Movements executed with the intent to move as quickly as possible are called “ballistic” actions (18). A true ballistic action is preprogrammed in that once the central command for it has been formulated and dispatched to the motoneurons, there is no modification of motoneuron activation based on peripheral feedback. This is distinguished from slower “ramp” actions, which are characterized by feedback-based modification of motoneuronal discharge (7). Ballistic actions may also be initiated differently by the brain(10), are associated with a high frequency discharge of the motoneurons (6), and may involve preferential activation of high threshold, fast twitch motor U (9). Ballistic actions often exhibit concurrent agonist and antagonist activation(coactivation); at other times a triphasic activation pattern (in which an agonist burst of activity is followed by an antagonist burst, followed by a second agonist burst) is seen (5). Finally, when a ballistic movement is done on a background of tonic muscular activity, a brief reduction or cessation of agonist muscle activation may occur at the onset of the action (18). This is known as agonist premovement depression (20) or premovement silence(4), respectively, and may act to increase the velocity of the movement (18), possibly by bringing all motoneurons into a nonrefractory state immediately before phasic activation(13).
If the neuronal mechanisms responsible for and associated with ballistic actions could be amplified by practice or training, then enhanced ballistic performance might be expected in athletes who regularly perform these actions. Furthermore, electromyography might reveal some of the amplified neuronal mechanisms, such as more pronounced agonist muscle activation, an altered pattern of antagonist-agonist coactivation, or a greater occurrence of agonist premovement depression. In this study, ballistic performance and muscle activation were studied in highly trained karate athletes, whose training consists of performing ballistic type actions (punching, kicking(8,14)).
Subjects. Nine men, trained in Chito-Ryu karate for a minimum of 10 yr and having achieved a minimum karate rank of shodan (first degree black belt), formed the ballistic-trained group. They had a mean training experience of 16.3 ± 4.5 yr and included two shodan, one nidan (second degree), five sandan (third degree), and one godan (fifth degree) practitioners. The control group included 13 recreationally active men who had not undertaken any specific upper body training or whole body strength training. The physical characteristics of the two groups are shown in Table 1. All subjects participated with informed written consent, and the study carried the approval of the McMaster University Ethics Committee.
Apparatus. Subjects did isometric and ballistic elbow extension actions on a specially designed arm manipulandum (19). Briefly, this apparatus was constructed and instrumented with a torque transducer and potentiometer to record torque and displacement during isometric and ballistic elbow extension, while electromyographic (EMG) activity of elbow extensor and flexor muscles was recorded. Subjects sat on a seat with the upper arm to be tested flexed forward ≈90° (0° = arm at side) and supported on a horizontal plate extending to a point just proximal to the olecranon process. The supinated forearm was strapped to a separate support plate fixed directly to the steel rotational axis of a loading wheel, such that the elbow joint approximated this rotational axis. For the isometric actions and the start of the ballistic actions, the forearm was positioned vertically by setting its support plate perpendicular to the horizontal plane. This positioning resulted in an elbow joint angle of≈90° (180° = full extension). For the isometric actions, the plate was locked in this position; for the ballistic actions, it was free to rotate about its axis. Subjects placed the other arm in a relaxed fashion on their lap during all testing. On ballistic action trials, subjects extended the elbow fully from the initial position through a range of ≈90-100° before the back of the hand contacted a soft foam striking pad, the latter to prevent accidental excessive hyperextension of the elbow joint.
For ballistic movements against load, weights (equivalent to 10% of maximal voluntary isometric contraction peak torque) were attached to the loading wheel with cording and damped against oscillation with surgical tubing. Amplified torque sensor and displacement potentiometer signals, along with amplified (bandpass 3 Hz to 10 kHz) EMG signals were fed into a 12 bit A/D(analog to digital) converter (Dataq Electronics) and then into a microcomputer sampling at 1250 Hz and running CODAS data acquisition software(Dataq Electronics, Akron, OH).
Isometric maximal voluntary contraction (MVC). Isometric maximal voluntary contractions (MVC) were done first to determine isometric peak torque and also to set loads equal to 10% of this value for the ballistic actions. Three maximal isometric elbow extension actions were done. Subjects were allowed 1 min recovery periods between trials. After the determination of MVC peak torque (average of three trials), the arm manipulandum was released to allow free movement and the subjects were given ≈3-5 min rest before the next phase of the experiment.
Ballistic actions. From the MVC peak torque values, ballistic“load” values equal to 10% of isometric MVC (L10) peak torque were determined. Previous research (17) had indicated that a 10% MVC load induces an acceptable background agonist premovement activation level for the investigation of agonist premovement depression. The load was set with weights affixed to a rope attached to the loading wheel. When the load was set, the lock on the forearm support plate was released, subjects maintained a brief isometric phase to support the set load and then did the ballistic elbow extension actions. Thus, subjects were subjected to a“preload” condition in which an isometric action with a torque equal to the preload had to be maintained before the start of the ballistic action. Ballistic actions were done with L10 and with no added load (L0). It was recognized, however, that the ballistic actions with L0 and L10 were not only opposed by the inertia of the set loads, but also by that of the forearm, light support plate, and light alloy loading wheel. Thus, L0 refers to no weight attached to the rope. The start position for the ballistic actions was verified by the investigator via on-line potentiometer values before each ballistic action.
Ten maximal ballistic actions were done by each subject in each load condition. By random assignment, a subject did the L0 or L10 condition(trials) first. Subjects were given ≈30-45 s rest periods between trials, and ≈1-2 min rest periods between the load conditions. Whereas subjects were given a verbal “go” signal before each isometric MVC trial, ballistic performance was self paced. Subjects were told that the recording device was being activated but were instructed to extend their arms at maximal velocity on their own cue. Subjects were encouraged to “extend the arm as fast as possible” and to “contract explosively and ballistically” before each movement. Also, subjects were asked to leave their arms in the extended position after the action until requested by the investigator to return to the start position.
Electromyography (EMG). The skin over the biceps and triceps brachii of the appropriate arm was shaved, abraded with steel wool, cleaned with alcohol, and prepared with 5 EMG (bipolar configuration: two biceps, two triceps, one ground) electrodes (pediatric electrocardiography electrodes: Red Dot, 3 M). Agonist and antagonist activation (EMG) was recorded during MVC and ballistic actions.
Data analysis and statistical methods. MVC. MVC trials were analyzed for peak torque and maximum biceps and triceps integrated EMG(IEMG) and peak EMG (EMGpeak) (from full wave rectified and smoothed signal) activity.
Ballistic actions. Ballistic movement trials were averaged for each subject and experimental condition (L0 or L10) and analyzed for peak torque, peak rate of torque development, peak acceleration, peak velocity, movement time, and biceps and triceps EMG activity (IEMG and EMGpeak). EMG recordings were full-wave rectified and integrated (IEMG) and smoothed using a 25-point moving average function. EMGpeak was obtained directly from the full-wave rectified EMG signal, while the IEMG was subjected to further analysis. Ballistic and isometric IEMG values were divided by their activation durations to determine an average EMG (AEMG). The ballistic EMG values were also normalized to the isometric MVC maximum values for each subject. This normalization procedure was applied to agonist AEMG and EMGpeak. Antagonist AEMG was expressed as a percentage of the corresponding agonist AEMG. The torque and displacement recordings were filtered using a moving average function (10 points for torque, 30 for velocity; CODAS software, Dataq Electronics) and differentiated to provide values of instantaneous rate of torque development and peak velocity, respectively. The velocity recordings were again smoothed with a 30-point moving average function and differentiated to provide a signal for determination of instantaneous peak acceleration. Occasionally, analysis of ballistic trials would reveal a flexion countermovement immediately before the ballistic elbow extension. Also, in the L10 condition some subjects would sporadically do a slow ramp rather than a ballistic movement. Thus, 23 L0(10.4% of all trials) and 34 L10 trials (15.5% of all trials) showing these characteristics were discarded from the analysis.
Agonist EMG premovement depression (PMD). Ballistic L10 trials were analyzed (unrectified and rectified EMG) for PMD occurrence, duration, and movement potentiation. The occurrence of PMD was determined by visual inspection of recordings and was taken as a clear reduction in EMG amplitude before phasic activation. The term agonist premovement depression was applied instead of premovement silence (13) because the former more closely reflects the documented premovement changes in agonist activation, encompassing both complete quiescence and/or marked depression in the tonic agonist EMG immediately before the phasic ballistic discharge (seeFig. 1A).
Statistical analysis. One-factor (group) and 2-factor (group, load) ANOVA, as well as Student's t-tests, were used to analyze the data, with statistical significance set at P ≤ 0.05. Descriptive statistics include mean ± SD or SE.
Physical characteristics. There were no significant differences between groups in age, height, and biceps and triceps skinfolds. The karate subjects had greater body mass and arm girth than control subjects(Table 1).
Isometric peak torque. Control subjects produced a peak torque of 49.3 ± 7.9 N·m (mean ± SD), significantly smaller than the karate athletes' value of 65.2 ± 16.0 N·m (P < 0.005)
Ballistic action performance. A sample recording of EMG, torque, and angular displacement during a ballistic action with the 10% load is shown in Figure 1. Greater peak torque, rate of torque development, and movement time occurred with L10 than L0, whereas greater peak acceleration and velocity were produced at L0 (main effect for load,P < 0.001). Group comparisons are given below.
Peak torque. Karate athletes produced greater peak torque at both L0 and L10 (group main effect P < 0.001) (top panel ofFig. 2). There was a significant group × load interaction (P < 0.001), indicating that the absolute difference between the groups was greater at L10. The peak torque achieved at L0 and L10 was also expressed as a percentage of the peak isometric torque(normalized peak torque). The karate group attained a higher percentage at L10(P < 0.05) but not L0 (Fig. 2, bottom panel).
Rate of torque development (RTD). Karate athletes produced greater rate of torque development (RTD) at both L0 and L10 (main effect for group, P < 0.001) (top panel of Fig. 3). There was no group × load interaction (P = 0.065).
Peak acceleration. Karate athletes produced greater peak acceleration at both L0 and L10 (main effect for group, P < 0.03)(bottom panel of Fig. 3). There was a significant group× load interaction (P < 0.04), indicating that the absolute difference between groups was greater at L0.
Peak velocity and movement time. There were no group main effects or group × load interactions (Fig. 4).
Summary of ballistic performance. Group differences in ballistic performance variables are summarized in Fig. 5, which indicates the percentages by which the karate group's values exceeded the control group's values. The largest difference was in rate of torque development (RTD, ≈50%), followed by peak torque (PT) at L10 (≈40%). The group difference (not significant) in movement time is not shown in the figure; the karate group's values were 9 and 5% shorter than the control group's at L0 and L10, respectively.
Electromyography. Agonist activation. Agonist activation did not differ significantly between karate and control groups in isometric or ballistic actions; that is, there was no group difference in average integrated electromyographic activity (AEMG), or in peak electromyographic activity (EMGpeak) (Table 2). In correspondence with normalizing the ballistic action torque in relation to isometric values (bottom panel of Fig. 2), ballistic EMG values were also normalized to isometric values (ballistic/isometric). There were no significant differences between groups in normalized EMG(Table 3).
Agonist premovement depression (PMD). PMD occurred sporadically, being present in 10% of all possible trials in the karate group and in 5.4% of all possible trials in the control group. Three of nine karate athletes showed PMD, with an occurrence range of 20 to 50%, while four of 13 control subjects showed PMD with an occurrence range of 10 to 40%. The duration of PMD in the karate group was 57.8 ± 16.3 ms and 45.4 ± 12.9 ms in the control group (NS). Since agonist PMD occurred only sporadically, there were few trials on which to attempt a detailed analysis of any potentiating effects of PMD. PMD movement potentiation is most likely to affect initial force production via motor unit synchronization; therefore, three mechanical variables most sensitive to the initial ballistic discharge were selected for analysis. In a within-subject analysis, ballistic peak torque, peak RTD, and peak acceleration values observed in the presence of PMD were compared with those seen when PMD was absent. As the number of “PMD trials” was quite small, the value with PMD occurrence was considered to be significantly different at the P ≤ 0.05 level if it was more than 2 SDs larger than the nonPMD mean value. As shown in Table 4, some movement potentiation was observed, but only in the kinetic variables (mostly peak torque [PT]). No PMD movement potentiation was observed in the three karate athletes examined.
Antagonist coactivation. Antagonist coactivation determined from the AEMG values was greater in the karate group in isometric actions, but there were no group differences in ballistic actions (Fig. 6). The results for EMGpeak were the same (data not shown).
Karate athletes had superior ballistic performance, indicated by greater ballistic action peak torque, peak rate of torque development, and acceleration compared with untrained control subjects. However, if the“bottom line” in ballistic performance is peak velocity attained, the karate athletes were not superior to the control subjects in this measure. The small amounts by which the karate group exceeded the control group in peak velocity at L0 (6.3%) and L10 (3.3%) were not significant. The lack of statistical significance may have been a result of the small sample sizes as well as the small differences; however, given the respective reproducibility of the peak velocity measures, a significant 5% difference should have been able to be detected with the number of subjects tested(19). The apparent paradox between the karate group's superior peak torque and especially acceleration, but not superior peak velocity, is resolved by considering that velocity (v) is equal to the product of acceleration (a) and time (t); that is,v = a × t. Although acceleration(a) was 15 and 9% greater in the karate group at L0 and L10, respectively, movement time (t), although not significantly different between the two groups, was on average 9 and 5% shorter. For the given angular displacement (elbow joint range of movement), the greater acceleration (a) was offset by the shorter movement time(t), with the result that peak velocity was only slightly (6.3 and 3.3% at L0 and L10) and not significantly greater in the karate group. This consideration underscores the difficulty in improving peak velocity with training, even when training succeeds in improving peak torque and acceleration.
In comparing the performance of the two groups, it should also be recognized that subjects' ballistic tests were done with samerelative but different absolute loads. The karate athletes, whose isometric peak torque (MVC) on average exceeded the control group's by 32%, did the L10 test with a 32% greater absolute load. If the karate group had used the same absolute load as the control group, it would have represented a relative load of only 7.6% of the karate group's isometric peak torque (MVC). Given the inverse relation between relative load and velocity shown in the top panel of Fig. 4, the karate group would have been expected to produce a greater peak velocity with the control group's L10 (10% MVC load). This would be an advantage in certain aspects of karate performance.
The karate athletes' greater isometric strength (peak torque) was likely related to their greater body mass and especially upper arm circumference(Table 1). In view of the frequently observed high velocity specific training response (2), however, it was expected that the between group difference in peak torque would be greater for the ballistic than isometric actions. This expectation was fulfilled in that the karate group exceeded the control group to a greater extent in ballistic action peak torque at L10 (40.2%) than in isometric peak torque (32.3%)(Fig. 5). The expectation was also tested by expressing the ballistic action peak torque values as a percentage of the isometric values; i.e., normalized torque values ([L0 or L10 peak torque/isometric peak torquel] × 100). It was found that the karate group's normalized value was significantly larger for L10 (bottom panel of Fig. 2). Therefore, there was evidence of velocity/load specific superior performance in the karate group.
It might also be expected that the karate group would exceed the control group more (greater percentage difference) in L0 than L10 peak torque since karate punching actions seem to be most similar to the L0 test. (It should be noted that L0 was not a true “zero” load because the subject had to overcome in the inertia of the forearm, the light aluminium forearm support plate, and the light alloy wheel.) Contrary to this expectation, the karate group did not exceed the control group more in peak torque with L0 (30.3%) than isometric peak torque (32.3%) (Fig. 5). Furthermore, the between group difference in the normalized L0 peak torque was not significant (bottom panel of Fig. 2). However, it should be noted that with L0, the karate group's normalized peak torque of 8.2± 3.4 was 20.6% greater than the control group's value of 6.8 ± 1.8 (see also bottom panel of Fig. 2). In normalized peak torque with L10, the karate group's value of 43.6 ± 6.2 was 12.8% greater than the control group's value of 38.7 ± 4.4. Expressed in these terms, the between group difference appears to be greater at L0 (20.6% vs 12.8%). The lack of a significant group difference in L0 normalized peak torque may be partly explained by the greater intersubject variability in L0 values; the coefficients of variation ([SD/mean]×100) were 41.5 and 26.5% at L0 vs 14.2 and 11.4% at L10 for the karate and control groups, respectively.
The karate group's superior ballistic performance may have derived from neural and/or muscular adaptations to karate training. We tested for three possible neural adaptations: 1) increased agonist activation; 2) decreased antagonist coactivation; 3) more frequent occurrence of agonist premovement depression. No evidence of neural adaptation was found. The only significant difference between groups was the karate group's greater antagonist coactivation in the isometric actions. However, the normalized (to isometric) agonist activation was on average 43% and 25% greater (nonsignificant) in the karate group with L0 and L10, respectively (Table 3), corresponding to 21% (nonsignificant) and 13% greater (significant) normalized peak torque at L0 and L10 (Fig. 2). The lack of significance in the agonist measures, like the L0 normalized peak torque, was possibly a result of the larger intersubject variability in these measures(i.e., large coefficients of variation). Therefore, neural adaptation cannot be ruled out on the basis of our negative results. In addition to the aforementioned variability, the EMG measures used may have failed to detect possible neural adaptations such as a high motoneuron firing rate at the onset of ballistic actions (6) or increased motor neuron excitability (12).
The karate group's greater antagonist/agonist activation ratio in isometric actions would seem a counterproductive neural adaptation because increased antagonist (biceps) activation would reduce net extensor torque. A decrease in antagonist activation would appear more favorable, and this has occurred in the early phase of an isometric strength training program(3). On the other hand, antagonist coactivation may preserve joint integrity when large torques are generated, as would occur in strength trained athletes (1). Moreover, in karate, postures are often assumed with the goal of preventing perturbations inany direction. In this situation a strong degree of antagonist coactivation might be a beneficial neural adaptation to training. The karate and control groups had the same degree of antagonist coactivation in the ballistic actions; perhaps the benefits of increasing or decreasing coactivation were offsetting. In rapid reciprocal knee extensions and flexions, sprinters exhibit greater coactivation than distance runners(15). Thus, adaptations in coactivation may vary considerably depending on the motor task.
A novel aspect of the study was the measurement of agonist premovement depression (PMD). The occurrence of agonist PMD is highly correlated with high peak acceleration and rate of torque development(4,13,18); therefore, the experimental protocol provided optimal conditions for PMD occurrence. In fact, PMD occurrence was infrequent. The rate of occurrence in the karate (10% of trials) and control (5%) groups fell on the low end of the extremely wide range reported in the literature (11,16). Moderately trained karate athletes had an average PMD occurrence rate of≈27% of trials, but with the large intersubject variability(17) typical of this phenomenon(11).
PMD may operate by a supra-spinal neural switching mechanism, potentially amplifying ballistic performance by allowing for a near synchronous motor unit discharge (11,13). PMD occurrence may be a learned motor response (11). Volitional control over PMD can be acquired with specific biofeedback training (16). In this light, it is surprising that PMD occurrence did not differ between a very highly trained group habituated to ballistic movement performance, and an untrained, recreationally active control group. On the contrary, agonist PMD occurred so infrequently in both groups that determination of movement potentiation could be conducted in only a few subjects. This analysis revealed significant potentiation in peak torque (3 of 4 subjects) and peak RTD (2 of 4 subjects) in some control subjects. However, there was no significant movement potentiation in any of the karate subjects who showed agonist PMD.
In summary, karate athletes possessed superior ballistic performance. Within the limitations of the electromyographic methods used, the superior performance was not related to greater agonist activation, altered antagonist coactivation, or more frequent occurrence of agonist premovement depression.
1. Barrata, R., M. Solomonow, B. H. Zhou, D. Letson, R. Chuinard, and R. D'ambrosia. Muscular coactivation: the role of the antagonist musculature in maintaining knee stability. Am. J. Sports Med.
2. Behm, D. G. and D. G. Sale. Velocity
specificity of resistance training. Sports Med.
3. Carolan, B. and E. Cafarelli. Adaptations in coactivation after isometric resistance training. J. Appl. Physiol.
4. Conrad, B., R. Benecke, and M. Goehmann. Premovement silent period in fast movement initiation. Exp. Brain Res.
5. Cooke, J. D. and S. H. Brown. Movement-related phasic muscle activation. II. Generation and functional role of triphasic pattern.J. Neurophysiol.
6. Desmedt, J. E. and E. Godaux. Ballistic contractions in man: characteristic recruitment pattern of single motor units of the tibialis anterior muscle. J. Physiol.
7. Desmedt, J. E. and E. Godaux. Voluntary motor commands in human ballistic movements. Ann. Neurol.
8. Funakoshi, G. Karate-do Kyohan
(2nd Ed.). Kodansha Int.: Tokyo, 1990.
9. Grimby, L. and J. Hannerz. Firing rate and recruitment order of toe extensor motor units in different modes of voluntary contraction.J. Physiol.
10. Hamada, I. Correlation of monkey pyramidal tract neuron activity to movement velocity
in rapid wrist flexion movement. Brain Res.
11. Moritani, T. Neuromuscular adaptations during the acquisition of muscle strength, power and motor tasks. J. Biomech.
26(Suppl.1): 95-107, 1993.
12. Mortimer, J. A. and D. D. Webster. Dissociated changes of short-and long-latency myotatic responses prior to a brisk voluntary movement in normals, in karate experts, and in Parkinsonian patients. In:Advances in Neurology: Motor Control Mechanisms in health and disease
, Vol. 39, J. E. Desmedt (Ed.). New York: Raven, 1983, pp. 541-554.
13. Mortimer, J. A., P. Eisenberg, and S. S. Palmer. Premovement silence in agonist muscles preceding maximum efforts. Exp. Neurol.
14. Nakayama, M. Dynamic Karate
. Kodansha International: Tokyo, 1983.
15. Osternig, L. R., J. Hamill, J. Lander, and R. Robertson. Coactivation of sprinter and distance runner muscles in isokinetic exercise. Med. Sci. Sports Exerc.
16. Walter, C. B. Voluntary control of agonist premotor silence preceding limb movements of maximal effort. Percept. Mot. Skills
17. Zehr, E. P. and D. G. Sale. Ballistic elbow extension movements in moderately trained karate practitioners: peak torque
, and the agonist pre-movement silence. Med. Sci. Sports Exerc.
18. Zehr, E. P. and D. G. Sale. Ballistic movement: muscle activation and neuromuscular adaptation. Can. J. Appl. Physiol.
19. Zehr, E. P. and D. G. Sale. Reproducibility of ballistic movement. Med. Sci. Sports Exerc.
20. Zehr, E. P., D. G. Sale, and J. J. Dowling. Agonist EMG
PMD does not occur as a naturally acquired learned motor response in karate-trained subjects. Med. Sci. Sports Exerc.
Keywords:©1997The American College of Sports Medicine
TORQUE; VELOCITY; ACCELERATION; EMG