Journal of Behavioral and Brain Science, 2013, 3, 137-149
http://dx.doi.org/10.4236/jbbs.2013.31013 Published Online February 2013 (http://www.scirp.org/journal/jbbs)
Qualitative Comparison between Rats and Humans in
Quadrupedal and Bipedal Locomotion
Taisei Hosoido1, Futoshi Mori2*, Keita Kiyoto2, Takashi Takagi2, Yukari Sano2,
Megumi Goto2, Katsumi Nakajima3, Naomi Wada2
1Osaka Veterinary Referral Center, Osaka, Japan
2Laboratory of Veterinary System Science, Yamaguchi University, Yamaguchi, Japan
3Department of Physiology, Kinki University School of Medicine, Osaka-Sayama, Japan
Received December 10, 2012; revised January 10, 2013; accepted January 17, 2013
ABSTRACT
Bipedal (Bp) locomotion is one of the most characteristic motor behaviors in human beings. Innate quadrupedal (Qp)
four-legged animals also often walk bipedally. The walking posture, however, is significantly different between the two.
This suggests that although both hav e a potential to walk bipedally, however, the human has a body scheme suitable for
Bp locomotion, probably its sk eletal system. The skeletal system includes the lumbar lordosis, sacral kyphosis, a round
pelvis, a large femur neck angle, short feet, and so on. To verify this hypoth esis, we compared kinematic and EMG ac-
tivities between rats and humans during Qp and Bp locomotion on a treadmill belt. The rat is a representative Qp animal,
but it is able to acquire Bp walkin g cap ability with motor learning. Although th e mobile ranges of th e hind limb jo int are
different during each locomotor pattern between rats and humans, both showed replicable flexion and extension excur-
sion patterns fo r each joint depending on the locomotor phase. There are many phase-locked EMG bursts between rats
and humans during the same walking task and these are observed in the proximal rather than the distal muscles. This
suggests that both rats and humans utilize similar neuronal systems for the elaboration of Qp and Bp locomotion. It was
interesting that both subjects showed more muscle activities during non-natural locomotor patterns; Qp < Bp for rats
and Bp < Qp for humans. This indicates that rat Bp and human Qp walking need more effort and we may be able to find
its reason in their skeletal system.
Keywords: Human; Rat; Evolution; Posture; Locomotion; Bipedal; Quadrupedal
1. Introduction
Most terrestrial vertebrates walk with the limbs on the
ground and terrestrial locomotion can be divided into
quadrupedal (Qp) and bipedal (Bp) locomotion. The ma-
jority of mammals walk quadrupedally, while humans
walk bipedally. As Aristotle and Darwin showed, the up-
right standing posture and Bp walking are the most basic
and essential features of human beings. For this, we have
developed a unique skeletal system including the lumbar
lordosis, sacral kyphosis, a round pelvis, large femur
neck angle; short feet, and so on. The lumbar lordosis in
humans is a critical component in the ability to achieve
upright and bipedal locomotion [1]. It is shown that such
skeletal adaptation suitable for human like Bp walking
was evolved relatively recent [2]. As a result, humans are
a unique mammal that can adapt their Bp walking pat-
terns to the various walking circumstances. However,
many other innate Qp four-legged animals (Qp animals)
can also stand upright and walk bipedally without the
skeletal characteristics observed in humans. Bears and
even cats and dogs can stand on their hindlimbs and walk
bipedally if trained for a circus attraction. It is also well
known that some species of genetically Qp nonhuman
primates walk bipedally in the wild. In this sense, the Bp
locomotion itself is by no means limited to humans or
some nonhuman primates.
What are the triggers for bipedalism? During Qp walk-
ing, the forelimbs are needed. Therefore, Qp animals sup-
port their body weight, maintain body balance and pro-
duce propulsive force with their limbs. Ho wever, Qp ani-
mals often use their forelimbs not only for the execution
of locomotion, but also for other purposes. These include
gripping and collecting food, hanging from branches,
digging up soil to find food and nestin g. These functio nal
usages of the forelimb may have already existed when
mammals appeared on the earth (Jurassic period) and
reinforced the different functional roles between fore-
limbs and hindlimbs. The functional differences are often
*Corresponding a uthor.
C
opyright © 2013 SciRes. JBBS

T. HOSOIDO ET AL. 138
observed in the morphology. In this sense, one can easily
observe skeletal differences between forelimbs and hind-
limbs in the most primitive mammals, Monotr emes , on
earth. These are also observed in other relatively old
mammals such as the Insectivora and Marsupials. For
further development of such specific forelimb function,
the forelimbs must be freed from locomotion by standing
up.
It has been considered that human beings evolved from
Qp animals. Because locomotor control mainly consists
of skeletomuscular and neuronal systems in both Qp
mammals and Bp human, the evolutional transition from
Qp to Bp walking may be primarily found in the changes
in these systems [3]. Human infants start walking quad-
rupedally and then acquire stable Bp capability as they
grow up. One possible reason why the walking pattern of
human infants smoothly transfers from Qp to Bp walking
is that the infants already have a skeletomuscular sys-
tem suitable for Bp walking at birth. It is interesting that
infants develop lordo sis around th e time that they acquire
bipedal locomotion. Furthermore, the facts described
above suggest that reorganization of the central nervous
system (CNS) for bipedalism is not essentially required
at birth in human beings. Moreover, nonhuman primates
including chimpanzees cannot fully extend their hip and
knee joints when they walk bipedally, and have a “bent-
hip and bent-knee” walking posture [4]. This is because
the skeletal structure around their ischia is not adequate
for full extension of the hip joint. These observations
strongly suggest th at innately Qp mammals d o not have a
skeletal system suitable for the elaboration of Bp walking,
but use a inherent neuronal system similarly for both Qp
and Bp locomotion. However, the defining factor in the
shift from Qp walking to Bp walking remains to be elu-
cidated. Here, we test the hypothesis that four legged
mammals and humans have their similar inherent neu-
ronal systems for the elaboration of both Qp and Bp
walking. To do this, we compared their kinematic char-
acteristics and recorded electromyographic activities from
different types of muscles during Qp and Bp walking.
Rats are originally Qp animals, but we had showed that
they are able to acquire Bp walking capability by train ing
[5].
2. Materials and Methods
2.1. Subjects
Twenty-five rats, of both sexes were used and 5 healthy
male, human subjects with no history of motility disor-
ders volunteered for this study. We established rat Bp
walking model (RBWM) to study the neuronal control
mechanisms of Bp walking and the effects of Bp walking
on the central nervous system [5,6]. One of the charac-
teristic features in our model is that the forelimbs, which
are often amputated when establishing a Bp animal mo-
del, were preserved. The detailed bipedal training method
was described elsewhere [5]. Briefly, 3 weeks after birth,
the rats were housed in a large rectangular glass contai-
ner that required them to use an upright posture to reach
water for 2 to 3 week s. Water was provided throu gh drip
bottles that were placed on the top of the container. Then,
they began training using a bipedal-walking training de-
vice. This device moves the drip bottle at a constant
speed, and the rat follo ws the bottle to obtain water while
maintaining an u pright posture. Each rat underwent train-
ing for 30 mins/day and 7 days/week for at least 3 months .
Water was given to the rats from the drip bottle installed
on the bipedal-walking training device or treadmill dur-
ing the training periods. All procedures confirmed to the
guidelines of the Animal and Human Investigation Com-
mittee and the Ethics Committee at Yamaguchi Univer-
sity. Informed, written consent was obtained from each
human subject. The experiments were performed accord-
ing to the Declaration of Helsinki.
2.2. Behavioral Task and Data Analysis
2.2.1. Qp and Bp Wal k ing Tasks
The RBWM and humans participated in the following
two walking tasks.
Qp walking: Both RBWM and human subjects walk ed
on the treadmill belts quadrupedally at a comfortable
speed for at least one minute. Each collection consisted
of 5 - 10 stable consecutive strides as a minimum number.
For humans, all subjects were instructed to walk on their
hands and feet and were given time to adapt to this Qp
walking pattern on the treadmill belt before recording.
Bp walking: Both subjects walked on a treadmill belt
bipedally at a comfortable speed for at least a half minute.
In RBWM, each collection consisted of at least 3 - 5 sta-
ble consecutive strides as a minimum number. For hu-
man subjects, all subjects were given time to adapt their
Bp walking on the treadmill belt before recording.
2.2.2. D ata Recordi ng
The right side lateral views of the Qp and Bp walking in
both RBWM and humans were recorded using a high
speed video-camera with sampling rates of 200 f/s (HAS-
220) and 2000 - 1000 f/s, respectively. These frames
were captured and stored on a computer and used for fur-
ther analyses (Movias Pro, NAC, Japan). Prior to all re-
cording sessions, circle markers were placed on each rat
and human’s skin at the vertebrae at L3 (a), hip (b), knee
(c), ankle (d), and metatarsal joints (e), and the hip, knee,
and ankle joints were defined as follows, hip joint: angle
a-b-c, knee joint: angle b-c-d, ankle joint: angle c-d-e
(Figure 1).

T. HOSOIDO ET AL. 139
a a
a
a
b
b
d
b
b
e
e
c
d
e
d
e
d
c
c c
A-1 B-1
B-2
A-2
2.2.3. Data Analysi s
Stick figures, which connected the circle marks on the
skin and eye, ear and toe, were made from each frame.
From these images, the onsets of swing (SW) and stance
(ST) phases were determined by visual inspection. The
hip, knee, and ankle joints were measured for the angle
excursion graphs and angle-angle cyclographs. Raw ki-
nematic data were averaged for ~10 step cycles. EMG
data were synchronized with a high speed camera using a
200 Hz signal. Raw EMG bursts were rectified and aver-
aged for 10 - 15 step cycles and averaged EMG wave-
forms were made for each muscle to compare the EMG
activity patterns to the locomotor cycle. EMGs recorded
from humans were also rectified and averaged. The an-
gular excursion graphs and the duration of EMG bursts
were demonstrated as % of cycle; onset of the SW phase
(=0%) and termination of the ST phase of the right hind-
limb (=100%). To detect commonly observed EMG ac-
tivity patterns between Qp and Bp in a single subject, and
between RBWM and humans for the same walking pat-
tern, the qualitative EMG comparison was made. For this,
we focused mainly on the temporal component such as
the onset and/or the duration of EMG activity during a
step.
Figure 1. The position of circle markers and joint angles of
RBWM (A) and humans (B) during Qp (1) and Bp (2)
walking. The arrows indicate the position of the virtual
COG and its direction.
Electromyographic (EMG) activity was recorded with
video recordings. For RBWM, 6 paired bipolar wire
electrodes (AM Systems Inc, USA) with 0.5 - 1.0 mm
recorded area were implanted into muscles on both sides
from the Longissimus lumborum (L) at L1-3, the gluteus
medius (Gm), the rectus femoris of the quadriceps (Q),
the semitendinosus of the Hamstring (H), the lateral head
of the gastrocnemius (Gs), and the tibialis anterior (TA)
under anesthesia with Ketamin-HCl (initial injection: 40
- 60 mg/kg, additional injections: 15 - 25 mg/kg/h) or
Isoflurane mixed with oxygen gas. The position of the
EMG electrodes was confirmed by electrical stimulation
of the appropriate electrodes. The wire electrodes were
connected with bio-amplifiers via a cable and amplified
(30 - 3000 KHz, ×2000; Nihon Koden, Japan). In hu-
mans, the EMGs were recorded from the same muscles
as those recorded in the rats. EMG bursts were recorded
using bipolar surface electrodes placed on the skin (Ni-
hon Koden, Japan). In both RBWM and human subjects,
EMGs during Qp and Bp walking were recorded alter-
nately in a same recording session.
3. Results
Both RBWM and humans could walk seamlessly on the
treadmill belt. Figures 2 and 3 show serial drawings of a
single step cycle of Qp and Bp walking in RBWM and
humans, respectively. The tail of RBWM was not drawn
in Figure 2.
3.1. Angular Changes at the Hip, Knee and
Ankle Joints in RBWM and Human
Figure 4 shows representative serial angular excursions
during three step cycles in the hip, knee and ankle joints
of RBWM (A) and humans (B) during Qp (1) and Bp (2)
walking on a level treadmill belt. All three joints in both
RBWM and humans showed replicable across step cycles,
but the angular excursion graphs of the same joint in the
two grou ps were quite d i ff erent.
1 2 3 4 5 6 7 8
A
1 2 3 4 5 6 7 8
B
Figure 2. Seri al drawings of Qp (A) and Bp (B) walking in RBWM during a single step cycle. The numbers on each drawin g
indicate the order of the steps.

T. HOSOIDO ET AL. 140
1 2 3 4 5 6 7 8
A
1 2 3 4 5 6 7 8
B
Figure 3. Serial drawings of Qp (A) and Bp (B) walking in human subjects during a single step cycle. The numbers on each
drawing indicate the order of the steps.
A-1
A-2
B-1
(deg)
(deg)
(deg)
B-2
hip
knee
ankle
180
150
120
90
60
rH
180
150
120
90
60
rH
Figure 4. Repr esentative three step cycl e angular changes in hip, knee and ankle joints in RBWM (A) and humans (B) during
Qp (1) and Bp (2) walking. Gait diagrams of the right hindlimb are shown at the bottom of each graph; the swing phase is
shown by white bars and stance phase by gr ay bar s.
3.1.1. Qp Wa lking in RBWM (Figure 4A-1)
Hip joint showed smoo th flexion and ex tension tran sition
during a single step cycle. The maximum flexion and ex-
tension were observed during the early and late ST phase,
respectively. The mean mobile range of this joint was
about 75˚ - 90˚. From the initiation of the SW phase,
knee joint continued to move toward extension until the
transition period to ST phase and showed maximum ex-
tension. It then moved toward flex ion until the end of the
ST phase. The mean mobile range of this joint was ab out
50˚ - 100˚. Ankle joint had a wider mobile range than
other two joints throughout the single step cycle. From
the late SW phase, this joint moved toward extension
until the early ST phase. Th en, it flex ed fo r a short period
and again moved toward extension until the late part of
the ST phase, making two humps at the beginning (small)
and the late part (large) of the ST phase, after which it
started to flex toward the SW phase. The mean mobile
range of this joint was about 50˚ - 105˚.
3.1.2. Qp Walking in Human (Figure 4B-1)
From the end of SW phase, hip joint moved toward ex-
tension until the end of the ST phase, making a small
hump at the beginning of this phase. It then rapidly
flexed until the end of SW. The mean mobile range of
this joint was 80˚ - 150˚. Knee joint showed a basically
similar angular excursion pattern to that of the hip joint.
The maximum flexion and extension of this joint were
observed at the mid to late SW phase and at the late ST
phase, respectively, and these appeared earlier than those
of the hip. At the end of the ST phase, it started quick
flexion until the mid SW. The mean mobile ran ge of this
joint was about 80˚ - 145˚. Ankle joint moved through a
single step cycle similar to that during Bp walkin g. There
were two humps, i.e., a small one at the early and a large
one at the late phase of ST, and these were similarly ob-
served in RBWM Qp walking. Between these humps
during the ST phase, this joint first moved toward flexion
until the mid ST phase an d then extend ed until the end of
the ST phase. The mean mobile range of this joint was
about 85˚ - 110˚.
3.1.3. Bp Walking in RBWM (Figure 4A-2)
Hip joint showed a small range of movement during a

T. HOSOIDO ET AL. 141
single step cycle. At the end of the ST phase, this joint
showed maximum extension and then continued to flex
until the beginning of the next ST phase. The mean mo-
bile range of this joint was about 85 ˚ - 105˚, and this was
as small as that during Bp walking. In contrast to the
small movement range of the hip joint, knee joint showed
saw-blades with clear extension and flexion during a step
cycle. Throughout all phases of SW, this joint moved
towards extension. The maximum extension of this joint
was observed at the early phase of ST. Then, it showed
gradual flexion until the end of the ST phase. At this
timing, this joint showed maximum flexion at around 40˚.
The range of motion of this joint was from 40˚ to 80˚.
Ankle joint showed similar excursion curve to that during
Qp walking, with two humps at the early and late phases
of ST. The mobile range of this joint during Bp walking
was smaller than that during Qp walking (65˚ - 95˚).
3.1.4. Bp Walking in Human (Figure 4B-2)
From the late ST phase, hip joint moved towards flexion
until the mid SW phase, followed by an extension until
the end of the ST phase with a small hump at the begin-
ning of the ST phase. This joint showed maximum ex-
tension and flexion at the end of the ST and at the end of
SW phase, respectively. The mobile range of this joint
was 150˚ - 180˚. At the late phase of ST, knee joint
started to flex and showed maximum flexion at the early
phase of SW. It then rapidly moved toward extension
until the end of th e SW phase. The joint then remained at
around 160 o until the end of the ST phase. This constant
joint angle period was not observed in RBWM, which
showed a gradual decrease during this phase. The mobile
range of this joint was 125˚ - 165˚. Ankle joint showed
similar angular excursion to that observed during Qp
walking. From the end of the ST phase, this joint exhib-
ited rapid extension until the mid SW phase. It then
flexed until the end of SW and again started to extend
during the early ST phase. Finally, towards the end of ST,
it showed a gradua l extension.
3.2. Angle-Angle Cyclographs of Knee-Hip and
Ankle-Knee Relationships
The cyclographs show the average angle-angle relation-
ships between two joints such as the knee and hip, and
ankle and knee joints in RBWM (Figure 5) and humans
(Figure 6) during Qp (A) and Bp (B) walking. In general,
these figures showed that RBWM walked with their
A-1 A-2
B-1 B-2
hip
(deg.)
180
150
120
90
60
30
180
150
120
90
60
30
knee
(deg.)
180
150
120
90
60
30
180
150
120
90
60
30
hip
(deg.) knee (deg.)
30 60 90 120 150 18030 60 90 120 150 180
ankle (deg.)
knee (deg.)
30 60 90 120 150 18030 60 90 120 150 180
ankle (deg.)
knee (deg.)
Figure 5. Representative averaged angle-angle cyclographs of hip-knee and knee-ankle joints in RBWM during Qp (A) and
Bp (B) walking. For each group (A-1, A-2, B-1, B-2), the left and right side cyclograph shows the relationship between hip
and knee, and knee and ankle, respectively. The upward and downward open triangles indicate the onset of the SW (gray line)
and ST (black line) phase, respectively. The curved gray-colored arrows in the graphs show the direction of angle-angle joint
movement.

T. HOSOIDO ET AL. 142
A-1 A-2
B-1 B-2
hip
(deg.)
knee
(deg.)
180
150
120
90
60
30
180
150
120
90
60
30
hip
(deg.)
knee
(deg.)
30 60 90 120 150 18030 60 90 120 150 180
ank l e ( deg.)knee (deg.)
30 60 90 120 150 18030 60 90 120 150 180
ank l e ( deg.) knee (deg.)
180
150
120
90
60
30
Figure 6. Representative averaged angle-angle cyclographs of hip-knee and knee-ankle joints in humans during Qp (A) and
Bp (B) walking. For each group (A-1, A-2, B-1, B-2), the left and right side cyclograph shows the relationship between hip
and knee, and knee and ankle, respectively. The upward and downward open triangles indicate the onset of the SW (gray line)
and ST (black line) phase, respectively. The curved gray-colored arrows in the graphs show the direction of angle-angle joint
movement.
hindlimb joints rather flexed, while humans walked with
their hindlimbs extended.
3.2.1. Q p Locom ot i on: Knee-Hip Joint Relation s h i p
(Figures 5A-1 vs 6A -1)
In RBWM, it was clearly observed that Qp walking was
achieved with less hip joint movement. After the initia-
tion of the SW phase, the hip joint started to flex, while
the knee joint simultaneously started to extend, until the
early ST phase. Then, the hip joint started to extend and
the knee joint started to flex until the end of ST. The re-
lationship between th ese two join ts was o bserved lin early.
In contrast to the hip joint of RBWM, that of humans
showed a wider range of movement. After initiation of
the SW phase, both hip and knee joints started to flex si-
multaneously. At the late SW phase, the maximum flex-
ion of the hip and knee joint was observed. During this
phase, the cyclograph curve was characterized by an
“arc” shape. After that, both started to extend from the
end of SW toward the late phase of ST. The maximum
extensions of the hip and knee joints were observed dur-
ing the ST phase. In the late ST phase, the knee showed
rapid flexion, while the hip reversed from extension to
flexion. The cyclograph moved in an anti-clockwise di-
rection in both RBWM and humans.
3.2.2. Qp Locomotion: Ank l e-Knee J oi nt Rel a ti ons hi p
(Figures 5A-2 vs 6A -2)
For RBWM, the cyclograph curve was characterized by a
“crescent moon” shape during a single step cycle. At the
late ST phase, the knee joint started to extend and the
ankle started to flex, respectively. The maximu m flexion
of the ankle joint was observed at the late phase of SW.
Then, both knee and ankle joints simultaneously started
to extend until the early ST phase. After that, the knee
and ankle joints started to flex until the mid ST. The knee
joint continued to flex until the late ST phase, while the
ankle joint rapidly extend ed. At the late phase of ST, the
knee and ankle started to flex and extend, respectively.
For humans, the cyclograph was characterized by an
“oval” shape. This is why both joints move in a similar
way during a step cycle. After initiation of the swing
phase, both started to flex and the curve moved in a
clockwise direction. In this relationship, the cyclographs
of both RBWM and human moved in a same clockwise
direction.

T. HOSOIDO ET AL. 143
3.2.3. Bp Locomotion: Knee-Hip Joint Relationship
(Figures 5 B -1 vs 6B-1 )
In RBWM, the cyclograph showed a “flat oval” shape
with a longer horizontal (knee) axis. This shape was
made by the rapid extension-flexion movement of the
knee joint with less that of the hip joint. After initiation
of the SW phase, the knee joint showed exten sion for th is
period, while the hip joint was weakly extended. Then,
the knee joint started to flex until the late part of the ST
phase, with weak hip extension. In humans, the cyclo-
graph curve was characterized by a “triangle” shape dur-
ing a one step cycle. This triangle shape cyclograph in-
dicated that both the hip and knee joints showed similar
flexion and extension cycles throughout a single step
cycle. At the early SW phase, both hip and knee joints
simultaneously flexed. During mid to late SW phase, the
hip joint remained constant and the knee joint was ex-
tended. From the beginning of the ST phase, the hip joint
showed gradual extension, while the extended knee joint
was preserved. Then, at the late ST phase, both the hip
and knee joints showed gradual flexion until the initia-
tion of the SW phase. In both RBWM and human sub-
jects, the cyclograph directions were anti-clockwise.
3.2.4. Bp Locomotion: Ankle-Knee Joint Relationship
(Figures 5 B -2 vs 6B-2 )
The cyclograph curve in RBWM was expressed by a
“figure of eight”. During the SW phase, the knee joint
rapidly extended, while the ankle joint showed gradual
extension. After the onset of the ST phase, both the knee
and ankle joint started to flex up to two-thirds of the ST
phase. Then, the ankle began to extend, while the knee
joint maintained flexion in the late ST phase. For humans,
the cyclograph between the knee and ankle showed a
“figure of eight” shape which was similar to that observed
during RBWM Bp walking. At the early phase of SW,
both the knee and ankle were flexed. Then, the knee joint
was rapidly extended, while the ankle joint remained co n-
stant. Subsequently, both the knee and ankle started to
flex at the beginning of ST. During the late ST phase, the
knee showed flexion while the ankle showed extension.
3.3. EMG Burst Patterns between RBWM and
Humans in Qp and Bp Walking
Figure 7 illustrates ensemble-averaged EMG waveforms
through a single step cycle between different subjects in
the same walking pattern. As in Figure 7, EMG compo-
nents showing similar onset and/or duration between
RBWM and humans during Qp and Bp walking are indi-
cated by the black and white arrows in each averaged
EMG.
3.3.1. Longissimus Lumborum
Although no common EMG bursts were observed in ei-
ther RBWM or humans during Qp walking, a burst at the
early ST phase was observed in both RBWM and hu-
mans during Bp walking.
3.3.2. Gluteus M ed ius
During Qp walking, this muscle showed small and large
bursts at the end of SW and at the late phase of ST, re-
spectively, in both RBWM and humans. During Bp
Qp-RBWM
Gm
0 50 100%
L
50 100%0 50 100% 50 100%
Bp-RBWM Bp-Human Qp-Hu man
L
Gm
Q
Q
HH
GsGs
TATA
0.05V
Figure 7. Ensemble-averaged EMG waveforms through a single step cycle in RBWM (n = 5) and humans (n = 3) during Qp
and Bp walking. The vertical black bar on the right of each graph shows an amplitude of 0.05 V. L: Longissimus lumborum
at L1-3, Gm: gluteus medius, Q: rectus femoris of the quadriceps, H: semitendinosus of the Hamstring, Gs: lateral head of
the gastrocnemius, TA: tibialis anterior.

T. HOSOIDO ET AL. 144
walking, there was a burst EMG activity at the mid-ST in
both RBWM and humans.
3.3.3. Quadric e p s
During Qp walking, two bursts were observed at the tran-
sition period from SW to ST and at the late phase of ST,
however, the amplitudes were different.
3.3.4. Hamstrings
During Qp walking, the burst EMG activity was detected
at the late ST phase in both RBWM and human subjects,
while a burst component was found at the early phase of
ST during Bp walking.
3.3.5. G astroc n em ius
In both RBWM and humans, there were burst activities
at the late phase of ST. During Bp walking, an in-
creased EMG activity was observed at the mid ST in
both RBWM and humans.
3.3.6. Tibialis Anterior
This muscle showed a variety of activity patterns be-
tween RBWM and human subjects in either Qp or Bp
walking.
3.4. EMG Activity between Qp and Bp Walking
in RBWM and Humans
Figure 8 was prepared to detect common features of
EMG activity during Qp and Bp walking in RBWM and
humans, respectively. Overall, these results show that the
muscle activities during the subject’s natural walking
pattern were less active than those during the non-essen-
tial walking pattern, i.e. Qp < Bp for rat and Bp < Qp for
human.
3.4.1. Longissimus Lumborum
In Qp-RBWM, there seem to be two burst activities dur-
ing a single step cycle. They were observed at the transi-
tion period from the SW to ST phase and at the late part
of ST. In Bp-RBWM, the EMG showed tonic activity
during a step cycle with three burst activities. A burst
EMG was obtained at the transition period from the SW
to ST phases in both Qp and Bp walking. In human Qp,
this muscle showed basically tonic activity with three
bursts. One of these bursts was observed at the SW phase,
and the other two were seen during the ST phase, respec-
tively. The burst observed at the early ST phase during
Bp walking was also obtained during Qp walking.
3.4.2. Gluteus M ed ius
In Qp-RBWM, this muscle showed small and large
bursts at the en d of SW and at the transition period from
ST to SW phase, respectively. Except for these active
periods, this muscle was rather silent. In Bp walking, this
muscle showed three EMG bursts during a single step
cycle. These were observed at the transition period from
SW to ST, the mid-ST and the late ST phases. The burst
at the late ST phase during Qp and Bp were similarly
observed in RBWM. In humans, there were two burst
activities and one burst activity during Qp and Bp walk-
ing, respectively. The onset timing and the duration were
different.
3.4.3. Quadric e p s
In RBWM, the EMG burst activities were different be-
tween Qp and Bp walking. During Qp walking, two burst
activities were observed at the end of the SW (larger) and
the mid-ST (smaller) phases. This muscle was activated
at the end of ST during Bp walking. In human Qp, there
were two bursts; one during SW and the other during the
ST phase. A small burst was observed at the early phase
of ST during Bp walking. This muscle showed different
burst patterns and duration in humans.
3.4.4. Hamstrings
In RBWM during Qp walking, two EMG bursts were
clearly observed at the late SW and at the end of ST
phases, respectively. There were EMG bursts activated at
the same timing during Bp walking. In humans, there
were two EMG burst activities during Qp walking, while
there was no clear burst during Bp walking.
3.4.5. G astroc n em ius
In Qp-RBWM, there were two bursts at the transition
from SW to ST phases and during ST phase. On the other
hand, there were three burst activities du ring Bp walking .
These were observed at the transition period from the
SW to ST, the mid-ST and the end of the ST phases. The
first and the last ones were activated at the same timing
in Qp-RBWM. In humans, two bursts were observed
during Qp walking, while a single burst was seen during
Bp walking. This burst was activated at the same timing
during a step cycle (early phase of ST) during Qp walk-
ing.
3.4.6. Tibialis Anterior
In RBWM, this muscle showed a simple burst at the end
of ST during both Qp and Bp walking. In humans, there
were two burst activities in both Qp and Bp walking. The
burst which appeared at the transition from the SW to ST
phase was observed in both Q p an d B p wa lking.
3.5. Comparison of Averaged EMG Activity
during Qp and Bp Waling in RBWM and
Human
To compare each muscle activity during a single step

T. HOSOIDO ET AL. 145
cycle between Qp and Bp walking quantitatively, the
area of averaged EMG activity of each muscle and the
ratio between non-natural/natural walking, i. e. Bp/Qp in
RBWM and Qp/Bp in humans was calculated ( Table 1).
The area of averaged EMG activity was defined as the
shaded area shown in Figure 8 top-right. In both subjects,
5 of 6 muscles examined showed larger areas dur ing non-
natural walking than during natural walking, i.e. small
EMG during Qp in RBWM and small EMG during Bp in
humans. The highest ratio was observed in the muscles
of H in RBWM and of Q in humans. The average values
of area comparisons were 203.8 and 240.1 in RBWM and
humans, respectively. This suggests that the EMG activ-
ity during non-natural walking was about two times
higher than during each subject’s natural walking.
4. Discussion
This study examined the kinematic characteristics and
spatiotemporal patterns of muscle activ ity during Qp and
Bp walking in rats which had acquired the ability to walk
bipedally and humans. We focused to show the locomo-
tor and EMG similarities and differences observed be-
tween rats and humans during Qp and Bp walking. Our
results show a number of similarities between rat and hu-
man walking, but they also revealed several features spe-
cific to humans.
4.1. General Differences during Qp and Bp
Walking
We consider that the change in the skeletal system is one
of the determining factors in the development of human
like upright standing and Bp walking in humans. For
both upright standing and the execution of Bp walking, it
is necessary to bring the center of gravity (COG) close to
the hip joint and acqu ire the ability to raise the upp er part
of the body (Figure 1B-2). To do this, humans have a
unique skeletal system that is not observed in other
four-legged animals. Such skeletal features including the
lumbar lordosis, sacral kyphosis, round pelvis, and so on
enable humans to walk bipedally against gravity [1]. For
example, lordosis stabilizes the upper body over the
lower limbs in humans by positioning the trunk’s center
of mass above the hip. In human females, substantially
proliferated lordosis is observed during pregnancy to
Table 1. Comparison of averaged EMG activity (area) during
Qp and Bp walking in RBWM (n = 5) and human subjects
(n = 3).
RBWM Human subjects
QpBpBp/Qp (%) Qp Bp Qp/Bp (%)
L 22.858.2255.3 101.7 66.9 152.0
Gm11.717.9153.0 151.3 93.9 161.1
Q 12.8 5.643.8 344.2 57.5 598.6
H 8.344.0530.1 155.3 67.2 231.1
Gs17.027.6 162.4 95.3 112.5 84.7
TA6.99.1131.9 457.1 145.3 314.6
Ave13.327.1203.8 217.5 90.6 240.1
Qp-RBWM
Gm
0 50 100%
L
50 100%0 50 100%50 100%
Bp-RBWM Bp-Human
Qp-Human
L
Gm
QQ
H
H
Gs
Gs
TA TA
0.05V
Figure 8. Rearranged ensemble-averaged EMG waveforms through a single step cycle in RBWM (n = 5) and humans (n = 3)
during Qp and Bp walking to compare the effect on the walking pattern in each subject. The arrows and the black bar have
the same meaning as those in Figure 7.

T. HOSOIDO ET AL. 146
compensate for the fetal load [7]. From these facts, it
should be emphasized that these skeletal characteristics
help such spatial body orientation for upright body pos-
ture. The typical Qp ungulates are not able to walk bi-
pedally. This is because their COGs are always main-
tained at around the front of the animal’s body during
walking and this position is cl early different from that of
humans. Therefore, the bipedalism must strongly relate
with the position of COG. In this sense, humans are a
unique mammal which can adapt their Bp walking to
various circumstances while maintaining the COG di-
rectly at the hip joint.
It is well known that primates can acquire the ability to
walk bipedally and even clear obstacles on a walking
path with motor learning [8-10]. This indicates that they
are able to bring their COG near to the hip joint due to
their acquired skeletal similarities to humans such as the
lordosis in various situations [10-12]. However, their
walking posture is not exactly the same as that of humans.
This is due to the characteristic flexed body posture and
called “bent-hip and bent-knee” walking posture [4]. As
a result, their upper bodies are rather forward-bent. In our
rat model, the rats could walk bipedally, but their body
axis orientation was forwardbent, suggesting that the
COG is not exactly on the hip joint (Figures 1A-2 and
2B). This walking posture surprisingly resembles to that
of chimpanzees and other nonhuman primates. Why do
they walk with such a posture? There are many kinds of
four-legged mammals in nature and they have adapted
their body structures su itable to their liv ing environments
during evolution. From this point of view, it is possible
that their bodies are not suitable for erect posture due to
skeletal constraints such as the different orientation of
the ischium between primates and humans [13]. Substan-
tially, the key to the skeletal structure for human-like
walking posture can be found at around the pelvic girdle
[14].
4.2. Motor Patterns during Qp Walking in Rats
and Humans
Qp walking is a natural locomotor pattern for the rat, but
it may be a challenging locomotor task for humans [15-
17]. In this locomotor task, our findings revealed both
similarities and differences between rats and humans.
Among the similarities is the rhythmical and replicable
hindlimb movement across consecutive step cycles whic h
reflect the existence of a common rhythm generating sys-
tem in both subjects [18]. The cyclographs showed that
both knee-hip and ankle-knee relationships during Qp
walking were in anti-clockwise and clockwise directions,
respectively, in rats (Figure 5) and humans (Figure 6).
These results suggest that the rhythmical hindlimb mo-
tion and its alignment were similarly controlled between
rats and humans for the smooth and seamless execution
of Qp walking.
We also found several differences in the mobile range
of the hindlimb joint and the amplitude of each muscle
during walking. The mobile ranges of the hip and knee
joints during Qp walking in RBWM (Qp-RBWM) were
smaller than that during Qp walking in humans (Qp-
Human), and these two joints were generally flexed
(Figure 4). On the other hand, the range of ankle in Qp-
RBWM was lar ger th an that in Qp-Hu man. Ank le p lanta r
flexion during Qp-RBWM and Qp-Human was observed
twice in a single step cycle. This angular excursion was
well coupled with the EMG burst observed in Gs, which
is the ankle plantar flexor muscle (Figure 4). The Q mu-
scle in both RBWM and humans showed EMG bursts
during the extension phase of the ipsilateral hindlimb
knee joint and these bursts were observ ed during the SW
phase of the same limb during Qp walking. Moreover,
the burst activities observed in Gm and H were common
between RBWM and humans, suggesting the proximal
muscles are also used in a similar manner in both rats and
humans (Figure 7). These results suggest that both sub-
jects recruited common motor patterns in the hindlimbs
during Qp walking, while having their own mobile range
of hindlimb joints.
4.3. Motor Patterns during Bp Walking in Rats
and Humans
RBWM showed stable BP walking, but the Bp walking
pattern was not exactly the same as that of humans (Fig-
ure 2B). In this locomotor task, a clear difference was
observed at the shortened period of the SW phase in
RBWM. The ratio between the period of the SW:ST
phases during Bp walking was about 1:9 in RBWM and
3:7 in humans; however, the ratio itself was preserved
with replicable angular excursions which reflect the ex-
istence of a rhythm generator in rats and humans (Figure
4, [18]).
Conversely to Qp walking, many muscles were highly
activated during Bp walking in RBWM (Figures 7 and
8). However, it should also be indicated that all of these
muscles were phase-dependently activated [16]. For ex-
ample, in Q muscle during Bp walking, the activity was
transiently observed from the end of ST to the SW
phases and this was for both extension of knee joints and
flexion of hip joints in RBWM. In contrast to the RBWM
Q EMG activity, that of humans was seen from the end
of SW to the mid-ST phase, and seems to be used for
knee extension and maintenance. In fact, Q is a bi-func-
tional muscle used to flex the hip and extend kne e joints,
and these results suggest that both rat and human subjects
could use this muscle for Bp walking. The activity o f the
H muscle in Bp-RBWM was maintained throughout the

T. HOSOIDO ET AL. 147
ST phase (early > late), while human H muscle showed a
different activity pattern during the same period. This H
muscle activity and the lower activity of Q muscle
caused knee flexion in RBWM during the ST phase. This
knee flexion lowered the COG of Bp walking rats and
may have forcibly terminated stable Bp walking on the
treadmill belt. Certainly, the knee joint gradually flexed
during the ST phase in RBWM (Figure 4A-2). To cor-
rect this walking posture, it is necessary to bring the
lowered COG higher to the proper level. The small EMG
burst activity of the H muscle observed at the end of the
ST phase in the ipsilateral limb must be recruited for this.
Continuous knee flexion during the ST phase also forced
initiation of the SW phase of the contralateral limb and
the short SW phase was indispensable for the next step
cycle. It was interesting that the activity pattern of this H
muscle was not exactly the same between RBWM and
human subjects during Bp walking, but the largest EMG
amplitude of this muscle was observed at the transition
from the SW to ST phase. The same EMG activity com-
ponent is present in other muscles such as the L, Gm and
Gs. Four of six muscles examined, showed surprisingly
similar activation patterns and/or components during Bp
walking in both rats and human sub j ects. Moreover, three
of these muscles are activated in both Qp and Bp walking.
These results may indicate that both subjects have a
similar control system for the different walking patterns.
4.4. Motor Patterns between Qp and Bp Walking
in Rats and Humans
Between Qp and Bp walking, we also found many simi-
larities and differences in the characteristic EMG bursts
in both RBWM and human subjects, respectively (Fig-
ure 8). The Gm muscle is the extensor of the hip joint.
The largest EMG burst of this muscle was mainly ob-
served at the end of the ST phase in both locomotor
patterns. This reflected that the ipsilateral hindlimb is
pushing the walking surface before toe off. The hip joint
excursion in both Qp and Bp walking was quite small
(Figure 4) and this seems to reflect the smaller EMG
burst activity of this muscle in RBWM. During Bp walk-
ing, the other two bursts were observed at the beginning
of ST and the mid-ST phases and were probably recrui-
ting for the extension of the hip joint and the main-
tenance of its joint angle during Bp walking. In humans,
the EMG activities of the Gm muscle in both Bp and Qp
walking resemble those of the Q muscle. This suggests
that these two muscles are co-functioning in humans. In
Bp walking, Gm muscle bursts observed during the ST
phase extended the angle of the hip joint from 165˚ to
180˚. It was noteworthy that the Gm EMG burst was
terminating when the ipsilateral knee started to flex to-
gether with hip flexion. On the other hand, during Qp
walking, Gm EMG bursts were observed twice at the end
of SW and the ST phase, respectively. This pattern was
similarly observed in the Q muscle, suggesting the Gm
muscle co-activates with the Q muscle. The latter activity
was observed during extension of the ip silateral hip joint.
The former was observed when the hip joint showed
flexion and the kn ee jo in t started to ex tend. It see med un-
likely to be related to hip joint extension, but this might
be a characteristic EMG burst observed when human
subjects are walking on a treadmill belt quadrupedally
(hand-foot crawling).
A clear difference was also observed in the Q muscle
in humans. The burst of this muscle mainly corresponded
to the flexion of the knee joint. It is considered that the Q
muscle EMG burst during the ST phase of the ipsilateral
hindlimb inhibits excessive knee flexion during Qp walk-
ing. EMG bursts of Q in Bp-human were small and
mainly observed during early ST phase. During Qp walk-
ing, the Q EMG burst was also observed at the ST phase
and had large amplitude. However, the function of this
EMG burst during Qp walking seems to be different. The
EMG burst during Bp maintained the angle of the knee
joint at around 165˚, while Q EMG bursts during the ST
phase in Qp walking were observed during the extension
phase of the ipsilateral knee joint. These facts indicate
that the human subjects could appropriately recruit each
muscle’s functional role for different walking patterns.
Between Qp and Bp walking, five of six and three of six
muscles showed similar EMG burst activities in rats and
humans, respectively. In rats, this result suggests that
their Qp and Bp walking is achieved by similar muscular
activations. In contrast to the results from rats, humans
tend to use each muscle selectively to fit different walk-
ing patterns. This may be due to the recruitment of vo-
luntary control in Qp walking that is not a natural pattern
for adult hu mans [16].
4.5. Expected Energy Consumption between Qp
and Bp in Rats and Humans
For rats and adult humans, the Bp and Qp walking pat-
terns were not natural, respectively. The general high ac-
tivation of each muscle during Bp-RBWM and during
Qp-Human in the present study implies that RBWM and
humans need a lot of energy during their Bp and Qp
walking, respectively. In fact, the EMG amplitude during
Qp-RBWM was smaller than that during Bp-RBWM,
and that during Bp-human was smaller than that during
Qp-human (Figure 8). It is not possible to calculate
energy consumption from EMG activity, however, it
might be possible to argue that the increased activation
should increase the cost. From this point, we had cal-
culated the area of averaged each muscle EMG activity
as shown in Figure 8 (top right dark shaded area). The

T. HOSOIDO ET AL. 148
results are summarized in Table 1 . Five of six muscles in
both RBWM and humans showed about a 1.5 to 6 times
larger area in Bp and Qp walking, respectively. The ave-
rage comparison showed about a 2 to 2.4 times larger
area in the non-natural walking patterns. Although the
areas of Q in RBWM and Gs in humans were smaller
during the non-natural walking patterns, the majority of
the muscles that underwent EMG recording showed si-
milar trends. This suggests that the non-natural walking
patterns was made by a lot of efforts and probably re-
quired more energy than for the natural walking pat-
terns. The smaller areas in Q and Gs, however, indicate
that these are probably not necessary or are less affected
in both Bp and Qp in RBWM and humans. This is be-
cause a clear EMG burst was observed in both muscles
and this burst induced clear joint movement during dif-
ferent walking patterns. Nakatsukasa et al. [19,20] showed
that Bp-trained nonhuman primates consume less meta-
bolic energy during their natural Qp walking than Bp.
This suggests that RBWM Bp and human Qp walking
may demand more precise neuronal control, such as cor-
tical control, for unfamiliar body posture requirements
and/or intersegmental coordination as compared to their
habitual quadrupedal and bipedal walking patterns.
As described above, energy consumption is the one of
the key determinants in th e selection of a walkin g pattern
for animals [21]. Sockol et al. [13] investigated the cost
of transport in five chimpanzees between Qp and Bp
walking. They found that an individual comparison
showed a significant difference between Qp and Bp
walking in both juvenile males (n = 2) and adult females
(n = 3). For two juveniles, the cost of transport was sig-
nificantly lower during Qp walking and one of the adult
chimpanzees showed a similar result. Surprisingly, one
of the other two female adults showed a significantly
lower cost of transport during Bp walking (Bp lower
energy consumption chimpanzee). They also compared
the kinematic characteristics and found that the Bp lower
energy consumption chimpanzee extended her hips and
knees to a similar degree during Qp and Bp walking,
while the others showed more flexed limb orientation
during Bp walking. Even this chimpanzee, however, bent
her body axis rather forward and this suggests that the
body axis, or at least th e hip an d knee joints, functio nally
interlock during walking as in a hinge. The anatomical
difference in ischium between chimpanzee and human is
clear; the former is oriented distally, while the latter is
dorsally oriented. This spatial d ifference redu ces the abi-
lity of the hamstring, which produces an extensor mo-
ment when the femur is extended, relative to the p elvis in
chimpanzees. This biomechanical constraint seems to be
one of the main reasons preventing chimpanzees from
walking like humans. From these, it was also suggested
that slight kinematic changes induce large differences in
walking cost and are consistent with previous studies
showing that differences in posture can affect cost [22,
23]. For pregnant human females, the fetal load moves
her COG forward and thus bends her posture forward.
This destabilized Bp standing posture is usually com-
pensated by enhanced lumbar lordosis and translating the
COG backward [7]. Without this skeletal compensation,
such an instable posture must be counteracted by the
back and proximal lower limb muscles, requiring much
more energy than a stable posture. This is reasonable be-
cause the energy consumption must be smaller when the
COG is on the hip joint than when it is far from the hip.
From these facts, it is clear that the skeletal system also
affects the energy consumption during walking for indi-
vidual animals.
5. Conclusion
Overall, this study provides new data on EMG activities
and kinematics during Qp and Bp walking in both rats
and humans. Rats that acquired Bp walking capability
and humans showed several common kinematics and
EMG burst patterns during the same walking task. This
suggests that they share common neuronal control sys-
tems, at least at the spinal cord level. The differences in
posture during Bp walking may be due to the skeletal
constraints that do not allow four-legged mammals in-
cluding nonhuman primates to extend their hip and knee
joints and even the body axis. Due to this skeletal char-
acteristic, the elaboration of Bp walking requires a lot
more energy than for Qp walking in rat.
6. Acknowledgements
We thank all the student members of our laboratory for
their dedicated daily rat training and technical support.
This work was partly supported by Grants-in-Aid for
Priority Areas “Emergence of Adaptive Motor Function
through Interaction between Body, Brain and Environ-
ment” and challenging Exploratory Research from the
Ministry of Education, Culture, Sports, Science and Tech-
nology Japan to Mori F.
REFERENCES
[1] J. T. Robinson, “Early Hominid Posture and Locomo-
tion,” University of Chicago Press, Chicago, 1973.
[2] D. Schmitt, “Insights into the Evolution of Human Bipe-
dalism from Experimental Studies of Humans and Other
Primates,” Journal of Experimental Biology, Vol. 206, No.
9, 2003, pp. 1437-1448. doi:10.1242/jeb.00279
[3] D. C. Dunbar, F. B. Ho rak, J. M. Macpher son, D. S. R u sh-
mer, “Neural Control of Quadrupedal and Bipedal Stance:
Implications for the Evolution of Erect Posture,” Ameri-
can Journal of Physical Anthropology, Vol. 69, No. 1,
1986, pp. 93-105. doi:10.1002/ajpa.1330690111

T. HOSOIDO ET AL. 149
[4] J. T. Stern Jr. and R. L. Susman, “The Locomotor Anat-
omy of Australopithecus Afarensis,” American Journal of
Physical Anthropology, Vol. 60, No. 3, 1983, pp. 279-317.
doi:10.1002/ajpa.1330600302
[5] N. Wada, Y. Toba, W. Iwamoto, M. Goto, H. Miyata, F.
Mori and F. Morita, “Investigation and Characterization
of Rat Bipedal Walking Models Established by a Training
Program,” Brain Research, Vol. 1243, 2008, pp. 70-77.
doi:10.1016/j.brainres.2008.09.034
[6] T. Hosoido, M. Goto, Y. Sano, F. Mori, K. Nakajima, F.
Morita and N. Wada, “Hoffman Reflex in a Rat Bipedal
Walking Model,” Neuroscience Letters, Vol. 505, No. 3,
2011, pp. 263-267. doi:10.1016/j.neulet.2011.10.035
[7] K. K. Whitcome, L. J. Shapiro and D. E. Lieberman,
“Fetal Load and the Evolution of Lumbar Lordosis in Bi-
pedal Hominins,” Nature, Vol. 450, No. 7172, 2007, pp.
1075-1078. doi:10.1038/nature06342
[8] E. Hirasaki, N. Ogihara, Y. Hamada, H. Kumakura and M.
Nakatsukasa, “Do Highly Trained Monkeys Walk Like
Humans? A Kinematic Study of Bipedal Locomotion in
Bipedally Trained Japanese Macaques,” Journal of Hu-
man Evolution, Vol. 46, No. 6, 2004, pp. 739-750.
doi:10.1016/j.jhevol.2004.04.004
[9] F. Mori, K. Nakajima, A. Tachibana, C. Takasu, M. Mori,
T. Tsujimoto, H. Tsukada and S. Mori, “Reactive and An-
ticipatory Control of Posture and Bipedal Locomotion in
a Nonhuman Primate,” Progress in Brain Research, Vol.
143, 2004, pp. 191-198.
doi:10.1016/S0079-6123(03)43019-7
[10] A. Tachibana, F. Mori, C. A. Boliek, K. Nakajima, C.
Takasu and S. Mori, “Acquisition of Operant-Trained Bi-
pedal Locomotion in Juvenile Ja panese Monke ys (Macaca
Fuscata): A Longitudinal Study,” Motor Control, Vol. 7,
No. 4, 2003, pp. 388-410.
[11] S. Hayama, M. Nakatsukasa and Y. Kunimatsu, “Monkey
Performance: The Development of Bipedalism in Trained
Japanese Monkeys,” Acta Anatomica Nipponica, Vol. 67,
No. 3, 1992, pp. 169-185.
[12] H. Preuschoft, S. Hayama and M. M. Günther, “Curva-
ture of the Lumbar Spine as a Consequence of Mechani-
cal Necessities in Japanese Macaques Trained for Biped-
alism,” Folia Primatologica (Basel), Vol. 50, No. 1-2,
1988, pp. 42-58. doi:10.1159/000156333
[13] M. D. Sockol, D. A. Raichlen and H. Pontzer, “Chim-
panzee Locomotor Energetics and the Origin of Human
Bipedalism,” Proceedings of National Academy of Sci-
ences of United States of America, Vol. 104, No. 30, 2007,
pp. 12265-12269. doi:10.1073/pnas.0703267104
[14] M. W. Grabowski, J. D. Polk and C. C. Roseman, “Di-
vergent Patterns of Integration and Reduced Constraint in
the Human Hip and the Origins of Bipedalism,” Evolu-
tion, Vol. 65, No. 5, 2011, pp. 1336-1356.
doi:10.1111/j.1558-5646.2011.01226.x
[15] M. Hildebrand, “Symmetrical Gaits of Primates,” Ameri-
can Journal of Physical Anthropology, Vol. 26, No. 2,
1967, pp. 119-130. doi:10.1002/ajpa.1330260203
[16] M. J. MacLellan, Y. P. Ivanenko, G. Cappellini, F. S. La-
bini and F. Lacquaniti, “Features of Hand-Foot Crawling
Behavior in Human Adults,” Journal of Neurophysiology,
Vol. 107, No. 1, 2012, pp. 114-125.
doi:10.1152/jn.00693.2011
[17] W. A. Sparrow, “Creeping Patterns of Human Adults and
Infants,” American Journal of Physical Anthropology,
Vol. 78, No. 3, 1989, pp. 387-401.
doi:10.1002/ajpa.1330780307
[18] N. Dominici, Y. P. Ivanenko, G. Cappellini, A. d’Avella,
V. Mondi, M. Cicchese, A. Fabiano, T. Silei, A. Di Paolo,
C. Giannini, R. E. Roppele and F. Lacquaniti, “Locomo-
tor Primitives in Newborn Babies and Their Develop-
ment,” Science, Vol. 334, No. 6058, 2011, pp. 997-999.
doi:10.1126/science.1210617
[19] M. Nakatsukasa, N. Ogihara, Y. Hamada, Y. Goto, M.
Yamada, T. Hirakawa and E. Hirasaki, “Energetic Costs
of Bipedal and Quadrupedal Walking in Japanese Ma-
caques,” American Journal of Physical Anthropology,
Vol. 124, No. 3, 2004, pp. 248-256.
doi:10.1002/ajpa.10352
[20] M. Nakatsukasa, E. Hirasaki and N. Ogihara, “Energy
Expenditure of Bipedal Walking Is Higher than That of
Quadrupedal Walking in Japanese Macaques,” American
Journal of Physical Anthropology, Vol. 131, No. 1, 2006,
pp. 33-37. doi:10.1002/ajpa.20403
[21] K. Steudel, “The Work and Energetic Cost of Locomotion.
I. The Effects of Limb Mass Distribution in Quadrupeds,”
Journal of Experimental Biology, Vol. 154, No. 1, 1990,
pp. 273-285.
[22] T. S. Carey and R. H. Crompton, “The Metabolic Costs of
‘Bent-Hip, Bent-Knee’ Walking in Humans,” Journal of
Human Evolution, Vol. 48, No. 1, 2005, pp. 25-44.
doi:10.1016/j.jhevol.2004.10.001
[23] R. H. Crompton, L. Yu, W. Weijie, M. Günther and R.
Savage, “The Mechanical Effectiveness of Erect and
‘Bent-Hip, Bent-Knee’ Bipedal Walking in Australopith-
ecus Afarensis,” Journal of Human Evolution, Vol. 35,
No. 1, 1998, pp. 55-74. doi:10.1006/jhev.1998.0222