- Email: [email protected]
Examination of trial-independent characteristics of body kinematics in response to similar postural perturbations
Gait and Posture 7 (1998) 110 – 116
Examination of trial-independent characteristics of body kinematics in response to similar postural perturbations Ge Wu Department of Physical Therapy, Uni6ersity of Vermont, Burlington, VT 05405, USA Received 1 April 1996; received in revised form 25 June 1997; accepted 23 July 1997
Abstract This study investigated the variability of certain kinematic features describing body segmental movement in response to a series of consecutive, translational movements of the supporting base along the anterior-posterior direction of the subject. These kinematic features included the amplitude and timings of the angular displacement, velocity and acceleration of the head, trunk, thigh and shank. A group of 21 subjects was tested, with their ages ranging from 22 to 77 years. The results showed that the amplitudes changed significantly from the first trial to the later trials in most of the variables observed. However, there existed certain features, other than the amplitude, that were invariant over consecutive trials. Specifically, the time at which the initial peak amplitude occurred, the onset time was found to be similar between trials. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Adaptation; Trial-independence; Time-varying; Dynamic posture; Kinematics
1. Introduction Human postural control during upright stance is a complicated problem. Although there has been tremendous amount of effort over the past years in exploring the fundamental mechanism of human postural control, there has always been a critical question that relates to the experimental design in human postural studies. That is: are the strategies used to control upright balance reproducible after repeated exposure to similar postural perturbations [1,2]? In the past, understanding this question has been mainly limited to the changes in response size of muscular activities, based on the observations of adaptive attenuation/facilitation during stance or balance regulations in humans [3–9]. For example, in an early experiment on human dynamic postural control, functional stretch reflex in the leg muscle was found to either decrease its magnitude with consecutive movements of the supporting surface when it destabilized the posture, or increase its magnitude when it stabilized the posture [3,10,11]. Keshner et al. [3] also reported adaptive at0966-6362/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S 0 9 6 6 - 6 3 6 2 ( 9 7 ) 0 0 0 3 3 - 7
tenuation in the neck muscles. Many experiments have shown that postural adaptation usually occurs within three or four consecutive exposures to the same dynamic stimulation [3,10,11]. Besides adaptive attenuation/facilitation observed during human stance or balance regulations, are there certain biomechanical variables observable in dynamic human postural movement that are trial-independent? The term trial-independence in this study is defined as, no change over repeated exposures to the same dynamic postural stimulation. To date, few studies have addressed this trial-independent issue directly in human upright stance [12], although trial-independence has been reported in studies relating to other activities, such as active joint dynamics in a non weight bearing condition [13] and the control patterns (i.e. the timing) during locomotion [14]. Recently, McClenaghan et al. [15] explored the spectral signature of the ground reaction forces during a static stance task and found high reliability between multiple trials of each individual subject. All these findings suggest that there may exist certain parameters, especially the timing of the control
G. Wu / Gait and Posture 7 (1998) 110–116
strategies, that are trial-independent in maintaining a balanced upright posture. The mastery of trial-independent characteristics of human postural response after repeated exposure to similar postural perturbations is inevitably important because they are experimentally reproducible. In the past, in order to avoid the influence of adaptation on the experimental outcome, a frequently used approach is to precondition the subjects, that is, before the real experiment begins, the subjects need to practice for a period of time in the same environment as the one in which the data will be collected [16]. However, one of the potential problems with preconditioning is fatigue, which is non-stationary in the motor control system [12]. The other problem with preconditioning is missing the earliest, unpracticed trial which is most likely to be most relevant to the functional ability to balance in daily life, where we rarely get a chance to practice our response to a specific external perturbation. The purpose of this study was to investigate whether or not there are certain features of human postural control strategies that are trial-independent. The postural control strategy in this study was quantified by the movement of each individual body segment (i.e. head, trunk, thigh and shank) in response to a sequence of similar dynamic postural balance tasks. In the past, postural control has been represented by spontaneous body sway [17], electromyography of selected muscles [18], and gait patterns [19]. However, since whole body balance is obtained by proper coordination of single body segments, the movement of individual segments in a balance task will provide in-depth information about the functioning of the postural control system. In this study, the trial-dependence of the following characteristics of individual body segment movement was examined: the peak amplitude, the peak time and the onset time. Based on the results of other studies, using either spontaneous body sway or electromyography during posture or gait, it was hypothesized that the peak magnitude of body movements would change with increasing number of trials experienced, while both timing characteristics (i.e. onset time and peak time) would not show signs of adaptation. Visual, movement direction and the surface type of the supporting surface were altered to create a randomized sequence of perturbations.
111
control, vestibular function and muscle strength. Based on these tests, those selected were subjects free of skeletal and muscular problems that affect normal, daily life activities, significant loss of binocular vision, abnormal vestibular function, and medications likely to have an impact on the peripheral or central nervous system or muscular function. Before testing, each subject was asked to read and sign a consent form agreeing to the condition and expectations of the experiment.
2.2. Apparatus The human body was modeled as four rigid segments in the sagittal plane: the head, trunk and arms, thigh and shank (Fig. 1). The kinematic variables being studied included angular displacement, angular velocity and linear accelerations in the horizontal and vertical directions of each of the four body segments. The velocity and acceleration were directly measured using integrated kinematic sensors (IKS [20]), one on each body segment. Each IKS included one angular rate sensor (Watson, Eau Claire, WI) and two single axial linear accelerometers (Kistler, Amhurst, NY) arranged orthogonally. Another accelerometer was attached to a movable platform to measure its acceleration in the moving direction. The outputs from the IKSs and the platform accelerometer were collected at 120 Hz for 3 s.
2. Methodology
2.1. Subjects A total of 21 subjects (12 males and 9 females) participated in this experiment. They ranged from 22 to 77 years of age. All subjects were examined by a registered nurse to evaluate their fine and gross motor
Fig. 1. Schematic diagram illustrating the experimental configuration. Symbols a – n, a – t and v are the normal and tangential accelerations of a body segment, and a – p is the linear acceleration of the platform. Body segments are labeled as h for head, ta for trunk and arms, th for thigh, and s for shank.
112
G. Wu / Gait and Posture 7 (1998) 110–116
Postural perturbation was applied to the subject through a platform movement in the anterior-posterior direction of the subject. The platform movement was driven by an electromagnetic actuator with a control unit to alter the direction of the movement. The maximum acceleration was 8 m/s2, speed 40 cm/s and total displacement 8 cm.
2.3. Procedures During the experiment, each subject was tested for a total of 12 conditions, four times each. They included two visual conditions (vision/no vision) ×2 movement directions (platform moving forward/backward) × 3 surfaces (standing on hard/soft/reduced surfaces). The vision was either allowed or blocked by a pair of shuttered goggles which remained open most of the time to allow normal vision. For the blocked vision condition, the shutters were automatically closed at about the same time as the onset of the platform movement. The soft surface was a piece of foam (Foamaex Polyurethane foam, 5 cm thick, compliance of 0.086 kPa − 1) that covered the entire surface of the platform. The reduced surface was made of two pieces of wooden blocks (3.6 cm thick), that were cut into the shape of, and were attached to the rear half of the shoe bottom. In order to minimize subject’s anticipation of the platform movement, all these conditions were provided randomly by block design. That is, each block included a randomized sequence of all the conditions. Before each trial, the subject was instructed to stand quietly on the movable platform, with feet comfortably separated in the lateral direction, arms folded across the chest and eyes looking at a stationary object 25 m in front of them. In order to distract the subject’s attention to the upcoming platform movement, they were given a random number between 50 and 100, and were asked to continuously subtract three as quickly and accurately as possible, until the platform movement started. In response to the platform movement, the subject was instructed to try to maintain the upright balance by moving any part of the body except the arms.
2.4. Data analysis The perturbation onset time was first determined based on the first point at which the platform acceleration was above the mean, plus three times the standard deviation of the baseline level. The accelerations of each body segment prior to the platform movement onset were used to determine the initial orientation angle, which was then used for calculating the angular displacement of the body segment during the postural perturbation period, by numerical integration of the angular velocity. The segmental angular displacement
Fig. 2. (a) and (b) are the means and S.D. of the platform acceleration in forward and backward directions, respectively (n = 504). (c), (d) and (e) are time trajectories of shank acceleration, angular velocity and displacement, respectively. Two trials (the first, heavy line, and the fourth, light line) for standing on normal surface, eyes open, and moving forward. The parameters Pmax, Tp and To are indicated for the first trial. Time zero corresponds to the onset of the platform movement.
was also used to dynamically calibrate the accelerometers due to the effect of the gravity acceleration, in order to obtain real (kinematic) accelerations [20]. The following parameters were calculated for each kinematic variable in each trial: first maximum peak value (Pmax); first peak time (Tp) defined as the time at which the magnitude reached to a maximum for the first time; and onset time (To) determined using the same criterion for determining the onset of the platform movement. These temporal features are illustrated in Fig. 2. The means of all 12 conditions for each of the four trials were first computed for each subject. They were then normalized by subtracting the average value of the four trials. These normalized values are referred to in this study as relative values. This normalization procedure would remove the effect due to variation among the subjects. Friedman’s Test was then conducted on these relative values to identify differences among the four trials. If the test proved to be significant (PB
G. Wu / Gait and Posture 7 (1998) 110–116
113
0.05), Scheffe’s multiple comparisons test was made to detect which differences among the trials were significant (P B 0.05). The analysis was done for all the trials of all subjects, whether or not a stepping occurred. Since all the steppings occurred 1 s after the platform movement, they would not affect the temporal parameters extracted.
3. Results First of all, the repeatability of the platform movement was examined since variations in perturbation could contribute to variations in response. The means and standard deviations of the platform acceleration in both directions were first calculated and are shown in Fig. 2a, b. The maximum variations occurred during the transition from acceleration to deceleration of the platform. The variations at both positive and negative peaks were B 0.06 m/s2. Moreover, there were no systematic changes in the peaks and the timings of the acceleration over trials. The temporal trajectories of all the variables were consistent for most of the subjects. An example is shown in Fig. 2c–e, based on one subject, while standing on a normal surface, eyes open and moving forward. The first and the fourth trials are overlaid.
3.1. First maximum peak amplitude (Pmax) Fig. 3 shows the means and91 S.D. of Pmax for the angular displacement, angular velocity and linear accelerations of the head, trunk, thigh and shank. Statistical analyses on these variables indicated that the difference
Fig. 4. Means and 91 S.D. of Tp (n =1008) for the angular displacement, angular velocity, and linear accelerations (horizontal and vertical) of the head, trunk, thigh and shank. Only those trials that are significantly different are indicated in each corresponding figure.
in Pmax between the first and the later trials was significant for almost all the kinematic variables measured, except for the head angular displacement. For most of the cases, the significant difference occurred at about the third trial. In all cases, no significant difference was found between the third and fourth trial. With respect to the variation, the maximum standard deviation in the displacement was 0.02 rad ( 13% of mean angular displacement), B12°/s in the angular velocity of the trunk, thigh and shank ( 15% of mean angular velocity), and B0.2 m/s2 in acceleration ( 16 and 25% of the mean horizontal and vertical accelerations, respectively). A relatively larger variation in the vertical acceleration was mainly due to the fact that the major body movement was in the anterior/posterior direction, resulting in a relatively small acceleration in the vertical direction.
3.2. First peak time (Tp)
Fig. 3. Means and 91S.D. of Pmax (n= 1008) for the angular displacement, angular velocity, and linear accelerations (horizontal and vertical) of the head, trunk, thigh and shank. Only those trials that are significantly different are indicated in each corresponding figure.
Similarly, the means and9S.D. of Tp for the angular displacement, angular velocity and linear accelerations of the head, trunk, thigh and shank are shown in Fig. 4. For most of the cases, the Friedman’s test determined that there was not enough evidence to suggest a difference among the mean of the trials. The only exceptions were the accelerations of the shank and thigh. Further analysis showed that the significant changes found in the shank and thigh accelerations occurred between the first and the fourth trials only. It should be noted that the variations for all four trials were similar, except for the shank where the first trial varied slightly more than the later trials. Nevertheless, the maximum variation was B 70 ms, which was 16% of the mean peak time.
114
G. Wu / Gait and Posture 7 (1998) 110–116
3.3. Onset time (To) The results, as shown in Fig. 5, indicate non-significant differences between the first and the later trials in most of the parameters, except for the head angular velocity and trunk acceleration. For the trunk acceleration, the first trial was found to be significantly different from all later trials, whereas for the head angular velocity, the first trials were significantly different from the third and the fourth trials. The standard deviations in displacement, velocity and horizontal acceleration were B 20 ms, 10% of the mean onset time. Note that the shank and thigh were, in general, smaller. Again, a relatively larger variation in the vertical acceleration (40 ms) accounted for 20% of the mean onset time.
4. Discussion Postural adaptation, as quantified by the size (or magnitude) of postural muscle activation, has been well documented with the external postural perturbation being either sinusoidal [4,11] or transient [3,10]. Although the variables investigated in this study are mainly related to the kinematics, similar results to those postural adaptation observations should be expected, since the body movement (as described by the kinematics) is well correlated with the joint torques (as described by the muscle force or activation). Consistently, it is found in this study that the magnitudes of all but one of the kinematic variables observed were significantly affected by the number of trials that the subjects went through. Furthermore, the magnitude adaptation
Fig. 5. Means and91 S.D. of To (n= 1008) for the angular displacement, angular velocity, and linear accelerations (horizontal and vertical) of the head, trunk, thigh and shank. Only those trials that are significantly different are indicated in each corresponding figure.
found in this study is also consistent with the findings by others, that the adaptation process occurred within four trials [11]. To date, few investigations have been directed towards trial-independence in human dynamic posture. Consequently, the main question that this study intended to answer is whether or not there are variables (or features), other than the magnitude of the movement (or muscle activity) that are invariant over consecutive trials. Evidently, the results of this study have demonstrated that there are certain characteristics in the kinematic variables that are independent from the adaptation process. They include the peak time, the onset time and the head maximum displacement. It is interesting to observe in this study, that although the magnitude of the postural responses can be adapted when multiple trials with similar conditions are experienced by the subject, the timing at which the first peak value occurs does not seem to be adaptive. Since Tp reflects the timing of the compensation movement, these results provide some insight into the mechanism of postural adaptation. In the past, two opposite hierarchical models of adaptive strategy have been proposed. Bernstein first proposed a pre-established pattern control theory, which states that the movement specific pattern is established always prior to the initiation of movement [21]. In response to unexpected postural perturbation, the motor system gradually adapts its pattern in such a way that the executional errors will be reduced in the subsequent trials [22]. In contrast, another theory proposes that the adaptation occurs after the initiation of movement [23,24]. In doing so, it would be expected that the compensatory response time is delayed to allow additional time to alter the internal control pattern. However, the finding in this study, that Tp of almost all body segmental movement does not change significantly over consecutive trials, seems to support the former adaptation strategy: the adaptation occurs and completes before the subsequent movement arrives. The above proposition is further ascertained by the trial-independent characteristic of the initial movement onset time found in this study. It has been suggested that the initial body movement mainly represents the passive response to the postural perturbation [25]. From the motor control point of view, the initial passive motion of the body provides essential stimulation to the peripheral sensors whose outputs are then integrated to execute a corresponding compensatory movement. Therefore, the fact that initial movement time is trial-independent indicates that the time period between the initial stimulation to the peripheral sensory systems and the initiation of compensatory movement is a constant for every trial. This seems to contradict the latter adaptation theory, yet agree with the former one.
G. Wu / Gait and Posture 7 (1998) 110–116
Among all five body segments investigated in this study, the head maximum movement was the only one that was not trial-dependent. This distinctive behavior of the head seems to suggest that the head orientation has a different function than the others in maintaining a dynamic balance. In fact, Pozzo et al. [26] have found that, during various locomotor tasks, the head movement is composed of the standard posture (or postural fixation), whereas the lower limb movement behaves like an actuator [26]. This standard posture is further proposed to be used as a reference frame for maintaining dynamic equilibrium. The findings in this study indicate that this reference frame is well stabilized from the very first trial. It is the movement of the actuator (or other part of the body) that changes its magnitude at later trials. The findings in this study on the trial-independency of kinematic variables describing body movement during a dynamic posture have implications for both empirical and theoretical works. For example, the magnitude adaptation usually occurs at about the second trial and is more observable in the velocity and accelerations of each body segment than in the displacement. The knowledge of this time course is important when a time-varying model of the postural control system is attempted. On the other hand, the temporal features (such as the first peak time) of these kinematic variables are primarily controlled and show no sign of adaptation. Therefore, these characteristics can be extracted from the early trials of the experiment. It should be noted that human postural control is a complicated dynamic system. Many factors that have not been investigated in this study may possibly affect the temporal characteristics of the body movement. These factors include, for example, the characteristics of the external perturbation, the consciousness and the physical conditions of the individual subject. Moreover, as the results show, the adaptive characteristics of each of the parameters examined in this study are different, therefore, one should be cautious when the results are applied to other cases. Finally, it should be emphasized that all the parameters examined in this study are based on the early phase of the postural responses. More investigations are needed in order to address the trial independence of the temporal features in later part of the postural responses. In summary, this study investigated the trial-dependent characteristics of body segmental kinematic features during dynamic posture. These kinematic features included the amplitude and timings of the angular displacement, angular velocity and linear accelerations of the head, trunk, thigh and shank. The results from 21 healthy subjects showed that the amplitudes changed significantly from the first trial to the later trials, in most of the variables. However, there existed certain features, other than the amplitude, that were trial-inde-
115
pendent. Specifically, the time at which the initial peak amplitude occurred and the onset time were found to be similar between trials.
Acknowledgements The author would like to thank Mr Weifeng Zhao and Mrs Mary Becker for subject recruitment and data collection. This work was supported by a grant from the Whitaker Foundation and by a National Institute of Health grant No lR29AG1160201A2.
References [1] Johansson R, Magnusson M. Human postural dynamics. Crit Rev Biomed Eng 1991;18(6):413– 37. [2] Hanke TA, Rogers MW. Reliability of ground reaction force measurements during dynamic transitions from bipedal to singlelimb stance in healthy adults. Phys Ther 1992;72(11):810–6. [3] Keshner EA, Allum JHJ, Pfaltz CR. Postural coactivation and adaptation in the sway stabilizing responses of normals and patients with bilateral vestibular deficit. Exp Brain Res 1987;69:77 – 92. [4] Berger W, Discher M, Trippel M, Ibrahim IK, Dietz V. Developmental aspects of stance regulation, compensation and adaptation. Exp Brain Res 1992;90:610 – 9. [5] Hansen PD, Woollacott MH, Debu B. Postural responses to changing task conditions. Exp Brain Res 1988;73(3):627–36. [6] Maki BE, Whitelaw RS. Influence of expectation and arousal on center-of-pressure responses to transient postural perturbations. J. Vestibular Res 1993;3(1):25 – 39. [7] Horak FB, Diener HC, Nashner LM. Influence of central set on human postural responses. J. Neurophysiol 1989;62(4):841–53. [8] McIlroy WE, Malci BE. Adaptive changes to compensatory stepping response. Gait Posture 1995;3(1):43 – 55. [9] Stelmach GE, Teasdale N, Di Fabio RP, Phillips J. Age related decline in postural control mechanisms. Int J Aging Hum Dev 1989;29(3):205– 23. [10] Nashner LM. Adapting reflexes controlling the human posture. Exp Brain Res 1976;26:59 – 72. [11] Dietz V, Tripple M, Ibrahim IK, Berger W. Human stance on a sinusoidally translating platform: Balance control by feedforward and feedback mechanisms. Exp Brain Res 1993;93:352–62. [12] Kearney RE, Hunter IW. System identification of human joint dynamics. Crit Rev Biomed Eng 1990;18(1):55 – 87. [13] Kearney RE, Weiss PL, Morier R. Qantitative analysis of ankle joint dynamics: Inter-subject and intra-subject reliability. Clin Biomech 1990;5(4):205 – 17. [14] Marsden CD. The mysterious motor function of the basal ganglia. Neurology 1982;32:514 – 39. [15] McClenaghan BA, Williams H, Dickerson J, Thombs L. Spectral signature of forces to discriminate perturbation in standing posture. Clin Biomech 1994;9:21 – 7. [16] Horak FB, Nashner LM, Diener HC. Postural strategies associated with somatosensory and vestibular loss. Exp Brain Res 1990;82:167 – 77. [17] Sinclair AJ, Nayak USL. Age-related changes in postural suray. Compr Ther 1990;16(9):44 – 8. [18] Woollacott MH, Hosten CV, Ro¨sblad B. Relation between muscle response onset and body segmental movements during postural perturbations in humans. Exp Brain Res 1988;72:593– 604.
G. Wu / Gait and Posture 7 (1998) 110–116
116
[19] Wolfson L, Whipple R, Amerman P, Tobin JN. Gait assessment in the elderly: A gait abnormality rating scale and its relation to falls. J Gerontol Med Sci 1990;45:M12–9. [20] Wu G, Ladin Z. The Kinematometer—An integrated kinematic sensor for kinesiological measurements. ASME Trans Biomech Eng 1993;115(1):53–62. [21] Bernstein N. Coordination and Regulation of Movements. New York: Pergamon Press, 1967. [22] Welford AT. On sequencing of action. Brain Res 1974;71:381 – 92.
.
[23] Pew RW. Levels of analysis in motor control. Brain Res 1974;71:393 – 400. [24] Sears TA. The afferent regulation of learnt movements. Brain Res 1974;71:465 – 73. [25] Alexander NB, Shepard N, Gu MJ, Schultz A. Postural control in young and elderly adults when stance is perturbed: Kinematics. J Gerontol Med Sci 1992;47(3):M79– 87. [26] Pozzo T, Berthoz A, Lefort L. Head stabilization during various locomotory tasks in humans I. Normal subjects. Exp Brain Res 1990;82:97 – 106.
.
Examination of trial-independent characteristics of body kinematics in response to similar postural perturbations Ge Wu Department of Physical Therapy, Uni6ersity of Vermont, Burlington, VT 05405, USA Received 1 April 1996; received in revised form 25 June 1997; accepted 23 July 1997
Abstract This study investigated the variability of certain kinematic features describing body segmental movement in response to a series of consecutive, translational movements of the supporting base along the anterior-posterior direction of the subject. These kinematic features included the amplitude and timings of the angular displacement, velocity and acceleration of the head, trunk, thigh and shank. A group of 21 subjects was tested, with their ages ranging from 22 to 77 years. The results showed that the amplitudes changed significantly from the first trial to the later trials in most of the variables observed. However, there existed certain features, other than the amplitude, that were invariant over consecutive trials. Specifically, the time at which the initial peak amplitude occurred, the onset time was found to be similar between trials. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Adaptation; Trial-independence; Time-varying; Dynamic posture; Kinematics
1. Introduction Human postural control during upright stance is a complicated problem. Although there has been tremendous amount of effort over the past years in exploring the fundamental mechanism of human postural control, there has always been a critical question that relates to the experimental design in human postural studies. That is: are the strategies used to control upright balance reproducible after repeated exposure to similar postural perturbations [1,2]? In the past, understanding this question has been mainly limited to the changes in response size of muscular activities, based on the observations of adaptive attenuation/facilitation during stance or balance regulations in humans [3–9]. For example, in an early experiment on human dynamic postural control, functional stretch reflex in the leg muscle was found to either decrease its magnitude with consecutive movements of the supporting surface when it destabilized the posture, or increase its magnitude when it stabilized the posture [3,10,11]. Keshner et al. [3] also reported adaptive at0966-6362/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S 0 9 6 6 - 6 3 6 2 ( 9 7 ) 0 0 0 3 3 - 7
tenuation in the neck muscles. Many experiments have shown that postural adaptation usually occurs within three or four consecutive exposures to the same dynamic stimulation [3,10,11]. Besides adaptive attenuation/facilitation observed during human stance or balance regulations, are there certain biomechanical variables observable in dynamic human postural movement that are trial-independent? The term trial-independence in this study is defined as, no change over repeated exposures to the same dynamic postural stimulation. To date, few studies have addressed this trial-independent issue directly in human upright stance [12], although trial-independence has been reported in studies relating to other activities, such as active joint dynamics in a non weight bearing condition [13] and the control patterns (i.e. the timing) during locomotion [14]. Recently, McClenaghan et al. [15] explored the spectral signature of the ground reaction forces during a static stance task and found high reliability between multiple trials of each individual subject. All these findings suggest that there may exist certain parameters, especially the timing of the control
G. Wu / Gait and Posture 7 (1998) 110–116
strategies, that are trial-independent in maintaining a balanced upright posture. The mastery of trial-independent characteristics of human postural response after repeated exposure to similar postural perturbations is inevitably important because they are experimentally reproducible. In the past, in order to avoid the influence of adaptation on the experimental outcome, a frequently used approach is to precondition the subjects, that is, before the real experiment begins, the subjects need to practice for a period of time in the same environment as the one in which the data will be collected [16]. However, one of the potential problems with preconditioning is fatigue, which is non-stationary in the motor control system [12]. The other problem with preconditioning is missing the earliest, unpracticed trial which is most likely to be most relevant to the functional ability to balance in daily life, where we rarely get a chance to practice our response to a specific external perturbation. The purpose of this study was to investigate whether or not there are certain features of human postural control strategies that are trial-independent. The postural control strategy in this study was quantified by the movement of each individual body segment (i.e. head, trunk, thigh and shank) in response to a sequence of similar dynamic postural balance tasks. In the past, postural control has been represented by spontaneous body sway [17], electromyography of selected muscles [18], and gait patterns [19]. However, since whole body balance is obtained by proper coordination of single body segments, the movement of individual segments in a balance task will provide in-depth information about the functioning of the postural control system. In this study, the trial-dependence of the following characteristics of individual body segment movement was examined: the peak amplitude, the peak time and the onset time. Based on the results of other studies, using either spontaneous body sway or electromyography during posture or gait, it was hypothesized that the peak magnitude of body movements would change with increasing number of trials experienced, while both timing characteristics (i.e. onset time and peak time) would not show signs of adaptation. Visual, movement direction and the surface type of the supporting surface were altered to create a randomized sequence of perturbations.
111
control, vestibular function and muscle strength. Based on these tests, those selected were subjects free of skeletal and muscular problems that affect normal, daily life activities, significant loss of binocular vision, abnormal vestibular function, and medications likely to have an impact on the peripheral or central nervous system or muscular function. Before testing, each subject was asked to read and sign a consent form agreeing to the condition and expectations of the experiment.
2.2. Apparatus The human body was modeled as four rigid segments in the sagittal plane: the head, trunk and arms, thigh and shank (Fig. 1). The kinematic variables being studied included angular displacement, angular velocity and linear accelerations in the horizontal and vertical directions of each of the four body segments. The velocity and acceleration were directly measured using integrated kinematic sensors (IKS [20]), one on each body segment. Each IKS included one angular rate sensor (Watson, Eau Claire, WI) and two single axial linear accelerometers (Kistler, Amhurst, NY) arranged orthogonally. Another accelerometer was attached to a movable platform to measure its acceleration in the moving direction. The outputs from the IKSs and the platform accelerometer were collected at 120 Hz for 3 s.
2. Methodology
2.1. Subjects A total of 21 subjects (12 males and 9 females) participated in this experiment. They ranged from 22 to 77 years of age. All subjects were examined by a registered nurse to evaluate their fine and gross motor
Fig. 1. Schematic diagram illustrating the experimental configuration. Symbols a – n, a – t and v are the normal and tangential accelerations of a body segment, and a – p is the linear acceleration of the platform. Body segments are labeled as h for head, ta for trunk and arms, th for thigh, and s for shank.
112
G. Wu / Gait and Posture 7 (1998) 110–116
Postural perturbation was applied to the subject through a platform movement in the anterior-posterior direction of the subject. The platform movement was driven by an electromagnetic actuator with a control unit to alter the direction of the movement. The maximum acceleration was 8 m/s2, speed 40 cm/s and total displacement 8 cm.
2.3. Procedures During the experiment, each subject was tested for a total of 12 conditions, four times each. They included two visual conditions (vision/no vision) ×2 movement directions (platform moving forward/backward) × 3 surfaces (standing on hard/soft/reduced surfaces). The vision was either allowed or blocked by a pair of shuttered goggles which remained open most of the time to allow normal vision. For the blocked vision condition, the shutters were automatically closed at about the same time as the onset of the platform movement. The soft surface was a piece of foam (Foamaex Polyurethane foam, 5 cm thick, compliance of 0.086 kPa − 1) that covered the entire surface of the platform. The reduced surface was made of two pieces of wooden blocks (3.6 cm thick), that were cut into the shape of, and were attached to the rear half of the shoe bottom. In order to minimize subject’s anticipation of the platform movement, all these conditions were provided randomly by block design. That is, each block included a randomized sequence of all the conditions. Before each trial, the subject was instructed to stand quietly on the movable platform, with feet comfortably separated in the lateral direction, arms folded across the chest and eyes looking at a stationary object 25 m in front of them. In order to distract the subject’s attention to the upcoming platform movement, they were given a random number between 50 and 100, and were asked to continuously subtract three as quickly and accurately as possible, until the platform movement started. In response to the platform movement, the subject was instructed to try to maintain the upright balance by moving any part of the body except the arms.
2.4. Data analysis The perturbation onset time was first determined based on the first point at which the platform acceleration was above the mean, plus three times the standard deviation of the baseline level. The accelerations of each body segment prior to the platform movement onset were used to determine the initial orientation angle, which was then used for calculating the angular displacement of the body segment during the postural perturbation period, by numerical integration of the angular velocity. The segmental angular displacement
Fig. 2. (a) and (b) are the means and S.D. of the platform acceleration in forward and backward directions, respectively (n = 504). (c), (d) and (e) are time trajectories of shank acceleration, angular velocity and displacement, respectively. Two trials (the first, heavy line, and the fourth, light line) for standing on normal surface, eyes open, and moving forward. The parameters Pmax, Tp and To are indicated for the first trial. Time zero corresponds to the onset of the platform movement.
was also used to dynamically calibrate the accelerometers due to the effect of the gravity acceleration, in order to obtain real (kinematic) accelerations [20]. The following parameters were calculated for each kinematic variable in each trial: first maximum peak value (Pmax); first peak time (Tp) defined as the time at which the magnitude reached to a maximum for the first time; and onset time (To) determined using the same criterion for determining the onset of the platform movement. These temporal features are illustrated in Fig. 2. The means of all 12 conditions for each of the four trials were first computed for each subject. They were then normalized by subtracting the average value of the four trials. These normalized values are referred to in this study as relative values. This normalization procedure would remove the effect due to variation among the subjects. Friedman’s Test was then conducted on these relative values to identify differences among the four trials. If the test proved to be significant (PB
G. Wu / Gait and Posture 7 (1998) 110–116
113
0.05), Scheffe’s multiple comparisons test was made to detect which differences among the trials were significant (P B 0.05). The analysis was done for all the trials of all subjects, whether or not a stepping occurred. Since all the steppings occurred 1 s after the platform movement, they would not affect the temporal parameters extracted.
3. Results First of all, the repeatability of the platform movement was examined since variations in perturbation could contribute to variations in response. The means and standard deviations of the platform acceleration in both directions were first calculated and are shown in Fig. 2a, b. The maximum variations occurred during the transition from acceleration to deceleration of the platform. The variations at both positive and negative peaks were B 0.06 m/s2. Moreover, there were no systematic changes in the peaks and the timings of the acceleration over trials. The temporal trajectories of all the variables were consistent for most of the subjects. An example is shown in Fig. 2c–e, based on one subject, while standing on a normal surface, eyes open and moving forward. The first and the fourth trials are overlaid.
3.1. First maximum peak amplitude (Pmax) Fig. 3 shows the means and91 S.D. of Pmax for the angular displacement, angular velocity and linear accelerations of the head, trunk, thigh and shank. Statistical analyses on these variables indicated that the difference
Fig. 4. Means and 91 S.D. of Tp (n =1008) for the angular displacement, angular velocity, and linear accelerations (horizontal and vertical) of the head, trunk, thigh and shank. Only those trials that are significantly different are indicated in each corresponding figure.
in Pmax between the first and the later trials was significant for almost all the kinematic variables measured, except for the head angular displacement. For most of the cases, the significant difference occurred at about the third trial. In all cases, no significant difference was found between the third and fourth trial. With respect to the variation, the maximum standard deviation in the displacement was 0.02 rad ( 13% of mean angular displacement), B12°/s in the angular velocity of the trunk, thigh and shank ( 15% of mean angular velocity), and B0.2 m/s2 in acceleration ( 16 and 25% of the mean horizontal and vertical accelerations, respectively). A relatively larger variation in the vertical acceleration was mainly due to the fact that the major body movement was in the anterior/posterior direction, resulting in a relatively small acceleration in the vertical direction.
3.2. First peak time (Tp)
Fig. 3. Means and 91S.D. of Pmax (n= 1008) for the angular displacement, angular velocity, and linear accelerations (horizontal and vertical) of the head, trunk, thigh and shank. Only those trials that are significantly different are indicated in each corresponding figure.
Similarly, the means and9S.D. of Tp for the angular displacement, angular velocity and linear accelerations of the head, trunk, thigh and shank are shown in Fig. 4. For most of the cases, the Friedman’s test determined that there was not enough evidence to suggest a difference among the mean of the trials. The only exceptions were the accelerations of the shank and thigh. Further analysis showed that the significant changes found in the shank and thigh accelerations occurred between the first and the fourth trials only. It should be noted that the variations for all four trials were similar, except for the shank where the first trial varied slightly more than the later trials. Nevertheless, the maximum variation was B 70 ms, which was 16% of the mean peak time.
114
G. Wu / Gait and Posture 7 (1998) 110–116
3.3. Onset time (To) The results, as shown in Fig. 5, indicate non-significant differences between the first and the later trials in most of the parameters, except for the head angular velocity and trunk acceleration. For the trunk acceleration, the first trial was found to be significantly different from all later trials, whereas for the head angular velocity, the first trials were significantly different from the third and the fourth trials. The standard deviations in displacement, velocity and horizontal acceleration were B 20 ms, 10% of the mean onset time. Note that the shank and thigh were, in general, smaller. Again, a relatively larger variation in the vertical acceleration (40 ms) accounted for 20% of the mean onset time.
4. Discussion Postural adaptation, as quantified by the size (or magnitude) of postural muscle activation, has been well documented with the external postural perturbation being either sinusoidal [4,11] or transient [3,10]. Although the variables investigated in this study are mainly related to the kinematics, similar results to those postural adaptation observations should be expected, since the body movement (as described by the kinematics) is well correlated with the joint torques (as described by the muscle force or activation). Consistently, it is found in this study that the magnitudes of all but one of the kinematic variables observed were significantly affected by the number of trials that the subjects went through. Furthermore, the magnitude adaptation
Fig. 5. Means and91 S.D. of To (n= 1008) for the angular displacement, angular velocity, and linear accelerations (horizontal and vertical) of the head, trunk, thigh and shank. Only those trials that are significantly different are indicated in each corresponding figure.
found in this study is also consistent with the findings by others, that the adaptation process occurred within four trials [11]. To date, few investigations have been directed towards trial-independence in human dynamic posture. Consequently, the main question that this study intended to answer is whether or not there are variables (or features), other than the magnitude of the movement (or muscle activity) that are invariant over consecutive trials. Evidently, the results of this study have demonstrated that there are certain characteristics in the kinematic variables that are independent from the adaptation process. They include the peak time, the onset time and the head maximum displacement. It is interesting to observe in this study, that although the magnitude of the postural responses can be adapted when multiple trials with similar conditions are experienced by the subject, the timing at which the first peak value occurs does not seem to be adaptive. Since Tp reflects the timing of the compensation movement, these results provide some insight into the mechanism of postural adaptation. In the past, two opposite hierarchical models of adaptive strategy have been proposed. Bernstein first proposed a pre-established pattern control theory, which states that the movement specific pattern is established always prior to the initiation of movement [21]. In response to unexpected postural perturbation, the motor system gradually adapts its pattern in such a way that the executional errors will be reduced in the subsequent trials [22]. In contrast, another theory proposes that the adaptation occurs after the initiation of movement [23,24]. In doing so, it would be expected that the compensatory response time is delayed to allow additional time to alter the internal control pattern. However, the finding in this study, that Tp of almost all body segmental movement does not change significantly over consecutive trials, seems to support the former adaptation strategy: the adaptation occurs and completes before the subsequent movement arrives. The above proposition is further ascertained by the trial-independent characteristic of the initial movement onset time found in this study. It has been suggested that the initial body movement mainly represents the passive response to the postural perturbation [25]. From the motor control point of view, the initial passive motion of the body provides essential stimulation to the peripheral sensors whose outputs are then integrated to execute a corresponding compensatory movement. Therefore, the fact that initial movement time is trial-independent indicates that the time period between the initial stimulation to the peripheral sensory systems and the initiation of compensatory movement is a constant for every trial. This seems to contradict the latter adaptation theory, yet agree with the former one.
G. Wu / Gait and Posture 7 (1998) 110–116
Among all five body segments investigated in this study, the head maximum movement was the only one that was not trial-dependent. This distinctive behavior of the head seems to suggest that the head orientation has a different function than the others in maintaining a dynamic balance. In fact, Pozzo et al. [26] have found that, during various locomotor tasks, the head movement is composed of the standard posture (or postural fixation), whereas the lower limb movement behaves like an actuator [26]. This standard posture is further proposed to be used as a reference frame for maintaining dynamic equilibrium. The findings in this study indicate that this reference frame is well stabilized from the very first trial. It is the movement of the actuator (or other part of the body) that changes its magnitude at later trials. The findings in this study on the trial-independency of kinematic variables describing body movement during a dynamic posture have implications for both empirical and theoretical works. For example, the magnitude adaptation usually occurs at about the second trial and is more observable in the velocity and accelerations of each body segment than in the displacement. The knowledge of this time course is important when a time-varying model of the postural control system is attempted. On the other hand, the temporal features (such as the first peak time) of these kinematic variables are primarily controlled and show no sign of adaptation. Therefore, these characteristics can be extracted from the early trials of the experiment. It should be noted that human postural control is a complicated dynamic system. Many factors that have not been investigated in this study may possibly affect the temporal characteristics of the body movement. These factors include, for example, the characteristics of the external perturbation, the consciousness and the physical conditions of the individual subject. Moreover, as the results show, the adaptive characteristics of each of the parameters examined in this study are different, therefore, one should be cautious when the results are applied to other cases. Finally, it should be emphasized that all the parameters examined in this study are based on the early phase of the postural responses. More investigations are needed in order to address the trial independence of the temporal features in later part of the postural responses. In summary, this study investigated the trial-dependent characteristics of body segmental kinematic features during dynamic posture. These kinematic features included the amplitude and timings of the angular displacement, angular velocity and linear accelerations of the head, trunk, thigh and shank. The results from 21 healthy subjects showed that the amplitudes changed significantly from the first trial to the later trials, in most of the variables. However, there existed certain features, other than the amplitude, that were trial-inde-
115
pendent. Specifically, the time at which the initial peak amplitude occurred and the onset time were found to be similar between trials.
Acknowledgements The author would like to thank Mr Weifeng Zhao and Mrs Mary Becker for subject recruitment and data collection. This work was supported by a grant from the Whitaker Foundation and by a National Institute of Health grant No lR29AG1160201A2.
References [1] Johansson R, Magnusson M. Human postural dynamics. Crit Rev Biomed Eng 1991;18(6):413– 37. [2] Hanke TA, Rogers MW. Reliability of ground reaction force measurements during dynamic transitions from bipedal to singlelimb stance in healthy adults. Phys Ther 1992;72(11):810–6. [3] Keshner EA, Allum JHJ, Pfaltz CR. Postural coactivation and adaptation in the sway stabilizing responses of normals and patients with bilateral vestibular deficit. Exp Brain Res 1987;69:77 – 92. [4] Berger W, Discher M, Trippel M, Ibrahim IK, Dietz V. Developmental aspects of stance regulation, compensation and adaptation. Exp Brain Res 1992;90:610 – 9. [5] Hansen PD, Woollacott MH, Debu B. Postural responses to changing task conditions. Exp Brain Res 1988;73(3):627–36. [6] Maki BE, Whitelaw RS. Influence of expectation and arousal on center-of-pressure responses to transient postural perturbations. J. Vestibular Res 1993;3(1):25 – 39. [7] Horak FB, Diener HC, Nashner LM. Influence of central set on human postural responses. J. Neurophysiol 1989;62(4):841–53. [8] McIlroy WE, Malci BE. Adaptive changes to compensatory stepping response. Gait Posture 1995;3(1):43 – 55. [9] Stelmach GE, Teasdale N, Di Fabio RP, Phillips J. Age related decline in postural control mechanisms. Int J Aging Hum Dev 1989;29(3):205– 23. [10] Nashner LM. Adapting reflexes controlling the human posture. Exp Brain Res 1976;26:59 – 72. [11] Dietz V, Tripple M, Ibrahim IK, Berger W. Human stance on a sinusoidally translating platform: Balance control by feedforward and feedback mechanisms. Exp Brain Res 1993;93:352–62. [12] Kearney RE, Hunter IW. System identification of human joint dynamics. Crit Rev Biomed Eng 1990;18(1):55 – 87. [13] Kearney RE, Weiss PL, Morier R. Qantitative analysis of ankle joint dynamics: Inter-subject and intra-subject reliability. Clin Biomech 1990;5(4):205 – 17. [14] Marsden CD. The mysterious motor function of the basal ganglia. Neurology 1982;32:514 – 39. [15] McClenaghan BA, Williams H, Dickerson J, Thombs L. Spectral signature of forces to discriminate perturbation in standing posture. Clin Biomech 1994;9:21 – 7. [16] Horak FB, Nashner LM, Diener HC. Postural strategies associated with somatosensory and vestibular loss. Exp Brain Res 1990;82:167 – 77. [17] Sinclair AJ, Nayak USL. Age-related changes in postural suray. Compr Ther 1990;16(9):44 – 8. [18] Woollacott MH, Hosten CV, Ro¨sblad B. Relation between muscle response onset and body segmental movements during postural perturbations in humans. Exp Brain Res 1988;72:593– 604.
G. Wu / Gait and Posture 7 (1998) 110–116
116
[19] Wolfson L, Whipple R, Amerman P, Tobin JN. Gait assessment in the elderly: A gait abnormality rating scale and its relation to falls. J Gerontol Med Sci 1990;45:M12–9. [20] Wu G, Ladin Z. The Kinematometer—An integrated kinematic sensor for kinesiological measurements. ASME Trans Biomech Eng 1993;115(1):53–62. [21] Bernstein N. Coordination and Regulation of Movements. New York: Pergamon Press, 1967. [22] Welford AT. On sequencing of action. Brain Res 1974;71:381 – 92.
.
[23] Pew RW. Levels of analysis in motor control. Brain Res 1974;71:393 – 400. [24] Sears TA. The afferent regulation of learnt movements. Brain Res 1974;71:465 – 73. [25] Alexander NB, Shepard N, Gu MJ, Schultz A. Postural control in young and elderly adults when stance is perturbed: Kinematics. J Gerontol Med Sci 1992;47(3):M79– 87. [26] Pozzo T, Berthoz A, Lefort L. Head stabilization during various locomotory tasks in humans I. Normal subjects. Exp Brain Res 1990;82:97 – 106.
.