What can studying musicians tell us about motor control of the hand?


Dr Alan H. D. Watson, School of Biosciences, Cardiff University, Museum Avenue, Cardiff CF10 3US, UK. E: watsona@cardiff.ac.uk


Most standard accounts of human anatomy and physiology are designed to meet the requirements of medical education and therefore consider their subject matter from the standpoint of typical rather than outstanding levels of performance. To understand how high levels of skill are developed and maintained, it is necessary to study elite groups such as professional athletes or musicians. This can lead to the rediscovery of arcane knowledge that has fallen into neglect through a lack of appreciation of its significance. For example, although variability in the muscles and tendons of the hand was well known in the nineteenth and early twentieth centuries, it is through recent studies of musicians that its practical significance has become better appreciated. From even a cursory acquaintance with the training methods of sportsmen and women, dancers and musicians, it is clear that sophisticated motor skills are developed only at the cost of a great deal of time and effort. Over a lifetime of performance, musicians arguably spend more time in skill acquisition than almost any other group and offer a number of unique advantages for the study of motor control. Such intensive training not only modifies cortical maps but may even affect the gross morphology of the central nervous system. There is also evidence that in certain individuals this process can become maladaptive. Recent studies of musicians suggest that intensive training can lead to the appearance of ambiguities in the cortical somatosensory representation of the hand that may be associated with the development of focal dystonia; a condition to which musicians are particularly prone. The realization that changes in cortical maps may underlie dystonia has led to the development of new approaches to its treatment, which may ultimately benefit musicians and non-musicians alike.


Over the last few decades, a growing interest in the biology of sporting performance in elite athletes has stimulated new research into whole body physiology, tissue structure and gross anatomy that has led to significant advances both in training and in the effective treatment of injury. Although it has made less impact in the field of neuroscience, recent research on another elite group, instrumental musicians, suggests that it may provide considerable insights into motor control and higher sensory processing. The degree of motor control demanded of musicians makes them excellent subjects for the study of skill acquisition and the exploration of the limits of performance. Indeed as a group they present some unique advantages in this respect. Many begin their training at a very early age, often before the age of 7 years, when the circuitry of the central nervous system is still very plastic. To perform at the highest level, their motor skills must be honed by many hours of daily practice, but as the activity involved is not energetically demanding, physical fatigue is not generally a limiting factor. As a consequence, intensive training can be maintained over almost the entire lifespan. Practice sessions are almost entirely devoted to skill acquisition and are not diluted by the extensive periods of strength conditioning typical of athletic training. In keyboard and string players, motor activity is focused primarily on the muscles which control the hand. The object of the trained movements is to depress the keys or strings of the instrument. These are discrete events and can provide a precise means of assessing outcome as the duration, timing and force applied can easily be measured. The fingers are mapped at high resolution in the primary sensory cortex and the hand is also well represented in the motor cortex, so changes in the organization of these regions are easy to detect against the relatively constant representations of regions of the body with contiguous cortical representations. Interestingly, in some instances these changes appear to be maladaptive, and observations of the cortex of musicians suffering from focal dystonia have contributed new theories about the origin of this condition, which are discussed later.

The object of this review is to consider some of the insights into the motor control of the hand that have been gained, or placed in a clearer context, by the study of musicians. As this is a relatively new research area, the current literature is still quite small. Many of the studies involve relatively few subjects who may play a variety of instruments that make different demands on the hands. Consequently, variability between subjects is a significant factor that should be taken into consideration when evaluating the results. Nevertheless, the information that has emerged not only has implications for the training of musicians themselves, but also for other groups who seek to develop advanced physical skills.

Anatomical limitations to independent finger control

The hand evolved primarily for grasping, but in modern life it is used for a great deal more and the demands made of it in instrumental performance can push its design specifications to their limits. Standard descriptions of the actions of the muscles controlling the hand can give a misleading impression of the degree to which the fingers can be controlled independently. A number of factors conspire to limit this. For example, the muscle bellies of flexor digitorum profundus and flexor digitorum superficialis and of extensor digitorum communis that act on different fingers are partially fused, so contraction in any one will produce some passive movement of the others. The degree of fusion varies between individuals and in the case of the flexors, slips of muscle may even cross between the deep and superficial muscle masses. Another significant constraint is the degree to which the subdivisions of the muscle can be controlled independently by the nervous system. Studies of the flexor digitorum profundus and extensor digitorum communis have revealed that there is frequently a strong synchronization between the motorneurones controlling different muscle bellies (Schmied et al. 1993, 2000; Reilly et al. 2004). This is most marked for the slips supplying adjacent fingers and is particularly strong in those acting on the ring and little fingers. With practice, however, it is possible to increase or decrease the degree of synchronization, which will enhance the level of control that can be exerted over individual fingers. This observation is consistent with biomechanical studies of pianists, which demonstrate that professionals not only exhibit more independence of finger movement than amateurs but are also able to control the duration and force of the movements much more precisely (Parlitz et al. 1998). The amateurs showed a considerable degree of inappropriate coactivation not only of fingers, but also between the fingers and the thumb. The question of whether synchrony exists between activity of motor neurones innervating muscles of the fingers and of the thumb has not been investigated physiologically, but it would be consistent with what is known of the descending cortical control of movement (see below).

Nineteenth- and early twentieth-century descriptions of human anatomy paid considerable attention to anatomical variation. Although variations in the paths of nerves and blood vessels that can pose problems for surgery have retained a place in contemporary textbooks, descriptions of variations in muscles and tendons are often neglected. Though these may appear esoteric and of little relevance among the general population, they can be highly significant for those who must carry out skilled movements that lie close to the limit of what is physically possible. In some individuals, certain muscles or tendons may be absent entirely, or their form or attachments may differ from the norm. For example, the flexor digitorum superficialis tendon to the little finger is missing in about 5% of hands (Miller et al. 2003). The intrinsic muscles of the hand make a major contribution to finger dexterity and the independence of finger movement, so it is particularly significant that variation is frequently observed in the attachments of the lumbricals (Fig. 1). These muscles allow the two terminal (interphalangeal) joints of the fingers to be straightened while the knuckle (metacarpo-phalangeal) joint is flexed. As many as 50% of hands do not show the ‘standard’ pattern (Mehta & Gardner, 1961; Perkins & Hast, 1993). In up to one-third of hands, the tendon of the third lumbrical divides to insert into both the ring and the middle fingers (Fig. 1A), whereas in a small number of cases there is no lumbrical insertion on the little finger at all. Therefore, regardless of the degree of training, not all musicians are capable of the same finger movements. Some practical examples of the problems this produces for pianists and how they be overcome are discussed by Beauchamp (2003b,c).

Figure 1.

Variability in muscle insertions and tendons of the hand. The lumbrical muscles provide one of the commonest sources of anatomical variation in the hand. The standard pattern is shown in A, although this is probably present in less than 40% of hands. The four lumbrical muscles arise from the tendons of flexor digitorum profundus and insert into the radial side of each finger. The inset shows variations involving double tendon insertions. The incidence of these is of the order of 35–40% for the 3rd lumbrical and 5–10% for the 4th. (B) Tendinous linkage between the tendons of flexor pollicis longus and flexor digitorum profundus of the index finger (Linburg–Comstock syndrome). This can cause problems for pianists in particular.

Connections between tendons running to different fingers are a significant feature of the hand. The best known example of this is the extensive pattern of linkages between the tendons of the common extensor muscle of the fingers, which create particular problems for independent extension of the ring finger. In addition, the tendon of extensor digiti minimi, which is usually entirely separate, is sometimes connected to the common extensor tendon of the little finger (Allieu et al. 1998) This makes it impossible to straighten this digit when the others are flexed. In one study, this anomaly was found in 18% of individuals examined while in a further 34%, the extensor digiti minimi was absent entirely (Baker et al. 1981). What is less well appreciated is that within the carpal tunnel, extensive connections often exist between the tendons of the deep flexor muscle of the fingers (Leijnse et al. 1997). These are either fine tendinous linkages or adherent sheets of tenosynovium, both of which may be very resistant to stretch. The tension put on these structures during playing makes them potential sites of pain and inflammation. Some attempt has been made to model the effect of such tendon linkages mathematically in the hope that this may ultimately help to identify which finger exercises are capable of improving performance and which are either useless or potentially damaging (Leijnse et al. 1992, 1993). For example, attempts to develop the ability to raise the ring finger as high as the others despite the restrictions imposed by tendon linkages may be a common source of injury (Brown, 2000). One reason why the small intrinsic muscles are so important for hand dexterity is that they can be controlled individually to a considerable degree. Based on a knowledge of their anatomy and functional roles, strategies involving these muscles have been proposed by some piano teachers to minimize the impact of the limitations imposed by tendon linkages (Beauchamp, 2003a).

Although one would expect that the movements of the thumb would be quite independent of those of the fingers, it is not unusual to find an anomalous linkage between the tendon of flexor pollicis longus and the deep flexor tendon of the index finger (Linburg–Comstock syndrome – see Fig. 1B). Information drawn from several studies indicates an incidence of between 20 and 35% for this anomaly in the general population, with about one-quarter of the individuals exhibiting it in both hands (Allieu et al. 1998; Miller et al. 2003). Under these circumstances it is impossible to flex the thumb without inducing a flexion of the index finger and in a small proportion of instrumentalists this can lead to pain or difficulty in playing (Miller et al. 2003). This is one of the few tendon linkages for which surgical section is both feasible and generally beneficial (Allieu et al. 1998).

Cortical control of the hand

In order to appreciate the results of studies on the effect of musical performance on cortical activity, it is first necessary to review how the cortical motor areas are organized. The notion that the primary motor cortex contains a detailed somatotopic map of the body has persisted in many textbooks despite a steady flow of evidence to the contrary (Sanes & Schieber, 2001). Although there is certainly a rough somatotopic order which distinguishes regions involved in the control of muscles in the face, hand, upper limb, trunk and lower limb, detailed topographical maps cannot be identified within these subdivisions. Even Penfield, whose work is often assumed to provide one of the main bulwarks of this theory, stated that the map represented by the motor homunculus ‘cannot give an accurate indication of the specific joints in which movement takes place, for in most cases movement appears at more than one joint simultaneously.’ (Penfield & Rasmussen, 1950, p. 56; Sanes & Schieber, 2001). Regions of cortex that can activate a particular muscle do not all lie at a single location, but are scattered across a small area of motor cortex. As a result, although the region of motor cortex controlling the hand is relatively easy to define, the representations of the muscles moving different fingers or individual joints overlap to a considerable degree (Schieber & Hibbard, 1993). This would in any case be expected if only because several muscles acting primarily on the fingers not only move several joints (including the wrist), but also have slips running to more than one finger (e.g. flexor digitorum profundus and superficialis and extensor digitorum communis). Although the muscle slips may have their own motor pools, we have already seen that there can be considerable synchrony in the firing of motorneurones in different pools (Reilly et al. 2004).

Physiological studies indicating that single pyramidal cells within the primary motor cortex can control several muscles are supported by anatomical observations that their axons often send terminal branches to several motor pools, which sometimes lie within different spinal segments (Futami et al. 1979; Shinoda et al. 1986; Kuang & Kalil, 1990). Electrical stimulation of the primary motor cortex using pulses that are longer or more intense than those needed to produce twitches in single or small groups of muscles evokes complex but well co-ordinated movements. It appears that a given cortical neurone may drive a muscle or set of muscles only during one particular type of movement, and remain silent when the same muscles are used in a different context. This has led to the hypothesis that the cortical map represents not individual muscles or joint movements but a set of limb trajectories (Graziano et al. 2002).

Activity in the primary motor cortex is driven or influenced by connections from a number of other cortical regions (Rothwell, 1994). These regions and their interrelationships are summarized in Fig. 2. Sensory information is received by the primary motor cortex directly from the somatosensory cortex as well as from sensory association areas such as the superior parietal region, which sits immediately behind it and plays a role in integrating the sensory information used in the planning of motor activity. In addition to these sensory streams, the primary motor cortex also receives inputs from the supplementary motor cortex and the premotor areas. Both have direct connections to the motor pools in the brainstem and spinal cord, so they can act on them directly as well as through the motor cortex. Like the primary motor cortex, they receive information from sensory association areas. Visual information reaches the premotor areas along two processing paths. The more ventral stream carries information on the shape and position of objects and is used to direct reaching and grasping behaviour. The more dorsal stream is active when visual and other sensory signals trigger a movement but do not guide it. The supplementary area, by contrast, is concerned with movement that is self-generated rather than triggered by external cues. Into this category fall many of the movements required for playing an instrument and the finger tapping tasks discussed below in the context of motor learning. It also controls sequences of movements replayed from memory. The significance of the roles of these different motor areas to musical performance and skill acquisition will be obvious. When a new set of movements is first being learned in humans, an area of the cortex that lies just anterior to the supplementary motor cortex (the presupplementary area) is briefly active (Sakai et al. 1999). Although connected to it, the supplementary motor cortex shows much less activity during this initial period, but once learning is complete it becomes active when the motor sequences are re-enacted. The premotor cortex is also very active during the initial stages of learning (Toni et al. 1998). With further practice, the replaying of these sequences becomes fully automatic. Activity in the premotor cortex declines whereas that in the supplementary cortex is maintained (Wu et al. 2004) and that in the primary motor cortex increases (Karni et al. 1995). The resulting cortical activation pattern for automatic motor activity is essentially that seen in Fig. 3(B). This posterior drift in cortical activation during learning will be encountered again when we discuss the contribution of the different motor areas of the brain in the context of musical experience.

Figure 2.

(A,B) The cortex seen from two different aspects to show the major areas discussed in the paper. Their positions should be compared with the sites the cortical activity seen in Fig. 3. (C) This diagram summarizes some of the roles of, and interconnections between the regions shown in A and B.

Figure 3.

Patterns of cortical activity determined by functional MRI (fMRI) analysis [blood oxygen level dependent (BOLD) signal]. (A,B) A comparison between cortical activity in amateur and professional violinists asked to finger the left-hand movements needed to perform the first 16 bars of a Mozart violin concerto while the right arm was kept as relaxed as possible. The left hand is controlled by the right (R) hemisphere. Red areas represent increased BOLD signals compared with rest, and green, decreased BOLD signals. Both groups show similar patterns of activity, but this is more focused within discrete areas in the professionals. The approximate position of the central sulcus, which forms the boundary between the primary motor and sensory areas, is indicated by the dashed line. The strongest activity straddles this line in a region that represents the sensory and motor representation of the left hand. Activity within the premotor (PMA) and supplementary motor areas (SMA) is indicated. (C–F) A comparison between cortical activity in a violinist mentally rehearsing a Bach partita (C,E) with activity during a bilateral finger tapping exercise (D,F). The finger tapping followed a regular pattern at about 2 Hz and was non-musical. Shades from red to yellow indicate increasing levels of BOLD signal compared with rest. The dashed line represents the central sulcus. Note that during the imagined playing, there is no activity in the primary motor and sensory areas, although there is marked activity in the premotor (PMA), supplementary motor (SMA) and superior parietal (SP) areas. During the finger tapping exercise, the hand regions of the primary motor and primary sensory areas are strongly activated. A,B: reprinted from Lotze et al. 2003. Neuroimage 20: 1817–1829; C–F: reprinted from Langheim et al. 2002. Neuroimage 16: 901–908. Both reproduced by permission of Elsevier. Labelling of SMA, PMA, SP and central sulcus added by present author.

The effect of musical performance training on cortical activation

Patterns of brain activity differ very considerably between professional and amateur musicians even when playing quite simple pieces of music. In one study, professional violinists who typically played around 30 h a week and had an average of 30 years of experience were compared with amateurs who played only 1 h a week and had about 10 years of training (Lotze et al. 2003). Their cortical activation patterns were recorded using functional magnetic resonance imaging while they performed the left-hand finger movements required to play a short extract of a Mozart concerto. Examples of these are shown in Fig. 3(A,B). As would be expected, there were many similarities between the brain responses of the professionals and the amateurs. Both showed cortical activity in primary sensory and primary motor areas that represented the hand. However, the activity in the professional group was much more tightly focused spatially, and in the primary motor cortex was more intense and confined to the right side of the brain (the side which controls the left hand), whereas in the amateurs it was more diffuse and present on both sides. The stronger signal from the hand area of the cortex in the professional group may in part be a reflection of an increase in its cortical representation (Amunts et al. 1997; Schlaug, 2001) as well as a consequence of the automatic nature of the movements. The area around the supplementary and premotor regions of the cortex were active in both groups, although more so in the amateurs. The supplementary, premotor and primary motor regions, together with some other frontal areas, and the left side of the cerebellum (the side receiving proprioceptive input from the active hand) that were also active in amateurs are particularly heavily involved in the acquisition of complex motor skills before they become fully automatic. In professional players, therefore, it appears that many of the complex motor programmes required for executing the movements used for playing have become integrated and refined so that they arise fully formed and thus are controlled mainly from the primary motor cortex. High levels of activity in the basal ganglia, which is often seen at an early stage in the formation of motor programmes, was found only in the amateurs, again probably reflecting their lower level of proficiency (Seitz et al. 1990; Seitz & Roland, 1992). This may also account for the greater level of activity in the right side of the cerebellum in the amateurs. By contrast, in the professionals, particularly those who started their training in early life, there was an increased level of activity in a small region of the cerebellum on the same side as the active left hand. This may be correlated with the observed structural changes seen in the cerebellum of musicians (Schlaug, 2001; Gaser & Schlaug, 2003).

Performance-induced structural changes in the motor areas of the brain

Because of the great demands made on their manual dexterity, keyboard players are among the musicians most often studied in the search for changes in brain structure and function that are related to musical performance. The primary motor and sensory areas of the cortex are obvious places to search for such modifications because they contain topographical body maps in which distortions, should they exist, will be readily detectable. Most individuals show a greater dexterity with one hand or the other, i.e. they are either right- or left-handed, and brain-imaging studies have revealed that this is reflected in the depth of the central sulcus (Fig. 2) that lies along the posterior edge of the primary motor cortex on the opposite side of the brain (Amunts et al. 1996). Because there is a greater development of grey matter within the primary motor cortex on the side controlling the dominant hand, it bulges out more, making the sulcus deeper. Left/right asymmetry is seen only in the hand region of the primary motor cortex, and not in the part lying immediately inferior to it that controls the muscles of the face. The playing of keyboard instruments requires an almost equal dexterity in both hands so it has been hypothesized that the difference between the depth of the central sulcus on the left and right side should be less in keyboard players than in the general population. This indeed appears to be the case, at least for male players (who are the only group studied so far) and can be attributed to an increase in the area of the motor cortex on the side controlling the non-dominant hand (Amunts et al. 1997; Schlaug, 2001). This anatomical observation correlates with a greater symmetry in finger dexterity between the two hands in pianists (Jancke et al. 1997). The depth of the sulcus on both sides of the brain of keyboard players shows some correlation with the age at which they started to learn to play (Amunts et al. 1997). There was, however, no correlation with the total number of years of playing at the time the study was carried out, indicating that the morphological changes reflect a plasticity in brain structure that occurs only early in life.

The two cerebral hemispheres are linked by a large sheet of transverse fibres known as the corpus callosum (Fig. 2), and it might be envisaged that the requirement to co-ordinate the two hands in instrumentalists could lead to changes in its structure. The corpus callosum matures slowly. Although it grows most rapidly in the first decade of life it continues to increase in size until the mid-twenties (Pujol et al. 1993). The period of most rapid growth coincides with the time in early childhood when motor co-ordination is developing. Sensory deprivation during this period can result in a reduction in the size of parts of the corpus callosum. A comparison has been made of the size of the callosum between a population of musicians and one of non-musicians in the 18–35 years age range. In two separate studies of male musicians, the anterior part of the corpus callosum was found to have a greater cross-sectional area than in the control groups (Schlaug et al. 1995; Lee et al. 2003). This region carries connections between the primary sensory and primary motor areas of the two hemispheres, as well as between the premotor and supplementary motor regions. Its larger size in musicians may therefore be a reflection of the changes in the primary motor cortex already discussed. When the motor cortex on one side of the brain is active, the equivalent region on the opposite side is usually inhibited. In pianists and guitarists, it has been reported that this inhibition is reduced but it is unclear if, or how, this is related to the changes in the corpus callosum. It has been hypothesized that the reduced inhibition may improve the ability to co-ordinate the timing of movements of the two hands (Ridding et al. 2000; Nordstrom & Butler, 2002). In contrast to the situation seen in male musicians, one study of female instrumentalists found no increase in the dimensions of the corpus callosum compared with a control group of non-players (Lee et al. 2003).

Since the middle of the nineteenth century, the cerebellum has been regarded as being associated with the control of motor activity and motor learning (Larsell & Jansen, 1967). Imaging studies of the cerebellum in living subjects have revealed that its volume (as a percentage of total brain volume) appears to be greater in male instrumentalists than in non-musicians (Schlaug, 2001). Interestingly, unlike some other morphological changes seen in musicians, this is said to be related not to the age at which music training began but to the intensity of current long-term practice (Hutchinson et al. 2003). The timescale over which this volume change takes place is unknown. The bulk of the cerebellum is made up of white matter; however, it appears that the increase in cerebellar volume is not due to changes in this alone. The relative size of the region of the cerebellar cortex involved in control of the left hand is also positively correlated with the degree of musical training (Gaser & Schlaug, 2003). As all of the musicians in this study were right handed, this represents an increased representation of the non-dominant hand. In females, however, Gaser & Schlaug (2003) found that relative to overall brain volume, the size of the cerebellum appeared comparable with that of the male musician group regardless of musical experience.

Learning to play the music

Very high levels of motor control are required by professional musicians not only for accurate rendition of the music, but also to enable it to be played with expression – a rather intangible but generally instantly recognizable element of performance which is realized by subtle manipulation of timing and dynamics. These skills are not acquired easily, and by the age of 20 years an expert player will typically have carried out in the order of 10 000 h of practice (Sloboda et al. 1996).

The mechanism of motor learning

The studies that are of most relevance to understanding how the technical mastery of an instrument is achieved are based on the learning of simple patterns of human finger movement. The results suggest that we learn to carry out such patterns of movement in several stages (Karni et al. 1995, 1998). An initial phase of fast learning takes place over a period of minutes when a new pattern of movement is performed for the first time. This is followed by a subsequent period of consolidation during the 6–8 h after the activity has ended. Subsequent training sessions produce additional increments of improvement but these become progressively smaller until an upper level of proficiency is reached that may be retained for months or years (Karni et al. 1995). To break through this ceiling then requires a considerable increase in effort. Some notions of the mechanisms underlying these processes have been gleaned from imaging brain activity during this type of learning. When the task is carried out for the first time, activity is seen in the primary motor cortex controlling the active hand. When repeated, the level of cortical activity is initially reduced due to habituation, but when the finger sequence has been repeated many tens of times, the signal in the primary motor cortex becomes larger, and then remains at the same level during subsequent training sessions (Karni et al. 1998). It is during this latter stage of motor learning that the new synaptic connections that underlie the reorganization of the motor cortical map are made (Kleim et al. 2004). Of course, as we have already seen, the primary motor cortex is not the only part of the brain involved in motor learning. Activity is also present in areas such as the premotor and supplementary motor cortex, as well as in the basal ganglia and the cerebellum (Hund-Georgiadis & von Cramon, 1999). Activity in subcortical structures occurs particularly during the early phases of motor skill acquisition and declines as greater competence is achieved.

Significantly, it has also been shown that cortical activity during the initial stages of learning a finger-tapping task differs between pianists and non-musicians. In pianists there is much less activity in the supplementary motor and premotor areas, and greater activity in the primary motor cortex (Haslinger et al. 2004). Right from the outset, the pianists appear to be showing a pattern of activity that the non-musicians take some time to achieve, which presumably reflects their previous intensive training in the control of fine finger movement. The supplementary motor cortex is thought to be involved in the control of sequential movements carried out in the absence of visual feedback. It is more active when the task is complex and so the reduced activity in this region in the pianists may imply that the task is less demanding for them. These differences in the patterns of cortical activity resemble those we have already seen between amateur and professional musicians (Lotze et al. 2003) (Fig. 3A).

Although the left hemisphere controls the right hand and vice versa, there is some evidence that training of the non-dominant hand may also cause activation of the ipsilateral motor cortex. This suggests that training one hand may improve the performance of the other through interhemispheric communication (Hund-Georgiadis & von Cramon, 1999). Although such transference has not been observed in all studies and therefore remains controversial, we shall see later that focal dystonia in one hand can quickly appear in the other if it is used to carry out tasks formerly assigned to the dystonic one. If, as has been proposed, this type of dystonia is an effect of overtraining, its transference to the other hand would be consistent with a bilateral effect of training on the sensory–motor cortex.

Instrumental rehearsal in practice

In the initial stages of learning a new piece of music, reading through the score may be used to gain a clear idea of the notes and expression marks in the absence of the distraction of the physical challenges of playing. This can take the form of studying the overall structure of the piece and/or a careful examination of single phrases or their component note sequences. Music has its own grammar, and when based on conventional harmonies or chordal progressions, only certain sequences are to be expected. Recognizing this should increase the probability of playing the notes accurately at the first attempt and, in so doing, establishing the correct motor programme from the outset. Technical exercises of various types, including those based on scales, are designed to inculcate just such motor subroutines. We have already seen that finger movements must be repeated many times to generate the initial increased cortical response that underlies the first stage of motor learning. If several incorrect variants are played initially, this will at the very least slow the consolidation of the correct motor sequence, and at worst lead to the firm establishment of an incorrect variant, which may persist for some time as a learned alternative to the correct one (especially if the incorrect sequence is easier to perform). The importance of the initial stages of learning is borne out by observations of how high-level performers actually practice. One study recorded how a pianist set about learning and memorizing a movement from Bach's Italian concerto from first seeing the piece until a performance-level of execution was achieved. Different short segments of one awkward eight-bar sequence were repeated more than 150 times during the first practice session, and a further 50 times in the second session (Chaffin & Lemieux, 2004). In subsequent practice sessions, it was never singled out for special treatment, indicating that the motor sequence had been effectively mastered. This demonstrates the efficacy of this highly focused approach to motor learning often known as ‘deliberate practice’. Animal studies also suggest that focused attention plays a significant role in determining the efficacy of motor practice (Schmied et al. 2000). It therefore appears that the pianist had intuitively arrived at a strategy that is in accord with the results of more objective scientific studies.

Mental rehearsal

For many professional musicians, mental rehearsal is an integral part of their performance preparation. In neurobiological terms, mental rehearsal and preparation can be seen to encompass several distinct elements. First, there is the interpretation of the score in terms of an internal representation of sound (i.e. mentally hearing the music when reading the score). Second, there is the committing of the score to memory, which includes not only the notation on the stave, but also the marks of expression. Playing from memory reduces the cognitive load of performance and allows greater attention to be given to assessment of the sound being produced. Only when memorization has been achieved does the final stage of using the score to support a virtual rehearsal of the movements required to perform it become possible. In its most advanced form, it may even be used to explore different options for expressive interpretation. For this to be possible, the brain must create an internal image not only of the movements but also of the precise effect they will have on the sound. Although they are combined holistically, each of these tasks requires a different set of mental skills; we will concentrate primarily on the mental rehearsal of playing movements. This assumes a complete familiarity with and mastery of the instrument, as the imagined movements must remain accurate in the absence of auditory feedback, and is therefore an option open only to advanced players (Lotze et al. 2003). This form of physical imagery is not unique to musicians; it is widely used in sports that involve complex stereotyped movements. High-jumpers, for example, can often be seen mentally practising using minimal or reduced movements before a crucial jump. Perhaps surprisingly, the effectiveness of mental practice in improving aspects of the dynamics of movement, such as the accuracy of a movement trajectory, has been verified experimentally (Yaguez et al. 1998). In one study, the effect of mental rehearsal was compared with physical practice of the same simple musical passage on the keyboard (Pascual-Leone, 2001). The subjects practised mentally (i.e. rehearsing the activity in the absence of any movement or muscle activity) or physically for 2 h daily over 5 days. The effect of this on the representation of the hand in the motor cortex was assessed by comparing the efficacy of transcranial magnetic stimulation in eliciting movement of the trained and untrained hand. Although the section of the primary motor cortex that drives the muscles of the hand remained silent in the group carrying out only mental rehearsal, the size of the areas that were able to activate the muscles moving the fingers increased, while their threshold for activation was reduced. This was accompanied by a demonstrable improvement in the accuracy of motor performance, although it was not as great as that achieved through physical practice. However, at the end of the study, a single session of physical practice in the mental rehearsal group was sufficient for them to reach parity of performance with the physical rehearsal group. This result is in general agreement with those of other less physiologically rigorous studies, which suggest that while mental practice is better than no practice, it is not as good as actual practice (Gabrielsson, 1999). Professional musicians who play a great deal can be prone to overuse injury, so one advantage of mental practice is that it can be used to refine or improve performance while reducing this risk.

Clearly one source of information missing in mental rehearsal is exteroceptive and proprioceptive feedback. The less experienced the player, the more important this feedback will be, but in experienced players in whom the movements of the fingers have become fully automatic, its importance may be reduced. There is also a lack of auditory feedback as there is no tangible output from the virtual activity. The effect of its absence on performance accuracy has been investigated in experienced pianists playing on a silent keyboard (Finney & Palmer, 2003). For substantial excerpts of previously learned music, there was no significant difference in error rate between playing with or without sound. In simple sight reading tests, however, although the absence of auditory feedback had no effect on performance from the score, it did have a deleterious effect on the accuracy of the music when it was subsequently repeated from memory, indicating that it remains important for the initial stages of learning even in experienced players.

With the exception of the primary motor and sensory areas, many of the other cortical regions that are normally active during playing are also active during virtual practice (Langheim et al. 2002; Lotze et al. 2003; Meister et al. 2004). Functional imaging studies have revealed activity in premotor and supplementary motor areas (Fig. 3C–F). The premotor area was active when the physical performance of a melody was silently re-created in the mind, though not when the melody was simply recalled. Perhaps surprisingly, there was no activity in the primary auditory area during these experiments, despite the fact that virtual rehearsal generally requires a vivid mental realization of the sound associated with the virtual ‘movements’. This is in contrast to imagining a scene, which does produce activity in visual cortical areas. However, when the rehearsal involved real hand movements, activity was present in the right primary auditory cortex and left auditory association cortex even if the hand was not in contact with the instrument (Lotze et al. 2003). This suggests that some link exists between the primary motor and auditory areas (Bangert & Altenmuller, 2003). The right primary auditory cortex is the main region involved in the perception of pitch, harmony and timbre, and its level of activity during silent practice with actual finger movements is greater in professional musicians than in amateurs. A link in the opposite direction between the primary auditory and primary motor cortices has also been demonstrated. A study of advanced piano students demonstrated that listening to a piece of keyboard music with which they were already familiar caused involuntary activity in the primary motor cortex even though no contraction of muscles moving the fingers took place (Haueisen & Knosche, 2001). Activity in the motor cortex occurred in the region controlling a finger just before the note it would have played was sounded and so it mirrored the activity that would have been required for playing. No such response was seen in a control group of similarly experienced singers who were not pianists. This type of connection would undoubtedly support the ability to play music by ear.

Occupational focal dystonia in musicians – a consequence of maladaptive cortical remodelling?

Dystonia refers to abnormal involuntary and uncontrollable muscle contractions that are often seen in the hand, or in muscles of the neck or face. It is said to be focal when it affects only a single muscle or small group of muscles. This is generally task specific, particularly affecting activities that have habitually been carried out for prolonged periods as part of a patient's occupation. Occupational focal dystonia is twice as common in men than in women, and most typically appears in the fourth decade of life (Wynn Parry & Tubiana, 1998; Hochberg & Hochberg, 2000; Lim et al. 2001), although it can occur earlier (Jankovic & Shale, 1989). Among musicians it therefore tends to develop in mid-career in individuals who already have a sound and well-established technique. The consequences may not initially be obvious to the listener, and the condition sometimes take years to develop fully; however, in some cases the dystonia becomes debilitating in a matter of months (Tubiana, 2000). The incidence of focal dystonia among instrumentalists appears to be of the order of 1 : 200–1 : 500 (Lim et al. 2001), which is about ten times the level seen in the general population (Wynn Parry & Tubiana, 1998; Pullman & Hristova, 2005). Dystonia is most commonly reported among pianists and guitarists, although the hands of other string players and of woodwind and brass players may also be affected. In addition, focal dystonia of the embouchure is seen in wind players (Frucht et al. 2001).

Symptoms of focal dystonia in musicians

The onset of dystonia is usually gradual and painless. This may start as a feeling of heaviness in the fingers but it soon becomes more debilitating. In rapid passages, some of the fingers may appear to lag behind the others, causing poor performance in pieces that have been executed without problems for decades (Hochberg & Hochberg, 2000; Tubiana, 2000). As the condition progresses, the affected fingers may become hyperflexed, with the ultimate consequence that in pianists the outer surface of the nail strikes the key instead of the pad (Candia et al. 2002). In other cases, the recalcitrant fingers may become involuntarily extended so that they cannot be brought down to touch the key or string. The later stages of dystonia are sometimes characterized by a fierce simultaneous contraction of antagonistic flexor and extensor muscles of the digits that causes fatigue and pain. Interestingly, in patients who play more than one instrument (e.g. piano and violin), the dystonia may affect the playing of only one. Activities such as typing, which one might think would share many similarities with keyboard playing may, initially at least, remain unaffected (Wynn Parry & Tubiana, 1998). The typical site of the dystonia varies depending on the instrument. Wynn Parry & Tubiana (1998) report that in pianists, it is the right hand that is most frequently dystonic (70%), in which case the most commonly affected fingers are those closest to the little finger. When the left hand is involved, the affected digits tend to be on the other side of the hand. Among guitarists, it is the index and ring fingers of the right hand that are most likely to become dystonic, whereas in violinists, three-quarters of cases involve the left hand.

Theories of the origin of dystonia

A full discussion of the theories of the origin of focal dystonia are beyond the scope of this review, but an account of these can be found in Lim et al. (2001). Some researchers have proposed a strong causal link between focal dystonia and repetitive strain injury (Byl & McKenzie, 2000), although the relationship with injury remains unclear. Although some anecdotal evidence has been put forward to support the idea that a variety of injuries may trigger dystonia (Jankovic & Shale, 1989), a lack of objectivity in the way the data have been collected, and an absence of any rationale to explain the theory, undermines its credibility. In addition, the average age of onset for overuse injuries among musicians is in the mid-twenties, whereas for dystonia it is in the mid- to late thirties (Tubiana, 2000).

There is a growing belief that focal dystonia is not fundamentally a malfunction of peripheral motor control but has its origins in the central nervous system. Studies of the cortex of monkeys and humans with focal dystonia have revealed a change in the organization in the primary sensory cortex. In monkeys that develop dystonia after being trained to carry out repetitive hand movements to obtain food, there is a marked degradation of the sensory map of the hand. The near simultaneous stimulation of adjacent fingers during tasks that require close attention appears to be a significant factor in these changes. The receptive fields of cortical neurones receiving information from the fingers can be enlarged by a factor of 10–20. This causes considerable overlap between the cortical representation of different fingers and between the front and back of the hand (Byl & Melnick, 1997). Receptive fields may grow to cover several finger joints. Investigations of musicians and non-musicians with focal dystonia of the hand reveal similar maladaptive changes in the cortical mapping (Fig. 4). The finger representations of the dystonic hand are much closer together, and may either overlap or be in random order (Bara-Jimenez et al. 1998; Elbert et al. 1998; Byl et al. 2000). There is also evidence of a degradation in the mapping of the non-affected hand of dystonic patients, although this is not as marked as that of the affected hand (Elbert et al. 1998). Braille readers who use several fingers to read text also often show a disordered or fused cortical representation of these fingers which is reflected in the representation of the other hand, even though it is not used for reading (Sterr et al. 1998). This suggests that the sensory maps on each side of the brain are not fully independent and may help to explain why dystonia sometimes appears quite quickly in the previously non-affected hand when it takes over tasks previously carried out by the dystonic one (Lim et al. 2001).

Figure 4.

Mapping of the hand in the primary sensory cortex of an organist with focal dystonia. On the side of the cortex receiving sensory information from the non-dystonic hand, the representation of the fingers are well spaced. By contrast, in the sensory representation of the dystonic hand, the finger representations overlap significantly. D1 = thumb, D5 = little finger. The circles and squares indicate the mean location of the magnetic dipoles from magneto-encephalographic recordings. These have been superimposed on an MRI image of a coronal section through the somatosensory cortex. Reprinted from Elbert et al. (1998) Neuroreport 9: 3571–3575 with permission from Lippincott, Williams and Wilkins.

Focal dystonia is a disorder of motor control, so if the idea that it is caused by changes in the sensory map is valid, it should theoretically be possible to demonstrate some reconfiguration of the motor cortex. In dystonic patients it is has not been possible to detect detailed changes in the motor map of the hand, although given what we already know of the organization of the primary motor cortex, this is perhaps to be expected. There are, however, disturbances in the organization of the sensory–motor maps within the basal ganglia (Delmaire et al. 2005). Changes are also seen in other motor areas of the brain in dystonic patients, including musicians (Ibanez et al. 1999; Lim et al. 2001, 2004). For example, during the execution of tasks that induce dystonic symptoms, activity in the posterior part of the supplementary motor cortex is reduced below normal levels whereas that in the primary motor and sensory cortex is increased. One consequence of the intensive training needed to perfect the rapid and continuous finger movements required by instrumentalists is a reduction in the inhibition normally seen between the primary motor areas on each side of the brain (Nordstrom & Butler, 2002). Paradoxically, the inhibition is thought to be part of a mechanism that allows the individual muscles acting on the fingers to be controlled independently. The reason why it is reduced in normal musicians is unknown but as a similar pattern is seen in dystonic nonmusicians, this may predispose them to dystonia. This may be another factor underlying the rapid development of dystonia in the non-affected hand when it is used to compensate for the one initially affected.

Treating focal dystonia in musicians

Currently there is no treatment regime for focal dystonia in musicians that significantly and reliably restores the ability to play the instrument at a high level. A number of drugs are used routinely to control the symptoms, the major classes being benzodiazepines, anticholinergics and antidopaminergics (Hochberg et al. 1990). In addition, the chronic muscle contractions that can develop may be treated by direct injection of Botulinum toxin into them. In musicians in particular, the dosage and site of injection is critical if the muscle activity required for playing is to be preserved. Although this treatment can have a positive effect for some musicians, in only a minority of cases is it sufficiently successful to allow playing of a professional standard to be maintained (Altenmuller, 2003; Schuele et al. 2005).

The recent discovery of the changes in cortical mapping that are associated with occupational focal dystonia has suggested new treatments based not on drugs but on sensory and/or motor re-education. Although there is no consensus on the optimum strategy for this approach and although it is still uncertain how successful it will prove to be in the long term, a number of preliminary studies have been carried out to investigate its potential benefits.

The tasks used for sensory retraining are designed to require attention as this appears important for the remoulding of sensory maps in the cortex. In one study a group of patients with ‘writer's cramp’ (a form of focal dystonia) were taught to read Braille as a means of improving sensory discrimination (Zeuner et al. 2002). After an initial training period of 8 weeks some continued daily practice for up to 6 months. Tactile spatial acuity improved significantly, which suggested that some reconfiguration of the sensory map was taking place, as did motor performance (in this case, writing); however, those who stopped the therapy reverted quite quickly to their original state. In a second study, subjects who suffered from a variety of occupational dystonias were trained in various discriminative tasks that required the tactile identification of objects and patterns by the hands and fingers. This was accompanied by a programme aimed at reducing the aberrant motor activity of dystonia and improving general fitness and posture (Byl & McKenzie, 2000). Improvements in sensory discrimination and motor accuracy were again evident, although movement remained slower than normal. There have been only a few studies directed specifically at musicians, who are probably one of the most challenging groups in terms of rehabilitation. These used an approach known as constraint-induced movement therapy (Candia et al. 2002, 2003; Taub et al. 2002), which was originally developed for the rehabilitation of stroke victims. One study involved six pianists, two guitarists, two flautists and an oboist, all of whom suffered from dystonia of the hand. In each case the main finger(s) which were being used to compensate for the dystonic one were immobilized with splints, leaving the dystonic one to cope on its own. During a period of up to 1 year, the dystonic finger was required to carry out repetitive movements demanding the co-ordination of several muscles, for 1 h or more each day. After an initial 8-day period of intensive training, many (but not all) of the subjects showed signs of improvement as measured on a subjective self-assessment scale. However, maintaining and augmenting this improvement required that the treatment be continued regularly for many months. Pianists and guitarists who persisted with the training gained the most from this approach and some reached a level close to normality, but the wind players derived little or no benefit from it. In a further study of ten dystonic musicians, a mechanical device composed of two keys was used to provide an objective assessment of the effectiveness of movement-constraint therapy (Candia et al. 2003). Improvements in finger control of key depression were statistically correlated with the degree of restoration in the structure of the somatosensory map of the dystonic hand.

Concluding remarks

It is in the nature of musical performance that succeeding generations have pushed the levels of virtuosity to ever greater heights. This inevitably reveals the limits of the motor system both from an anatomical and from a physiological point of view. By re-assessing the anatomy of the muscles and tendons of the hand, we can begin to appreciate that variations in anatomy between individuals will have a significant effect on the possible fingerings which can be used to play certain passages. The combination of functional MRI with its high spatial resolution and magnetoencephalography (which provides excellent temporal resolution) has made investigation of the effects of training on cortical activity patterns a realistic proposition. As we have seen, by comparing instrumentalists of a professional standard with amateurs and with non-musicians, it has been possible with these techniques to observe the effects of hand training on activity patterns within the regions of the cortex devoted to motor control. We have concerned ourselves here primarily with the motor system, but the same methods have been used to study the somatosensory system and have revealed changes in the representation of the hand within the primary sensory cortex of non-dystonic instrumentalists (Elbert et al. 1995; Hashimoto et al. 2004). As one might expect, brain imaging techniques have also been applied extensively to studies of auditory processing of both music and sound in musicians and in those with impairments in musical processing (Deutsch, 1999; Peretz & Zatorre, 2003). Many of the factors that make musicians such excellent subjects for studies of motor control apply equally to these studies of sensory processing. Indeed, among human subjects who lack sensory, cognitive or motor deficits, they are one of the most interesting groups with specialist skills in which to study patterns of brain activation.