss
2
, Dale
W. Chapman
1,2
1Physiology Discipline, Australian Institute of Sport, Canberra, Australian Capital Territory, Australia,
2Centre for Exercise and Sport Science Research, School of Medical & Health Sciences, Edith Cowan
University, Joondalup, Western Australia, Australia, 3Institute of Sport, Exercise & Active Living, College of
Sport & Exercise Science, Victoria University, Footscray, Victoria, Australia, 4Skill Acquisition Discipline,
Australian Institute of Sport, Canberra, Australian Capital Territory, Australia
*Israel.halperin@ausport.gov.au
Abstract
Training in front of mirrors is common, yet little is known about how the use of mirrors effects
muscle force production. Accordingly, we investigated how performing in front of a mirror
influences performance in single and multi-joint tasks, and compared the mirror condition to
the established performance effects of internal focus (IF) and external focus (EF) instruc-
tions in a two part experiment. In the single-joint experiment 28 resistance-trained partici-
pants (14 males and 14 females) completed two elbow flexion maximal voluntary isometric
contractions under four conditions: mirror, IF, EF and neutral instructions. During these tri-
als, surface EMG activity of the biceps and triceps were recorded. In the multi-joint experi-
ment the same participants performed counter-movement jumps on a force plate under the
same four conditions. Single-joint experiment: EF led to greater normalized force production
compared to all conditions (P0.02, effect-size range [ES] = 0.46–1.31). No differences
were observed between neutral and mirror conditions (P= 0.15, ES = 0.15), but both were
greater than IF (P<0.01, ES = 0.79–1.84). Surface EMG activity was comparable across
conditions (P0.1, ES = 0.10–0.21). Multi-joint experiment: Despite no statistical difference
(P= 0.10), a moderate effect size was observed for jump height whereby EF was greater
than IF (ES = 0.51). No differences were observed between neutral and mirror conditions
(ES = 0.01), but both were greater than IF (ES = 0.20–22). The mirror condition led to supe-
rior performance compared to IF, inferior performance compared to EF, and was equal to a
neutral condition in both tasks. These results provide novel and practical evidence concern-
ing mirror training during resistance type training.
Introduction
Over the past two decades a large body of research has investigated the effects of attentional
focus conditions on motor learning and performance [1,2]. Specifically, the effects of
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 1 / 15
a11111
OPEN ACCESS
Citation: Halperin I, Hughes S, Panchuk D, Abbiss
C, Chapman DW (2016) The Effects of Either a
Mirror, Internal or External Focus Instructions on
Single and Multi-Joint Tasks. PLoS ONE 11(11):
e0166799. doi:10.1371/journal.pone.0166799
Editor: Nir Eynon, Victoria University, AUSTRALIA
Received: July 14, 2016
Accepted: November 3, 2016
Published: November 29, 2016
Copyright: ©2016 Halperin et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and Supporting Information files.
Funding: The authors received no specific funding
for this work.
Competing Interests: The authors have declared
that no competing interests exist.
instructions that elicit an internal focus (IF) or external focus (EF) of attention on exercise per-
formance have been commonly compared. IF leads individuals to focus on a specific body part
or muscle group, whereas EF leads individuals to focus on the intended effects of their move-
ments on the environment. Generally, research has reported that EF enhances motor learning
performance, when compared with IF instructions, and compared to neutral instructions,
which are deprived of an internal or external point of reference [2]. For example, instructing
participants to focus on the movement of their wrist during a basketball shot hinders accuracy,
when compared with focusing on the basket [3]. Superior performance with EF is observed
with tasks requiring large power output, such as long jump [4], sprint running [5], and in tasks
requiring maximal force such as single joint elbow flexion [6], and multi-joint exercises, such
as the isometric mid-thigh pull [7]. While preference for instructions/focus conditions has
been shown to have a small effect on performance [8,9], the benefits of EF are consistent
across tasks, skill level, and age groups [2].
Physical training is commonly performed in front of mirrors in numerous environments,
such as fitness gyms, martial arts and dancing studios. Despite the mirrors apparent popular-
ity, the few studies to investigate the influence of mirrors on motor performance have reported
mixed results [10–17]. For example, Bennett and Davids [11] observed that novice and inter-
mediate level powerlifters benefited from performing the squat exercise in front of a mirror
when asked to descend to a very precise and optimal depth. In contrast, advanced powerlifts
remained unaffected by the mirror when completing the same task. Furthermore, while studies
have found a mirror improves static balance performance in both young [15] and old [16]
adults, other investigators have not observed differences between mirror and no mirror
conditions [18]. In regards to dancing, practicing in front of mirrors enhanced learning and
performance of a dance sequence among experienced dancers [12], but hindered dancing per-
formance with untrained participants [13]. These experiments demonstrate the inconsistent
findings on the effects of mirror training on motor learning and performance. Furthermore,
all of the investigated tasks required movement accuracy, precision and balance, yet many gym
goers perform motor tasks that require maximal muscular tension in front of mirrors, such as
a barbell squat and biceps curls. However, to the best of our knowledge there are currently no
studies examining the effects of mirrors on performance during such tasks.
Conflicting findings also exist as to what emotions and perceptions training in front of a
mirror elicits. For example, studies have found that exercising in front of mirrors increase self-
efficacy [19], have no effect on self-efficacy [20], lead to a self-conscious negative body imagine
[21], and elicit negative feelings [22]. An analysis of interviews with dancers reported that mir-
rors may be a necessary tool to improve dancing technique [23]. Yet within this study the
dancers also stated that mirrors can lead to body objectification due to comparisons of oneself
to the image in the mirror. It is interesting to consider that the potential self-conscious
response elicited by mirrors is also associated with IF, which is known to hinder motor learn-
ing and performance [24]. In summary, the relevant literature concerning how mirrors may
affect emotions, perceptions, and feelings during physical training is conflicting. Given that
mirrors may influence perceptions and emotions as a result of visual feedback during exercise,
it is also of interest to understand how exercise in front of mirrors may affect one’s attentional
focus and overall performance. Finally, since females were the participants in the majority of
described studies above, it is of interest to compare the effects of mirrors on perceptions as
well as on performance between the genders.
It is plausible that looking at a mirror focuses one’s attention to the body part or muscle
groups being observed, and elicits IF. Conversely, since the body part being observed in the
mirror is external to the self the use of mirrors may elicit EF. Thus we sought to directly inves-
tigate this question using a two part study design. Specifically, the goals of these experiments
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 2 / 15
were fourfold: the first was to compare the effects of four sets of instructions: IF, EF, Neutral
and Neutral with the addition of a mirror, on maximal voluntary isometric contraction
(MVIC) and electromyography (EMG) activity of the elbow flexors, as well as on counter-
movement jumping performance. The second purpose was to examine if participants’ prefer-
ence for instructions/focus conditions was matched with their performance outcome, as
indicated by previous studies [8,9]. The third was to understand whether the use of a mirror is
perceived as either IF or EF by participants by use of a questionnaire. Finally, since most mir-
ror studies used females as participants, a comparison was made between male and female par-
ticipants since the use of mirrors has a possible gender effect.
Methods
Experimental rationale
The chosen experimental design sought to account for what we perceived to be two substantial
confounders of reported literature. The use of a dual experiment design with single and multi-
joint tasks were firstly chosen as it was of interest to compare tasks that vary in their degrees
of freedom of coordination or central nervous system input. That is, will a single-joint task
requiring significantly less degrees of freedom relative to the multi-joint task result in different
outcomes. Secondly we were concerned with issues of experimental validity versus ecological
validity. In the single-joint experiment participants were only able to observe the contracting
muscle group with the mirror, thereby increasing the study’s internal validity and attributing
any observed effects to the mirror. This design substantially increases the internal validity as
any effects on force production can be attributed to the mirror with a greater degree of cer-
tainty. However, this design also has low ecological validity, as it is unlikely that people train
and practice in their natural environments while only being able to observe the contracting
muscles with a mirror. Thus, in the multi-joint experiment we sought to compensate for the
low ecological validly of single-joint experiment, and implement a jumping protocol in which
participants were able to observe movement of their whole body within the mirror as they
performed the requested task. Hence, the multi-joint experiment design represents training
environments to a greater extent which enhances the degree of both external and ecological
validity.
Participants
Twenty-eight resistance-trained participants volunteered for both experiments (14 males and
14 females, age: 26±5 y, weight: 70±11 kg). All participants had performed resistance training
at least twice a week for the past year, and participated in various sporting activities such as
Soccer, Rugby and Judo once to three times a week. Participants were provided with a carefully
presented verbal description of the study, so as to not compromise the study design. Thereaf-
ter, each athlete provided written informed consent. The study was approved by the Australian
Institute of Sport Ethics Committee and conformed to the declaration of Helsinki for human
research. The individual in Fig 1 has given written informed consent (as outlined in the PLOS
consent form) to publish these figures.
Participants attended the laboratory on a single occasion with the maximal voluntary iso-
metric contraction (MVIC) single-joint experiment performed first followed by the counter-
movement jump (CMJ) multi-joint experiment, after which they completed the questionnaires
for both experiments. On arrival each participant was familiarised with the MVIC testing
protocol, and thereafter completed the experimental session as described below. During a 5
minute resting period between experiments each participant was familiarised with the CMJ
testing protocol, and thereafter completed the CMJ experimental session as described below.
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 3 / 15
Single-joint experiment
Procedures. All data collection was carried out in a quiet room by the same two investiga-
tors, in which the same investigator provided the instructions to all participants. Participants
were informed about the importance of maintaining a straight gaze during all trials (other
than the mirror condition) with the goal of eliminating possible vision confounders as a result
of the instructions. In an attempt to control for gaze, a large mono-tone board was placed in
front of participants in both tests to block out potential vision confounders that they may
focus on otherwise (Fig 1A).
Prior to initiating the MVIC test, sEMG electrodes were attached to participant’s biceps
and triceps brachii muscles. Each participant was then seated on the preacher curl bench (RM,
China) with the seat height adjusted so that the elbow joint was at a 90˚ angle during each iso-
metric contraction and a strap was secured around their wrist which was attached to a force
transducer (Fig 1A). Participants then performed a warm up consisting of ten elbow flexion
contractions at an intensity equal to ~50% of their perceived maximum (work to rest ratio of
2/2 s) and one 3 s contraction at an intensity equal ~80% of their perceived maximum. There-
after participants rested for two minutes and then completed two baseline MVICs, lasting 3 s,
separated by 30 s of rest. Instructions for the baseline contractions were the same as the Neu-
tral instructions which were “Attempt to produce as much force as you possibly can”. After
completion of the second contraction participants were given two minutes of rest. Participants
then received one of four instructions in a randomized order prior to completing two MVICs
per condition separated by 30 s of rest. Two minutes of rest were provided between each
instructional condition.
The instructions provided to each participant are described in Table 1 for each of the inves-
tigated conditions. In the Mirror condition a 0.2 x 0.08 m mirror was installed ~0.6 m away
from the participants at eye level (Fig 1A). The size of the mirror was constrained so it only
allowed participants to see their elbow flexors contracting. To reduce the possibility of partici-
pants focusing on the EMG electrodes, a small skin coloured wrap was placed around them.
Other than the single instructional sentence no other guidelines, encouragement, verbal or
visual feedback were provided.
Table 1. Instructions.
Single-joint
IF Attempt to produce as much force as you possibly can while focusing on contracting
your arm muscles as hard and as fast as you can
EF Attempt to produce as much force as you possibly can while focusing on pulling the
strap as hard and as fast as you can
N Attempt to produce as much force as you possibly can
M Attempt to produce as much force as you possibly can while looking at yourself in the
mirror
Multi-joint
IF Attempt to jump as high as you can while focusing on contracting your leg muscles as
hard and as fast as you can
EF Attempt to jump as high as you can while focusing on pushing of the ground as hard and
as fast as you can
N Attempt to jump as high as you can
M Attempt to jump as high as you can while looking at yourself in the mirror
The instructions provided to each participant in the single and multi-joint experiments for the Internal Focus
(IF), External Focus (EF), Neutral (N) and Mirror (M) investigation conditions.
doi:10.1371/journal.pone.0166799.t001
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 5 / 15
Maximum voluntary isometric contraction (MVIC). Subjects were seated on a preacher
curl device with their upper arm supported and elbow flexed at 90˚. Secured around the
wrist was a padded strap attached by a high-tension wire to a load cell (capacity 200 kg,
Sensitivity = 10.2μV/N, Vishay, Australia) to measure elbow flexion forces. All force data were
sampled at a rate of 1000 Hz by a personal computer via an 8 channel data acquisition system
(PowerLab, ADInstruments, Australia) operated by Labchart software (ADInstruments, Aus-
tralia) sampling at 1000 Hz, allowing direct measurement of force-time characteristics. Mean
force was determined for all contractions. The mean was determined over a 2-s window
defined as 0.5 s after the initiating of the contraction and 0.5 before it ended. Due to the
expected large inter-subject variability between genders in maximal force production all mean
force values were normalized to the baseline condition, and thus reported and analysed as a
percentage.
Electromyography (EMG). Surface electromyography (sEMG) recording electrodes
(Viasys, USA) were placed approximately 3 cm apart over the proximal, lateral segment of the
biceps brachii and over the lateral head of the triceps brachii. Skin preparation included shav-
ing and cleansing of the area with an isopropyl alcohol swab and allowing to air dry prior to
placement of the electrodes. sEMG was collected using a 8 channel data acquisition system
(PowerLab, ADInstruments, Australia) with Labchart software (ADInstruments, Australia)
sampling at 1000 Hz with a 2 MOimpedance, common mode rejection ratio >110 dB min
(50/60 Hz), and noise >5μV. A bandpass filter (10–500 Hz) was applied prior to digital con-
version. Using the same 2 s window as the force analysis, mean root mean square (RMS) of the
sEMG was determined using a window width of 50 ms and then a mean value was calculated.
Analysis of these values was performed in two separate ways; first, they were normalized to
baseline and reported as a percentage and second, the absolute mV Biceps brachii values were
divided by absolute mV Triceps brachii to provide a co-contraction ratio.
Multi-joint experiment
Procedures. The warm up for this experiment included low-intensity cycling for 5 min-
utes, followed by 5 minutes of self-selected dynamic stretching. Participants were positioned
on a force plate while holding a lightweight (0.4 kg) aluminium bar across their shoulders. As
an extension of the warm up, participants completed 10 submaximal CMJ equal to ~50% of
their perceived maximal height, and then one CMJ equal to ~80% of their perceived maxi-
mum. Similar to the single joint experiment, participants then completed two CMJs separated
by 30 s per condition and two minutes of rest between conditions. The instructions provided
to each participant in this experiment are described in Table 1 for each of the investigated con-
ditions. In the Mirror condition a 1.76 by 0.56 m mirror was placed ~2 m away from the centre
of the force plate (Fig 1B) whereas in all other conditions a large mono-tone board was placed
in front of participants to block out potential vision confounders. In contrast to the more con-
trolled routine of the single-joint experiment in which participants could only see their elbow
flexors, in this experiment participants were free to choose what they would look at in the mir-
ror. Other than the single instructional sentence no other guidelines, encouragement, verbal
or visual feedback were provided. Finally, after the completion of this experiment, participants
answered a questionnaire on their preferred instruction and reported if the neutral-mirror
instruction elicited a stronger EF or IF response for both the single and multi-joint experi-
ments (see below).
Countermovement jumps. The countermovement jump (CMJ) trials were completed on
a commercially available portable force plate (9290AD Quattro Jump, Kistler, Switzerland).
Additionally, a single linear position transducer (Ballistic Measurement System, Fitness
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 6 / 15
Technology, Adelaide, Australia) was mounted directly above the participant and utilised to
directly measure displacement via a tether attached to the centre of the aluminium pole held
by the participant across their shoulders during each CMJ trial. The force plate and a linear
position transducer were synchronised and interfaced with a personal computer via an 8 chan-
nel data acquisition system (PowerLab, ADInstruments, Australia) with Labchart software
(ADInstruments, Australia) sampling at 1000 Hz, allowing direct measurement of force-time
characteristics. Ground reaction forces and linear position transducer were analysed using
Labchart software and custom macros. Prior to all data collection, the force plate was cali-
brated using a range of known loads and the linear position transducer was calibrated using a
two point calibration process and a known distance. The utilisation of the aluminium bar
across the shoulders eliminated arm swing from the movement and thus our outcome mea-
sures provide a reflection of only lower body performance capabilities and not the general ver-
tical jumping capacity. Due to the expected large inter-subject variability between genders in
maximal jump height, and due to a possible order effect resulting from completing a repeated
number jumps, all mean maximal jump values (cm) were normalized to baseline condition,
and thus reported and analysed as a percentage.
Common procedures
Questionnaires. Participants answered a two part questionnaire (S1 File) after the com-
pletion of the multi-joint experiment. Participants were asked to rank the four listed instruc-
tions in accordance with their preference for eliciting their best performance, with 1 being the
most preferred and 4 being the least preferred. Participants were then asked to report if the
mirror instructions were perceived as more of an IF or EF. This was achieved by having partic-
ipants mark a vertical line over a 20 cm horizontal line which had EF listed on the left side, and
IF on right side. The distance of the drawn vertical line from the midpoint was then measured
with a ruler to provide a quantification of how strongly a participant rated the mirror condi-
tion as either IF or EF.
Statistical analysis. In the single joint experiment data from the two MVICs completed in
each of the five conditions were averaged and used for further analysis. To examine if signifi-
cant differences existed between the first and second contractions in each of the five condi-
tions, a two-way Analysis of Variance (ANOVA) with repeated measures was conducted
(conditions [5] x MVICs [2]). A two-way ANOVA with repeated measures was used to com-
pare the mean normalized forces and EMG activity, between the four conditions, and to inves-
tigate if a gender effect exists (instructions [4] x gender [2]). An additional two-way ANOVA
with repeated measures was used to compare the order of preferences on normalized force
production, and to investigate if a gender effect exists (instruction preferences [4] x gender
[2]) on normalized force production. If the assumption of Sphericity was violated, the Green-
house—Geisser correction was employed with an LSD post hoc test if a main effect was identi-
fied or paired t-tests with a Boferroni correction if an interaction was identified. In the multi-
joint experiment the data from the two CMJs completed in each of the five conditions were
averaged and used for further analysis. Using a similar statistical approach as in the single joint
experiment, a two-way ANOVA with repeated measures was conducted to investigate if the
first jump differed from the second jump within each condition (conditions [5] x jump height
[2]). A two-way ANOVA with repeated measures was used to compare the mean normalized
jump height, peak vertical concentric force and peak concentric velocity, between the four
conditions, and to investigate if a gender effect exists (instructions [4] x gender [2]). An addi-
tional two-way ANOVA with repeated measures was used to compare the order of preferences
on jump height, and to investigate if a gender effect exists (instruction preferences [4] x gender
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 7 / 15
[2]) on jump height. A Chi-square test of independence was used to examine differences
between the number of participants who rated the Mirror condition as either IF or EF. Statisti-
cal significance was accepted as P<0.05 for all tests. Furthermore, 95% confidence intervals
(CI) of the mean percent differences and Cohen deffect sizes (ES) were reported when appro-
priate. The magnitudes of these ES were classified as trivial (0–0.19), small (0.20–0.49),
medium (0.50–0.79) and large (0.80 and greater) using the scale advocated by Cohen [25].
Results
Single-joint experiment
No significant interaction was identified between the first and second MVIC in each of the five
conditions (P= 0.289). The mean (±SD) absolute force (N) produced in the MVIC in each of
the conditions were: EF (268±74 N), IF (240±72 N), Neutral (260±74 N), Mirror (255±72 N)
and IF (240±72 N) (S2 File). In this experiment a main effect for instruction type was identified
(P<0.001), however no significant interactions were identified between gender and instruc-
tion (P= 0.741). Specifically, participants produced significantly greater normalized mean
force in EF compared to IF (P<0.001; ES = 1.31; CI 95% [6.3, 15.6%]), Neutral (P= 0.028;
ES = 0.46; CI 95% [0.5, 6.3%]) and Mirror (P= 0.017; ES = 0.67; CI 95% [0.9, 8.9%]). When
compared to IF, greater normalized force was produced in the Neutral (P<0.001; ES = 0.98;
CI 95% [4, 11.2%]), and Mirror conditions (P<0.001; ES = 0.79; CI 95% [2.7, 9.3%]), however,
no differences were observed between Neutral and Mirror conditions (P= 0.392; ES = 0.14; CI
95% [-2, 5.1%]) (Fig 2A). No significant differences were discerned between the four condi-
tions in normalized sEMG activity of biceps brachii (P0.972), triceps brachii (P= 0.588),
or co-contraction ratio (absolute mV activity of biceps brachii/ triceps brachii) (P= 0.979).
The lack of statistical significant sEMG differences were accompanied by small effect sizes
(ES0.12).
There were no significant interactions (P= 0.445) or main effects (P= 0.226) for the partici-
pants (n = 25) preferences of instructions on normalized force production (Table 2). That is,
the most preferred instruction did not elicit greater force production compared to the least
preferred. However, there was a moderate effect (ES = 0.32) for the greatest forces to be associ-
ated with the most preferred (EF), compared with other instruction. No differences were seen
between the 2
nd
, 3
rd
and 4
th
ranked instructions (ES0.01). No significant differences (X
2
=
1.01, P= 0.297) were observed between the number of participants who rated the Mirror con-
dition as an EF (n = 15), compared to participants who rated it as an IF (n = 10). The strength
of the participants’ perception of how the mirror instructions compared to IF and EF is illus-
trated in Fig 3A.
Multi-joint experiment
A significant interaction was identified between the first and second jump completed across
conditions (P= 0.042), however the Bonferroni corrected t-test post hoc comparisons were
not statistically different within conditions. The mean (±SD) absolute jump heights for all
conditions were as follows: EF (38.2±7.4 cm), Neutral (37.6±7.6 cm), Mirror (37.7±7.9 cm)
and IF (37.2±7.0 cm) (S2 File). No significant interactions between gender and instructions
(P= 0.346), or a main effect for instructions were identified (P= 0.101). However, despite the
lack of statistical differences, the results of this experiment followed a similar pattern to the sin-
gle-joint experiment in which EF led to greater jump height compared to IF (ES = 0.48; CI
95% [0.36, 4.3%]) and to slightly higher jump heights compared to Neutral (ES = 0.27; CI 95%
[-0.41, 3.0%]) and Mirror conditions (ES = 0.26; CI 95% [0.36, 4.3%]) (Fig 2B). No differences
were observed between Neutral and Mirror conditions (ES = 0.01; CI 95% [-1.81, 1.92%], but
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 8 / 15
Fig 2. Normalized Performance Measures. (A) MVIC data from single-joint experiment. (B) CMJ data from multi-
joint experiment. Note: each square represents data from a single participant and the black horizontal lines
represent the group average per condition.
doi:10.1371/journal.pone.0166799.g002
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 9 / 15
compared to IF, slightly greater jump heights were observed with both Neutral (ES = 0.22; CI
95% [-0.46, 2.90%]) and Mirror (ES = 0.20; CI 95% [-0.57, 3.21%]) conditions. No significant
or meaningful differences were identified between the four conditions for normalized peak
force (P= 0.402), mean force (P= 0.670) and mean velocity (P= 0.447). The lack of significant
differences was accompanied by small effect sizes (ES0.19). However, peak velocity was
statistically significant between conditions (P= 0.018) with EF resulting in greater peak veloci-
ties compared to IF (P= 0.01; ES = 0.19; 95% [0.015–0.108 ms
-2
) and compared to Mirror
(P= 0.02; ES = 0.13; 95% [0.007–0.092 ms
-2
). The Neutral instructions lead to significantly
greater peak velocities compared to Mirror (P= 0.014; ES = 0.12; 95% [0.010–0.085 ms
-2
) and
IF (P= 0.037; ES = 0.17; 95% [0.004–0.116 ms
-2
).
There were no significant interactions (P= 0.680) in participant instruction preferences
(Table 2), however, a main effects for conditions (P= 0.038) was identified. Note that question-
naire data is missing from three participants. The differences between the most and least
preferred instructions were not matched for jump height performance. That is, jumping
performance did not follow the rank of preferred instructions. The third most preferred
instruction elicited greater jump heights compared to all other preferences. Specifically, signifi-
cantly greater jump heights were found compared to the second most preferred instruction
(P= 0.029; ES = 0.43; CI 95% [0.5, 4.2%]), the fourth (P= 0.028; ES = 0.52; CI 95% [0.3,
4.9%]). A significant difference (X
2
= 4.83, P= 0.027) was observed between the number of
participants who rated the Mirror condition as an EF (n = 19), compared to participants who
rated it as an IF (n = 5). It should be noted that one participant perceived the Mirror as neither
EF or IF. The strength of the participants’ perception of how the mirror instructions compared
to IF and EF is illustrated in Fig 3B.
Discussion
The primary goals of the two experiments were to examine how performance of an isometric
single-joint, and a dynamic multi-joint tasks would be affected by performing in front of a
mirror; and compare the mirror performance results to the well investigated EF and IF
instructions. Elbow flexion forces were greater with EF and the lowest with IF, and a similar
trend, albeit not statistically significant, was found with jump heights. Furthermore, in both
studies performance in the Mirror conditions were comparable to the Neutral condition.
That is, both the Mirror and Neutral conditions were lower than EF but greater than IF.
The secondary goals of these experiments were to investigate if participants’ preferences of
Table 2. Instructional preferences.
Preference rankings External Neutral Mirror Internal
Single-joint
1
st
12 3 5 5
2
nd
6 5 8 6
3
rd
7 5 6 7
4
th
0 12 6 7
Multi-joint
1
st
15 2 5 3
2
nd
3 9 3 10
3
rd
5 7 7 6
4
th
2 7 10 6
Participants’ preferences of the four instructions in both experiments.
doi:10.1371/journal.pone.0166799.t002
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 10 / 15
Fig 3. Perception of instructions. Strength of the participants’ perception of the mirror instructions
compared to IF and EF in the single-joint (A) and multi-joint (B) experiments.
doi:10.1371/journal.pone.0166799.g003
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 11 / 15
instructions match their performance; to descriptively analyse if participants perceived the
mirror as EF or IF; and examine if a gender effect would be observed. The stated preferences
of instructions were not matched with either elbow flexion forces or with jump performance;
the majority of participants perceived the mirror to elicit an external focus, although the
strength of perception of the mirror differed widely between participants and experiments.
Finally, no gender effect was observed in either experiment.
The differences in performance observed in this study are aligned with previous work, in
which EF leads to superior performance and IF results in inferior performance [2]. Further-
more, within this study the mirror condition did not seem to result in a reduction or improve-
ment in performance. These results support some [18,21], but not all [16,17], studies
investigating the effect of mirrors on motor activities. Note that the majority of studies to date
that have investigated the influence of a mirror on performance during a motor task have com-
pared it solely to a Neutral condition [11,13,15,18]. In contrast, in the present study, the use
of a mirror was also compared to EF and IF conditions which extend our understanding of
how mirrors affect performance in a relation to the well-established focus conditions. Further,
while previous mirror studies have investigated outcome measures such as balance [16], accu-
racy [17], movement economy [14] and motor learning [15], to our knowledge no study inves-
tigated a maximal force and jumping tasks as in the present study. Investigating such tasks is
important as both trained and untrained participants commonly perform resistance training
exercises in front of mirrors in gym environments. While the presence of a mirror may be of
value in movement tasks requiring accuracy, such as squat depth assessment [11], our study
indicates that the mirror does not provide meaningful benefits in activities requiring maximal
force and in jumping performance. Interestingly, however, sEMG of both the agonist and
antagonist muscle groups did not differ between any conditions in the single-joint experiment
which is in contrast to previous attentional focus research on the elbow flexors [6,26]. These
contrasting findings may be in part due to differences in signal normalization techniques, as
well as the difference in implemented contraction types. Whereas in the present study isomet-
ric contractions were used, dynamic contractions were used in the two previous studies.
The results from both experiments indicate a lack of relationship between the preference of
instruction and performance outcomes. That is, irrespective of how participants ranked their
preferences for the four instructions in both studies, force and jump height remained unaf-
fected. This supports the previous work of Wulf et al. [27], in which a balance task was com-
pleted with fewer errors with EF irrespective if participants preferred IF or EF. Other authors
have reported that participants’ preferences of IF and EF influence their performance to some
extent in tasks requiring accuracy, such as dart throwing, billiards and basketball throws [8,9,
28]. However, within the studies investigating the relationship between attentional instructions
and participants preferences, the benefits of EF persist despite situations where use of a non-
preferred focus condition was imposed. That is, performance of participants who preferred EF
but were asked to use their non-preferred IF suffered to a greater extent compared to those
who preferred IF but were asked to use their non-preferred EF [8,9,28]. Thus, while prefer-
ences of focus conditions can account for some of the effects on performance, it seems as if
performance is affected to a greater extent by the type of focus instruction adopted.
Similar to other studies [8,27], participants in both experiments generally ranked EF as
their most preferred focus conditions (Table 2). Whereas a considerable range of perceptions
were reported regarding the degree to which the Mirror condition elicited IF and EF in the sin-
gle-joint experiments (Fig 3A), most participants in the multi-joint experiment reported that
the Mirror was perceived more as an EF rather than IF (Fig 3B). This observation is interesting
as mirrors can be expected to either; 1) elicit a self-conscious response and thereby lead to IF,
or 2) to shift participants focus away from themselves as they observe the mirror and thereby
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 12 / 15
elicit an EF. We speculate that the variation of individual response may be reflective of training
and life experiences. Future investigations should seek cohorts of participants that could be
initially classified on sporting skill level or experience in an environment to continue to refine
our understanding on the use of mirrors as an instruction focus tool.
The constrained action hypothesis proposed by Wulf et al. [29] provides an explanation of
our observed differences in performance between EF and IF in both experiments. In this
regard, it is hypothesized that EF allows participants to self-organize in an automatic manner
and perform the task unconstrained by conscious control. Conversely, IF disrupts the automa-
ticity of performance, making participants conscious of their movements. Although not
overtly evident in the single joint experiment, performing an MVIC required participants to
stabilise and synchronise their shoulder and trunk muscles as they performed the contraction.
The requirement to synchronise and coordinate numerous body parts and muscle groups to
elicit optimal performance is more evident in the multi-joint experiment. Thus, we speculate
that IF leads participants to focus on a single component of a complex movement task, which
reduces the contribution of other body parts and muscle groups, thereby hindering perfor-
mance. In contrast, EF allows participants to organize the relevant contributors around the
motor task without neglecting any one of the contributors in a more natural organisation of
the motor pathway.
Our observation that the mirror condition was more neutral in the performance effect can-
not be neatly explained by the constrained action hypothesis. However, given the inter-individ-
ual perception of the mirror condition as either IF and EF, we speculate that the constrained
action hypothesis can account for both the negative and positive effects as a function of the mir-
rors perception as IF or EF. In cases in which the mirror elicit a negative effect then the use of
a mirror is inducing a partial IF response, and in contrast, when participants focus on what
they observe in the mirror as external to the self, a partial EF response results. Future studies
should utilize specific IF and EF instructions as participants observe their movement within a
mirror to enhance our understanding of the constrained action hypothesis.
It would be remiss if we did not consider the impact of our imposed experiment design
constraints on the observed outcomes. An important consideration within this study was our
decision to not counterbalance the order of the two experiments due to logistical constrains
which could have led to an order effect or bias participants’ expectations. While we did seek to
compare the magnitude of response between conditions, the smaller effect sizes observed in
the multi-joint experiment could be related the order in which the experiments were con-
ducted. There is also the possibility that participants did not receive adequate familiarization
with the motor tasks. Particularly, there were some inter-individual differences related to par-
ticipants experience with the jumping task. These experience differences between participants
could partially account for the smaller effects observed in the second experiment. Finally, the
preference questionnaires for both experiments were conducted only after the completion of
the multi-joint experiment. Thus, this elapsed time between the completion of the single-joint
experiment and the questionnaires completion could have somewhat skewed the results. How-
ever, not doing so would have compromised the efficacy of the second experiment.
Conclusion
We have reported that EF leads to superior performance in both a single and multi-joint tasks
compared to all conditions, and that IF leads to inferior performance in such tasks. The Mirror
condition led to inferior performance compared to EF, superior performance compared to IF
and was comparable to the Neutral conditions. A lack of relationship between participants’
preferences of instruction type to performance outcomes was observed, as well as a wide range
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 13 / 15
of responses pertaining to how the Mirror condition was perceived in relation to IF and EF.
Finally, the effects were similar between males and females. We emphasised internal validity in
the single-joint experiment and external and ecological validity in the multi-joint experiment.
Since the results followed a similar pattern in both experiments, we consider these findings to
be robust. These results are of practical relevance given the popularity of training in front of
mirrors in studios and gyms, and also expand our understanding of how focus conditions
influence performance.
Supporting Information
S1 File. Questionnaire.
(PDF)
S2 File. Absolute values for both experiments.
(XLSX)
Acknowledgments
The authors would like to thank the participants for giving their time to take part in this study.
Author Contributions
Conceptualization: IH SH DC DP CA.
Data curation: IH SH.
Formal analysis: IH SH DC.
Investigation: IH SH.
Methodology: IH SH DC DP CA.
Project administration: IH SH.
Software: IH SH.
Supervision: DC CA.
Visualization: IH DC.
Writing – original draft: IH SH DC DP CA.
Writing – review & editing: IH SH DC DP CA.
References
1. Marchant DC. Attentional focusing instructions and force production. Frontiers in Psychology. 2011; 26
(1):210.
2. Wulf G. Attentional focus and motor learning: a review of 15 years. Int J Sports Psychol. 2013; 6(1):77–
104.
3. Zachry T, Wulf G, Mercer J, Bezodis N. Increased movement accuracy and reduced EMG activity as
the result of adopting an external focus of attention. Brain Res Bull. 2005; 67(4):304–9. doi: 10.1016/j.
brainresbull.2005.06.035 PMID: 16182938
4. Porter JM, Anton PM, Wikoff NM, Ostrowski JB. Instructing skilled athletes to focus their attention exter-
nally at greater distances enhances jumping performance. J Strength Cond Res. 2013; 27(8):2073–8.
doi: 10.1519/JSC.0b013e31827e1521 PMID: 23222082
5. Ille A, Selin I, Do M-C, Thon B. Attentional focus effects on sprint start performance as a function of skill
level. J Sport Sci. 2013; 31(15):1705–12.
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 14 / 15
6. Marchant DC, Greig M, Scott C. Attentional focusing instructions influence force production and muscu-
lar activity during isokinetic elbow flexions. J Strength Cond Res. 2009; 23(8):2358–66. doi: 10.1519/
JSC.0b013e3181b8d1e5 PMID: 19826287
7. Halperin I, Williams K, Martin DT, Chapman DW. The effects of attentional focusing instructions on
force production during the isometric mid-thigh pull. J Strength Cond Res. 2016; 30(4):919–23 doi: 10.
1519/JSC.0000000000001194 PMID: 27003451
8. Marchant DC, Clough PJ, Crawshaw M, Levy A. Novice motor skill performance and task experience is
influenced by attentional focusing instructions and instruction preferences. Int J Sports Psychol. 2009; 7
(4):488–502.
9. Ehrlenspiel F, Lieske J, Ru¨bner A. Interaction between preference and instructions for a focus of atten-
tion in billiards. Percept Mot Skills. 2004; 99(1):127–30. doi: 10.2466/pms.99.1.127-130 PMID:
15446637
10. Sewall LP, Reeve TG, Day RA. Effect of concurrent visual feedback on acquisition of a weightlifting
skill. Percept Mot Skills. 1988; 67(3):715–8. doi: 10.2466/pms.1988.67.3.715 PMID: 3226820
11. Bennett S, Davids K. The manipulation of vision during the powerlift squat: Exploring the boundaries of
the specificity of learning hypothesis. Res Q Exerc Sport. 1995; 66(3):210–8. doi: 10.1080/02701367.
1995.10608835 PMID: 7481082
12. Dearborn K, Ross R. Dance learning and the mirror: comparison study of dance phrase learning with
and without mirrors. Journal of Dance Education. 2006; 6(4):109–15.
13. Radell SA, Adame DD, Cole SP. Effect of teaching with mirrors on ballet dance performance. Percept
Mot Skills. 2003; 97(3):960–4.
14. Eaves DL, Hodges NJ, Williams AM. Energetic costs of incidental visual coupling during treadmill run-
ning. Med Sci Sports Exerc. 2008; 40(8):1506–14 doi: 10.1249/MSS.0b013e31817057a8 PMID:
18614938
15. Rougier P, editor [How mirror feedback improves undisturbed upright stance control]. Annales de read-
aptation et de medecine physique: revue scientifique de la Societe francaise de reeducation fonction-
nelle de readaptation et de medecine physique; 2002.
16. Vaillant J, Vuillerme N, Janvy A, Louis F, Juvin R, Nougier V. Mirror versus stationary cross feedback in
controlling the center of foot pressure displacement in quiet standing in elderly subjects. Arch Phys Med
Rehabil. 2004; 85(12):1962–5. PMID: 15605333
17. Lynch JA, Chalmers GR, Knutzen KM, Martin LT. Effect on performance of learning a Pilates skill with
or without a mirror. J Bodyw Mov Ther. 2009; 13(3):283–90. doi: 10.1016/j.jbmt.2008.09.003 PMID:
19524854
18. Notarnicola A, Maccagnano G, Pesce V, Di Pierro S, Tafuri S, Moretti B. Effect of teaching with or with-
out mirror on balance in young female ballet students. BMC research notes. 2014; 7(1):426.
19. Katula JA, McAuley E. The mirror does not lie: Acute exercise and self-efficacy. Int J Behav Med. 2001;
8(4):319–26.
20. Lamarche L, Gammage KL, Strong HA. The effect of mirrored environments on self-presentational effi-
cacy and social anxiety in women in a step aerobics class. Psychol Sport Exerc. 2009; 10(1):67–71.
21. Radell SA, Adame DD, Cole SP. Effect of teaching with mirrors on body image and locus of control in
women college ballet dancers. Percept Mot Skills. 2002; 95(3):1239–47.
22. Ginis KAM, Jung ME, Gauvin L. To see or not to see: Effects of exercising in mirrored environments on
sedentary women’s feeling states and self-efficacy. Health Psychol. 2003; 22(4):354. PMID: 12940391
23. Radell SA, Keneman ML, Adame DD, Cole SP. My body and its reflection: a case study of eight dance
students and the mirror in the ballet classroom. Res in Dance Educ. 2014; 15(2):161–78.
24. McKay B, Wulf G, Lewthwaite R, Nordin A. The self: Your own worst enemy? A test of the self-invoking
trigger hypothesis. Q J Exp Psychol. 2015; 68(9):1910–19.
25. Cohen J. A power primer. Psychol bull. 1992; 112(1):155. PMID: 19565683
26. Greig M, Marchant D. Speed dependant influence of attentional focusing instructions on force produc-
tion and muscular activity during isokinetic elbow flexions. Hum Movement Sci. 2014; 33:135–48.
27. Wulf G, Shea C, Park J-H. Attention and motor performance: Preferences for and advantages of an
external focus. Res Q Exerc Sport. 2001; 72(4):335–44. doi: 10.1080/02701367.2001.10608970 PMID:
11770783
28. Weiss SM, Reber AS, Owen DR. The locus of focus: The effect of switching from a preferred to a non-
preferred focus of attention. J Sport Sci. 2008; 26(10):1049–57.
29. Wulf G, McNevin N, Shea CH. The automaticity of complex motor skill learning as a function of atten-
tional focus. Q J Exp Psychol. 2001; 54(4):114
The Effects of Mirror, Internal or External Instructions on Single and Multi-Joint Tasks
PLOS ONE | DOI:10.1371/journal.pone.0166799 November 29, 2016 15 / 15
