Exp Brain Res (2006) 168: 337–347 DOI 10.1007/s00221-005-0099-6

R ES E AR C H A RT I C L E

Jin-Hoon Park Æ George E. Stelmach

Effect of combined variation of force amplitude and rate of force development on the modulation characteristics of muscle activation during rapid isometric aiming force production Received: 13 August 2004 / Accepted: 24 May 2005 / Published online: 3 December 2005
Springer-Verlag 2005

Abstract Studies of rapid target-directed limb movements have suggested that various control schemes can be defined by the modulation pattern of the muscle activity. The present study was aimed to address the question regarding the extent to which a simultaneous control of force amplitude, and rate of force development influences the modulation characteristics of muscle activation associated with producing rapid isometric aiming forces at the elbow joint. The subjects were instructed to produce rapid isometric force pulses to three different force amplitudes (15, 35, and 55% of their maximal voluntary contractions) under systematically varied force-rate conditions ranging from a fast and accurate force-rate to the fastest force-rate possible. The results showed that larger force amplitudes were achieved by increasing the rate of force development (d F/d t) while the time to peak force remained relatively constant. The magnitude of the electromyographic (EMG) burst systematically increased as a function of force amplitude at all force-rate conditions. The primary finding was that the characteristic of the EMG burst duration associated with different force amplitudes showed a significant difference among forcerate conditions. Under a fast and accurate force-rate condition, the duration of the agonist burst increased linearly with force amplitude. A gradual transition into a fixed duration of the agonist burst then was observed over the remaining three force-rate requirements. With increasingly faster force-rates, there were no changes in the agonist burst duration over three force amplitudes. These results indicate that the combined variations in force amplitude and force-rate examined relative to the most rapid force-rate influence the control patterns for the muscle activation during the fast isometric force production. Changes in the EMG modulation patterns J.-H. Park Æ G. E. Stelmach (&) Motor Control Laboratory Department of Kinsiology, Arizona State University, P.O. Box 870404, Tempe, AZ 85287-0404, USA E-mail: [email protected] Tel.: +1-480-9659081 Fax: +1-480-9658108

observed are likely due to the constraints imposed by muscle contractile properties. Keywords Electromyogram Æ Isometric force Æ Strategies Æ Elbow Æ Human

Introduction The control of rapid target-directed limb movements has generally been demonstrated to consist of two components: an initial ballistic movement to rapidly drive the limb toward the target and corrective adjustments at the final phase of the movement to accurately position the limb at the desired position (Meyer et al. 1988; Woodworth 1899). More distinct characterization of rapid limb movements aimed to a target has been possible by the use of the electromyographic (EMG) signal from the muscle activity associated with the movement. Alternating triphasic activity of sets of opposing muscles that act in a push–pull manner has been commonly described as the pattern accompanying fast aiming movements. An initial burst of the agonist muscle is known to quickly accelerate the limb in the desired direction, followed by an antagonist burst to decelerate the movement and, depending on the nature of task, a second agonist burst to stabilize the limb in the target zone (Wachholder and Altenburger 1926). This type of phasic EMG pattern in the generation of fast limb movement has been reported to be necessary in order to counteract the intrinsic neuromuscular constraints such as slow contraction and relaxation properties of the muscle as well as time delays inherent in sensory feedback processes (Ghez and Gordon 1987; Partridge 1965). It has been suggested that the nervous system achieves various characteristics of movements (e.g., duration, speed, or load) by appropriately adjusting the activation intensity, duration, and timing of the individual EMG bursts of the triphasic pattern (Brown and Cooke 1981; Mustard and Lee 1987). Many investigators have utilized the notion of pulse-height/pulse-width

338

modulation of motoneuronal excitations applied to the agonist and antagonist muscles to identify the underlying rules used by the nervous system to modulate these EMG parameters with relation to variations in movement parameters. For example, using rapid single-joint movement paradigms, Gottlieb et al. (1989a, 1989b) proposed a ‘‘dual-strategy’’ control hypothesis based on the distinguished modulation pattern of these excitation pulses across a variety of movement conditions. They found that movements of different amplitudes or with various external loads, whose task constraints are not imposed by explicit control of speed, are achieved by varying the width of activation pulses. These are shown as prolonging the duration of the agonist burst at a relatively uniform rate of rise. This pattern of EMG activity has been termed a ‘‘speed-insensitive’’ (SI) strategy. When the intended kinematic actions are directly or indirectly related to constraints on speed or movement duration by task instructions or accuracy requirements (on the other hand, such aspects of the task are controlled by a constant duration of the agonist burst) the rate of EMG rise is modulated by the activation-pulse intensity. This response pattern has been referred to as a ‘‘speed-sensitive’’ (SS) strategy. Thus, the rate of EMG rise has been utilized to determine the control strategy (i.e., pulse height and pulse width). Furthermore, similar control patterns were demonstrated in the generation of isometric forces at the elbow, indicating that both movement and isometric force control may share common neural control principles over the modulation of the activation pulses to the muscles for a particular task (Corcos et al. 1990). Although Gottlieb and his colleagues have provided consistent organizing principles for the modulation patterns of muscle activation to control movements, the distinct pattern of muscle activity defined by the dualstrategy hypothesis has been challenged by many subsequent studies. For example, Hoffman and Strick (1990, 1993) observed that even though subjects performed the same classes of movements (e.g., movement type controlled by an SI strategy), the modulation patterns of the first agonist burst were changed depending on the range of the task examined. These authors used the peak amplitude and duration of the EMG burst as an indication of pulse-height and pulse-width modulation, respectively. Subjects varied the magnitude of the first agonist burst while keeping its duration approximately fixed (i.e., an SI modulation) for rapid wrist movements with short duration or light-weight load. But they progressively increased the duration of the agonist burst (i.e., a shift to an SS modulation) for the identical type of movements with longer duration or heavy load. These authors suggested that the distinct pattern of muscle activity observed is primarily due to the force characteristics specified by the task rather than changes in the control strategies determined by the implicit or explicit characteristics of speed constraints of the task. Similar results with the biomechanical constraints at different joints were found by Pfann et al. (1998).

It has also been suggested that the control patterns that the nervous system applies to modulate EMG parameters in relation to desired task characteristics are not simply specified by the dual-strategy control model when more than a single task parameter is employed together. In a study by Khan et al. (1999), for example, subjects performed rapid single-joint aiming movements of different target sizes that were opposed by inertial loads, thus requiring to control the two task parameter at the same time. The results showed that when the size of the target was large, inertial load had no effect on EMG slopes indicating the use of an SI control strategy. However, under the small target, EMG rise rates were varied as a function of inertial load reflecting the use of an SS strategy. Also, Monohar et al. (1998) examined the effects of accuracy (target size) and force amplitude on the control of rapid isometric force production. They observed that manipulating target size was associated with the SS control pattern when the target size of the percentage maximum voluntary contractions (MVC) was relatively small (e.g., 4 and 8%) during isometric pulse contractions. However, the modulation of muscle activity beyond a certain point in target size remained unchanged implying an upper limit to the dual-strategy hypothesis and the force level interacted with the size of the target. Taken together, these experimental data illustrated limitations in the rules for modulation patterns of muscle activation proposed by Gottlieb and colleagues when the tasks demand a simultaneous control of more than one parameter of the task. The purpose of this study was to examine changes in the modulation characteristics of the EMG bursts when two task parameters are concurrently controlled in an isometric task. Subjects were instructed to produce rapid isometric force pulses to different force amplitudes (15, 35, 55% of their maximum force) under various target force-rate conditions by systematically manipulating the time to peak force (ranging from submaximal to maximal speed) and therefore, required simultaneous adjustment of both force amplitude and the rate at which force increases. According to the dual-strategy model, force amplitude in isometric tasks is equivalent to the distance in movement. Hence, it is expected to be controlled by varying the EMG burst duration with rising along a relatively constant slope (SI strategy). However, it is hypothesized that this control pattern of muscle activation associated with scaling of force amplitude should be altered by the range of force rates examined relative to the most rapid force-rate. The range of force-rate required is near the level of maximal force-rate that one is capable of executing (e.g., fastest force-rate). Consequently, when the time to peak force approaches the shortest duration such as the one resulting from a brief twitch pulse, the scaling of force amplitude would be accomplished by modulating pulse height. In this process pulse duration is relatively kept constant because of a limited capacity of the motor control system to vary pulse duration. Recent studies support the hypothesis that there exists an ineffective

339

range of control for the pulse-width modulation. Pfann et al. (1998) showed that when the task demands a brief activation of the muscle to decrease muscle forces over short range of movement, the movement times approach the lower bound of muscle contraction. For instance, during a muscle twitch time, the nervous system produce such forces by switching to the pulse-height modulation control because further reduction of pulse width cannot be accomplished by duration modulation of the muscle (Gottlieb et al. 1996). The results of this study demonstrate that the modulation characteristics of muscle activation during the rapid force production of different force amplitudes are highly dependent on the instructed force-rate requirements of the task.

Materials and methods Subjects Ten neurologically normal individuals (five males and five females; mean age 22 years, range 19–26 years) participated in the experiment. They were right-hand dominant and naive to the experimental task and specific purposes of the study. This study was approved by the institutional review board of the Arizona State University. Informed consent was obtained from each subject before participation in the experiment.

Apparatus The subject was seated in a rigid chair facing a computer monitor. The height of chair was adjusted to maintain the right arm abducted to approximately 90 while the forearm was strapped in an immobilized aluminum manipulandum with the elbow flexed to 90 in a horizontal plane. The subject was instructed to grasp a vertical metal handle affixed near the end of the manupuladum with the wrist in a neutral position with regard to pronation and supination. The position of the handle was adjusted for each subject so that the elbow was centered over the axle of rotation. A strain gauge (Gamma F/T transducer, manufactured by ATI Industrial Automation Inc) positioned under the axis of elbow rotation measured the voluntary isometric force of elbow flexion. The targets and applied force were presented to the subject on a computer monitor positioned at eye level, 50 cm away. Electromyographic (EMG) surface electrodes (Bortec Biomedical Ltd, Canada) were applied over the bellies of the biceps brachii and the lateral head of triceps in parallel with the muscle fibers to record muscle activity. The pre-amplified EMG signals near the electrodes (·2 k) were further amplified between 500 and 2 k with a frequency response of 10– 1,000 Hz. The force and EMG data were recorded at a 500 Hz sampling rate and stored on a hard disk for later offline analysis.

Procedure After a few practice trials to familiarize the subjects with the device and tasks, MVC of elbow flexion and extension were performed for three trials each to compute the pre-determined submaximal target force amplitude: 15, 35, and 55% of the subject’s MVC. The target width was 4% of MVC (one reference line at 2% below and the other at 2% above each target force amplitude). Then, the subject was told to produce an isometric force pulse of elbow flexion to each target force amplitude in a fastaccurate manner without any attempt to correct their force output once initiated. A customized computer algorithm then generated a one-half-cycle sinusoidal waveform with time ranging from the initiation of force production to peak force. The program further scaled the waveform into three different slopes with higher rates by systematically reducing its duration, resulting in visual templates with four different rates of force developments for a given target force amplitude: the subject’s fast and accurate force-rate (F&A), two times faster than the F&A force-rate (1/2 F&A), four times faster than the F&A force-rate (1/4 F&A), and the fastest force-rate possible (Fastest). Then, this routine was repeated for the other target force amplitudes. Within a block of the fast and accurate force-rate condition, for example, the templates of the F&A force-rate generated for each target force amplitude were used. Thus, the templates for each target force amplitude within a force rate condition differed in force rise rates. During testing phase, as shown in Fig. 1, the subject was instructed to produce a single pulse force to a given target force amplitude while trying to match the slope of visual templates presented on the computer monitor without adjustment of force responses during the course of force development. After achieving the target force, they were told to passively return the force to the baseline. The experimenter monitored the subject’s performance of each trial. When the subject made an apparent correction of response during their attempts, the trial was repeated. These error trials usually occurred earlier in testing session, but the rate was less than 5% of their overall responses. The visual template and subjectproduced force trajectory were displayed in real time and remained for 1 s on the computer monitor after each experimental trial. The testing session consisted of four blocks of trials. Within each block, the force amplitude was randomized and the force-rate was blocked. The order of blocks was counterbalanced across subjects. Each subject performed 30 trials per block: 10 trials at each target force amplitude. A rest interval of 10 s was provided between trials and 1 min between blocks. Data analysis Signals from the force transducer were digitally conditioned with a fourth-order Butterworth filter having a

340

b

a

F&A

Force (N)

200

1/2 F&A

1/4 F&A

Fastest

Target force

150 100

Template

50 0 0

100 200 300 400 500 600

Time (ms)

Fig. 1 a Illustration of the screen display during a trial. The gray traces represent visual template used to show target force amplitude and rate of force development for the F&A force-rate condition. The black trace is an example of subject-produced force trajectory. b Visual templates with three different force-rates used for higher force-rate requirements

low-pass cutoff frequency of 50 Hz before trial-by-trial data analysis. For the MVC data, the highest force value over the trials was used as a reference for the determination of the target force amplitude. A customized computer algorithm searched the recorded EMG signal from each MVC trial and found a 100-ms interval (50 sampled points). The area of EMG amplitude over each 100-ms epoch was then divided by the integration time (100 ms) and the highest average EMG value was used for the normalization of the EMG amplitude for each muscle. The first-time derivative of force (d F/d t) was computed in each trial using the three-point-difference algorithm written in MATLAB M-files. The d F/d t trace was filtered with a second-order low-pass Butterworth filter with a cutoff frequency of 10 Hz to reduce noise. The maximum value of d F/d t trace was taken as peak d F/d t. The onset of force production was determined with reference to the peak value of d F/d t in each trial. From the point of peak d F/d t, a backwards search was performed to find the first point in the d F/d t trace that was 5% of peak d F/d t. From this point, if next values of d F/d t remained below 5% of peak d F/d t for at least 20 ms (ten sampled points), then the landmark was taken to indicate the onset of force production. Peak force was located by searching forward for the peak d F/d t to find the last point that was 5% of peak d F/d t dt before the values became negative. Time to peak force was defined as the time from the onset until peak force. The trials with more than 2 SD of force overshoot mean were not included in further analysis, which accounted for three trials across all subjects. The digitized EMG signal was full-wave rectified and filtered with fourth-order high-pass Butterworth filter with a 5-Hz cutoff frequency, and then smoothed with a 10-ms moving-average window. We generally followed the methods used by Gottlieb et al. (1989a) for the determination of EMG variables. Onset and end of the EMG burst were defined using a computer algorithm that scanned each sampled EMG data to find the point that was greater than a threshold (three times the mean of the baseline EMG activity calculated during the rest interval). If a point exceeded the threshold and the next ten sampled EMG values (20 ms) were sustained beyond

5% of peak EMG amplitude, then it was taken to represent the onset of the EMG burst. The reason for adopting this algorithm to locate the onset of the EMG burst was to identify the time point at the beginning of the main component of the EMG bursts. The end of EMG burst was identified by the point that was less than the threshold and sustained below 5% of peak magnitude of the EMG burst at least for 20 ms. The duration was measured as the time period from the onset to the termination of the EMG burst. In addition, a visual inspection was performed to ensure the accurate location of these points for each trial and correction was made when needed. The initial rate of the EMG rise (Q30) was calculated by the integral of the area over the first 30 ms relative to onset of the EMG burst. Before quantifying the EMG signal, all EMG amplitudes were normalized with respect to the peak EMG amplitude obtained from MVC trials for each muscle. The first trial from each block was rejected from further analysis. To examine the effects of force amplitude and forcerate, a 3 (15, 35, and 55% of MVC) · 4 (F&A, 1/2 F&A, 1/4 F&A, and Fastest) analysis of variance (ANOVA) with repeated measures on both factors was performed. Post hoc comparisons between means were evaluated using Duncan’s multiple-range test. A simple main-effects analysis was performed when there was a significant interaction effect. An alpha level of P