Practice and transfer effects during fast single-joint elbow movements in individuals with Down syndrome.

Individuals with Down syndrome have been reported to exhibit various types of motor impairment.[1-6] They have delays in motor performance at a very young age and display atypical sequences of motor development.[6-9] They have longer reaction times[10-12] and sometimes utilize different strategies to control their movements[5,13] as compared with individuals who are neurologically normal.

In general, the movements of individuals with Down syndrome can be described as "clumsy," characteristics of which are slowness and low efficacy. Latash and Corcos[14] asked individuals with Down syndrome to perform unpracticed elbow flexion movements in a horizontal plane, under the instruction to move "as fast as possible." These authors reported that the movements of individuals with Down syndrome were typically slow and variable. Some movements in a series of trials demonstrated characteristic features of comparably slow movements of individuals who are neurologically normal.[15-17] The velocity profile of these movements was characterized by a bell-shaped curve, in which the movements were initiated by a burst of muscle activation in the agonist muscles (elbow flexors) and a delayed phasic burst of muscle activity in the antagonists (elbow extensors). In the same series, however, trajectories for movements were not smooth and included multiple bursts of activity in both agonist and antagonist muscles.

There are two views on the underlying causes of the deficits in motor performance in individuals with Down syndrome. According to the first view, basic organic abnormalities in the motor control system occur with Down syndrome.[18-20] These abnormalities may be related to the pathology or diminished volume of the cerebellum and other brain structures.[21,22] Therefore, one explanation for the "clumsiness" of the movements of individuals with Down syndrome may be an inability of these individuals to activate their muscles in a neurologically normal manner. Several studies have supported this view. For example, subjects with Down syndrome displayed a smaller magnitude of maximum torque and electromyographic (EMG) activity than individuals with other types of mental retardation and individuals who are neurologically normal.[20] Subjects with Down syndrome did very poorly when a time constraint was imposed on different tasks.[23] When encouraged to go fast, they merely pressed harder on a tap-pad or tracing surface.[18,23] They also could not modulate the rate of change of their grip force when asked to lift objects with different frictional surfaces. Instead, they prolonged the duration of the grip force.[24]

The alternative view suggests that all the apparent differences in the motor performance of individuals with Down syndrome and the general population are due to the suboptimal performance of an otherwise intact motor control system.[25] This later view suggests that there is room for improvement if adequate practice is provided.[26-29] On the basis of this optimistic view, we hypothesized that with training, individuals with Down syndrome would be able to increase the intensity with which they activate their motoneuron pools, generating more force and producing kinematic changes similar to those observed in individuals who are neurologically normal.[30,31]

The present series of experiments was designed to address these ideas. We studied the effects of prolonged practice of a simple motor task (a fast, unidirectional single-joint movement) on different indexes of motor performance in individuals with Down syndrome. The results support the view that extensive practice can significantly improve motor performance.

Method

Subjects

Four male and four female subjects with Down syndrome took part in the experiments. The chronological age and gender of each subject are presented in the Table. None of the eight subjects were institutionalized, and all were right-handed. Only one subject (S1) had a job. The subjects and their parents gave informed consent prior to participating in the experiments.

Table. Chronological Age and Gender of Subjects

Subject     Gender     Age(y)
S1          F          35
S2          F          16
S3          F          30
S4          F          19
S5          M          16
S6          M          20
S7          M          21
S8          M          15

Experimental Protocols

The subjects sat in a chair with their right forearms positioned on a low-friction horizontal manipulandum (moment of inertia=0.086 Nm[Xs.sup.2]/ rad), which was rigid and could rotate in a horizontal plane. The elbow joint was flexed 90 degrees (complete elbow extension=0 [degrees]) and aligned with the axis of rotation of the manipulandum. The shoulder was abducted 90 degrees in such a way that the subject could perform elbow flexion and extension movements in the horizontal plane. Then, the forearm was strapped in the manipulandum. A computer monitor, positioned in front of the subject, continually displayed a cursor. On the left side of the computer monitor, two sets of narrow bars specified the target size. The width of these bars corresponded to 6 degrees. Another bar on the right side of the monitor screen specified the initial position. The distance between the initial position and the target on the computer monitor was proportional to the distance the subject had to move.

During the pretest and the posttest, the subjects performed the following tests. The first test consisted of three trials of maximal isometric voluntary contractions in both flexion and extension under the instruction to push (or pull) "as strongly as possible." The elbow joint position was 90 degrees. The interval between the isometric trials was approximately 30 seconds. The second test was four series of isotonic (ie, without apparent change in constant external load) movements over 18, 36, 54, and 72 degrees from an initial elbow position of 55 degrees. The third test was two series of isotonic movements over 18 and 36 degrees; the initial position of the elbow was 73 degrees. During the second and third tests, the subjects were asked to move "as fast as possible." The last test was a series of isotonic movements over 36 degrees at a "comfortable speed" from an initial position of 55 degrees. Six trials were recorded in each series of isotonic movements.

Between the pretest and the posttest, subjects had 10 training sessions of isotonic movements, which were divided into 10 blocks of 11 trials each. These movements were performed at a 36-degree target distance from an initial position of 55 degrees. The total number of practice trials was 1,100. During the first day, subjects performed the pretest, followed by the first training session. Then, the subjects had two more training sessions per day for 4 days (days 2, 8, 9, and 15) after the pretest. Finally, on day 16, the subjects performed the last training session, followed by the posttest.

The use of any technical words in the instructions was avoided. The experimenter tried to use expressions familiar to the individuals by asking what they called different parts of the apparatus. Some examples of the instructions used were "Pull the armrest (manipulandum) toward you flexion direction) as strongly as you can" for isometric flexion and "Move as fast as possible but do not overshoot the red lines (target distance) too much" for isotonic movements. A strong and concomitant verbal reinforcement was given if the subject followed the instruction. For the set of isotonic movements, the reinforcement was based on the peak velocity of the movement performed by the subject. If the subject increased speed, the feedback took the form of encouragement: "Now you moved faster than before. This is fantastic!" However, if the subject started to play during the trials, he or she was asked to stop playing.

At the beginning of each trial, the subject was asked to relax his or her muscles and to move after hearing a computer-generated sound together with the experimenter's verbal command "GO!" During the experiment, the subjects were told that the purpose of the study was to teach them to move as quickly as possible and that they should not pay attention to accuracy. Reaction time also was not emphasized in these experiments.

Mechanical measurements. Elbow angle was measured by a capacitative transducer mounted on the axis of rotation of the manipulandum. Elbow acceleration was measured by a piezoresistive accelerometer, which was mounted 46.7 cm from the center of rotation at the distal end of the manipulandum. The accelerometer axis of maximal sensitivity was oriented to measure tangential acceleration. The acceleration, torque, and angle signals were digitized with 12-bit resolution at a rate of 1,000 Hz. Velocity was derived by integration of the acceleration signal after low-pass filtering at 25 Hz. The torque was measured by a strain-gauge transducer and filtered at 25 Hz.

Movement time was defined as the interval from the first acceleration deflection, which was visually determined (line A in Fig. 1), to the projected end of deceleration line C in Fig. 1). The projected end of deceleration was determined by linearly extrapolating the deceleration to 0, from a point where it had fallen to 50% of its negative peak.[32] Movement time was divided into two intervals: acceleration time and deceleration time. Acceleration time (see D in Fig. 1) was defined as the interval of time from the onset of acceleration until it first crossed 0 (see line B in Fig. 1). Deceleration time (see E in Fig. 1) was defined as the interval from the end of acceleration time to the projected end of deceleration.

We defined symmetry as the acceleration time divided by the deceleration time. Peak acceleration, peak deceleration, and peak velocity were defined as the maximum value of each of these variables. Finally, we defined the degree of overshooting as the maximum position achieved during the movement (Fig. 1) subtracted from the target position to which the subject was asked to move.

Electromyographic measurements. Electrocardiographic disposable pediatric electrodes (self-adhesive)(*) were placed over the bellies of one agonist (biceps brachii muscle) and one antagonist (lateral head of the triceps brachii muscle). These EMG signals were amplified (x 1,600) and bandpass filtered (60-500 Hz). Each signal was digitized at the rate of 1,000 Hz with 12-bit resolution. After that, it was full-wave rectified and smoothed by a 10-millisecond moving average filter. The first 800 milliseconds of the data from this signal was displayed on a computer screen, and the onsets of the agonist and antagonist bursts of activity were visually estimated for each trial. The onset of the agonist muscle burst was defined as the first time the EMG signal rose above the baseline (see line F in Fig. 1). All the trials were aligned for averaging according to this time.

The first trial of each series of isotonic movement was rejected. Because we compared movements performed at different distances, the trials in which the subject overshot or undershot the target by more than 10 degrees were rejected for purposes of analysis. We also rejected from further analysis the trials in which the onset of agonist or antagonist muscle burst of activity was ambiguous to identify. The total number of trials analyzed for each condition for each subject was about 4 ([plus or minus] 1) for the pretest and 5 ([plus or minus] 1) for the posttest.

For further quantification, we integrated the signal of the agonist muscle burst of activity of each movement over two time intervals, the first 30 milliseconds after onset of the agonist muscle burst of activity (see [Q.sub.30] in Fig. 1) and from the onset of the agonist muscle burst of activity to the first zero crossing of the acceleration (see [Q.sub.acc] in Fig. 1). The antagonist muscle burst of activity was integrated over the time interval from onset of the agonist muscle burst of activity to the projected end of deceleration (see [Q.sub.dec in Fig. 1). The value [Q.sub.30] is assumed to represent the intensity with which the motoneuron pools of muscles are activated and is related to the number and the size of motor units recruited. An increase in the value of [Q.sub.30] is used as a measure to correspond to a change in the slope of the agonist EMG trace because the integration time was constant (30 milliseconds). This method is analyzed in Appendix 1 of the article by Gottlieb and colleagues.[32] The value [Q.sub.acc] corresponds approximately to the first agonist EMG burst of activity.

Third, [Q.sub.30], [Q.sub.acc], and [Q.sub.dec] of each session were normalized by dividing them by the value of the EMG signal from the maximal voluntary contraction (MVC) measured from the same session, allowing their comparison across sessions. The EMG signal of the MVC was calculated from the trial in which the torque value was the largest. We integrated this EMG signal from the interval between 500 and 1,000 milliseconds after onset of the agonist muscle burst of activity. The normalized quantities [Q.sub.30], [Q.sub.acc], and [Q.sub.dec] are referred to in this article as [Q.sup.*.sub.30], [Q.sup.*.sub.acc], and [Q.sup*.sub.dec]. It was not possible to have a consistent measurement of torque from two subjects (S1 and S8) during the pretest; therefore, their EMG data ([Q.sub.30], [Q.sub.acc], and [Q.sub.dec]) were not normalized and were discarded from quantitative analysis.

The antagonist muscle activity started with a small component of activity, which usually began 20 to 30 milliseconds after onset of the agonist muscle burst of activity and continued until the onset of a larger, and abrupt, increase in activity. We defined the time of the onset of this sustained abrupt rise of the antagonist muscle burst of activity (see line G in Fig. 1) as the antagonist latency (see H in Fig. 1), and it was chosen because it varies with distance and is related to movement time.[32] The following procedure was used to help identify the onset of the antagonist muscle EMG burst of activity in each trial. First, the averaged burst of activity of the antagonist muscle was plotted for a set of trials, from 100 milliseconds before onset of the antagonist muscle burst of activity to 800 milliseconds after onset of the burst of activity. Subsequently, the onset of the antagonist muscle burst of activity was determined for this averaged record. This value was used to help visually estimate the onset of the antagonist muscle burst of activity in each individual trial.

Three subjects (S1, S4, and S8) had a pattern of co-contraction during the pretest. That is, they activated the agonist and the antagonist muscles simultaneously. At the posttest, these three subjects did not have this pattern of co-contraction. For each of these three subjects, however, we discarded antagonist latency data from quantitative analysis because these data were not identified at the pretest. The inclusion of the antagonist latency data for these three subjects, in a quantitative analysis, would lead to a larger change in latency than is reflected in the data set as a whole.

Experimental Design and

Statistical Analysis

Repeated-measures analyses of variance (ANOVAs) were used to analyze changes in peak velocity, peak acceleration and deceleration, acceleration and deceleration time, [Q.sup.*.sub.30], [Q.sup.*.sub.acc], [Q.sup.*.sub.dec], and antagonist latency. Confidence intervals were also used in the analysis of peak velocity, using a modified Bonferroni procedure. The overall error rate was set at 0.1.[33]

Results

Voluntary Changes in

Movement Speed

Before training, the subjects demonstrated the ability to move at different speeds. This ability was preserved at the posttest. In addition, movement speed increased dramatically. The data in Figure 2 depict angle, velocity, acceleration, and both the agonist (biceps brachii muscle) and antagonist (lateral head of the triceps brachii muscle) EMG signals for elbow flexion movements. The data are from a 36-degree movement performed by one subject "as fast as possible" (solid line) and "at a comfortable speed" (broken line). During the pretest, the EMG and kinematic traces of the movements of four of the eight subjects rose more steeply under the instruction to move "as fast as possible" as opposed to the instruction to move "at a comfortable speed." During the posttest, the slopes of the EMG and kinematic traces of the movements of all eight subjects rose more steeply under the instruction to move "as fast as possible."

Effect of Practice on Movement

Performance

Practice improved performance over all the movement distances, as shown in Figure 3 in which we present the data of the slowest subject (S8). The subject was asked to move "as fast as possible" over four different distances. The slopes of the kinematic and EMG profiles rose more sharply during the posttest than during the pretest. That is, there was an increase in these slopes with training, both at the practiced distance (36 [degrees]) and at the other distances. After training, peak velocity, peak acceleration, peak deceleration, agonist and antagonist muscle activation, and the slope of the agonist burst of activity were higher, with a concomitant decrease in movement time. These general EMG and kinematic patterns of improvement were observed for all eight subjects, including those subjects for whom the data of the EMG signals were not quantified.

Peak Velocity

With training, the subjects doubled their movement speed for the distances they practiced and even for those they did not practice. Figure 4A depicts peak velocity averaged across subjects for movements over four different distances. A two-way repeated-measures ANOVA with practice (prepractice compared with postpractice) and movement distance (18 [degrees], 36 [degrees], 54 [degrees], and 72 [degrees]) showed an improvement with practice (F=91.48; df=1,7; P=.0001) and distance (F=83.87; df=3,21; P=.0001). There was also a combined effect of practice and distance (F=13.212; df=3,21; P=.0001). To clarify the source of this interaction, we calculated the difference among the peak velocities for the pretest and posttest for all four distances, as well as confidence intervals. These data are presented in Figure 4B. The difference among the pretest and posttest means was different from zero for all four distances, but the difference was greater for the 36-, 54-, and 72-degree movements than for the 18-degree movements. The relative change from the pretest to the posttest was also calculated (ie, the difference between pretest and posttest was divided by the pretest value for all distances). These data are presented in Figure 4C and, as shown in the figure, this ratio remained almost constant, demonstrating that, after training, the subjects on average doubled their peak velocity for each movement distance.

Movement Time

As subjects improved their movement speed, they spent less time performing the movements. The acceleration profile of their movements remained symmetrical. The data in Figure 5 depict the averaged movement time for all eight subjects with Down syndrome. Movement time was divided into two components, acceleration time and deceleration time, which were analyzed separately. Figure 6 depicts the averaged data for both acceleration time and deceleration time for the pretest and posttest. A three-way repeated-measures ANOVA was performed to assess the effect of practice (pretest versus posttest), acceleration profile symmetry (acceleration time versus deceleration time), and distance. There was a decrease in movement time due to practice (F=19.96; df=1,7-1 P=.0029), acceleration and deceleration time were symmetrical (F=0.152; df=1,7; P=.7078), and there was an effect due to distance (F=18.91; df=3,21; P=.0001). Finally, none of the interactions were significant: practice versus acceleration symmetry (F=0.51; df=1,7; P=.4968), practice versus distance (F=1.62; df=3,21; P=.2156), acceleration symmetry versus distance (F=2.36; df=3,21; P=.1010), and practice versus acceleration symmetry versus distance (F=0.49; df=3,21; P=.6927).

Peak Acceleration and Peak

Deceleration

The increment in movement speed with training was also observed in a symmetrical increase of both peak acceleration and peak deceleration. The data in Figure 7 depict the average of both peak acceleration and peak deceleration at pretest and posttest, for all eight individuals with Down syndrome, across four distances. A three-way repeated-measures ANOVA was performed to assess the effect of practice on both peak acceleration and peak deceleration. The subjects performed their movements at higher peak acceleration and deceleration as a result of practice (F=27.53; df=1,7; P=.0012). The acceleration and deceleration peaks were symmetrical (F=0.20; df=1,7; P=.6661). There was a main effect due to distance (F=12.99; df=3,21; P=.0001). There was also an interaction between acceleration symmetry (peak acceleration versus peak deceleration) and distance (F=4.37; df=3,21; P=.0153) as well as between practice (pretest versus posttest) and distance (F=7.63; df=3,21; P=.0012). There was no interaction for acceleration symmetry versus practice (F=2.27; df=1,7; P=.1757) or for acceleration symmetry versus practice versus distance (F=0.90; df=3,21; P=.4520).

Overshoot

The improvement in motor performance with practice for the kinematic variables (ie, peak velocity, movement time, peak acceleration, and peak deceleration) did not result in less accuracy as measured by the degree of overshooting. Except for three trials in which one subject undershot by less then 2 degrees, the subjects overshot the target in all trials. The averaged overshoot across all subjects was 5.40, 5.25, 4.50, and 3.25 degrees, respectively, for the movement distances of 18, 36, 54, and 72 degrees. A two-way repeated-measures ANOVA with practice (pretest versus posttest) and movement distance (18 [degrees], 36 [degrees], 54 [degrees], and 72 [degrees]) showed a significant effect due to distance (F=3.45; df=3,21; P=.0352). Effects due to practice were not found (F=1.08; df=1,7; P=.3339), and there was no interaction between practice and distance (F=22.30; df=3,21; P=.8793).

Agonist and Antagonist

Electromyographic Signals

After training, the subjects increased the EMG quantity of both agonist and antagonist muscles. The data in Figure 8A are the averaged, normalized EMG signals of six individuals for the agonist ([Q.sup.*.sub.acc]) and antagonist ([Q.sup.*.sub.dec]) muscles. A two-way repeated-measures ANOVA was performed for both agonist ([Q.sup.*.sub.acc]) and antagonist ([Q.sup.*.sub.dec]) activity. For [Q.sup.*.subacc], there was an effect due to practice (F=13.22; df=1,5; P=.0150). There was also an effect for distance (F=9.34; df=3,15; P=.0010), but there was no interaction between practice (pretest versus posttest) and distance (F=2.37; df=3,15; P=.1388). For antagonist activity [Q.sup.*.sub.dec]), there was an effect due to practice (F=7.94; df=1,7; P=.0372), but no effect due to distance (F=1.17; df=3,15; P=.3532). There was no interaction between practice and distance (F=0.93; df=3,15; P=.4505).

Agonist Electromyographic

Slope

Individuals learned with training how to activate their muscles with an increased intensity, as shown by an increase in the slope of the initial component of the agonist EMG traces ([Q.sup.*.sub.30]). The data in Figure 8B depict the average of the normalized EMG quantity [Q.sup.*.sub.30] for six subjects at the pretest and at the posttest. A two-way repeated-measures ANOVA showed that with practice there was an increase in the slope of the agonist EMG traces (F=16.62; df=1,5; P=.0096). The effect due to distance was not significant (F=0.78; df=3,15; P=.4900), and the interaction between practice (prepractice versus postpractice) and distance was also not significant (F=0.57; df=3,15; P=.6422).

Antagonist Latency

Overall, the subjects learned how to activate their antagonist muscles earlier as they moved faster. Three out of eight subjects, however, displayed patterns of co-contraction, at the pretest, in which the EMG burst of activity of the antagonist muscle could not be separately identified. We therefore did not include these data in our statistical analysis. For the other five subjects, a two-way repeated-measures ANOVA showed an effect due to practice (F=37.62; df=1,4; P=.0036) and distance (F=20.05; df=3,12; P=.0001), and no interaction (F=3.112; df=3,12; =.0667). Because there is a strong relationship between movement time and antagonist latency in individuals who are neurologically normal, we have examined antagonist latency versus movement time at the pretest (broken fine in Fig. 9) and at the posttest (solid line in Fig. 9) for five individuals with Down syndrome. With practice, there was a decrease in both antagonist latency and movement time at all four distances.

Effect of Different Initial

Positions on Movement

Performance

The subjects were also able to transfer what they learned at one trained initial position to another at which they were not trained. A three-way repeated-measures ANOVA was performed on peak velocity to quantify the effect of different initial positions on the performance of practiced movements over different distances. The three factors were different initial position (55 [degrees] versus 73 [degrees]), practice (pretest versus posttest), and movement distance (18 [degrees] and 36 [degrees]). These data are depicted in Figure 10 and represent the averaged peak velocity for the eight individuals with Down syndrome. There was no effect due to initial position (F=0.631; df=1,7; P=.4530). There was, however, an effect due to both practice (F=50.414; df=1,7; P=.0002) and distance (F=30.525; df=1,7; P=.0009), and there was there was an interaction between practice and movement distance (F=22.68; df=1,7; P=.0021). There was no interaction between initial position and practice (F=1.351; df=1,7; P=.2832), but an interaction was found between initial position and distance (F=7.087; df=1,7; P=.0324). Finally, there was no effect of the interaction between practice, initial position, and distance (F=1.579; df=1,7; P=.2493).

Discussion and Conclusions

Practice Enhances Motor

Performance in Individuals

With Down Syndrome

All eight individuals with Down syndrome demonstrated substantial changes in performance as a result of practicing a simple elbow flexion task under reproducible conditions. The enhancement in motor performance showed that individuals with Down syndrome can activate their muscles in a neurologically normal manner and produce similar kinematic and myoelectric changes to those observed in individuals who are neurologically normal.[30,31] This was reflected by improvements in all kinematic variables associated with the movements.

After practice, subjects doubled their peak movement speed Fig. 4A), moving at 185.9 [degrees], 320 [degrees], 400 [degrees], and 430 [degrees]/S, respectively, for the 18-, 36-, 54-, and 72-degree distances. These peak velocities are approximately 30% below those previously reported for neurologically normal individuals who were trained using a similar experimental protocol.[30,31] The explanation for this slower velocity in performance may be due to individual differences among the subjects in the studies (ie, gender, age, amount of practice, life-style). The individuals who were neurologically normal were all male university graduate students, each of whom had 300 more practice trials than the subjects with Down syndrome in this study.

The improvement in peak velocity was highly correlated with a decrease in movement time (Fig. 5). A decrease in movement time with practice was reported in other studies of individuals with developmental disabilities[34,35] and specifically for those with Down syndrome.[28] Performance improvement, however, was much higher in our study (50% and more, Fig. 5) than in the study by Kerr and Blais,[28] in which the improvement in movement time ranged from 8% to 25% depending on the degree of practice and the amount of overshooting allowed in performing the movements.

The first explanation for the Large decrease in movement time is related to the nature of the task. In our experiment, the task (single-joint elbow flexion) was very simple and required little cognitive involvement, whereas the task in the study by Kerr and Blais[28] was very complex (discrete pursuit tracking) and placed great emphasis on cognition. It is reasonable that individuals with neurological impairments perform better on the simplest task. The second explanation relates to different levels of motivation offered to the subjects in both experiments. In our experiment, the strong encouragement to more faster may have played an important role in the enhancement of motor performance.

Improvement in the performance on one task, however, does not necessarily constitute learning. Schmidt[36] has pointed out that to be considered learned, the enhancement obtained during the practice of one task should be well generalized or transferred to variations on this task. This ability to transfer what is learned was observed in two task situations. First, the improvement in motor performance of all subjects with Down syndrome observed at the trained distance (36 [degrees]), as measured by the kinematic and EMG variables, was also observed at nontrained distances (18 [degrees], 54 [degrees], and 72 [degrees). After training, for example, the subjects were able to double the initial peak velocity of their movements across all the distances (Fig. 4C). Second, when asked to move from different initial positions, individuals with Down syndrome performed as well as they had done at the trained initial position Fig. 10).

We would also argue that to be considered a learned task, the transfer of the performance improvement should not occur at the expense of decreased accuracy. Besides performing at a high level and improving rapidly, individuals with Down syndrome also moved with remarkable accuracy. Kerr and Blais[28] reported that individuals with Down syndrome put more emphasis on accuracy than speed, and our results, before practice, support this view. We should note, however, that with appropriate training, individuals with Down syndrome can maintain accuracy and increase the speed of their movements. After training, the subjects in our study did not increase the degree of overshooting or the variability of peak velocity (Fig. 4A).

The Motor Control System

of individuals With Down

Syndrome is Functionally Intact

With training, the individuals with Down syndrome in this study were able to increase the intensity with which they activated their motoneuron pools, recruiting a larger number of motor units, generating more force, and moving faster. Because maximum inertial torque is proportional to peak acceleration, the subjects increased their level of inertial torque in an isotonic condition by 400% (Fig. 7). These findings do not support the idea that individuals with Down syndrome necessarily have a decreased ability to activate their motoneuron pools[20] or that they cannot generate appropriate levels of force.[20,24] Our findings support die view that individuals with Down syndrome can improve some aspects of their motor performance with practice of the task [26-29] or with therapeutic intervention.[37-39] The lack of opportunity to practice is, we believe, a major cause of the "clumsiness" and slowness of die movements of individuals with Down syndrome. Individuals with neurological impairments usuaUy receive suboptimal stimulation, have a history of failure, and are poorly reinforced to overcome their own limitations. We simply tried to make the individuals believe that they could exceed their own limits and move faster. We encouraged them to compete with themselves. When we said "Fantastic, you are doing a great job, but you can do better!" dley often answered in a loud voice with a big smile on their face "Yes, I'm gonna do it." The friendly and encouraging enviromnent we tried to create is difficult to describe in terms of experimental methods, even if it produces a great impact on the results.

Mechanisms Underlying Motor

Performance Enhancement

According to the dual-strategy hypothesis, [32] during the performance of fast single-joint movements, the central control signals to activate muscles (alpha-motoneuron pools) can be described as rectangular excitation pulses of fixed height (intensity) and duration. The motoneuron pools are assumed to perform low-pass filtering of the central control signals leading to experimentally observed EMG bursts of muscle activation, the development of joint torque, and eventually joint kinematics.

As the name "dual-strategy hypothesis" suggests, it includes two control strategies: speed-insensitive and speed-sensitive. The first strategy is used when there are no changes in externally or internally imposed constraints upon movement time, for example, when the subjects are asked to move "as fast as possible" over different distances. The control patterns for this strategy involve modulation of the duration of muscle EMG bursts of activity, not their intensity. These patterns of muscle activation lead to coinciding initial segments of agonist EMG and joint acceleration, and a delayed antagonist burst of activity as distance increases. This is exactly what we observed when the individuals with Down syndrome were asked to move "as fast as possible" over different distances (Fig. 3). The capability of some individuals with Down syndrome to modulate the duration in which they activate their muscles has been previously reported.[14] The additional finding in this report is that individuals with Down syndrome were also able to activate their antagonist muscles later for longer movements. This finding can be seen for the pretest in Figure 9 and is very similar to previously reported findings for who are neurologically normal.[32]

The speed-sensitive strategy is used when the subject is adjusting movement speed based on changes in externally or internally imposed con straints upon movement time, for example, when the subject is asked to move over a fixed distance at different speeds. These movements are controlled by excitation pulses in which height (intensity) is modulated, so that agonist EMG patterns and acceleration traces diverge at the very beginning of the movements. When asked to move at two different speeds over the same target distance, individuals with Down syndrome used the speed sensitive strategy to control their movements (Fig. 2). The kinematic and EMG slopes that accompany their movements diverged immediately after movement onset for both the pretest (four out of eight individuals) and the posttest (all eight individuals). The results of our study also suggest that the effect of extensive training was to enable the subjects to increase the level at which they activate their muscles beyond what was initially their maximum. On the basis of the size principle,[40] we would argue that with training the subjects learned how to increase the number or the size of motor units recruited. This could be observed by comparing the slope ([Q*.sub.30]) of the EMG traces between the pretest and the posttest (Fig. 8B). The values of [Q.*.sub.30] approximately doubled.

With training, the individuals with Down syndrome were also able, as predicted by the speed-sensitive strategy, to activate their antagonist muscle earlier (Fig. 9). The decrease in antagonist latency can have two potentially different consequences. This decrease can lead to an overlap in the agonist and antagonist activity, which is not the most energy-efficient mechanism for generating a movement. Minimizing energy expenditure or any other variable, however, may not be a part of the task demands of performing the most rapid movements possible. On the other hand, muscle coactivation can help increase joint stability, which could be important for the performance of rapid, accurate movements.[25] Decreased antagonist activation latency also results in earlier limb deceleration, thus reducing the time necessary to perform the movement.

Symmetry

Unlike neurologically normal individuals who decreased their movement time as a function of practice by proportionally shortening deceleration time more than acceleration time,[30,31] our subjects with Down syndrome did so by decreasing both acceleration and deceleration time in the same proportion. Approximately equal acceleration times and deceleration times were used by subjects with Down syndrome for die pretest and the posttest (Fig. 6) and are used by naive subjects before training.[30,31] The only exception was for the symmetric acceleration pattern for the longest distance (72[degrees]) at the pretest, which was less symmetrical than for the other distances. The symmetrical way in which our subjects performed their movements, even after training, was the only qualitative difference observed between subjects with Down syndrome and individuals who are neurologically normal. It remains to be determined whether this symmetry can be changed with specific training and whether changes in symmetry can enhance motor performance of individuals with Down syndrome beyond the level reported in this study.

Study Limitations

We are ruling out the possibility that the improvement in motor performance reported here may be due to a ceiling effect or to an improved comprehension of die task. We have shown elsewhere[4] that all eight individuals with Down syndrome in this study gradually improved their motor performance over training sessions. This improvement was best described by a typical learning curve. In addition, our task was very simple and the subjects understood the instructions well enough to move at different subject selected speeds even before training.

The implication that training can eliminate mate all deficits in the motor control mechanisms (ie, poor coordination and the inability to modulate grip force) of individuals with Down syndrome during the performance of more complex tasks is still waiting to be tested. We are not ruling out organic or psychological problems as the cause of some altered motor control mechanisms of individuals with Down syndrome reported in the literature. Our sample was not randomly selected, and there is a possibility that individuals with Down syndrome in other studies were considerably more handicapped in some aspects of their development than the subjects in our experiment. Our sample also was small, and the data reported should be considered preliminary.

We also did not include a matched control group, which limited between-subject comparisons. The problem with this procedure is the selection of attributes to be matched (eg, chronological age or mental age). The major point, however, is that comparative studies between the performance of individuals with Down syndrome and that of control subjects implicitly assume that the subjects come to the test with the same prior task-related experience. This assumption is not true, and prior experience may play an important role in the motor performance of individuals with Down syndrome.

Practical and Theoretical

Implications

The following practical implications can be derived from this study. First, individuals with Down syndrome should be provided with a stimulating environment in which they can learn how to perform motor skills. Second, comparisons between neurologically normal and neurologically impaired populations should take account of prior experience of both groups.[42,43] An unfair comparison can create misconceptions and produce biases in the way we teach motor skills.

One example of a misconception is the belief that an organic difference (eg, proportionately smaller cerebellar weight) underlies a behavioral deficit (eg, poor balance). The problem with this point of view [20-22] is that the correlation between organic dysfunction and behavioral deficit is not necessarily causal. The movements of individuals with Down syndrome did not present any of the signs of cerebellar dysfunction reported by patients with cerebellar deficits.[44] Individuals with Down syndrome did not exhibit movements that are characterized by prolonged acceleration time. Their movements were characterized by symmetry between acceleration time and deceleration time. Under the instruction to move "as fast as possible," individuals with Down syndrome did not seem to increase the duration of the agonist muscle activation (Fig. 2). Unlike patients with cerebellar dysfunction,[44] after training, individuals with Down syndrome were able to increase the intensity with which they activate their muscles, producing considerable agonist activity. As a consequence, they generated high levels of inertial torque, as measured by the increase in peak acceleration. Given our experimental conditions, the degree of overshooting of their movements did not present any characteristics of hypermetria.

Another major characteristic of patients with cerebellar lesions is a normal short-latency response, followed by a delay of long-latency reflexes.[45] Because of that, Shumway-Cook and Woollacott[5] suggested that the poor balance of individuals with Down syndrome could be due to cerebellar pathology. All eight individuals with Down syndrome in our study, however, showed EMG reactions in response to changes in instructions typical of preprogrammed reactions.[13] That is, their long-latency reflexes were at normal onset.

The danger of these misconceptions is not just the implications they produce at a theoretical level, but their influence at a practical level. If therapists and parents believe that because of an organic difference (eg, disproportionately lower cerebellar weight) individuals with Down syndrome will be unable to improve their motor performance, then there is no room for teaching and for therapy. Finally, because of the ability to improve their motor performance very quickly and at a high level, without losing initial accuracy, individuals with Down syndrome might be good candidates for several jobs that require these abilities. They may qualify for assembly-line jobs that require accuracy, speed, and repetition.

Acknowledgments

We thank the eight individuals who participated in this study and their families, and we thank Dr Jacques Lempers for his important suggestions during this work. We also acknowledge the help and support of Dr Gerald L Gottlieb.

References

[1.] Carr J. Six weeks to twenty-one years old: a longitudinal study of children with Down's svndrome and their families. J Child Psychol Psychiatry. 1989;30:187. [2.] Carr J. Mental and motor development in young mongol children. J Ment Defic Res. 1970;14:205-220. [3.] Cowie VA. A Study oftbe Early Development of Mongols. Oxford, England: Pergamon Press; 1970. [4.] Hartley XY. A summary of recent research into the development of children with Down's syndrome. J Ment Defic Res. 1986;30:1-14. [5.] Shumway-Cook A, Woollacott MH. Dynamics of postural control in the child with Down syndrome. Phys Ther. 1985;65:1315-1322. [6.] Haley SM. Postural reactions in infants with Down syndrome: relationships to motor milestone development and age. Phys Ther. 1986; 66:17-22. [7.] Harris SR. Down syndrome. In: Campbell SK, ed. Pediatric Neurologic Physical Therapy. Edinburgh, Scotland: Churchill Livingstone; 1984:169-204. [8.] Lydic JS, Steele C. Assessment of the quality of sitting and gait patterns in children with Down's syndrome. Phys Ther. 1979;59:1489-1494. [9.] Parker AW, Bronks R, Snyder CW Jr. Walking patterns in Down's syndrome. J Ment Defic Res. 1986;30:317-330. [10.] Anson JG. Down syndrome: neuromotor programming and fractionated reaction time. In: Latash ML, ed. Motor Control in Down Syndrome. Chicago, Ill: Latash; 1989:6-11. [11.] Berkson G. An analysis of reaction time in normal and mentally deficient young men: I, II, and III. J Ment Defic Res. 1960;4(RT):51-77. [12.] Lincoln AJ, Courchesne E, Filman BA, et al. Neuropsychological correlates of information processing by children with Down syndrome. Am J Ment Defic. 1985;89:403-414. [13.] Latash ML, Almeida GL, Corcos DM. Preprogrammed reactions in individuals with Down syndrome: the effect of instruction and predictability of the perturbation. Arch Phys Med Rehabil. 1993;74:391-399. [14.] Latash ML, Corcos DM. Kinematic and electromyographic characteristics of single-joint movements in Down syndrome individuals. AM J Ment Retard. 1991;96:189-201. [15.] Bouisset S, Lestienne F. The organization of a simple voluntary movement as analyzed from its kinematic properties. Brain Res. 1974; 71:451-458. [16.] Freund H, Budingen HJ. The relationship between speed and amplitude of the fastest voluntary contractions of human arm muscles. Exp Brain Res. 1978;31:1-12. [17.] Corcos DM, Gottlieb GL, Jaric S, et al. Organizing principles underlying motor skill acquisition. In: Winters J, Woo S, eds. Multiple Muscle Systems: Biomechanics and Movement Organization. New York, NY: Springer-Verlag New York Inc; 1990:251-267. [18.] Frith U, Frith CD. Specific motor disabilities in Down's syndrome. J Child Psychol Psychiatry. 1974;15:293-301. [19.] Parker AW, Bronks R. Gait of children with Down syndrome. Arch Phys Med Rehabil. 1982; 61:345-351. [20.] Davis WR, Sinning WE. Muscle stiffness in Down syndrome and other mentally handicapped subjects: a research note. Journal of Motor Behavior. 1987;19:130-134. [21.] Crome LC, Cowie V, Slater E. A statistical note on cerebellar and brain stem weight in mongolism.J Ment Defic Res. 1966;10:69-72. [22.] Woollacott MH, Shumway-Cook A. The development of postural and voluntary motor control system in Down's syndrome children. In: Wade MG, ed. Motor Skill Acquisition and tbe Mentally Handicapped Issues in Research and Training. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1986:45-71. [23.] Henderson SE, Morris J, Ray S. Performance of Down's syndrome and other retarded children on the Cratty Gross Motor Test. Am J Ment Defic. 1981;85:416-424. [24] Cole KJ, Abbs JH, Turner GS. Deficits in the production of grip force in Down Syndrome. Dev Med Child Neurol. 1988;30:752-758. [25.] Latash ML. Motor control in Down syndrome: the role of adaptation and practice. J Dev Phys Disab. 1992;4:227-261. [26.] Kanode JO, Payne VG. Effects of variable practice on retardation and motor schema development in Down syndrome subjects. Percept Mot Skills. 1989;69:211-218. [27.] Kerr R, Blais C. Down syndrome and extended practice of a complex motor task. Am J Ment Defic. 1987;90:313-318. [28.] Kerr R, Blais C. Directional probability information and Down syndrome: a training study. Am J Ment Retard. 1988;92:513-538. [29.] Youn G, Youn S. Influence of training and performance IQ on the psychomotor skill of Down syndrome persons. Percept Mot Skills. 1991;73:1191-1194. [30.] Corcos DM, Jaric S, Agarwal GC, et al. Principles for learning single joint movements, I: enhanced motor performance by practice. Exp Brain Res. 1993;94:499-513. [31.] Jaric S, Corcos DM, Agarwal GC, et al. Principles for learning single joint movements, II: generalizing a learned behavior. Exp Brain Res. 1993;94:514-521. [32.] Gottlieb GL, Corcos DM, Agarwal GC. Organizing principles for single joint movements I: a speed-insensitive strategy. J Neurophysiol. 1989;62:342-357. [33.] Shott S. Statistics for Health Professionals. Philadelphia, Pa: WB Saunders Co; 1990. [34.] Hoover JH, Wade MG, Newell KM. Training moderately and severely mentally retarded adults to improve reaction and movement times. Am J Ment Defic. 1981;85:389-395. [35.] Wade MG, Hoover JH, Newell KM. Training reaction and movement times of moderately and severely retarded persons in aiming movements. Am J Ment Defic. 1984;89:174-179. [36.] Schmidt RA. Motor Control and Learning: A Behavioral Empbasis. 2nd ed. Champaign, Ill: Human Kinetics Publishers Inc; 1988. [37.] Stratford B, Ching YY. Responses to music and movement in the development of children with Down's syndrome. J Ment Defic Res. 1989; 33:13-24. [38.] Edwards SJ, Yuen HK. An intervention program for a fraternal twin with Down syndrome. Am J Occup Ther. 1990;94:488-498. [39.] Purdy AH, Deitz JC, Harris SR. Efficacy of two treatment approaches to reduce tongue protrusion of children with Down syndrome. Dev Med Child Neurol. 1987;29:469-476. [40.] Henneman E, Somien G, Carpenter DO. Excitability and inhibility of motoneurons of different sizes. J Neurophysiol. 1965;28:599-620. [41.] Almeida GL. Practice, Transfer and Performance Enhancement of Fast Single-Joint Movements in Indiv:duals Witb Down Syndrome. Ames, lowa: lowa State University; 1993. Doctoral dissertation. [42.] Worringham CJ. Inferring motor deficits in Down syndrome: the role of practice. In: Latash ML, ed. Motor Control in Down Syndrome. Chicago, Ill: Latash; 1989:47-49. [43.] Newell KM. Down's syndrome and motor control: comments and notes. In: Latash ML, ed. Motor Control in Down Syndrome. Chicago, Ill: Latash; 1989:43-46. [44.] Hallett M, Berardelli A, Matherson J, et al. Physiological analysis of simple rapid movements in patients with cerebellar deficits. J Neurol Neurosurg Psychiatry. 1991;53:124-133. [45.] Nashner LM, Shumway-Cook A, Marin O. Stance posture control in select groups of children with cerebral palsy: deficits in sensory organization and muscular coordination. Exp Brain Res. 1983;49:393-409.

Invited Commentary

As more has been learned about the differences in neuroanatomical structures between individuals with Down syndrome and "neurologically normal" individuals, new questions have been raised about motor control and functioning. In particular, the identification of disproportionately lower cerebellar weights in individuals with Down syndrome has led to speculations that problems noted in balance, coordination, and muscle tone are of a cerebellar nature.[1-3] In particular, Shumway-Cook and WoollaCott[2] found that postural responses to loss of balance were slow in young children (1-6 years of age) with Down syndrome, and they concluded that these responses were inefficient for maintaining stability. They also stated that children with Down syndrome had poor spatiotemporal coupling between multiple muscle groups that act together. Furthermore, children with Down syndrome demonstrated inappropriate postural responses during rotational perturbation trials.[2]

Nashner[4] also found that abnormalities of stance balance control seen in individuals with Down syndrome were consistent with problems noted in individuals with cerebellar pathology. In a longitudinal study of children with Down syndrome,[3] my associates and I found deficits in the areas of visual motor coordination, bilateral coordination, balance, and response speed in individuals with Down syndrome as measured by the Bruininks-Oseretsky Test of Motor Proficiency.[5] Additionally, we had previously reported that children with Down syndrome performed poorly on measures of running speed, balance, strength, visual motor control, and overall gross motor and fine motor skills in comparison with children without Down syndrome but of comparable chronological and mental ages.[6]

Almeida et al are to be commended on examining yet another aspect of motor control in individuals with Down syndrome. As they stated, individuals with Down syndrome are often described as "clumsy" due to their decreased efficacy of movement resulting from either a pathology within the cerebellum or an inability to activate their muscles in a "neurologically norrnal" manner, or perhaps a combination of both of these factors. The authors have demonstrated that individuals with Down syndrome can activate their muscles in a normal manner for one particular task if adequate practice is provided.

There are several variables, however, that must be considered before the results of this study can be generalized to motor learning in individuals with Down syndrome. Although the authors discuss the possible effects of varying cognitive, academic, and functional levels on performance, they failed to identify the cognitive, academic, and functional levels of the individuals in the sample. There is also no information given regarding previous therapeutic interventions with these subjects. A more complete description of these characteristics would allow a better understanding of the overall functioning of the sample and would likely affect the interpretation of the data. These descriptors give essential information because individuals with Down syndrome demonstrate marked variability in cognitive and motor abilities.[3,7,8] This variability is thought to be related to a number of factors, including gene dosage, gender, muscle tone, and environmental effects.[8-10] Given the simple noninvasive task that was used and the relative brevity of the study, a larger sample size should have strengthened the application of the findings. These identified sources of variation in the abilities of individuals with Down syndrome could be taken into account with larger sample sizes, and, as the authors state, this study should be viewed as a pilot study.

Additionally, the authors discarded information during the study that should have been discussed. For example, the first trial of each series of isotonic movements was rejected as well as trials in which the agonist or antagonist onset of activity was ambiguous to identify. Analysis of this information may have led to a different interpretation of the findings. Furthermore, electromyographic data on two subjects were not quantitatively analyzed due to the presence of an inconsistent measurement of torque. Of interest would have been a discussion of why this inconsistency occurred in 25% of the sample.

Although the authors have presented information that seems to indicate that motor learning can take place with practice in individuals with Down syndrome, more meaningful clinical information would have been provided if a "real-life" task had been examined instead of a simple "pushpull" task. Patterns used for movement are specific to the task, and one cannot assume that the movement patterns used for one task will be used for another task.[11] Selection of a "real-life" task would have provided data on a functional task rather than an experimental task that presents itself infrequently in the day-to-day life of individuals. The suggestion by the authors that individuals with Down syndrome may qualify for assembly-line jobs is questionable because many assembly-line tasks involve response speed, body rotation, elbow flexion and extension, and visual motor coordination - all of which are identified problematic areas for individuals with Down syndrome.[2-4, 8, 12]

Furthermore, the authors have examined one aspect of cerebellar functioning in individuals with Down syndrome (ie, sequencing of) muscle contractions). Damage to the spinocerebellum results in an abnormal sequence, with agonist muscle activity prolonged and timing of antagonist contraction for limb deceleration delayed.[13] The authors did not examine cerebrocerebellar or vestibulocerebellar functions. Both lesion studies and cell recordings have documented that deficits in the cerebrocerebellum lead to delays in the initiation of movement, hypotonia, and motor incoordination.[13] Deficits in the vestibulocerebellum demonstrate a lack of postural stability.[13] The assumption by the authors that organic differences in individuals with Down syndrome do not affect other aspects of motor functioning such as balance, bilateral coordination, or visual motor coordination seems premature given that they examined only one area of cerebellar functioning. Perhaps the differences noted in the motor functioning seen in individuals with Down syndrome are due to dysfunction in differing areas of the cerebellum. Perhaps further studies will be forthcoming from the authors using more functional tasks in exarnining a variety of cerebeflar functions.

As a longtime advocate of early intervention programs for children with developmental disabilities, I certainly concur with the authors that individuals with Down syndrome should be provided with a stimulating environment in which they can leam to perform motor skills. My associates and I demonstrated through longitudinal studies that children with Down syndrome continue to progress in their motor skills in such an environment, although at a slower rate than the "neurologically normal" individual.[3]

[References]

[1.] Crome I. Pathology of Down's disease. In: Hilliard LT, Kirman BD, eds. Mental Deficiency. 2nd ed. Boston, Mass: little, Brown & Co Inc; 1965. [2.] Shumway-Cook A, Woollacott MH. Dynamics of postural control in the child with Down syndrome. Phys Ther. 1985;65:1315-1322. [3.] Connolly BH, Morgan SB, Russell FF, Fulliton WL. A longitudinal study of children with Down syndrome who experienced early intervention programming. Phys Ther. 1993;73:170-179. [4.] Nashner LM, Grimm RJ. Analysis of multiloop dyscontrols in standing cerebellar patients. In: Desmedt JE, ed. Cerebellar Motor Control in Man: Cerebral Event-Related Potentials (Programs in Clinical Neurophysiology). Basel, Switzerland: S Karger AG, Medical and Scientific Publishers; 1977;4:300-319. [5.] Bruininks RH. Bruininks-Oseretsky Test of Motor Proficiency Examiner's Manual. Circle Pines, Minn: American Guidance Service Inc; 1978. [6.] Connolly BH, Michael BT. Performance of retarded children, with and without Down syndrome, on the Bruininks-Oseretsky Test of Motor Proficiency. Phys Ther. 1986;66:344-348. [7.] Buckley S. Attaining basic educational skills: reading, writing and number. In: Lane D, Stratford B, eds. Current Approaches in Down's Syndrome. London, England: Holt, Rinehart & Winston; 1985:315-343. [8.] Carr J. Six weeks to twenty-one years old: a longitudinal study of children with Down's syndrome and their families. J Child Psychol Psychiatry. 1988;29:401-431. [9.] Reed RB, Pueschel SM, Schnell RM, Cronk CE. Interrelationships of biological, environmental and competency variables in young children with Down syndrome. Applied Research in Mental Retardation. 1980; 1:161-174. [10.] Kunit DM, Neve RL. Inborn errors of morphogenesis in Down syndrome. In: Pueschel SM, Tingey C, Rynders JE, et al, eds. New Perspectives on Down Syndrome. Baltimore, Md: Paul H Brookes Publishing Co Inc; 1987:81-91. [11.] Giuliani CA. Disorders in motor synergies, initiation, and termination of movement. In: Montgomery PC, Connolly BH, eds. Motor Control and Physical Therapy. Chattanooga, Tenn: Chattanooga Group Inc; 1991:111-119. [12.] Shea AM. Motor Development in Down Syndrome. Cambridge, Mass: Harvard Univertity; 1987. Dissertation. [13.] Newton RA. Neural systems underlying motor control. In: Montgomery PC, Connolly BH, eds. Motor Control and Pbysical Tberapy. Chattanooga, Tenn: Chattanooga Group Inc; 1991:31-44.