Changes in Perception Accompany the Development of Music Performance Skills

Abstract

In two experiments we examined the extent to which musicians identify discrepancies between their intentions and their playing during individual practice. In the first experiment, 60 musicians representing four levels of skill development practiced a familiar piece from their own repertoire for 5 min while being audio recorded. They then listened to their recorded practice and pressed a computer key to mark moments of discrepancy between what they had intended while practicing and what they heard on the recordings. The mean rates of key presses did not differ among the four participant groups, although there were large within-group variances. In the second experiment, 13 high school and 11 expert participants from Experiment 1 returned 2 years later and listened to their original recordings, this time marking moments of discrepancy between what would be their current intentions and what they heard on their 2-year-old recordings. High school participants marked significantly more discrepancies after 2 years than they had in Experiment 1, but the mean rate of key presses among experts did not increase between Experiment 1 and Experiment 2. These results support the notion that the precision of performance goals and the acuity of perceptual discrimination are central features of musical expertise.

Keywords

errors music performance performance monitoring error monitoring

Errors are an inherent and necessary part of learning in all dimensions of human activity. In order to effectively encode and refine procedural memories (i.e., memories of how to do things), learners must experience attempts to accomplish goals, perceive the discrepancies between the results of their attempts and their intended outcomes, and adjust their behavior to accommodate those discrepancies (Herzfeld & Shadmehr, 2014; Seidler et al., 2013; Wu et al., 2014). This broad view of error in learning encompasses instances in which ongoing behavior is amenable to adjustments during the course of action (balancing on a bicycle or adjusting the pitch of a sustained tone in music) and instances in which errors are uncorrectable once the target behavior is executed (e.g., pressing a key on a computer keyboard or piano).

It seems that many novice music learners conceive of errors only as discrete events, not realizing, perhaps, that nearly all movements that unfold over time involve ongoing adjustments intended to reach discernible goals with increasing precision (Maidhof, 2013). Discussions about performance quality often frame errors in terms of their discrete consequences (falling off a bicycle, playing a note out of tune), which does not acknowledge the motor system’s use of an ongoing feedback loop capable of updating motor commands both during movement and in future attempts (Chen et al., 2013; Katahira et al., 2008).

Extant research on musicians’ practice suggests that particular aspects of practice structure and learner cognition may contribute to the efficiency and effectiveness of musicians’ time in the practice room, both for experts (Chaffin et al., 2003; Chaffin & Imreh, 1997, 2001, 2002; Duke et al., 2009; Gruson, 1988; Maynard, 2006; Williamon et al., 2002) and for developing musicians (Hallam, 2001; McPherson, 2005; Miksza, 2007; Rohwer & Polk, 2006); novice and skilled musicians who systematically manage errors demonstrate more effective practice. Empirical examinations of highly skilled musicians in the practice room suggest that expert musicians are led by a clear, artistic image of the final goal that facilitates the perception and correction of errors (Chaffin et al., 2003). These behavioral indicators of expert music learning are consistent with the results of research in other skill domains, which also show that learners who experience error making and manage errors during self-directed active practice perform better than do learners who avoid errors during the learning process (Dyre et al., 2017; Heimbeck et al., 2003; Huang et al., 2008; Wills et al., 2007).

The central nervous system receives sensory feedback that informs the generation of motor commands sent to the musculoskeletal system. When the brain generates a motor command, it also creates an efference copy: an encoded representation of expected sensory feedback that will result from the upcoming movement. As musicians’ movements unfold, the central nervous system receives proprioceptive, tactile, and auditory feedback that then is compared to the expected feedback represented in the efference copy. When there is a discrepancy between learner expectations and what actually comes about, this prediction error signals the central nervous system to modify the movement parameters to more closely approximate the intended goal (Wolpert & Ghahramani, 2000). To refine motor control, then, learners must perceive discrepancies between intentions (expectations) and outcomes. Thus, the effectiveness of practice is wholly dependent on the clarity and precision of learners’ expectations and the accuracy of their perceptions.

What varies among the performances of novices and experts is often more than simply the number of errors that occur but how quickly adjustments are made that render errors imperceptible to observers (Chen et al., 2008; Kruse-Weber & Parncutt, 2014). Chen and colleagues (2008), for example, observed that trained cellists make small positional changes in finger shape after initial contact with the string to correct the resulting pitch. Cellists who demonstrate higher perceptual ability regarding pitch also demonstrate higher performance accuracy in shifting motions, which is consistent with the notion that skilled string musicians sample the auditory effects of their movements and make physical adjustments moment to moment.

Neurophysiological investigations have shown that skilled musicians with extensive training can correct or minimize errors even before the auditory onset of an errant note (Maidhof, 2013; Ruiz et al., 2009, 2011). Maidhof and colleagues (2009) found that changes in skilled pianists’ brainwave potentials (event-related potentials, or ERPs) indicate that detection of error in rapid movement sequences actually precedes the performance of key-press errors and that incorrect key presses are performed slightly later and more softly (slower key velocity) than correct key presses. These neurological data are consistent with behavioral observations of music practice, which indicate that experts often stop or slow down in anticipation of potential errors (Duke et al., 2009).

In this article, we define errors as moments of discrepancy between movement goals (intentions) and outcomes, a definition consistent with literature examining errors in procedural memories (Shadmehr et al., 2010). This definition also has important implications for skill learning in music, as the clarity of a performer’s intentions necessarily determines the performer’s self-perceptions of accuracy in fulfilling those intentions. If, for example, a young musician’s proximal goal is to remember the C-natural (as opposed to C-sharp) in a brief exercise, their perception of accuracy will depend primarily on whether they actually played the correct pitch (or an approximation of the correct pitch). In contrast, a musician whose goal it is to connect beautifully the C-natural to the note that follows will base their perception of accuracy not only on playing the correct pitch but also on aspects of articulation and inflection that may not be a part of novices’ thinking. Failing to meet the intended goal in either case may be considered an error by the learner, again in the sense that there is a discrepancy between the performer’s intentions and outcomes, but the more experienced musician will perceive discrepancies between what they are doing and this more precise and elaborate goal.

We hypothesize that as musicians develop expertise, they develop increasing levels of auditory and physical discrimination that then shift their intentions and expectations to increasing motor refinement and the formation of more precise and elaborate goals. We sought to determine whether the development of increasing skill and discrimination leads to a concomitant decrease in the frequency of error (as defined by the performer) during practice or whether perceived error rates during practice remain relatively constant across the range of skill levels, perhaps as a result of increasing levels of discrimination and precision of musical intention.

We designed two experiments to examine musicians’ perceptions of discrepancies in their own practice. In Experiment 1, high school, college, and expert musicians recorded brief practice sessions and then immediately listened to the recordings and pressed a computer key to mark moments of discrepancy between what they had intended while practicing and what they heard on their recordings. In the second experiment, conducted 2 years later, the high school and expert participants from Experiment 1 performed the same task with their original recordings, pressing the computer key to mark moments of discrepancy between what they would intend if this were a recording of their current playing and what they heard on their 2-year-old recordings.

Experiment 1

In this experiment, we examined the extent to which musicians at different levels of experience identify discrepancies between their intentions and their actual playing during individual practice.

Method

Participants

We recruited 60 musicians who played either violin, viola, cello, or bass; 15 were high school students who had played their primary instrument for at least 3 consecutive years (experience, M = 9.7 years; age, M = 16.2 years; 10 female), 15 were undergraduate music majors (experience, M = 10.7 years; age, M = 20.1 years; 9 female), 15 were graduate performance majors (experience, M = 16.7 years; age, M = 25.8 years; 10 female), and 15 were expert musicians working in either an orchestra or university position (experience, M = 32.7 years; age, M = 40.1 years, 7 female).¹

Procedure

Participants were recruited by email from professional and educational music organizations within the community; all volunteered to participate in the study and received no compensation for their participation. The Institutional Review Board of the University of Texas at Austin approved all procedures.

We tested participants in individual sessions that were scheduled at their convenience. We asked participants to bring with them to the test session an étude or piece that they were practicing at the time that was prepared beyond the note-reading stage but was not yet performance ready. After a brief orientation to the testing room and a warm-up period, we recorded participants practicing their selected piece for approximately 5 min using a MacBook computer running QuickTime Player 10.4.

Immediately following the 5-min practice period, participants listened to the audio recording of their practicing through Bose QuietComfort 2 Acoustic Noise Cancelling headphones and, as they listened, pressed a designated computer key each time they heard a discrepancy between what they had intended to do while practicing and what actually had occurred. We asked them to press the key for all discrepancies, regardless of whether they had noticed the discrepancies while they were practicing or only while they were listening to the recording.

We used SCRIBE 4 behavior analysis software (Duke & Stammen, 2010) to calculate the rate of key presses per minute, which served as the primary dependent measure of the study. We calculated the rate of discrepancies for each participant by dividing the total number of key presses by the total amount of time spent practicing (most sessions were a few seconds longer or shorter than 5 min).

Participants listened to their recordings only once, after which we asked them to describe (a) whether they felt they had marked accurately the discrepancies they heard, (b) the extent to which the discrepancies they identified were heard during practice or only while listening to the recording, (c) the nature of the discrepancies they identified, and (d) how long they had been working on the piece.² These questions were presented informally in a way that allowed participants to respond freely about their experience.

The primary author was present throughout the testing procedure for all participants and took written notes of participants’ responses. After a preliminary review of the participants’ responses about the nature of the discrepancies they identified, we grouped their responses into five categories: intonation (statements pertaining to precise finger or hand placement, playing in tune, or shifting in tune), tone (statements about quality of sound, bow control, or articulation), expression (statements about phrasing, inflection, vibrato, and dynamics), notes (statements about playing the correct notes, memorizing the correct notes, or creating playable fingerings), and timing (statements about tempo, fitting rhythms to a tempo, and coordinating left and right hands).

A trained reliability observer read verbatim transcripts of one third (n = 20) of the interviews and coded participants’ responses using the same category system. Overall reliability between the primary author’s codes and the observer’s codes of the verbatim transcripts was 87%.³

One important variable that may have affected the rates of perceived discrepancies among the four groups of participants is the suitability of the repertoire that each performer practiced for the study. On the basis of their performance during each 5-min practice recording, the first author, an expert string player and teacher, rated the suitability of each participant’s repertoire using a 5-point Likert-type scale, with 1 representing not well suited and 5 representing very well suited. Another expert string teacher (naive to the purpose of the study) also independently listened to all 60 audio recordings and evaluated the repertoire’s suitability using the same scale.

The level of agreement between the primary author’s ratings of repertoire suitability and the expert observer’s ratings was high: We both indicated that the repertoire played by all experts, all graduate students, all undergraduate students, and 12 of the 15 high school students was well suited to their apparent level of technical proficiency (giving each recording a rating of either 4 or 5 on the scale described previously). Of the three high school students whose pieces were thought to be less well suited to their technical proficiency, we both rated the suitability in the middle of the scale (giving a rating of 3 on the scale described previously). We both agreed that there were no participants playing repertoire that was not well suited to their technical proficiency (there were no ratings of 1 or 2).

Results

Accuracy of the task

In order to confirm that participants believed that they had completed the task accurately, we asked at the end of the listening session whether the key presses accurately reflected the number of discrepancies that participants had perceived while listening. Fifty-three of the 60 participants indicated that their key presses accurately reflected the discrepancies they had heard. Nearly half the participants volunteered that their key presses were slightly delayed, which was not surprising given the nature of the task.

Only seven of the 60 participants indicated that there were small inaccuracies in their records of key presses,⁴ so we included all of the data as recorded. One expert participant was omitted from all key-press rate analysis due to a technical error while collecting data.

Mean rates of key presses

We compared the rates of perceived discrepancies (rates of computer key presses) among the four groups of participants in a one-way analysis of variance (ANOVA) and found no significant differences in mean key-press rates among the four experience levels, F(3, 55) = 0.28, p = .84. The overall group means were nearly identical among the groups, although there were large variances within each group. Figure 1 presents the individual data points for each participant in each experience category and the mean values for each experience level.

Figure 1.

Key-press rate per minute by participant experience level. Bars represent group means. Point locations along the y-axis represent the key-press rates of individual participants.

Because experienced musicians often work on pieces they already have performed, we were interested in whether the participants in the present study were working on new repertoire or were reviewing repertoire that had been learned and performed previously. Six of the experts and nine of the graduate students had selected repertoire for the study that they were learning for the first time. The remaining nine expert participants and six graduate students had selected music that they had performed at some time in the past and were “relearning.”

Since precisely half (n = 15) of the graduate students and experts combined practiced new repertoire and half (n = 15) were relearning a familiar piece, we decided to compare the rates of discrepancies between these two serendipitous conditions. We examined just the data on rates of perceived discrepancies between two groups: graduate students and experts who were working on a piece for the first time, and graduate students and experts working on repertoire that they had performed previously. Using an independent-samples t test, we found no difference in mean rates of key presses between participants relearning a piece (M = 5.52, SD = 2.83) and participants who were learning a piece for the first time (M = 5.18, SD = 3.92), t(27) = 0.27, p = .79.

We were also interested in whether the amount of time participants had practiced their pieces prior to the testing session would affect rates of perceived discrepancies. We grouped the participants using the following categories: 1 to 7 days preparation (n = 16), 8 to 60 days preparation (n = 15), 61 to 365 days preparation (n = 13), and relearning repertoire that was performed previously (n = 16). We compared the rates of perceived discrepancies in a one-way ANOVA and again found no significant difference in mean rates of key presses among the four levels of preparation, F(3, 55) = 0.39, p = .76.

Participants estimated the percentage of discrepancies they heard while listening to the recording that they had not heard while practicing. The experts’ mean estimate was lower than the means for the other three groups; see Supplementary Table S1 (included with the online version of the article). The fact that nearly all participants reported hearing most of these discrepancies while they were practicing supports the notion that participants understood the nature of the task and that their reported discrepancies are in fact indicative of their perceptions while practicing.

We also looked at the variability in the time intervals between key presses by calculating each participant’s coefficient of variation among the key-press intervals and found no significant differences among the four groups; see Supplementary Table S2 (included with the online version of the article). For nearly all participants, key presses were distributed relatively evenly across each practice session.

Types of perceived discrepancies

After listening to the recordings and marking points of discrepancy, participants described the types of discrepancies they heard while listening and during practice. We coded each participant’s comments according to the categories they mentioned; most participants made comments in two or more categories. We examined the distribution of discrepancy types among the four experience levels in a chi-square analysis. The numbers of discrepancies of each type in each experience category are presented in Supplementary Table S3 (included with the online version of the article). We adjusted the number of categories in the statistical analysis to accommodate small numbers of observations in several cells. The resulting categories used in the chi-square test were intonation, tone, and other. We found no significant relationship between the experience level and the types of discrepancies participants described, χ²(6, N = 122) = 0.91, p = .99.

Discussion

Our results indicate that the number of discrepancies string musicians perceive in their own practice sessions does not vary systematically across levels of experience and expertise. Given current understanding of prediction error in skill learning (Diedrichsen et al., 2005, 2010; Seidler et al., 2013), we posit that as musicians learn over time, increasing levels of auditory and physical discrimination shift performers’ intentions and expectations to increasing levels of refinement. Thus, experts’ detection of discrepancies between intentions and outcomes may reflect clearer intentions, more vivid expectations, and finer levels of discrimination than are defined by less skillful players, though this conjecture must be subjected to further empirical scrutiny.

Comments from an expert violinist and a high school violinist in the current study also illustrate this point. Both indicated that the discrepancies they identified were “always about intonation,” and both demonstrated similar rates of key presses while listening to their recordings. The high school student’s intonation was generally worse than that of the expert, and he seldom made pitch adjustments after the onset of a mistuned note. The expert’s intonation errors were not only smaller in magnitude than those of the high school student, but inaccuracies were corrected immediately following note onsets. Although both musicians perceived intonation discrepancies at similar rates, the quality of their intonation varied with respect to accuracy at note onset and speed of correction when errors occurred. The expert’s intonation was clearly superior to the high school student’s, but their perceptions of their own work were quite similar with respect to the number of intonation discrepancies they perceived.

The primary dependent variable in the present study was not the number of errors in participants’ practicing but rather the rate at which participants perceived discrepancies between what they had intended and what they heard while practicing repertoire that was well suited to their technical capabilities. It is notable that among the most experienced performers in the sample (expert musicians and graduate students), the rates of perceived discrepancies did not differ between those who were learning a piece for the first time and those who were relearning a piece that they had performed at some time in the past. For these musicians, the mean rate of perceived discrepancies was not dependent on the familiarity of the repertoire.

These data suggest that with respect to practice sessions, the progression from novice to expert is characterized not primarily by a diminution in the rate of error making but rather by an apparent increase in the elaboration and refinement of intentions and expectations. Data from Experiment 1 suggest that as musicians develop increasing levels of auditory discrimination, they perceive finer discrepancies between what they intend and what they do. These revised expectations guide practice behavior and lead to increasing levels of motor refinement over time.

Experiment 2

Approximately 2 years after the completion of Experiment 1, we recruited 24 of the Experiment 1 participants and asked them to repeat the task of identifying discrepancies between their intentions and their actual playing in the 2-year-old recordings. In Experiment 2, participants again indicated points of discrepancy but now between what they believed would be their current intentions and their practice on their 2-year-old recording.

Method

Participants

We contacted the 15 high school musicians and the 15 expert musicians who had participated in Experiment 1, approximately 2 years after their first testing session (test interval, M = 1.8 years, range = 1.5–2.3 years). Thirteen of the original 15 student participants (experience, M = 11.2 years; age, M = 17.8 years; 9 female) and 11 of the original 15 expert musicians (experience, M = 35.0 years; age, M = 41.2 years; 6 female) agreed to participate in Experiment 2.

Five of the student participants were in high school at the time of Experiment 1 and were still in high school at the time of Experiment 2. The remaining student participants were in high school at the time of Experiment 1 and had begun college classes at the time of Experiment 2 (five as music majors, one as an undeclared major with the intent of becoming a music major, and two as nonmusic majors). All expert musicians from Experiment 1 still were performing professionally at the time of Experiment 2.

All participants volunteered to participate in the study and received no compensation for their participation. The institutional review board of the University of Texas at Austin approved all procedures.

Procedure

We tested participants in individual sessions that were scheduled at their convenience. Participants listened to audio recordings of their own 5-min practice sessions that had been recorded 2 years earlier (during Experiment 1) and, as before, pressed a designated computer key to mark each time they heard a discrepancy between what would be their current musical intentions and what they heard on the recording. All procedures were identical to those of Experiment 1.

The primary author was present throughout the testing procedure for all participants and took written notes of participants’ responses. We again coded participants’ responses about the nature of the discrepancies they perceived. Five of six categories (intonation, tone, expression, notes, and timing) were established in Experiment 1 after examining participants’ open-ended responses about the nature of discrepancies they perceived in their practice. In Experiment 2, many participants, in addition to identifying aspects of playing that led to their key presses, offered comments about how they were practicing during the recordings. Thus we added the sixth category (practice) to our coding system. Most participants made comments in two or more categories.

A trained reliability observer read verbatim transcripts of 20% of the interviews (n = 8) and coded participants’ responses. Overall reliability was 88%.⁵ The same reliability observer again coded participants’ comments, this time from the primary author’s written notes for all (N = 24) of the participants’ statements. Overall reliability was 94%.⁶

Results

Accuracy of the task

We asked participants at the end of the listening session whether the key presses accurately reflected the number of discrepancies that they had perceived while listening. All 24 participants indicated that their key presses accurately reflected the discrepancies they had heard.

Five of the 24 participants volunteered that key presses were slightly delayed (i.e., they pressed the key after a few moments had passed since detecting the discrepancy). Two of the 24 participants indicated that there were small inaccuracies in their records of key presses.⁷ Given the small numbers of key-press errors that these participants described, we included all of the data as recorded.

Mean rates of key presses

We compared participants’ key-press rates from Experiment 1, just after they had recorded their practice, to their rates in the current experiment and calculated a difference score for each participant. Figure 2 shows each participant’s key-press rate in Experiments 1 and 2. All but one student identified more discrepancies in Experiment 2 than they had identified in Experiment 1, whereas nearly all expert musicians identified discrepancies in Experiment 2 at approximately the same rate as they had in Experiment 1. Two expert participants identified many more discrepancies in Experiment 2 than they had in Experiment 1.

Figure 2.

Mean key-press rate per minute collected during Experiment 1 and Experiment 2. Each gray/black bar pair represents data from an individual participant; gray bars represent participants’ rates of key presses collected during Experiment 1, and black bars represent participants’ rates of key presses collected 2 years later, during Experiment 2. Participants in each group (students, experts) are placed in order of increasing mean key-press rates from left to right. Outliers are indicated with asterisks.

We calculated individual key-press rate difference scores between Experiments 1 and 2 and found three outliers among the participants (whose difference scores were 1.5 times the interquartile range above the third quartile or below the first quartile), one student participant with a difference score of −8.35 key presses per minute, and two expert participants with difference scores of 11.28 and 15.69 key presses per minute.⁸ We excluded these participants’ data from all subsequent analyses.

We compared the mean rates of key presses by student and expert participants taken from Experiment 1 and 2 using a two-factor mixed-design ANOVA with repeated measures on time point. We found no significant difference between the mean rates of key presses by students and experts overall, F(1, 19) = 1.53, p = .23. We found a significant difference between the mean rates of key presses attributable to time points, F(1, 19) = 12.19, p = .002, η_p² = .39, and a significant interaction between group and time point, F(1, 19) = 12.89, p = .002, η_p² = .40. Experts identified discrepancies at highly similar rates during Experiments 1 and 2 (M_Exp1 = 4.65, SD_Exp1 = 2.62; M_Exp2 = 4.61, SD_Exp2 = 3.22), but students identified significantly more discrepancies during Experiment 2 than they had during Experiment 1 (M_Exp1 = 4.46, SD_Exp1 = 2.33; M_Exp2 = 7.80, SD_Exp2 = 3.48). In both experiments, within-group variances were high. Figure 3 shows the mean key-press rates from Experiments 1 and 2 for student and expert participants.

Figure 3.

Participants’ rates of key presses in Experiment 1 and 2 by experience level. Gray bars represent group means. Points represent individual participants’ mean key-press rates per minute.

We examined individual participants’ patterns of key presses and compared these to the patterns of key presses recorded in the first experiment, 2 years earlier. We found that for all participants, key presses were distributed throughout the practice sessions, and results from Experiment 1 and 2 were very similar.⁹

After completing the key-press task, participants gave their impression about the types of discrepancies they heard on the recording. We coded each participant’s comments according to the categories they mentioned; most participants made comments in two or more categories. The numbers of discrepancies of each type in each experience level are presented in Supplementary Table S4 (included with the online version of the article). As with Experiment 1, student and expert participants identified similar types of discrepancies.

Discussion

In Experiment 2, the high school and expert participants returned 2 years after making a practice recording and listened to their original recordings, this time marking moments of discrepancy between what would be their current intentions and what they heard on the recordings. High school participants marked significantly more discrepancies after 2 years, but the mean rate of key presses among experts did not increase. As less experienced musicians developed over time, they identified more discrepancies between what they would have intended and their performance in their 2-year-old recording.

These data are consistent with the notion that the development of musical expertise involves not only increasing levels of physical skill but also commensurately increasing levels of intentional precision and refinement of auditory discrimination. This interpretation is in keeping with our current understanding of motor learning: As learners attempt to accomplish tangible goals, the sensory feedback they receive about the effects of their movements serves to modify motor commands and update procedural memories as movements unfold and during future attempts to accomplish similar goals. This raises an important issue about musical development, namely, that a major inhibitor of young learners’ progress may be a lack of refinement in goal setting. The motor system can obtain only as much information from a given movement as the movement’s goal is clear in the mind of the learner. When the learner’s goals are clearer, there is a greater opportunity to gain more meaningful information from each iteration of the movement.

The data from this experiment are consistent with the hypothesis that refinement of physical skills occurs in concert with refinement of performance goals and with increasingly astute auditory perceptions. These results contribute not only to the formulation of a model of expertise in music but also to pedagogical practice. It is often the case that music teachers focus in individual lessons on giving clear instructions and pointed feedback intended to modify various aspects of students’ playing. Less often are learners directed to think carefully and explain clearly their intentions about what they set out to accomplish in each performance trial (Duke, 2012).

General Discussion

The development of music performance skills requires the refinement of both motor behavior and perceptual skills through an iterative process of goal setting, performance, and self-evaluation. Technical facility and musical artistry require the formulation of vividly clear performance goals and high levels of auditory and physical discrimination. Put more simply, artist-level performers have a precise idea of what they want to accomplish, and they keenly identify discrepancies between their goals and what they do moment to moment. After eliminating most of the discrepancies encountered during practice, artist-level performers are able to play in a way that may seem flawless to even the most avid listener. In this sense, learning to practice effectively seems better characterized not as a decrease in error making but as an increase in intentional clarity and perceptual acuity.

This is consistent with a feed-forward model of motor control originally suggested by Robinson (1973), which hypothesized that the central nervous system compares efferent predictions to afferent sensory feedback, weighing these variables with regard to uncertainty. Discrepancies between what was expected and what is experienced signal a prediction error, which updates the motor command through a process mathematically represented by a Kalman filter (Kalman, 1960).

When taken together, results from both of our experiments indicate that two important variables, clarity of musical intentions and perceptual acuity, differentiate artist-level musicians from less experienced musicians. This seems especially notable in light of the fact that music performance instruction at all levels often focuses more on what musicians do and less on what musicians intend to do or what musicians perceive about what they do (Colprit, 2000). Thus, a major impediment in developing skill is learners’ ability to know precisely what they intend to do and to know precisely whether they actually have done it.

It is understandable that the beginning stages of skill development in any domain must focus initially on learners’ reaching approximations of what ultimately will become highly refined movements. But how learners define goals and focus attention has everything to do with the effectiveness of practice over time, and directing learners’ attention to defining goals and evaluating performance outcomes is the purview of the teacher. The goal of moving the bow, for example, may focus only on the bow arm, where the bow contacts the string, and the angle of the bow relative to the string. But this movement produces not only proprioceptive, haptic, and visual feedback but auditory feedback (the ultimate goal of the bow movement), as well. If novice learners remain focused on the movements themselves, rather than the sounds those movements produce, it is understandable that their progress will be limited compared to that of novice learners who focus also on the sound of the instrument (Duke et al., 2011).

Novice learners embarking on their musical studies often expect to play increasingly difficult repertoire over time, but more important is an understanding that over time, they will develop increasingly more refined expectations. What varies among the performances of novices and experts is often more than the number of obvious errors that occur but rather the depth of what musicians hear, which in turn informs what they do.

In both experiments, we observed a great deal of variability in the key-press rates among participants in each group. It is notable that in Experiment 1, when participants were listening to themselves practicing a piece of their choice, the rates of key presses were unrelated to how long participants’ had been working on their pieces. Similarly, key-press rates were not related to whether participants were relearning a piece that they had performed previously. The ratio of each participant’s repertoire level relative to their skill level seems like a probable source of variability, as does the length of time participants had been working on the repertoire they brought to the testing session, but the data were not consistent with these possibilities. This finding certainly warrants further attention.

The results of Experiment 2 must be considered carefully in light of the small sample size and the omission of outliers in the statistical analysis. It seems important that this type of longitudinal comparison be conducted with other participants to confirm the results we obtained; however, data from this current study support the idea that as musicians gain increasing levels of skill, there is a commensurate increase in auditory discrimination. These results are consistent with investigations comparing musicians’ playing ability with their ability to detect small differences in recorded performances that have indeed found a correlation between these two variables (Hamilton et al., 2019).

Our results contribute not only to the formulation of a model of expertise in music performance but also to pedagogical practice. Music teachers often set goals of giving clear instructions and pointed feedback (telling) intended to modify various aspects of their students’ playing. Equally important, though, is asking learners to think carefully and explain clearly their intentions about what they set out to accomplish in each performance trial. For students to practice like their teachers, they must establish a level of acceptability for their own playing that is consistent with their teachers’ refined, artistic image and subsequently must perceive small discrepancies between what they are doing and this artistic image.

As musicians develop expertise, their standards of acceptability become increasingly refined, which in turn leads to successful attempts toward increasingly refined tangible goals. It is not simply that expert musicians make fewer errors but that they have a rich and vivid standard of acceptability along with the perceptual skills to perceive very fine discrepancies between these refined intentions and the outcomes of their movements, an important message for developing musicians to understand.

Supplemental Material

Supplementary_Materials_Revised – Supplemental material for Changes in Perception Accompany the Development of Music Performance Skills

Supplemental material, Supplementary_Materials_Revised for Changes in Perception Accompany the Development of Music Performance Skills by Lani M. Hamilton and Robert A. Duke in Journal of Research in Music Education

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

The appendices are available in the online version of the article.

Notes

Author Biographies

Lani M. Hamilton is assistant professor of music education at the University of Missouri–Kansas City Conservatory. Her research interests include music practice, motor skill development, and the perception and monitoring of errors.

Robert A. Duke is professor of music and human learning at the University of Texas at Austin. His research interests include the formation and refinement of procedural memories, skill development, and expert thinking.

References

Chaffin

Imreh

(1997). “Pulling teeth and torture”: Musical memory and problem solving. Thinking & Reasoning, 3, 315–336. https://doi.org/10.1080/135467897394310

Chaffin

Imreh

(2001). A comparison of practice and self-report as sources of information about the goals of expert practice. Psychology of Music, 29, 39–69. https://doi.org/10.1177/0305735601291004

Chaffin

Imreh

(2002). Practicing perfection: Piano performance as expert memory. Psychological Science, 13, 342–349. https://doi.org/10.1111/j.0956-7976.2002.00462.x

Chaffin

Imreh

Lemieux

A. F.

Chen

(2003). “Seeing the big picture”: Piano practice as expert problem solving. Music Perception: An Interdisciplinary Journal, 20, 465–490. https://doi.org/10.1525/mp.2003.20.4.465

Chen

Woollacott

Moore

G. P.

(2013). Stochastic aspects of motor behavior and their dependence on auditory feedback in experienced cellists. Frontiers in Human Neuroscience, 7, 419. https://doi.org/10.3389/fnhum.2013.00419

Chen

Woollacott

M. H.

Pologe

Moore

G. P.

(2008). Pitch and space maps of skilled cellists: Accuracy, variability, and error correction. Experimental Brain Research, 188, 493–503. https://doi.org/10.1007/s00221-008-1380-2

Colprit

E. J.

(2000). Observation and analysis of Suzuki string teaching. Journal of Research in Music Education, 48, 206–221. https://doi.org/10.2307/3345394

Diedrichsen

Hashambhoy

Rane

Shadmehr

(2005). Neural correlates of reach errors. Journal of Neuroscience, 25, 9919–9931. https://doi.org/10.1523/JNEUROSCI.1874-05.2005

Diedrichsen

White

Newman

Lally

(2010). Use-dependent and error-based learning of motor behaviors. Journal of Neuroscience, 30, 5159–5166. https://doi.org/10.1523/JNEUROSCI.5406-09.2010

10.

Duke

R. A.

(2012). Their own best teachers: How we help and hinder the development of learners’ independence. Music Educators Journal, 99(2), 36–41. https://doi.org/10.1177/0027432112458956

11.

Duke

R. A.

Cash

C. D.

Allen

S. E.

(2011). Focus of attention affects performance of motor skills in music. Journal of Research in Music Education, 59, 44–55. https://doi.org/10.1177/0022429410396093

12.

Duke

R. A.

Simmons

A. L.

Cash

C. D.

(2009). It’s not how much; it’s how: Characteristics of practice behavior and retention of performance skills. Journal of Research in Music Education, 56, 310–321. https://doi.org/10.1177/0022429408328851

13.

Duke

R. A.

Stammen

(2010). Scribe 4 (Version 4.1.1). Learning and Behavior Resources.

14.

Dyre

Tabor

Ringsted

Tolsgaard

M. G.

(2017). Imperfect practice makes perfect: Error management training improves transfer of learning. Medical Education, 51, 196–206. https://doi.org/10.1111/medu.13208

15.

Gruson

L. M.

(1988). Rehearsal skill and musical competence: Does practice make perfect? In Sloboda

J. A.

(Ed.), Generative processes in music: The psychology of performance, improvisation, and composition. (pp. 91–112). Oxford University Press.

16.

Hallam

(2001). The development of expertise in young musicians: Strategy use, knowledge acquisition and individual diversity. Music Education Research, 3, 7–23. https://doi.org/10.1080/14613800020029914

17.

Hamilton

L. M.

Duke

R. A.

Simmons

A. L.

Allen

S. E.

(2019, March). Perception of timing, pitch, and loudness differences in artist-level performances: Differences among musicians with varied levels of expertise. Paper presented at the 23rd Meeting of the Clifford K. Madsen International Symposium on Research in Music Behavior, Estes Park, CO, United States.

18.

Heimbeck

Frese

Sonnentag

Keith

(2003). Integrating errors into the training process: The Function of error management instructions and the role of goal orientation. Personnel Psychology, 56, 333–361. https://doi.org/10.1111/j.1744-6570.2003.tb00153.x

19.

Herzfeld

D. J.

Shadmehr

(2014). Motor variability is not noise, but grist for the learning mill. Nature Neuroscience, 17, 149–150. https://doi.org/10.1038/nn.3633

20.

Huang

V. S.

Shadmehr

Diedrichsen

(2008). Active learning: Learning a motor skill without a coach. Journal of Neurophysiology, 100, 879–887. https://doi.org/10.1152/jn.01095.2007

21.

Kalman

R. E.

(1960). A New approach to linear filtering and prediction problems. Journal of Basic Engineering, 82(1), 35–45. https://doi.org/10.1115/1.3662552

22.

Katahira

Abla

Masuda

Okanoya

(2008). Feedback-based error monitoring processes during musical performance: An ERP study. Neuroscience Research, 61, 120–128. https://doi.org/10.1016/j.neures.2008.02.001

23.

Kruse-Weber

Parncutt

(2014). Error management for musicians: an interdisciplinary conceptual framework. Frontiers in Psychology, 5. https://doi.org/10.3389/fpsyg.2014.00777

24.

Maidhof

(2013). Error monitoring in musicians. Frontiers in Human Neuroscience, 7. https://doi.org/10.3389/fnhum.2013.00401

25.

Maidhof

Rieger

Prinz

Koelsch

(2009). Nobody is perfect: ERP effects prior to performance errors in musicians indicate fast monitoring processes. PLOS ONE, 4(4), e5032. https://doi.org/10.1371/journal.pone.0005032

26.

Maynard

(2006). The role of repetition in the practice sessions of artist-teachers and their students. Bulletin of the Council for Research in Music Education, 167, 61–72.

27.

McPherson

G. E.

(2005). From child to musician: skill development during the beginning stages of learning an instrument. Psychology of Music, 33, 5–35. https://doi.org/10.1177/0305735605048012

28.

Miksza

(2007). Effective Practice. Journal of Research in Music Education, 55, 359–375. https://doi.org/10.1177/0022429408317513

29.

Robinson

D. A.

(1973). Models of the saccadic eye movement control system. Kybernetik, 14(2), 71–83. https://doi.org/10.1007/BF00288906

30.

Rohwer

Polk

(2006). Practice behaviors of eight-grade instrumental musicians. Journal of Research in Music Education, 54, 350–362. https://doi.org/10.1177/002242940605400407

31.

Ruiz

M. H.

Jabusch

H.-C.

Altenmüller

(2009). Detecting wrong notes in advance: Neuronal correlates of error monitoring in pianists. Cerebral Cortex, 19, 2625–2639. https://doi.org/10.1093/cercor/bhp021

32.

Ruiz

M. H.

Strübing

Jabusch

H.-C.

Altenmüller

(2011). EEG oscillatory patterns are associated with error prediction during music performance and are altered in musician’s dystonia. NeuroImage, 55, 1791–1803. https://doi.org/10.1016/j.neuroimage.2010.12.050

33.

Seidler

R. D.

Kwak

Fling

B. W.

Bernard

J. A.

(2013). Neurocognitive mechanisms of error-based motor learning. Advances in Experimental Medicine and Biology, 782. https://doi.org/10.1007/978-1-4614-5465-6_3

34.

Shadmehr

Smith

M. A.

Krakauer

J. W.

(2010). Error correction, sensory prediction, and adaptation in motor control. Annual Review of Neuroscience, 33, 89–108. https://doi.org/10.1146/annurev-neuro-060909-153135

35.

Williamon

Valentine

(2002). Shifting the focus of attention between levels of musical structure. European Journal of Cognitive Psychology, 14, 493–520. https://doi.org/10.1080/09541440143000221

36.

Wills

A. J.

Lavric

Croft

G. S.

Hodgson

T. L.

(2007). Predictive learning, prediction errors, and attention: Evidence from event-related potentials and eye tracking. Journal of Cognitive Neuroscience, 19, 843–854. https://doi.org/10.1162/jocn.2007.19.5.843

37.

Wolpert

D. M.

Ghahramani

(2000). Computational principles of movement neuroscience. Nature Neuroscience, 3, 1212–1217. https://doi.org/10.1038/81497

38.

H. G.

Miyamoto

Y. R.

Castro

L. N. G.

Ölveczky

B. P.

Smith

M. A.

(2014). Temporal structure of motor variability is dynamically regulated and predicts motor learning ability. Nature Neuroscience, 17, 312–321. https://doi.org/10.1038/nn.3616

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.23 MB