Trick Train your CNS, get stronger

There’s some truth in the idea that training stability can make you stronger. Provided that we are talking about the right kind of stability and the right kind of strength. (Around 2.400 words, estimated reading time 12-15 min.)

If you have ever tried to lift unstable stuff you probably noticed that it feels heavier than it is.

How you feel is not always a lie and there’s more to it than a trick of your brain. A simple-minded explanation would be that, as physics has it, moving stuff is heavier because the weight is a function of mass and acceleration. And if I was in a charitable mood I’d probably give some credit to BOSU® and Swiss ball lovers for clever exploits of physics.

Tyler-Durden-clever though.


The acceleration added by micro-movement amounts to nothing, but the micro-movements may be enough to destabilize the joints and cause micro-trauma and possibly injuries. And given that BOSU® lovers also have a fetish for ‘whole-body’ movements they can wreak havoc everywhere.

Which brings me to today’s topic.

I toyed with the idea of rationalizing my hatred for BOSU® and other non-trademarked unstable surfaces for a while. But the subtleties undermining the use of unstable platforms actually support the use of something else. Something interesting.

Part of that something is kinda speculative and I’m going to dump it somewhere else. The rest, I can develop here, but I need a bit of a buildup because that’s how you make crazy ideas look a little less crazy.

Or reinventing the wheel less like, well, reinventing the wheel.

Stability2 and more stability2

When I laid down the basics of mechanical stability from the last post you may have noticed how shy I was about joints.

That was intentional. First, because I’ve always intended to give joint stability2 its own treatment. Second, because the joints I’m the most interested in (the shoulder and hip joints) are missing a biomechanical model with the same level of detail as the Bergmark-McGill model of the spine.

Therefore, everything I could have said about the joints would have been rehashing old stuff about the spine or peddling speculations about the hips and shoulders. And it was not a day for speculations, but today is.

It is the scientific use of the imagination, but we have always some material basis on which to start our speculation.

Sherlock Holmes

Now, I’ve been nurturing ideas about the hip and shoulder for quite a while and I’ve recently updated them when I finally did my homework about biomechanics and quit trusting my gut feeling. They mesh well with some solid recent research but the biomechanics is actually the missing part of the picture. And so I’ll go for the “scientific use of imagination”. It’s not like I’m not in good company.

Bergmark’s model of joint stability2

Don’t worry, I’m not getting into a lengthy exposition of stability2 all over again. Obviously, I assume that the details of my last post on the topic are still fresh in your mind. If they aren’t you should get back to it.

Under the above assumption, the following diagram from (Bergmark 1989:27) should not seem too alien and the caption should be self-explanatory.[1]

Something that I left out of the last post because it might have been a bit confusing without an elaboration I did not have time for, is this: the total stiffness of the muscles stabilizing2 a joint is sometimes below the critical value for stable equilibrium. Here’s how Bergmark explains this relative to the ankle joint:

“[A] stiffness […] somewhat, but not much, lower than the critical value […] implies such a slow magnification of a disturbance that there will be ample time for the nervous system to arrange necessary corrections.”

Bermark, The Stability of the Lumbar Spine, p. 29

The reason stiffness is maintained under the critical value for stable equilibrium is obviously efficiency: a higher average stiffness would require more energy than momentary corrections. Besides, a higher average stiffness could not be maintained without a significant alteration of muscle fibers ratio.

But let’s not get sidetracked (there’s an aside about muscle fibers below anyway) and let’s get to the second point.

Failure at low loads

In 1996, Janek Cholewicki and Stuart McGill extended Bergmark’s approach to the in vivo spine, feeding the model with data about several loading tasks not covered by Bergmark’s study and estimating in vivo muscle stiffnesses for those tasks.[2]

But I am actually more interested in the hypotheses they put forward than in the empirical data they provided. Particularly interesting is a hypothesis about the risk of injury associated with low load: Cholewicki and McGill hypothesized that the risk of injury is high with lower and higher loads and lower with moderate loads.

The hypothesis is of course formulated for spinal joints. But it is based on (1) Bergmark’s General model of joint stability2, and: (2) well-entrenched hypotheses about muscles and central nervous system (CNS) wiring that apply across the human body. And thus we can generalize the hypothesis to other joints like shoulder, hip, and knee joints.

I’ll leave the details of the justification of this hypothesis (and its generalization) for a short aside, that you can skip if you want to get to my last point.

‘Postural’ muscles. It is usually taken for granted that postural muscles have a greater proportion of slow-oxidative muscle fibers (Type I, ‘slow twitch’) while power-generating muscles have a greater proportion of fast-glycolytic fibers (Type II, ‘fast twitch’, see here for a primer). The truth is, as usual, more complicated. First, there is a lot of inter-individual variations. Some are genetic and some epigenetic: for the latter, the make-up of muscle fibers is at least in part the result of how the muscles are used (muscle fiber conversion seems to occur as an effect of training). Second, calling a muscle ‘postural’ is more a matter of convention than a hard fact of physiology, let alone biomechanics. For instance, the rectus abdominis, the obliques and transverse are major contributors to posture; they do so not only by maintaining muscle tone but also by operating real-time corrections (the ‘anti-rotation’ function is the delight of ‘functional movement’ gurus, see item 14 on that comprehensive list). Accordingly, they have been found to be made up of about 55% of Type I and 45% of Type II fibers about 40 years ago (this seems to still be the reference).

Fast corrections. The possibility of fast-twitch postural muscles should by now seem fairly logical: in order to optimize the resources allocated to posture, it can be efficient to maintain stiffness below the critical value for stable equilibrium if the CNS and the muscles involved can compensate fast enough. However, and unfortunately, the CNS does not always fully anticipate torque-delta that could occasion failure. Which is how Cholewicki and McGill explain why a fatigued spine can buckle while its owner is picking a pencil at the end of the day (Cholewicki & McGill 1996:9). This kind of trouble could also occur at other joints. For instance, you could injure your shoulder reaching out to the kids on the back seat of the car. You, not me. Because I don’t own a car. And because my shoulder is fucked up enough already, and I would let the kids tear each other apart.

Proactive error corrections

In their 1996 paper, Cholewicki and McGill note that “there is evidence of co-contraction of antagonistic trunk muscles during most daily activities, which is energetically and mechanically costly” (p. 2). This co-contraction is evidence of real-time error-correction by the CNS when the mechanical stiffness of the passive structure is insufficient.

[C]o-contractions increase when people prepare for unexpected and sudden loading such as an unknown weight dropping in a hand-held bucket.

Cholewicki and McGill, Mechanical stability of the in vivo lumbar spine, 1996

Now, you probably remember that Bergmark’s hypothesis that muscles have intrinsic stiffness was motivated by considerations pertaining to the reaction time of the CNS: in short, the CNS would be too slow to protect joints reactively. Then again, the CNS is central for a reason, and can use all the resources of the body. Including expectations based on visual data, that in turn result in proactive measures for error correction.

Interestingly, the existence of proactive co-contractions of antagonists goes a long way to explain the myth of a ‘Law of Irradiation’ (in particular the part where it is supposed to overrule the stretch reflex). Posture-correcting co-contractions overrides the stretch reflex (that inhibits an antagonist), so there’s clearly something here, and I’ll come back to it in the future.

And now for practical applications, and if you follow me on Instagram, you should be guessing where I’m heading.


Since Cholewicki and McGill’s study, the hypothesis that joint failure happens at the extreme of the loading curve has been thoroughly investigated.

Not only has it garnered considerable empirical support but it has become textbook science. Accordingly, the tests used today to predict the risk of spinal injury are tests of muscle endurance, not muscle strength (and certainly not range of motion).

That is, however, only one side of the story. Proactive error-correction involves co-contractions of muscles with both slow-twitch and fast-twitch fibers. Thus, proactive error-correction also relies on the ability to contract powerfully. And it cannot occur repeatedly unless fast-twitch fibers are trained to relax and recover faster. That would be a “power-endurance” of sorts, not strength-endurance.

That being said, I’ll forsake terminological nitpicking. Instead, I’ll hypothesize that there is something like stability2-endurance with two components:

  • Strength-endurance: the ability to maintain mechanical stability (stability2) through muscle tone, predominantly for moderately demanding tasks.
  • Power-endurance: the ability to rapidly correct instability2 through phasic contractions and to recover so as to be able to do it again, predominantly for very-short duration tasks

So posited, strength-endurance would matter more for the low-end of the spectrum (long duration effort relatively far from maximal effort) while power-endurance would matter more for the high-end of the spectrum (repeated corrections close to maximal effort).

In practice, strong(wo)men loaded carries qualify as near-maximal strength-endurance and girevoy as power-endurance but it’s a topic another day.[3] For today, I’ll just suggest a simple trick to train stability2. I just need one more tiny bit of theory.

Motor learning

It is easy to forget that a muscular effort is both a mechanical and a neurological event. Accordingly, strength has both neurological and mechanical components. The effect of an “unknown weight dropping in a hand-held bucket” is a prime example of how interlocked those events are.

Now, someone who buys into the Hebbian theory of learning (after Donald E. Hebb, author of The Organization of Behavior [1949]) will hypothesize that the neurological component of strength can be improved by building neural pathways. For the record, I buy Hebbian learning lock, stock, and barrel, for professional reasons.

The Hebbian theory is often summed up by the slogan “neurons wire together if they fire together”, so training the neural component of strength is basically teaching the neurons you want to fire together to do just that. Hebbian learning would, for instance, explain why goblet squats and pressurized kettlebell swings (so-called ‘hardstyle’) transfer to squatting and deadlifting.

In a nutshell, even if you train them far from maximal strength parameters, you train your CNS to fire your trunk and leg extensors in the same way as you would want it to do so in the squat or the deadlift. You’re not reinforcing the mechanical component of strength, but definitely its neurological component. Next time you squat or deadlift, you will be better at coordinating all those muscles.

Trick Train your CNS

For all the above reasons, I contend that the following drill is actually useful for kettlebell lifters and perhaps for athletes who need to stabilize submaximal weight quickly with one hand.

Not the part where I drop it though.

In fact, for the anecdote, I had a stronger activation in the trunk and shoulder stabilizers with that one than with the one in the second video of this post, although the weight was about 4kg for the water jug and 12kfg for the kettlebell:

After reading Cholewicki & McGill, I understand better why. And I bet you do too. The weight difference here matters less than the inability of the CNS to anticipate its shifts (the sloshing of the water). Anticipations are easier with a kettlebell but I bet you could make them harder closing your eyes.

And that’s why I’d program no-kettlebell kettlebell jug-ing, but that’s the crazy stuff that I’ll dump somewhere else.


Louie Simmons is a genius

The crown achievement of this post was showcasing my Instagram and that’s Louie Simmons’ fault.

Without Louie Simmons and his Westside Barbell method, I might have invented concurrent training. Instead, I’ve just been reinventing the wheel, because my water jug is to kettlebell lifting what Louie’s bamboo bar is to powerlifting.

However, my point is a little more general than that. I’m willing to bet my shirt (although I obviously don’t wear it that often) that adding “stabilization” exercises to the routine of a beginner would make them stronger when they finally hit the big weights.

Of course, these exercises must be specific. No-kettlebell jug-ing may be fun for a powerlifter but thy’re not going to cut it specificity-wise. Bandbell bench press do. Then again, dips, push-ups and pull-ups have an often-neglected stability2 component, and they are great to prep people to hit the powerlifts.

Didn’t I mention reinventing the wheel?



[1]^Again, full reference: Anders Bergmark (1989) Stability of the lumbar spine, Acta Orthopaedica Scandinavica, 60: sup. 230, 1-54, DOI: 10.3109/17453678909154177.

[2]^ I didn’t say “tested the model” and that’s intentional. Cholewicki and McGill spend the whole final section of their paper explaining that the model cannot, in fact, be tested and that their study is a validation but not a test. Cholewicki, J., & McGill, S. M. (1996). Mechanical stability of the in vivo lumbar spine: implications for injury and chronic low back pain. Clinical biomechanics, 11(1), 1-15.

[3]^There’s not much research pertaining to the biomechanics of girevoy (kettlebell studies only consider short sets). For stong(wo)man, McGill, McDermott & Fenwick, found that the “Superyoke” event required more power than the hip extensor can develop and thus that strongmen had to recruit the quadratus lumborum to assist with gait. The same study found that yoke carry imposed higher compressive loads on the spine (due to co-contraction) than Atlas stones, in spite of the spine being close to neutral in the former and rounded in the latter. Obviously, Superyoke blurs the line between strength-endurance and power-endurance. McGill, S. M., McDermott, A., & Fenwick, C. M. (2009). Comparison of different strongman events: trunk muscle activation and lumbar spine motion, load, and stiffness. The Journal of Strength & Conditioning Research, 23(4), 1148-1161.doi: 10.1519/JSC.0b013e318198f8f7

[4]^ Whatever

2 Comments Add yours

  1. lmds1 says:

    Interesting, interesting. Do you think this could apply to circus dumbbell by lifting something wacky and even more unstable over your head? Or is circus dumbbell already unstable and that’s what makes it harder than a regular dumbbell to work with?

    1. Thanks for the appreciation. On the top of my head I’d say yes to both. The circus dumbbell is definitely unstable in itself and you could train overhead stability_2 by making it even worse. And your question gives me some ideas as to how 😉

Leave a Reply