Exp Brain Res (2009) 199:117–126 DOI 10.1007/s00221-009-1979-y
R ES EA R C H A R TI CLE
Music performance anxiety in skilled pianists: eVects of social-evaluative performance situation on subjective, autonomic, and electromyographic reactions Michiko Yoshie · Kazutoshi Kudo · Takayuki Murakoshi · Tatsuyuki Ohtsuki
Received: 5 June 2009 / Accepted: 5 August 2009 / Published online: 22 August 2009 © Springer-Verlag 2009
Abstract Music performance anxiety (MPA), or stage fright in music performance, is a serious problem for many musicians, because performance impairment accompanied by MPA can threaten their career. The present study sought to clarify on how a social-evaluative performance situation aVects subjective, autonomic, and motor stress responses in pianists. Measurements of subjective state anxiety, heart rate (HR), sweat rate (SR), and electromyographic (EMG) activity of upper extremity muscles were obtained while 18 skilled pianists performed a solo piano piece(s) of their choice under stressful (competition) and non-stressful (rehearsal) conditions. Participants reported greater anxiety in the competition condition, which conWrmed the eVectiveness of stress manipulation. The HR and SR considerably increased from the rehearsal to competition condition reXecting the activation of sympathetic division of the autonomic nervous system. Furthermore, participants showed higher levels of the EMG magnitude of proximal muscles (biceps brachii and upper trapezius) and the co-contraction of antagonistic muscles in the forearm (extensor digitorum communis and Xexor digitorum superWcialis) in the competition condition. Although these responses can be interpreted as integral components of an adaptive biological system that creates a state of motor readiness in an unstable or unpredictable environment, they can adversely inXuence
M. Yoshie (&) · K. Kudo · T. Murakoshi · T. Ohtsuki Department of Life Sciences, Graduate School of Arts and Sciences, The University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan e-mail:
[email protected] M. Yoshie Japan Society for the Promotion of Science, 8 Ichibancho, Chiyoda-ku, Tokyo 102-8472, Japan
pianists by disrupting their Wne motor control on stage and by increasing the risk of playing-related musculoskeletal disorders. Keywords Psychological stress · Motor control · Music performance anxiety · Pianist · Electromyography · Autonomic nervous system
Introduction Music performance anxiety (MPA), more commonly referred to as stage fright, has always been aZicting musicians all around the world. In a large survey involving 2,212 musicians, MPA was found to be the most frequent non-musculoskeletal medical problem among the respondents, with more than 20% using beta-blockers before signiWcant public performances (Fishbein et al. 1988). On top of its prevalence, MPA can even threaten a musician’s career, because an excessive level of MPA sometimes leads to actual impairment of performance skills (Yoshie et al. 2008b, 2009). So far, research has examined the contribution of situational factors to the arousal of MPA (Papageorgi et al. 2007; Wilson 1997). The Wndings suggested that performance circumstances leading to a strong sense of exposure, including solo against group performance (Cox and Kenardy 1993), public performance against practice (Fredrikson and Gunnarsson 1992; LeBlanc et al. 1997), and evaluative against non-evaluative performance (Abel and Larkin 1990; Brotons 1994; Craske and Craig 1984; Hamann and Sobaje 1983; Yoshie et al. 2008a) are likely to increase MPA and associated physiological arousal. Thus, it seems reasonable to suppose that the combination of these situational factors will induce the greatest levels of
123
118
MPA. Furthermore, these studies normally used an experimentally devised performing situation where participants’ performance outcome would never aVect their future career prospects. In a real musical world, however, every single performance in public can either boost or wreck musicians’ career, which causes them higher levels of psychological stress. In the present study, therefore, we employed an actual piano competition to observe the stress responses of musicians performing in a realistic environment. Some of the previous MPA studies (Abel and Larkin 1990; Brotons 1994; Craske and Craig 1984; Fredrikson and Gunnarsson 1992; LeBlanc et al. 1997) have eVectively demonstrated the changes in physiological arousal in stressful performances mostly through the measurements of heart rate (HR). However, the relationship between physiological arousal and performance has been seen as equivocal (Yoshie et al. 2009), with a variety of predictions ranging from the inverted-U hypothesis (Yerkes and Dodson 1908), zones of optimal functioning (ZOF) model (Hanin 1978; Yoshie et al. 2008b) to catastrophe theory (Hardy and ParWtt 1991). To better understand the eVects of MPA on performance quality, it would be highly beneWcial to examine the motor process mediating the arousal–performance relationship. Recent studies have accumulated evidence indicating that the induction of psychological stress or negative emotions leads to activation of the motor system. Using a functional magnetic resonance imaging (fMRI), Butler et al. (2007) showed elevated activation of dorsal basal ganglia in response to an experimentally induced state of conscious fear. Transcranial magnetic stimulation studies (Baumgartner et al. 2007; Hajcak et al. 2007; Schutter et al. 2008) have also demonstrated enhanced motor-evoked potentials during the presentation of fearful or unpleasant emotional stimuli indicating increased corticospinal motor tract excitability associated with negative emotions. These phenomena may be understood as representing a state of motor readiness to facilitate defensive behaviors under a potential threat to survival, which has traditionally been known as the Wght-or-Xight response (Cannon 1915). These Wndings have also been conWrmed in experiments combining electromyography (EMG) with simple motor tasks. Such studies consistently reported increased EMG activity in the upper extremity muscles associated with cognitive or emotional stress using a wide range of tasks such as Wnger tapping (Bloemsaat et al. 2005), Wnger elevation (Finsen et al. 2001), computer work (van Galen et al. 2002; Visser et al. 2004; Wahlström et al. 2002, 2003), hand and shoulder exertions (Au and Keir 2007), and ball stroking (Matsumoto et al. 2001). Predictably, elevated muscle activity under stress has often been accompanied by greater force production, as reXected in the increases of isometric key-pressing force (Bloemsaat et al. 2005), grip- and click-
123
Exp Brain Res (2009) 199:117–126
forces applied on a computer mouse (Visser et al. 2004; Wahlström et al. 2002), maximal isometric force produced by the wrist and Wnger extensor muscles (Coombes et al. 2006), pinch grip force (Coombes et al. 2008), and speed of a stroked ball (Matsumoto et al. 2001). Interestingly, our recent experiment (Yoshie et al. 2008a) showed that these Wndings hold for a piano-playing task as well. In the experiment, we attempted to manipulate participants’ stress levels by simply scoring their arpeggio performance in a laboratory setting, and the resulting increase in state anxiety coincided with elevated EMG activity in the upper extremity muscles and stronger keystroke force. Because striking the keys with appropriate force levels is a fundamental skill for pianists, increased muscle activity stemming from overall activation of the motor system might account for the loss of Wne motor control in stressful performance situations. In the present study, we examined whether such EMG activation is observed also during performances of classical piano works in a real competition. We hypothesized that participants would respond to the social-evaluative stressor with higher muscle activity and less muscle relaxation. Psychological stress can aVect not only levels of EMG activity, but also its coordination patterns. Research has indicated that stress or anxiety increases the co-contraction levels of antagonistic muscles in the upper extremity (Meulenbroek et al. 2005; van Galen et al. 2002; Weinberg and Hunt 1976). Normally, the levels of co-contraction and joint stiVness become lower as motor learning proceeds and the contribution of feedforward component increases (Fujii et al. 2009a, b; Furuya and Kinoshita 2008; Osu et al. 2002). When put under psychological stress, however, the CNS may be required to heighten joint stiVness through cocontraction to maintain movement accuracy in the face of deteriorated signal-to-noise ratio in the motor system (Gribble et al. 2003; van Galen et al. 2002; van Gemmert and van Galen 1997). In this regard, increased co-contraction levels under stress may be interpreted as a strategic means to adapt muscle activity to central information processing demands at the expense of physiological eYciency. Based on these Wndings, we hypothesized that the co-contraction level of antagonistic muscles in the forearm would also be higher in pianists executing over-learned movements under intense social-evaluative stress.
Methods Participants We recruited pianists for participation in the competition experiment by distributing application forms and informational brochures describing qualiWcation requirements to potential applicants at music schools and piano clubs.
Exp Brain Res (2009) 199:117–126
Eventually, 18 highly trained pianists (7 men and 11 women, mean age § SD = 26.7 § 6.3 years) with a mean of 20.4 (SD = 6.2) years of playing experience participated in the experiment. Written informed consent was obtained from all the participants, and the study was approved by the Ethics Committee of the Graduate School of Arts and Sciences, the University of Tokyo. Experimental task and conditions We asked participants to perform a solo piano piece(s) of their choice requiring considerable playing skills, such as piano sonatas of Beethoven, ProkoWev, and Rachmaninov. Participants played the whole piece(s) or the Wrst part of the piece (mean duration § SD = 6.5 § 1.4 min) from memory on an acoustic grand piano. Participants attended two sessions on two separate days: the rehearsal and competition. In both conditions, we instructed participants not to drink caVeinated or alcoholic beverages within 24 h before their performance. In the rehearsal condition, each participant individually visited a small music practice room, and played the piece(s) on a Yamaha C3 grand piano (Yamaha, Hamamatsu, Japan) with no experimenter present in the room. In the competition condition, we held an actual piano competition in a concert hall, and participants performed the piece(s) on a Steinway D-274 full concert grand piano (Steinway & Sons, NY, USA) in front of a large audience (N = 45.6 § 3.3) and Wve professional judges (four professional concert pianists and a musicologist). After each performance, the performer received warm applause from the audience. In the awarding ceremony following the competition, a judge gave the certiWcates of commendation and cash rewards of 20,000, 10,000, and 5,000 JPY (»200, 100, and 50 USD) to the recipients of the Wrst, second, and third prize, respectively. In both the practice room and concert hall, temperature and humidity were carefully controlled. Data acquisition and analysis Performance measures In both conditions, participants’ performances were digitally recorded at 44.1 kHz with two microphones, which were placed above the piano and connected to a handy audio recorder H4 (Zoom, Tokyo, Japan). The total duration of each performance was calculated based on the audio recordings. With respect to the rehearsal condition, the Wve judges evaluated the recorded performances. In the competition condition, participants were rendered invisible to the judges so that visual information would not aVect their evaluation, and the jury directly evaluated participants’ live performances. The jury scored performances on ten items,
119
with four items concerning technical aspects (i.e., accuracy, technical dexterity, tempo and rhythm, and memory) and six items concerning artistic aspects (i.e., artistry, interpretation, expressiveness, structural strength, melodic and harmonic balance, and tone quality). The scale ranged from 1 (poor) to 10 (excellent), leading to the total score ranging from 10 to 100. To examine the inter- and intra-rater reliability of performance evaluation scores, we computed intraclass correlation coeYcients (ICC) for the rankings of total scores. The performance scores demonstrated substantial levels of inter-rater reliability, with ICC (2, 5) of 0.80 and 0.85 (Ps < 0.001) for the rehearsal and competition conditions, respectively. To conWrm that the jury was capable of consistently evaluating recorded performances and live performances, we asked three members of the jury to rate recorded performances of the competition condition once again well after the experiment. Then, we computed ICC (1, 1) for each judge using his/her score rankings determined for recorded and live performances of the competition condition. The ICC (1, 1) for the three judges were 0.77, 0.76, and 0.70 (Ps < 0.001) showing suYcient levels of intra-rater reliability. Thus, we used the total scores that were averaged across the judges to compare performance quality in the competition condition with that in the rehearsal condition. We also calculated the mean scores of the technical and artistic items separately (technical and artistic scores). Self-reported measure The level of state anxiety was assessed just prior to each participant’s performance with the visual analog mood scale (VAS), which is capable of measuring emotional states in a quick, reliable, and relatively sensitive manner (Cella and Perry 1986). The VAS is a 100-mm continuous scale ranging from 0 (not anxious at all, the left end) to 100 (extremely anxious, the right end). We asked participants to place a vertical line bisecting the 100-mm line to indicate the perceived level of anxiety at the moment. The VAS data were quantiWed by measuring the distance between the left end and the vertical line (mm). Autonomic measures Beat-to-beat HR was monitored continuously throughout experimental sessions using a wireless signal transmission device with electrocardiogram precision (Polar S810i by Polar Electro Oy, Kempele, Finland) at a sampling rate of 1,000 Hz. The receiver unit of the HR monitor, which resembled a large wrist watch, was hung from an adjustable belt worn round the waist in order not to disturb the performing pianists. The consecutive inter-beat (RR) intervals were automatically stored on the receiver. The RR data
123
120
Exp Brain Res (2009) 199:117–126
were carefully edited by visual inspection and elimination of measurement artifacts. We computed the mean HR (bpm) during performance (i.e., from the onset of the Wrst note to the end of last note) based on the corrected data. During each piano performance, we also gauged the amount of sweat evaporated from the sole of the left foot with a sweat rate (SR) meter (TS100 by Techno Science, Tokyo, Japan) utilizing the ventilated capsule technique. The SR data were sampled at 1,000 Hz with an MP150 data acquisition system (Biopac Systems, Inc., CA, USA) interfaced with a personal computer. We calculated the mean SR amplitude (mg/min/cm2) for each condition. Figure 1a, b shows the typical examples of HR and SR time series, respectively. EMG measures During each piano performance, surface EMG activity was recorded from the left extensor digitorum communis (ED), Xexor digitorum superWcialis (FD), biceps brachii (BB), and upper trapezius (TR) muscles. Bipolar Ag/AgCl disposable electrodes (10 mm diameter) were placed at the estimated motor point of each target muscle with a 20-mm center-to-center distance. At each electrode position, the skin was cleaned using alcohol to reduce skin resistance. During EMG recording, special care was taken to prevent the connecting cables from disturbing the performing pianist. The EMG signals were ampliWed 500 times and sampled at 1,000 Hz together with SR using the MP150 system. EMG signals were Wrst notch Wltered to remove the 50 Hz (local power line frequency) interference. Then, the signals Rehearsal
Competition
a: HR
120
bpm
160
80
b: SR
0.2
mg/min/cm 2
0.4
0 1 min
Fig. 1 Typical examples of heart rate (HR, a) and sweat rate (SR, b) data. The graphs show the time series obtained during the Wrst 5 min of a participant’s performance (Beethoven: Piano Sonata No. 23 in F minor, Op. 57, 1st Mov.). The left column shows the data recorded in the rehearsal condition and the right column in the competition condition
123
were high-pass Wltered at 20 Hz, full-wave rectiWed, and low-pass Wltered at 50 Hz. Subsequently, baseline noise (i.e., mean resting EMG amplitude of pre- and post-performance phases) was removed from the signals according to Kudo and Ohtsuki (1998). To normalize these EMG data for each muscle for each participant, maximal voluntary contraction (MVC) data were obtained by asking participants to perform maximal isometric force production (5 s) after their piano performance in each condition. Participants were verbally encouraged to achieve maximal force at designated joint angles. During MVC tasks for the ED and FD muscles, the wrist joint was kept at 180°, and for the BB muscle, the elbow joint was kept at 90°. In these trials, an assisting person applied the highest possible load to help participants produce maximal Xexion or extension force. As for the TR muscle, we adopted the sustained shoulder shrug as the MVC task. An MVC value was determined as the highest mean EMG amplitude observed during the MVC task, which was obtained with a 300 (for the TR muscle) or 1,000 ms (for the other three muscles) window moving in steps of 1 ms. EMG signals recorded during performance were normalized relative to these MVC values (Fig. 2a). Prior to EMG analyses, we selected appropriate EMG data by applying systematic criteria. We Wrst performed the Smirnov–Grubbs tests (P < 0.05, with Bonferroni correction) on the percent changes in the mean resting EMG amplitude from the pre- to post-performance phase to exclude EMG data with substantial changes in baseline noise level (probably due to the detachment of electrodes). In the same way, we excluded EMG data with substantial changes in the MVC values from the rehearsal to competition condition. The selection process required us to exclude the EMG data of three participants for the TR muscle and the data of two participants for the other three muscles. The EMG magnitude was quantiWed by computing the mean EMG amplitude (%MVC) during performance. We also examined the frequency distribution of EMG signals to evaluate the level of muscle relaxation. Figure 2b shows examples of the relative frequencies that are cumulatively plotted as a function of the EMG activity level (%MVC). Here, the relative frequency at a given EMG activity level indicates the relative duration in which EMG activity was below the threshold level [muscle relaxation ratio (MRR), in % total performance duration]. Because approximately 90% of the data points fell below 50% MVC for all the muscles, we examined the MRRs at the thresholds ranging from 5 to 50 with intervals of 5% MVC. In addition, we estimated the co-contraction level of antagonistic muscles in the forearm (i.e., ED and FD muscles) based on the relative diVerence signals (RDS) as proposed by Heuer (2007). The rectiWed and low-pass Wltered (6 Hz) EMG signals were scaled to a mean of 1, and for
Exp Brain Res (2009) 199:117–126
121
Rehearsal
Competition
a 0
mV
10
Snd
FD
40 0 60
BB 0 60
TR 0
100 ms
%MVC
0 80
%MVC
20
%MVC
40
ED
%MVC
-10
Relative Frequency (%)
b 100 80 60 40 20 0 0
10
20
30
40
50
0
10
20
30
40
50
EMG Activity Level (%MVC)
50 0 -50
2
ED
n.u.
Snd
mV
c
4 2 0
FD
n.u.
0
According to Fujii et al. (2009b), we evaluated the co-contraction level by calculating the standard deviation (SD) of RDS for each piano performance [reciprocal contraction index (RCI)]. A higher RCI value indicated the stronger tendency toward reciprocal activity, whereas a lower RCI value indicated the stronger tendency toward co-contraction of the antagonistic muscles. Statistical analysis We conducted paired t tests to examine the signiWcance of diVerences in the total performance score, scores for each performance evaluation item, VAS score, autonomic measures, and RCI between the rehearsal and competition conditions. Because the data of mean EMG amplitudes did not fulWll the prerequisite for the application of parametric analyses (i.e., normal distribution), Wilcoxon-matched pairs signed-rank tests were used to determine the signiWcance of the diVerences between conditions. As for the two performance subscores, we Wrst applied an one-way multivariate ANOVA (MANOVA) to examine the eVect of condition, and then conducted paired t tests for the technical and artistic scores separately. Regarding the MRR data, we Wrst conducted a repeated-measures MANOVA with two repeated factors: condition (2 levels: rehearsal and competition) and threshold (10 levels: 5, 10, 15, 20, 25, 30, 35, 40, 45, and 50% MVC). Then multiple two-way ANOVAs were further performed on the MRR data of each muscle. When the condition £ threshold interaction eVect was found to be signiWcant, we conducted one-way ANOVAs separately for each of the 10 levels of threshold to determine the simple main eVect of condition. P value of
