Interactions Between Posture and Locomotion: Motor Patterns in
Humans Walking With Bent Posture Versus Erect Posture
R. GRASSO, M. ZAGO, AND F. LACQUANITI
Human Physiology Section of the Scientific Institute Santa Lucia and the University of Rome “Tor Vergata”,
00179 Rome, Italy
INTRODUCTION
The issue of the relationship between posture and locomotion is of great theoretical and experimental relevance (see
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.
288
Burleigh et al. 1994; Lacquaniti et al. 1997; Massion 1992;
Mori 1987; Winter 1991; Zernicke and Smith 1996). Neurophysiological studies indicate that the control of posture and
locomotion are interdependent at many different levels of the
CNS, from the motor cortex to the basal ganglia, the brain
stem, and the spinal cord. Thus basal ganglia are connected
reciprocally with ponto-mesecephalic structures (the peduncolopontine nucleus) belonging to the mesencephalic locomotor
region (an area known to modulate spinal locomotion oscillators) (Shik and Orlovsky 1976), on the one side, and with the
axial motor cortex, on the other side. These structures participate in the preparation and initiation of locomotion by providing the appropriate spatial frameworks necessary to incorporate postural adjustments in the locomotion process (GarciaRill 1986; Grasso et al. 1999). Also when specific areas in the
hypothalamus or the brain stem are stimulated, changes in
posture are triggered just before step initiation (Mori 1987;
Mori et al. 1978, 1983, 1989). Interactions between pathways
controlling posture and gait exist even at the level of spinal
premotor interneurons (Jankowska and Edgley 1993). In sum,
it has been suggested that the “activation of setting mechanisms in the level of postural muscle tone and that of the spinal
stepping generator are not separate phenomena” (Mori 1987).
If the interrelationship between posture and locomotion is
well recognized, the implications vis-à-vis the issue of the
motor pattern generation are not well understood (see, however, Zernicke and Smith 1996). In particular, it is not clear
whether or not the setting of different task-dependent postural
configurations of the body should affect the waveforms of the
basic motor patterns that are rhythmically output during walking. This question hinges on the more general problem of the
nature of the control waveforms output by central locomotor
networks (see Lacquaniti et al. 1999; Zernicke and Smith
1996).
It is generally thought that the multisegment motion of
mammals locomotion is controlled by a network of coupled
oscillators (central pattern generators or CPGs) (see Grillner
1981; Pearson 1993; Rossignol 1996). Although it often is
assumed that CPGs control patterns of muscle activity, an
equally plausible hypothesis holds that they control patterns of
limb segment motion instead (Bianchi et al. 1998b; Borghese
et al. 1996; Grasso et al. 1998; Shen and Poppele 1995). This
hypothesis is based on the following observations (for a review, see Lacquaniti et al. 1999). In walking the waveforms of
limb segment angular motion are much simpler and more
consistent than the corresponding waveforms of muscle activity, both in man (Borghese et al. 1996; Grasso et al. 1998) and
0022-3077/00 $5.00 Copyright © 2000 The American Physiological Society
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
Grasso, R., M. Zago, and F. Lacquaniti. Interactions between
posture and locomotion: motor patterns in humans walking with bent
posture versus erect posture. J. Neurophysiol. 83: 288 –300, 2000.
Human erect locomotion is unique among living primates. Evolution
selected specific biomechanical features that make human locomotion
mechanically efficient. These features are matched by the motor
patterns generated in the CNS. What happens when humans walk with
bent postures? Are normal motor patterns of erect locomotion maintained or completely reorganized? Five healthy volunteers walked
straight and forward at different speeds in three different postures
(regular, knee-flexed, and knee- and trunk-flexed) while their motion,
ground reaction forces, and electromyographic (EMG) activity were
recorded. The three postures imply large differences in the position of
the center of body mass relative to the body segments. The elevation
angles of the trunk, pelvis, and lower limb segments relative to the
vertical in the sagittal plane, the ground reaction forces and the
rectified EMGs were analyzed over the gait cycle. The waveforms of
the elevation angles along the gait cycle remained essentially unchanged irrespective of the adopted postures. The first two harmonics
of these kinematic waveforms explain ⬎95% of their variance. The
phase shift but not the amplitude ratio between the first harmonic of
the elevation angle waveforms of adjacent pairs was affected systematically by changes in posture. Thigh, shank, and foot angles covaried
close to a plane in all conditions, but the plane orientation was
systematically different in bent versus erect locomotion. This was
explained by the changes in the temporal coupling among the three
segments. For walking speeds ⬎1 m s⫺1, the plane orientation of bent
locomotion indicates a much lower mechanical efficiency relative to
erect locomotion. Ground reaction forces differed prominently in bent
versus erect posture displaying characteristics intermediate between
those typical of walking and those of running. Mean EMG activity
was greater in bent postures for all recorded muscles independent of
the functional role. The waveforms of the muscle activities and
muscle synergies also were affected by the adopted posture. We
conclude that maintaining bent postures does not interfere either with
the generation of segmental kinematic waveforms or with the planar
constraint of intersegmental covariation. These characteristics are
maintained at the expense of adjustments in kinetic parameters, muscle synergies and the temporal coupling among the oscillating body
segments. We argue that an integrated control of gait and posture is
made possible because these two motor functions share some common
principles of spatial organization.
POSTURAL GEOMETRY AND GAIT COORDINATION
the kinetic and kinematic signals rather than their mean value.
According to our previous hypothesis, locomotion is controlled
in a kinematic space (see Lacquaniti et al. 1999). If so, one
would expect that the changes in posture should leave the
kinematic waveforms essentially unchanged. The interrelationship between posture and locomotion should manifest itself
instead on the phase delay between the motion of different limb
segments; that is, the control parameter that is set centrally
according to the specific task demands. (As we noted in the
preceding text, the phase delay varies as a function of changes
in walking speed or reversal of walking direction.) An additional implication of our hypothesis is that the conservation of
kinematic templates across changes in body posture can occur
only at the expense of a reorganization of muscle patterns and
synergies.
METHODS
General procedures have been previously described (Bianchi et al.
1998b; Borghese et al. 1996). Kinematic data were obtained by means
of the ELITE system (Ferrigno et al. 1990). Four 100-Hz TV cameras
were spaced on the recording side of the walkway to enhance spatial
accuracy. After three-dimensional (3D) calibration, the SD spatial
accuracy of the system was better than 1.5 mm (Bianchi et al. 1998b).
The position of selected points on the side of the dominant lower limb
was recorded by attaching the infrared reflective markers to the skin
overlying the following landmarks: gleno-humeral joint (GH), anterior superior iliac spine (ASIS), posterior superior iliac spine (PSIS),
greater trochanter (GT), lateral femur epicondyle (LE), lateral malleolus (LM), and fifth metatarso-phalangeal joint (VM). VM marker
was placed on the shoe after verifying the correspondence on the bare
foot. ASIS and PSIS coordinates were averaged to obtain ilium (IL)
position. The body was modeled as an interconnected chain of rigid
segments: GH-IL for the trunk, IL-GT for the pelvis, GT-LE for the
thigh, LE-LM for the shank, and LM-VM for the foot. The limb axis
was defined as GT-LM.
Ground reaction forces were recorded at 500 Hz by means of a
multicomponent force platform (0.6 ⫻ 0.4 m, Kistler 9281B), placed
approximately at the center of the walkway. EMG activity was recorded by means of surface electrodes from the gluteus maximus
(GM), biceps femoris (long head) (BF), rectus femoris (RF), vastus
lateralis (VL), lateral gastrocnemius (GCL), and tibialis anterior (TA).
EMG signals were preamplified (100 ⫻) at the recording site, digitized, and transmitted to the remote amplifier via 15-m optic fibers.
These signals were band-pass filtered (10-Hz high-pass and 200-Hz
low-pass, 4-pole Bessel filters), and sampled at 500 Hz. Sampling of
kinematic, force, and EMG data were synchronized.
Protocol
Experiments were approved by the Ethics Committee of Santa
Lucia Institute and conformed with the Declaration of Helsinki on the
use of human subjects in research. Five healthy volunteers (2 females,
3 males, 21–36 yr age range) participated after giving informed
consent. Before the recording session, the dominant lower limb of
each subject was determined according to standard criteria (VandenAbeele 1980). All subjects proved to be right-dominant. Before the
experiment, we asked our subjects to adopt static KF and KT postures
on the force platform and computed the instantaneous position of the
center of pressure of the net reaction forces during 20-s records
(which were repeated 2 times). Fifteen seconds of steady state data
were averaged to obtain the mean anteroposterior displacement of the
center of pressure relative to RE condition, which was 0.38 ⫾ 0.48
and 0.43 ⫾ 0.66 cm (mean ⫾ SD) forward for KF and KT, respectively. This small displacement is not statistically significant.
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
cat (Shen and Poppele 1995). Moreover the temporal changes
of the elevation angles of the lower limb segments do not
evolve independently of each other, but they are tightly coupled together (Bianchi et al. 1998b; Borghese et al. 1996).
When the elevation angles are plotted one versus the others,
they describe a regular loop constrained close to a plane,
common to both the stance and swing phase. The specific
orientation of the plane of angular covariation reflects the
phase relationships between the elevation angles of the segments and therefore the timing of the intersegmental coordination. The phase delay shifts systematically with increasing
speed both in man (Bianchi et al. 1998b) and cat (Shen and
Poppele 1995). Because in man this phase-shift is correlated
with the net mechanical power output over a gait cycle (Bianchi et al. 1998b), we hypothesized that the control of kinematic
phase can be used by the nervous system for limiting the
overall energy expenditure with increasing speed (Bianchi et
al. 1998a). Finally, we observed that a reversal of the direction
of walking from forward to backward involves the same waveforms (time-reversed) of the elevation angles as in forward
gait, with a simple reversal of the delay in the phase coupling
between limb segments, at the expense of a reorganization of
the patterns of muscle activity (Grasso et al. 1998).
How does a change in walking posture affect these locomotor patterns? Although we normally walk with an erect posture,
we can as easily walk stooped (as it happens in a low tunnel).
Here we compared normal erect walking (regular, RE) with
two different styles of bent walking, namely knee-flexed walking (KF) and knee-flexed plus trunk-flexed walking (KT). RE
walking is a unique feature of human locomotion (nonhuman
primates normally walk with a bent posture). Its evolutionary
history indicates highly specific adaptations of the skeletal and
muscular apparatus (Crompton et al. 1998; Spoor et al. 1994).
Also, erect posture is mechanically efficient in humans because
the center of body mass vaults over the supporting limb like an
inverted pendulum, thereby limiting energy expenditure by
means of an exchange of the forward kinetic energy with the
gravitational potential energy (Cavagna et al. 1977). Maintaining KF and KT posture while walking may interfere with the
pendulum mechanism. McMahon et al. (1987) showed that in
subjects running knee-flexed, reaction moments acting on the
knee are increased and the effective vertical spring stiffness of
the legs is decreased relative to normal running. We introduced
condition KT in addition to KF to assess the specific role of
trunk orientation in the generation of locomotor patterns. In
fact it has been proposed that the trunk may act as a reference
in the control of posture and movement (Darling and Miller
1995; Massion et al. 1997; Mouchnino et al. 1993). This role
then could be disrupted when the trunk is flexed.
In looking for the effects of changes of walking posture on
the locomotor patterns, we keep in mind that there exist inevitable mechanical consequences of a bent posture. In particular,
KF and KT walking must involve an offset in the mean value
of several kinematic and kinetic variables as compared with RE
walking. For instance, because the legs are flexed, the mean
elevation angles of the limb segments in the former tasks will
be generally different from those measured in the latter task.
Also the mean level of muscle activity will be increased
because of the reduced mechanical advantage of bent limbs
(Biewener 1990). With regard to the issue of locomotor pattern
generation, however, the key point concerns the waveform of
289
290
R. GRASSO, M. ZAGO, AND F. LACQUANITI
Subjects then were instructed to walk (with their shoes on) with the
arms folded on the chest, at a freely chosen, roughly constant speed
within the ELITE sampling volume. They were encouraged to vary
the speed across trials. Because no additional constraints were used,
the speed range overlapped but did not coincide exactly in different
experiments. To avoid modifications of the natural step length, subjects were asked not to pay attention whether or not they stepped onto
the force platform. Walks were performed in one of the three different
postures (RE, KF, and KT), randomly alternated from trial to trial.
paths that can be fitted by a plane that is computed by means of
orthogonal regression (Bianchi et al. 1998b; Borghese et al. 1996). In
each trial, we computed the covariance matrix R of the ensemble of
time-varying elevation angles over the gait cycle, after subtraction of
their respective mean value. Eigenvalues ␭ and eigenvectors u are
computed by factoring the covariance matrix R from the set of
original signals by using a singular value decomposition algorithm
such that
Data analysis
where U and ⌳ are the eigenvector and eigenvalue matrices, respectively, and T denotes matrix transpose. The sum of the eigenvalues is
equal to the sum of the variances of the original signal waveforms.
The three eigenvectors u1 ⫺ u3 of R, rank-ordered on the basis of the
corresponding eigenvalues, correspond to the orthogonal directions of
maximum and minimum variance in the sample scatter. The first two
eigenvectors u1 ⫺ u2 lie on the best-fitting plane of angular covariation. The third eigenvector (u3) is the normal to the plane and defines
the plane orientation in the position-space of the elevation angles. For
each eigenvector, the parameters uit, uis, and uif correspond to the
direction cosines with the positive semi-axis of the thigh, shank and
foot angular coordinates, respectively. The direction cosines of the
plane normal provide a measure of the orientation of the plane. In
particular, u3t denotes the direction cosine with respect to the thigh
axis.
RESULTS
General gait parameters
Walking flexed was performed easily in both KF and KT
conditions. The subjects had no difficulty in maintaining that
posture for the few seconds required to walk the 10-m path. We
analyzed a total number of 463 gait cycles at speeds within the
range 0.2 ⫼ 2.0 m s⫺1. The range of speeds 0.5 ⫼ 1.5 m s⫺1
was represented evenly across subjects and conditions.
The well-known monotonic increase of step length with
walking speed was not significantly affected by postural geometry. The slopes of the intraindividual regression lines were
0.48 ⫾0.11 m per 1 m s⫺1 speed increment, 0.44 ⫾ 0.33, and
0.33 ⫾ 0.38 for RE, KF, and KT, respectively (the intercepts
were 0.75 ⫾0.09 m, 0.84 ⫾ 0.41, 0.83 ⫾ 0.37). However the
scatter was larger for KF and KT than for RE (correlation
coefficients were 0.97 ⫾ 0.02, 0.85 ⫾ 0.10, and 0.69 ⫾ 0.38,
respectively). Repeated-measures ANOVA across the three
conditions yielded nonsignificant outcomes for the comparisons between regression parameters from the three conditions.
On average, the duration of the stance phase was ⬃60% of the
gait cycle in all conditions (61.9 ⫾ 3.9%, 62.1 ⫾ 2.7%, and
62.8 ⫾ 2.6% for RE, KF, and KT, respectively) and decreased
with increasing speed (on average by 6.5% of gait cycle
duration per 1 m s⫺1). The latter relationship was less consistent in KF and KT than in RE.
Geometric postural features
Figure 1 (top) shows stick-diagram series from one subject
walking at approximately the same speed (⬇1 m s⫺1) with the
three different postures. As implied by the task, postural geometry differs drastically across the three conditions. In KF
and KT, the lower limbs are flexed and the trunk is tilted, more
so in KT. The vertical oscillations of the trunk and pelvis
segments are less pronounced in KF and KT than in RE. The
position of the CM is displaced forward and downward in KF
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
Three-dimensional kinematic data were filtered with an optimal
low-pass FIR filter with automatic bandwidth selection (D’Amico and
Ferrigno 1990). The angle of elevation of the ith segment in the
sagittal plane with respect to the vertical was computed as ai ⫽
arctan[(xd ⫺ xp)/(yp ⫺ yd)], subscripts p and d denoting proximal and
distal endpoints of the segment, respectively, and x and y the horizontal and vertical coordinates in the sagittal plane, respectively.
Elevation angles are positive in the forward direction relative to the
vertical (i.e., when the distal marker falls anterior to the proximal).
Gait cycle (T) was defined as the time interval between two successive
maxima in the time series of the limb axis elevation, step length (S) as
the linear translation of GT marker during T, and average speed as
V ⫽ S/T. Stance phase was defined as the interval during which the
vertical reaction force exceeded 7% of body weight. The maximum
elevation of the limb axis slightly precedes the heel touch-down
(Borghese et al. 1996). Different trials from each subject were ensemble-averaged after time-interpolation of the kinematic data over T
to fit a normalized 1,000-points time base. Postural geometry was
measured by taking the mean, maximum and minimum value of the
elevation angle of each segment over T. The x, y coordinates of the
location of the center of body mass (CM) are computed as the mean
of x, y coordinates of the respective centers of mass of seven body
segments: the H.A.T. plus the thigh, shank and foot of the right and
left limbs. H.A.T. is comprised of head, folded arms, trunk, and pelvis
and is assumed to be one rigid link (corresponding to GH-GT segment) (see Winter 1991). The center of mass of each of these segments was derived from anthropometric measurements taken on the
subject and the geometric models based on the gamma-scanner
method (Zatsiorsky et al. 1990). Motion of the side of the body
contralateral to the recording one was estimated by time-shifting the
recorded data by T/2, under the assumption of symmetrical gait
(Bianchi et al. 1998b).
EMGs were rectified numerically and low-pass filtered (in both
time directions to avoid tail and phase distortions) by means of a
Butterworth filter, with cutoff at 25 Hz. Cross-correlation functions
(CCF) between pairs of muscles were computed as previously described (Grasso et al. 1998).
Intersegmental coordination among the lower limb segments (thigh,
shank, and foot) was evaluated starting from elevation angle waveforms both in the time domain (1) and in position space (2).
1) Because of the periodic structure of the elevation angle waveforms over the gait cycle (Bianchi et al. 1998b; Borghese et al. 1996),
we decomposed such waveforms into their Fourier series components.
The series was truncated at the 10th harmonic which had a frequency
corresponding to the highest low-pass cutoff frequency in the data
series.
The amplitude transfer ratio (Gpd) and phase shift (⌽pd) between
the corresponding Fourier harmonics from the elevation angles of
adjacent limb segments p and d were computed, respectively, as
Ak(d)/Ak(p) and as ⌰k(d) ⫺ ⌰k(p) where Ak and ⌰k are the amplitude
and phase respectively of the kth order Fourier series component.
2) The intersegmental coordination was evaluated in position space
as previously described (Borghese et al. 1996). Briefly, the changes of
the elevation angles of the thigh, shank, and foot covary linearly
throughout the gait cycle. The thigh-shank-foot 3D loops describe
R ⫽ U⌳U T
POSTURAL GEOMETRY AND GAIT COORDINATION
291
and more so in KT so that it tends to lie outside of the body.
(CM in the 1st stick diagram from each of the 3 series is
displayed as a F in Fig. 1). Note that the hip is raised above the
supporting foot in midstance like an inverted pendulum in RE
but that it remains level or is even lowered in midstance in KF,
a situation closer to that occurring in running than in normal
walking. On the other hand, the trajectory described by the foot
tip is similar across the three postures. Figure 1, bottom, shows
the instantaneous location of the CM as a function of time from
the gait cycles displayed in the top. The prominent oscillation
typical of regular gait is dampened drastically in KF, indicating
that the impact shock with the ground is attenuated by the
lower limb joints. The decrease in the oscillation of the CM in
KF relative to RE was more evident at higher speeds.
Table 1 reports the mean value (over the gait cycle, from all
trials of all subjects) of the elevation angles of each limb
segment in the sagittal plane, and the mean value of the length
of the limb axis. The mean orientation of the body segments
changed according to the instructions. Thus in bent postures,
the trunk was flexed anterior, the thigh was elevated in the
forward direction, and the shank in the backward direction in
TABLE 1. Mean elevation angle for the recorded body segments
and limb axis length
Segment
RE
KF
KT
Trunk
Pelvis
Thigh
Shank
Foot
Limb axis
1.4 ⫾ 3.4
13.5 ⫾ 5.4
1.2 ⫾ 3.4
⫺15.2 ⫾ 3.2
48.7 ⫾ 6.2
754 ⫾ 59
⫺12.8 ⫾ 9.3
10.9 ⫾ 5.8
22.1 ⫾ 5.3
⫺31.5 ⫾ 5.4
46.2 ⫾ 8.6
665 ⫾ 50
⫺54.5 ⫾ 12.7
6.9 ⫾ 6.9
28.1 ⫾ 3.6
⫺27.7 ⫾ 8.2
49.6 ⫾ 8.2
651 ⫾ 38
Values are means ⫾ SD. Angles are in degrees. Limb axis length is in mm.
RE, regular walking; KF, knee-flexed walking; KT, knee- and trunk-flexed
walking.
comparison with erect posture. The changes in these parameters were not independent of each other. The mean elevation of
the trunk and thigh changed in a correlated manner across all
subjects and postures, as did thigh and shank (r ⫽ ⫺0.73 and
0.77, respectively). As expected, the mean limb length was
significantly shorter in KF and KT than in RE (F1,460⫽194.3,
P ⬍ 0.001, with no difference between KF and KT, Scheffé’s
post hoc test).
Kinematic waveforms
Figure 2 shows the average waveforms (⫾SD, data from all
trials and subjects) of the elevation angles in the three conditions. These waveforms are plotted versus normalized time,
expressed as a percentage of gait cycle duration. The elevation
angle of the trunk and pelvis changed little during the gait
cycle. All other elevation angles displayed the typical biphasic
shape, which has been described thoroughly in previous articles (Bianchi et al. 1998b, Borghese et al. 1996, Grasso et al.
1998). Briefly, the lower limb segments rotate backward during
stance and forward during swing.
The new finding is that the segmental kinematic waveforms
have a very similar shape in the three conditions (RE, KF, and
KT) even though they may differ in offset or amplitude. The
SD bands are rather narrow but more so in the RE condition.
Table 2 shows the correlation coefficients (mean ⫾ SD, data
from all subjects) computed between the individual average
waveforms of each experimental condition. The correlation
was always very high, confirming that the shape of the elevation angle waveforms was maintained across the three walking
conditions in all subjects.
Figure 3 shows the range (⫾SD) of the values of the elevation angles during one gait cycle for the different tasks. For all
limb segments, the overall extent of the angular excursion
changed only slightly.
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
1. A: stick figures from 1 representative subject walking at approximately the same speed (1.07, 1.01, and 1.08 m s⫺1) in
the 3 different postures. Sticks are displayed in position space at a time resolution of 0.01 s. ●, position of the center of body mass
(CM) relative to the stick diagram. B: vertical position of CM from the gait cycles shown above is displayed as a function of time.
FIG.
292
R. GRASSO, M. ZAGO, AND F. LACQUANITI
Intersegmental coordination—time domain
The first two harmonics of the Fourier series expansion
account together for ⬎95% of the experimental variance of
each kinematic waveform (Table 3), and the shape of the
reconstructed waveforms faithfully reproduces the main features of the original ones. The first harmonic alone explains
most (⬎70%) of the original waveform variance.
TABLE 2. Correlation coefficients between elevation angle
waveforms
RE-KF
RE-KT
KF-KT
Thigh
Shank
Foot
0.87 ⫾ 0.05
0.94 ⫾ 0.04
0.94 ⫾ 0.08
0.98 ⫾ 0.01
0.95 ⫾ 0.04
0.98 ⫾ 0.02
0.98 ⫾ 0.01
0.98 ⫾ 0.03
0.95 ⫾ 0.06
Values are means ⫾ SD. n ⫽ 5.
Figure 4 shows the average time-normalized waveforms of
thigh, shank, and foot elevation angles from one subject. To
superimpose all traces on the same scale, the mean value was
subtracted from each one. The time changes of thigh elevation
lead those of shank elevation (the time delay between the 2
minima is indicated by the shaded areas in the figure). However, the time lead is consistently shorter in KF and KT than in
RE. In contrast, the amplitude of segmental excursions does
not vary substantially across the three conditions.
The amplitude transfer ratio (G) and the phase shift (⌽)
between pairs of adjacent lower limb segments (thigh-shank
and shank-foot) was quantified by means of harmonic analysis
(see METHODS) and is shown in Fig. 5 for the first harmonic
(mean ⫾ SD, from all subjects). One-way ANOVA showed
that the changes of the two gains across conditions were
marginally significant (0.01 ⬍ P ⬍ 0.05 and P ⫽ 0.05 for Gts
and Gsf, respectively). The difference between RE and KT was
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
FIG. 2. Time-normalized waveforms of
the elevation angles for all recorded body segments. For each subject and condition, we first
selected trials in the speed range 0.5–1.5 m
s⫺1; then we normalized the waveform time
scales from 0 to 100% of the gait cycle. We
ensemble averaged the time-normalized
waveforms as grouped by subjects and conditions. Finally, we computed the mean (thick
lines) and SD (thin lines) from the ensembleaverage waveforms of all subjects pooled together, for each conditions. x axis is percent
cycle length. 0% corresponds to the time of
maximum elevation in the limb axis (from hip
to lateral malleolus), which slightly precedes
the time of heel touch down. y scale in degrees. Schematic stick diagrams at the bottom
delimit stance phase.
POSTURAL GEOMETRY AND GAIT COORDINATION
293
FIG. 3. Ranges of excursion for thigh, shank, and foot elevation angles in the 3 experimental conditions [regular walking (RE),
knee-flexed walking (KF), and knee- and trunk-flexed walking (KT)]. Origin of the x axis indicates the vertical. Positive values
indicate forward elevation. - - -, mean value over the gait cycle. Data are from all subjects and experiments.
Planar covariation of limb elevation angles
We have found previously that in normal erect posture the
temporal changes of the elevation angles of the limb segments covary along a plane common to both the stance and
the swing phase (Bianchi et al. 1998b, Borghese et al. 1996,
Grasso et al. 1998). We now report that a planar law holds
also for walking with a bent posture. Figure 6 shows the gait
loops from the subject displayed in the stick diagram of Fig.
1. Each loop is obtained by plotting the elevation angle of
the thigh, shank, and foot one versus the others (3D position
TABLE
3.
space). Paths progress in time in the counter-clockwise
direction, heel strike and toe off phases corresponding to the
top and bottom of the loop, respectively. In all conditions,
the gait loop lies close to one plane. The grids correspond to
the (least-squares) best-fitting planes and to their intersection with the cubic wire frame of the angular coordinates.
The proportion of variance explained by the three eigenvalues ␭1, ␭2, and ␭3, from all trials is shown in Table 4. In all
conditions, the planar regression (the first 2 eigenvalues
together) accounts for ⬎98% of the data variance. However,
the loop tends to be contracted along its long axis in the two
flexed postures. This is reflected by a significantly lower
percent variance explained by ␭2 in KF and KT with respect
to RE (Table 4, F2,460⫽88.4, P ⬍ 0.001).
The orientation of the plane of angular covariation is not the
same in the three conditions, but it shows a rotation about the long
axis of the loop in KF and KT relative to RE. The direction cosine
of the plane normal with the thigh axis (u3t) is the parameter most
sensitive to this type of rotation (see Bianchi et al. 1998b). Figure
7 shows the value of u3t for the overall data set. This parameter
displayed a systematic increment for the two flexed postures with
respect to the regular one (0.22 ⫾ 0.07 for RE, 0.61 ⫾ 0.18 for KF
and 0.67 ⫾ 0.15 for KT, F2,447 ⫽ 509.1, P ⬍0.001). The difference between KF and KT was small but significant (Scheffé’s
post hoc test, P ⬍ 0.01).
Patterns of contact forces with the ground
Figure 8 shows the time-normalized vertical (top) and
longitudinal (bottom, anterior force is positive) ground reaction forces for the speed range 0.5–1.5 m s⫺1 from 11
trials of the same subject as in Fig. 1. In RE the time course
of vertical forces exhibits the classical pattern, with two
Harmonic analysis of kinematic waveforms
RE
KF
KT
Segment
First
First ⫹ Second
First
First ⫹ Second
First
First ⫹ Second
Thigh
Shank
Foot
95.8 ⫾ 1.2
79.3 ⫾ 2.4
71.1 ⫾ 3.3
99.3 ⫾ 0.3
98.0 ⫾ 0.9
97.4 ⫾ 0.9
90.8 ⫾ 2.8
82.1 ⫾ 3.0
71.8 ⫾ 4.6
98.0 ⫾ 2.3
97.6 ⫾ 0.8
96.1 ⫾ 1.9
91.9 ⫾ 3.1
87.5 ⫾ 1.2
75.2 ⫾ 1.6
98.9 ⫾ 0.7
99.0 ⫾ 0.4
98.2 ⫾ 0.5
Percent of total variance explained by the first 2 harmonics in individual waveforms (n ⫽ 5). Values are means ⫾ SD.
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
significant, whereas KF could not be distinguished from either
RE or KT (Scheffé’s post hoc test). The relative increase of Gsf
means that the amplitude of foot movements relative to that of
the shank tended to increase in bent postures relative to the
erect posture. The opposite tendency is showed by Gts.
For the phase shifts ⌽ts and ⌽sf the changes from RE to KF
and KT were prominent and highly significant (P ⬍ 0.0001). In
RE, the first harmonic of the thigh elevation waveform led
consistently that of shank elevation (by 53.3 ⫾ 5.1°, corresponding to 14.8 ⫾ 1.4% of T) and the shank led slightly the
foot (by 15.2 ⫾ 10.8°, 3.1 ⫾ 1.2% of T). In KF the phase shift
between the thigh and shank dropped (to 15.2 ⫾ 10.8°, 4.2 ⫾
3.0% of T) and the shift between shank and foot increased
approximately twofold (27.6 ⫾ 6°, 7.7 ⫾ 1.7% of T).
Ratios and phase shifts for the second harmonic displayed
little changes across the conditions for the shank-foot pair,
whereas for the thigh-shank pair only KF differed from either
RE or KT (which were similar to each other). Higher harmonics contributed little to the original waveform shape, and gain
and phases between adjacent segments were related inconsistently to the walking condition.
294
R. GRASSO, M. ZAGO, AND F. LACQUANITI
FIG. 4. Elevation angle waveforms (after subtraction of the mean value from each trace) from 1 representative subject. Shaded
areas help evaluate the time shift between thigh and shank. Negative peak of the thigh leads that of shank by 17% of cycle in RE.
Lead decreases to 10 and 9% of the cycle in KF and KT, respectively. Note that the waveform amplitude is rather constant across
the 3 conditions.
rameters: slope ⫽ ⫺6.9% per 1 m s ⫺1, ⫺7.2 and ⫺8.2 for RE,
KF, and KT; intercept ⫽ 65.6, 68.7, and 68.5%, for RE, KF,
and KT; correlation coefficients ⫽ 0.91, 0.83, and 0.61 for RE,
KF, and KT.
EMG activity
Figure 9, left, shows the EMG ensemble average of 10 gait
cycles from one subject walking at 1 m s⫺1 in each of the three
conditions. In all examined muscles, the mean activity tended
to be greater in KF and KT than in RE. GM mean integrated
EMG over the gait cycle increased 117 and 230% from RE to
KF and KT, respectively, RF increased 271 and 174%, VL 418
and 189%, BF 152 and 460%, TA 72 and 35%, and GCL 278
and 145%. Not only the mean amplitude but also the pattern of
activity differed across conditions. RF displayed prominent
activity in the early swing phase of KT condition compared
with RE. BF displayed a prominent activity during the stance
phase of KF and even more so in KT, whereas in RE this
muscle is typically active during the swing phase mainly.
Task-dependent changes in the activity profile also were observed in the other tested muscles. As a consequence, the
squared correlation coefficients between postural conditions
tend to be lower than those computed for the corresponding
FIG. 5. A: mean amplitude transfer ratio (G) between the 1st harmonics of adjacent lower limb segments (ts ⫽ thigh:shank, sf ⫽
shank:foot). Differences are marginally significant (ANOVA, P ⫽ 0.05). Post hoc tests revealed that only the difference between
KT and RE is significant (P ⬍ 0.05). B: mean phase shift (⌽) between the 1st harmonics of adjacent lower limb segments.
Differences are highly significant (P ⬍ 0.001). Left scale in degrees, right scale in percent of cycle length. Error bars represent 1
SD in both A and B. Data are from all subjects.
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
main peaks (one in early stance and the second one in late
stance) separated by a trough in midstance due to partial
unloading from the contralateral limb (see Borghese et al.
1996; Winter 1991). The first peak and the trough become
more pronounced with increasing speed. By contrast, in KF
and KT the profile of vertical forces tends to be monophasic,
with peaks and trough much less pronounced than in RE.
Moreover this pattern does not vary with increasing speed as
much as in RE (the slopes and r coefficients of the regression line relating the walking speed to the peak force are
109 ⫾ 25 N m⫺1 s, r ⫽ 0.81 for RE, 40 ⫾ 10 N m⫺1 s, r ⫽
0.77 for KF, and 41 ⫾ 25 N m⫺1 s, r ⫽ 0.53 for KT). The
amplitude of the plateau in bent postures is lower than the
two peaks displayed in RE especially at higher speeds.
As for the longitudinal reaction force, in RE it exhibits an
ordered amplitude increase with speed both in the deceleration
(after heel touch-down) and the acceleration (before toe takeoff) phase (r ⫽ 0.94 for the peak force vs. speed linear
regression). In KF and KT, this force has a lower peak amplitude than in RE and the profiles are much more variable across
trials (r ⫽ 0.46 and 0.41 for KF and KT, respectively). Also the
stance phase duration displays a less consistent relation with
speed as compared with RE. The linear regression between
stance percent duration and speed yielded the following pa-
POSTURAL GEOMETRY AND GAIT COORDINATION
295
FIG. 6. Mean gait loops from 1 subject across
the 3 experimental conditions. A loop represents
1 gait cycle and is obtained by plotting the
average thigh waveform vs. the shank and foot
waveforms (after mean values subtraction).
Paths progress in time in the counterclockwise
direction, heel contact and toe-off phases corresponding approximately to the top and bottom of
the loop, respectively. Interpolation planes result
from orthogonal planar regression fitting. Note
that the loop lies close to the plane in all conditions but it displays different orientations in KF
and KT vs. RE.
DISCUSSION
Kinematic waveforms
The purpose of the present study was to verify how walking
with a bent posture affects the generation of gait motor patterns. In bent postures (KF and KT), the lower limbs and the
trunk are flexed as compared with the erect posture (RE). In
spite of the drastic changes in postural geometry, the oscillations of each limb segment relative to the vertical are very
similar across different postures.
The adaptation of the intersegmental coordination to the
different postures is achieved by changing the phase delays
between the motion of different limb segments. Thus the phase
lead of thigh oscillations relative to shank oscillations is systematically shorter with bent postures than with an erect posture, whereas the phase lead of the shank relative to the foot
TABLE 4. Percent variance explained by the eigenvalues of the
elevation angle covariance matrix
␭1
␭2
␭3
RE
KF
KT
87.2 ⫾ 2.8
11.6 ⫾ 2.7
1.2 ⫾ 0.4
90.0 ⫾ 3.7
8.4 ⫾ 3.9
1.6 ⫾ 0.6
91.5 ⫾ 2.3
7.1 ⫾ 2.3
1.4 ⫾ 0.5
Values are means ⫾ SD.
oscillations is longer in the former than in the latter. As a result,
there is an approximately constant phase décalage (of ⬃20°
corresponding to ⬇5% of the gait cycle) between adjacent limb
segments in the proximal-to-distal direction in bent walking.
The behavior of a flexed lower limb then can be equated to that
of a rolling wheel with the pivot corresponding to the knee
joint.
Somewhat similar results have been reported previously in
the comparison of level walking versus crouched walking in
cats (Trank et al. 1996). (The latter is an attitude typical of
stalking behavior in this species.) The profiles of hindlimb
kinematics were rather similar between the two forms of locomotion either for the hip, knee, ankle, or the metatarso-phalangeal joint angles (see Fig. 3 of Trank et al. 1996) in spite of
prominent changes in their mean values and ranges of excursion. The timing of motion reversals (from flexion to extension) differed at some joints.
Therefore postural adaptation acts on the same variable,
intersegmental kinematic phase, which has been shown previously to be centrally modified as a function of changes in
walking speed (Bianchi et al. 1998b) or walking direction
(Grasso et al. 1998). Intersegmental phase plays a role of
global control variable similar to that previously advocated for
the network of coupled oscillators involved in the generation of
locomotion (see Grillner 1981; Pearson 1993; Rossignol 1996).
According to the kinematic view we have exposed in the
introduction, each unit oscillator would directly control a limb
segment, alternately generating forward and backward oscillations of the segment. Intersegmental coordination would be
achieved by coupling unit oscillators with a variable phase.
Variable coupling could result, for instance, by changing the
synaptic strength (or polarity) of the relative spinal connections. Supraspinal centers may drive or modulate functional
sets of coordinating spinal interneurons to generate different
walking modes. We shall take up the issue of the mechanisms
of postural modulation of the coordinating network in a subsequent section.
Planar covariation of elevation angles
The changes in the elevation angles of the thigh, shank, and
foot covary along a plane common to both the stance and the
swing phase. More than 98% of the data variance is explained
by the planar regression in all postural conditions. However,
the orientation of the plane differs across conditions: the plane
is rotated about the long axis of the loop with bent postures as
compared with the erect posture. The direction cosine of the
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
changes in the elevation angles (right). This was true for all
subjects.
Not only the patterns of activation of individual muscles but
also the time sequence of activation of different muscles, or
muscle synergies, varied substantially as a function of body
posture. To quantify this kind of synergies, we computed the
CCF between pairs of EMG ensemble averages. The CCFs for
the three tasks are superimposed in Fig. 10. The patterns of
muscle synergies differ across conditions as indicated by the
different shape of the CCF. For instance, in RE and KF the
peak in the pair RF-VL occurs at ⬃0% of the gait cycle,
indicating agonistic activity. By contrast, in KT the peak occurs at ⬃40% lead, indicating quasi-antagonistic activity in
these two muscles. Most other pairings of muscles exhibit
significant time shifts in either the maximum or the minimum
of the CCF in KF and KT conditions relative to the RE
condition. This is shown by the scatter plot of these values in
Fig. 11. Several data points (13/30 and 18/30 for the maximum
and minimum, respectively) fall outside the ⫾10% cycle duration band (shaded area).
296
R. GRASSO, M. ZAGO, AND F. LACQUANITI
FIG. 8. Compressed arrays of vertical (A) and longitudinal (B) ground reaction force profiles for the 3 experimental conditions
in 1 subject as a function of walking speed (y axis). Eleven trials at speeds ranging from 0.5 to 1.5 m s⫺1 (y axis) are included.
Time scale of each profile was normalized over the gait cycle. x axis represents percent of cycle length. Forces (z axis) are expressed
in Newton (N).
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
FIG. 7. Direction cosine u3t from the overall data set. This parameter represents the orientation of the normal to the best-fitting
plane relative to the thigh axis. Its theoretical range goes from ⫺1 to 1. Values in KF and KT are systematically higher than in RE.
Initials refer to the 5 subjects.
POSTURAL GEOMETRY AND GAIT COORDINATION
297
values of u3t for KF and KT are systematically greater than the
corresponding values for RE. This result is consistent with the
observation that the level of muscle activity (and therefore of
overall energy expenditure) is systematically higher with bent
posture than with erect posture (see next section).
Adaptation of limb kinetics to bent postures
plane normal with the thigh axis (u3t) is the parameter most
sensitive to this type of rotation (see Bianchi et al. 1998b). We
have shown previously that changes of u3t reflect changes in
the phase coupling between the limb segment elevation angles
(Bianchi et al. 1998b). Therefore the demonstration of changes
of u3t in KF and KT relative to RE is consistent with the
observation that the adaptation of locomotion to these postural
changes affects the intersegmental phase coupling.
We have shown previously that the parameter u3t generally
correlates with the mechanical energy expenditure during erect
walking: the greater is the value of u3t in any given trial, the
greater is the corresponding net mechanical power output over
a gait cycle (Bianchi et al. 1998a). Here we found that the
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
FIG. 9. Left: ensemble averages of electromyographic (EMG) activity from
1 subject walking at the speed of 1 m s⫺1 are superimposed for RE (shaded
envelopes), KF (dotted) and KT (thick). GM, gluteus maximus; RF, rectus
femoris; VL, vastus lateralis; BF, biceps femoris; TA, tibialis anterior; and
GCL, gastrocnemius lateralis. Right: ensemble averages of elevation angles
from the same data set are superimposed for RE (thin solid), KF (dotted), and
KT (thick solid). x axes are percent of gait cycle. Stick diagrams at the bottom
delimit stance phase. Insets: squared correlation coefficients between the
indicated pairs.
The patterns of ground reaction forces and muscle activity
change systematically as a function of walking posture. The
time profile of the vertical reaction forces is multiphasic in
erect posture but monophasic and of lower amplitude in bent
postures (Li et al. 1996). In the latter case, in fact, the impact
shock with the ground is attenuated, especially by the knee
joint (McMahon et al. 1987). The monophasic profile is reminiscent of that measured during erect running. However, in
bent walking the percent duration of stance is much longer than
in running, so that McMahon et al. (1987) pointed out that
knee-flexed locomotion is somewhat intermediate between
walking and running. The peak amplitude of the longitudinal
forces is greater in erect than in bent walking. In general,
reaction forces are more variable in bent postures.
The level of muscle activity is systematically higher with
bent posture than with erect posture. When the limb joints are
flexed and displaced away from the main axis of the limb, the
mechanical advantage of the muscles to support body weight
during stance and to flex the limb during swing is reduced in
comparison with that of the erect posture, and more muscle
activity is needed to generate the appropriate joint torques. An
increment of muscle activity also is required because the recovery of mechanical energy by means of the pendulum mechanism is less effective, and more muscular work needs to be
done to move the center of body mass and to swing the limbs.
In erect walking, in fact, the center of body mass oscillates
above the supporting limb like an inverted pendulum, thereby
limiting energy expenditure by means of an exchange of the
forward kinetic energy with the gravitational potential energy
(Cavagna et al. 1977). By contrast, in bent postures the center
of body mass tends to be displaced downward and forward, and
its oscillations are reduced because the legs cannot fully extend
(see Fig. 1). Thus the exchange of kinetic and potential energy
is more limited. Indeed, human knee- and hip-flexed walking
has been found to produce in-phase fluctuations in potential
and kinetic energies (Li et al. 1996) rather than the out-ofphase fluctuations typical of erect locomotion.
Not only the amplitude but also the pattern of activation of
individual muscles, as well as the time sequence of muscle
synergies, vary substantially as a function of body posture.
This was shown by considering the cross-correlation functions
of pairs of EMG ensemble averages. In particular we observed
considerable shifts in either the maximum or the minimum of
several cross-correlation functions, indicating the transition
from an agonistic activity in a given postural condition to a
quasi-antagonistic activity in the same pair of muscles in a
different postural condition (see Figs. 10 and 11).
Trank et al. (1996) also found that muscle patterns for cats
walking knee-flexed display some changes with respect to
normal walking. The burst duration for three primary knee,
ankle, and digit flexor muscles are longer and EMG amplitude
has often a higher amplitude according to the increased range
and duration of flexion during the swing phase. Furthermore
298
R. GRASSO, M. ZAGO, AND F. LACQUANITI
FIG. 10. Cross-correlograms between pairs of EMG data
from 1 subject. Correlation values are plotted as a function of
time delay (expressed as percent of the gait cycle). A positive
value of cross-correlation at a positive time delay indicates
that the activation of the column muscle leads that of the row
muscle, whereas a positive cross-correlation at a negative
delay indicates a time lag of the column muscle relative to the
row muscle.
to normal walking also have been described for down-slope
(Smith et al. 1998) and up-slope walking (Carlson-Kuhta et al.
1998). The extensive investigation of different forms of locomotion in cats (normal forward, backward, crouched, up slope,
FIG. 11. Abscissas correspond to the lead (positive values) or lag (negative values) of the maximum (left) or the minimum
(right) of the cross-correlation functions (CCF) in RE condition for the indicated muscle pairs. Ordinates correspond to the values
for KF (unfilled symbols) or KT (filled symbols). Data points on the diagonal line indicate no change between conditions, and the
shaded area depicts the ⫾10% cycle duration band. Time delays are expressed as percent of the gait cycle.
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
two muscles that show mainly swing-related activity in normal
walking (digit extensors) have distinct stance-related bursts in
crouched walking, similar to what we found for BF. Similar
changes in the functional role of specific muscles with respect
POSTURAL GEOMETRY AND GAIT COORDINATION
Integrated control of gait and posture
Postural adaptation conserves the general kinematic waveforms with an appropriately tuned intersegmental phase and
allocates specific patterns of muscle activity as a function of
the required kinematic coordination. Integrated control of gait
and posture is made possible because these two motor functions share some common organizational principles (Lacquaniti et al. 1997; Massion 1992). First, the frame of reference for
the kinematic coordination for both postural responses and
locomotion seems to be anchored to the vertical. Second, a
control of the position of the center of body mass for static or
dynamic equilibrium is involved in both gait (Cavagna et al.
1977) and posture (Massion 1992). Also the planar law of
intersegmental kinematic coordination applies to both tasks.
The planar law involved in postural responses has been reported in previous work (Lacquaniti and Maioli 1994a,b). The
changes in the geometric configuration of the forelimbs and
hindlimbs in cats pitched by variable tilts of the support platform lie close to one plane. This planar covariation is not
affected by adding loads that shift the animal’s center of mass.
Intersegmental coordination of lower limbs and trunk also has
been described for human postural responses evoked by perturbations of the support platform (Nashner and McCollum
1985) or by axial bending movements (Massion 1992). It is not
surprising to find that similar laws of intersegmental coordination apply to the control of posture and locomotion. Locomotion must assure a forward progression compatible with dynamic equilibrium, adapting to potentially destabilizing factors
(e.g., changes in body posture or load, uneven terrain, obstacles, etc.) in an anticipatory fashion by means of coordinated
synergies of upper limbs, trunk, and lower limbs (Dietz et al.
1987; Hirschfeld and Forssberg 1991).
The concept of an integrated control of posture and locomotion also stems from neurophysiological data. The stimulation
of specific areas in the brain stem and hypothalamus in freely
moving cats causes the animal to adopt different locomotor and
postural styles (Mori et al. 1989). Flexed locomotion can be
evoked in cats by stimulating the lateral hypothalamus (Mori et
al. 1989). Also selective lesions in Deiters’ nucleus result in a
significant decrease in extensor muscle activity during locomotion (Orlovsky 1972). Repetitive stimulation of medial longitudinal fasciculus neurons may disrupt fictive locomotor
rhythms (Floeter et al. 1993; Gossard et al. 1996). The link
between posture and gait control is also mediated through
afferent feedback. Thus Shik and Orlovsky (1976) proposed
that vestibulospinal and reticulospinal pathways carrying information from both the sensory feedback and from the state of
the spinal circuitry may operate during locomotion to “sculpt”
the output patterns generated in the segmental CPGs (see also
Armstrong 1988; Hasan and Stuart 1988).
A specific role in setting the spatial framework for the
control of the postural geometry of the trunk and the coordination of lower limb segments is provided by the basal ganglia
(Garcia-Rill 1986). A recent study addressed this issue in
patients with Parkinson’s disease (Grasso et al. 1999). Patients
could be switched ON by means of either a D1-D2 receptor
agonistic drug (apomorphine) or by globus pallidum internum
(GPi) electrical stimulation. It was found that the inclination of
the trunk with respect to the vertical, the waveforms of the
elevation angles, and the planar law of angular covariation
change all in parallel in the transition from the OFF to the ON
condition.
We thank D. Prissinotti for skillful technical help with the experiments.
The financial support of Telethon-Italy (Grant 1159) is gratefully acknowledged.
Address for reprint requests: F. Lacquaniti, IRCCS Santa Lucia, via
Ardeatina 306, 00179 Rome, Italy.
Received 6 May 1999; accepted in final form 14 September 1999.
REFERENCES
ARMSTRONG, D. B. Review lecture: the supraspinal control of mammalian
locomotion. J. Physiol. (Lond.) 405: 1–37, 1988.
BIANCHI, L., ANGELINI, D., AND LACQUANITI, F. Individual characteristics of
human walking mechanics. Pflügers Arch. 436: 343–356, 1998a.
BIANCHI, L., ANGELINI, D., ORANI, G. P., AND LACQUANITI, F. Kinematic
coordination in human gait: relation to mechanical energy cost. J. Neurophysiol. 79: 2155–2170, 1998b.
BIEWENER, A. A. Biomechanics of mammalian terrestrial locomotion. Science
250: 1097–1103, 1990.
BIZZI, E., MUSSA-IVALDI, F. A., AND GISZTER, S. Computations underlying the
execution of movement: a biological perspective. Science 253: 287–291,
1991.
BORGHESE, N. A., BIANCHI, L., AND LACQUANITI, F. Kinematic determinants of
human locomotion. J. Physiol. (Lond.) 494: 863– 879, 1996.
BOSCO, G., RANKIN, A., AND POPPELE, R. A representation of passive hindlimb
postures in cat spinocerebellar activity. J. Neurophysiol. 76: 715–726, 1996.
BURLEIGH, A. L., HORAK, F. B., AND MALOUIN, F. Modification of postural
responses and step initiation: evidence for goal-directed postural interactions. J. Neurophysiol. 72: 2892–2902, 1994.
CARLSON-KUHTA, P., TRANK, V. T., AND SMITH, J. Forms of forward quadrupedal locomotion. II. A comparison of posture, hindlimb kinematics, and
motor patterns for upslope and level walking. J. Neurophysiol. 79: 1687–
1701, 1998.
CAVAGNA, G., HEGLUND, N. C., AND TAYLOR, C. R. Mechanical work in
terrestrial locomotion: two basic mechanisms for minimizing energy expenditure. Am. J. Physiol. 233 (Regulatory Integrative Comp. Physiol. 2):
R243–R261, 1977.
CROMPTON, R. H., YU, L., WEIJIE, W., GUNTHER, M., AND SAVAGE, R. The
mechanical effectiveness of erect and “bent-hip bent-knee” bipedal walking
in Australopithecus afarensis. J. Hum. Evol. 35: 55–74, 1998.
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
and down slope) (reviewed in Zernicke and Smith 1996) has
made clear that the coupling between muscle groups is an
important functional variable and may be modified centrally
depending on the behavioral context. Even robust stance-sustaining synergies, which rely on the excitatory coupling between hip and knee extensor units, may be switched off in
particular cases (i.e., in down-slope walking). To explain these
transitions to different synergies without dropping the schema
of an array of joint-related CPG units (Grillner 1981), Zernicke
and Smith (1996; see also Prochazka 1996) proposed that the
rule for setting interunit connections may be issued by supraspinal control following conditional statements related to
the behavioral goal.
Our view is that, in response to the kinematic reference
signals, the appropriate muscle synergies are determined in a
flexible manner to fit the current mechanical constraints. Indeed, the relationship between muscle activity and whole body
mechanics seems too variable and complex to allow a direct
control of the latter without an intermediate kinematic representation. Neurophysiological data indicate that intermediate
kinematic representations may be found in both premotor
(Bizzi et al. 1991) and proprioceptive sensory areas of the
spinal cord (Bosco et al. 1996) as well as in somatosensory and
motor cortices (Helms Tillery et al. 1996; Lacquaniti et al.
1995; Scott and Kalaska 1997).
299
300
R. GRASSO, M. ZAGO, AND F. LACQUANITI
MCMAHON, T. A., VALIANT, G., AND FREDERICK, E. C. Groucho running.
J. Appl. Physiol. 62: 2326 –2337, 1987.
MORI, S. Integration of posture and locomotion in acute decerebrate cats in
awake, freely moving cats. Prog. Neurobiol. 28: 161–195, 1987.
MORI, S., KAWAHARA, K., AND SAKAMOTO, T. Supraspinal aspects of locomotion in the mesencephalic cat. Symp. Soc. Exp. Biol. 37: 445– 468, 1983.
MORI, S., NISHIMURA, H., KURAKAMI, C., YAMAMURA, T., AND AOKI, M.
Controlled locomotion in the mesencephalic cat: distribution of facilitatory
and inhibitory regions within pontine tegmentum. J Neurophysiol. 41:
1580 –1591, 1978.
MORI, S., SAKAMOTO, T., OHTA, Y., TAKAKUSAKI, K., AND MATSUYAMA, K.
Site-specific postural and locomotor changes evoked in awake, freely moving intact cats by stimulating the brainstem. Brain Res. 505: 66 –74, 1989.
MOUCHNINO, L., AURENTY, R., MASSION, J., AND PEDOTTI, A. Is the trunk a
reference frame for calculating leg position? Neuroreport 4: 125–127, 1993.
NASHNER, L. M. AND MCCOLLUM, G. The organization of human postural
movements: a formal basis and experimental synthesis. Behav. Brain Sci. 8:
135–172, 1985.
ORLOVSKY, G. N. The effect of different descending systems on flexor and
extensor activity during locomotion. Brain Res. 40: 359 –371, 1972.
PEARSON, K. G. Common principles of motor control in vertebrates and
invertebrates. Annu. Rev. Neurosci. 16: 265–297, 1993.
PROCHAZKA, A. The fuzzy logic of visuomotor control. Can. J. Physiol.
Pharmacol. 74: 456 – 462, 1996.
ROSSIGNOL, S. Neural control of stereotypic limb movements. In: Handbook of
Physiology. Exercise: Regulation and Integration of Multiple Systems.
Washington, DC: Am. Physiol. Soc., 1996, sect. 12, p. 173–216.
SCOTT, S. H. AND KALASKA, J. F. Reaching movements with similar hand paths
but different arm orientations. I. Activity of individual cells in motor cortex.
J. Neurophysiol. 77: 826 – 852, 1997.
SHEN, L. AND POPPELE, R. E. Kinematic analysis of cat hindlimb stepping.
J. Neurophysiol. 74: 2266 –2280, 1995.
SHIK, M. L. AND ORLOVSKY, G. N. Neurophysiology of locomotor automatism.
Physiol. Rev. 56: 456 –501, 1976.
SMITH, J. L., CARLSON-KUTHA, P., AND TRANK, T. V. Forms of forward
quadrupedal locomotion. III. A comparison of posture, hindlimb kinematics,
and motor patterns for downslope and level walking. J. Neurophysiol. 79:
1702–1716, 1998.
SPOOR, F., WOOD, B., AND ZONNEVELD, F. Implications of early hominid
labyrinthine morphology for evolution of human bipedal locomotion. Nature 369: 645– 648, 1994.
TRANK, T. V., CHEN C., AND SMITH, J. L. Forms of forward quadrupedal
locomotion. I. A comparison of posture, hindlimb kinematics, and motor
patterns for normal and crouched walking. J. Neurophysiol. 76: 2316 –2326,
1996.
VANDEN-ABEELE, J. Comments on the functional asymmetries of the lower
extremities. Cortex 16: 325–329, 1980.
WINTER, D. A. The Biomechanics and Motor Control of Human Gait: Normal,
Elderly and Pathological. Waterloo, Ontario: Waterloo Biomechanics Press,
1991.
ZATSIORSKY, V., SELUYANOV, V., AND CHUGUNOVA, L. In vivo body segment
inertial parameters determination using a gamma-scanner method. In: Biomechanics of Human Movement: Applications in Rehabilitation, Sports and
Ergonomics, edited by N. Berme and A. Cappozzo. Worthington, OH:
Bertec, 1990, p. 186 –202.
ZERNICKE, R. F. AND SMITH, J. L. Biomechanical insights into neural control of
movement. In: Handbook of Physiology. Exercise: Regulation and Integration of Multiple Systems. Washington, DC: Am. Physiol. Soc., 1996, sect.
12, p. 293–330.
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 17, 2017
D’AMICO, M. AND FERRIGNO, G. Technique for the evaluation of derivatives
from noisy biomechanical displacement data using a model-based bandwidth-selection procedure. Med. Biol. Eng. Comput. 28: 407– 415, 1990.
DARLING, W. G. AND MILLER, G. F. Perception of arm orientation in threedimensional space. Exp. Brain Res. 102: 495–502, 1995.
DIETZ, V., QUINTERN, J., AND SILLEM, M. Stumbling reactions in man: significance of proprioceptive and pre-programmed mechanisms. J. Physiol.
(Lond.) 368: 149 –163, 1987.
FERRIGNO, G., BORGHESE, N. A., AND PEDOTTI, A. Pattern recognition in 3D
automatic human motion analysis. ISPRS J. Photogrammetry Remote Sensing 45: 227–246, 1990.
FLOETER, M. K., SHOLOMENKO, G. N., GOSSARD, J. P., AND BURKE, R. E.
Disynaptic excitation from the medial longitudinal fasciculus to lumbosacral
motoneurons: modulation by repetitive activation, descending pathways,
and locomotion. Exp. Brain Res. 92: 407– 419, 1993.
GARCIA-RILL, E. The basal ganglia and the locomotor regions. Brain Res. Rev.
11: 47– 63, 1986.
GOSSARD, J. P., FLOETER, M. K., DEGTYARENKO, A. M., SIMON, E. S., AND
BURKE, R. E. Disynaptic vestibulospinal and reticulospinal excitation in cat
lumbosacral motoneurons: modulation during fictive locomotion. Exp. Brain
Res. 109: 277–288, 1996.
GRASSO, R., BIANCHI, L., AND LACQUANITI, F. Motor patterns for human gait:
backward versus forward locomotion. J. Neurophysiol. 80: 1868 –1886,
1998.
GRASSO, R., PEPPE, A., STRATTA, F., ANGELINI, D., ZAGO, M., STANZIONE, P.,
AND LACQUANITI, F. Basal ganglia and gait control: apomorphine administration and internal pallidum stimulation in Parkinson’s disease. Exp. Brain
Res. 126: 139 –148, 1999.
GRILLNER, S. Control of locomotion in bipeds, tetrapods, and fish. In: Handbook of Physiology. The Nervous System. Motor Control. Bethesda, MD:
Am. Physiol. Soc., 1981, sect. 1, vol. 2, part 1, p. 1179 –1236.
HASAN, Z. AND STUART, D. G. Animal solutions to problems of movement
control: the role of proprioceptors. Annu. Rev. Neurosci. 11: 199 –223, 1988.
HELMS TILLERY, S. I., SOECHTING, J. F., AND EBNER, T. J. Somatosensory
cortical activity in relation to arm posture: nonuniform spatial tuning.
J. Neurophysiol. 76: 2423–2438, 1996.
HIRSCHFELD, H. AND FORSSBERG, H. Phase-dependent modulations of anticipatory postural activity during human locomotion. J. Neurophysiol. 66:
12–19, 1991.
JANKOWSKA, E. AND EDGLEY, S. Interactions between pathways controlling
posture and gait at the level of spinal interneurones in the cat. Prog Brain
Res. 97: 161–171, 1993
LACQUANITI, F., GRASSO, R., AND ZAGO, M. Motor patterns for walking. News
Physiol. Sci. 14: 168 –174, 1999.
LACQUANITI, F., GUIGON, E., BIANCHI, L., FERRAINA, S., AND CAMINITI, R.
Representing spatial information for limb movement: role of area 5 in the
monkey. Cereb. Cortex 5: 391– 409, 1995.
LACQUANITI, F. AND MAIOLI, C. Coordinate transformations in the control of
cat posture. J. Neurophysiol. 72: 1496 –1515, 1994a.
LACQUANITI, F. AND MAIOLI, C. Independent control of limb position and
contact forces in cat posture. J. Neurophysiol. 72: 1476 –1495, 1994b.
LACQUANITI, F., MAIOLI, C., BORGHESE, N., AND BIANCHI, L. Posture and
movement: coordination and control. Arch. Ital. Biol. 135: 353–367, 1997.
LI, Y., CROMPTON, R. H., ALEXANDER, R. M., GUNTHER, M. M., AND WANG,
W. J. Characteristics of ground reaction forces in normal and chimpanzeelike bipedal walking by humans. Folia Primatol. (Basel) 66: 137–159, 1996.
MASSION, J. Movement, posture and equilibrium: interaction and coordination.
Prog. Neurobiol. 38: 35–56, 1992.
MASSION, J., POPOV, K., FABRE, J. C., RAGE, P., AND GURFINKEL, V. Is the erect
posture in microgravity based on the control of trunk orientation or center of
mass position? Exp. Brain Res. 114: 384 –389, 1997.