Secondary Logo

Journal Logo


On the Functional Aspects of Variability in Postural Control

van Emmerik, Richard E.A.; van Wegen, Erwin E.H.

Author Information
Exercise and Sport Sciences Reviews: October 2002 - Volume 30 - Issue 4 - p 177-183
  • Free



We argue that the traditional perspective in biology in general and movement science in particular that tended to universally associate noise and variability with performance decrements and pathology is no longer tenable. Tools and methodologies that have emerged from the dynamical systems perspectives (e.g., nonlinear dynamics and chaos theory) are discussed in the context of coordination and postural control. We will focus our discussion primarily on the implication of these concepts for understanding postural instability due to aging and movement disorders, with special emphasis on Parkinson’s disease. After an overview of key issues regarding the role of variability in motor control, theoretical and empirical evidence will be presented demonstrating that variability can play a functional role in the detection and exploration of postural stability boundaries. Finally, it is argued that research on the role of movement variability in exploring these stability boundaries can add to our understanding of factors underlying loss of balance due to aging and disease.


In data obtained from biological systems, there will always be variability within and between measurements. The observed variability, however, can be fundamentally of two different forms, namely noise due to measurement error and variation due to inherent dynamics of the system (3). Measurement noise is typically regarded as independent of and additive to the dynamics of interest. Typical examples of the sources contributing to measurement noise are equipment noise, electrical interference, and movement artifacts. Traditional filtering techniques in biomechanics and motor control focus on separating this measurement noise from the original signal. Dynamical variability, on the contrary, arises from variation within the system to be studied and no clear separation can be obtained between the “original” signal and this kind of variation. This form of variability emerges from the underlying nonlinearities in the system and can play a critical role in pattern formation and perception in biology (2,4).

The majority of research in motor control and biomechanics has treated variability as principally emerging from measurement noise. However, more recent developments in nonlinear dynamics and chaos theory have emphasized the functional aspects of variability intrinsic to nonlinear systems (2). It is recognized that in coordination tasks involving multiple elements or degrees of freedom (e.g., limb segments, joints, muscles, motor units), a similar task outcome can be obtained in a variety of different configurations of these elements (5). This capability in biological organisms to produce a variety of solutions to a particular (movement) task offers flexibility to deal with unexpected or changing constraints and is a major source of variability in movement patterns. Variability is now considered an essential element in revealing what movement parameters are under direct CNS control (5).

In traditional perspectives on motor learning and development increased movement skill has been associated with decreased variability. Similarly, increased variability is typically regarded to identify the detrimental aspects associated with aging and disease. Current perspectives, however, have offered insights that aging and many diseases are associated with a loss of complexity and variability (7,11). The degree of complexity is typically associated with the number of system elements and their functional interactions. Research supporting this change in perspective on the role of variability has emerged from a large variety of different domains, such as loss of variability in heart rate intervals, EEG signals in Alzheimer’s disease, and loss of variability in gait due to aging and Huntington’s disease (7). In movement coordination, increased variability in the phase relations between segments is considered a necessary element before a transition to a new pattern can emerge (as in the transition between gait patterns) (4). In much of the clinical research on locomotion, increased variability in traditional gait parameters such as stride duration and stride length is often used as a marker of pathology and/or decreased stability. On the other hand, reduced variability in the coordination dynamics during locomotion has now been associated with an inability to transition from one movement pattern to another in Parkinson’s disease (12). A schematic overview of the difference in perspectives on variability is given in Figure 1.

Figure 1
Figure 1:
Changing perspectives on the role of variability in the control and coordination of movement. The figure is intended to schematically present the main message of the paper, namely the change in perspective on the role of variability in biological systems. The functionality of variability may also be dependent on task goals and system dynamics (11).

The schematic in Figure 1 reflects the general change in perspective regarding the role of variability in health and disease. It is not intended to convey the message that variability is always beneficial or under all conditions reflects healthy functioning. Indeed, a recent challenge of the general association of higher variability and complexity with healthy functioning systems has been made (11). This challenge proposes that in aging complexity might increase or decrease, dependent upon the type of movement dynamic (see Fig. 2). But it is also possible that differences emerge when assessing different type of variables, such as movement outcome (a pattern drawn in handwriting) versus performance variables (underlying joint, segmental, and muscular coordination) (5).

Figure 2
Figure 2:
A, B: Center of pressure patterns emerging from different postural tasks. A: Oscillatory movement in anterior-posterior direction. B: Sway due to postural lean. C: Schematic of postural patterns and their different dynamics in state space. In all examples the state space represents the anterior-posterior (vertical) and medio-lateral (horizontal) dimensions of the individual’s center of pressure. Note that, although the patterns in A and B have approximately the same amount of postural sway, the pattern in A more resembles the dynamic of the stable limit cycle in C. Traditional assessments focus on the amount of postural sway, leading to the potentially erroneous conclusion that both patterns signify equally stable states. AP, anterior-posterior; ML, mediolateral.

In the following sections, we will discuss the implications of these emerging concepts on variability and how variability is an essential component in assessing mechanisms underlying loss of balance with aging and disease.


In the literature on postural control, the degree of movement of an individual’s center of pressure on a force platform has been used to identify postural instability, with more movement indicating a higher degree of instability (14). The displacement of the center of pressure represents the point of application of the reactive forces under the feet. In the literature on postural control and aging, the typical finding is of a larger degree of excursion and greater variability of the center of pressure in older subjects as compared with younger subjects. A common conclusion from these patterns would be that the older subjects’ balance is less stable. However, this assumed relation between increased amount of postural sway and loss of stability does not take into account (a) the structure of these postural patterns, and (b) the fact that upright posture is almost never an isolated task but integrated or “nested” within other task goals (such as opening doors or picking up an object).

From nonlinear and complex systems principles, it is now evident that the structure of the variation in postural patterns needs to be taken into account in stability assessments (8). In Figure 2 are presented known dynamics that will have marked influence on the degree of stability and adaptability of the postural system. In the random pattern there is no preferred region of the center of pressure in state space (in this example created by the anterior-posterior and mediolateral dimensions of the center of pressure). In the fixed point there is attraction toward a very specific region in state space. The limit-cycle type dynamic results in postural states being “attracted” toward a cyclic pattern in state space. This type of dynamic shows resistance to perturbations but is less flexible and adaptable when a change in postural state is needed. The chaotic pattern shows both stable attraction to a region in state space and variability. These dual features are now associated with the higher pattern complexity that is reflective of healthy systems. As an example, increased heart rate variability is considered an important indicator of healthy heart function, reflecting a degree of complexity in organization in which disruptions can be compensated for and changes in rhythm imposed more easily (2,7).

In addition, the control and maintenance of upright posture is, in most natural activities, nested within other task goals of the organism. Upright stance is a fundamental prerequisite in, for example, opening doors, picking up objects, and performing “measurements” or observations of different aspects of the environment. For example, increased sway could arise from exploratory activity that may or may not be destabilizing (9,14). For exploratory behavior not to interfere with the required movement pattern (the so-called “dual control problem”), variability due to exploratory dynamics occurs at smaller time scales (lower amplitude and higher frequency) than the actual movements (9). However, in most postural control research these higher-frequency components have traditionally been filtered out and eliminated as unwanted sources of noise.

An example of these different time scales and their role in postural control is illustrated in Figure 3. The center of pressure (COP) patterns in Figure 3 exhibit higher-frequency characteristics than the center of mass (COM). The higher frequencies in the COP are a necessary aspect to control or correct the COM position (15). These higher frequency aspects in the COP, however, do not only reflect compensatory activity, but could also result from exploratory activity that may contribute to postural stability.

Figure 3
Figure 3:
Top: ML center of pressure (COP) and center of mass (COM;thick line) trajectories during quiet stance in a healthy subject. Bottom left: power spectral density (PSD) plot of the COP trajectory. Bottom right: PSD of the COM trajectory. Power in arbitrary units (au).

Different approaches have emerged to investigate properties of these different time-scale windows and their effects on adaptive movement control. Collins and colleagues have applied “stabilogram-diffusion” analysis to assess postural center of pressure patterns (1). This analysis technique can reveal the time scales over which particular postural strategies show “persistence” (i.e., remain invariant) or show “antipersistence” (i.e., show a change). Changes in these time scales have been observed in older adults and patients with Parkinson’s disease compared with younger healthy individuals, suggesting a more lenient postural strategy with aging and disease (1).

The different time scales underlying postural control have also been divided into the temporal aspects that reflect control and modification of actuators (relatively long) and those that reflect observability and information gain (relatively short) (9). From this perspective, short duration, high-frequency temporal aspects are not only reflective of corrective actions as in (1) but can reveal exploratory behavior that can be useful to change ongoing (slower) control of actuators.

A critical element in the proposal that variability plays a role in facilitating adaptive postural control is that it allows for the exploration of the limits of current states and boundaries between different postural configurations. In the next section, we will discuss current research on postural control with a special emphasis on stability boundaries, variability, and exploration.


In postural control it is recognized that the goal is to maintain the body within the boundaries of the base of support. For quasistatic stance on a fixed surface, this requires the COM to remain within the boundaries of the feet. These boundaries can mark the limits of a particular postural configuration (e.g., upright stance) for which a change in postural state (e.g., a step, a fall, a reach to hold onto a supporting object) is not required and perturbations can be reversed. Most quantitative measures of upright stance, however, do not include boundaries or thresholds in the assessment of postural stability (14).

A central requirement for the robust and adaptive control of human movement is the ability to perceive the range of effectiveness of a variety of action systems. Action systems can be operationalized within the same musculoskeletal segments, as can be seen for example in ankle and hip compensatory strategies. However, these action systems could also involve overlapping or nested segments, such as in the interaction of postural and manual control. Detection of effectiveness of stability boundaries implies observability of one’s own postural state relative to these boundaries. Riccio and colleagues have developed an approach that emphasizes the functional role of variability in the detection of these stability boundaries (9,14). Including stability boundaries in assessments of postural stability introduces a framework to interpret the role of variability in the maintenance of posture. Our argument is that proposed links between variability of postural sway and instability are incomplete without an assessment of these boundaries or thresholds for postural strategies.

Experimental support has recently been provided for the suggestion that the variability and stability of stance should be considered in reference to limits of the base of support at the feet (10,13,14). In these experiments, the center of pressure was used to investigate the role of stability boundaries, as it reflects the interaction of the organism with the environment through the transfer of forces to the substantial support surface. The data in Figure 4 represent center of pressure trajectories obtained from forward, neutral, and backward static lean conditions. From Figure 4, it appears that the older subject shows different adaptations to the lean conditions than the younger subject. The older individual shows an increase in the center of pressure variability under maximal forward and backward leans whereas the younger subject decreases center of pressure variability toward the stability boundary under these lean conditions. In the older subject, the higher variability in the lean conditions could have more severe consequences, such as the center of mass crossing the boundary of support, leading to either a correction or a fall. These data support the hypothesis put forward by Vaillancourt and Newell (11) that the functionality of variability is task dependent.

Figure 4
Figure 4:
Example graphs of center of pressure trajectories during no lean (middle traces) and maximal forward (top traces) and backward lean (bottom traces). Circular region represents hypothetical area in which control can be lenient with minimal corrections based on feedback. Note that the older subject increases variability toward the stability boundaries, whereas the younger subject decreases variability. Both subjects had similar foot length. Dashed lines represent limits of the base of support in AP direction based on subject’s foot length.

This application of boundary-relevant measures to postural control is in our view extremely important for global questions related to loss of stability and falling. Indeed, adaptability in complex systems might emerge from the ability to control behavior with respect to these (stability) boundaries. The stability boundaries concept will not only provide a framework for assessing the role of variability in postural control per se, but might also help to elucidate the role of different forms of variability and their impact on postural control. The controllability of this variability might be a critical factor in motor control and perception.

So far, we have only discussed aspects of the spatial proximity to stability boundaries. Spatial proximity measures, however, do not take into account the rate and direction with which the center of pressure is moving toward the stability boundary. The ability to perceive the spatio-temporal proximity to a stability boundary has been proposed as an important determinant for postural stability (9). In visual perception research, distance or depth information can be obtained through information in the optic flow on the retina via time-to-contact. Time-to-contact is the time that remains before an approaching object “hits” the observer or before a surface is contacted and reflects position and velocity information. An object that is threatening to hit you (such as an approaching ball) would be characterized by a symmetrical outflow of the optic pattern from the center of the ball, signaling the closing in of the ball to the target. Lee and colleagues have shown that animals and humans use time-to-contact information based on this optic flow to guide the timing of purposeful actions, such as the closing of the wings in diving gannets and the onset of braking when driving an automobile (6). These researchers have also assessed the role of optic flow in the form of time-to-contact in the “moving room” paradigm. In this experimental setup, the “walls” can be moved while the floor is stationary. The moving walls provide large optic flow patterns (such as in Imax theaters) that result in vision overriding other sources of sensory information (such as vestibular, muscle spindle, cutaneous, etc.), inducing postural sway or even loss of balance in young children and individuals with neurological and vestibular deficits.

More recently, the time-to-contact approach has been extended to sensory and perceptual aspects of posture other than the visual system (10,13,14). As there is no actual “contact” in the postural case, we prefer to label this temporal margin the “time-to-boundary” (14). The time-to-boundary information in upright stance can be obtained from distance and velocity of the center of pressure with respect to the base of support provided by both feet. Figure 5 depicts an example of a time series of the time-to-boundary to the anterior stability boundary.

Figure 5
Figure 5:
Left panel: schematic of derivation of time-to-boundary. Time-to-boundary is obtained by the ratio of the distance, d, to the boundary and the instantaneous velocity, v, for every data point. Right panel: a time series of assembled temporal margins to the AP stability boundary. In the analysis an average of respective minimum values is obtained as these minima reflect changes in postural strategies (see (13) for a more detailed description).

The spatio-temporal margin of the center of gravity or center of pressure with respect to these stability boundaries might be a key variable for upright postural control. Techniques outlined by Schöner and colleagues using the so-called “uncontrolled manifold” analysis could reveal the importance of this time-to-boundary variable for postural control (5). Research along these lines should elucidate whether these boundaries are in fact perceivable, and if variation in task dimensions (e.g., reaching distance or platform perturbations) leaves aspects of boundary-relevant dynamics (such as minimum time-to-boundary threshold) relatively invariant. Support for the invariance of boundary-relevant measures comes from observations that time-to-boundary measures do not appear to be less than 200–300 ms (9,13,14).


Postural instability is one of the major symptoms in Parkinson’s disease. A relatively large body of literature has emerged on assessing postural stability changes in Parkinson’s disease using the position and variability of the center of pressure. A review of these studies yields equivocal results regarding these positional and variability changes in the center of pressure in these patients (13). Although differences in study methodology and patient population could partially contribute to these different findings, it is also possible that these variables alone are not sufficient to assess postural stability problems in patients with Parkinson’s disease. Based on the current body of literature, we argue that postural control measures that reflect the relationship between the organism and task-relevant stability boundaries are a necessary addition to postural stability research in aging and disease.

Several studies have examined the time-to-boundary of the center of pressure to the stability boundary (10,13,14). A reduction in time-to-stability boundaries has been observed in older adults as compared with younger adults, indicating that these measures could be useful for the study of populations that suffer from postural instability (10,13). In one of our studies comparing older and younger healthy individuals (14), subjects in the older group did not show the increase in variability in center of pressure as usually observed in older populations (Fig. 6 B). These older subjects did, however, show reductions in mediolateral temporal margins (see Fig. 6 C). Patients showed further reductions in lateral stability margins compared with healthy older adults (see Fig. 6 C). In addition, patients with Parkinson’s disease had significantly lower time-to-boundary variability, even though they had larger mediolateral center of pressure variability (Fig. 6 d). The reduced time-to-boundary variability could be indicative of a postural control system with reduced complexity and loss of adaptability. These conclusions are in line with suggestions of increased variability being a hallmark of healthy, adaptable physiological systems (7). However, they also indicate that both a decrease and an increase in variability and complexity may describe changes in a behavioral system due to aging or disease, depending on the specific dynamics of the system that is investigated (11).

Figure 6
Figure 6:
Changes in mediolateral COP due to aging and Parkinson’s disease. A: average center of pressure position; B: variability of center of pressure; C: average minimum time-to-boundary (TtB); D: variability of minimum TtB. % = % of foot width.Error bars reflect both between-subject variation as well as within-subject variation due to averaging a variety of lean conditions. * Indicates statistically significant difference at P = 0.05.

These results suggest that variability measures of the center of pressure alone might not be sufficient to reveal changes in the postural system that may lead to a loss of balance. This suggestion emerges from the observation that the older individuals clearly show changes in boundary-relevant measures that are not revealed by traditional center of pressure variability variables. In addition, the postural instability observed in many patients with Parkinson’s disease could be related to reduced time to the lateral stability boundaries and not merely be the result of an increase in mediolateral sway variability.


The main argument in this review is consistent with a growing body of literature that stresses the beneficial and functional aspects of variability in biological systems. These developments are in stark contrast to previous perspectives in which decreased variability was universally associated with increased competence, skill, and health. The path to frailty or disease (7), therefore, cannot be identified solely by increased variability in fundamental variables reflecting biological function. Future research on behavioral and physiological systems should focus on the role of variability and system complexity, as not all essential system variables might show the same patterns of complexity and variability change.

Although from the research on posture and balance presented it is suggested that variability is important in the detection of stability boundaries, more direct evidence of the functionality of variability and its role in exploratory behavior is needed. However, the research clearly shows the importance of stability boundaries in the assessment of postural instability due to aging and disease. For the treatment of balance-impaired populations such as older adults and patients with Parkinson’s disease, this notion can have a large impact: Instead of focusing on the reduction of postural variability, treatment could be aimed at increasing adaptability by stimulation of exploratory behavior with respect to stability limits.


We would like to thank Gary Riccio for his contributions to the work presented here. We are also grateful for valuable input from Karl Newell, Jane Kent-Braun, and an anonymous reviewer. The research presented in this paper was supported in part by NIAAA grant 1R43aa12663–01 and Whitaker Foundation grant RG-99–0097.


1. Collins, J.J., De Luca, C.J. Burrows, A. and Lipsitz. L.A. Age-related changes in open-loop and closed-loop postural control mechanisms. Exp. Brain Res. 104: 480–492, 1995.
2. Glass, L. Synchronization and rhythmic processes in physiology. Nature. 410: 277–284, 2001.
3. Kantz, H., and Schreiber. T. Nonlinear Time Series Analysis. Cambridge, UK: Cambridge University Press, 1997.
4. Kelso, J.A.S. Dynamic Patterns: The Self-Organization of Brain and Behavior. Cambridge, Massachusetts: MIT Press, 1995.
5. Latash, M.L., Scholz, J.P. and Schöner. G. Motor control strategies revealed in the structure of motor variability. Exerc. Sport Sci. Rev. 30: 26–31, 2002.
6. Lishman, J.R., and Lee. D.N. The autonomy of visual kinaesthesis. Perception. 2: 287–294, 1973.
7. Lipsitz, L.A. Dynamics of stability: the physiologic basis of functional health and frailty. J. Gerontol. A Biol. Sci. Med. Sci. 57: B115–B125, 2002.
8. Newell, K.M., van Emmerik, R.E.A. Lee, D. and Sprague. R.L. On postural stability and variability. Gait Posture. 4: 225–230, 1994.
9. Riccio, G.E. Information in movement variability. About the qualitative dynamics of posture and orientation. In: Variability and Motor Control, edited by Newell K.M. and Corcos. D.M. Champaign, IL: Human Kinetics, 1993, pp. 317–357.
10. Slobounov, S.M., Moss, S.A. Slobounova, E.S. and Newell. K.M. Aging and time to instability in posture. J. Gerontol. 53: B71–B78, 1998.
11. Vaillancourt, D.E., Newell. K.M. Changing complexity in human behavior and physiology through aging and disease. Neurobiol. Aging. 23: 1–11, 2002.
12. van Emmerik, R.E.A, Wagenaar, R.C. Winogrodzka, A. and Wolters. E.C. Identification of axial rigidity during locomotion in Parkinson disease. Arch. Phys. Med. Rehabil. 80: 186–191, 1999.
13. van Wegen, E.E.H., van Emmerik, R.E.A. Wagenaar, R.C. and Ellis. T. Stability boundaries and lateral postural control in Parkinson’s disease. Motor Control. 3: 254–269, 2001.
14. van Wegen, E.E.H., van Emmerik, R.E.A. Riccio, G.E. G.E. Postural orientation: age-related changes in variability and time-to-boundary. Hum. Mov. Sci. 21: 61–84, 2002.
15. Winter, D.A. A.B.C. (Anatomy, Biomechanics and Control) of Balance During Standing and Walking. Waterloo, Ontario: Graphic Services, 1995.

variability; postural stability; aging; Parkinson’s disease; complexity

©2002 The American College of Sports Medicine