Gait & Posture 40 (2014) 556–560

Contents lists available at ScienceDirect

Gait & Posture

journal homepage: www.elsevier.com/locate/gaitpost

Visual availability, balance performance and movement complexity in

dancers
Ruth Muelas Pérez a, Rafael Sabido Solana b,*, David Barbado Murillo b,
Francisco Javier Moreno Hernández b
a
Professional Dance School of Castilla y León, Burgos, Spain
b
Sport Research Centre, Miguel Hernández University of Elche, Elche, Spain

A R T I C L E I N F O A B S T R A C T

Article history: Research regarding the complex fluctuations of postural sway in an upright standing posture has yielded
Received 28 February 2013 controversial results about the relationship between complexity and the capacity of the system to
Received in revised form 12 May 2014 generate adaptive responses. The aim of this study is to compare the performance and complexity of two
Accepted 30 June 2014
groups with different levels of expertise in postural control during a balance task. We examined the
balance ability and time varying (dynamic) characteristics in a group of 18 contemporary dancers and 30
Keywords: non-dancers in different visual conditions. The task involved maintaining balance for 30 s on a stability
Balance ability
platform with opened or closed eyes. The results showed that dancers exhibited greater balance ability
Complexity
Dance
only in open eyes task than non-dancers. We also observed a lower performance in both groups during
Visual information the test with closed eyes, but only dancers reduced their complexity in closed eyes task. The main
conclusion is that the greater postural control exhibited by dancers depends on the availability of visual
information.
ß 2014 Elsevier B.V. All rights reserved.

1. Introduction Keshner showed the benefits of a balance-training program, in

Postural control during the maintenance of an upright standing
posture is a fundamental motor act that provides the basis for
locomotion and most other movement tasks [1]. The postural
control system regulates the body’s postural sway during upright
standing through the complex interaction of somatosensory, visual
and vestibular sensory feedback networks, numerous brain
regions, and the musculoskeletal system [2–4]. Complexity is
defined as the number of system components and coupling
functions (interactions) among the components [1]. This complex-
ity can be observed in the upright standing posture through
fluctuations of postural sway [4–6] and has increasingly led
scientists to analyze postural stability through non-linear mathe-
matical tools [7–11].
The results observed through this type of analysis have allowed
scientists to relate lower complexity levels to a worse performance
[12] related to aged and unhealthy systems [13]. Haran and
* Corresponding author at: Sport Research Centre of Elche, Miguel Hernández
University, Avda. de la Universidad s/n, 03202 Elche, Spain. Tel.: +34 965 22 24 37;
fax: +34 965 22 24 56.
E-mail address: rsabido@umh.es (R. Sabido Solana).
which the unhealthy participants improved their postural control,
and the complexity of postural sway was increased [14].
By applying different levels of difficulty, depending on the
availability of visual information, studies have shown a reduced
performance associated with less postural sway complexity when
the subjects kept their eyes closed [11,15]. However, some authors
state that, depending on the specialization of the sample in
proprioceptive tasks, the decrease in performance will generally be
significant. Some results have shown that gymnasts only worsen
their performance on stability tasks with closed eyes compared to
another group of athletes [16].
However, other studies appear to be inconsistent with this
relationship between performance and system complexity. Some
age-related studies have observed a greater complexity of the
system related to worse performances on postural control tasks
[9,10]. High levels of complexity may indicate that the system is
becoming less sustainable. This assumption is close to the
traditional interpretation of variability as a measure of disorder
and noise [10].
The relationship between complexity and performance in
balance tasks has been previously analyzed in dancers. Stin
et al. have observed that young dancers exhibit greater postural
control with greater complexity compared to other groups of

http://dx.doi.org/10.1016/j.gaitpost.2014.06.021
0966-6362/ß 2014 Elsevier B.V. All rights reserved.
 R. Muelas Pérez et al. / Gait & Posture 40 (2014) 556–560
participants without experience in balance skills [7]. However,
Schmit et al. did not observe differences in postural control
between dancers and track athletes [11]. These authors observed
differences in behavioral complexity between both groups with
greater complexity or irregularity in the postural stability of
dancers. Schmit et al. argued that there is a qualitative difference,
rather than quantitative, in the balance task between these two
groups [11].
Because of the controversy in the results, the aim of this study
was to evaluate the relationship between complexity and
performance through a comparison of two groups of different
levels of expertise in postural control (dancers and non-dancers),
and two levels of availability of visual information during a balance
task. The following are the hypotheses of this study: (1) expert
dancers will show greater complexity and a better performance
during the balance task than non-dancers in both visual conditions,
and (2) both groups will show a greater complexity with a better
performance in the balance task when visual information is
available.
2. Method
2.1. Participants
Eighteen undergraduate dancers (all females) from the Spanish
Royal Conservatory of Dance and thirty healthy young women
without any experience in dance participated in the study. All
dancers were specifically trained in contemporary dance and ballet
for a minimum of five years. The remaining thirty women served as
a control group and were not explicitly trained in balance tasks.
The participants signed a written informed consent document
prior to the experimental session. Table 1 shows the descriptive
data, including age, weight and height of the sample. The study
protocol was approved by the ethics committee of Extremadura
University.
2.2. Performance task and apparatus
The performance task used to measure balance ability was a
stabilometer (Model 16020, Lafayette Instrument Inc., Lafayette,
IN), in which the tilt angle, recorded by a SMEG330 electro-
goniometer (1-KHz data collection rate) represented the criterion
measure. The tilt angle was the participant’s error score reflecting
deviation (medio-lateral) from the target horizontal platform
position (08).
The stabilometer platform (0.66 m  1.08 m  0.025 m) was
placed 0.16 m from the frame and 0.22 m from the floor. The range
of the stabilometer was set to 208 from a horizontal position. The
stabilometer task has been shown as a valid and reliable measure of
balance [17].
2.3. Procedure
The participants were asked to stand barefoot on the platform
maintaining stability from the horizontal position for 30 s in two
visual conditions: open eyes (OE) and closed eyes (CE). The order of
the conditions was randomized between the participants.
Table 1
Descriptive and anthropometric data of all participants.
Group Age (years) Weight (kg) Height (cm)
(average  SD) (average  SD) (average  SD)
Dancers 23.32  2.58 65.73  7.96 171.91  7.02
Non-dancers 22.23  1.79 65.94  10.53 169.97  7.56
557
The participants were instructed to adopt a shoulder-width
stance with their arms held at their sides. The participants were
further directed not to speak during the trials. At the beginning of
each trial, the participants assumed the aforementioned stance.
Data collection was initiated after the participants felt comfortable
and ready. The participants were allowed to rest for 3 min between
the conditions.
To ensure that there was no rest of vision during the closed eyes
condition, all participants placed an ocular mask on their face.
2.4. Data analysis and reduction
Data obtained from electrogoniometer was subsampled to
100 Hz. To evaluate the performance in postural stability, the
absolute error of the tilt platform was measured as the average of
the absolute distance (AE) to the horizontal angle of the platform.
In addition, we assessed participant’s balance control trough the
standard deviation (SD) and mean velocity (MV).
Non-linear time series analysis was applied to the angular
displacement of the platform. The complexity of the postural sway
dynamics was calculated by two methods: Sample Entropy
(SampEn) and Permutation Entropy (PE). Higher values of SampEn
thus represent lower repeatability of vectors of length m to that of
m + 1, which marks a lower predictability of future data points and
a greater irregularity within the time series. SampEn was
performed using the following input parameters for the analysis
algorithms: 0.15 for tolerance (r) (in proportion to the SD of the
signal) and 2 for vector length (m). The selection of these values
was based on the procedures suggested by Cavanaugh et al. [18].
We have included PE to reduce the influence of the magnitude of
the time series, and therefore the influence of the tolerance
window parameter. Permutation entropy is independent of the
data magnitude because it measures the entropy of sequences of
ordinal patterns derived from m-dimensional delay embedding
vectors [19]. PE was performed using 5 for vector length (m).
In order to assess the robustness of SampEn and PE method, we
have applied them modifying input parameter: SampEn was
applied on angular displacement signal using different r (0.15, 0.20
and 0.25 in proportion to the SD of the signal) and m (2, 3 and 4). PE
was applied modifying m from 4 to 6. Higher r and m values
increased entropy output. Nevertheless its influence seems similar
in all conditions and does not affect result interpretation.
2.5. Statistical analysis
Normality of the data distribution was evaluated by calculating
asymmetry, kurtosis and the Kolmogorov–Smirnov method with
the Lilliefors correction. All variables were normally distributed. A
mixed-design analysis of variance (ANOVA) was performed to test
the mean differences between the two groups (dancers and
control), vision conditions (repeated measures factors) and
interactions. A post hoc analysis with a Bonferroni adjustment
was used for multiple comparisons. Pearson’s correlation coeffi-
cient tested for correlations between the variables of the present
study. Significance was established at p < 0.05. SPSS 16.0 (SPSS Inc,
Chicago, IL, USA) was used for all statistical procedures.
3. Results
An example of a medio-lateral deviation of the platform tilt angle from the target
horizontal position (08) for dancers and non-dancers in both visual conditions is
shown in Fig. 1. Higher medio-lateral deviations can be observed in the CE condition
compared to OE condition. The data from the dancer performance show lower
deviation than non-dancer mainly in the OE condition.
Table 2 shows the relationship between the AE, SD and MV of the tilt platform
and the vision conditions. All the participants performed significantly better in the
balance task in the OE condition compared to the CE condition. Both groups
[(Fig._1)TD$IG]
558 R. Muelas Pérez et al. / Gait & Posture 40 (2014) 556–560

Fig. 1. An example of the medio-lateral deviation of the platform tilt angle from the target horizontal position (08) for dancers and non-dancers in both visual conditions.

increased their values in AE, SD and MV when standing on the platform in the CE
condition. Regarding the differences between the groups (Table 3), the dancers
showed better postural stability (lower absolute error and standard deviation) than
non-dancers only in the OE condition (p < 0.05). Furthermore, MV was higher in
dancers in both conditions (p < 0.05).
The analysis of the complexity of the tilt platform excursion showed significant
main effects for the visual condition in SampEn and for group in PE (Table 2).
Nevertheless, the interaction effects indicate that the visual conditions affect the
groups differently. Pairwise comparisons showed lower entropy values for dancers
in CE condition compared with OE condition. In addition, dancers exhibited lower
entropy than non-dancers in CE condition. The results indicate that dancers exhibit

Table 2
Mixed ANOVA testing mean differences between the dancers and non-dancers Table 3
(group) and the repeated measures factors of visual condition (vision) and Mean and standard deviations of absolute error along the medio-lateral axis (AE),
interaction. standard deviation (SD), mean velocity (MV), sample entropy (SampEn) and
permutation entropy (PE) for dancers and non-dancers in the two visual conditions.
F1,46 P
N Open eyes Closed eyes
AE Vision 48.187 .001
Group 3.313 .075 AE ND 30 11.45  1.11a 12.57  1.30B
Vision  group 2.256 .140 D 18 10.61  1.13 12.34  1.24B

SD Vision 37.538 .001 SD ND 30 12.85  1.60A 13.74  1.61B

Group 6.758 .013 D 18 11.45  1.02 13.41  0.81B
Vision  group 6.845 .012
MV ND 30 40.82  11.69a 43.94  11.11A
MV Vision 7.793 .008 D 18 33.67  5.86 36.01  3.45b
Group 7.962 .007
Vision  group .120 .731 SampEn ND 30 0.094  0.030 0.082  0.037a
D 18 0.096  0.028 0.058  0.024B
SampEn Vision 19.108 .000
Group 2.288 .137 PE ND 30 0.58  0.12 0.60  0.10A
Vision  group 5.804 .020 D 18 0.54  0.06 0.44  0.07B
a
Different from the dancers using the Bonferroni correction. Pairwise compari-
PE Vision 3.064 .087
son, p < .05.
Group 13.324 .000 A
Different from the dancers using the Bonferroni correction. Pairwise
Vision  group 24.708 .000
comparison, p < .01.
B
AE, absolute error; SD, standard deviation; MV, mean velocity; SampEn, sample Different from the open eyes condition using the Bonferroni correction.
entropy; PE, permutation entropy. Pairwise comparison, p < .01.
 R. Muelas Pérez et al. / Gait & Posture 40 (2014) 556–560
lower complexity, lower entropy values, when visual information is not available
(Table 3).
The results of the correlation analysis performed on the sample of non-dancers
showed that there is a positive relationship between AE (r = .465, p < .05), SD
(r = .504, p < .01) and MV (r = .786, p < .01) in the OE and CE conditions. Thus, the
higher the performance stability test in OE condition, the greater the performance
stability test in CE condition. However, the performance of dancers between the two
stability tests is not significantly correlated. The better performance of the dancers
in the OE condition is not associated with a higher yield in the test with CE.
Comparing the absolute error of the platform angle with the complexity
measured through SampEn, the results in the dancer group showed negative
correlations between the AE and complexity of the tilt angle for both conditions.
This behavior is exhibited in both the OE (r = .605, p  .01) and CE conditions
(r = .711, p  .01). Similar correlations has been observed between SampEn and SD
in OE condition (r = . 483, p < .05) and CE condition (r = .807, p < .01). Thus, as
the complexity increases, the performance improves, regardless of the availability
of visual information. The non-dancers also showed these negative correlations in
both conditions between AE and SampEn (OE: r = .579, p < .01; CE: r = .594,
p < .01), and between SD and SampEn (OE: r = .602, p < .01; CE: r = .648,
p < .01).
4. Discussion
The results are consistent with other studies in which
performance on the stability test in the OE condition was better
than the performance in the CE condition [11,15,20]. Moreover,
and partially confirming the first hypothesis, the group of dancers
performed better only on the balancing task with OE. These results
may be because of the increased specialization of dancers in tasks
in which postural control is regulated through visual feedback
[11,21,22]. Rist states that dancers are used to regulate their
posture through visual information from the initial moment of
learning [23]. Teasdale et al. claim that a specialization in learning
leads to a specialization in the information channel used in
execution [24]. Thus, the greater the visual information used
during the learning process, the greater the decrease in perfor-
mance when vision is removed [25]. This idea was also observed by
Gerbino et al., who observed highly significant reductions in the
balance performance of dancers in the CE condition [26].
Our results show that the deprivation of visual information
matches the performance of both groups in the balance task. These
results confirm the dependence of dancers on visual information to
regulate their posture, as argued by other authors in several studies
[21,27]. Kavounoudias et al. suggested that superficial plantar
mechanoreceptors provide the CNS with information relative to
the position of the body with respect to the vertical reference [28].
Exclusion of the vision would render the CNS rely on vestibular and
proprioception systems. Perrin et al. argued that dance training
strengthens the accuracy of proprioceptive inputs in a lower way
than other sports [27], developing specific modalities of balance
not transferable to posture control in CE situations [21].
In relation to the entropy values, opposed to the hypothesis, the
group of dancers shows lower behavioral complexity than non-
dancers during the tasks only in the CE condition. Indeed,
comparing both visual conditions, only the dancers entropy values
decrease in the CE condition, which is in contrast to the group of
non-dancers, in which no difference was observed. Movement
instabilities, in the form of critical fluctuations, or temporary losses
in stability, are exhibited when a transition from one coordination
pattern to another may be about to occur [29]. As argued
previously, under CE condition dancers exhibit a shift in
sensorimotor dominance from vision to proprioception reflected
in a different coordination pattern. The reduced complexity
observed in the dancer system may reflect an altered coordination
strategy characterized by lower degrees of freedom in an adapting
process to an unfamiliar task [30].
Consistent with the above-mentioned, Schmidt et al. argue that
differences in the postural stability behavior of a dancer compared
to other (physically active) groups are not in their better
559
performance but in their system flexibility to adapt to different
situations of postural control [11]. Thus, the differences would be
more qualitative than quantitative. For this reason, we might have
only observed differences between both groups in their behavioral
complexity during the balance tasks in CE conditions, where
difference in the performance between the groups practically
disappears. Sensory deprivation increases the difficulty of the task,
and the group of dancers changed their behavior, freezing their
degrees of freedom in a reduced effectiveness of their strategy to
stabilize and prevent falls, thus resulting in lower values of entropy
[31].
Regarding the relationship between complexity and perfor-
mance, we have found contradictory results. On the one hand,
correlation analysis in both groups showed that the participants
with higher complexity exhibited higher performance regardless
of the availability of visual information. On the other hand, even
though dancers exhibited better performance than non-dancers in
OE, they there was not differences in complexity between groups.
In addition, the decrease in performance of non-dancers in the CE
condition compared to OE condition was not related to lower
behavioral complexity. In this regard, the relationship between
performance and complexity is not linear or is mediated by other
variables [1]. Some authors suggested that an increase in entropy
variables is not necessarily synonymous with an increase in
physiologic complexity [13] and the relationship between entropy
variables and physiologic complexity is not linear. In this sense,
according with Stergiou and Decker, there would be inverse U-
shape relationship between entropy variables and complexity [32].
In summary, the differences in balance ability between dancers
and non-dancers are only meaningful in the OE condition, thus
indicating that the best performance in the former group is coupled
with the use of information from vision to control balance
adjustments. The deprivation of visual information increases the
difficulty of the task, which is more significant in dancers because
of their specialization in learning through visual information.
Decreases in the complexity of postural control in the dancers
entail a decrease in performance. These results may be observed
because of specific strategies developed to control a novel balance
tasks. The non-linear analysis methods allow us to explain both the
differences observed between the different levels of difficulty in
the balance tasks and those observed between the different levels
of expertise in the participants.
Conflict of interest statement
There are no financial or personal relationships between any of
the authors and other people or organizations that could
inappropriately influence this study.
References
[1] Vaillancourt DE, Newell KM. Changing complexity in human behavior and
physiology through aging and disease. Neurobiol Aging 2002;23:1–11.
[2] Winter DA, Patla AE, Frank JS. Assessment of balance control in humans. Med
Prog Technol 1990;16:31–51.
[3] Palmieri RM, Ingersoll CD, Stone MB, Krause BA. Center-of-pressure param-
eters used in the assessment of postural control. J Sport Rehabil 2003;11:51–
66.
[4] Manor B, Costa MD, Hu K, Newton E, Starobinets O, Kang HG, et al. Physiologi-
cal complexity and system adaptability: evidence from postural control dy-
namics of older adults. J Appl Physiol 2010;109:1786–91.
[5] Lipsitz LA. Dynamics of stability: the physiologic basis of functional health and
frailty. J Gerontol Ser A: Biol Sci Med Sci 2002;57:115–25.
[6] Thurner S, Mittermaier C, Ehrenberger K. Change of complexity patterns in
human posture during aging. Audiol Neurootol 2002;7:240–8.
[7] Stins JF, Michielsen ME, Roerdink M, Beek PJ. Sway regularity reflects atten-
tional involvement in postural control: effects of expertise, vision and cogni-
tion. Gait Posture 2009;30:106–9.
560 R. Muelas Pérez et al. / Gait & Posture 40 (2014) 556–560
[8] Roerdink M, Haart MD, Daffertshofer A, Donker SF, Geurts ACH, Beek PJ.
Dynamical structure of center of pressure trafectories in patients recovering
from stroke. Exp Brain Res 2006;174:256–69.
[9] Duarte M, Sternad D. Complexity of human postural control in young and older
adults during prolonged standing. Exp Brain Res 2008;191:265–76.
[10] Borg FG, Laxåback G. Entropy of balance – some recent results. J Neuroeng
Rehabil 2010;7:38.
[11] Schmit JM, Regis DI, Riley MA. Dynamic patterns of postural sway in ballet
dancers and track athletes. Exp Brain Res 2005;163(3):370–8.
[12] Lamouth CJ, Van Lummel RC, Beek PJ. Athletic skill level is reflected in body
sway: a test case for accelometry in combination with stochastic dynamics.
Gait Posture 2009;29(4):546–51.
[13] Goldberger AL, Peng CK, Lipsitz LA. What is physiologic complexity and how
does it change with aging and disease? Neurobiol Aging 2002;23(1):23–6.
[14] Haran FJ, Keshner EA. Sensory reweighting as a method of balance training for
labyrinthine loss. J Neurol Phys Ther 2008;32:186–91.
[15] Donker SF, Lebdet A, Roerdink M, Savelsberg JP, Beck PJ. Children with cerebral
palsy exhibit greater and more regular postural sway than typically develop-
ing children. Exp Brain Res 2008;184(3):363–70.
[16] Vuillerme N, Teasdale N, Nougier V. The effects of expertise in gimnastics on
propioceptive sensory integration in human subjects. Neurosci Lett
2001;311(2):73–6.
[17] Murray JF. Construction of a stabilometer capable of indicating the variability
of non-level performance. Percept Mot Skills 1982;55:1211–5.
[18] Cavanaugh JT, Mercer VS, Stergiou N. Approximate entropy detects the effect
of a secondary cognitive task on postural control in healthy young adults: a
methodological report. J Neuroeng Rehabil 2007;4:42.
[19] Frank B, Pompe B, Schneider U, Hoyer D. Permutation entropy improves fetal
behavioural state classification based on heart rate analysis from biomagnetic
recordings in near term fetuses. Med Biol Eng Comput 2006;44:179–87.
[20] Rogind H, Simonsen H, Era P, Bliddal H. Comparison of Kistler 9861A. Force
Platform and Chattecx Balance System for measurement of postural
sway: correlation and test–retest reliability. Scand J Med Sci Sport
2003;13(2):106–14.
[21] Hugel F, Cadopi M, Kohler F, Perrin P. Postural control of ballet dancers: a
specific use of visual input for atistic purposes. Int J Sports Med
1999;20(2):86–92.
[22] Schanfein L, Rietdyk S. The relationship between center of pressure displace-
ment and estimated instability of dancers and non-dancers while in a moving
room. In: Proceedings for the North American congress on biomechanics. MI:
Ann Arbor; 2008, August.
[23] Rist R. Dance science. The Dancing Times 1991, December;243.
[24] Teasdale N, Stelmach GE, Breunig A. Postural sway characteristics of the
elderly under normal and altered visual and support surface conditions. J
Gerontol 1991;46B:238–44.
[25] Proteau L. On the specificity of learning and the role of visual information in
movement control. In: Proteau L, Elliot D, editors. Vision and Motor Control,
North Holland: Amsterdam; 1992. p. 67–103.
[26] Gerbino PG, Griffin ED, Zurakowski D. Comparison of standing balance be-
tween female collegiate dancers and soccer players. Gait Posture 2007;26(4):
501–7.
[27] Perrin P, Deviterne D, Hugel F, Perrot C. Judo, better than dance, develops
sensorimotor adaptabilities involved in balance control. Gait Posture 2002;15:
187–94.
[28] Kavounoudias A, Roll R, Roll JP. The plantar sole is a ‘‘dynamometric map’’ for
human balance control. NeuroReport 1998;9:3247–52.
[29] Schöner G, Haken H, Kelso JAS. A stochastic theory of phase transition in
human hand movement. Biol Cybern 1987;53:247–57.
[30] Davids K, Button C, Bennet S. Dinamical skill acquisition. A constraints-led
approach. Champaign: Human Kinetics; 2007.
[31] Amoud H, Snoussi H, Hewson DJ, Duchêne J. Intrinsic Mode Entropy for
postural steadiness analysis. In: Proceedings for the 4th European conference
of the international federation for medical and biological engineering, vol. 22;
2008. p. 212–5.
[32] Stergiou N, Decker LM. Human movement variability, nonlinear dynamics, and
pathology: is there a connection? Hum Mov Sci 2011;30(5):869–88.