Teach in #24, Lombard's Paradox: What people said...


Dear CGA:

                      Two things about Lombard's paradox and Dr. Kirtley's reluctance to cover
                      the main muscles and their actions.

                      First, several years ago I placed a very large and bolded statement in my
                      motor control course notes that the classic muscle action charts in
                      anatomy and kinesiology books are partially correct at best, and woefully
                      inaccurate at worst.  Much data has been presented in the past 5-10
                      years or so (e.g., see Herzog's work) on task-specific muscle activation,
                     compartmentalization, and more; all of which suggest that our classic notions
                      of muscle actions need rethinking.

                      Second, and in line with this thinking, was a presentation I attended
                      about using fMRI during real life tasks. The presenter showed leg scans
                      of an individual doing submax squats and discussed at length the
                      technology and color-coding system depicting the level of muscle
                      activity.  None of that interested me.  The only thing I remember was
                      that the vastus muscles were lit like a Christmas tree and the rectus femoris
                      was dark with inactivity (I can't remember what the hamstrings were doing).

                      One more explanation for Lombard's paradox?

                      Jeff
                      --
                     Jeffrey C. Ives, Ph.D.
                      Associate Professor
                      Dept. Exercise &  Sport Sciences                Email: jives@ithaca.edu
                      Ithaca College                                  Phone: 607-274-1751
                      Ithaca, NY  14850    USA                                Fax: 607-274-7055



                      Hello All also, and happy new year.

                      Chris, I can't begin to comment authoritatively about your real question
                      here, but want to say that in the late 1960s, in my orthopedic residency
                      at Pitt, I learned--can't remember from whom--that if the leg/foot were
                      fixed[e.g. planted] that the gastroc heads could pull the femoral condyles
                      posteriorly--it made intuitive sense to me, and still does.
                           I also feel very leery of accepting computer modeling [of specific
                      postulated actions such as this] and think the models must be, as you
                      say, " extremely difficult to construct with any confidence." It makes
                      more sense to me to work toward imaging actual muscles in action, eg w
                      MRI when they get it 'miniaturized' etc.  Surgically we see variations in
                      muscle bulk, attachments, tendon width/thickness etc--how much do
                      these modify motions?

                     Mary Williams Clark MD
                      Pediatric Orthopedics
                      Sparrow Regional Children's Center
                      Suite 145, SPB
                      1200 E. Michigan Ave
                      Lansing MI 48909-7980
                      517-364-5434



Hello all,
   These paradoxical joint accelerations have come up in some of my
computer modeling work.  Like Felix Zajac before us, Scott Delp and I found
that the gastrocnemius could extend the knee under certain conditions (J.
Biomech 29:723-733, 1996).  We also found that the rectus femoris could
extend the hip and that the hamstrings could flex the hip.  Each of these
biarticular muscles produced moments that were consistent with their
anatomical descriptions, but accelerations that (under certain conditions)
appeared to be paradoxical.

   Many of the people to whom we presented these findings were skeptical of
them, and we were curious about whether these accelerations were artifacts
of our modeling assumptions, so we decided to construct a physical
model.  A movie (3 MB .avi) showing the physical model results may be found at

http://www.celos.psu.edu/kines488

The movie shows a horizontal double pendulum that was constructed to
represent the thigh and shank (same lengths, but 1/17 of the mass and
moment of inertia).  A wire that passes behind the knee (roughly
representing the moment arms of semimembranosus) is pulled on, and it is
clear that this action initially produces hip flexion rather than hip
extension.  The way that this paradoxical acceleration arises is through a
reaction force at the knee that passes anterior to the hip.  The existence
and magnitude of this phenomenon depends on the configuration of the
system, segment inertial parameters, and musculotendon geometry.  I
sometimes use the following thought experiment to explain this: Imagine
that you are sitting down and that your hamstrings attach near your ankle
instead of near your knee.  In this configuration the hamstrings still
produce a hip extension moment, but it would not be hard to imagine that a
hamstrings contraction would produce hip flexion rather than extension.

Best regards,
Steve Piazza
Assistant Professor
Departments of Kinesiology, Mechanical Engineering,
Bioengineering, and Orthopaedics & Rehabilitation
Center for Locomotion Studies
29 Recreation Building
The Pennsylvania State University
University Park, PA  16802



Dear Steve,

That's a great demonstration and great video!

I wonder if I could pin you down a bit and ask the following questions:

1. Does this behavior only apply to two-joint muscles - the Zajac review
seems to me to imply that it can happen with uni-articular muscles too?

2. Is this the correct explanation for Lombard's paradox, or do you
think that is a separate issue?

3. Is there any way to predict this behavior without resorting to a full
biomechanical model/simulation? For example, I wonder if we could come
up with some basic rules for when it might be expected to happen?

Chris
--
Dr. Chris Kirtley MD PhD



Chris,
   Thanks for your comments - I have tried to answer your questions below.

>I wonder if I could pin you down a bit and ask the following questions:
>
>1. Does this behavior only apply to two-joint muscles - the Zajac review
>seems to me to imply that it can happen with uni-articular muscles too.

A uni-articular muscle can't produce an acceleration that doesn't agree
with its moment at the joint it spans (e.g., soleus can't produce ankle
dorsiflexion), but it is possible for uni-articular muscles to produce
motions at joints they do not span.  For example, hip flexors may produce
knee flexion during swing phase, and a rapid kicking motion produced by
vasti action would also tend to plantarflex the ankle.  I scanned through
the Zajac review to find the wording to which you referred, but I didn't
see it.

>2. Is this the correct explanation for Lombard's paradox, or do you
>think that is a separate issue?

It seems like a separate issue, at least in the sit-to-stand example you
have on the teach-in web site.  I haven't done a simulation of sit-to-stand
to answer this question definitively, but it seems to me that hamstrings
activation would produce a downward reaction force on the femur at the hip
and thus act as a knee flexor at the beginning of the motion.  A guess as
to why hamstrings activation occurs is that it might help to stabilize the
knee.

>3. Is there any way to predict this behavior without resorting to a full
>biomechanical model/simulation? For example, I wonder if we could come
>up with some basic rules for when it might be expected to happen?

Unfortunately I don't have a good answer to this question.  Even if you did
have a full-blown, muscle-actuated model, it's worth noting that there are
many possible reasons that the behavior of such a model might not be
realistic: joints are typically modeled as frictionless, soft tissues such
as skin are usually not modeled, etc.  Perhaps the basic rules you are
looking for could take the form of characteristic muscle-induced
acceleration profiles for common activities (such as normal gait), but this
wouldn't be helpful if you were trying to determine whether paradoxical
accelerations were present in a pathological gait pattern.

Steve Piazza
Assistant Professor
Departments of Kinesiology, Mechanical Engineering,
Bioengineering, and Orthopaedics & Rehabilitation
Center for Locomotion Studies
29 Recreation Building
The Pennsylvania State University
University Park, PA  16802



Dear all,

Can I just comment on Stephen's model (Paolo Selber can actually do this
with a folded bit of paper and a piece of thread but there must be
something about precisely how he makes the folds because despite a desk
strewn with the little folded strips they never seem to work for me).

There seems to be a perception that this is "magic" or at the very best
that it can only be understood using complex mathematical techniques. I
think, at least qualitatively, that the average mind can understand what is
happening if we think about it in the correct way.

The important thing to recognise is that the hip is the joint connecting
the whole limb (and not just the thigh) to the pelvis. In considering what
is happening at the hip we thus have to consider the movement of the centre
of mass of the whole limb and not just the thigh. If the knee flexes then
the CM of the whole limb moves posteriorly with respect to the thigh.
Consider doing this by isolated contraction of a uni-articular knee flexor
(short head of biceps). No moment is exerted at the hip so Newtons Laws say
that the centre of mass cannot move anteriorly or posteriorly. The only way
this can occur is if, at the same time as the limb's CM moves posteriorly
with respect to the thigh, the thigh itself moves anteriorly by a
compensating amount, i.e. the hip must flex. The short head of the biceps
is thus a hip flexor in this configuration!

Exactly the same is happening in Stephen's model except the hamstrings are
also exerting a small extensor moment at the hip. This is the only external
moment on the limb so the CM of the whole limb must move backwards. The
knee flexion however causes a larger backwards moment of the CM with
respect to the thigh than is required by the hip moment so the hip flexes
as described above (but less than it would do if the knee moment arose
purely from a uni-articular knee flexor). Because of the geometry of the
knee joint and the hamstrings (and possibly because of the stop limiting
knee flexion in the physical model), as the knee flexes more and more,
further knee flexion brings about a smaller change in the
anterior-posterior position of the limb's CM and the effect of the extensor
moment exerted at the hip starts to dominate. The initial hip flexion has
thus reverted to hip extension by the end of the experiment.

Note that this explanation is based purely on consideration of the movement
of the CM of the whole limb. It is not dependent on any considerations of
segment accelerations and would happen if the movements were performed
infinitesimally slowly. A fuller analysis must include inertial terms, and
of course in real life nothing happens in the horizontal plane so gravity
is also acting. The fact that Stephen's model does behave as predicted by a
consideration of the changing position of the CM within the limb suggests
that this is the key factor in explaining this "paradoxical" behaviour.

Anyone who is more deeply into the maths than I am like to comment?

Richard Baker PhD CEng
Gait Analysis Service Manager and Director of Research
Hugh Williamson Gait Laboratory, Royal Children's Hospital, Victoria 3052,
Australia
Tel: +61(0)3 9345 5354, Fax: +61(0)3 9345 5447

Adjunct Associate Professor,  La Trobe University
Honorary Senior Fellow, University of Melbourne



Dear Chris,
This is really an important challenge for us: how do we tell this
story to the kids (if we have biomechanics classes) and how do we tell
this to the docs (if we do clinical gait analysis).
The point, as already stated is that the limbs are connected: the
heel bone is connected to the leg bone, the leg bone is connected to the
thigh bone, etcetera. It makes a great difference whether muscle action is
studied in the open chain, as anatomy textbooks do without word of
mouth, or in the closed chain. In the open chain, e.g. when you are
seated in a chair with your right leg over your left, activity of your right
soleus plantarflexes the foot, which is should according to the textbooks.
When you are standing on the right foot, however, the footsole rests on
the immovable floor, and the only thing soleus can do is to raise the
whole body (minus right foot). It can thus be said that right soleus
moves your left little finger! (among other things). Two more examples of
soleus function are found in Zajac's first paper, in cycling and in the
stance phase of walking.

When you think of it, these open and closed chains, there is not
a very exact distinction. In the open chain example, knee extension also
moves the foot and can even give foot plantarflexion.

Thus: the action of a muscle STRONGLY depends on the relative
configuration of the limbs and on the interaction with the environment
(e.g. the floor) and, to a lesser degree and mainly in fast movements, on
the mass distribution.

What Lombard has to do with it is, as it seems, mainly that he
has already anticipated such effects, not so much  that he has predicted
them. On this subject I can only recommend Art Kuo's masterly chapter
10 "The action of two-joint muscles: the legacy of W.P. Lombard." in
"Classics in Movement Science" ed. by Latash and Zatsiorsky, Human
Kinetics (2001).  It is a tough piece of math, but worth while.
In an isometric task, it is found that the action of a muscle,
which of course consists of of a force on the environment, has in a
certain configuration a definite direction. This direction is constant and
the magnitude of the delivered force is linearly proportional to the
respective muscle force. This is not so difficult to see, I wrote a paper on
this isometric case in J.Biomech 34:1085-1089 (2001). Kuo makes a major
step further. In a movement the result of the action, usually the
acceleration of  segments, has also in a certain configuration a definite
direction(s). This direction is constant and the magnitude is again
linearly proportional to the respective muscle force. If you look very
carefully, you can see this also in Zajac I's formula's. This direction is not
the same as the force direction in the isometric case, which is relatively
easy to predict,  because of the different segment masses. Monoarticular
muscles around some joint all have the same force or movement
direction. Biarticular muscles have directions between the monoarticular
ones, which vary with the ratio of their moment arms. Their presence
thus provides the body with additional force or movement directions.
Finally, about Jeff Ives remark:

Second, and in line with this thinking, was a presentation I attended
about using fMRI during real life tasks. …… The only thing I remember
was that the vastus muscles were lit like a Christmas tree and the rectus
femoris was dark with inactivity.

I can refer to a very recent finding by Anand Nene et al.: "Is rectus
femoris really a part of quadriceps?" Gait & Posture 16: S121 (2002). The
answer, obtained with intramuscular EMG, is: "……functionally RF is a
different muscle and does not belong with Quadriceps."

Yours,

At Hof
Institute of Human Movement Sciences &
Laboratory of Human Movement Analysis AZG
University of Groningen
A. Deusinglaan 1, room 321
postal address:
PO Box 196
NL-9700 AD GRONINGEN
THE NETHERLANDS
Tel:   (31) 50 363 2645
Fax:   (31) 50 363 3150
e-mail: a.l.hof@med.rug.nl
http://www.ppsw.rug.nl/~ibw/



The question I have relates to the validity of the EMG recordings. Anand
Nene presented a beautifully crafted and completely convincing study at
ESMAC last September (Abstract in Gait and Posture 16;suppl 1:ps120-121)
comparing surface EMG of rectus and the vasti with fine wire EMG of the
same muscles during the same exercises (walking in this case).

Surface EMG showed clear monophasic activity of the vasti (late swing and
early stance) and biphasic activity of the rectus (with the vasti and also
in late stance and early swing). Most of us have been taught that this is
normal rectus and vasti activity and if you look through most of the
standard texts tracing surface EMG data to Inman, Sutherland, Winters or
others this pattern is clearly described.

However fine wire showed monophasic activity in both groups. Vasti were
active in late swing and early stance as expected. Rectus however was ONLY
active during late stance and early swing. Comparison of the data from fine
wire and surface EMG made a convincing case that the late swing/early
stance activity is in fact cross-talk from the vasti being picked up by the
rectus electrode. I've got a lot of respect for the use of EMG at Enschede
and if Anand is getting this I'm pretty sure that we all are. True to form
Perry is the only one who hasn't been hoodwinked by those dodgy surface
electrodes. Page 96 of Gait Analysis: Normal and Pathological function
clearly shows monophasic rectus activity at the same phase of the gait
cycle as Anand recorded it.

Anand's abstract is pretty heavily oriented to muscle function (rectus'
primary function during gait is as a hip flexor) but the message I take out
of it is more related to electrode function (surface EMG, even when
performed carefully, does not reliably record rectus activity). The
relevance to the Lombard paradox is clear. Is the observation a result of
inaccurate recordings of rectus activity? A combination of activity in
vasti (extending the knee) with hamstrings (augmenting hip extension once
the knee is stabilised by the vasti) is a perfectly sensible combination of
muscles to use to perform this activity (I assume the gluts are working as
well). Certainly any examination of the department's skeleton shows that
with the hip flexed 90 degress the hamstrings are ideally suited to act as
hip extensor and there's no real need to resort to SIMM to identify this.

How did Lombard record rectus activity? (The copy of the American Journal
of Physiology for January 1907 that normally sits on the shelf behind me is
inexplicably missing!) Do we believe him?

Richard Baker PhD CEng
Gait Analysis Service Manager and Director of Research
Hugh Williamson Gait Laboratory, Royal Children's Hospital, Victoria 3052,
Australia
Tel: +61(0)3 9345 5354, Fax: +61(0)3 9345 5447

Adjunct Associate Professor,  La Trobe University
Honorary Senior Fellow, University of Melbourne



 

Back to Teach-in