The Analytic Fitness™ Dictionary – Functional Movement

This entry of the Analytic Fitness™ Dictionary takes a deep dive in the murky waters of functional movement (around 3.600 words, estimated reading time: 17-19 min).

The term ‘functional’ in phrases like ‘functional movement’ and ‘functional training’ belongs in principle to kinesiology, the science of movement, but there is no consensus about its meaning.

Because there is no litmus test for what is scientific and what isn’t (see here), consensus rules how definitions of terms art are established in science. Thus, the lack of consensus about the definition of terms of art (which is more common than you’d think) typically triggers self-reflective efforts within scientific fields.

There is no consensus among fitness professionals about the meaning of ‘functional’ but that’s of little consequence. As much as they like to pretend that they are, the majority of fitness professionals are not experts in kinesiology in any relevant sense. For instance, in the sense that would make their disagreement interesting.

Fortunately, the notion of ‘functionality’ can be reconstructed from other notions of kinesiology and related disciplines on a piecemeal basis, modulo a modicum of analytical effort. Two examples would be the following:

  • Sports performance: a movement pattern is functional relative to some sport S if and only if reinforcing that movement pattern transfers to the performance in S.
  • Biomechanics: a movement is functional relative to a subset of the joints of the body if and only if the joints involved move relative to one another as they would in some natural movement pattern.

Under this type of reconstruction, “functional” becomes a parametric concept: the context provides some parameter (a particular athletic performance, a subset of joints’ relative positions, etc.) to which functionality becomes relative.

The “parametric” notion accounts for how experts use the terms (e.g. Joe DeFranco, for athletic field performance, or Stuart McGill in Low Back Disorders [276-285], for rehabilitation exercises). But we might, however, want a notion of functional movement that goes the extra conceptual mile, for instance by being:

  • general enough to explain the overlap between special concepts (eg: sport-functional, joint-functional, etc.);
  • normative enough to inform choices of exercises or training protocols based on what counts as “moving well” of “being a functional human being”;
  • reductive enough to require as fewer tests of functionality markers as possible.

As it turns out, there’s a proposal on the market that does just that, the so-called Functional Movement Screen™ , or FMS™, and there’s actually legit research about it. Still, I mean “on the market” literally, because it’s trademarked, so don’t keep your hopes up. In fitness-related matters, where there’s money, there’s bullshit, and FMS™ is no exception to that rule.

Still, FMS™ did not occur in a vacuum and tried to put some order in the piecemeal parametric approach, so let’s begin with that one.

Normal Function, Bottom-Up

‘Functionality’ is easily defined on a piecemeal basis but some generality across contexts would be more useful.

Recall sport performance and biomechanics: both agree that loaded carries are ‘functional’ but on the surface, the reasons are not the same. It’s worth spending some time on the example of loaded carries, as loaded carries illustrate how generality can be obtained ‘bottom-up’ that is, by successive abstraction from the data (bottom) to yield an integrative theoretical model (up).

First, it is somewhat obvious that sports performance lies beyond ‘normal’ function. The proverbial devil is hiding in the details of how ‘normal’ is defined here, so let’s just acknowledge that sports push the envelope of what the human body can do. But let’s fix intuitions with an example of loaded carries: my occasional training buddy Hugo Karlsson (@captainhugokarlsson on Instagram) performing a 20m yoke walk with 340kg (with apologies for the poorly framed video, shot by yours truly).

Second, ‘normal’ joint function cannot be maintained on overdrive if it is not already there under ‘normal’ conditions. Hence, there’s a case to make that biomechanics should serve as the foundation for ‘functionality’ in sport-performance contexts. (If you wonder why I chose biomechanics over physiology, there’s an aside about that at the end of the section.)

The opposite core muscles must function with the gluteal muscle to create walking proficiency. 

Stuart McGill, Low Back Disorders (3E), p. 126

By the same token, the study of extreme conditions can inform about normal conditions. In fact, studying extreme loaded carries in world-class strongman competitions informed biomechanists about one of the most basic movements of the human body, gait. For instance, here are two seemingly unrelated facts:

  1. Paralysis of the quadratus lumborum muscle impairs gait function: this has long been a minor mystery for gait biomechanists, who assumed that pelvis stabilization was guaranteed by hip abductors only (which the quadratus lumborum is not).
  2. Hip adductors are not strong enough for yoke carry: under extreme loads, hip adductors moment is insufficient to support gait function (the movement may require up to 150% of the strength hip abductors can develop).

Studying strongmen showed that they made up for the strength deficit with the obliques and quadratus lumborum, which McGill concluding: “The insight obtained demonstrated that the opposite core muscles must function with the gluteal muscle to create walking proficiency.” (Low Back Disorders, 3rd. Ed., p. 126). Below is an illustration based on McGill’s adapted to match the yoke carried by Hugo (you’ll know what I mean if you own the book).

The case of gait is an example of bottom-up elucidation of a motor function that can, down the line, inform choices of rehabilitation and training protocols. Again, the “bottom” is the data and the “up” is the unifying theoretical model. The theoretical consequence is that a theoretical model of normal gait cannot isolate hip abduction function from spinal lateral stabilization no more than a model of supra-normal gait. The practical implication is that any protocol aimed at re-patterning (rehabilitation) or improving (training) gait-like motor patterns must include some sort of trunk stability training.

Why not physiology? I didn’t consider physiology as a possible basis for the notion of ‘functionality’ but it would seem a natural choice. After all, basic physiology is usually the only science personal trainers have some minimal training in. The reason is that the kind of textbook physiology fitness professionals are familiar with is much less general than textbook biomechanics (and if you wonder about my fixation on textbook science, go check there). It looks at the origins and insertions of muscles in our skeleton and hypothesizes what movement should result from the contraction of said muscles. By contrast, biomechanics looks at the mechanical forces that our musculoskeletal structure as a whole must counteract in order to maintain its integrity while performing a given movement, then hypothesizes which muscles should contribute to maintaining said integrity. Furthermore, biomechanics has introduced a concept of stability (from mechanical stability) that even some biomechanists are still struggling with, and you can expect that physiology still has to fully integrate that. As a concrete example of biomechanics-vs.physiology , I’ll give you the discussion of the function of rotatores and intertransversarii muscles, summarized in McGill (2016), Low Back Disorders, pp. 71-72. Based on their insertions and attachments between vertebrae, theses small muscles are often billed in physiology textbooks as contributors to axial twisting (rotatores) and to lateral bending (intertransversarii) of the spine. But their billed function is about as accurate as the billed height of pro wrestlers. In fact, these muscles are too weak to perform these function alone, and as far as one can tell from in vivo data, they actually do not seem to activate much during torsion and flexion. Instead, they seem to serve a “position transducers” sending feedback to the CNS about the relative positions of the vertebrae, which in turn helps adjust the stiffnesses of other stronger muscles. The Wikipedia entry rotatores sums this biomechanical insight rather nicely, while the entry on intertransversarii is still somewhere in the middle between biomechanics and basic textbook physiology, noting that they are too weak to perform lateral bending alone. As for an example of physiological orthodoxy, you can peruse this lesson from, where the free excerpt ends up with: “Based on […] the location of the rotatores muscles, you probably guessed that these muscles function to move the spine and back. (If this was your guess, then you were right!)”. Bottom line: stick to Wikipedia, it’s cheaper and (at least in that case) more accurate. Also, biomechanics 1-physiology 0.[1]

Normal Function, Top Down

A ‘bottom-up’ approach cannot begin without tons of data, but the only requirement for a ‘top-down’ approach is an armchair.

Obviously, the value of top-down speculations depends heavily on whoever is sitting in the armchair: the stronger their empirical knowledge base, the better their educated guesses. In fact, the fate to the best top-down proposal for the notion of functionality was sealed from the start due to a weak-ass empirical knowledge base (if your guess is “not enough biomechanics”, you are right).

But I’m getting ahead of myself, so let’s backtrack a bit.

The bottom-up approach to functionality can easily achieve piecemeal generality. It picks a function (for instance, walking), studies the data all across the board (from pathological to normal to supra-normal) and progressively abstracts what is required to perform this function well.

But one may also want a higher level of generality and normativity. For instance, one may want to define “moving well” in general and deduce from it what “walking well” is. Or characterize a “functional human being” as one capable of moving well in some defined sense, and deduce from that what a “functional walker” would be.

Functional Movement Screen™ (FMS™) is so far the best top-down attempt at a general notion of movement functionality. A spoiler alert is in order tho: it’s “the best” because it’s the only one, and it has turned out not to be a particularly good one.

FMS™ originates with the efforts of a handful of physical therapists to organize the chaotic use of ‘functional’ by other physical therapists and make good money out of it. Proceeding from the intuition that a loss of function is correlated with injuries, its proponent wanted an evaluation method that would function (pun intended) as:

  • a diagnostic test indicative of lingering consequences of past injuries (including cumulative trauma leading to chronic dysfunction, such as gluteal amnesia), useful for choosing rehabilitation and protocols and corrective exercise;
  • a prognostic test indicative of risk of future injuries, in order to assess the need for prevention protocols, or (again) for corrective exercise.

From there, the idea was simply to pick a handful of easy-to-define fundamental movements and motor patterns, test them, and see how well the test worked. Here, “easy-to-define” means that there would be more-or-less immediate consensus on what constitutes an adequate performance of those movements and motor patterns. The approach was developed in the late 1990s, was systematically codified in the early 2000s, and publicized in a legit scholarly paper in 2006.

There’s a lot to say about the interplay between the conceptual underpinnings of FMS™ and the empirical research, but I’ll cut the long story short: the proponent of FMS™ did not elaborate much on their choice of patterns and put it into practice, and left to other researchers the tasks to generate data. Eventually, they looked at the data, gave a facelift to their initial approach to fit the conclusions based on the data, and considered themselves vindicated by the data (but we’ll see later about that).

Now, I must walk a fine line here because, as the “™” in FMS™ gives away, information about it is always that close to copyright infringement. But there’s a chart summing up the FMS™ tests and claims that I found here and there, which seems to indicate that the guys behind FMS™ are not overly proactive with cease-and-desist letters, so I’ll roll the dice and reproduce it.

As I said, initially, proponents of the FMS™ approach did not offer much support for their choices of movement patterns. But everybody seemed to agree from the get-go that the words “fundamental” and “functional” were not abused, so there was no harm (see the aside about Expert consensus at the end of this section for some elaboration).

Then, empirical studies started to pile up and apparently validated the approach for injury prevention. Whatever the criteria for choosing the motor patterns had been, the choice itself seemed to be vindicated.  The need for clarifying the criteria more or less disappeared and the authors of the original paper took stock in a self-congratulatory update of the 2006 paper in 2014 (see the aside about FMS™ pragmatism at the end of this section for some elaboration).

Unfortunately, success was short-lived. A couple of years later, new meta-analyses of the data on FMS™ revealed that the testing method is not in fact reliable for injury prevention (see the aside The Meta-analysis of FMS™ at the end of this section for additional details). Since predictive reliability was the only articulated support for the choice of fundamental patterns, there was not much left to warrant the reduction of “functional” to “functional according to FMS™”.

And just like that, the best proposal on the market lost its footing. Then again, the sales did not seem to suffer that much, which will bring us to the conclusion in about 3 minutes if you read the asides, or right away if you skip it keep it for later.

Expert consensus. The original FMS™ paper, “Pre-Participation Screening: The Use of Fundamental Movements as an Assessment of Function”, was published as a two-part series in the North American Journal of Sport Physical Therapy. The online version includes the following note from the editors: “The Functional Movement Screen™ is the registered trademark of with profits from the sale of these products going to Gray Cook and Lee Burton. The Editors of NAJSPT emphasize (and the authors concur) that the use of fundamental movements as an assessment of function is the important concept to be taken from Part I and Part II of this series and can be performed without the use of the trademarked equipment.” The note is absent from the nearly-identical 2014 paper “Functional movement screening: the use of fundamental movements as an assessment of function”, published as a two-part series in the International Journal of Sport Physical Therapy but it can be assumed that the editors agreed with the authors that FMS™ had been vindicated by empirical studies carried between 2006 and 2014.

FMS™ pragmatism. The 2014 FMS™ paper acknowledges that the term “functional” does not have a fixed meaning even among physical therapists, but dodges the bullet of a definition in two steps. The first step acknowledges difficulties but downplays them by appealing to an easy-to-agree upon belief: “Function is a common term in current physical therapist practice, and what is defined as functional varies greatly between patients and clients. […] It is the belief of the authors that many athletes and individuals are performing high-level activities despite being inefficient in their fundamental movements; thus, without knowing it, these individuals are attempting to add fitness to dysfunction. […] Thus, the authors suggest that screening an individual’s fundamental movements prior to beginning a rehabilitative or strength and conditioning program is important.” (Cook et al., 2014, p. 397). The second step claims empirical success for the FMS™ approach by endorsing explicitly a particular empirical criterion for evaluating its adequacy, namely the prevention of injuries: “[T]he ultimate test of any pre-employment or pre-placement screening technique is its effectiveness in identifying individuals at the highest risk of injury.” (Cook et al., 2014, p. 407). The empirical studies seemed at the time of the publication to validate FMS™ based on this criterion. Hence, the strategy was enough to dispense with the need to make explicit the assumptions that had led Cook and Burton to hypothesize that their reduction of ‘functionality’ to a handful of motor patterns. Of course, the meta-analysis of the data made the whole strategy collapse.

The Meta-analysis of FMS™. Cook et al. (2014) wholeheartedly endorsed early studies reporting a correlation between FMS™ scores and decreased risk of injury, which led to the statement that “the ultimate test of any preemployment or preplacement screening technique is its effectiveness in identifying individuals at the highest risk of injury” (p. 407) absent from the original article (Cook et al., 2006). This move left FMS™ vulnerable to a re-evaluation of the empirical data. Since the publication of (Cook et al., 2014), three meta-analyses have been published with partially conflicting results. Dorrel et al. (2015) found that earlier studies did not support the predictive value of FMS™ due to methodological limitations. Bonazza, Smuin, Onks, Silvis, and Dhawan (2017) concluded that FMS™ tests have high intra- and inter-rater reliability and FMS™ scores above the cut-off for ‘functionality’ were good predictors of lower risk of injury. Finally, Moran et al. (2017) reached the same conclusions as Dorrel et al. and addressed the discrepancies between (Dorrel et al., 2015) and (Bonazza et al., 2017). Thus, their conclusions are stronger and can be taken as the final say: “the level of evidence for the strength of association between FMS composite scores and subsequent injury is not sufficient to support the use of FMS composite score as an injury prediction tool.” (Moran et al., 2017, p. 1668).

Wrapping Up: Let’s Not Be Too Reductive

The whole FMS™ story is a cautionary tale about the simplistic reduction of complex notions and how little the fitness industry actually cares about legit scientific research even when it kickstarts it.

As noted by Cook et al. (2014), the use of the notion of ‘function’ arose in the context of physical therapy and its use remained anarchistic even after the proposed unification under the FMS™ umbrella. Even though Physical therapy is applied science, the lingering methodological issues with its conceptual basis in basic textbook physiology cast some doubt about the possibility of an alternative underlying unifying concept of ‘functionality’.

Still, I’m not one to throw the baby with the proverbial bathwater, however small the baby or murky the bathwater. So I’ll give some credit to physical therapists and assume that variations in uses of the term ‘functional’ own to context-sensitivity more than to bullshitting gone wild uncontrolled logomachy.

The most recent meta-analysis of the FMS™-related data suggests that the notion of functionality is in need of further conceptual analysis. While I agree with the suggestion, I am somewhat skeptical of the specifics which, I’m afraid, would only add to the existing logomachy [2]. I’d rather look in the rear-view mirror than speculate about potentialities

And in fact, the rear-view mirror might very well be all we need. To see this, let’s first remark that (1) the FMS™ reduction relies essentially on two insights about functionality:

  • Functionality is correlated with stability. Out of 7 items of the FMS™ test, 5 are stability tests (Deep Squat, Hurdle Step, Inline Lunge, Trunk Stability Push Up, and Rotary Stability).
  • Functionality is correlated with range of motion. Out of 7 items of the FMS™ test, 3 are ROM tests (Deep Squat, Shoulder Mobility, and Active Straight Leg Raise).

Second, let’s remark that (2) FMS™ is a somewhat typical instance of confusion between ‘whole-body stability’ and ‘joint stability’, as diagnosed by McGill and common among physiotherapists (and further clarified by the distinction between stability1 and stability2): 4 out of 5 of the stability tests are more concerned with balance than with mechanical stability (the Trunk Stability Push Up is legit).

Hence, a natural suggestion is to re-evaluate ‘functionality’ substituting mechanical stability to balance. This won’t vindicate the FMS™ reduction: the test is already known not to be reliable, and there’s no other reason to keep it. But it would validate the notion of ‘functional screen’ and could potentially organize the anarchistic use of the term in physical therapy contexts, and may even recover the missing dimension of athletic performance.

I’ll explore the topic further in the Stability entry of the Analytic Fitness™ Dictionary. For now, suffices to say the desiderata for a useful notion of functionality crowd out reductionism: there is no other methodological way than bottom-up. A conceptual foundation in mechanical stability would recover much of the balance and ROM components of functionality, but it wouldn’t warrant a reduction to a short list of items.

Still, as I implied in the introduction, most uses of ‘functional’ and ‘functionality’ are just standard issue fitness-industry bullshit (in the technical sense) aimed at selling various special equipment or exercise programs. For all its scientific veneer, FMS™ falls under that description too. I don’t expect conceptual clarity to put an end to that.


P.-S.: Functional fitness bullshit makes for endless entertainment and I’ll leave you with yet another random example of that, just for the fun of it.

View this post on Instagram

Demonstration what training with the Suples Ball is all about at the end of my recent workshop in Houston at @uproar.athletics . The Ball may look intimidating at first, but I can assure you in just a few minutes of step-by-step progressions, I can teach most anyone to learn the foundations of it and how to condition cardio, speed, and enhance coordination abilities. There are many ways to slow the movements down, chunk into component parts, and make adjustments to elements such as the strap lengths to simplify the movements, making them more accessible. The old saying of “what’s the best form of exercise?…The one you’re willing to do regularly” applies perfectly to Ball training. It is unique and a total blast to work with. 💪🔥 I have a very special event weekend with myself and inventor @ivanr.i coming up soon! Message me if you’d like more information, all are welcome 🤗 Also, use code “teamsalemi10” when you order any Ball or Bag on my site for 10% off. #suples #suplesball #speed #cardio #coordination #fitness #movement #conditioning #move #trainworkout #grappling #boxing #jiujitsu #muythai

A post shared by @ mike.salemi on


[1]^Honesty forces me to disclose that I have not set up a ‘risk-free’ account on, and therefore not read the full lesson either. Therefore, I cannot guarantee that the author(s) do not eventually nuance their initial statements with more accurate biomechanical information for paying customers. Then again, Wikipedia is more accurate, and free.

[2]^The proposal made by by Moran et al. (2017) is is summed up in the Research implications section of their paper, as follows: “Given the complexity of injury aetiology, investigators who seek to model the risk of future injury should apply multivariate analysis and predictor variables such as `movement competency’ (or similarly named constructs) need to be justified from a stronger theoretical basis. The theoretical construct addressed by the FMS, labelled as both `movement competency’ or `movement quality’, has undergone limited scholarly development, and its relationship with similar conceptual constructs, such as physical literacy, requires explication.” (Moran et al. 2017, p. 1667) The main issue with this proposal is that the notion of ‘physical literacy’ is another possible case of bullshitting gone wild uncontrolled logomachy (see this very interesting review). Still, the argument is somewhat more complex than what I’d dare to burden the readers of that blog with so I’ll leave things hanging there. But I have fewer qualms about burdening readers of scholarly journals an I have a scholarly paper submitted about these conceptual issues (and arguing for the alternative sketched in this conclusion) so I may return to the topic in the future.

Leave a Reply