N. Kobayashi et al.

Neuroscience Letters 399 (2006) 141–146
Posterior-anterior body weight shift during stance period studied by
measuring sole-floor reaction forces during healthy and
hemiplegic human walking
Nobuyoshi Kobayashi a,∗ , Tateo Warabi a , Masamichi Kato a , Kiichi Kiriyama a ,
Toshikazu Yoshida a , Susumu Chiba b
a
Clinical Brain Research Laboratory, Toyokura Memorial Hall, Sapporo Yamanoue Hospital, Yamanote 6-9-1-1, Sapporo 063-0006, Japan
b Department of Neurology, School of Medicine, Sapporo Medical University, Minami-1, Nishi 16, Sapporo 060-8543, Japan
Received 7 December 2005; received in revised form 4 January 2006; accepted 24 January 2006
Abstract
Posterior-anterior body weight shift during stance phase of human overground locomotion was investigated by recording sole-floor reaction
force from five anatomically discrete points with strain gauge transducers of 14 mm diameter attached firmly to the sole of bare foot. At first the
subject was asked to walk straight on the laboratory floor at his/her preferred velocity. Then the subject was asked to walk curved path of about
1 m radius. For kicking off the body at the end of stance phase, sole-floor reaction force from 3rd metatarsal was stronger than 1st metatarsal or
5th metatarsal during the straight walking, thus body weight shift is represented from heel to 3rd metatarsal line. When walking along a curved
path, two types of strategies were recognized; a group of subjects walked leaning to inner leading foot during stance period as judged by stronger
forces recorded from 5th metatarsal combined with stronger force from 1st metatarsal of outer trailing foot. Another group of subjects showed
almost the same patterns either in the straight and curved walking, suggesting the subjects changed direction of the foot during the immediately
previous swing phase to the tangent direction of the curve and placed the foot without leaning the body weight to either direction. Hemiplegic
patients showed strikingly different distribution of sole-floor reaction forces from the five points; strongest forces were recorded from 3rd and 5th
metatarsals combined with reduced reaction force from heel, therefore characteristic y-vector patterns were observed.
© 2006 Elsevier Ireland Ltd. All rights reserved.
Keywords: Straight walking; Circular walking; Sole-floor reaction force; Healthy human subjects; Hemiplegic patient
Locomotion along a straight path, either on laboratory floor or
on treadmill, is studied in usual set-up of experiment. Previous reports from our laboratory [9,10,16,17] describe various
aspects of human locomotion investigated by recording solefloor reaction force from anatomically discrete five points of
sole.
In everyday life humans are frequently faced with changes in
path direction, which can be either anticipated or unexpected.
Courtine and Schieppati [2,3] investigated human walking along
a curved path, and mentioned that the increased duration of the
stance phase of the inner foot was accompanied by leaning of the
trunk towards the inner side of the walking path [2]. Furthermore
the authors reported that the decreased duration of the stance
∗
Corresponding author. Tel.: +81 11 621 1200; fax: +81 11 644 0435.
E-mail address: yamanoue@alles.or.jp (N. Kobayashi).
0304-3940/$ – see front matter © 2006 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.neulet.2006.01.042
phase of the outer limb was probably related to the obligatory
overall higher velocity of the outer limb, given the longer path to
be traveled with respect to the inner limb [3]. Hase and Stein [8]
investigated the mechanisms involved in rapid turning during
human walking, and found two walking strategies, spin turn and
step turn, were used depending on which leg was placed in front
for breaking.
During walking along a curved path in the previous study
from our laboratory posterior-anterior body weight shift during
stance phase was estimated by obtaining y-vector which was
calculated from force curves of calcaneus and 3rd metatarsal
[10]. The aim of the present study is to investigate how
humans “select” locomotor pattern under (1) straight walking, (2) clockwise turning, and (3) anti-clockwise turning. For
this purpose, posterior-anterior shift of body weight during
stance phase was systematically studied by obtaining y-vector
from (1) calcaneus and 3rd metatarsal, (2) calcaneus and 1st
142
N. Kobayashi et al. / Neuroscience Letters 399 (2006) 141–146
metatarsal, and (3) calcaneus and 5th metatarsal in the healthy
subjects.
Furthermore, posterior-anterior body weight shift during
stance phase of hemiplegic patients was also investigated compared with healthy subjects.
There are two subject groups: healthy control groups and
hemiplegic patients. (1) Healthy control subjects are 10 adults
without known neurological disorders or discrepancies between
right and left lower limbs that could affect the experimental outcome (5 males: 21–31 years, average 24.2 ± 4.0 years; 5 females:
21–33 years, average 23.6 ± 5.2 years). These subjects are the
same subjects who were studied on different aspects of locomotion in an earlier publication [10]. (2) Five patients showed
hemiplegic limbs due to cerebral hemorrhage (47–71 years, 3
male and 2 female).
Both groups of the subjects were fully informed of the purpose and procedures including estimated time of the measurement and consented voluntarily to participate the measurement.
The study was in accordance with Declaration of Helsinki [18]
and was approved by Ethics Committee of Yamanoue Hospital.
The method to record the sole-floor reaction force was fully
documented in the earlier publications [9,16] from our laboratory. Therefore, the method is briefly described here. Strain
gauge load cells (NEC, San’Ei, type 9E01-L42-500N, diameter
14 mm, thickness 4 mm, measuring range 0–500 Newton) were
used. The load cells were securely adhered with double faced
adhesive tape on the skin of five points of the sole: (1) medial
process of calcaneus, (2) head of 5th metatarsal, (3) head of 3rd
metatarsal, (4) head of 1st metatarsal and (5) middle of great
toe. The subject wore a cotton socks to secure the load cells.
Leading wires from the gauges were attached on foot and leg
with surgical tapes and connected to the power supply (about
50 gr) which was held around waist with a belt. Outputs from
the load cells were fed to multichannel pen oscillograph (NEC
San’Ei Rectiholy 8K20) through amplifiers and low-pass filters
(frequency range; 0–200 Hz), simultaneously stored on magnetic tapes. The analogue outputs from the gauges were digitized
with a microcomputer (NEC San’Ei Signal Processor 7T18T) at
a sampling frequency of 250 Hz and were fed into a personal
computer.
At first the subjects were asked to walk on the laboratory floor
with the preferred velocity for 10 m. This walking was repeated
for five or six times in order to collect experimental data from the
five points along with obtaining the average walking velocity.
The subjects were asked to maintain gaze straight ahead with
head position about 15◦ downward during straight walking, as
in our previous study [16].
Then the subjects were asked to walk on circular path with
about the same speed. For walking along circular path, the
radius of curvature was about 1 m, and the subject was asked
to walk both clockwise and counter-clockwise for about 20
steps each. No specific instruction was given to the subjects
regarding gaze or head or body orientation during circular
walking. One subject complained feeble vertigo during circular walking, then measurement was instantly stopped for this
subject. For the hemiplegic patients, only straight walking was
investigated.
Method of calculation of posterior-anterior balance (or yvector) of body shift was described in an earlier publication from
our laboratory [9,10]. As described above, the analogue outputs from the gauges were digitized at a sampling frequency of
250 Hz and were stored in a personal computer. After measurement session three or four y-vectors were obtained off-line by
subtracting digitized force curve of the calcaneus from digitized
force curve of (1) 3rd metatarsal (C → M3), (2) 1st metatarsal
(C → M1) and (3) 5th metatarsal (C → M5) for the healthy subjects and in some patients (4) great toe (C → G). x-vector was
evaluated by subtracting digitized force curve of 5th metatarsal
from digitized force curve from 1st metatarsal.
The preferred walking velocities were 3.5 km/h for two subjects, 3.6 km/h for one subject, 4.0 km/h for five subjects and
4.5 km/h for two subjects.
In our earlier reports, posterior-anterior body weight shift
was discussed by obtaining y-vector by subtracting force curves
Fig. 1. Distribution of sole-floor reaction force from the five points of one
representative step of straight walking at 100 ms intervals. Time points were
selected since touchdown of calcaneus. Stance phase proceeds from the left
lowest (100 ms) and kicks off from the right-upper (700 ms) foot prints. Lengths
of columns represent relative strength of sole-floor reaction force (e.g. 17.1 kg at
calcaneus of 100 ms). Subject, T.M. male 20 years, height 176 cm body weight
60 kg. Walking velocity 4.0 km/h.
N. Kobayashi et al. / Neuroscience Letters 399 (2006) 141–146
of calcaneus from force curves of 3rd metatarsal [9,10,17].
Rationale to estimate posterior-anterior body weight shift from
obtaining y-vector by subtracting force curve of calcaneus from
force curve of 3rd metatarsal during straight floor walking was
investigated in detail, as a basis for comparing straight walking
and circular walking. Chronological distribution of sole-floor
reaction force from the five points of one representative step is
shown in Fig. 1. At the initial stage of the stance phase, sole-floor
reaction force from calcaneus dominates (100 and 200 ms), then
force from 5th metatarsal increases (200–400 ms), followed by
strongest force from 3rd metatarsal (500 and 600 ms) and finally
stance phase ends (700 ms). The period from 0 to 300 ms roughly
corresponds to rearfoot phase of y-vector and period from 400
to 700 ms corresponds to forefoot phase.
In the following figures (Figs. 2–4A) mean peak values
(n = 15 steps) of sole-floor reaction force from the five points dur-
143
ing stance period are illustrated. As examples from two subjects
are illustrated in Fig. 2 (straight) and Fig. 3 (straight), sole-floor
reaction force from 3rd metatarsal is statistically significantly
(p < 0.01) stronger than force of either 1st metatarsal or 5th
metatarsal. This relation was confirmed from data obtained from
all the 10 healthy subjects. This relation was also observed in
unaffected side of a patient illustrated in Fig. 4A. Based on these
data it is a fair conclusion that posterior-anterior body weight
shift directed from heel to 3rd metatarsal in the subjects studied in this project. Hence, it is reasonable to obtain y-vector by
calculating from force curves of calcaneus and 3rd metatarsal.
The lower part of y-vector was due to the impact of the heel,
hence designated as the rearfoot phase, and the upper part coincides with the period when the body shifts forward and push off,
therefore it was designated as the forefoot phase [9]. From the
investigation of temporal patterns of force curves of calcaneus
Fig. 2. (A) Peak force values (kg) recorded from the five points on the sole are illustrated as relative heights of columns with mean values ± S.D. (n = 15 steps) at the
corresponding points. (B) Averaged y-vectors (n = 10 steps) of C → M1, C → M3 and C → M5 are illustrated with heavy line and x-vector is shown with dotted line.
Relative values of areas of forefoot phase of y-vector compared to the largest value are indicated as percentage for all the records. Left column shows data of outer
foot during circular walking, middle column shows data of straight walking and right column shows results from inner foot during circular walking. Same subject
as Fig. 1.
144
N. Kobayashi et al. / Neuroscience Letters 399 (2006) 141–146
Fig. 3. Peak force values from the five points (A) and y-vectors (B) as illustrated in Fig. 2. Subject; Y.K. female, 22 years, height 156 cm, weight 48 kg. Walking
velocity 4.0 km/h.
and 3rd metatarsal, it is said amplitude of lower going rearfoot
phase is about 10% smaller than the peak value of the calcaneus,
while the upper going forefoot phase the peak value correspond
to peak value of 3rd metatarsal [16]. As can be seen in the middle columns of Figs. 2 and 3B, amplitudes of forefoot phases of
C → M3 is larger than either C → M5 or C → M1 for straight
walking. These observations were supplemented by comparing
areas of the forefoot phase of y-vector at C → M1, C → M3 and
C → M5. When area of C → M3, which is the biggest among
C → M1, C → M3 and C → M5, was taken as 100, C → M5 was
65.3% and C → M1 was 56.7% in Fig. 2B straight. For the subject in Fig. 3B straight, C → M5 was 37.1% and C → M1 was
18.2%, as compared with C → M3 which showed the largest
area. In Figs. 2–4B, relative values of areas of forefoot phase
of y-vector as compared to the strongest point are indicated as
%. These findings indicate that posterior-anterior body weight
shift is reasonably represented in line from calcaneus to 3rd
metatarsal during straight walking.
Successful recording was made on 9 among the 10 subjects;
the reason being mentioned above.
Compared with straight walking the process of changing path
consists of decelerating the forward motion, rotating the body
and stepping out toward the new direction.
Two types of walking strategy were recognized. As is illustrated in Fig. 2A, one type of subject shows sole-floor reaction
force from 5th metatarsal increased (p < 0.01) in inner foot as
compared to straight walking while forces from 3rd and 1st
metatarsals decreased. These changes are reflected in the patterns of three y-vectors shown in Fig. 2B. In the inner foot, the
amplitude of forefoot phase of C → M5 y-vector is much larger
than straight and outer foot as compared in terms of amplitudes
as well as areas of the forefoot phase. Also amplitudes of the
rearfoot phase at C → M3 and C → M1 increased and duration
elongated, indicating forward body weight sway in C → M3 and
C → M1 directions delayed. These data show that the subject
adjusts the body weight shift by leaning to inner side during
stance phase. In the outer foot, rearfoot phase is not so different
in the three y-vectors, while the forefoot phase of C → M3 is
the largest. This means that in the trailing (or outer) foot leaning to the inner side is compensated to some extent. This type
N. Kobayashi et al. / Neuroscience Letters 399 (2006) 141–146
Fig. 4. Peak force values (mean ± S.D.) from the five points of left (unaffected
side) and right foot (affected side) which showed typical equinovarus deformity
(A), and y-vector of C → M1, C → M3 and C → M5 of both sides (B). Patient;
A.M., female, 63 years old, height 151 cm, body weight 55.7 kg. Walking velocity about 2.5 km/h.
of walking was observed in five subjects (three male and two
female).
The other type of walking strategy shown by a subject is
illustrated in Fig. 3. This subject shows almost no recognizable changes in either inner or outer foot as compared with
straight walking. Change of direction was apparently planned
and executed during the immediately previous swing phase. This
strategy is easier and more stable for the subject because the base
of support while changing direction is much more symmetric in
the standing foot. This type of walking strategy was observed
on four subjects (two males and two females).
Central locomotor pattern generator in mammals such as carnivores and rodents is located mainly in the lumbosacral spinal
cord [12]. Although normal walking is automatic, it is not necessarily stereotyped. Central pattern generators are quite flexible.
Three important types of sensory information are used to regulate stepping; somatosensory input from the receptors of muscle
145
and skin, input from the vestibular apparatus (for controlling
balance), and visual input. Motor cortex is involved in the control of precise stepping movements in visually guided walking.
The cerebellum fine-tunes the locomotor pattern by regulating
the timing and intensity of descending signals [15].
Recent researches on the locomotor function of human subjects with a complete spinal cord injury at either lower cervical
or upper thoracic cord [1,4–7], and of human infants [13,14]
strongly suggest that essentially the same mechanisms exist
in human spinal cord as do in carnivores or rodents. Lamb
and Yang [11] argued that a common locomotor pattern generator controls walking in all different directions in human
infants.
The present experimental results suggest that humans adjust
their locomotor pattern generator either for walking on straight
as well as curved paths by presumably inputs from supraspinal
structures, although the mechanisms are yet to be clarified.
Fig. 4 illustrates data obtained from a patient. This female
patient of 63 years old suffered subcortical hemorrhage at left
hemisphere 15 months earlier. She showed Brunnstrom stage
III spastic hemiplegia for both upper and lower limbs of right
side, and showed equinovarus deformity during standing as
well as walking. The patient walked straight at about 2.5 km/h
without using a cane, however she could not walk curvilinear
path. To change walking direction she once stopped walking
and resumed walking after changing direction of her body. As
can be seen from the figure, unaffected left foot showed the
pattern which could not be differentiated from the healthy subjects illustrated in Figs. 2 (straight) and 3 (straight). On the
contrary, affected right side reveal characteristic pattern of distribution of sole-floor reaction forces in the sole; extremely weak
force from the calcaneus and unusually strong force from 5th
metatarsal (Fig. 4A). These characteristic distribution of force
is reflected on the pattern of y-vector obtained from C → M3
and C → M5 metatarsals, as are illustrated in Fig. 4B. The most
characteristic feature of this patient is that rearfoot phase of yvector is lacking. Among the other four hemiplegic patients, two
patients could not walk curvilinear path as the patient described
in Fig. 4, and the remaining patients could walk curvilinear path
but with difficulty. Distribution of the peak values from the five
points differed among the hemiplegic patients, hence patterns
of y-vector are different. This point is in preparation for other
report.
Total peak values of sole-floor reaction forces of the five
points are 52.6 kg for the affected side and 62.0 kg for the
unaffected side, and the difference was statistically significant
(p < 0.01) in this patient. This tendency was observed in all
the present subjects. Stance period of the unaffected side was
1490.6 ± 93.1 ms (n = 15 steps) and that of the affected side was
1231.3 ± 84.5 ms, and the difference was statistically significant
(p < 0.01). This difference was observed in all the hemiplegic
patients (n = 10, including this patient) and the difference was
statistically significant (paired t-test, p < 0.01). There was no difference in cadence of the unaffected and affected sides (p = 0.88).
From these data it can be said that the hemiplegic patients
use the affected side with short stance period and weaker solefloor reaction force, alongside with compensatory use of the
146
N. Kobayashi et al. / Neuroscience Letters 399 (2006) 141–146
unaffected side which showed longer stance period and stronger
reaction force.
Acknowledgements
The authors would like to express their gratitude to Professor R. Makino of Ohio State University for improving English.
Skillful technical assistance with data collection and illustration
by Ms. Ayano Sasaki is greatly acknowledged.
References
[1] J.A. Beres-Jones, S.J. Harkema, The human spinal cord interprets
velocity-dependent afferent input during stepping, Brain 127 (2004)
2232–2246.
[2] G. Courtine, M. Schieppati, Human walking along a curved path. I.
Body trajectory, segment orientation and the effect of vision, Eur. J.
Neurosci. 18 (2003) 177–190.
[3] G. Courtine, M. Schieppati, Human walking along a curved path. II.
Gait features and EMG patterns, Eur. J. Neurosci. 18 (2003) 191–205.
[4] V. Dietz, G. Colombo, L. Jensen, L. Baumgartner, Locomotor capacity
of spinal cord in paraplegic patients, Ann. Neurol. 37 (1995) 574–582.
[5] V. Dietz, R. Mueller, Degradation of neuronal function following a
spinal cord injury: mechanisms and countermeasures, Brain 127 (2004)
2221–2231.
[6] V. Dietz, R. Mueller, G. Colombo, Locomotor activity in spinal man:
significance of afferent input from joint and load receptors, Brain 125
(2002) 2626–2634.
[7] S.J. Harkema, S.L. Hurley, U.K. Patel, P.S. Requejo, B.H. Dobkin, R.
Edgerton, Human lumbosacral spinal cord interprets loading during stepping, J. Neurophysiol. 77 (1997) 797–811.
[8] K. Hase, R.B. Stein, Turning strategies during human walking, J. Neurophsyiol. 81 (1999) 2914–2922.
[9] K. Kiriyama, T. Warabi, M. Kato, T. Yoshida, N. Kobayashi, Progression of human body sway during successive walking studied by
recording sole-floor reaction forces, Neurosci. Lett. 359 (2004) 130–
132.
[10] K. Kiriyama, T. Warabi, M. Kato, T. Yoshida, N. Kobayashi, Mediallateral balance during stance phase of straight and circular walking of
human subjects, Neurosci. Lett. 388 (2005) 91–95.
[11] T. Lamb, J.F. Yang, Could different directions of infant stepping be controlled by the same locomotor central pattern generator? J. Neurophysiol.
83 (2000) 2814–2824.
[12] G.N. Orlovsky, T.G. Deliagina, S. Grillner, Neuronal control of locomotion. From mollusk to man, Oxford University Press, Oxford, 1999.
[13] M.Y.C. Pang, J.F. Yang, The initiation of the swing phase in human
infant stepping: importance of hip position and leg loading, J. Physiol.
528 (2000) 389–404.
[14] M.Y.C. Pang, J.F. Yang, Interlimb co-ordination in human infant stepping, J. Physiol. 533 (2001) 617–625.
[15] K. Pearson, J. Gordon, Locomotion, in: E.R. Kandel, J.H. Schwartz,
T.M. Jessell (Eds.), Principles of neural sciences, fourth ed., McGrawHill, Co, New York, 2000, pp. 737–755.
[16] T. Warabi, M. Kato, K. Kiriyama, T. Yoshida, N. Kobayashi, Analysis of human locomotion by recording sole-floor reaction forces
from anatomically discrete points, Neurosci. Res. 50 (2004) 419–
426.
[17] T. Warabi, M. Kato, K. Kiriyama, T. Yoshida, N. Kobayashi, Treadmill walking and overground walking of human subjects compared
by recording sole-floor reaction force, Neurosci. Res. 53 (2005) 343–
348.
[18] World Medical Association, Declaration of Helsinki on Ethical principles for medical research involving human subjects. 2004 http://www.
wma.net/.