Exp Brain Res (2003) 151:318���329 DOI 10.1007/s00221-003-1501-x R E S E A R C H A R T I C L E John E. Misiaszek Early activation of arm and leg muscles following pulls to the waist during walking Received: 18 December 2002 / Accepted: 9 April 2003 / Published online: 3 June 2003 # Springer-Verlag 2003 Abstract Many studies have investigated the compensa- tory reactions in humans elicited during walking when the support surface is perturbed. This has led to the descrip- tion of characteristic responses generated in the muscles of the legs and torso, and recently the arms. The present study aimed to investigate the compensatory reactions elicited when balance was challenged by a perturbation applied to the waist, to determine to what extent balance corrective responses are generalized across perturbation modalities. A second aim was to characterize the arm responses elicited by the perturbations applied to the waist. We measured muscle activity of the left arm and leg following application of backward pulls of the waist while the subjects walked on a motorized treadmill. This resulted in robust activation of tibialis anterior and vastus lateralis, with co-activation of soleus and biceps femoris also evident when perturbations were applied at heel strike. These early responses occurred with a distal to proximal temporal organization. The responses in the leg muscles displayed a phase-dependent modulation in amplitude, decreasing in amplitude later in the stance phase. Leg muscle responses were not evident during the swing phase, except for the end of swing, just prior to heel strike. Arm muscle responses were observed in all subjects however, the pattern of the arm responses varied considerably between subjects. Generally, shoulder mus- cles were more likely to respond than elbow muscles, at latencies consistent with the leg responses. Two important conclusions are drawn from the present study. First, the responses evoked in the legs with a pull to the waist are very similar to what has been reported for perturbations of the support surface, despite the very different locus of the perturbation. This suggests that balance control during walking may be achieved by preprogrammed reactions or synergies, which are triggered by multiple sensory cues. Second, rapid arm actions are integrated with these leg responses. However, the arm responses are more flexible, likely reflecting the fewer constraints imposed upon the actions of the arms, compared to the legs, during normal locomotion. Keywords Human �� Gait �� Balance �� Interlimb coordination �� Dynamic equilibrium Introduction Successful locomotion in humans requires moving the body in the desired direction while maintaining equilib- rium. In contrast to the gains made in the study of stepping control, far less research has been directed to understanding the neural control of balance during gait. Reactive balance control following external perturbations involves coordinated activation of musculature of the legs, torso, arms and neck. To date, a handful of studies have investigated the compensatory responses involved in correcting a slip-like perturbation to the feet during treadmill (Berger et al. 1984 Dietz et al. 2001) or overground (Marigold and Patla 2002 Tang et al. 1998 Tang and Woollacott 1999) locomotion, primarily report- ing on the reactions elicited in the musculature of the legs and to a lesser extent the torso. The general findings of these studies indicate that: (1) anterior-posterior support surface translations lead to a typical pattern of muscle activity, for example, ���slips��� applied at heel strike result in activation of tibialis anterior at short latency, (2) these patterns of muscle activity are phase dependent, and (3) there is usually a complex bilateral activation of leg muscles. However, these observations have been derived from a similar experimental paradigm and support surface translation. During natural locomotion balance can be challenged by several factors. Reactive balance control is required not only for perturbations applied to the feet, such as with a slip or trip, but also for perturbations applied to the mass of the body. To date, reactive balance J. E. Misiaszek ()) Department of Occupational Therapy and Centre for Neuroscience, University of Alberta, 2���64 Corbett Hall, Edmonton, Alberta, T6G 2G4, Canada e-mail: john.misiaszek@ualberta.ca Tel.: +1-780-4926042 Fax: +1-780-4924628

control during locomotion has been limited to the findings from perturbations applied to the feet. It remains unclear whether the findings from this one experimental paradigm would generalize to other perturbations of balance during locomotion. In a recent study, we demonstrated that perturbations applied to the waist of humans while walking on a treadmill generated a short-latency compensatory reaction in the leg muscles that was consistent with responses reported by others for support surface translations (Misi- aszek et al. 2000). This suggested that a similar corrective strategy was utilized to regain balance following both forms of balance perturbation. However, a more thorough description of the compensatory reactions elicited by a perturbation of the waist throughout the walking cycle is necessary to determine to what extent the observations arising from support surface translation studies can be used to describe balance reactions in general. In our previous study, we also noted that subjects typically produced movements of the arms following the backward perturbation at the waist (previously unreported observations). Tang and Woollacott (1999) reported exaggerated arm reactions in older adults, compared to younger adults, following support surface translations during locomotion. Marigold and Patla (2002) also reported an arm elevation strategy employed by subjects following slips. Therefore, it is evident that reactions in the arms form an important part of the corrective strategy evoked by balance disturbances during locomotion in humans. McIlroy and Maki (1995) demonstrated that arm reactions are elicited at a similar short latency to the leg reactions following support surface translation during standing. These authors suggested that a common balance recovery strategy generated the synchronous reactions elicited in the arms and legs. Recently, Dietz et al. (2001) demonstrated that early reactions in the muscles of the arms and legs followed rapid deceleration of a treadmill belt in walking subjects. In a later study, Marigold et al. (2003) also demonstrated that activation of arm muscles following an unexpected slip at heel strike occurred rapidly, with a similar early latency to the responses elicited in the leg muscles. In the present study, we investigated whether similar early reactions are elicited in the arm muscles during walking following backward pulls of the waist. Specifically, it was hypothesized that backward pulls applied to the waist throughout the step cycle would result in activation of the arm and leg muscles at similar latencies. Such a result would indicate that the responses elicited in the arm muscles are not related to the disturbance in the motion of the legs per se, as the perturbation would be initiated at the waist. Rather, it would suggest that stabilizing actions of the arms occur simultaneously with those of the legs regardless of the source of the instability. A portion of these results has been reported in an abstract (Misiaszek 2001). Materials and methods Experiments were performed using ten healthy young (21���28 years old) adult volunteers. The subjects ranged in height from 155 to 188 cm and ranged in weight from 50 to 98 kg. Subjects provided written consent of their participation. The procedures were approved by the University of Alberta Health Research Ethics Board. The subjects were equipped with a belt (a standard, padded, sport climbing harness) that fit snugly at the level of the iliac crest. A pair of cables was fastened to the belt and led behind the subject to a single handle, which pulled evenly on both cables. The cable was approximately 3 m in length. This helped ensure that the small changes in the position of the handle did not produce large changes in the angle of pull. In addition, the experimenter was able to observe the sagittal view of the subject and the setup throughout the experiment to ensure that the height of the handle remained relatively constant. Backward displacements of the waist were achieved by manually pulling the handle. A force transducer was placed in series with the cables to record the onset and force profile of the disturbance. Electromyographic (EMG) activity was record- ed from the soleus (SOL), tibialis anterior (TA), vastus lateralis (VL) and biceps femoris (BF) of the left leg and the anterior deltoid (AD), posterior deltoid (PD), biceps brachii (BB) and triceps brachii (TB) of the left arm. A pair of Ag/AgCl surface electrodes was placed over the bellies of each of the muscles, 2 cm apart and parallel to the predicted path of the muscle fibres. The raw signals were preamplified and band-pass filtered (30 Hz ��� 10 kHz) using Grass P511 preamplifiers (Astro-Med, Inc., West Warwick, R.I., USA). Electrogoniometers (Penny and Giles, Santa Monica, Calif., USA) placed across the ankle, knee and elbow recorded joint angles. In addition, reflective markers placed on the shoulder and wrist were recorded to videotape at 30 fps. Selected portions of the video were later digitized and the position of the markers for each frame selected using custom-written software. The position of the wrist relative to the shoulder was then estimated as a gross indicator of arm action. Force-sensitive resistors (Interlink Electronics, Camarillo, Calif., USA) placed in the shoes were used to record contact of the heel and toes. For two subjects, a uniaxial accelerometer (Model 1210 Analog Accelerometer, Silicon Designs Inc., Issaquah, Wash., USA) was affixed to the subject���s head. The accelerometer was placed on the head such that changes in anterior- posterior accelerations would be detected. This alignment also detected changes in acceleration associated with changes in the pitch of the head. All signals were digitized online at 1000 Hz and stored directly to hard disk using a custom-written LabView v.5 data acquisition routine and a National Instruments data acquisition card (PCI-MIO-16E-4, National Instruments, Austin, Tex., USA). Subsequently, the EMG signals were digitally full-wave rectified and low-pass filtered at 50 Hz (4th order Butterworth filter), while the force, accelerometer and electrogoniometer signals were digitally low-pass filtered at 25 Hz prior to data analysis. Protocol Subjects walked on a motorized treadmill at a self-selected pace. Subjects were permitted to walk on the treadmill prior to the start of the experiment until they were comfortable. The treadmill was equipped with safety rails to the front and sides of the subject, approximately 45 cm from the lateral edge of the arms of the subjects on either side and approximately 75 cm to the front of the subject. Transient force pulses, approximately 350 ms in duration and 20% body weight (BW) in peak magnitude, were periodically applied through the belt worn by the subject. The experimenter was provided with an oscilloscope record of the cable force to allow control of the force applied. Trials were subsequently screened off- line so that only those trials with a peak force within 5% of the target force and duration within 50 ms of the target duration were included in the analysis. In addition, post hoc analysis of the slope of the rise in the force profile revealed no significant differences in this attribute across bins (one-way repeated measures analysis of 319