CGA FAQ: Central Pattern Generators (CPGs)

Hi every one
I am debating the validity of the following statement (on the nay side):
Human locomotor is a products of central patterns generators (CPG).
In other word I do not agree with the above statment.
What do you think of the  CPG concept? Does it exist ? what is the evidence?
Any idea how to convene others that CPG is not valid concept in human being.
Looking for your input, please before Nov 30th.

Mohammad Alkhazim Alghamdi, MS, PT
Queen's University
Kingston, ON  K7M 2W8
Tel (613) 544 9531


I will take the yea side (with caveats) in regards to the existence of
locomotor CPG in humans.

First, the evidence for CPGs for locomotion in animals from fish through
primates is very strong.  To my knowledge, there is no direct evidence of
their existence in humans; the evidence is only by inference.  As in my
previous posting, I point to the paper by Whelen (1996) who reviewed the
cat locomotor work.  CPGs for locomotion in higher animals do not appear to
be comprised of a discrete set of neurons, such as observed in the
stomatogastric ganglion's control of peristalsis, rather the network is
distributed through several levels of the cord. Furthermore, CPGs are
modifiable and are very low in the locomotion hierarchy.   Locomotion does
have a very well defined hierarchically organized circuitry which utilizes
the spinal cord CPGs to produce the agonist-antagonist muscle activity
which results in the swing-stance limb movements.
        The medullary reticular formation locomotor region has been described as
the 'head ganglion' of locomotion -- a region where descending commands
converge before traveling to the cord.  The mesencephalic locomotor region
perhaps has the greatest control over locomotion, as progressively more
stimulation can induce the animal to walk, walk faster, then transition to
trotting, then galloping.  The cerebral cortex appears to play a minimal
role in locomotion, since decorticated animals locomote with minimal, if
any difficulty (locomotor behavior has been described in anencephalic
children).  There are other identified regions involved in locomotor
circuitry but these are a few of the critical domains.

        I feel that the most dramatic human evidence for a constrained neural
network that controls normal locomotion comes from the 1998 paper by
Visintin et al.  In that paper the results of a randomized study of 100
rehabilitation patients with cerebral cortex strokes were presented.  Half
of the patients were treated with traditional PT and half were treated with
body weight support gait training done over a treadmill.
        As expected, the traditionally treated patients improved during the six
weeks of rehab from a walking speed of 0.17 m/s to 0.25 m/s.  The BWS
trained patients improved significantly more from 0.18 m/s to 0.34 m/s.
The most dramatic effect was when the patients were re-evaluated three
months after discharge -- after three months of not having therapy.  The
traditionally trained patients were walking 0.30 m/s while the BWS trained
patients improved to 0.52 m/s !  (For US therapists, like myself, those
numbers translate to final walking speeds of 0.7 mph for the traditionally
trained and 1.2 mph for the BWS trained, this is not a small difference, it
is a HUGE improvement.)
        It is my position that when gait training patients with stroke using
traditional PT the focus is on teaching the patient cortical means of
locomotion, bypassing the normal brainstem control mechanisms. Cortical
control constrains locomotion to slower speeds, perhaps by requiring enough
time for afference to reach the cortex.  Cortical control also increases
the attentional demands for locomotion.  I would posit that by using PWB
gait training the patients were allowed to re-activate (for lack of a
better term) the normal brainstem mechanisms of locomotion.  In these
patients the brainstem was not damaged so more normal locomotion could be
        Hesse et al's work (1994 ) supported this finding in a paper that used a
similar training paradigm for patients who were an average of three months
post stroke, who, after three weeks of traditional PT were non-ambulatory
(they required the assistance of two PTs to walk).   In the three patients
reported, after only three weeks of body weight support gait training
their gait improved to where they needed only intermittent physical support
or only verbal cues.  Such a rapid effect would not have been expected
unless the body weight support gait training was able to recruit a
mechanism that was not utilized by traditional PT -- perhaps a mechanism
already present for locomotion which was undamaged by their cortical injury.
        Other work which supports the presence of a CPGs in humans includes the
body wight support treatments for walking following spinal cord injury, the
pharmacological stimulation of locomotion, and the sensory electrical
stimulation paradigms.

        In wrapping up, I think that the presence of spinal CPGs for locomotion
will be found in humans, in some form.  Although CPGs may be present, they
are very low in the hierarchical control of locomotion.  Furthermore, while
there is central neural control of locomotion, those control mechanisms do
utilize sensory afference to modulate locomotor output.  Furthermore,
locomotion is constrained or augmented by the mechanics of the total
system, including intrinsic factors, such as muscle strength or limb
length, and extrinsic factors, such as the coefficient of friction of the
locomotor surface.

Paul H.

Hesse S, Bertelt C, Schaffrin A, Malezic M, Mauritz KH. Restoration of gait
in nonambulatory hemiparetic patients by treadmill training with partial
body-weight support. Arch Phys Med Rehabil 1994; 75:1087-1093.

Visintin M, Barbeau H, Korner-Bitensky N, Mayo NE. A new approach to
retrain gait in stroke patients through body weight support and treadmill
stimulation. Stroke 1998; 29: 1122-1128.

Whelan, PJ 1996 Control of locomotion in the decerebrate cat. Progress in
Neurobiology 49:481-515.   (This is presently my favorite review paper on
the neurology of locomotion)

Paul D. Hansen, Ph.D., P.T.
  Fircrest Physical Therapy
  1105 Regents Blvd., Suite C
  Fircrest, WA  98466

Voice:  253.565.7796
Fax:    253.565.7836

Hi, Dear all,

It's nice to read Dr. Hansen's post, surely I take the yea side also, to the
existence of the mechanism of CPGs. There are quite a few significant
contributions on vertebrate locomotion, including human locomotion, governed
by locomotor CPGs, such as:

Grillner, S., "Locomotion in vertebrates: central mechanisms and reflex
interaction", Physiol. Rev., 55: 247-304, 1975.
Grillner, S., "Neurobiological bases of rhythmic motor acts in vertebrates",
Science, 228: 143-149, 1985.
Pearson, K.G., "Common principles of motor control in vertebrates and
invertebrates", Annu. Rev. Neurosci., 16: 265-297, 1993.

Just as Dr. Golubitsky et al. argued, although current neurophysiological
techniques are unable to isolate such circuitry (CPGs) among the intricate
neural connections of complex animals, but the indirect experimental evidence
for their existence is strong. His group developed a very interesting,
mathematically strict, general modular network for retrieval the mechanism of
CPGs underlying legged locomotion, see:

Golubitsky, M., Stewart, I., Buono, P.L., and Collins, J.J., "A modular network
for legged locomotion", Physica D, 115: 73-86, 1998.

Zhijun Yang

Caixa Postal 68511
CEP: 21945 - 970
Rio de Janeiro - RJ

Dear Colleagues:

This discussion has tweeked my interest, largely because of a paper I just finished reading. The paper was published in the J.
Exp. Biol. just 2 days ago by Robert Full and Daniel Koditschek and outlines a "neuromechanical hypothesis" to legged
locomotion. As someone who initially trained as a biomechanist, and who has subsequently become quite interested in motor
control, this paper addresses several issues which I have believed for a long time, but have seen little discussion of in the
scientific literature.

Traditionally, there have been two approaches to studying locomotion: (1) neurophysiology, and (2) engineering and
biomechanics. These approaches, however, have only rarely (if at all) crossed paths. However, locomotion is a behavior that is
expressed by a whole organism, which means that in order to fully understand the behavior that is being expressed, one must
consider BOTH the neurophysiological structures AND the mechanical structures of the animal. Full and Koditschek quote
another paper by Mark Raibert and Jessica Hodgins (1993) as follows:

Many researchers in neural motor control think of the nervous system as a source of commands that are issued to the body as direct orders. We believe that the mechanical system has a mind of its own, governed by the physical structure and laws of physics. Rather than issuing commands, the nervous system can only make suggestions which are reconciled with the physics of the system and the task [at hand].

Full and Koditschek then go on to give several examples from the literature, and conclude that:

During slow, variable-frequency locomotion tasks requiring precise stepping, the nervous system probably dominates control by way of continuous feedback... [However,] the dynamics of the mechanical system most probably begins to dominate at intermediate and fast speeds.

In my own research (currently submitted for publication), we found that diabetic patients with severe peripheral neuropathy were able to maintain stable walking patterns (for at least 10 minutes continuously). Our results demonstrated that peripheral sensory feedback is not ESSENTIAL for producing stable locomotion in humans. Most likely, this result occurred because these patients were able to substitute information from other sensory sources (i.e. vision), or because locomotion at
self-selected speeds is largely a mechanical behavior as Full and Koditschek suggest, or both. My bet is on both.

Jim Collins (and others as well) developed some very nice coupled-oscillator models of CPG output that provide insight into
gait transition mechanisms (at least in quadrupedals), but these CPG models still do not produce robust and adaptive locomotor behavior. On the opposite end of the spectrum, passive dynamic walking machines (Garcia et al., 1998) have been constructed which produce stable locomotor behavior with NO actuators or nervous system at all. Gentaro Taga, on the other hand, was able to produce robust locomotor behavior by coupling some form of CPG model to a mechanical model of a walking human.

My point in bringing this up is that I believe there is indeed a substantial amount of literature supporting the evidence of CPGs in
quadrupeds and there are likely CPGs of some sort in humans as well. However, these neural circuits are intimately
inter-connected within a mechanical system that must obey the laws of mechanics, regardless of what the CNS may or may not
WANT it to do. Thus, if we want to understand how locomotor patterns are generated and controlled, we must think both
mechanically as well as neurophysiologically.


Collins, J.J. and I.N. Stewart, Coupled Nonlinear Oscillators and The Symmetries Of Animal Gaits. Journal of Nonlinear
Science, 1993. 3: 349-392.

Collins, J.J. and S.A. Richmond, Hard-Wired Central Pattern Generators For Quadrupedal Locomotion. Biological
Cybernetics, 1994. 71: 375-385.

Full, R.J. and D.E. Koditschek, Templates and Anchors: Neuromechanical Hypothesis of Legged Locomotion. Journal of
Experimental Biology, 1999. 202(23): 3325-3332.

Garcia, M., A. Chatterjee, A. Ruina, and M. Coleman, The Simplest Walking Model: Stability, Complexity, and Scaling.
Journal of Biomechanical Engineering, 1998. 120(2): 281-288.

Golubitsky, M., I. Stewart, P.-L. Buono, and J.J. Collins, A Modular Network for Legged Locomotion. Physica D, 1998.
115: 56-72.

Taga, G., A Model of the Neuro-Musculo-Skeletal System for Human Locomotion I: Emergence of Basic Gait. Biological
Cybernetics, 1995. 73(2): 97-111.

Taga, G., A Model of the Neuro-Musculo-Skeletal System for Human Locomotion II: Real-Time Adaptability Under Various
Constraints. Biological Cybernetics, 1995. 73(2): 97-111.

Taga, G., A Model of the Neuro-Musculo-Skeletal System for Anticipatory Adjustment of Human Locomotion During
Obstacle Avoidance. Biological Cybernetics, 1998. 78(1): 9-17.

        Jonathan Dingwell, Ph.D.
        Postdoctoral Research Associate

        Rehabilitation Institute of Chicago
        345 East Superior, room 1401
        Chicago, Illinois, 60611
        Phone:    (312) 238-1233      ***  NOTE: NEW PHONE #  ***
        FAX:      (312) 908-2208

Hi Mohammad,

It sounds like you have already made up your mind and you are just looking
for information to support your view that humans do not have a locomotor

But if you are open-minded about the idea, I suggest that you read the
following article, which to my mind,  provides the best evidence yet for a
locomotor CPG in man:  Calancie B, Needham-Shropshire B, Jacobs P, Willer
K, Zych G, Green BA.  Involuntary stepping after chronic spinal cord
injury. Evidence for a central rhythm generator for locomotion in man.
Brain.  1994; 117(Pt 5):1143-59.

Clearly, it is difficult to obtain evidence of CPG-related locomotor
activity in humans (and primates) with complete SCI.  But to my mind this
does mean it's not there, it simply has more to do with the inhibitory
mechanisms that have evolved in higher animals that make it difficult to
elicit this activity (See: Fedirchuk B, Nielsen J, Petersen N, Hultborn H.
Pharmacologically evoked fictive motor patterns in the acutely spinalized
marmoset monkey (Callithrix jacchus).  Exp Brain Res.  1998 Oct;
For further reading I would refer you to the following articles:

Pinter MM, et al; Gait after spinal cord injury and the central pattern
generator for locomotion. (Spinal Cord, 1999 Aug)

Dimitrijevic MR, et al; Evidence for a spinal central pattern generator in
humans. (Ann N Y Acad Sci, 1998 Nov 16)
McKay WB, et al; Assessment of corticospinal function in spinal cord
injury using transcranial motor cortex stimulation: a review. (J
Neurotrauma, 1997 Aug)

Dimitrijevic MR, et al; Motor control physiology below spinal cord injury:
residual volitional control of motor units in paretic and paralyzed
muscles. (Adv Neurol, 1997)



Edelle "Edee" Field-Fote, PhD, PT
University of Miami School of Medicine, Division of Physical Therapy
5915 Ponce de Leon Blvd, 5th FL; Miami, FL 33146
& The Miami Project to Cure Paralysis
M-W-F: 305-585-7970 (The Miami Project)
T-Th: 305-284-4535 (Division of PT)
FAX:  305-284-6128

I would suggest to consider also the papers from the Forssberg group at
Karolinska (e.g. Forssberg, 1985 Exp. Brain Res., Hirschfeld and Forssberg,
1992 J. Neurophys.).

They make a clear distinction between Human gait and Locomotion. Human
mature gait is plantigrade and characterised uniquely by a marked heel
strike in front of the body with toe in the air, followed by toe contact
and heel lift with toe still in contact. Moreover there is a marked knee
flexion-extension in the swing phase. There is a secondary locomotor
pattern which can be termed infant stepping where there is no ankle
extension at the end of the stance and flexion-extension at the knee is not

Forssberg reports of anencephalic infants able to produce similar patterns
of infant stepping; moreover, the stepping movement patterns are already
present as spontanous coordinated kicking movements in supine infants. All
of these postulate that CPG are present at a very low level in the spinal
cord, are innate, but they can generate only stepping locomotion.

From Parkinsonians and hemiplegic patients some evidence has been acquired
that alterated sensory feed-back produce alterated locomotion (e.g. toe
walking), which suggests that brain stem CPG can be largely modified.

Their picture is that CPG in the brain stem have been acquired
phylogenically, and are genetically hardwired in the brain stem ciruits.
These are the patterns which are used by lower vertebrates for locomotion.
In Humans a higher form of locomotion developed late epigenetically, over a
larger set of motor structures.

I would very much like to see the video of the patients of Dr. Hansen after
recovery and see their re-acquired locomotion pattern

Best regards.

N. Alberto Borghese
Laboratory of Human Motion Study and Virtual Reality
Istituto Neuroscienze e Bioimmagini, CNR
LITA, via Fratelli Cervi, 93 - 20090 Segrate (Milano) - Italy
Tel. +39-02-21717.544 office   Fax  +39-02-21717.558 

A brief note regarding patient videos.  Unfortunately, at the present time
my patients have not granted me permission to show their videos publicly.
Data regarding the kinematics and kinetics following body weight support
gait training in hemiplegia has been presented by Visintin et al and Hesse
et al.  Their results show that following BWS gait training hemiplegic gait
moves toward "normal" in traditional kinematic and kinetic measures.

Also, I would like to second Alberto Borghese's recommendation of the
Forssberg group's papers;  they are excellent discussions of locomotor

Hesse S, Bertelt C, Schaffrin A, Malezic M, Mauritz KH. Restoration of gait
in nonambulatory hemiparetic patients by treadmill training with partial
body-weight support. Arch Phys Med Rehabil 1994; 75:1087-1093.

Visintin M, Barbeau H. The effects of body weight support on the locomotor
pattern of spastic paretic patients. Can J Neurol Sci 1989; 16:315-325.

Visintin M, Barbeau H. The effect of parallel bars, body weight support and
speed on the modulation of the locomotor pattern of spastic paretic gait. A
preliminary communication. Paraplegia 1994; 32:540-553.

Paul H.

Paul D. Hansen, Ph.D., P.T.
  Fircrest Physical Therapy
  1105 Regents Blvd., Suite C
  Fircrest, WA  98466

Voice:  253.565.7796
Fax:    253.565.7836

Back to FAQ