Neuroscience 164 (2009) 822– 831

EFFECTIVE UTILIZATION OF GRAVITY DURING ARM DOWNSWING IN KEYSTROKES BY EXPERT PIANISTS S. FURUYA,a,c* R. OSUb AND H. KINOSHITAa

because of the motion of the linked segments) are utilized to facilitate the intended movement at the distal joints (e.g. Dounskaia, 2005; Furuya and Kinoshita, 2007, 2008b; Hirashima et al., 2007; Putnam, 1993). Furuya and Kinoshita (2008b) have shown that during piano keystrokes, muscular torques that produce elbow extension and wrist flexion are smaller and interaction torques at these joints are greater among highly trained pianists compared with novice piano players. They have also demonstrated that experts, but not novices, are able to maintain repetitive keystrokes at a designated level of loudness for more than 30 min. These findings suggest that the effective use of motion-dependent interaction torques could facilitate physiological efficiency of upper-limb movements during piano performance. Gravity is another non-muscular force that inevitably affects our bodily movements and has been studied in the context of its compensation in vertical arm movements (Atkeson and Hollerbach, 1985; Papaxanthis et al., 2003) as well as in trunk and leg movements during walking (White et al., 2008) and sit-up (Cordo et al., 2006). In contrast, less is understood about the active use of gravity during natural upper-limb movement. Dennerlein and colleagues have studied the use of gravity when professional typists strike keys (Dennerlein et al., 1998; Kuo et al., 2006). They found that the intrinsic and extrinsic muscles of the finger flexors were not activated prior to the production of flexion torque at the metacarpophalangeal (MP) joint. They therefore concluded that there is a free-fall drop of the finger to complement finger-flexion muscular force during the initial phase of skilled typing. By contrast, studies of vertical arm-reaching actions in ordinary individuals demonstrated that descending arm motion was commonly preceded by a clear burst of agonist muscular activity at the shoulder joint, indicating the limited use of gravity in this population (Berret et al., 2008; Papaxanthis et al., 2003). These findings thus suggest that the effective use of gravity could be specific to motor control in skilled vertical limb movement. The use of gravity to reduce the physiological cost of muscular work while striking piano keys, a technique called “weight playing,” has long been a serious concern of players and educators worldwide (Hmelnitsky and Nettheim, 1987; Matthay, 1905). In 1930, Bernstein and Popova attempted to determine its feasibility by computing torques due to muscles and gravity at the upper limb joints when expert pianists struck a key at varied tempi (Kay et al., 2003). They found that at fairly slow tempi, the amount of decrease in elbow flexion muscular torque was near the value of gravitational torque acting at the elbow joint. Be-

a

Graduate School of Medicine, Osaka University, Toyonaka, Osaka 5600043, Japan b

ATR Computational Neuroscience Laboratories, Kyoto 6190288, Japan c Research Center for Human Media, Kwansei Gakuin University, Sanda, Hyogo 6691337, Japan

Abstract—The present study investigated a skill-level-dependent interaction between gravity and muscular force when striking piano keys. Kinetic analysis of the arm during the downswing motion performed by expert and novice piano players was made using an inverse dynamic technique. The corresponding activities of the elbow agonist and antagonist muscles were simultaneously recorded using electromyography (EMG). Muscular torque at the elbow joint was computed while excluding the effects of gravitational and motion-dependent interaction torques. During descending the forearm to strike the keys, the experts kept the activation of the triceps (movement agonist) muscle close to the resting level, and decreased anti-gravity activity of the biceps muscle across all loudness levels. This suggested that elbow extension torque was produced by gravity without the contribution of agonist muscular work. For the novices, on the other hand, a distinct activity in the triceps muscle appeared during the middle of the downswing, and its amount and duration were increased with increasing loudness. Therefore, for the novices, agonist muscular force was the predominant contributor to the acceleration of elbow extension during the downswing. We concluded that a balance shift from muscular force dependency to gravity dependency for the generation of a target joint torque occurs with long-term piano training. This shift would support the notion of non-muscular force utilization for improving physiological efficiency of limb movement with respect to the effective use of gravity. © 2009 IBRO. Published by Elsevier Ltd. All rights reserved. Key words: motor control, non-muscular force, physiological efficiency, musicians, pianists.

Nearly half a century ago, Bernstein (1967) proposed that a key feature of highly skilled motor action is the effective use of non-muscular forces to reduce the work of muscles. Recently, researchers have successfully demonstrated that in multi-joint limb movements, motion-dependent interaction torques (rotational forces that arise at one joint *Correspondence to: S. Furuya, Graduate School of Medicine, Osaka University, Health and Sports Science Building, 1-17 Machikaneyamachou, Toyonaka, Osaka 5600043, Japan. Tel: ⫹81-6-6850-6034; fax: ⫹81-6-6850-6030. E-mail address: [email protected] (S. Furuya). Abbreviations: ANOVA, analysis of variance; EMG, electromyography; f, forte; MP, metacarpophalangeal; MVC, maximum voluntary contraction; p, piano; SD, standard deviations; SPLs, sound pressure levels.

0306-4522/09 $ - see front matter © 2009 IBRO. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.neuroscience.2009.08.024

822

S. Furuya et al. / Neuroscience 164 (2009) 822– 831

cause tendons, muscles and connective tissues of the elbow joint produce small resistive forces against gravity, a muscular torque value near the gravitational torque value suggests that the pianists were dropping their arms without the use of extension muscular torque at slow tempi. One problem with these findings was that the computed muscular torques did not exclude interaction torques. Our recent study demonstrated that interaction torques have a substantial effect on the elbow rotation produced in piano keystrokes performed by expert pianists (Furuya and Kinoshita, 2008b). In addition, the decrease in muscular torques during arm downswing does not exclusively represent the use of gravity because it can also occur by contracting the agonist muscles. Thus, monitoring the agonist muscular activity of the target joint movement (as described by Dennerlein et al., 1998) is essential. The purpose of the present study was to investigate the skill-dependent gravity-muscular force interaction that occurs at the elbow joint during natural downward swing motion of the arm. To this aim, we compared the kinematics and kinetics of the preparatory lift and downward movement of the arm, as well as their associated muscular activities during piano keystrokes in expert and novice piano players. Kinetic analysis was made by the use of a four-link upper limb segmental model in order to accurately estimate the elbow joint muscular torque by excluding the effect of both interaction torque arising from the surrounding joints and gravitational torque (Furuya and Kinoshita, 2008b). We hypothesize that expert players will make a greater use of gravity with less reliance on the agonist (the triceps) muscular work, along with development of elbow extension muscular torque to generate the arm-descending phase. Therefore, there may be a longer duration of the phase of producing extension muscular torque without apparent triceps activity. Novices, on the other hand, will rely heavily on active agonist muscular work to develop their elbow extension muscular torque while utilizing gravity to a lesser extent. We also hypothesize that a greater use of gravity during stronger keystrokes to produce louder sound, and therefore a greater level of skill-dependent gravity-muscular force interaction may occur when generating a louder sound than when producing a weaker sound.

EXPERIMENTAL PROCEDURES Participants Seven active expert pianists (three males and four females, mean age⫾SD⫽24.3⫾3.2 years) with more than 15 years of classical piano training and seven novice piano players (three males and four females, age⫽21.0⫾4.6 years) with less than 1 year of piano training participated in the present study. All of the expert pianists had won awards at domestic and/or international classical piano competitions. In order to exclude the possibility that any differences in the keystroke motions across pianists could be attributed to a technique that had been intensively taught at a certain piano school, participants who had been taught to play the piano by different instructors were selected. The novice participants were selected from those who had hand size large enough to be able to easily perform octave keystrokes. The mean value of the hand span (the distance from the tip of the thumb to the tip of the little finger in a stretched hand) for the novice participants was 195⫾15 mm, which did not differ from 194⫾9 mm for the experts (F(1,6)⫽0.024;

823

P⫽0.88). These values were also clearly larger than the center-tocenter distance (164 mm) of the octave keys. All participants were right-handed, as determined by the Edinburgh MRC Handedness Inventory (Oldfield, 1971). We paid our best efforts for minimizing the participant’s suffering and discomfort during the experiment. In accordance with the declaration of Helsinki, the experimental procedure was explained to all participants and each participant signed a written informed consent. The study was approved by the local ethics committee at Osaka University.

Experimental apparatus and key-striking task The experimental apparatus consisted of a Yamaha U1 upright piano, an 8-channel telemetric electromyography (EMG) system (Nihon Koden Co., WEB-5000), two 2-D position sensor systems (C5949, Hamamatsu Photonics Co., Japan), a sound-level meter (NA-27, Rion Co., Japan), and a stereo sound amplifier. In the G3-key, a strain-gauge miniature uniaxial force transducer was installed at its distal end for accurately estimating joint torque during key depression (see the “Appendix”). The resolution of the transducer was 0.02 N, and the natural frequency of the unloaded force transducer was 1 kHz DC. The force signal was amplified using a strain gauge amplifier (DMP602B, Kyowa Co., Japan). The sound-level meter was placed 1 m above the keyboard and collected sound signals at a sampling frequency of 900 Hz. The experimental task was a right-hand octave keystroke, a simultaneous strike of the 35th (G3) key by the thumb and the 47th (G4) key by the little finger. The keys were 164 mm apart. This movement task was chosen to induce a whole arm movement, as well as to minimize the mediolateral and pronation–supination movements of the hand and arm during the keystroke to permit a 2-D kinematic analysis (Furuya and Kinoshita, 2008b). In the experiment, the participant began by lightly touching the fingertips of the right hand on the keys at the initial position, lifting his/her right arm/hand to a self-determined height at a self-determined speed, striking the keys in a short tone production (a “staccato” touch) at a designated level of tone, lifting the hand and arm again as a followthrough to a self-determined height, and returning to the initial position. This sequential motor action is defined as a “keystroke trial”. The left arm and hand were kept relaxed and placed on the side of the body while the trunk was in an upright position with minimum movement. Based on our previous study, four target sound pressure levels (SPLs) of 103, 106.5, 110, and 113.5 dB were chosen in this study, which roughly corresponded to the loudness of piano (p), mezzo–piano (mp), mezzo forte (mf), and forte (f), respectively (Furuya and Kinoshita, 2008b). These relatively high levels of SPL were due to high background noise coming from the computers and amplifiers, and the use of an ordinary nonsound proof experimental room (Kinoshita et al., 2007). For each participant, kinematic, EMG, and simultaneous sound data were collected from thirty successful keystroke trials at each of the target SPLs with an approximately 10 s trial-to-trial interval. The target SPL was a piano sound pre-recorded on a minidisk, which was presented from a set of speakers placed on the top of the piano. With the experimenter providing feedback regarding the difference in the produced and given SPLs, each participant practiced the task until he/she could reduce the errors to within 0.9 dB of the target SPL before data collection. The duration of this practice was seemingly longer for the novices compared with the experts.

Data-acquisition procedures The movement of the right upper limb in the sagittal plane was recorded using one of the position sensor cameras (sampling frequency⫽150 Hz) located 3.5 m from the right side of the participant. The spatial resolution in this camera setting was 0.8 mm in the sagittal plane. The LEDs for the cameras were mounted on the skin over the tip of the little finger and the centers of the MP

824

S. Furuya et al. / Neuroscience 164 (2009) 822– 831

(hand), styloid process (wrist), head of radius (elbow), and coracoid process (shoulder) joints. The movement of the finger-tip of the thumb was not measured, because in a pilot study an LED placed on the edge of the thumb sometimes hit the key, which could interfere natural keystroke motion. To minimize measurement errors, the rotational center of each joint was carefully estimated. For this, we video-recorded the joint movement after attaching the LEDs on the skin to visually check whether each LED was positioned at the center of rotation of the target joint. The data were digitally smoothed at a low-pass cutoff frequency of 10 Hz using a second-order Butterworth digital filter. The angular displacements at the MP, wrist, elbow and shoulder joints, as well as that of the little finger relative to the key surface, were then numerically calculated using an inner product method. Because there were no position data at the proximal-interphalangeal (PIP) and distal-interphalangeal (DIP) joint centers in the present experiment, we approximated the MP joint angle as the angle formed by the vectors from the MP joint center to the fingertip of the little finger and from the MP joint center to the wrist joint center. The G3-key kinematics were recorded using another position sensor camera located 0.65 m left of the key (spatial resolution⫽0.16 mm) and an LED placed on the key surface. The onset of the key-descending movement (“the finger-key contact time”) was determined when the calculated vertical velocity of the key exceeded 5% of its peak value. The movement of the G4 key was not measured due to difficulty in placing a close-up view camera on the right side of the piano without interfering with the kinematic recording of the hand movement by the other far-view camera. In a pre-test of octave keystrokes performed at various loudness levels, a significant spatiotemporal synchrony of the G3 and G4 keys had been confirmed (r⬎0.76). The electric activities of the right side of the triceps brachii medial head and the biceps brachii muscles were recorded with the EMG system. The biceps brachii muscle, rather than the brachioradialis muscle, was chosen in this study because our preliminary experiment of one expert and one novice player showed no clear activity of the brachioradialis muscle during keystroke motion. A previous study also failed to demonstrate activity in the brachioradialis muscle during multi-joint arm movements in the vertical plane (Papaxanthis et al., 2003). Pairs of Ag/AgCl surface disposal electrodes were placed at the estimated center of the belly of each target muscle with a 20 mm center-tocenter difference. Electrode position was carefully determined to minimize cross talk from adjacent muscles. At each electrode position, the skin was shaved, abraded, and cleaned using isopropyl alcohol to reduce source impedance. The EMG signals were amplified (5000⫻) and sampled at 900 Hz using an A/D converter interfaced with a personal computer. The signals were then digitally high-pass filtered with a cut-off frequency of 20 Hz and then root-mean squared. In order to normalize these EMG data for each muscle for each participant, maximum voluntary contraction (MVC) EMG data were obtained for each muscle by asking the participant to perform maximum flexion or extension isometric force production against a stationary object for a 5 s period. Each subject was verbally encouraged to achieve maximal force at a designated joint angle. During a MVC trial for the biceps and triceps muscles, the elbow joint was kept at 90°. A percentage MVC value was then calculated using the mean value of the middle 3 s period MVC data.

Data analysis For a graphic representation of the kinematic, kinetic and EMG data, mean curves of 30 keystroke trials at each loudness level for each subject were generated using a point-by-point averaging process based on a predetermined reference point for each trial. Fig. 1 shows an example of time history curves of the raw biceps and triceps EMGs, and vertical hand and key positions during a single keystroke performed at forte by one representative novice

Fig. 1. An example of the time history curves of raw EMGs recorded from the triceps and biceps muscles, and those of vertical positions of the hand and key during a single keystroke trial at the f loudness by one representative novice player. The duration of arm downswing was defined from the time of maximum hand position to the finger-key contact time.

participant. The reference point used was the time of the fingerkey contact, which was set as the time zero point in the present study (Fig. 1). For statistical analysis of the data, temporal, kinetic, and EMG variables were also computed from the data from each keystroke trial. One temporal variable was the movement duration of the period from the time of maximum hand position to the time of finger-key contact. This corresponds to the period of the arm downswing (Fig. 1). The time of finger-key contact was determined when the key-descending velocity exceeded 5% of the maximum. Using the measured kinematic data together with the anthropometric data of each participant, inverse dynamics equations were used to compute the time-varying elbow-joint muscular torque values during the entire phase of keystroke motion. To minimize the influence of interaction torques produced by surrounding joint motions on the computed muscular torque, the upper extremity was assumed to be four interconnected rigid links (upper arm, forearm, hand, and finger). Measured data and body segment parameters for Japanese individuals were used for the upper arm, forearm, and hand segments (Ae et al., 1992). The finger segment was approximated by uniform cylindrical tubes with ellipsoidal cross-sections, and the masses and moments of inertia were estimated using the density of water (Dennerlein et al., 1998; Kuo et al., 2006). The effect of body movement on these kinetic computations was not taken into account because it was nearly negligible. The muscular torque can be separated into static and

S. Furuya et al. / Neuroscience 164 (2009) 822– 831 dynamic components (Gottlieb et al., 1996). The static muscular torque is a portion that counteracts the effect of gravitational torque in maintaining the limb posture and, thus, it can have a magnitude equivalent to that of the gravitational torque but act in the opposite direction. The latter, on the other hand, is a portion used for limb movements and is the residual muscular torque after removing gravitational torque. In the following sections of this paper, this dynamic muscular torque is referred to as “muscular torque” for simplicity (see Appendix for details). Note that the positive and negative signs of muscular torque indicate production of flexion and extension torque by elbow muscles, respectively. The kinetic and EMG data were then quantified to evaluate the relation between the muscular activity and muscular torque. For the EMG data, the amount of muscular activity relevant to the production of muscular torque was quantified as follows. For each participant, the EMG data were normalized to a MVC value by using mean amplitude value of root-mean squared EMG data obtained from the middle 3 s period of each MVC trial, and then full-wave rectified and low-pass filtered at 15 Hz, yielding linear envelopes of each muscle EMG. For the triceps (agonist) muscle, the peak activation during the arm downswing period was computed for each keystroke and then averaged across trials (“Peak triceps activity”). To quantify the amount of reduction in the antigravity activity for elbow extension in the biceps (antagonist) muscle, we evaluated the deactivation of the biceps (antagonist) muscle by subtracting the minimum value of the biceps activation during the downswing phase from the maximum value obtained during the preparatory arm lift (“Amount of decrease in biceps activity”). For the muscular torque data, the peak value of extension muscular torque during the arm-downswing period was computed. The onset times of muscular activity (Tburst) of the triceps and biceps were also computed by employing the algorithm designed by Santello and McDonagh (1998) in order to evaluate our hypothesis that the triceps muscular contractions are initiated earlier in novices than in experts. We further computed the amount of elbow extension muscular torque produced by decreasing the biceps anti-gravity activity and increasing the triceps activity using the following equations.

␶GRA⫽

兰

T1⫹␦

MUS (t) dt,

T0

␶MUS⫽

兰

T2 T1⫹␦

MUS (t) dt,

where ␶GRA and ␶MUS are the impulse values of elbow extension muscular torque produced by gravity and agonist muscles, respectively, during arm downswing. T0 is the time at the initiation of downswing, T1 is Tburst of the triceps muscle, T2 is the time at the finger-key contact (the end of downswing), and ␦ is the electromechanical delay of the triceps muscle during concentric contraction. In the present study, we set ␦ as 26 ms based on the previous report by Norman and Komi (1979). To confirm that the participants achieved the required keystriking task, we computed several acoustical, kinematic, and kinetic variables. These included actual sound pressure level of radiated tone, peak key depressing force, peak key descending velocity, horizontal finger-tip position relative to the closer end of the key at the time of finger-key contact, maximum hand position during the preparatory arm-lift, and impulse of joint torque for accelerating elbow extension (elbow extension net torque) during the arm-downswing period.

Statistical data analysis Using the group (a between factor) and loudness values (a within factor) as independent variables, a two-way ANOVA with repeated measures was performed for each of the dependent variables. Statistical significance was set at P⬍0.05.

825

Fig. 2. Time history curves of the root-mean squared and normalized EMGs for the triceps brachii and biceps brachii muscles, the computed muscular torque and the angular velocity at the elbow joint at the forte (pink line) and piano (green line) loudness levels in one representative expert (left panel) and novice (right panel) player. The curves represent the average of 30 keystrokes. The yellow bar represents the average period of arm downswing. The dotted vertical lines indicate the average time of the highest hand position (a) and the finger-key contact (b).

RESULTS Time history curves Representative mean time history curves of the muscular activities of the upper arm (biceps and triceps) muscles, elbow muscular torque and angular velocity at the forte and piano loudness by one of the experts and one of the novices are shown in Fig. 2. In the preparatory arm-lift phase, both the expert and the novice showed an increase in biceps muscular activity immediately prior to increases in flexion muscular torque and angular velocity at the elbow joint. Toward the end of the arm-lift phase, biceps muscular activity and elbow flexion muscular torque continued to decrease. Prior to the initiation of the arm downswing, both players exhibited an increase in elbow extension torque, followed by an increase in elbow extension velocity. The magnitude of the elbow extension torque and velocity was greater at the loudness level of forte than at piano. During the arm downswing, a clear difference in the muscular activation pattern was observed between the expert and novice players. The novice exhibited triceps muscular activa-

826

S. Furuya et al. / Neuroscience 164 (2009) 822– 831

Fig. 3. The group means of the onset time of muscular activities (the squares) during arm downswing in keystroke for the triceps (top panel) and biceps (bottom panel) muscles. The group means of the onset time of elbow extension (the triangles) were also plotted. Error bars represent⫾one SD. Time zero indicates the time of finger-key contact. The filled and open symbols indicate expert and novice players, respectively.

tion that increased until the time of finger-key contact, but the expert did not. Muscular deactivation of the biceps was observed for both the expert and the novice, but the magnitude of this decrease in biceps activity was greater when generating a louder tone only for the expert. Immediately before the finger-key contact time, the coactivation of biceps and triceps muscles appeared for both players. The curves of muscular torque and angular velocity showed that during the downswing phase the expert had a smaller extension muscular torque and greater angular velocity than the novice. Onset time of muscular activity (Tburst) Fig. 3 shows the group means of Tburst of the biceps and triceps muscles at different loudness levels. ANOVA revealed significant main effects of group (F(1,12)⫽16.34, P⫽0.002) and loudness (F(3,36)⫽13.3, P⬍0.001) for the triceps Tburst. Their interaction effect was, however, nonsignificant. The biceps Tburst was near zero for both groups across all loudness levels, and thus neither group nor loudness effect were significant. The earlier triceps Tburst value for the novices than the experts with no difference in the biceps Tburst value between the two groups confirmed earlier initiation of reciprocal elbow muscular activation for the novice players. To examine the possibility that the observed group difference in Tburst at the triceps could be attributed to the difference in the duration of the arm downswing, the triceps Tburst relative to the downswing duration was also compared between the experts and novices (given in parentheses in Fig. 3). ANOVA using the percent value data

again confirmed a significant group effect of the relative Tburst value for the triceps muscle (F(1,12)⫽13.13, P⫽ 0.003), but no group⫻loudness interaction effect. To examine the duration during which the arm downswing was produced by gravity, we first computed the time at which the elbow extension movement was initiated relative to the finger-key contact time at each loudness level (see triangle marks in Fig. 3). Then we computed the duration between this onset time of elbow extension and Tburst. The mean values and SDs of this duration for all experts were 126⫾30, 111⫾24, 91⫾49, and 82⫾52 ms at p, mp, mf, and f loudness, respectively. The corresponding values for the novices were 84⫾38, 68⫾30, 43⫾27, and 12⫾23 ms, respectively. ANOVA revealed significant main effects of group (F(1,12)⫽10.17, P⫽0.008) and loudness (F(3,36)⫽17.12, P⬍0.001), but their interaction effect was insignificant. The greater mean value of this duration for the experts indicated that they used gravity to descend the forearm for a longer period of time than the novices. Correlation between the muscular activity and joint torque To examine the skill-level-dependent gravity-muscular force interaction with respect to tone loudness, we compared loudness-dependent changes in elbow agonist and antagonist muscular activity in relation to muscular torque in the expert and novice players. Fig. 4 shows the correlation of the peak extension muscular torque during arm downswing with the amount of decrease in biceps activity (the upper panels) and with the peak triceps activity (the lower panels) during downswing for one representative expert

S. Furuya et al. / Neuroscience 164 (2009) 822– 831

827

Fig. 4. The relation between the amount of decrease in the biceps activity and the peak elbow muscular torque (top panels) and between the peak triceps activity and the peak elbow muscular torque (bottom panel) in one representative expert (left panels) and novice (right panels) player. Dots with blue, orange, green, and red indicate p, mp, mf, and f loudness, respectively.

(left figures) and one novice (right figures) player across all loudness levels. For all of the experts, the correlation associated with the biceps was significant, but that with the triceps was not. In contrast, the correlation associated with the triceps was significant for all novice players, but that with the biceps was significant for only two of the novices. The group means of the correlation values with the decrease in biceps activation and the increase in triceps activation for the experts were 0.57⫾0.09 and 0.12⫾0.06, respectively, whereas those for the novices were 0.23⫾0.19 and 0.63⫾0.09, respectively. ANOVA revealed that the experts had a higher correlation value for the biceps (F(1,12)⫽19.84, P⬍0.001) and a lower correlation for the triceps than did the novices (F(1,12)⫽139.74, P⬍0.001). Elbow extension torque produced by gravitational and muscular forces Fig. 5 shows the group means of the ␶GRA (Fig. 5A) and ␶MUS (Fig. 5B) values at each loudness level. ANOVA

revealed a significant group⫻loudness interaction effect for each of these variables (␶GRA: F(3,36)⫽5.58, P⫽0.003; ␶MUS: F(3,36)⫽23.06, P⬍0.001). The interaction effect for ␶GRA indicated that with the generation of louder sound, the experts had a larger mean value than the novices. The interaction effect for ␶MUS, on the other hand, indicated that the novices had a larger value than the experts during the production of louder sound. The main effect of group was also significant for the ␶MUS (F(1,12)⫽57.41, P⬍0.001), indicating that the experts had smaller muscular torque produced by the triceps activity compared to the novices. Sound pressure level, key reaction force, key descending velocity, horizontal position of the finger-tip at the finger-key contact, hand height, and joint torque for elbow extension Table 1 lists the group means and SDs of task-related acoustic, kinematic, and kinetic parameters at each loudness level, and the results of ANOVA. There was a significant loudness effect for the means of these variables

828

S. Furuya et al. / Neuroscience 164 (2009) 822– 831

Fig. 5. The group means of ␶GRA (A) and ␶MUS (B) for elbow extension during a keystroke at each loudness level. The error bars represent⫾one SD. * P⬍0.05, ** P⬍0.01.

except for the horizontal position of the finger-tip at the finger-key contact time. Neither the group main effect nor the effect of the interaction between the group and loudness was significant, confirming the achievement of task requirement for both subjects.

DISCUSSION The use of gravity to effectively reduce muscular work for the production of elbow extension muscular torque, independent of interaction torque arising from the surrounding joints and gravitational torque, was investigated using expert and novice players of the piano during the downward key-striking motion in the present study. The findings demonstrated a clear skill-level difference in coordination of upper arm muscular activity in relation to the production of elbow extension muscular torque. Across all levels of sound loudness examined, no triceps activity was noted until just before the end of arm downswing for the experts. Instead, clear deactivation of their anti-gravity muscle was observed, allowing the predominant production of elbow extension torque by gravity. In addition, the amount of this deactivation was accentuated when producing a louder tone, indicating that the experts used greater amount of gravity to increase movement speed of arm descent. The use of gravity to facilitate movement of a target body segment has been also reported when trained typists were typing a computer key, in which no muscular activity of the finger flexors was found prior to the production of finger flexion torque (Dennerlein et al., 1998; Kuo et al., 2006). However, the loudness-dependent modulation of the amount of gravity utilization for arm descent in the current expert pianists clearly highlighted the difference in motor skills of striking motions between the skilled typists and pianists. Typing a computer key does not typically require graded exertion of finger force. On the other hand, a piano keystroke requires precise control of the key-striking force

because it determines the loudness of the produced tone (Kinoshita et al., 2007). We previously found that this adjustment of key-striking force was associated with modulation of the elbow extension muscular torque (Furuya and Kinoshita, 2008b). Therefore, to elicit the aimed tone loudness via a gravity-driven keystroke, pianists must have acquired motor skills to precisely control the timing and amount of deactivation of the anti-gravity muscles while they are lengthening. However, previous studies showed that muscular torque produced by the lengthening contraction (eccentric contraction) is highly sensitive to changes in muscular activation signals (e.g., Duchateau and Enoka, 2008). Effective use of gravity in striking the key, therefore, requires more complex motor skills with respect to force control for pianists than for typists. In contrast to the motor control revealed for the expert players, clear activation of the triceps muscle was commonly observed in the middle of the downswing for the novices. In addition, in response to an increase in loudness, this muscular activity started earlier, and its magnitude became greater. Because there was no coactivation of the biceps muscle during the corresponding period, the triceps activity observed cannot be considered as a part of stiffening mechanism for the elbow joint against upcoming key-reaction force. These findings therefore indicate an agonist contribution to the production and regulation of muscular torque for descending the arm. Similar to this, some studies that examined muscular activity in ordinary individuals during vertical arm movement showed that the triceps activity during arm descent scaled with movement speed (d’Avella et al., 2008; Flanders and Herrman, 1992). The novice piano players in the current study, therefore, most likely used a rudimentary muscular activation pattern developed through daily experience of the arm-descending motion for accelerating arm-downswing when striking the keys.

0.7 50.0*** The numbers in the parenthesis indicate the standard deviation. * P⬍0.05, ** P⬍0.01, *** P⬍0.001.

⫺0.15 (0.04) ⫺0.12 (0.05)

⫺0.23 (0.05)

⫺0.28 (0.06)

⫺0.14 (0.05)

⫺0.16 (0.05)

⫺0.21 (0.07)

⫺0.28 (0.11)

6.4 ⫻ 10⫺4

0.2 43.2*** 0.1 99.8 (27.3) 78.7 (22.8) 65.2 (18.5) 53.1 (18.3) 93.1 (16.7) 60.4 (24.4) 52.0 (29.0)

77.4 (17.9)

2.7 2.6 0.2 0.09 825.3*** 189.4*** 423.6*** 0.13 3.9 4.4 0.9 0.81 114.0 (0.5) 20.2 (5.1) 211.2 (10.8) 16.2 (8.4) 109.9 (0.5) 12.3 (2.0) 186.9 (10.5) 15.4 (4.1) 107.2 (0.5) 7.3 (1.8) 159.5 (11.6) 15.1 (5.7) 103.1 (1.0) 4.8 (1.5) 117.2 (14.9) 15.2 (6.4) 113.0 (1.0) 23.1 (3.4) 216.9 (14.9) 12.1 (8.2) 109.0 (0.9) 16.0 (1.7) 192.6 (16.8) 12.3 (7.7) 105.8 (0.7) 8.1 (1.2) 160.0 (15.1) 11.7 (9.0) 101.9 (0.8) 4.9 (1.1) 121.5 (15.6) 12.3 (9.3)

Sound pressure level (dB) Peak key reaction force (N) Peak key velocity (mm/s) Horizontal finger-tip position at the time of finger-key contact (mm) Maximum hand position during preparatory armlift Impulse of elbow extension net torque (Nm)

Group⫻ Loudness Loudness Group p mp

mf

f p

mp

mf

f

ANOVA results (F-value) Novices Experts Variables

Table 1. Mean and SD values of sound pressure level, key reaction force, key-descending velocity, horizontal finger-tip position at the time of finger-key contact, maximum hand position during preparatory arm-lift, and elbow extension net torque during arm downswing

S. Furuya et al. / Neuroscience 164 (2009) 822– 831

829

A free-fall of the arm is a simple motor task even for novices. One may therefore wonder why they are unable to execute free-fall movement during striking the piano keys. There seem to be at least three reasons. First, eccentric muscular contraction of the anti-gravity muscles during gravity-driven arm downswing could constrain the accurate production of the desired key-striking velocity. We found that the production of elbow extension muscular torque was predominantly accompanied by a lengthening (eccentric) contraction of the biceps for experts and by a shortening (concentric) contraction of the triceps for novices. Previous studies reported that the force produced by eccentric muscular contraction is more variable compared to that produced by concentric contraction (Christou and Carlton, 2002; Enoka, 1996). The findings suggest that when utilizing gravity to descend the forearm and strike a key at a designated sound loudness, the eccentric contraction of the biceps demands a more precise motor command than that for the concentric contraction of the triceps. Indeed, Fang et al. (2001) showed that the production of a target level of muscular torque by eccentric contraction of the biceps, compared with concentric contraction, elicited greater cortical activation related to movement planning and execution, as well as feedback signals from the peripheral systems. This finding implies a more complex force control requirement for the eccentric contraction. In addition, precise modulation of the amount of eccentric contraction of the biceps for adjusting the loudness of tone should be another complex motor control problem, as argued above. Indeed, ␶GRA and ␶MUS, the measures of concentric and eccentric muscular contraction, indicated that the novices relied more on concentric contraction of the triceps and less of eccentric contraction of the biceps for these loudness adjustment. The difference in force production between these two contractile modes may, therefore, inhibit novices from using gravity when striking the keys. Second, the variability of the movements caused by external perturbation as well as by intrinsic noise, such as a signal-dependent noise, can also perturb the production of the targeted arm-descending velocity. Previous studies have reported that appropriate coactivation can effectively reduce these perturbing effects by elevating joint stiffness (Gribble et al., 2003; Osu et al., 2004) and thus facilitate the production of the desired arm-descending velocity. Other studies (Franklin et al., 2008; Osu et al., 2002; Thoroughman and Shadmehr, 1999) have also provided evidence that, at an early stage of motor learning, feed forward muscle coactivation is a common strategy used in coping with unexpected deviations in produced movement, which gradually decrease with training. The novices’ strategy in the present study, therefore, may reflect this coactivation strategy because a muscular-driven arm downswing involves smaller biceps deactivation and larger triceps activation and, thus, higher coactivation than the gravity-driven downswing. Third, the need to instantaneously develop an adequate level of joint stiffness in the upper limb to compensate the reaction force from finger-key collision can also be

830

S. Furuya et al. / Neuroscience 164 (2009) 822– 831

a constraint for novices in selecting a complete free-fall of the arm for piano keystrokes. Our recent study found that at the end of arm downswing, both expert and novice pianists exhibited an increase in force of muscles surrounding the elbow joint, and amount of this coactivation was graded in proportional to tone loudness (Furuya and Kinoshita, 2008a). This indicated that the elevation of the target joint stiffness appropriately compensated the keyreaction force. However, according to Park and Stelmach (2006), exerting a targeted muscular force at a faster rate produces a more variable force. Therefore, a vigorous key-striking motion by active agonist muscular contraction simplifies this problem for novices because joint stiffness control can be initiated in advance during the downswing. Indeed, the novices showed earlier initiation of the triceps activity when production of larger coactivation is needed at louder tone condition, which may reflect their strategy of avoiding an abrupt increase in joint stiffness. One may consider the possible use of gravity at joints other than the elbow. For the wrist, we previously found that the wrist flexion muscular torque significantly developed after occurrences of corresponding wrist muscular activation for both expert and novice players (Furuya and Kinoshita, 2008a,b), implying that the use of gravity for wrist joint motion in piano keystrokes was quite small if any. For the shoulder, there is multiple intrinsic and extrinsic muscles crossing over this joint, causing difficulties in obtaining experimental evidence of the utilization of gravity (which should include observations of no agonist muscular activation during shoulder movement). Nevertheless, there remains a possibility that the experts utilized gravity for shoulder extension. Earlier, Dennerlein et al. (1998) confirmed that there was no muscular activity of the agonist of the finger flexor in professional typists prior to the initiation of the finger flexion movement to strike the computer keys, and thus they concluded that gravity was used for initiating keystroke action. However, due to the lack of a comparison among individuals with different levels of skill, their findings were insufficient to determine whether this motor skill simply reflected the task nature of the key-striking motion or the subjects’ expertise. Therefore, the current finding of expertise-dependent utilization of gravity for driving the arm-downswing provides the first evidence that directly supports Bernstein’s idea with respect to gravity. The results also support the idea of weight play in the piano pedagogy, the active use of gravity to drop the arm for striking the keys (Kay et al., 2003). The use of gravity in keystrokes lessens the muscular effort for movement production because it complements the agonist muscular force in descending the limb. This is particularly essential in endurance-type motor tasks because even a subtle reduction in muscular effort during one key-striking action can eventually lead to a remarkable enhancement of physiological efficiency after a number of repetitive keystrokes. This idea is likely supported by the fact that expert pianists, but not novices, were able to maintain vigorous keystrokes for a prolonged duration (Furuya and Kinoshita, 2008b). Survey studies have shown that musculoskeletal disorders

in the upper extremities of computer keyboard users (Gerr et al., 2006) and pianists (Bragge et al., 2006; Furuya et al., 2006) are prevalent. The strategy used by the experts in the current study can be of help in preventing these injuries.

CONCLUSION In summary, a clear expert-novice difference in the activity of the muscles for the production of muscular torque at the elbow joint to drive a downswing motion of the arm in the present study indicated an expertise-dependent nature of gravity-muscular force interaction in piano keystrokes. The novices commonly used a muscular force-driven arm downswing so that a target tone could be constantly produced, whereas the experts relied heavily on a gravity-dependent drop of the arm while keeping the contribution of the muscular force and work to a minimum. These findings confirmed a training-dependent accuracy-efficiency tradeoff, as proposed by Osu et al. (2002) and, more recently, by Franklin et al. (2008). A similar learning mechanism might be involved in the acquisition of the current “weight-play” technique. ¨ LLER Acknowledgments—We thank to Prof. Eckart ALTENMU (Hanover University of Music and Drama) for his critical comments and helpful and constructive suggestions on an earlier version of this manuscript. A part of the present study was supported by Grant-in-Aid for Young Scientists (Start-up). This work was also partly supported by SCOPE, MIC and SRPBS, MEXT.

REFERENCES Ae M, Tang H, Yokoi T (1992) Estimation of inertia properties of the body segments in Japanese athletes. Biomechanism 11:23–33 (in Japanese with English abstract). Atkeson CG, Hollerbach JM (1985) Kinematic features of unrestrained vertical arm movements. J Neurosci 5(9):2318 –2330. Bernstein NA (1967) The coordination and regulation of movements. London: Pergamon. Berret B, Darlot C, Jean F, Pozzo T, Papaxanthis C, Gauthier JP (2008) The inactivation principle: mathematical solutions minimizing the absolute work and biological implications for the planning of arm movements. PLoS Comput Biol 4(10):e1000194. Bragge P, Bialocerkowski A, McMeeken J (2006) A systematic review of prevalence and risk factors associated with playing-related musculoskeletal disorders in pianists. Occup Med 56(1):28 –38. Christou EA, Carlton LG (2002) Motor output is more variable during eccentric compared with concentric contractions. Med Sci Sports Exerc 34(11):1773–1778. Cordo PJ, Hodges PW, Smith TC, Brumagne S, Gurfinkel VS (2006) Scaling and non-scaling of muscle activity, kinematics, and dynamics in sit-ups with different degrees of difficulty. J Electromyogr Kinesiol 16(5):506 –521. d’Avella A, Fernandez L, Portone A, Lacquaniti F (2008) Modulation of phasic and tonic muscle synergies with reaching direction and speed. J Neurophysiol 100(3):1433–1454. Dennerlein JT, Mote CD Jr, Rempel DM (1998) Control strategies for finger movement during touch-typing. The role of the extrinsic muscles during a keystroke. Exp Brain Res 121(1):1– 6. Dounskaia N (2005) The internal model and the leading joint hypothesis: implications for control of multi-joint movements. Exp Brain Res 166:1–16.

S. Furuya et al. / Neuroscience 164 (2009) 822– 831 Duchateau J, Enoka RM (2008) Neural control of shortening and lengthening contractions: influence of task constraints. J Physiol 586(Pt 24):5853–5864. Enoka RM (1996) Eccentric contractions require unique activation strategies by the nervous system. J Appl Physiol 81(6):2339 – 2346. Fang Y, Siemionow V, Sahgal V, Xiong F, Yue GH (2001) Greater movement-related cortical potential during human eccentric versus concentric muscle contractions. J Neurophysiol 86(4):1764 –1772. Flanders M, Herrmann U (1992) Two components of muscle activation: scaling with the speed of arm movement. J Neurophysiol 67(4):931–943. Franklin DW, Burdet E, Tee KP, Osu R, Chew CM, Milner TE, Kawato M (2008) CNS learns stable, accurate, and efficient movements using a simple algorithm. J Neurosci 28(44):11165–11173. Furuya S, Kinoshita H (2007) Roles of proximal-to-distal sequential organization of the upper limb segments in striking the keys by expert pianists. Neurosci Lett 421:264 –269. Furuya S, Kinoshita H (2008a) Organization of the upper limb movement for piano key-depression differs between expert pianists and novice players. Exp Brain Res 185:581–593. Furuya S, Kinoshita H (2008b) Expertise-dependent modulation of muscular and non-muscular torques in multi-joint arm movements during piano keystroke. Neuroscience 156(2):390 – 402. Furuya S, Nakahara H, Aoki T, Kinoshita H (2006) Prevalence and causal factors of playing-related musculoskeletal disorders of the upper extremity and trunk among Japanese pianists and piano students. Med Probl Perform Art 21:112–117. Gerr F, Monteilh CP, Marcus M (2006) Keyboard use and musculoskeletal outcomes among computer users. J Occup Rehabil 16(3):265–277. Gottlieb GL, Song Q, Hong DA, Corcos DM (1996) Coordinating two degrees of freedom during human arm movement: load and speed invariance of relative joint torques. J Neurophysiol 76:3196 –3206. Gribble PL, Mullin LI, Cothros N, Mattar A (2003) Role of cocontraction in arm movement accuracy. J Neurophysiol 89(5):2396 –2405. Hirashima M, Kudo K, Watarai K, Ohtsuki T (2007) Control of 3D limb dynamics in unconstrained overarm throws of different speeds performed by skilled baseball players. J Neurophysiol 97:680 – 691. Hmelnitsky I, Nettheim N (1987) Weight-bearing manipulation: a neglected area of medical science relevant to piano playing and overuse syndrome. Med Hypotheses 23:209 –217. Kay BA, Turvey MT, Meijer OG (2003) An early oscillator model: studies on the biodynamics of the piano strike (Bernstein and Popova, 1930). Motor Control 7:1– 45.

831

Kinoshita H, Furuya S, Aoki T, Altenmüller E (2007) Loudness control in pianists as exemplified in keystroke force measurements on different touches. J Acoust Soc Am 121(5 Pt 1):2959 –2969. Kuo PL, Lee DL, Jindrich DL, Dennerlein JT (2006) Finger joint coordination during tapping. J Biomech 39(16):2934 –2942. Matthay T (1905) The first principles of pianoforte playing. London: Longmans, Green, and co. Norman RW, Komi PV (1979) Electromechanical delay in skeletal muscle under normal movement conditions. Acta Physiol Scand 106(3):241–248. Oldfield RC (1971) The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9:97–113. Osu R, Franklin DW, Kato H, Gomi H, Domen K, Yoshioka T, Kawato M (2002) Short- and long-term changes in joint co-contraction associated with motor learning as revealed from surface EMG. J Neurophysiol 88(2):991–1004. Osu R, Kamimura N, Iwasaki H, Nakano E, Harris CM, Wada Y, Kawato M (2004) Optimal impedance control for task achievement in the presence of signal-dependent noise. J Neurophysiol 92(2): 1199 –1215. Papaxanthis C, Pozzo T, Schieppati M (2003) Trajectories of arm pointing movements on the sagittal plane vary with both direction and speed. Exp Brain Res 148(4):498 –503. Park JH, Stelmach GE (2006) 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. Exp Brain Res 168(3):337–347. Putnam CA (1993) Sequential motions of body segments in striking and throwing skills: description and explanations. J Biomech 26 (Suppl 1):125–135. Santello M, McDonagh MJ (1998) The control of timing and amplitude of EMG activity in landing movements in humans. Exp Physiol 83(6):857– 874. Thoroughman KA, Shadmehr R (1999) Electromyographic correlates of learning an internal model of reaching movements. J Neurosci 19(19):8573– 8588. White O, Bleyenheuft Y, Ronsse R, Smith AM, Thonnard JL, Lefèvre P (2008) Altered gravity highlights central pattern generator mechanisms. J Neurophysiol 100(5):2819 –2824.

APPENDIX Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.neuroscience.2009.08.024.

(Accepted 12 August 2009) (Available online 18 August 2009)