J Neurophysiol 93: 189 200, 2005.
First published August 25, 2004; doi:10.1152/jn.00221.2004.
A New Interpretation of Spontaneous Sway Measures Based on a Simple
Model of Human Postural Control
Christoph Maurer1,2 and Robert J. Peterka1
1
Neurological Sciences Institute, Oregon Health & Science University, Portland, Oregon; and 2Neurological University Clinic,
Freiburg, Germany
Submitted 4 March 2004; accepted in final form 18 August 2004
Maurer, Christoph and Robert J. Peterka. A new interpretation of
spontaneous sway measures based on a simple model of human
postural control. J Neurophysiol 93: 189 200, 2005. First published
August 25, 2004; doi:10.1152/jn.00221.2004. This study shows that
center-of-pressure (COP) traces that closely resemble physiologically
measured COP functions can be produced by an appropriate selection
of model parameters in a simple feedback model of the human
postural control system. Variations in the values of stiffness, damping,
time delay, and noise level determine the values of 15 sway measures
commonly used to characterize spontaneous sway. Results from
model simulations indicate that there is a high degree of correlation
among these sway measures, and the measures cluster into three
different groups. Only two principal components accounted for about
92% of the variation among the different sway measures analyzed.
This model can be used to formulate hypotheses regarding the cause
of postural control deficits reported in the literature. This is accom-
plished using a multidimensional optimization procedure to estimate
model parameters from a diverse set of spontaneous sway measures.
These model parameters describe physiologically meaningful features
of the postural control system as opposed to conventional sway
measures that provide only a parametric description of sway. To show
the application of this method, we applied it to published data of
spontaneous sway from elderly subjects and contrasted it to the data
of young healthy subjects. We found that modest increases in stiffness
and damping and a fairly large increase in noise level with aging could
account for the variety of sway measures reported in the literature for
elderly subjects.
INTRODUCTION
Upright stance is inherently unstable. Small deviations from
an upright body position result in a gravity-induced torque
acting on the body, causing it to accelerate further away from
the upright position. Corrective torque must be generated to
counter the destabilizing torque due to gravity. This process of
continuous small body deviations countered by corrective
torques creates a pattern known as spontaneous sway. The
mechanisms underlying spontaneous sway are not fully under-
stood, and controversy remains regarding the organization of
sensory and motor systems contributing to spontaneous sway.
Numerous authors have suggested that active feedback con-
trol mechanisms contribute to the maintenance of upright
stance (e.g., Fitzpatrick et al. 1996; Johansson and Magnusson
1991; Peterka and Benolken 1995; Peterka and Loughlin 2004;
van der Kooij et al. 1999). However, the relevance of feedback
mechanisms for postural control is still debated. Some authors
concluded from their experiments that corrective torque orig-
Address for reprint requests and other correspondence: C. Maurer, Neurolo-
gische Universitatsklinik, Neurozentrum, Breisacher Str. 64, 79106 Freiburg,
Germany (E-mail: maurer@nz11.ukl.uni-freiburg.de).
www.jn.org 0022-3077/05 $8.00 Copyright 2005 The American Physiological Society
inating from feedback control is insufficient for stabilizing the
body (Fitzpatrick et al. 1996). Others suggested additional
sources for corrective torque, like prediction (Morasso et al.
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017
1999; van der Kooij et al. 1999), or have proposed more
complex concepts like throw and catch movements (Loram and
Lakie 2002a), anticipatory saccadic control (Baratto et al.
2002), or open-loop and closed-loop control (Collins and De
Luca 1993).
However, recent studies have shown that a model based
primarily on a feedback mechanism with a 150- to 200-ms time
delay can account for postural control during a broad variety of
perturbations (Peterka 1995, 2002; Peterka and Benolken
1995; Peterka and Loughlin 2004) and can yield a spontaneous
sway pattern that resembles normal (Peterka 2000) or patho-
logical spontaneous sway (Parkinsons disease; Maurer et al.
2003). The simple feedback model we aim to investigate here
is consistent with models previously used to simulate upright
stance behavior in humans (Ishida and Imai 1980; Ishida and
Miyazaki 1987; Johansson et al. 1988; Peterka 1995, 2000,
2002; Peterka and Benolken 1995; Peterka and Loughlin
2004). In this model, the body is treated as a single-link
inverted pendulum. The angular deviation from upright stance
is detected by sensory systems, and this sensory information is
used to generate corrective torque that compensates for distur-
bances. Johansson et al. (1988) showed that stabilization of an
inverted pendulum requires two components of corrective
torque: one component proportional to the angular deviation
and the other proportional to the time derivative of the angular
deviation. When adding a third component proportional to the
mathematical integral of the angular deviation signal, Johans-
sons experimental data were even better explained. This
three-component controller is commonly referred to as a PID
controller (for proportional, integral, and derivative controller).
By including an internal random disturbance torque, this sim-
ple model generates sway patterns that resembles spontaneous
sway. Sway measures calculated from this model-generated
spontaneous sway can be compared with experimental sway
measures to determine the extent to which the model accounts
for experimental data.
Spontaneous body sway is most often described with mea-
sures based on movements of the center-of-pressure (COP)
measured with a force platform. In general, we can distinguish
time domain from frequency domain measures (Prieto et al.
1996). Time domain measures estimate a parameter associated
with either the displacement or the velocity of the COP trace.
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked advertisement
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
189
190 C. MAURER AND R. J. PETERKA
Frequency domain measures characterize the area or shape of
the power spectral density of the COP trace. In the past, many
studies characterized postural stability based on a single dis-
placement- or velocity-related measure (Baloh et al. 1998;
Brocklehurst et al. 1982; Era and Heikkinen 1985; Kolleger et
al. 1992; Overstall et al. 1977). Some studies included multiple
measures (Chiari et al. 2002; Laughton et al. 2003; Maki et al.
1990; Murray et al. 1975). Depending on the cause of the
postural instability, velocity-related sway measures were often
reported to separate stable postural control from reduced sta-
bility better than displacement-related sway measures. This
was shown for Parkinsons disease (Burleigh et al. 1995;
Maurer et al. 2003; Rocchi et al. 2002), peripheral neuropathy
(Dickstein et al. 2001; Horak et al. 2002; Uccioli et al. 1995),
and postural instability in elderly adults (Maki et al. 1990;
Prieto et al. 1996).
One of the more recent methods for analyzing COP time
series is the determination of a stabilogram diffusion function
(SDF). The SDF summarizes the mean square COP displace-
ment as a function of the time interval between COP compar-
isons (Collins and De Luca 1993). A typical SDF shows that
spontaneous sway is characterized by a two-part behavior,
which led the authors to assume that, over short time intervals,
the postural system is not controlled (i.e., it operates open-
loop), while at longer time intervals, there is active feedback
control (i.e., closed-loop control). However, Peterka (2000)
showed that a simple feedback mechanism combined with a
feedback time delay could generate realistic SDFs.
Only a few studies have compared a large number of
COP-based measures with the explicit aim of finding the most
significant measure to distinguish between different groups and
conditions (Baratto et al. 2002; Hufschmidt et al. 1980; Prieto
et al. 1996; Rocchi et al. 2004). Using measures of sway path
length, mean sway amplitude, mean sway frequency, and sway
area, Hufschmidt et al. (1980) reported differences of equally
high significance for both sway path and mean sway amplitude
when examining patients with cerebellar lesions and patients
with labyrinthine lesions. From a large variety of time and
frequency domain measures, Prieto et al. (1996) isolated mean
sway velocity as the one measure with the highest significance
for separating age groups as well as visual conditions (eyes-
closed/eyes-open). Based on principal component analysis,
Rocchi et al. (2004) provided a recommended set of sway
measures for characterizing anterior-posterior and medio-lat-
eral sway.
J Neurophysiol VOL
It is currently unknown why one sway measure or another
should be superior in distinguishing between normal and
pathological conditions. The statistical relationships between
alternative sway measures and between these measures and the
underlying dynamic properties of the postural control system
are also unknown. This study was undertaken to gain insight
into these relationships. Spontaneous body sway traces were
created using a simple postural control model that was previ-
ously shown to generate realistic SDFs (Peterka 2000) and was
shown to account for responses to various external stimuli
(Peterka 2002). In this study, we 1) show that numerous other
COP-related measures of spontaneous sway reported in the
literature for normal subjects are also compatible with this
model, 2) characterize the relationship between the sway mea-
sures and the sensitivity of the sway measures to the model
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017
parameters, 3) show that model parameters representing fun-
damental characteristics of the postural control system can be
derived from a set of spontaneous sway measures using an
optimization procedure, 4) apply this optimization procedure to
existing sets of spontaneous sway measures in elderly and
young subjects to determine how fundamental postural control
parameters differ between young and elderly subjects, and 5)
investigate how variations in the model architecture (active vs.
passive control models) influence the quality of the model
parameter estimates.
METHODS
Model description
We used a simple model of human postural control (Fig. 1) with all
inputs and outputs restricted to the sagittal plane (anterior-posterior
direction). We assumed that a neural control system senses a deviation of
the body away from an upright reference position and sends commands
to generate a corrective torque to resist the deviation of body position
away from upright. An inverted pendulum represents the human body
(Johansson et al. 1988; Peterka 2000) with parameters set to values
typical for an average adult male subject. Specifically, JB 66 kg/m2 is
the moment of inertia of the body about the ankle joint axis, mB 76 kg
is body mass excluding the feet, mF 2.01 kg is the mass of the feet,
dB 0.87 m is the height of the body center of mass (COM) above the
ankle joint axis, hF 0.085 m is the height of the ankle joint axis above
the ground, and dF 0.052 m is the horizontal distance between the
ankle joint axis and the COM of the feet.
The input to the body is the torque exerted about an axis through
both ankle joints. The torque consists of two components. One (Td, see
Fig. 1) is a random disturbance torque that contributes to the gener-
FIG. 1. Feedback control system model de-
scribing the control of human upright stance. JB,
moment of inertia of the body; mB, body mass
excluding the feet; mF, mass of the feet; dB,
height of the body center of mass above the ankle
joint; hF, height of the ankle joint above the
ground; dF, horizontal distance between the ankle
joint and center of mass of the feet; g, accelera-
tion due to gravity; Td, disturbance torque; Tc,
corrective torque; Ta, active torque; Tp, passive
torque; N, low-pass filter time constant; KP, gain
of the proportional part, KD, gain of the deriva-
tive part, and KI, gain of the integrative part of
neural controller; d, feedback time delay; KN,
noise level gain; Kpas, passive stiffness; Bpas,
passive damping; , body sway angle; xCOP, cen-
ter-of-pressure (COP) displacement; s, Laplace
transform variable.
93 JANUARY 2005 www.jn.org
 A NEW INTERPRETATION OF SPONTANEOUS SWAY 191

ation of spontaneous body sway patterns. The other is a control torque
(Tc) that corrects for the disturbance torque and the torque due to
gravity. For most simulations, Tc was generated entirely by the action
of a neural controller, i.e., Tc Ta in Fig. 1. The input to the neural
controller is the sensed body sway with a time delay. We assume that
the sensory systems provide an accurate measure of body sway
(sensory systems 1 in Fig. 1). The neural controller properties are
specified by the three constants KP, KD, and KI, which scale the
components of the control torque that are proportional to the deviation
from zero of the angular position, angular velocity, and time integral
of the angular deviation, respectively. For a subset of simulations, we
assumed a more complex model structure that included passive
stiffness (Kpas) and passive damping (Bpas) components. In that case,
Tc is a sum of Ta and the passive torque (Tp) determined by the
intrinsic ankle stiffness and damping. The outputs of the system are
body sway angle () and COP displacement (xCOP). The computation
The resulting center-normal model parameters were similar to the
ones used by Peterka (2000). The gain factors for the PID controller
were KP 16.7 Nm deg1 for the proportional part (stiffness), KD
4.83 Nm s deg1 for the derivative part (damping), and KI 0.60
Nm s1 deg1 for the integrative part. The feedback time delay (d)
was 0.171 s. The noise level gain (KN) was 462 Nm, which corre-
sponded to a standard deviation of the disturbance torque of 10.7 Nm.
A 30-s sample of an 800-s COP time series generated from the Fig. 1
model with the center-normal model parameters is shown in Fig. 2A.
The sample in Fig. 2A includes the body COP displacement, xCOP, and
COM displacement, xCOM dB sin().
Each of the five model parameters (KP, KD, KI, d, KN) were varied
in nine equal increments between 90 and 110% of their center-normal
values, and a simulation was performed for each possible combination
of the nine values of the five parameters, leading to 95 59,049
simulations.

Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017

of xCOP is based on a derivation by Barin (1989), which relates the
sagittal body displacement (xB) and the sagittal, vertical, and angular
body accelerations (xB, yB, B, respectively) to xCOP for an inverted
pendulum
(mBd B2 JB)B mB xB(g yB) mByBxB mBhFxB mFdFg
x COP
mB(g yB) mFg
where g 9.81 m/s2 is the acceleration due to gravity.
To simulate the continuous, small deviations of body sway around
an equilibrium position (Prieto et al. 1996), a disturbance torque (Td)
is added to the corrective torque (Tc). We are aware of the fact that a
major contribution to the overall noise in human postural control
might stem from the sensory organs (van der Kooij et al. 2001). Thus
it may be plausible to add noise to the sensory signal, too. However,
we tested the model with different sites of noise injection, and the
simulation results were very similar. For simplicity, we chose to use
only one noise source, which we inject as a random disturbance
torque. The disturbance torque is a low-pass filtered Gaussian noise
source with a first-order low-pass filter time constant of N 100 s for
all of the simulations presented in this paper. The use of this low-pass
filter led to a noise source with characteristics of a random walk
process in velocity whose spectral density declines with increasing
frequency for frequencies above the filter cutoff frequency (Milotti
2002). The resulting waveforms created sway characteristics similar
to those seen physiologically. In addition, the resulting sway measures
Sway measure calculations
Fifteen sway measures (Table 1) were calculated from each of the
800-s COP traces resulting from the 59,049 simulations. In the
(1)
following equations for calculation of the sway measures, n 80,000
is the number of data points included in the analysis, and T 800 s
is the period of time selected for analysis. To simplify the following
definitions, we referenced the COP time series to the mean, i.e., we
subtracted the mean COP position from the COP time series.
TIME-DOMAIN MEASURES. The measures described in this section
are the most commonly used measures of postural stability. We
adapted these measures from Prieto et al. (1996). Time domain
measures provide measures associated with either the displacement or
the velocity of the COP. The mean distance (MD) is the mean of the
absolute value of the COP time series and represents the average
distance of the COP from its mean
n
1
MD xCOP(i) (2)
n
i1
The root mean square distance (RMS) from the mean COP is equiv-
alent to the SD (SD) of the zero-meaned COP time series

were largely insensitive to the exact value of the filter time constant.
A higher value for the time-constant could be compensated for by a n
1
higher noise gain. RMS SD xCOP (i)2 (3)
n
i1

Model simulations The maximum distance (MAXD) is the peak-to-peak range of COP
values
Simulations were performed using Simulink version 5.0 of Matlab
6.5 (The MathWorks, Natick, MA). As a noise source, we used the MAXD maxxCOP minxCOP (4)
standard Matlab block band-limited white noise with zero mean,
unity variance, and with a correlation time of 0.01 s. The simulation The velocity of the COP time series was calculated by subtracting
duration was 800 s, and the period between model outputs was 0.01 s. consecutive positions of the COP path and multiplying by the sam-
The Dormand-Prince algorithm (ode5) with a fixed step size of 0.01 s pling rate (n/T)
was used for all simulations to solve the differential equations asso-
n
ciated with the model. xCOP(i) xCOPi 1 xCOPi (5)
Based on published sway data for young adults and using an T
optimization procedure to be detailed later, a center-normal set of The mean velocity (MV) is the average of the absolute value of the
model parameters was defined that provided a reference for investi- COP velocity
gating how variations in model parameters influenced sway measures.

Coefficients of variation (CVs) for sway measures were obtained from n1
the results of 100 repeated simulations, with each simulation using the 1
MV xCOP(i) (6)
same center-normal model parameters but a different random seed for n1
i1
the band-limited noise time series. CVs of the model parameters were
obtained by performing optimization procedures using the same 100 The root mean square velocity (RMSV) is the standard deviation of the
simulations with randomly varied noise seeds. COP velocity time series

J Neurophysiol VOL 93 JANUARY 2005 www.jn.org

192 C. MAURER AND R. J. PETERKA

FIG. 2. A: simulation results showing example time series

for the anterior-posterior COP displacement (thin line) and

Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017

center-of-mass (COM) displacement (bold line). Neural con-
troller parameters for this simulation were set to the center-
normal values of KP 16.7 Nm deg1, KD 4.83 Nm s
deg1, and KI 0.60 Nm s1 deg1. Feedback time delay
was set to d 0.171 s. Noise level gain (KN) was 462 Nm. B:
center-normal stabilogram diffusion function (SDF) calculated
from simulated COP data. Dashed lines show linear regression
fits used to estimate diffusion coefficients, DS and DL, and
critical point parameters, X2c and Tc.

RMSV 1
n1

n1

xCOP i 2
originally selected for their ability to characterize the area or shape
of the power spectral density of the COP, G( f ). We calculated the
one-sided spectral density function G( f ) via discrete Fourier trans-

i1
forms following the methods given by Bendat and Piersol (2000).

The spectral moments, k, which are needed to compute the

n1
n2
xCOP i 1 xCOP i2 (7) different frequency domain measures, were calculated for k 0, 1,
T n 1
2
i1 and 2 as

The mean frequency (MFREQ) is a frequency, in Hertz, of a sinusoi- m
dal oscillation derived from the mean distance MD and the mean k i f k Gi f (9)
velocity MV1 i1

MV where f 0.00125 Hz is the frequency increment and m 8,000 is

MFREQ
4 2 MD
FREQUENCY-DOMAIN MEASURES. The frequency domain measures
described here are adapted from Prieto et al. (1996). They were
1
Note that this formula, used by Prieto et al. (1996) for calculating the
one-dimensional MFREQ (i.e., the frequency of a sinusoid that has values of
MD and MV for its mean distance and mean velocity measures), was obtained
from Hufschmidt et al. (1980). The correct formula should read MFREQ
(8)
the total number of calculated discrete power spectral density esti-
mates, G(i f ), with frequencies ranging from f 1 f 0.00125
Hz to f 8,000 f 10 Hz. This frequency range covers virtually
the entire bandwidth of the COP signal. Specifically, Dichgans et al.
(1976) showed that 1% of the total power is located above 10 Hz.
The total power (POWER) is the integrated area of the power
spectrum
m n

MV/(2 MD) which means that MFREQ used here is smaller than the true POWER 0 Gi f 2
x COP(i) (10)
MFREQ by a factor of 4 2/(2 ) 0.90032. Because we compared our i1 i1
simulated sway measures to the values given by Prieto et al. (1996), we used
Eq. 8 for the calculation of MFREQ. However, the use of the incorrect formula Theoretically, if all power is accounted for, this is the mean square
did not affect the relationships found involving MFREQ, since correlations are value of the time series. This makes POWER redundant to RMS,
unaffected by a constant multiplier. which is the square root of the mean square value. Therefore to avoid
J Neurophysiol VOL 93 JANUARY 2005 www.jn.org
 A NEW INTERPRETATION OF SPONTANEOUS SWAY 193

TABLE 1. Glossary of sway measures and model parameters

Symbol Description Unit

FREQD 1
21
0 2
(14)

The frequency dispersion is zero for a pure sinusoid and increases

Sway measures with spectral bandwidth to a maximum of one.
Tc Critical point coordinate from the s For most simulation results reported in our study, the calculations
SDF (time)
X2c Critical point coordinate from the mm2
of the frequency domain measures were as described above. However,
SDF (amplitude) for the purpose of comparing our simulated sway measures to exper-
DS Short-term diffusion coefficient mm2/s imental values given by Prieto et al. (1996), we replicated the
from the SDF calculation methods used by those authors. Specifically, the total
DL Long-term diffusion coefficient mm2/s simulation time of 800 s was divided into 40 periods of 20 s, which
from the SDF were comparable to the time periods analyzed by Prieto et al. (1996).
RMS Root mean square distance mm Each 20-s COP trace was filtered with a first-order low-pass filter with
RMSV Root mean square velocity mm/s a cut-off frequency of 5 Hz. The spectral density functions were
MV Mean velocity mm/s calculated from 0.05 to 5 Hz, but the lowest two frequency compo-
MAXD Maximum distance (peak-to-peak) mm
MD Mean distance mm
nents at 0.05 and 0.1 Hz were not included in the calculation of the
frequency domain measures. Then, the 40 spectral density functions

Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017

MFREQ Mean frequency Hz
POWER Total power (integrated area of the mm2 were averaged, and the sway measures were calculated. The affected
power spectrum) sway measures were POWER, P50, P95, CFREQ, and FREQD in
P50 50% power frequency Hz Table 2 and Table 6.
P95 95% power frequency Hz
CFREQ Centroidal frequency Hz STABILOGRAM DIFFUSION FUNCTION MEASURES. SDFs were calcu-
FREQD Frequency dispersion lated following the methodology of Collins and De Luca (1993)
Model parameters
KI Integrative gain of the neural Nm s1 deg1

nm
1
controller x COP
2
m xCOPi m xCOP(i]2 (15)
nm
KP Proportional gain of the neural Nm deg1 i1
controller (active stiffness)
KD Derivative gain of the neural Nm s deg1 where symbolizes a calculation of the mean value, and m ranged
controller (active damping) from 0 to 1,000 corresponding to a time shift of 0 to 10 s between the
d Time delay of the neural controller s COP traces. At m 0, the SDF value is always zero. The short-term
KN Noise level gain Nm diffusion coefficient, DS, the long-term diffusion coefficient DL, and
Kpas Passive stiffness Nm deg1 the critical point coordinates, Tc and X2c, were computed from
Bpas Passive damping Nm s deg1 the results of two linear regressions to the short- and long-term
regions of the SDFs (see Fig. 2B). In most cases, a linear fit to the
Note that the parameters KI, KP, KD, Kpas, and Bpas must be expressed in
radian units rather than degrees when used in model simulations. short-term region of the SDF included data ranging from 0 to 0.5 s,
and the linear fit to the long-term region included data ranging from
the redundancy, we did not use POWER in some of the analysis 2 to 10 s. DS and DL correspond to one-half times the slope of the
procedures (correlation study and optimization procedures). However, respective linear fits to the SDFs. The critical point values were
we show simulated values for POWER because experimental mea- determined from the point of intersection of the linear regression fits
sures of POWER were available for young and elderly adults (Prieto to the short-term and long-term regions. In some cases, as the SDFs
et al. 1996). changed as a function of simulation parameters, the time periods
The 50% power frequency (P50), the median power frequency or included in the short-term and long-term regions were altered slightly
the frequency below which 50% of the total power is found, is uf, to maintain good quality linear fits to the data. An SDF calculated
where u is the smallest integer for which using the full 800-s COP time series from a simulation with center-
normal model parameters is shown in Fig. 2B. The SDF shows a
two-part functional form typical of experimental SDFs (Collins and

u
De Luca 1993).
Gi f 0.50 0 (11)
i1
Statistical procedures
The 95% power frequency (P95), the frequency below which 95% of
the total power is found, is vf, where v is the smallest integer for The statistical procedures were conducted with JMP 5.0.1a (SAS
which Institute, Cary, NC). Using results from the 59,049 simulations, we
calculated a standard correlation matrix, which gave a matrix of

v correlation coefficients that summarizes the strength of the linear
Gi f 0.95 0 (12) relationships between each pair of sway measures. Principal compo-
i1
nent analysis was applied to the correlation matrix. Principal compo-
nent analysis decomposes the correlation matrix by calculating a
The centroidal frequency (CFREQ), the frequency at which the linear combination of the original sway measures such that the first
spectral mass is concentrated, is the square root of the ratio of the principal component accounts for the maximum normalized variation
second to the zeroth spectral moment among the sway measures, the second principal component accounts
for the maximum variation not accounted for by the first component,
CFREQ 2
0
(13)
and so on. The principal components are given by the eigenvectors of
the correlation matrix, and the eigenvalue of each principal compo-
nent indicates the amount of variation accounted for by that principal
The frequency dispersion (FREQD), a unitless measure of the vari- component (Jolliffe 1986). All principal components with eigenval-
ability in the frequency content of the power spectral density, is given ues 1 are reported. A factor rotation process (varimax method) was
by applied to the principal components to identify factors that provide

J Neurophysiol VOL 93 JANUARY 2005 www.jn.org

194 C. MAURER AND R. J. PETERKA
insight into the relationships among the sway measures (Jackson
1991). The factor rotation process attempts to associate each sway
measure with a minimal number of factors.
Relationships between model parameters and sway measures
To understand why certain sway measures seem to be better than
others in distinguishing between normal and altered conditions, we
determined how the different sway measures changed as a function of
the model parameters KI, KP, KD, d, and KN. That is, we computed
the correlations between the changes in the parameter values and the
changes in the sway measures using all 59,049 simulations. In addi-
tion, we performed a sensitivity analysis by varying one parameter at
a time and keeping all other parameters at their center normal values.
Nine simulations were performed for each of the five parameters, with
the parameter varying from 90 to 110% in 2.5% increments. Sway
measures were calculated from each simulation. The normalized
change in the sway measure was compared with the normalized
change in parameter value.
Estimation of model parameters from sway measures
Another set of simulations was performed to investigate whether
the information contained in the different sway measures is sufficient
to determine the values of different model parameters. Preliminary
results showed that variations in gain of the integrative part of the
neural controller did not influence the sway measures to a large
degree. Therefore we kept the gain of the integrative part at KI 0.6
Nm s1 deg1 and varied the remaining parameters (KP, KD, d, and
KN) simultaneously and independently such that each parameter
assumed one of three possible values: 90, 100, and 110% of its
center-normal value. This procedure led to a total of 34 81 different
simulations, each with a different set of model parameter values.
Sway measures were calculated from the COP trace from each of
these simulations. The sway measures from each of these simulations
were used in an optimization procedure to obtain estimates of the
model parameters that could optimally account for each set of sway
measures. A comparison of the actual model parameters with the
estimated model parameters provided information about the ability of
this optimization procedure to accurately identify fundamental pos-
tural control parameters from a set of COP sway measures.
We used the Matlab Optimization toolbox function fminsearch to
find the minimum of a scalar error function of the sway measures,
starting at an initial estimate. The algorithm fminsearch uses the
simplex search method of Nelder-Mead (Lagarias et al. 1998). We
used the center-normal model parameters as the initial estimate. With
each iteration of the optimization procedure, the current model pa-
rameters were used in a simulation to generate a COP time series.
Sway measures were calculated from the COP time series, and a scalar
error function was evaluated. Then the optimization procedure made
changes to the model parameters, a new COP time series was
simulated, sway measures were calculated, and the error function was
re-evaluated. This sequence was repeated until the error function was
minimized. To normalize the error contribution from each sway
measure and to achieve a symmetrical error distribution around the
reference value, we used the error function
N
Mi Mi
E (16)
Mi Mi
i1
where n 14 is the number of sway measures, Mi are the reference
sway measures obtained using 1 of the 81 sets of model parameters,
and vMi is the estimated sway measure obtained with each iteration of
the optimization procedure.
We also applied this optimization procedure to the means of
experimental sway measures of young and elderly adults from previ-
J Neurophysiol VOL
ous studies (Collins and De Luca 1993; Collins et al. 1995; Prieto et
al. 1996). All experimental sway measures we used were originally
based on COP traces derived from force platform recordings during
20 30 s of quiet standing. Prieto et al. (1996) compared a variety of
time and frequency domain measures between a group of 20 healthy
young adults (2135 yr) and a group of 20 healthy elderly adults
(66 70 yr). Collins and De Luca (1993) described SDF-derived sway
measures from 25 healthy male subjects (19 27 yr). Collins et al.
(1995) presented SDF measures from 25 healthy elderly males (71 80
yr).
In a final set of optimizations, we considered alternative model
structures that included passive elements that contributed to the
corrective torque without time delay (Tp in Fig. 1). We tested three
different model versions that included passive dynamic contributions.
In all versions, the model parameters KP, KD, d, and KN were always
allowed to vary during the optimization. The passive dynamic ele-
ments Kpas and Bpas were either fixed or were allowed to vary. In one
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017
model version, the passive stiffness Kpas was set to 10.2 Nm deg1
(i.e., 90% of the stiffness necessary to provide minimal stabilization,
following the experimental results of Loram and Lakie 2002b), and
the passive damping Bpas was set to 0. In the second version, Kpas
was set to 10.2 Nm deg1, and the passive damping Bpas was set to
3.0 Nm s deg1 (i.e., values that were in equal proportion to the
active stiffness and damping, respectively, implemented in the center
normal model). The optimization procedure was used to identify the
other model parameters (KP, KD, d, and KN) that minimize the Eq. 16
error criterion. In the third version, Kpas and Bpas were included as free
parameters in the optimization.
RESULTS
We first show that the Fig. 1 model predicts sway measures
that agree with experimental data of young adults. Then we
present the relationships among sway measures predicted by
the model and the sensitivity of sway measures to variations of
model parameters. After verifying that our optimization pro-
cedure is able to reliably identify model parameters, we apply
the optimization procedure to experimental data from elderly
adults and investigate other model structures that include
passive stiffness and/or damping.
Center-normal sway measures
Table 2 (columns active control and experimental data)
shows that all sway measures derived from simulations with
the center-normal model parameters were well within the 1 SD
range of the mean experimental data taken from studies of
(Collins and De Luca 1993; their Table 5) and the results of
young healthy adults from Prieto et al. (1996). The CVs of the
sway measures were all 0.05 for all sway measures, indicat-
ing that the simulations yielded reliable estimates of the sway
measures. The CVs of the model parameters were 0.06,
indicating that model parameters obtained from the optimiza-
tion procedure were minimally affected by the particular noise
time series used in the optimization.
Correlations among sway measures
All sway measures aggregated into primarily three different
measurement groups, with high correlations within each group,
and in most cases, lower correlations between groups (Table 3;
Fig. 3A). The first group of the aggregated sway measures
contained the displacement-related measures (MAXD, RMS,
MD) and two measures from the SDF analysis (DL and X2c).
93 JANUARY 2005 www.jn.org
 A NEW INTERPRETATION OF SPONTANEOUS SWAY 195

TABLE 2. Simulations and experimental results (mean and SD

range) for young adults using 3 different model structures: 1) pure
active control, presented together with the CV resulting from
variations of the noise source, 2) passive stiffness, and 3) passive
stiffness and damping

Experimental Data
Passive
Active Passive Stiff. SD
Control CV Stiffness Damp. Mean Range

Sway
measures
Tc 0.96 0.045 0.94 0.97 0.97 0.511.43* FIG. 3. A: schematic representation of inter-relationships among sway mea-
X2c 12.9 0.047 12.7 12.7 12.69 1.9823.4* sures obtained from model simulations. Solid lines connecting different sway
DS 7.27 0.030 7.26 7.11 7.27 3.1611.4* measures indicate a correlation coefficient whose absolute value was 0.90
DL 1.56 0.034 1.51 1.39 2.51 0.754.27* between the measures. Dashed lines indicate a correlation coefficient whose

Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017

RMS 5.23 0.044 5.42 5.77 3.82 2.285.36 absolute value was 0.80, but not 0.90. Dividers labeled with 1, 2, and 3
MV 6.63 0.007 7.01 7.78 6.72 4.548.90 represent the aggregated groups of sway measures. B: sway measures are
MAXD 23.0 0.047 23.7 24.6 17.7 10.724.7 presented in the same order as in A, together with a representation of the
MD 3.40 0.043 3.54 3.80 3.1 1.814.39 absolute value of the vector components of factor 1 (white bars) and factor 2
MFREQ 0.30 0.044 0.30 0.31 0.42 0.270.57 (black bars) obtained from factor analysis.
POWER 10.7 0.039 10.9 11.2 16.8 5.0028.6
P50 0.35 0.024 0.35 0.35 0.3 0.210.39 Although the selection of correlation thresholds of 0.9 and 0.8
P95 1.0 0.031 1.05 1.0 0.98 0.671.30 for display in Fig. 3A is arbitrary, the results shown in Fig. 3A
CFREQ 0.54 0.025 0.55 0.54 0.54 0.400.67 provide a basis for understanding the inter-relationships among
FREQD 0.68 0.012 0.72 0.73 0.66 0.620.70 the sway measures predicted by the model shown in Fig. 1.
Model fit error 0.466 0.013 0.505 0.553
Model Principal component and factor analysis
parameters
KP 16.7 0.023 7.52 10.0 About 92% of the variability among all the different sway
KD 4.83 0.040 3.13 0.34 measures was explained by the first two principal components
d 0.171 0.054 0.166 0.177 (Table 4, columns 1 and 2). The third principal component
KN 462 0.035 522 653
Kpas 10.2 10.2
accounted for only 6% additional variability among the sway
Bpas 3.0 measures (eigenvalue 0.86).
Factor rotation led to one factor (rotated factor 1 in Table 4)
Numbers marked with a indicate outliers with respect to the SD range that contributed strongly to the displacement-related sway
of experimental data given by *Collins and De Luca (1993) and Prieto et al. measures (MAXD, RMS, MD) and by the SDF measures DL
(1996). The CVs (SD/mean) indicate the relative variability of the results from
simulations and optimizations of the sway measures and model parameters. and X2c (group 1 in Fig. 3 and Table 4). The second factor
Units of sway measures and model parameters are given in Table 1. (rotated factor 2 in Table 4) contributed strongly to the veloc-
ity-related sway measures (MV, RMSV) and by DS, FREQD,
The second group contained the velocity-related measures and P50 (group 2 in Fig. 3 and Table 4). Both factors contrib-
(MV, RMSV), as well as some frequency-related measures uted to the frequency-related sway measures P95, CFREQ, and
(FREQD, P50) and DS from the SDF. The third group con- MFREQ, and the SDF measure Tc (group 3 in Fig. 3 and
tained measures related to the frequency distribution of sway Table 4). The absolute values of factor loadings corresponding
(P95, CFREQ, MFREQ) and the SDF measure Tc. Sway to the 14 sway measures are shown in Fig. 3B. Comparison of
measure DL was an exception, in that DL showed a high Fig. 3B with the correlation results in Fig. 3A shows a similar
correlation with nearly all sway measures in groups 1 and 3. grouping of the sway measures.
TABLE 3. Correlations among sway measures derived from model simulations

DL X2c MD RMS MAXD P50 FREQD DS MV RMSV MFREQ CFREQ Tc

X c
2
0.93
MD 0.97 0.95
RMS 0.95 0.94 1.00
MAXD 0.77 0.85 0.90 0.92
P50 0.14 0.22 0.04 0.06 0.29
FREQD 0.37 0.01* 0.18 0.14 0.16 0.94
DS 0.20 0.12 0.04 0.08 0.43 0.85 0.92
MV 0.10 0.12 0.13 0.19 0.54 0.61 0.69 0.90
RMSV 0.12 0.10 0.11 0.17 0.52 0.60 0.69 0.88 1.00
MFREQ 0.88 0.71 0.76 0.73 0.41 0.38 0.61 0.55 0.53 0.55
CFREQ 0.86 0.65 0.73 0.69 0.36 0.47 0.69 0.62 0.56 0.58 0.99
Tc 0.87 0.72 0.72 0.68 0.37 0.40 0.63 0.58 0.54 0.55 0.94 0.95
P95 0.89 0.80 0.79 0.76 0.47 0.21 0.45 0.42 0.44 0.46 0.95 0.93 0.94

Except for the correlation coefficient marked with *(P 0.16 after Bonferroni correction), all correlations were significantly different from 0 (P 0.001).

J Neurophysiol VOL 93 JANUARY 2005 www.jn.org

196 C. MAURER AND R. J. PETERKA

TABLE 4. Principal components and rotated components of the derived from the stabilogram diffusion function (DS, DL,
correlation matrix of simulated sway measures X2c, and Tc) were very sensitive to changes in one or more
of the model parameters. In addition, P50 showed high sensi-
Principal Principal Rotated Rotated tivity to all parameter changes except KN. Other measures
Sway Component 1 Component 2 Factor 1 Factor 2
measures Eigenvector Eigenvector Components Components showed lower sensitivity to variations of all of the model
parameters. However, the lower sensitivity to parameter
Group 1 DL 0.33 0.16 0.98 0.16 changes did not necessarily lead to poorer correlations between
X2c 0.27 0.27 0.97 0.15
MD 0.29 0.25 0.99 0.08
these sway measures and some of the model parameters (Table
RMS 0.28 0.27 0.98 0.13 5). One reason is that the correlation coefficient characterizes
MAXD 0.18 0.38 0.85 0.48 the clustering of data points around a line representing the
Group 2 P50 0.14 0.33 0.02 0.82 relationship between two variables, irrespective of the scales
FREQD 0.22 0.30 0.21 0.88 the variables are measured on.
DS 0.18 0.38 0.03 0.98
MV 0.16 0.38 0.03 0.95 Figure 4 also shows that several sets of sway measures vary
RMSV 0.16 0.37 0.01 0.94 in similar ways with changes in model parameters. Sway

Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017

Group 3 MFREQ 0.35 0.03 0.82 0.55 measure groups with similar relationships are (MD, RMS, and
CFREQ 0.35 0.07 0.78 0.61 MAXD), (MV, RMSV), and (MFREQ, CFREQ).
Tc 0.34 0.05 0.79 0.57
P95 0.34 0.02 0.85 0.43
Eigenvalue 7.99 4.94 Determining model parameters from sway measures
Percent 57.1% 35.3% 51.8% 40.6%
Cum Percent 57.1% 92.4% 51.8% 92.4% Given the variety of inter-relationships among the model
parameters and sway measures (Fig. 4; Table 5), we predicted
The eigenvalues represent a partition of the total variation in the multivariate
sample. Percent lists the amount of variability covered by the principal
that model parameters could be estimated from a particular set
components and rotated factors as a percent of the total variability. CumPer- of sway measures using an optimization procedure. To test this
cent shows the cumulative percent of variation. The eigenvector components prediction, a reference set of sway measures was calculated for
and factor components represent the directions in parameter space that account a particular set of model parameters in which one or more of
for the data variation of the respective principal components and factors.
the parameters deviate by 10% from the center-normal pa-
Relationship between sway measures and model parameters rameters. An optimization procedure was performed, starting
from the center-normal values of the model parameters, to
Sway measures typically increased with increasing d, de- estimate the postural control parameters KP, KD, d, and KN
creased with increasing KD, and were unaffected by KI (Table that best explained the reference set of sway measures. Figure
5). Effects of KP and KN varied. Group 1 sway measures (DL, 5 shows the mean (with 95% CIs) results of 34 81 optimi-
X2c, MD, RMS, MAXD) were negatively correlated with KP zations, where the estimated parameter is shown as a function
and positively correlated with KN. In addition, except for DL, of the actual parameter used to generate the target sway
all group 1 measures were negatively correlated with KD and
measures. The graphs show that the optimization procedure
positively correlated with d. These findings are in line with the
was able to identify the actual postural control parameters
intuitive expectation that an increase in stiffness (KP) and/or a
decrease in noise level (KN) decrease the sway amplitude. With within a small error tolerance.
two exceptions (P50, FREQD), all group 2 sway measures (DS,
MV, RMSV) were positively correlated with KP, KN, and d, and TABLE 5. Correlations between sway measures and model
negatively correlated with KD. This again is in line with the parameters
intuition that an increase in stiffness and/or a decrease in
damping increases the mean sway frequency and thereby Neural Controller
increases sway velocity. While an increase in noise level Sway
Group Measure KI KP KD d KN
increases the overall sway velocity, it does not affect the
frequency range of sway (symbolized by P50). In fact, the first 1 DL 0.08 0.92 0.01* 0.01* 0.34
exception, P50, did not correlate with KN; the second excep- X2c 0.03 0.82 0.24 0.20 0.32
tion, FREQD, again did not correlate with KN and, in addition, MD 0.12 0.84 0.14 0.13 0.47
showed in all other relationships a reversed sign compared with RMS 0.15 0.81 0.16 0.15 0.50
MAXD 0.17 0.54 0.25 0.38 0.65
the rest of the group 2 measures. Group 3 sway measures were
2 P50 0.13 0.15 0.65 0.61 0*
almost exclusively correlated with KP. They did not correlate FREQD 0.18 0.42 0.63 0.60 0*
with KN, and the correlations with KD and d were low. The DS 0.01* 0.36 0.52 0.65 0.31
model parameter KI showed only weak correlations with the MV 0.01* 0.37 0.31 0.53 0.68
sway measures. Therefore as mentioned in METHODS, we con- RMSV 0.01* 0.39 0.29 0.53 0.68
sider it unlikely that information from sway measures could be 3 MFREQ 0.12 0.96 0.07 0.23 0*
used to characterize changes in the value of KI. CFREQ 0.12 0.93 0.13 0.32 0*
Tc 0.01* 0.90 0.17 0.31 0*
P95 0.04 0.93 0.14 0.29 0*
Sensitivity analysis
Except for the correlation coefficients marked with *, which indicates a P
The sway measures differed considerably in their sensitivity 0.05, all other correlations were significant (P 0.001, after Bonferroni
to changes in model parameters. Figure 4 shows that measures correction).

J Neurophysiol VOL 93 JANUARY 2005 www.jn.org

 A NEW INTERPRETATION OF SPONTANEOUS SWAY 197

Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017

FIG. 5. Comparison of actual model parameters to estimated postural con-
trol model parameters. Estimated parameter values were obtained using an
optimization procedure applied to the 14 sway measures that characterized
simulated body sway. Results are from 81 simulations where combinations of
actual model parameters were varied by 0, 10, or 10% from center-normal
values. Points are offset on the abscissa for clarity. Mean and 95% CIs are
shown for the estimated parameter values.

Determining the influence of passive stiffness

and/or damping
We found that introducing passive dynamics slightly wors-
ened the quality of the data fit. Specifically, the scalar error
function increased from 0.466 with no passive dynamics to
0.505 with passive stiffness (Kpas 10.2 Nm deg1; Table 2,
column passive stiffness) and to 0.553 with passive stiffness
and passive damping (Kpas 10.2 Nm deg1, Bpas 3.0 Nm
s deg1; Table 2, column passive stiff. damp.). This
increase in the scalar error function was mainly caused by a
slightly reduced ability of the models with passive stiffness
TABLE 6. Simulations and experimental results (mean and SD
range) for elderly adults
FIG. 4. Normalized variations in sway measures as a function of 10%
variations of model parameters KP, KD, KI, d, and KN about their center- Experimental Data
normal values. Sway measures and model parameters are normalized relative
to their center-normal values (see Table 2 for center-normal values). Simulated SD
Data Mean Range
Applying the optimization procedure to real datasets
Sway measure
We applied the optimization procedure for parameter iden- Tc 0.75 1.26 0.671.85*
X2c 20.6 27.4 7.5547.2*
tification to real experimental data sets for elderly adults to DS 14.7 12.8 4.1921.4*
determine whether model parameters could be identified that DL 1.89 1.86 0.233.49*
accounted for all of the experimental data. Specifically, we RMS 5.74 4.45 3.035.87
used a combination of sway measures given by Prieto et al. MV 10.9 13.5 8.2418.8
(1996) and by Collins et al. (1995) as the target set of sway MAXD 27.7 21.1 14.228.0
MD 4.52 3.63 2.464.80
measures. The optimization procedure identified postural con- MFREQ 0.58 0.66 0.490.84
trol system parameters of KP 19.7 Nm deg1, KD 6.18 POWER 14.9 28.1 13.143.1
Nm s deg1, d 0.165 s, and KN 693 Nm. This translates P50 0.50 0.39 0.270.51
to an 18% increase of KP, 28% increase of KD, 2.9% decrease P95 1.25 1.45 1.021.88
CFREQ 0.69 0.75 0.570.93
of d, and a 50% increase of KN with respect to the center- FREQD 0.68 0.67 0.630.71
normal parameter values that characterize the sway measures
Model fit error 0.686
of young adults (Table 2). Table 6 shows a comparison of the Model parameters
mean SD range of experimental measures collected from KP 19.7
Prieto et al. (1996) and Collins et al. (1995) and the simulated KD 6.18
sway measures achieved by setting the model parameters to the d 0.163
KN 693
values determined by the optimization procedure. We found
that all simulated sway measures were within 1 SD of the Experimental results taken from studies of *Collins et al. (1995) and Prieto et
experimental sway measures. al. (1996). Units of sway measures and model parameters are given in Table 1.

J Neurophysiol VOL 93 JANUARY 2005 www.jn.org

198 C. MAURER AND R. J. PETERKA
and/or passive damping to account for some of the experimen-
tal sway measures (RMS, FREQD) compared with a model
with only active stiffness and damping. Finally, we applied an
optimization procedure that allowed all six parameters (Kpas,
Bpas, KP, KD, d, KN) to vary, to determine if the optimization
procedure could identify any combination of Kpas and Bpas that
provided a lower error than the active feedback model. We
found that the values for Kpas and Bpas both converged to zero.
DISCUSSION
Our results showed that a very simple feedback model of
human postural control was able to reproduce realistic sway
behaviors. The sway measures obtained from model simula-
tions were well within the range of 1 SD of experimental data
published by Collins and De Luca (1993) and by Prieto et al.
(1996) for young healthy adults. Our results extend the previ-
ous finding of Peterka (2000) that showed that simulated
spontaneous body sway traces yield realistic SDFs. We show
here for the first time that numerous other COP-related sway
measures reported in the literature for normal subjects are also
compatible with this feedback control model.
Next, we showed that the different sway measures obtained
from the simulated COP traces aggregated into three groups
with high correlations within each group, while correlations
between groups were considerably lower. The first group
contained displacement-related measures, the second group
mainly velocity-related measures, and the third group mainly
frequency-related measures of sway. By deriving principal
components from the correlation matrix, we showed that about
92% of the variability among all different sway measures was
explained by only two principal components. Factor rotation of
the principal components led to factors that were closely
related to the sway measure aggregates (groups 13) derived
from the correlation matrix. The first factor mainly determined
the first group (displacement-related measures). The second
factor determined the second group (mostly velocity-related
measures). Both factors contributed to the frequency-related
sway measures (group 3).
By analyzing the relationships between sway measures and
the underlying model parameters, we showed that most of the
group 1related sway measures (specifically X2c, MAXD,
RMS, MD) were positively correlated with time delay (d) and
noise level (KN), but were negatively correlated with stiffness
(KP). Group 2related measures (specifically RMSV, MV, DS)
were positively correlated with KP, d, and KN. Together with
the sensitivity relationships displayed in Fig. 4, this correlation
matrix provides insight into why one sway measure might be
more or less reliable than another in distinguishing between
different categories of subjects. For example, an individual
sway measure would be unreliable if 1) the particular change in
a postural control parameter between subject categories had
little effect on that sway measure (i.e., low sensitivity, e.g.,
FREQD) or 2) if two or more postural control parameters
changed between subject categories and the effects on that
sway measure cancelled. Alternatively, an individual sway
measure would reliably distinguish between subject categories
if the particular change in the postural control parameter
between subject categories had a large effect on the sway
measures and/or there was an additive effect or at least no
J Neurophysiol VOL
cancellation of effects among the different changes of postural
control parameters.
We found that information contained in the relationships
among the different sway measures is sufficient to identify the
values of different postural control model parameters. By using
an optimization procedure, we demonstrated that it is possible,
in principle, to identify model parameters by analyzing the set
of 14 sway measures we investigated. This optimization pro-
cedure is able to accurately identify model parameters because
the set of sway measures includes a wide range of functional
relationships between model parameters and the individual
sway measures (Fig. 4).
It is likely that it is not necessary to include all 14 of the
sway measures we used in the optimization procedure. For
example, the sway measure FREQD showed very low sensi-
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017
tivity to parameter changes. One would also anticipate that
sway measures with high sensitivity to model parameter
changes would be particularly useful in identifying model
parameters. Interestingly, the parameters with some of the
highest sensitivities were parameters derived from SDFs (DS,
DL, X2c, Tc), suggesting that parameters derived from
stabilogram diffusion analysis should be more sensitive at
detecting changes in postural control dynamics than most other
sway measures. However, in identifying model parameters, it
is not only important that the sway measures are sensitive to
parameter changes, but also that the sway measures are not
highly correlated to one another.
We then showed that the optimization procedure of param-
eter identification is applicable to real datasets. We used a
combination of sway measures given by Prieto et al. (1996)
and by Collins et al. (1995) for elderly subjects, accepting the
fact that there were different individuals in these two subject
groups. Another limitation arose from the fact that we did not
know subjects body masses and moments of inertia. However,
the optimization procedure converged to a solution that found
that modest increases in stiffness (KP) and damping (KD) and a
fairly large increase in noise level (KN) compared with the
center-normal parameter set that corresponds to young adults
could account for the variety of sway measures reported in the
literature. Specifically, our model parameters predicted sway
measures that were well within 1 SD of the experimental data
collected by Prieto et al. (1996) and Collins et al. (1995). One
could speculate that the increased noise level in elderly sub-
jects is due to degradation of sensory function with age such
that actual body sway is less precisely encoded by sensory
systems. This increase in noise level is partially compensated
for by increases in both stiffness and damping. That is, in-
creases in stiffness and damping tend to reduce body sway
displacement that is caused by an increased noise level (RMS,
MD, MAXD measures in Fig. 4 all decrease with increasing KP
and KD). However, the model predicts that an increase in
stiffness and noise level both contribute to an increase in
velocity-related sway measures, consistent with the experimen-
tal finding of Prieto et al. (1996) that MV was the one measure
with the highest significance for separating age groups.
Finally, we determined whether the inclusion of passive
stiffness and/or damping components in the model could pro-
vide a better explanation for the experimental sway measures
of young adults than a model without passive components.
There is disagreement about the relative importance of active
versus passive contributions to corrective torque generation,
93 JANUARY 2005 www.jn.org
 A NEW INTERPRETATION OF SPONTANEOUS SWAY
with some studies showing a dominant role of active mecha-
nisms (Peterka 2002; Peterka and Loughlin 2004) and others a
dominant contribution from passive mechanisms (Loram and
Lakie 2002a, b). We showed that the optimization procedure
yielded a slightly better fit to experimental sway measures for
pure active control compared with control provided by a
combination of a smaller active and a larger passive compo-
nent. The decrease in fitting quality was shown by an increase
in the scalar error function and by a reduced ability of the
model with a passive component to predict the experimental
sway measures (i.e., RMS and FREQD were more than 1 SD
away from the experimental sway measures given by Prieto et
al. 1996 and Collins and De Luca 1993). When we included
passive stiffness and damping as freely varying model param-
eters in the optimization procedure, both passive stiffness and
passive damping converged to zero. However, the scalar error
function values of these different model structures had similar
values for all models tested. Therefore it is unlikely that results
from the optimization procedure can be used to determine the
relative contributions of active and passive components to
torque generation. Moreover, the poorer fit with the inclusion
of the passive components could be the result of using exper-
imental data from multiple sources.
In conclusion, this study shows that COP traces that closely
resemble physiologically measured COP functions can be pro-
duced by an appropriate selection of model parameters in a
simple feedback model of the human postural control system.
Variations in the values of stiffness, damping, time delay, and
noise level control the values of different sway measures,
leading to a characteristic pattern of sway measures. Selection
of appropriate model parameters can generate COP traces
whose sway measures match the measures reported for COP
traces recorded from young and elderly adults. Conversely, a
given set of sway measures can be used to estimate a set of
model parameters. These model parameters describe physio-
logically meaningful features of the postural control system as
opposed to conventional sway measures that provide only a
parametric description of sway. Although our modeling results
are consistent with a continuous feedback-regulation strategy
for upright stance control and are consistent with other recent
results (Peterka 2002; Peterka and Loughlin 2004), these re-
sults do not prove that this is the actual control strategy used by
the nervous system. That is, other control strategies might exist
that are also able to explain the statistical properties of spon-
taneous body sway. For example, the apparent simplicity of the
model could mask the potential complexity of sensory infor-
mation processing required for postural control during quiet
stance (Mergner et al. 1997, 2002; Peterka 2002). Neverthe-
less, it may be instructive to consider the implications for
differences in postural stability in terms of our models struc-
ture. The proposed model is quite simple, and yet is able to
account for the statistical properties of body sway as revealed
by the different sway measures.
GRANTS
This work was supported by National Institute on Aging Grant AG-17960
to R. J. Peterka.
REFERENCES
Baloh RW, Jacobson KM, Enrietto JA, Corona S, and Honrubia V.
Balance disorders in older persons: quantification with posturography.
Otolaryngol Head Neck Surg 119: 89 92, 1998.
J Neurophysiol VOL 93 JANUARY 2005
199
Baratto L, Morasso PG, Re C, and Spada G. A new look at posturographic
analysis in the clinical context: sway-density versus other parameterization
techniques. Motor Control 6: 246 270, 2002.
Barin K. Evaluation of a generalized model of human postural dynamics and
control in the sagittal plane. Biol Cybern 61: 3750, 1989.
Bendat JS and Piersol AG. Random Data: Analysis and Measurement
Procedures. New York: John Wiley, 2000.
Brocklehurst JC, Robertson D, and James-Groom P. Clinical correlates of
sway in old agesensory modalities. Age Ageing 11: 110, 1982.
Burleigh A, Horak FB, Nutt J, and Frank J. Levodopa reduces muscle tone
and lower extremity tremor in Parkinsons disease. Can J Neurol Sci 22:
280 285, 1995.
Chiari L, Rocchi L, and Cappello A. Stabilometric parameters are affected
by anthropometry and foot placement. Clin Biomech 17: 666 677, 2002.
Collins JJ and De Luca CJ. Open-loop and closed-loop control of posture: a
random-walk analysis of center-of-pressure trajectories. Exp Brain Res 95:
308 318, 1993.
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017
Collins JJ, De Luca CJ, Burrows A, and Lipsitz LA. Age-related changes
in open-loop and closed-loop postural control mechanisms. Exp Brain Res
104: 480 492, 1995.
Dichgans J, Mauritz KH, Allum JH, and Brandt T. Postural sway in
normals and atactic patients: analysis of the stabilizing and destabilizing
effects of vision. Agressologie 17: 1524, 1976.
Dickstein R, Shupert C, and Horak FB. Fingertip touch improves postural
stability in patients with peripheral neuropathy. Gait Posture 14: 238 247,
2001.
Era P and Heikkinen E. Postural sway during standing and unexpected
disturbance of balance in random samples of men of different ages. J
Gerontol 40: 287295, 1985.
Fitzpatrick R, Burke D, and Gandevia SC. Loop gain of reflexes controlling
human standing measured with the use of postural and vestibular distur-
bances. J Neurophysiol 76: 3994 4008, 1996.
Horak FB, Dickstein R, and Peterka RJ. Diabetic neuropathy and surface
sway-referencing disrupt somatosensory information for postural stability in
stance. Somatosens Mot Res 19: 316 326, 2002.
Hufschmidt A, Dichgans J, Mauritz KH, and Hufschmidt M. Some
methods and parameters of body sway quantification and their neurological
applications. Arch Psychiatr Nervenkr 228: 135150, 1980.
Ishida A and Imai S. Responses of the posture-control system to pseudoran-
dom acceleration disturbances. Med Biol Eng Comput 18: 433 438, 1980.
Ishida A and Miyazaki S. Maximum likelihood identification of a posture
control system. IEEE Trans Biomed Eng 34: 15, 1987.
Jackson JE. A Users Guide to Principal Components. New York: John
Wiley, 1991.
Johansson R and Magnusson M. Human postural dynamics. Biomed Eng 18:
413 437, 1991.
Johansson R, Magnusson M, and kesson M. Identification of human
postural dynamics. IEEE Trans Biomed Eng 35: 858 869, 1988.
Jolliffe IT. Principal Component Analysis. New York: Springer-Verlag, 1986.
Kollegger H, Baumgartner C, Wober C, Oder W, and Deecke L. Sponta-
neous body sway as a function of sex, age, and vision: posturographic study
in 30 healthy adults. Eur Neurol 32: 253259, 1992.
Lagarias JC, Reeds JA, Wright MA, and Wright PE. Convergence prop-
erties of the Nelder-Mead simplex method in low dimensions. SIAM J
Optimizat 9: 112147, 1998.
Laughton CA, Slavin M, Katdare K, Nolan L, Bean JF, Kerrigan DC,
Phillips E, Lipsitz LA, and Collins JJ. Aging, muscle activity, and balance
control: physiologic changes associated with balance impairment. Gait
Posture 18: 101108, 2003.
Loram ID and Lakie M. Human balancing of an inverted pendulum: position
control by small, ballistic-like, throw and catch movements. J Physiol 540:
11111124, 2002a.
Loram ID and Lakie M. Direct measurement of human ankle stiffness during
quit standing: the intrinsic mechanical stiffness is insufficient for stability.
J Physiol 545: 10411053, 2002b.
Maki BE, Holliday PJ, and Fernie GR. Aging and postural control. A
comparison of spontaneous- and induced-sway balance tests. J Am Geriatr
Soc 38: 19, 1990.
Maurer C, Mergner T, Xie J, Faist M, Pollak P, and Lucking CH. Effect
of chronic bilateral subthalamic nucleus (STN) stimulation on postural
control in Parkinsons disease. Brain 126: 1146 1163, 2003.
Mergner T, Huber W, and Becker W. Vestibular-neck interaction and
transformation of sensory coordinates. J Vestib Res 7: 347367, 1997.
www.jn.org
200 C. MAURER AND R. J. PETERKA
Mergner T, Maurer C, and Peterka RJ. A multisensory posture control
model of human upright stance. Progr Brain Res 142: 189 201, 2002.
Milotti E. 1/f noise: a pedagogical review. [Online]. arXiv.org e-Print archive.
http://arxiv.org/abs/physics/0204033 [2002].
Morasso PG, Baratto L, Capra R, and Spada G. Internal models in the
control of posture. Neural Networks 12: 11731180, 1999.
Murray MP, Seireg AA, and Sepic SB. Normal postural stability and steadiness:
quantitative assessment. J Bone Joint Surg Am 57: 510 516, 1975.
Overstall PW, Exton-Smith AN, Imms FJ, and Johnson AL. Falls in the
elderly related to postural imbalance. Br Med J 1: 261264, 1977.
Peterka RJ. Simple models of sensory interaction in human postural control.
In: Multisensory Control of Posture and Movement, edited by Hlavacka F
and Mergner T. New York: Plenum, 1995, p. 281288.
Peterka RJ. Postural control model interpretation of stabilogram diffusion
analysis. Biol Cybern 82: 335343, 2000.
Peterka RJ. Sensorimotor integration in human postural control. J Neuro-
physiol 88: 10971118, 2002.
Peterka RJ and Benolken MS. Role of somatosensory and vestibular cues in
attenuating visually induced human postural sway. Exp Brain Res 105:
101110, 1995.
J Neurophysiol VOL 93 JANUARY 2005
Peterka RJ and Loughlin PJ. Dynamic regulation of sensorimotor integration
in human postural control. J Neurophysiol 91: 410 423, 2004.
Prieto TE, Myklebust JB, Hoffman RG, Lovett EG, and Myklebust BM.
Measures of postural steadiness: differences between healthy young and
elderly adults. IEEE Trans Biomed Eng 43: 956 966, 1996.
Rocchi L, Chiari L, and Cappello A. Feature selection of stabilometric
parameters based on principal component analysis. Med Biol Eng Comput
42: 7179, 2004.
Rocchi L, Chiari L, and Horak FB. Effects of deep brain stimulation and
levodopa on postural sway in Parkinsons disease. J Neurol Neurosurg
Psychiatr 73: 267274, 2002.
Uccioli L, Giacomini PG, Monticone G, Magrini A, Durola L, Bruno E,
Parisi L, Di Girolamo S, and Menzinger G. Body sway in diabetic
neuropathy. Diabetes Care 18: 339 344, 1995.
van der Kooij H, Jacobs R, Koopman B, and Grootenboer H. A multisen-
sory integration model of human stance control. Biol Cybern 80: 299 308,
1999.
van der Kooij H, Jacobs R, Koopman B, and van der Helm F. An adaptive
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on October 19, 2017
model of sensory integration in a dynamic environment applied to human
stance control. Biol Cybern 84: 103115, 2001.
www.jn.org