- Email: [email protected]
Corrective postural responses evoked by completely unexpected loss of ground support during human walking
Gait & Posture 29 (2009) 483–487
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Corrective postural responses evoked by completely unexpected loss of ground support during human walking Masahiro Shinya a,b, Shinya Fujii a,b, Shingo Oda a,* a b
Laboratory of Human Motor Control, Graduate School of Human and Environmental Studies, Kyoto University, Yoshida-nihonmatsu-cho, Sakyo-ku, Kyoto 606-8501, Japan Research Fellow of the Japan Society for the Promotion of Science, Japan
A R T I C L E I N F O
A B S T R A C T
Article history: Received 12 June 2008 Received in revised form 18 November 2008 Accepted 19 November 2008
Understanding reactive responses to unexpected perturbation is fundamental to research on falls and their prevention. In this study, seven healthy young males walked along a walkway with and without a completely unexpected loss of ground support that was made by removing a wooden board (6.5 cm high) mounted on the walkway. Electromyography (EMG), ground reaction forces, and knee and ankle kinematics were recorded and comprehensively analyzed to investigate the corrective postural response to the perturbation. Three sequential strategies were observed. First, the fastest response was the reflexive muscle activity of the perturbed ankle, which we argue was evoked and enhanced by the absence of the expected somatosensory afferents at the expected heel contact. We also demonstrated that rapid soleus activity partially contributed to absorbing the impact of the actual touchdown. Second, after the touchdown, we argue that the central nervous system may reset the gait rhythm to permit continued walking by delaying the subsequent take-off. As a result, the duration of the total stance phase was identical to that recorded during normal walking. Third, we observed an adaptive locomotion to surmount the hole; both knees were more flexed than normal in order to allow the subject to withdraw the perturbed leg from the hole. ß 2008 Elsevier B.V. All rights reserved.
Keywords: Loss of ground support Electromyography Muscle response Unexpected perturbation Human walking
1. Introduction Every day, individuals must react to various unexpected perturbations to maintain balance and continue walking. Many studies have investigated the mechanisms of postural control in the context of unexpected perturbations during human walking; such as slipping [1,2], tripping [3,4], head tilt [5], sudden drop of the support surface [6], external force on the waist [7] or knee joint [8], and walking on an inclined platform [9] or compliant surface [10]. These studies have reported task-dependent, functionally appropriate muscle responses and strategies for any given perturbation. However, very few studies to date have considered postural responses to an unexpected loss of footing caused by a misstep into a hole, which is a potential cause of serious falls and injuries [11]. Recently, two studies were conducted to investigate the muscular and postural responses to an unexpected difference in ground level during human walking [12,13]. van der Linden et al. [12] reported that muscle responses similar to braking synergy [14] were triggered by the absence of the somatic afferent activation expected at heel contact when the subject missteps into
a hole unexpectedly. van Dieen et al. [13] measured the kinetics and kinematics of postural control upon an unexpected change in the ground level while stepping down, and they reported that the perturbed leg did not flex. These previous studies, however, were based on either electromyographic or mechanical data; to date, the necessary integration of muscular and postural responses has not been performed. Such an integrative approach would help us understand balance control in response to an unexpected loss of ground support. Furthermore, the fact that the subjects were able to anticipate the possibility of the perturbation, due to the warning before the experimental tasks, may have compromised these results. To investigate the overall postural control in response to a completely unexpected loss of ground support in daily life, we conducted a study in which participants were not informed of the impending drop in the walkway. We analyzed electromyographic data from the perturbed ankle, and we explored the resulting kinetics and kinematics. 2. Methods 2.1. Participants
* Corresponding author. Tel.: +81 75 753 6876; fax: +81 75 753 6876. E-mail address: [email protected] (S. Oda). 0966-6362/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2008.11.009
Seven males with no history of musculoskeletal or neurological disorders participated in this study. All of the subjects were right-footed. Their mean (SD) age, height, and weight were 22.9 0.7 years, 170.0 3.9 cm, and 69.3 6.7 kg,
484
M. Shinya et al. / Gait & Posture 29 (2009) 483–487
Fig. 1. Experimental apparatus. (A) Overhead view of the walkway showing the force plate and camera positions. (B) Side view.
respectively. They provided informed consent to undergo the experimental procedures, which were conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of the Graduate School of Human and Environmental Studies, Kyoto University, Japan (approval number: 19-H-8). 2.2. Protocol A walkway was built (1200 cm 150 cm 15 cm), and a force plate (OR6-72000, AMTI, 46.5 cm 51 cm 8.5 cm) was incorporated in the center (Fig. 1A). A wooden board (51 cm 46 cm 6.5 cm) was mounted on the force plate such that its top surface was flush with the walkway (Fig. 1B). A drop in the support surface (6.5 cm) could be created by removing the board. The board and force plate were covered with a thin black cloth to conceal the drop. Previous testing of the force plate with the wooden board and black cloth demonstrated accuracy of data acquisition. Before data collection, each participant determined his starting position and sufficiently practiced to ensure that his 8th step was collocated with the wooden board while walking at a comfortable speed and cadence. The experiment featured 10 trials. To collect baseline gait data, the wooden board was mounted on the force plate, and the support surface was not dropped in the first nine trials (control trials). On the 10th trial, the wooden board was removed (perturbed trial). Surface electromyography (EMG) electrodes were attached to the perturbed right ankle muscles—the medial gastrocnemius (MG), soleus (SOL) and tibialis anterior (TA). EMG signals were bandpass-filtered at 5–250 Hz, amplified, and sampled at 1000 Hz. The ground reaction forces were bandpass-filtered at 0.5– 300 Hz, amplified, and sampled at 1000 Hz. The kinematic data were collected using three cameras (Photoron; frame rate, 125 Hz). Light-reflecting markers were placed bilaterally on the fifth metatarsal, lateral malleolus, lateral femoral epicondyle, greater trochanter, iliac crest, acromion, lateral humerus epicondyle, lateral styloid process and fossa temporalis and at the center of the medial claviculars. Threedimensional positions of the markers were calculated using digitizing software (Frame-DIAS II, DKH). The estimated standard error of the direct linear transformation calculation was less than 1 cm in all three axes. These position data were used to calculate the knee angle (defined by the greater trochanter, lateral femoral epicondyle, and lateral malleolus) and ankle angle (defined by the lateral femoral epicondyle, lateral malleolus, and fifth metatarsal). 2.3. Data analysis Data were analyzed using custom-written Matlab code (The Mathworks, Inc., Natick, MA, USA). Kinematic data were low-pass-filtered (third-order Butterworth filter) at a cut-off frequency of 10 Hz and then interpolated and resampled at 1000 Hz, using a spline interpolation algorithm. For the control trials, foot contact (FC) was determined using a vertical ground reaction force (VGRF) threshold of 5 N. In the perturbed trial, the expected foot contact (EFC) was calculated when the vertical displacement of the right lateral malleolus fell 2 SDs below the mean of the control condition. Touch-down (TD) to the dropped surface was determined using a threshold of 5 N. For both the control and perturbed conditions, we defined midstance (MS) as the time when the anteroposterior ground reaction force changed sign from negative to positive. Take-off (TO) of the right foot was determined at a VGRF threshold of 5 N. EMG signals were high-pass-filtered at 5 Hz, full-wave
rectified, and low-pass-filtered using an 11-point boxcar moving average filter. The EMG response onset was defined as the time when the EMG activity of the perturbed trial exceeded those of the control trials by 2 SDs. EMG onsets and the duration between EFC and TD were compared using the one-way analysis of variance (ANOVA) and Tukey’s post-hoc test. To quantify the impact of FC in the control trials and TD in the perturbed trials, we calculated the impulse by integrating the VGRF during the 100-ms period after FC and TD, respectively. The integration of the rectified EMG signals (IEMG) during the 100-ms period after FC and TD was calculated as an index of ankle muscle activity after FC and TD, respectively. Pearson’s coefficient of correlation between the time from the onset of the SOL contraction to TD and the vertical impulse during the 100-ms period after TD was calculated. In the swing phase of the additional steps after the perturbation (9th and 10th steps in the perturbed condition), the maximum knee flexion angles of the left and right limbs, and the maximum ankle plantar-flexion of the left ankle were compared with those in the control condition. For the right ankle, angles at TO instead of the maximum value in the swing phase were compared between conditions because the entire swing phase of the right limb was not captured. The differences in the kinematic and temporal parameters between conditions were tested using the paired t test. An alpha level of 0.05 was considered significant for all statistics.
3. Results All participants performed nine control trials and one perturbed trial without any falls or injuries. In the perturbed trial, each subject lost his footing at the 8th step with his right leg. After touchdown to the dropped ground, all subjects took at least two additional steps (9th and 10th) and then stopped walking. Typical VGRF and EMG recordings from the lower leg muscles during the control and perturbed trials are shown in Fig. 2A. The VGRF and EMG activity traces were larger in the perturbed trial than in the control trial. The onset of EMG activity for the ankle plantar-flexor muscles (latencies from the EFC: MG, 97 16 ms and SOL, 117 23 ms) occurred before touchdown (128 10 ms); additionally, the onset of EMG activity for the dorsi-flexor (TA, 126 17 ms) occurred near TD (Fig. 2B). One-way ANOVA revealed significant differences among these four values [F(3, 24) = 4.71, p < 0.05]. A post-hoc Tukey’s test indicated that the EMG latency of the MG was significantly shorter than that of the TA and TD (p < 0.05). Greater EMG responses in both the ankle plantar-flexor and the dorsi-flexor were observed after TD (Table 1). The VGRF impulse during the 100-ms period after TD in the perturbed trial was larger than that after FC in the control trial. Some subjects started SOL activity very quickly, while others started after the TD (see error bar in Fig. 2B). This leads to the hypothesis that the inter-subject
M. Shinya et al. / Gait & Posture 29 (2009) 483–487
485
Fig. 2. Fast muscle responses to the unexpected loss of ground support. (A) Typical example of data recorded from a single subject in the control (gray) and the perturbed trial (black). The up arrows show the onset of muscle activity. (B) Mean and standard deviations of the latency associated with the electromyography response from EFC. The asterisk denotes a significant difference (p < 0.05) compared with the TA and TD. (C) Relationship between the onset of soleus activity and vertical impulse during the 100-ms period following TD in the perturbed trials. The time was aligned so that time 0 coincided with TD (vertical dotted line).
variation in SOL onset may explain variability in the TD impulse. Fig. 2C shows the relationship between the time from the SOL onset to TD and the vertical impulse in the 100-ms period after TD in the perturbed trials. The coefficient of correlation was 0.79 (p < 0.05), indicating that the faster the onset of SOL activity was, the smaller the TD impulse. Typical kinetic and kinematic data for the normal and perturbed gait are shown in Fig. 3. The unexpected loss of ground support delayed TD 128 10 ms more than EFC, while MS was not changed Table 1 Mean values of iEMG, kinetics, duration of walking phase, and joint angles during normal and perturbed walking. Variables iEMG MG (mV s) SOL (mV s) TA (mV s) Kinetics VGRF peak (%Weight) VGRF impulse (%Weight s) Duration of walking phase FC (TD) to MS (ms) MS to TO (ms) FC (TD) to TO (ms) Joint Angle L knee (8) R knee (8) L ankle (8) R ankle (8)
Control
Perturbed
P
15.5 11.6 4.2 1.3 8.2 7.4
37.3 22.7 24.5 5.8 63.9 29.3
<0.05* <0.001* <0.01*
109.4 5 .5 5.3 0.5
208.4 24.6 13.9 1.7
<0.001* <0.001*
(Table 1). Moreover, TO was also delayed in the perturbed trial, and the time interval between TD and TO in the perturbed trial was consistent with that between FC and TO in the control trial. After the perturbation, subjects continued walking with an alteration of the knee and ankle motions compared with normal walking. Both the left and right knees were more flexed, and the left ankle was more plantar-flexed during the swing phases of subsequent steps (Table 1). 4. Discussion Our data may be regarded as evidence of corrective postural responses to a completely unexpected loss of ground support during walking; the subjects were not warned of the possibility of the perturbation. Three sequential strategies for reestablishing balance were observed. (1) The reflexive muscle activities were evoked in the ankle plantar-flexors and dorsi-flexors. (2) The walking rhythm was reset during the perturbed stance phase. (3) The participants continued the adaptive locomotion to surmount the hole by flexing the right and left knees during the swing phase of subsequent steps before they stopped walking. 4.1. Reflexive muscle activity
*
373.5 23.3 335.7 29.3 709.2 26.5
219.0 43.5 457.7 65.4 676.7 62.1
<0.01 <0.01* 0.21
116 2.3 118 4.9 135 7.2 135 6.1
82 8.8 104 8.6 146 4.6 141 9.6
<0.001* <0001* <0.05* 0.18
iEMG and VGRF impulse were calculated during the 100-ms period after foot contact (FC) under the control condition and at touchdown (TD) under the perturbed condition. Joint angles are the maximum values during the swing phase, except in the case of the right ankle. The angle of the right ankle is the value at takeoff. * Significant difference between control and perturbed conditions.
Regarding the cutaneous input before the actual TD, the subjects touched only a thin cloth, which could not have evoked a muscle response. The observed EMG latency (MG, 97 ms) was also sufficiently short, indicating that the visual information did not trigger the EMG response [15]. Additionally, the somatosensory input associated with the actual TD could not possibly have served as the trigger because the EMG response was observed before the TD. Several previous studies have proposed that the difference between the actual and expected afferent inputs may evoke the reflexive muscle activity. In intact cats, the leg flexor muscles were rapidly activated to lift the paws soon after the foot entered a hole during walking [16,17]. These responses were absent in spinal cats,
486
M. Shinya et al. / Gait & Posture 29 (2009) 483–487
segments [13], but they might have contributed to absorbing the impact of the actual TD by stiffening the ankle joint [19]. In this study, the impulse after the TD decreased because the SOL activity was initiated earlier, suggesting that rapid contraction of the SOL might contribute to impulse absorption. Furthermore, simultaneous increases in muscle activity after TD were observed in all monitored muscles. Such muscle co-contraction might be a good strategy to cope with perturbations and to stabilize the ankle joint and the center of mass. 4.2. Phase reset of gait rhythm Resetting the gait rhythm is a good strategy to regain balance after various perturbations, such as stumbling during walking [4]. In-silico simulations also demonstrated the efficacy of resetting the gait rhythm in response to external impulsive perturbations to prevent falling [20,21]. In the present study, the latency between FC/EFC and MS was not altered by the perturbation, perhaps because the perturbed leg was nearly vertical at TD and consequently the forward horizontal and angular momentum could not be reduced [13]. This implies that the perturbation did not change the gait rhythm until MS. However, TO of the perturbed foot was delayed, and the duration of the stance phase was therefore the same as that in the control trial. The central nervous system reset the gait phase by delaying TO and maintaining the duration of the stance phase as a recovery strategy after the completely unexpected loss of ground support. 4.3. Adaptive locomotion to surmount the hole
Fig. 3. Typical examples of the ground reaction forces and the knee and ankle angles in the control (gray) and the perturbed (black) trial. The right leg is the support leg, while the left leg is the swing leg. Full knee extension and full ankle plantar-flexion occur at angles of 1808. The temporal events of the right leg are represented in the bottom of the graphs and vertical lines. For the control trial, FC: foot contact; MS: mid-stance; and TO: take-off. For the perturbed trial, EFC: expected foot contact; TD: actual touchdown; MS: mid-stance; and TO: take-off.
indicating that supraspinal structures contributed to the muscle reflexes. Humans also show reflexive muscle responses when encountering a difference in the support surface [12]. Since our results (ankle plantar-flexor contraction followed by dorsi-flexor contraction) were consistent with those of the previous study [12], the fastest response (MG) might be a muscle reflex evoked by the difference between the actual and expected somatosensory afferents via supraspinal structures. Larger muscle activities were also observed after the actual TD in both ankle plantar-flexors and dorsi-flexors. These muscle activities might be reflexes evoked by the afferent input accompanying the actual TD. It has been reported that, when one passes through a false floor, the stretch reflexes evoked from the impact with a lower, solid floor are twice as large as normal [18]. In the present study, the unexpected loss of ground support possibly enhanced the stretch reflex gain. How did the observed muscle responses contribute to postural control? It is conceivable that the braking synergy [14] was evoked by the completely unexpected loss of ground support in the present study, as van der Linden et al. [12] previously suggested. Given the electromechanical delay, the muscle activities before the actual TD did not change the positions or angles of the body
During the swing phase of subsequent steps, both the left and right knees were more flexed than during normal walking in the previous control trials. From the perspective of danger avoidance, it may be undesirable to keep one’s foot in a hole. It is well known that subjects will generally flex their knees to avoid obstacles during walking [22]. Thus, the knee flexion observed in our study might be a strategy for withdrawing the perturbed knee and securing toe clearance. Plantar-flexion of the left ankle during subsequent steps might be relevant for a cautious toe-first landing strategy instead of normal heel-first contact. Such adaptive strategies might be important to ensure security and to guarantee one’s ability to continue walking for several steps after the perturbation induced by the unexpected loss of ground support. Acknowledgements We would like to thank Dr. T. Kawabata for allowing us to use the EMG recording device. Conflict of interest We have no conflict of interests with other scientists or groups. References [1] Marigold DS, Patla AE. Strategies for dynamic stability during locomotion on a slippery surface: effects of prior experience and knowledge. J Neurophysiol 2002;88(1):339–53. [2] Tang PF, Woollacott MH, Chong RK. Control of reactive balance adjustments in perturbed human walking: roles of proximal and distal postural muscle activity. Exp Brain Res 1998;119(2):141–52. [3] Eng JJ, Winter DA, Patla AE. Strategies for recovery from a trip in early and late swing during human walking. Exp Brain Res 1994;102(2):339–49. [4] Schillings AM, van Wezel BM, Mulder T, Duysens J. Muscular responses and movement strategies during stumbling over obstacles. J Neurophysiol 2000;83(4):2093–102. [5] Dietz V, Horstmann GA, Berger W. Fast head tilt has only a minor effect on quick compensatory reactions during the regulation of stance and gait. Exp Brain Res 1988;73(3):470–6.
M. Shinya et al. / Gait & Posture 29 (2009) 483–487 [6] Nakazawa K, Kawashima N, Akai M, Yano H. On the reflex coactivation of ankle flexor and extensor muscles induced by a sudden drop of support surface during walking in humans. J Appl Physiol 2004;96(2):604–11. [7] Misiaszek JE. Early activation of arm and leg muscles following pulls to the waist during walking. Exp Brain Res 2003;151(3):318–29. [8] Dietz V, Colombo G, Muller R. Single joint perturbation during gait: neuronal control of movement trajectory. Exp Brain Res 2004;158(3):308–16. [9] Nieuwenhuijzen PH, Gruneberg C, Duysens J. Mechanically induced ankle inversion during human walking and jumping. J Neurosci Methods 2002; 117(2):133–40. [10] Marigold DS, Patla AE. Adapting locomotion to different surface compliances: neuromuscular responses and changes in movement dynamics. J Neurophysiol 2005;94(3):1733–50. [11] Berg WP, Alessio HM, Mills EM, Tong C. Circumstances and consequences of falls in independent community-dwelling older adults. Age Ageing 1997;26(4):261–8. [12] van der Linden MH, Marigold DS, Gabreels FJ, Duysens J. Muscle reflexes and synergies triggered by an unexpected support surface height during walking. J Neurophysiol 2007;97(5):3639–50. [13] van Dieen JH, Spanjaard M, Konemann R, Bron L, Pijnappels M. Balance control in stepping down expected and unexpected level changes. J Biomech 2007; 40(16):3641–9.
487
[14] Hase K, Stein RB. Analysis of rapid stopping during human walking. J Neurophysiol 1998;80(1):255–61. [15] Kandel ER, Schwartz JH, Jessell TM. Principles of Neural Science. Appleton & Lange; 2000. [16] Gorassini MA, Prochazka A, Hiebert GW, Gauthier MJ. Corrective responses to loss of ground support during walking. I. Intact cats. J Neurophysiol 1994;71(2):603–10. [17] Hiebert GW, Gorassini MA, Jiang W, Prochazka A, Pearson KG. Corrective responses to loss of ground support during walking. II. Comparison of intact and chronic spinal cats. J Neurophysiol 1994;71(2):611–22. [18] McDonagh MJ, Duncan A. Interaction of pre-programmed control and natural stretch reflexes in human landing movements. J Physiol 2002;544(Pt 3): 985–94. [19] Santello M. Review of motor control mechanisms underlying impact absorption from falls. Gait Posture 2005;21(1):85–94. [20] Yamasaki T, Nomura T, Sato S. Possible functional roles of phase resetting during walking. Biol Cybern 2003;88(6):468–96. [21] Nakanishi M, Nomura T, Sato S. Stumbling with optimal phase reset during gait can prevent a humanoid from falling. Biol Cybern 2006;95(5):503–15. [22] Lam T, Dietz V. Transfer of motor performance in an obstacle avoidance task to different walking conditions. J Neurophysiol 2004;92(4):2010–6.
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Corrective postural responses evoked by completely unexpected loss of ground support during human walking Masahiro Shinya a,b, Shinya Fujii a,b, Shingo Oda a,* a b
Laboratory of Human Motor Control, Graduate School of Human and Environmental Studies, Kyoto University, Yoshida-nihonmatsu-cho, Sakyo-ku, Kyoto 606-8501, Japan Research Fellow of the Japan Society for the Promotion of Science, Japan
A R T I C L E I N F O
A B S T R A C T
Article history: Received 12 June 2008 Received in revised form 18 November 2008 Accepted 19 November 2008
Understanding reactive responses to unexpected perturbation is fundamental to research on falls and their prevention. In this study, seven healthy young males walked along a walkway with and without a completely unexpected loss of ground support that was made by removing a wooden board (6.5 cm high) mounted on the walkway. Electromyography (EMG), ground reaction forces, and knee and ankle kinematics were recorded and comprehensively analyzed to investigate the corrective postural response to the perturbation. Three sequential strategies were observed. First, the fastest response was the reflexive muscle activity of the perturbed ankle, which we argue was evoked and enhanced by the absence of the expected somatosensory afferents at the expected heel contact. We also demonstrated that rapid soleus activity partially contributed to absorbing the impact of the actual touchdown. Second, after the touchdown, we argue that the central nervous system may reset the gait rhythm to permit continued walking by delaying the subsequent take-off. As a result, the duration of the total stance phase was identical to that recorded during normal walking. Third, we observed an adaptive locomotion to surmount the hole; both knees were more flexed than normal in order to allow the subject to withdraw the perturbed leg from the hole. ß 2008 Elsevier B.V. All rights reserved.
Keywords: Loss of ground support Electromyography Muscle response Unexpected perturbation Human walking
1. Introduction Every day, individuals must react to various unexpected perturbations to maintain balance and continue walking. Many studies have investigated the mechanisms of postural control in the context of unexpected perturbations during human walking; such as slipping [1,2], tripping [3,4], head tilt [5], sudden drop of the support surface [6], external force on the waist [7] or knee joint [8], and walking on an inclined platform [9] or compliant surface [10]. These studies have reported task-dependent, functionally appropriate muscle responses and strategies for any given perturbation. However, very few studies to date have considered postural responses to an unexpected loss of footing caused by a misstep into a hole, which is a potential cause of serious falls and injuries [11]. Recently, two studies were conducted to investigate the muscular and postural responses to an unexpected difference in ground level during human walking [12,13]. van der Linden et al. [12] reported that muscle responses similar to braking synergy [14] were triggered by the absence of the somatic afferent activation expected at heel contact when the subject missteps into
a hole unexpectedly. van Dieen et al. [13] measured the kinetics and kinematics of postural control upon an unexpected change in the ground level while stepping down, and they reported that the perturbed leg did not flex. These previous studies, however, were based on either electromyographic or mechanical data; to date, the necessary integration of muscular and postural responses has not been performed. Such an integrative approach would help us understand balance control in response to an unexpected loss of ground support. Furthermore, the fact that the subjects were able to anticipate the possibility of the perturbation, due to the warning before the experimental tasks, may have compromised these results. To investigate the overall postural control in response to a completely unexpected loss of ground support in daily life, we conducted a study in which participants were not informed of the impending drop in the walkway. We analyzed electromyographic data from the perturbed ankle, and we explored the resulting kinetics and kinematics. 2. Methods 2.1. Participants
* Corresponding author. Tel.: +81 75 753 6876; fax: +81 75 753 6876. E-mail address: [email protected] (S. Oda). 0966-6362/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2008.11.009
Seven males with no history of musculoskeletal or neurological disorders participated in this study. All of the subjects were right-footed. Their mean (SD) age, height, and weight were 22.9 0.7 years, 170.0 3.9 cm, and 69.3 6.7 kg,
484
M. Shinya et al. / Gait & Posture 29 (2009) 483–487
Fig. 1. Experimental apparatus. (A) Overhead view of the walkway showing the force plate and camera positions. (B) Side view.
respectively. They provided informed consent to undergo the experimental procedures, which were conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of the Graduate School of Human and Environmental Studies, Kyoto University, Japan (approval number: 19-H-8). 2.2. Protocol A walkway was built (1200 cm 150 cm 15 cm), and a force plate (OR6-72000, AMTI, 46.5 cm 51 cm 8.5 cm) was incorporated in the center (Fig. 1A). A wooden board (51 cm 46 cm 6.5 cm) was mounted on the force plate such that its top surface was flush with the walkway (Fig. 1B). A drop in the support surface (6.5 cm) could be created by removing the board. The board and force plate were covered with a thin black cloth to conceal the drop. Previous testing of the force plate with the wooden board and black cloth demonstrated accuracy of data acquisition. Before data collection, each participant determined his starting position and sufficiently practiced to ensure that his 8th step was collocated with the wooden board while walking at a comfortable speed and cadence. The experiment featured 10 trials. To collect baseline gait data, the wooden board was mounted on the force plate, and the support surface was not dropped in the first nine trials (control trials). On the 10th trial, the wooden board was removed (perturbed trial). Surface electromyography (EMG) electrodes were attached to the perturbed right ankle muscles—the medial gastrocnemius (MG), soleus (SOL) and tibialis anterior (TA). EMG signals were bandpass-filtered at 5–250 Hz, amplified, and sampled at 1000 Hz. The ground reaction forces were bandpass-filtered at 0.5– 300 Hz, amplified, and sampled at 1000 Hz. The kinematic data were collected using three cameras (Photoron; frame rate, 125 Hz). Light-reflecting markers were placed bilaterally on the fifth metatarsal, lateral malleolus, lateral femoral epicondyle, greater trochanter, iliac crest, acromion, lateral humerus epicondyle, lateral styloid process and fossa temporalis and at the center of the medial claviculars. Threedimensional positions of the markers were calculated using digitizing software (Frame-DIAS II, DKH). The estimated standard error of the direct linear transformation calculation was less than 1 cm in all three axes. These position data were used to calculate the knee angle (defined by the greater trochanter, lateral femoral epicondyle, and lateral malleolus) and ankle angle (defined by the lateral femoral epicondyle, lateral malleolus, and fifth metatarsal). 2.3. Data analysis Data were analyzed using custom-written Matlab code (The Mathworks, Inc., Natick, MA, USA). Kinematic data were low-pass-filtered (third-order Butterworth filter) at a cut-off frequency of 10 Hz and then interpolated and resampled at 1000 Hz, using a spline interpolation algorithm. For the control trials, foot contact (FC) was determined using a vertical ground reaction force (VGRF) threshold of 5 N. In the perturbed trial, the expected foot contact (EFC) was calculated when the vertical displacement of the right lateral malleolus fell 2 SDs below the mean of the control condition. Touch-down (TD) to the dropped surface was determined using a threshold of 5 N. For both the control and perturbed conditions, we defined midstance (MS) as the time when the anteroposterior ground reaction force changed sign from negative to positive. Take-off (TO) of the right foot was determined at a VGRF threshold of 5 N. EMG signals were high-pass-filtered at 5 Hz, full-wave
rectified, and low-pass-filtered using an 11-point boxcar moving average filter. The EMG response onset was defined as the time when the EMG activity of the perturbed trial exceeded those of the control trials by 2 SDs. EMG onsets and the duration between EFC and TD were compared using the one-way analysis of variance (ANOVA) and Tukey’s post-hoc test. To quantify the impact of FC in the control trials and TD in the perturbed trials, we calculated the impulse by integrating the VGRF during the 100-ms period after FC and TD, respectively. The integration of the rectified EMG signals (IEMG) during the 100-ms period after FC and TD was calculated as an index of ankle muscle activity after FC and TD, respectively. Pearson’s coefficient of correlation between the time from the onset of the SOL contraction to TD and the vertical impulse during the 100-ms period after TD was calculated. In the swing phase of the additional steps after the perturbation (9th and 10th steps in the perturbed condition), the maximum knee flexion angles of the left and right limbs, and the maximum ankle plantar-flexion of the left ankle were compared with those in the control condition. For the right ankle, angles at TO instead of the maximum value in the swing phase were compared between conditions because the entire swing phase of the right limb was not captured. The differences in the kinematic and temporal parameters between conditions were tested using the paired t test. An alpha level of 0.05 was considered significant for all statistics.
3. Results All participants performed nine control trials and one perturbed trial without any falls or injuries. In the perturbed trial, each subject lost his footing at the 8th step with his right leg. After touchdown to the dropped ground, all subjects took at least two additional steps (9th and 10th) and then stopped walking. Typical VGRF and EMG recordings from the lower leg muscles during the control and perturbed trials are shown in Fig. 2A. The VGRF and EMG activity traces were larger in the perturbed trial than in the control trial. The onset of EMG activity for the ankle plantar-flexor muscles (latencies from the EFC: MG, 97 16 ms and SOL, 117 23 ms) occurred before touchdown (128 10 ms); additionally, the onset of EMG activity for the dorsi-flexor (TA, 126 17 ms) occurred near TD (Fig. 2B). One-way ANOVA revealed significant differences among these four values [F(3, 24) = 4.71, p < 0.05]. A post-hoc Tukey’s test indicated that the EMG latency of the MG was significantly shorter than that of the TA and TD (p < 0.05). Greater EMG responses in both the ankle plantar-flexor and the dorsi-flexor were observed after TD (Table 1). The VGRF impulse during the 100-ms period after TD in the perturbed trial was larger than that after FC in the control trial. Some subjects started SOL activity very quickly, while others started after the TD (see error bar in Fig. 2B). This leads to the hypothesis that the inter-subject
M. Shinya et al. / Gait & Posture 29 (2009) 483–487
485
Fig. 2. Fast muscle responses to the unexpected loss of ground support. (A) Typical example of data recorded from a single subject in the control (gray) and the perturbed trial (black). The up arrows show the onset of muscle activity. (B) Mean and standard deviations of the latency associated with the electromyography response from EFC. The asterisk denotes a significant difference (p < 0.05) compared with the TA and TD. (C) Relationship between the onset of soleus activity and vertical impulse during the 100-ms period following TD in the perturbed trials. The time was aligned so that time 0 coincided with TD (vertical dotted line).
variation in SOL onset may explain variability in the TD impulse. Fig. 2C shows the relationship between the time from the SOL onset to TD and the vertical impulse in the 100-ms period after TD in the perturbed trials. The coefficient of correlation was 0.79 (p < 0.05), indicating that the faster the onset of SOL activity was, the smaller the TD impulse. Typical kinetic and kinematic data for the normal and perturbed gait are shown in Fig. 3. The unexpected loss of ground support delayed TD 128 10 ms more than EFC, while MS was not changed Table 1 Mean values of iEMG, kinetics, duration of walking phase, and joint angles during normal and perturbed walking. Variables iEMG MG (mV s) SOL (mV s) TA (mV s) Kinetics VGRF peak (%Weight) VGRF impulse (%Weight s) Duration of walking phase FC (TD) to MS (ms) MS to TO (ms) FC (TD) to TO (ms) Joint Angle L knee (8) R knee (8) L ankle (8) R ankle (8)
Control
Perturbed
P
15.5 11.6 4.2 1.3 8.2 7.4
37.3 22.7 24.5 5.8 63.9 29.3
<0.05* <0.001* <0.01*
109.4 5 .5 5.3 0.5
208.4 24.6 13.9 1.7
<0.001* <0.001*
(Table 1). Moreover, TO was also delayed in the perturbed trial, and the time interval between TD and TO in the perturbed trial was consistent with that between FC and TO in the control trial. After the perturbation, subjects continued walking with an alteration of the knee and ankle motions compared with normal walking. Both the left and right knees were more flexed, and the left ankle was more plantar-flexed during the swing phases of subsequent steps (Table 1). 4. Discussion Our data may be regarded as evidence of corrective postural responses to a completely unexpected loss of ground support during walking; the subjects were not warned of the possibility of the perturbation. Three sequential strategies for reestablishing balance were observed. (1) The reflexive muscle activities were evoked in the ankle plantar-flexors and dorsi-flexors. (2) The walking rhythm was reset during the perturbed stance phase. (3) The participants continued the adaptive locomotion to surmount the hole by flexing the right and left knees during the swing phase of subsequent steps before they stopped walking. 4.1. Reflexive muscle activity
*
373.5 23.3 335.7 29.3 709.2 26.5
219.0 43.5 457.7 65.4 676.7 62.1
<0.01 <0.01* 0.21
116 2.3 118 4.9 135 7.2 135 6.1
82 8.8 104 8.6 146 4.6 141 9.6
<0.001* <0001* <0.05* 0.18
iEMG and VGRF impulse were calculated during the 100-ms period after foot contact (FC) under the control condition and at touchdown (TD) under the perturbed condition. Joint angles are the maximum values during the swing phase, except in the case of the right ankle. The angle of the right ankle is the value at takeoff. * Significant difference between control and perturbed conditions.
Regarding the cutaneous input before the actual TD, the subjects touched only a thin cloth, which could not have evoked a muscle response. The observed EMG latency (MG, 97 ms) was also sufficiently short, indicating that the visual information did not trigger the EMG response [15]. Additionally, the somatosensory input associated with the actual TD could not possibly have served as the trigger because the EMG response was observed before the TD. Several previous studies have proposed that the difference between the actual and expected afferent inputs may evoke the reflexive muscle activity. In intact cats, the leg flexor muscles were rapidly activated to lift the paws soon after the foot entered a hole during walking [16,17]. These responses were absent in spinal cats,
486
M. Shinya et al. / Gait & Posture 29 (2009) 483–487
segments [13], but they might have contributed to absorbing the impact of the actual TD by stiffening the ankle joint [19]. In this study, the impulse after the TD decreased because the SOL activity was initiated earlier, suggesting that rapid contraction of the SOL might contribute to impulse absorption. Furthermore, simultaneous increases in muscle activity after TD were observed in all monitored muscles. Such muscle co-contraction might be a good strategy to cope with perturbations and to stabilize the ankle joint and the center of mass. 4.2. Phase reset of gait rhythm Resetting the gait rhythm is a good strategy to regain balance after various perturbations, such as stumbling during walking [4]. In-silico simulations also demonstrated the efficacy of resetting the gait rhythm in response to external impulsive perturbations to prevent falling [20,21]. In the present study, the latency between FC/EFC and MS was not altered by the perturbation, perhaps because the perturbed leg was nearly vertical at TD and consequently the forward horizontal and angular momentum could not be reduced [13]. This implies that the perturbation did not change the gait rhythm until MS. However, TO of the perturbed foot was delayed, and the duration of the stance phase was therefore the same as that in the control trial. The central nervous system reset the gait phase by delaying TO and maintaining the duration of the stance phase as a recovery strategy after the completely unexpected loss of ground support. 4.3. Adaptive locomotion to surmount the hole
Fig. 3. Typical examples of the ground reaction forces and the knee and ankle angles in the control (gray) and the perturbed (black) trial. The right leg is the support leg, while the left leg is the swing leg. Full knee extension and full ankle plantar-flexion occur at angles of 1808. The temporal events of the right leg are represented in the bottom of the graphs and vertical lines. For the control trial, FC: foot contact; MS: mid-stance; and TO: take-off. For the perturbed trial, EFC: expected foot contact; TD: actual touchdown; MS: mid-stance; and TO: take-off.
indicating that supraspinal structures contributed to the muscle reflexes. Humans also show reflexive muscle responses when encountering a difference in the support surface [12]. Since our results (ankle plantar-flexor contraction followed by dorsi-flexor contraction) were consistent with those of the previous study [12], the fastest response (MG) might be a muscle reflex evoked by the difference between the actual and expected somatosensory afferents via supraspinal structures. Larger muscle activities were also observed after the actual TD in both ankle plantar-flexors and dorsi-flexors. These muscle activities might be reflexes evoked by the afferent input accompanying the actual TD. It has been reported that, when one passes through a false floor, the stretch reflexes evoked from the impact with a lower, solid floor are twice as large as normal [18]. In the present study, the unexpected loss of ground support possibly enhanced the stretch reflex gain. How did the observed muscle responses contribute to postural control? It is conceivable that the braking synergy [14] was evoked by the completely unexpected loss of ground support in the present study, as van der Linden et al. [12] previously suggested. Given the electromechanical delay, the muscle activities before the actual TD did not change the positions or angles of the body
During the swing phase of subsequent steps, both the left and right knees were more flexed than during normal walking in the previous control trials. From the perspective of danger avoidance, it may be undesirable to keep one’s foot in a hole. It is well known that subjects will generally flex their knees to avoid obstacles during walking [22]. Thus, the knee flexion observed in our study might be a strategy for withdrawing the perturbed knee and securing toe clearance. Plantar-flexion of the left ankle during subsequent steps might be relevant for a cautious toe-first landing strategy instead of normal heel-first contact. Such adaptive strategies might be important to ensure security and to guarantee one’s ability to continue walking for several steps after the perturbation induced by the unexpected loss of ground support. Acknowledgements We would like to thank Dr. T. Kawabata for allowing us to use the EMG recording device. Conflict of interest We have no conflict of interests with other scientists or groups. References [1] Marigold DS, Patla AE. Strategies for dynamic stability during locomotion on a slippery surface: effects of prior experience and knowledge. J Neurophysiol 2002;88(1):339–53. [2] Tang PF, Woollacott MH, Chong RK. Control of reactive balance adjustments in perturbed human walking: roles of proximal and distal postural muscle activity. Exp Brain Res 1998;119(2):141–52. [3] Eng JJ, Winter DA, Patla AE. Strategies for recovery from a trip in early and late swing during human walking. Exp Brain Res 1994;102(2):339–49. [4] Schillings AM, van Wezel BM, Mulder T, Duysens J. Muscular responses and movement strategies during stumbling over obstacles. J Neurophysiol 2000;83(4):2093–102. [5] Dietz V, Horstmann GA, Berger W. Fast head tilt has only a minor effect on quick compensatory reactions during the regulation of stance and gait. Exp Brain Res 1988;73(3):470–6.
M. Shinya et al. / Gait & Posture 29 (2009) 483–487 [6] Nakazawa K, Kawashima N, Akai M, Yano H. On the reflex coactivation of ankle flexor and extensor muscles induced by a sudden drop of support surface during walking in humans. J Appl Physiol 2004;96(2):604–11. [7] Misiaszek JE. Early activation of arm and leg muscles following pulls to the waist during walking. Exp Brain Res 2003;151(3):318–29. [8] Dietz V, Colombo G, Muller R. Single joint perturbation during gait: neuronal control of movement trajectory. Exp Brain Res 2004;158(3):308–16. [9] Nieuwenhuijzen PH, Gruneberg C, Duysens J. Mechanically induced ankle inversion during human walking and jumping. J Neurosci Methods 2002; 117(2):133–40. [10] Marigold DS, Patla AE. Adapting locomotion to different surface compliances: neuromuscular responses and changes in movement dynamics. J Neurophysiol 2005;94(3):1733–50. [11] Berg WP, Alessio HM, Mills EM, Tong C. Circumstances and consequences of falls in independent community-dwelling older adults. Age Ageing 1997;26(4):261–8. [12] van der Linden MH, Marigold DS, Gabreels FJ, Duysens J. Muscle reflexes and synergies triggered by an unexpected support surface height during walking. J Neurophysiol 2007;97(5):3639–50. [13] van Dieen JH, Spanjaard M, Konemann R, Bron L, Pijnappels M. Balance control in stepping down expected and unexpected level changes. J Biomech 2007; 40(16):3641–9.
487
[14] Hase K, Stein RB. Analysis of rapid stopping during human walking. J Neurophysiol 1998;80(1):255–61. [15] Kandel ER, Schwartz JH, Jessell TM. Principles of Neural Science. Appleton & Lange; 2000. [16] Gorassini MA, Prochazka A, Hiebert GW, Gauthier MJ. Corrective responses to loss of ground support during walking. I. Intact cats. J Neurophysiol 1994;71(2):603–10. [17] Hiebert GW, Gorassini MA, Jiang W, Prochazka A, Pearson KG. Corrective responses to loss of ground support during walking. II. Comparison of intact and chronic spinal cats. J Neurophysiol 1994;71(2):611–22. [18] McDonagh MJ, Duncan A. Interaction of pre-programmed control and natural stretch reflexes in human landing movements. J Physiol 2002;544(Pt 3): 985–94. [19] Santello M. Review of motor control mechanisms underlying impact absorption from falls. Gait Posture 2005;21(1):85–94. [20] Yamasaki T, Nomura T, Sato S. Possible functional roles of phase resetting during walking. Biol Cybern 2003;88(6):468–96. [21] Nakanishi M, Nomura T, Sato S. Stumbling with optimal phase reset during gait can prevent a humanoid from falling. Biol Cybern 2006;95(5):503–15. [22] Lam T, Dietz V. Transfer of motor performance in an obstacle avoidance task to different walking conditions. J Neurophysiol 2004;92(4):2010–6.