Dynamic systems theory in human movement: Why movement variability is a good thing! Part 2

Movement variability and the concept of motor abundance

Often it is thought that there exists an “optimal” movement through the joint of interest.

Whilst I do see more in the human movement space challenging the prevailing paradigm of optimal movement, there is still an obsession with the idea that there is one way to move or the correct posture to maintain when you are doing X or Y. Often these are fused with ideas of damage through exposure to certain movement patterns.

Ultimately this view closes your options of value-based activities and (can) lead to lots of distress and frustration!

With the expansion of motor control theory and the use of alternative methods to measure it, the concept of an optimal movement pattern seems to wane, instead a spectrum of movement variability that represents behavioural richness of an organism seems to appear.

Elite athletes’ “behaviours” may appear stable i.e. the ability of Michael Jordan to shoot a 3 pointer with hair pin accuracy, but they are able to achieve this in an infinite number of ways.

I want to take some exerts directly from a paper by Nicholas Stergiou and Leslie Decker published in 2011.

Traditional linear measures, such as the standard deviation are measures of centrality and thus provide a description of the amount or magnitude of the variability around a central point from a human movement perspective, this approach in evaluating variability has led to believe that anything outside of the mean score is error.

This approach assumes that repeated repetitions of movements are random in nature and independent, i.e. not influenced by past or present… we now know this is simply not true. As we discussed in part 1, human movement is embodied and embedded.

Nicholas & Leslie present several seminal studies showing that during gait, when analysed with linear measures, there appear to be random variations from stride to stride. However, when you then run the same analysis of the gait with non-linear modelling, it appears that the healthy adult locomotor system possesses ‘‘motor memory’’, such that the fluctuations from one stride to the next display a subtle, ‘‘hidden’’ temporal structure.

The pair further allude to animal studies that have shown that when the intrinsic variability of gait was overridden (e.g., when a fixed pattern is imposed with no variability allowed), learning of a task was suboptimal relative to the condition when the training is assist-as-needed.

In humans, something similar happens. Lesions to the basal ganglia and other key areas associated with motor planning and execution appear to directly correlate with aberrant variability in gait performance when compared to healthy age matched controls… leading the author pair to suggest that human movement variability exists on a spectrum, too little or too much will impair performance or impact negatively on the organism.

Here in I (I really mean Nicholas and Lesley) present the goldilocks concept of human movement!

The inverted U-shape relationship. The presence of chaotic temporal variations in the steady state output of a healthy biological system. Shift the curve to the right and it becomes too rigid and predictable and motor performance suffers, shift it to the left and it becomes too random and too greater noise in the system and performance suffers. This has been echoed in similar papers, instead coined the bliss of motor abundance where the CNS increases the amount of relative variability under situations including novel tasks, fatiguing of the bodily systems and variable target locations (Latash, 2012).

The current state of the research around movement variability relative to task performance would suggest variability is inherently good but just as Goldilocks found out, too much or too little is detrimental to our performance and organism stability. We just don’t know where the upper and lower boundaries are at this point in time.

Acceptor states, transient states and the basin of opportunities for movement

Hopefully by now we agree that the body (and its internal control structures), our affective appraisal and bodily resonance, and the environment determine how many behavioural options the agent (you or I) has for completing a task. Now I want to introduce some new terms, taken again from dynamic systems theory, that we are going to use to further explore the boundaries of human movement and how we transition through movement states.

Human movement can be viewed from a dynamically shifting landscape; the state space represents the possible states for movement in any given context (Heino et al., 2023; Hiver et al, 2015) but due to the non-linear features of behaviour, the agent (us) does not visit all states with equal probability but rather is drawn to certain ones more than others.

An attractor state is a tendency for an agent to settle toward a particular critical pattern and is less likely to leave (Heino et al., 2023; Hiver et al, 2015). For example, during the approach to a stationary opponent, a rugby player planning a diagonal cut will drop his centre of mass and plant the outside foot opposite to the direction of intended travel. But this, by nature of movement being embedded and embodied, is going to be different each time and the ability to get into the cutting position is in part determined by the strength of the attractor state on the agent.

As described by Heino et al., (2023), valleys in a national park aptly represent all attractor states whilst the state space is the boundaries of the park itself.

Using the same visual concept, the pull of an attractor is the depth of the valley and thus a deeper attractor state means the agent is less likely to hop to another state. Different aspects of the person (psychological, physiological), environment and task may pull a valley deeper, making that attractor state more likely to attract. This would ultimately increase the ease or difficulty of transitioning to the desired state for that person. Thinking of the person following an ACL repair, the change in internal control structures of the knee due to loss of sensory fibres in the ligament itself may increase the depth of an attractor state that manifests as a position of greater knee extension and reduced foot dorsiflexion on impact of the outer foot during the diagonal cut. We know that this combination placed the ACL under increased strain.  

Quite the contrary to this, it is purported in the behavioural change research we can deepen attractor states we wish to move toward and make more shallow those we are trying to avoid. Two things I wish to emphasise here, firstly, just because we make the favourable one deeper and the less favourable one smaller, does not mean we are going to bounce to the deeper and hop out of the shallow one with ease. We are dealing with non-linear systems so we cannot predict what happens. We can only try and increase the probability of a desired attractor state being reached.

Secondly, it has been suggested that in embodied action (of the eye) we may never settle on one attractor state due to the ever-present feedback from proprioceptive-kinaesthetic coupling (Negrello & Pasemann, 2008). Instead, we jump from one attractor state to the next and is said we are in a constant transient state.

I am yet to find research relating to human movement that tests the concept of attractor states, much of it is theoretically based at this current point in time.

The limitations of movement screens for injury foresight but why we should still see how people move!

Recent meta-analysis’ have reported that neither the cut off score of 14 points (Trinidad-Fernandez et al., 2019) or FMS scores in general are consistently predictive of future injury in athletes (Moran et al., 2017; Asgari et al., 2021; Peterson et al., 2022; Bayrak, 2025; Dorrel et al., 2015). Although other reviews have tentatively identified relationships in favour of FMS for injury prediction in military personnel, they cite substantial inconsistency in findings of associations (Rhon et al., 2022, pre-peer review manuscript).

The above is not surprising given we now know the dynamical nature of embodied-embedded action (movement) and how the agent (us) does not have a pre-determined program for organising around the task and environment, instead we iteratively self-organise relative to the task and affordances. At least for now, we cannot seem to be able to accurately predict those who are going to get injured prospectively in the season by looking at how they move - but I still think we should use movements with our patients to see how they move.

Why?

Well to move and engage with our environment through our affordances for action, we must be able to produce force, without that, according to Newtons first and third law we will neither move anywhere and remain at rest (his first law) or be able to change direction (related to his third law via utilisation of vertical ground reaction force) in response to stimuli. So perhaps we view force (or our ability to produce force) as being fundamental to human movement – we go through our bodies, which produce force, to engage with our environment to achieve our valued actions.

By seeing how someone moves and by extension, produce force, we can see if they either close down or open up their opportunities (affordances) for action in their valued activities or sports. For example, most game sports athletes need to be able to change direction. For this, they need to change the output characteristics of their force component (direction, velocity, magnitude etc) to allow them to decelerate accelerate in different directions. During a movement screen we might have an athlete performing single leg hops, lunges and squats and notice they struggle to maintain an upright trunk during the lunge and have an extended ground contact time during repeated single leg hops with high amounts of trunk-hip flexion during contact. Without even looking at the athlete run, I would already have increased suspicions they may not have the requisite force qualities to be able to decelerate from a near max velocity run to perform a lateral cut.

We watch them run and notice as they decelerate, they get overspill of their trunk in the forward direction (they collapse their trunk forward over their centre of mass). This athlete suddenly has fewer affordances for action in their environment when they now need to make a diagonal cut around an opponent. The trunk overspill that has helped them stop is not wrong per se, it has helped them achieve their task of decelerating but now they are likely going to be more limited to perform a cut that is successful in getting round the opponent because their bodily position is in such a way that the force they produce will not propel them in their desired direction. Borrowing again from Leslie and Decker, you argue here that the movement output has become so chaotic that performance of subsequent is more likely to suffer.  

Taking a lesser sporting example, Dorris loves her gardening and takes care of her grandchildren Recently she has developed a painful hip and no longer trusts it, saying it feels a little wobbly! During a forward lunge we notice she collapses forward through her trunk to complete a cone sorting task on the floor, and she has limited knee flexion and extends her rear leg to use it as a strut and struggles to complete the task. Again, this is not a wrong movement. Her body is organising around the task with her affordances with the environment. She does it again but with a plinth by her side for support. She is now able to complete the task on the floor, and the movement resembles more of a lunge with an upright trunk and bend in the rear leg when she has a plinth close by for support. The increased environmental affordances lead to improved task performance as she can constrain her movement and reduce the degrees of freedom. In the first task, she does not have the requisite force characteristic to allow her to achieve the positions for effective utilisation of the task, again, she likely falls on the ‘chaotic’ end of the inverted U curve by Lesley and Decker.

Hopefully we can see that it is still imperative we see how people move and that this is most useful when it is focused on movements central to their valued tasks. After all, people seek health care advice when they are no longer able to achieve their valued activities via assimilative coping mechanisms!

We know that people get injured and enough robust research shows that variability of movement is good (to a point) but beyond the upper and lower boundaries, performance suffers, and injury is more likely to occur. We just do not know where the boundaries lie and if I were a betting man, we likely never will give the dynamical nature of human movement.

Change the input, change the context, change the task and build a resilient organism

Bringing the topic back to the concept of stability and proprioception, even though we now know it is entwined within general movement principles, the concept of attractor states and non-linear dynamics of movement emphasise we cannot be rigid in our approach to training!

We cannot change the state space, but we can try and increase the depth of our desirable attractor states.

If we seek to develop a rich behavioural state of movement, defined as a wide basin of state opportunities, then we must build this into our programs as opposed to attempting to chase some “optimal” pattern that does not appear to exist.

If we give our body lots of exposure through the three planes of movement, at a varying rate of force and movement velocities, with variable sensory inputs and cognitive loading, we are encouraging the greatest change in behavioural states across a spectrum.

One would hope this “richness” would translate to improvement in symptoms and/or performance during their tasks… obviously we need to balance “variability” (deemed as richness) with task specificity of what they will be exposed too but again understanding that the movement at the point of performance will organise around the perceived affordances.

Perhaps now is the time to revisit the challenge of the proprioception.

Knowing that a rich behavioural state is characterised by the capacity of the person to exhibit variability of movement in successful pursuit of their goal – remember variability is useless and perhaps detrimental if we can’t wield it properly – then the function of the proprioceptive aspect (let’s call it our internal control structures) would be to maintain stability in output characteristics of that movement.

Be that force, velocity, angle, visual-spatial characteristics, and respond according in a feed forward or feedback manner.

Thus, would it be more appropriate that the concept of movement stability has more meaning versus proprioception? Under these conditions it pertains more to the output being held in such a way to achieve the goal. Unlike proprioception, it is not embodied within movement itself but rather it is the desired output of the human movement system working successfully to achieve the task.

We can see in the figure below that I have made.

Solid black lines represent upper and lower boundaries of movement akin to inverted U theory by Lesley and Decker. Above the line like point A and below the lines like point B and performance suffers. Following appropriate training are increases our boundaries via deepening the attractor state and have the capacity to hold movement output characteristics to maximise movement output within the appropriate variability.

Exploring the ontology of movement from a PK-coupling and dynamic systems theory perspective.

Movement as we now know is an embodied-embedded behaviour that is the output of the dynamical relationship between the body, environment and task.

Going back to our valley of opportunities metaphor, structuring rehabilitation is far easier if we have a framework which we can work from. I think a combination of the PK-coupling and dynamic systems theory affords such an approach. The following is my opinion and thoughts on what is fundamental about human movement.

It is said in the embodied literature that we go through our bodies to interact with the environment and affordances for action (Tabor et al., 2017; Vaz et al., 2023). The very concept of affordances for action (in a state space) implies a viewing of those affordances from the perspective of a (conscious) body with intent for action. Therefore, the environment, synonymous here with state space, only exists if we perceive and feel it. The valley of affordances exists only with our perception through our bodies; if we do not exist in that valley then does the valley cease to exist? As there is nothing for perception-action coupling to occur with as no body exists in the environment for this to occur.

In my opinion, the body (perhaps even deeper than this, the consciousness) is fundamental. We go through our bodies, and by extension our environment, to achieve our valued actions. See my concept drawing below that illustrates my thoughts on this.

Solid line represents my idea that we go through our bodies to interact with the environment to achieve our task. The dashed lines illustrate the circular relationship between impression-expression that has been previously talked about above.

So, the three dimensions of the PK-coupling could be seen from a stepwise perspective. Reintegration of the body into the environment following injury can only occur once we have established a self-self-contingency and reembodied all bodily parts into the self to achieve full spatial-temporal awareness of the “us”. Perhaps this is why in some cases, surgery is required, the extent of damage to the internal control structures that allow for embodied action must first be put right to achieve full self-integration.

Just a thought!

Only then can we successfully begin to interact with the environment and begin the coupling of the self-environment. Now we integrate all bodily senses into our rehabilitation program, including bodily resonance through affective dimensions of impression-expression coupling. Focusing on behavioural richness as discussed above through trying to deepen as many desirable attractor states as possible is the goal in this dimension. Here we look to familiarise ourselves with the nuances of the desired task and fundamentals of what we think could be the action needed to achieve success of the task. I have no suggestions as to the timeframe spent in this contingency before moving to the next.

It would make sense, after gaining sufficient exposure to the (infinite) possibilities of attractor states to then begin to integrate the self-environment-other dimension. Here we look to familiarise ourselves with the nuances of the other person(s). How will they interact, what are the possible affective aspects that could drive certain resonance outputs? Ultimately, we don’t know… but we can try and strip it back to the absolute fundamentals (if that was a thing).

With this in mind, we need to consider how we integrate these affective dimensions into our rehabilitation. Lets get into that in part 3!