Training from Scratch (IV) – Stability (2)

‘Functional’ exercise pays lip service to biomechanics but forgets about mechanical stability, and that’s too bad because you can’t spell ‘biomechanics’ without ‘mechanics’. (Around 3.600 words, estimated reading time 18-20 min.)

In part (III) I argued that stability1 and balance get gently mixed up even by excellent people but with stability2 gentleness is about to go out the window.

Sability2 is mechanical stability. It is slightly more complicated to define than stability1 because it requires a handful of auxiliary notions that are not as straightforward as centers of mass and bases of support. Also, it is difficult to disentangle from stability1, just as stability1 is from balance.

But the main issue with mechanical stability is not sophistication. It’s ignorance.

And rest assured that I’m not tilting at windmills, beating up strawmen, or belittling otherwise skilled professionals. Joe PT Functional from part (III) is a fictional character and he is ignorant because I made him so. That’s a strawman. But what happened when the concept of mechanical stability was introduced in biomechanics?

The engineers who understood the mechanics did not have the biological or clinical perspective and the clinicians were hindered in the interpretation and implications of the engineering mechanics.

Stuart McGill, Low Back Disorders (third edition), p.156

If we’d ask Stuart McGill, his response would be sobering. And so I ask you: if engineers and clinicians had understanding issues, could you expect fitness professionals to grasp mechanical stability in a pinch?

I know I wouldn’t. And if you’d agree with me, you would not be wrong. In fact, the ‘hindered interpretation’ of mechanical stability is still widespread today among people who, honestly, should really know way better. Including prominent exercises scientists. That’s not even too difficult to verify, thanks to the ‘fair use’ provision of copyright laws (of which I’ll take full advantage).

But in part (III) I promised I’d be constructive and so I’ll confine the worst examples to asides and footnotes. Because today, I’ll close a cycle within this series, and given what’s coming next, planning for the future is more important than bitching complaining about the past.

But first, let’s get to the topic of the day, namely mechanical stability.

You can’t spell ‘biomechanics’ without ‘mechanics’

The engineering concept of mechanical stability, which I’ll refer to as stability2, was imported into biomechanics in the late 1980s.

The importer, Anders Bergmark, belonged at the time to the Department of Solid Mechanics at the Lund Institute of Technology (Lunds Tekniska Högskola, or LTH). In 1989, he published a thesis titled Stability of the lumbar spine: A Study in Mechanical Engineering.[1]

The subtitle is a dead giveaway of Bergmark’s intention: address the lumbar spine as a mechanical system.

“Wait!”, I can almost hear you thinking, “isn’t that the basic premise of biomechanics?”

Well, yes, it is. And still, Bergmark’s work is what McGill was talking about when he mentioned understanding issues. So your next question should be: “How come engineers and clinicians could have gotten it wrong?'”

And here we go.

A non-mechanical puzzle and its mechanical solution

Before Bergmark, textbook biomechanics could only answer so many questions about the spine or more generally the back. The reference model was the force couple model of the back. Although still popular in exercise science due to its simplicity[2] the model was severely limited:

The only (but anyhow very important) question that can be answered by using [the force couple] model is: What are the approximate magnitudes of the compression and shear force in the spine and the tensile force in the back muscles at different loading on the human body at different postures?

Anders Bergmark, The Stability of the Lumbar Spine, p.5

The model shows that the spine must withstand very high forces both parallel to its axis (compression) and perpendicular to it (shear), and those high forces suggest a role for muscle support. And thus, the model was supplemented by “assumptions about “non-mechanical” force distribution” (Bergmark, 1989:5), where “non-mechanical” means controlled by the central nervous system (CNS) playing the role of an active supporting system in a mechanical structure (aking to a servo-motor).

The spinal system would collapse (buckle) so quickly […] that nerve-muscle control would be too slow for corrections.

Anders Bergmark, The Stability of the Lumbar Spine, p.5

Anders Bergmark, The Stability of the Lumbar Spine, p.5

But here’s the snag: as fast as CNS control is, it is still too slow for the job. Bergmark, therefore, hypothesizes that active control is the last line of defense and that muscles must provide some passive support.

Before we dig into the details, here’s an extended quote from Bergmark’s introduction giving some contexts for the understanding issues mentioned by McGill.

Mechanical vs. ‘clinical’ stability. “It must be pointed out, that the term “stability” is often interpreted differently by physicians and physicist [sic]. In mechanics, stability is a well-defined concept, namely the ability of a loaded structure to maintain static equilibrium even at (small) fluctuations around the equilibrium position. If stability does not prevail, an arbitrarily small change of the position is sufficient to cause “collapse”, i.e. the structure moves further away from equilibrium.

For a physician the term “stability” is associated with “clinical stability”. According to White and Panjabi [Clinical Biomechanics of the Spine] (1978) the definition of “clinical stability of the spine” is as follows: “The ability of the spine under physiologic loads to limit patterns of displacement so as not to damage or irritate the spinal cord or nerve roots and, in addition, to prevent uncapacitating deformity or pain due to structural changes. Any disruption of the spinal components (ligaments, discs, facets) holding the spine together will decrease the clinical stability of the spine. When the spine looses enough of these components to prevent it from adequately providing the mechanical functions of protection, measures are taken to reestablish the stability.”

In mechanical terms, clinical stability is associated with the magnitude of the deformations when the spine is loaded. Thus the spine can be more or less clinically stable. Clinical stability can therefore be regarded as a continuously variable phenomenon.

Mechanical stability, on the other hand, is not continuously variable. The system is either stable or unstable (in rare cases it is indifferent, i. e. neither stable nor unstable).” (Bergmark, 1989:5)

A thought experiment

The spinal system must be mechanically stable in essentially the same way as purely passive engineering structures […] namely by possessing stiffnesses that are high enough.

Anders Bergmark, The Stability of the Lumbar Spine, p.5

Anders Bergmark, The Stability of the Lumbar Spine, p.5

Bergmark’s hypothesis is that mechanical stiffness contributes to the support system of the spine. Bergmark illustrates how muscle stiffnesses do the stability2 job with a thought experiment (didn’t I mention that they were useful in science?): a T-structure with a weight and a spring attached to it that stands vertically on a base when the force of the spring matches the pull of the weight (see below, reproduced from Bergmark 1989:25)

In both A and B, the T is standing because the potential energy of the spring (P) and of the weight (Q) cancel one another. But the spring in structure A is stiffer than the spring in B: it has a higher (potential) energy P than the spring in the structure B and thus could compensate for forces that would act in the same direction as Q.

The next stage of Bergmark’s thought experiment is to introduce a small disturbance in the system that adds up to Q (“a clockwise angular deviation Δφ”, depicted below), such that the potential energy of spring A could compensate for it, but not the potential energy of spring B. Explicitly, if the disturbance happened, the potential energy of both springs A and B would be converted in actual kinetic energy, but only spring A would pull the T back in place: with spring B, the T would still be slanted (clockwise).

The force Q tries to increase the disturbance angle Δφ, whereas the force P in the spring tries to move the system back to the equilibrium position Δφ = 0. Thus the question about stability and instability is reduced to the question about which one, P or Q, that wins.

Anders Bergmark, The Stability of the Lumbar Spine, p.25

Anders Bergmark, The Stability of the Lumbar Spine, p.25

So far, Bergmark’s thought experiment illustrates that both A and B are in equilibrium but only A is in stable2 equilibrium while B is in unstable2 equilibrium. In other words, that A is mechanically stable on it own, while B could only be made stable through non-mechanical control (a tightening of the screw).

And now for the last step of the thought experiment: substituting the spring with a muscle and identifying muscle stiffness with “passive stretching and spinal reflex tension modification” (Bergmark 1989:26). Passive stretching increases the potential energy stored in the muscle and is purely mechanical. The “spinal reflex” is the Liddell-Sherrington Reflex, or stretch reflex, and does not require CNS control. As such, it is not considered to be “active” (the equivalent of CNS control is tightening the screw by hand).

Bergmark’s thought experiment yields a general model of joint stability (I’ll come back to that in another post) and two special models: one for the ankle, and one for the spine. Bergmark’s model of the spine is the basis of McGill’s model reproduced below (from 2016:159) where the springs stand for stiffening structures that stabilize the lumbar spine in the back (a.) and the abdominal wall (b.).

In this representation, P is the force that stiffnesses have to counteract, and the spine buckles when energy P exceeds the potential energy due to muscle stiffnesses. Notice also that no single muscle matters: preventing buckling is a matter of coordinating the activity of a multitude or muscles.

A scientific revolution (?)

Bergmark’s hypothesis that muscles exhibit mechanical stiffness is deceptively simple and yet extremely powerful. It allows Bergmark to substitute the half-baked concept of ‘clinical’ stability with the well-defined concept of mechanical stability (stability2) with instant conceptual gains:

  1. ‘clinical’ stability assumes non-mechanical force distribution of obscure origins but stability2 is purely a mechanical concept and therefore, doesn’t (duh!);
  2. ‘clinical stability’ has degrees but stability2 is ‘all-or-nothing’ and therefore, hasn’t (a welcome mathematical feature);
  3. ‘clinical stability’ requires to track variations over time but stability2 is time-independent, and therefore doesn’t (another welcome mathematical feature)

By (1), (2) and (3) in other words, Bergmark obtains a full mathematical model of the stability2 of the spine where there was none. That’s the stuff scientific revolutions are made of.

Moreover, Bergmark does not limit himself to the spine. As I mentioned already, he proposes a blueprint for a general model of joint stability and provides a full mathematical model of the ankle joint stability2 as a proof-of-concept in two pages only (Bergmark 1989:27-29).

That seems awesome and beautifully clear. But clearly (too) something must have gone wrong somewhere (otherwise, McGill would have had nothing to complain about).

And indeed, something has.

Here’s why: by history-of-science standards, a concept with a 30-year history still qualifies as ‘newly introduced’. Newly introduced concepts are easily misunderstood, especially if they bear some relations with other, better-established concepts. And it is doubly the case with stability2, with both stability1 and “clinical stability” of the spine.

The successive editions of McGill’s Low Back Disorders (2002, 2007, 2016) offer multiple examples of lingering misunderstandings in clinical and performance circles, and of their consequences, but I picked just one illustration of the confusion between stability1 and stability2 (p. 158).

Now, since mere illustration does not qualify as fair use, I need to follow-up with some nitpicking some sort of critical discussion, and I’ll do that in the next section. But before I get to that, let’s wrap up this section with a few remarks about Bergmark’s math.

The math of stability2. “Clinical” stability would have required considering evolution over time and therefore differential equations with time derivatives. (cf. Mechanical vs. “clinical” stability). With the assumption that muscles have stiffness, the functional definition of stability2 for a given system is a characteristic function which takes n-uples of values as input, compares them (or the outcome of computations based on them) with certain other fixed values that characterize stability for that system, and returns ‘yes’ if the system characterized by the input verifies the concept, and ‘no’ if it does not. This is essentially (and mathematically) equivalent to a diagnosis program taking a one-time input and delivering a one-time output, rather than a monitoring program that has to be fed with data as time goes by and adjusts its assessment.

Critical values & optimization. In part (III), we saw that there is an optimal solution to a stability1 problem for a biped standing on one leg (and beyond, for any n-ped standing on n-1 limbs). Hence, a system S representing that biped (resp: n-ped) can be assigned a characteristic function comparing its state to a stable1 state of reference. Conversely, the existence of a characteristic function for stability2 of S depends on the solution to an optimization problem. In Bergmark’s thought experiment, there is a critical stiffness kcrit such that the system is stable iff k>kcrit, where k is the stiffness of the spring (for the geeks: kcrit= Q c/a2): when k verifies the equality for Δφ=0, the system also has a minimum potential energy, and k>kcrit thus characterizes the solution to an optimization problem. In the case of complex joints with multiple bridging muscles, k is a composite of the stiffnesses of the muscles and other tissues (ligaments, fascia, etc.)[3]

The limits of taxonomy

Nitpicking about scientific taxonomies is a time-honored philosophical pastime: Plato did it, Aristotle did it, both agree that Socrates had done it before them, and everybody else has done it ever since.

And I will sacrifice to the tradition, by showing that McGill’s taxonomy of “whole-body” vs. “spine” stability lends itself to criticism. Mostly because I have to, for otherwise, the pictures I pulled from his book would not fall under ‘fair use’. However, I won’t only be paying lip service to copyright laws, and this critical examination will actually be constructive.

But before I proceed, let’s insist on the obvious: everything that McGill writes about ‘stability’ soars loftily in the stratosphere of post-Bergmark biomechanics and casts a long shadow on the rest of exercise-science biomechanics which in comparison often flies just above the daisies.

Now, after the niceties, let’s bring on the nasties.

Nitpicking about stability

First, scroll back (up) to the picture of the biped of McGill’s fig. 6.5 (a.) and the quadruped of fig. 6.5. (b.) then come back here. Done? Good. Let’s call the biped Bob.

Bob can solve a stability1 problem by re-establishing a bipedal stance the same way Jane DM did in Thought Experiment #3 of part (III), and the same holds mutatis mutandis, for the unnamed quadruped. But both can also solve a balance problem by generating a force to counteract the force exerted on their structure by the pole. Hence:

Nitpicking claim #1: McGill’s “whole-body stability” is, in fact, a composite of stability1 and balance.

McGill’s distinction between ‘whole-body’ and ‘spine’ stability can, therefore, be refined, because whole-body stability may be further analyzed into stability1 and balance

Second, check the pictures below, borrowed from p. 19 of McGill and Marshall’s 2012 study on kettlebell snatches, swings, and loaded carries. The study concluded (among other things) on both theoretical and empirical grounds, that unilateral loaded carries challenged ‘spine’ stability qua stability2.[4]Let’s call the guy on the pictures Billy Bob.

Now, and clearly, both loaded carries cause a displacement of Billy Bob’s center of gravity compared to normal gait, that he has compensated for when stepping. This is, again clearly, a challenge to Billy Bob’s “whole-body stability” qua stability1.

Also, clearly again, the 16 kg kettlebell used in the study creates a force that offsets Billy Bob’s balance between steps, challenging his “whole-body stability” qua balance. But lo and behold! While the force results from the pull of gravity on the kettlebell, its direction is (roughly) the same as the force resulting from McGill’s assistant pushing Bob in fig. 6.5 (a., B.). Hence:

Nitpicking claim #2: McGill’s claim that Bob’s predicament “is not spine stability but whole-body stability” is incorrect.

Hence, McGill’s distinction between ‘whole-body’ and ‘spine’ stability cannot be sharply maintained, because challenges to whole-body stability, spine stability, and balance, are correlated.

What did I just say?

Wait! Didn’t I just shoot myself in the foot?

Haven’t I spent over 10,000 words across two posts establishing distinctions between stability1, stability2, and balance, and claiming practical implications of these distinctions?

Well, this I did, but that I didn’t.

Sure, what holds for McGill’s taxonomy holds also for mine, but it’s a conditional: if the distinction between “whole-body stability”, “spine stability” and balance cannot be maintained in practice, then the distinction between stability1, stability2, and balance, cannot be maintained in practice either.

Second, both taxonomies still have practical implications. A stability1 problem and a balance problem may overlap to the point of being and distinguishable until they are solved, but you don’t solve them the same way (see Thought Experiment #4 in part (III).

By the same token, it is perfectly sensible to maintain that Bob’s challenge is one of “whole-body stability” based on how we want Bob to meet the challenge and that Billy Bob’s challenges are with “spine stability” for the same reason. Or that Bob’s and Billy Bob’s challenge are, respectively, of stability1 and stability2. Or, pending epic-level nitpicking, something even more complicated.[5]

Moreover, the distinction also matters in practice when performance is measured. Often times, metrics for stability1 and balance are used to justify conclusions about what is, in fact, stability2, without actual measurement. A good grasp on the taxonomy is what makes the difference between extrapolating from incomplete data, as McGill and Marshall did (cf. footnote 4); and sloppily extending one’s conclusions, which is all too common in the literature on fall prevention in elderly adults (but that’s a topic for another day).

And with this, I can conclude for today.

Conclusion: Let’s build a brick shithouse

The taxonomy of stability1, stability2 and balance matters for science, but what about you?

Well, I said at the beginning that with stability2, gentleness would be out of the window, but I also promised to be constructive, so I kinda got myself in a box. But I can let someone else put the final nail in the coffin of ‘functional’ training (and trainers) and their ‘stability’ stuff:

Of course, De Franco sells books, DVDs, programs, supplements, etc., so bullshitting might be in his interest and on his agenda. But I’m not overly familiar with the rest of his work, so I’ll just assume that he’s not as full of shit dishonest, ignorant, or both, as others.

Again, I promised to be constructive so I won’t name the others. Or their publishing houses. (But one of them rhymes with “Dumbledore”.)

Back to De Franco’s argument then. It builds on the notion that, for athletic performance, ‘functionality’ is specificity. I made the same point in part (III), only more pedantically, but there was a point to the pedantry: extending the argument to activities of daily life (ADLs) is tricky. There’s case to be made that this trickiness caused the otherwise sensible idea of ‘functional training for ADLs’ to become what some philosophers of science would call a ‘degenerated research program’. Or a degenerated exercise program, if you will.

I’ll make that case in a scholarly article, one day, but for today, I’ll jump to the culprit: ‘functional training’ degenerated because stability1 was considered the paradigmatic function for ADLs. Now, if you substitute stability1 with stability2 as the paradigmatic function for ADLs, ‘functional’ training for ADLs collapses on athletic training, give or take.

  • ‘Give’: De Franco’s argument for loaded carries, which develop strength, endurance, stability{1,2} and athleticism (however it’s interpreted).
  • ‘Take’: athletic excellence is often at the expense of health markers (strength athletes compromise cardiovascular fitness, endurance athletes are vulnerable to sarcopenia).

With this remark, we reach a turning point: the foundations for ‘training from scratch’ are laid down, and all that’s left is wrapping up everything in a nice ‘functional’ system that can take fitness goals as input and spit out a program that would make anyone who follows it as stable as a brick shithouse.

And that’s what I’ll do in the next part. No less.



[1]^All quotes are from Bergmark’s introduction. The complete reference is: Anders Bergmark (1989) Stability of the lumbar spine, Acta Orthopaedica Scandinavica, 60: sup. 230, 1-54, DOI: 10.3109/17453678909154177.

[2]^It is still common nowadays not to go beyond the force couple model. For instance, the second edition of Zatsiorky & Kraemer Science and Practice of Strength Training (2006) still sticks to pre-Bergmark spine biomechanics, and lists the same White and Panjabi textbook quoted by Bergmark for the “old” definition of ‘clinical stability’ as the only reference textbook (actually the second edition from 1990, but it still only covers the force couple models). Worse, they still argue for some exercises and corrective measures that post-Bergmark biomechanics has debunked, for instance the still popular pelvic tilt (chap 7, p. 147), deconstructed by McGill (2016) in Low Back Disorders: Evidence-based Prevention and Rehabilitation, Third Edition, Champaign IL, Human Kinetics (p. 181, 225 and 239).

[3]^ For an explanation of stability2 in terms of minimal potential energy, see Bergmark (1989:25-26) or McGill (2016:ch.5).

[4]^ McGill, Stuart M & Marshall, Leigh W (2012), Kettlebell Swing, Snatch, and Bottoms-Up Carry: Back and Hip Muscle Activation, Motion, and Low Back Loads, The Journal of Strength & Conditioning Research, 26 (1): pp. 16-27, doi: 10.1519/JSC.0b013e31823a4063. Conclusions about total spinal load and spine stiffness are based on the combination of actual measures (“all muscles except the LEO [Left External Oblique] increased their activation with the bottoms-up carry”, p. 24) and theoretical arguments (“the sum of the muscle activities will probably be important in terms of total load on the spine and in terms of spine stiffness (not quantified in this study)”, ibid.), while joint compression and shear loads were measured and found “significantly greater in the bottoms-up position compared with that in the racked position.” (ibid.)

[5]^Given that “The magnitude of differences in muscle activation varied from 0.1% MVC [Maximal Voluntary Contraction] (between the racked and normal walking trials) to 14.3% MVC (between the bottoms-up and normal walking trials)” (McGill & Marshall, 2012:24), one could argue that Billy Bob’s problem is predominantly stability1 and secondarily balance in rack carries; and predominantly stability2 in bottoms-up.

Leave a Reply