הנזיר שחקר את המוח שלו

Neurobiology of Learning and Memory 78, 553–564 (2002) doi:10.1006/nlme.2002.4091

Imaging Brain Plasticity during Motor Skill Learning

Leslie G. Ungerleider,* Julien Doyon,*,† and Avi Kar

*Laboratory of Brain and Cognition, National Institute of Mental Health,

National Institutes of Health, Bethesda, Maryland 20892

†Department of Psychology, University of Montreal, Montreal, Quebec, Canada

‡ Faculties of Science and Education, Haifa University, Haifa, Israel

The search for the neural substrates mediating the incremental acquisition of skilled motor behaviors has been the focus of a large body of animal and human studies in the past decade. Much less is known, however, with regard to the dynamic neural changes that occur in the motor system during the different phases of learning. In this paper, we review recent findings, mainly from our own work using fMRI, which suggest that: (i) the learning of sequential finger movements produces a slowly evolving reorganization within primary motor cortex (M1) over the course of weeks and (ii) this change in M1 follows more dynamic, rapid changes in the cerebellum, striatum, and other motor-related cortical areas over the course of days. We also briefly review neurophysiological and psychophysical evidence for the consolidation of motor skills, and we propose a working hypothesis of its underlying neural substrate in motor sequence learning. q 2002 Elsevier Science (USA)

Key Words: motor sequence learning; motor cortex; procedural memory; consolidation; fMRI; humans.

In everyday life, we go about activities using a variety of motor skills that have been acquired gradually through practice and interactions with our environment. These include, for example, the use of smooth coarticulation of finger movements into a specific sequence (e.g., when playing a musical instrument like the piano), of regular multijoint movement synergies (e.g., during reaching and grasping of small objects), and of smoothly executed

We thank our collaborators, M. M Adams, P. Jezzard, F. Lalonde, G. Meyer, C. Rey-Hipolito, A. W. Song, and R. Turner. We also thank E. Gutierrez for technical assistance in preparing the manuscript. Some of the material included in this paper appears in A. Karni et al., 1998, Proceedings of the National Academy of Sciences

of the United States of America, 95, 861–868, and in J. Doyon and L. G. Ungerleider, 2002, Neuropsychology

of memory, 3rd ed., L. R. Squire and D. L Schacter (Eds.), Guilford Press, New York. This work was supported, in part, by grants from the Natural Sciences and Engineering Research Council of Canada and the Canadian Institutes of Health Research to J. D. and through funding from the NIMH-IRP to L.G.U. Address correspondence and reprint requests to Leslie G. Ungerleider, Laboratory of Brain nd Cognition, ational Institute of Mental Health, 10 Center Drive, Building 10, Room 4C104,  Bethesda, MD 20892-1366. Fax: (301) 402-0921. E-mail: ungerlel@intra.nimh.nih.gov. 553 1074-7427/02 $35.00 q 2002 Elsevier Science (USA) All rights reserved. 554 UNGERLEIDER, DOYON, AND KARNI

eye–body coordinated actions (e.g., as in playing sports such as golf). To study the cognitive processes and the neural substrates mediating our ability to learn such skilled behaviors in the laboratory, investigators have used experimental paradigms that fall into two categories: the first measures the incremental acquisition of movements into a wellexecuted

behavior (motor sequence learning), whereas the second tests our capacity to compensate for environmental changes (motor adaptation) (e.g., Karni et al., 1995, 1998;

Doyon et al., 1996; Shadmehr & Holcomb, 1997). Operationally defined, these two forms of motor skills refer to the process by which movements, produced either alone or in a sequence, come to be performed effortlessly through repeated practice. In both animals and humans, motor skill learning is usually measured by a reduction in reaction time and the number of errors and/or by a change in movement synergy and

kinematics (e.g., Doyon et al., 1997; Shadmehr & Holcomb, 1997; for reviews, see Karni, 1996; Doyon, 1997). For some skills, such as learning to play a new melody on a musical instrument, early learning can be facilitated using explicit knowledge (i.e., requiring thought). For most motor skills, however, motor performance is ultimately overlearned

to a point where it can be performed implicitly (i.e., without thought). As opposed to other forms of memory (e.g., episodic memory), these changes in performance are known to evolve slowly, requiring many repetitions over several training sessions (Karni, 1996). Indeed, psychophysical studies have demonstrated that the incremental acquisition of

motor skills follows two distinct stages: first, an early, fast learning stage in which considerable improvement in performance can be seen within a single training sessionand second, a later, slow learning stage in which further gains can be observed across several sessions (and even weeks) of practice (Nudo et al., 1996; Karni et al., 1998). In addition to these two stages, an intermediate phase corresponding to a consolidation period

of the motor routine has been proposed (Karni & Sagi, 1993; Brashers Krug et al., 1996; Karni et al., 1998). Finally, with extended practice, the skilled behavior is thought to become resistant to both interference and the simple passage of time. Once overlearned, a motor skill can thus be readily retrieved with reasonable performance despite long periods without practice. Based on work in animals and humans, several brain structures, including the striatum, cerebellum, and motor cortical regions of the frontal lobe, have been thought to be critical for the acquisition and/or retention of skilled motor behaviors (for reviews, see Karni, 1996; Doyon, 1997; Sanes& Donoghue, 2000). Anatomical studies have demonstrated that these structures form two distinct cortical–subcortical circuits: a cortico-striato-thalamocortical loop and a cortico-cerebello-thalamo-cortical loop (Middleton & Strick, 1997)

(Fig. 1). Evidence supporting the roles of these cortical–subcortical systems in motor skill learning has come from impairments found in patients with striatal dysfunction (e.g., in Parkinson’s or Huntington’s disease), with damage to the cerebellum, or with a circumscribed lesion involving frontal motor cortical areas (e.g., Doyon et al., 1997).

Further support has come from neurophysiological studies, as well as from lesion experiments, in rodents and nonhuman primates. More recently, modern brain imaging techniques, like positron emission tomography and functional magnetic resonance imaging (fMRI), have allowed us to identify the neural substrates mediating this type of memory in normal, healthy humans and to explore the functional dynamic changes that occur over the entire course of the acquisition process (for reviews, see Karni, 1996; Doyon, 1997; Doyon & Ungerleider, 2002). MOTOR SKILL LEARNING 555

FIG. 1. Diagram illustrating the major cortical and subcortical structures involved in motor skill learning and their interconnections. These structures are organized into two main circuits: a cortico-striato-thalamocortical loop and a cortico-cerebello-thalamo-cortical loop. Dynamic changes within these loops occur during motor sequence learning and motor adaptation. Cortical regions: SMA, supplementary motor area; PM, premotor cortex; M1, primary motor cortex. Thalamic nuclei: VLo, ventrolateral nucleus, oral division; X, area X; VLc, ventrolateral nucleus, caudal division; VPLo, ventroposterior nucleus, oral division.

In the following sections, we review recent findings, mainly from our own work using fMRI, which suggest that: (i) the learning of sequential finger movements produces a slowly evolving reorganization within primary motor cortex (M1) over the course of weeks and (ii) this change in M1 follows more dynamic, rapid changes in the cerebellum, striatum, and other motor-related cortical areas over the course of days. We also briefly

review neurophysiological and psychophysical evidence for the consolidation of motor skills, and we propose a working hypothesis of its underlying neural substrate in motor sequence learning.

SLOWLY EVOLVING REORGANIZATION OF M1

In the past decade, the learning of sequential finger movements—related to skills such as writing and typing—has become an important paradigm for studying the acquisition of motor skills using imaging techniques. These studies, however, were not designed to look at the effects of long-term training. To examine these effects, we used a simple finger-opposition task in which normal, healthy subjects were trained and tested over the course of several weeks and were scanned using fMRI at weekly intervals to image their brains (Karni et al., 1995). 556 UNGERLEIDER, DOYON, AND KARNI In this task, subjects were instructed to oppose the fingers of the nondominant hand to the thumb in one of two given sequences (Fig. 2A). The sequences were composed of five component movements or their mirror-reversed counterparts. Subjects were required

to tap each sequence, with no visual feedback, as accurately and rapidly as possible. While initial performance of the two sequences, in terms of speed and accuracy, did not differ (Figs. 2B and 2C), 10–20 min of daily practice, during which subjects were instructed to repeatedly tap a given sequence (the other sequence served as the unpracticed control)

in a rapid self-paced and accurate manner induced large gains in performance across weekly test sessions. Performance improvement reached an asymptote after about 3 weeks of training with more than a doubling of the initial rate (Fig. 2B). The improvement was specific to the trained hand and did not generalize to the performance of the control

sequence. These behavioral results suggest that a specific representation of the trained sequence of movements (rather than a representation of the individual component opposition movements) had developed as a function of training. In the scan sessions, we measured motor-activity-evoked signal changes at weekly intervals using a 4-T MRI system and a surface coil placed over M1’s hand representation in the central sulcus of the contralateral hemisphere. We focused on M1 because this area

has been indicated by studies in adult monkeys as a locus of manual skill learning, and it is thought to be important in the initiation of voluntary motor actions, especially those associated with fine manipulative abilities (Schieber, 1995). Additionally, we considereda possible analogy to the results of several basic perceptual tasks, in which primary cortical

representations have been shown to reorganize as a function of training and learning (Merzenich & Sameshima, 1993). During scanning, both the trained and the untrained control sequence were performed at a fixed, comfortable rate of 2 Hz, paced by the magnetic field gradient switch noise. Thus both rate and component movements were matched, and the only difference between the two sequences during scanning was the

difference in practice histories. The results showed that in the first scan session, performed before any training was given, a comparable extent of the contralateral M1 was activated by the execution of both sequences. However, by Session 4, which corresponded to 3 weeks of daily practice on the designated training sequence, and in all subsequent sessions,

the extent of activation evoked by the trained sequence in M1 was significantly larger compared to the extent of activation evoked by the control, untrained sequence (Fig. 2D).

It is important to note that in the initial, naive state the activation in M1 was somewhat patchy; it remained so by Session 4, but to a lesser degree. Control experiments showed that it did not extend beyond the hand representation itself, indicating that an expansion

of the total hand representation area had not occurred. Thus, the differential activation was accounted for by a subpopulation of pixels, in the hand area, that showed a significant response to performance of the trained sequence, but little or no response to the performance

of the untrained sequence. The more extensive activation evoked by the trained compared to the untrained sequence persisted in M1 weeks after training was discontinued (Fig. 2E). There was also no significant decrease in performance and, in fact, 1 year after training was stopped, there was still significant retention of the skill (results not shown here).

These imaging data suggest that long-term practice results in a gradually evolving, specific, and more extensive representation of the trained sequence of movements in M1. The results are compatible with the idea that motor practice induces the recruitment of MOTOR SKILL LEARNING 557 FIG. 2. The effects of long-term practice of a motor sequence. (A) The two sequences of finger-to-thumb opposition movements used in our study (Karni et al., 1995). In Sequence A the order of finger movements was 4, 1, 3, 2, 4 (numbering the fingers from index to little), and in Sequence B the order was 4, 2, 3, 1, 4, as indicated by the arrows (matched, mirror-reversed sequences). (B) Learning curves. Each curve depicts the performance of a single subject as a function of time. Pretraining is time point 0. Subjects reached asymptotic performance after about 3 weeks of training, at which point they had doubled the rate at which they could

perform the trained sequence. (C) Behavioral evidence for consolidation of the motor sequence. Number of correct sequences performed during a test interval of 30 s for the two sequences (randomly assigned to be the trained or the untrained control): before training, after a few minutes of externally paced performance of the

randomly assigned trained sequence, 24 h later, with no additional training in the interval. (D) Emergence of differential activation in M1 evoked by the trained compared to the untrained (control) sequence following 2 weeks of daily practice on the designated trained sequence. Data are from a single subject. (E) Maintained

differential activation 8 weeks later with no additional training in the interval. Sagittal sections in D and E are through the right hemisphere centered _35 mm from midline. A surface coil was used, which had the advantage of providing enhanced signal-to-noise ratios, but at the cost of resolution.

558 UNGERLEIDER, DOYON, AND KARNI

additional M1 units into a local network specifically representing the trained motor sequence. This interpretation is in agreement with the recent finding, in monkeys, of practicedependent plastic changes in the functional topography of M1. Nudo et al. (1996) found that following a few weeks of training on a task, which developed skilled manipulation,

the evoked-movement digit representation as well as the representation of task-related movement combinations in M1 were gradually expanded. Thus, M1 may code not just single movements, but rather complex movement sequences. This too is supported by the finding in monkeys that, following long-term practice, cocontracting muscles used in the task come to be represented together in motor cortex, with those movement combinations that were used more frequently in training more extensively corepresented (Nudo et al., 1996). That different subpopulations of neurons within a representational domain, such as the hand, participate in the representation of different movement sequences suggests

a potential for the learning of many parallel skills within a given representation, rather than an enlargement in the representation of a specific body part. We think it is likely that kinematic changes may explain why identical component movements are differentially represented in M1 when arranged in a trained sequence versus an untrained sequence. There is evidence from monkeys showing that fingers do

not move independent of each other and that each instructed movement is generated by combined activation of several muscles, ma ny acting on more than one digit (Schieber, 1995). Additionally, there is evidence demonstrating the complex overlapping representations of movements in maps of M1 (Schieber, 1995). Thus, the implementation of a sequence in M1 may be related to the representation of transitional movements (switching from one digit to the other) and temporally correlated movements (Nudo et al., 1996), which would be dependent on the particular temporal ordering of the component movements in the sequence. Our results (Karni et al., 1995, 1998) suggest that this order-specific aspect of the representation may be enhanced, extended, and consolidated by practice.

DYNAMIC PLASTICITY IN CEREBELLAR AND STRIATAL CIRCUITS

Although the evolution of a sequence-specific differential pattern of activation in M1 required extended practice over several weeks to be evident, changes in M1 activity were observed as early as the first scan session (Karni et al., 1995, 1998). In particular, there was an initial habituation-like effect, in which the second sequence performed in a set

(independent of whether it was to become the trained or untrained sequence) evoked a smaller response than the first sequence performed. Later in the first scan session, however, this effect reversed, such that the second sequence performed now evoked a larger response. We reasoned that these early changes in M1 activity might reflect dynamic events occurring elsewhere in the motor system, perhaps on the order of days instead of weeks. To test this possibility,we designed another fMRI experiment of motor sequence learning

using a version of the serial reaction time task, in which healthy subjects were required to press as quickly as possible one of four buttons corresponding to the location of a red circle that appeared on a screen (Fig. 3A) (Doyon & Ungerleider, 2002). The stimuli were either presented in an unpredictable order (random condition) or followed a repeating 10-

item sequence of movements that was taught to each subject explicitly prior to scanning (learning condition). Subjects were scanned at 1.5 T over three separate sessions with intervening 1-h periods of practice of the 10-item sequence administered just prior to the

MOTOR SKILL LEARNING 559

FIG. 3. (A) Diagram illustrating the materials and stimuli used in the motor sequence learning task. Sequence learning was tested using a modified, computerized version of the Serial Reaction Time task. The stimuli consisted of four blue boxes that were aligned in a horizontal row and of a red circle that appeared above one

of the boxes on each trial. These stimuli were generated using dedicated software (SuperLab) that allowed measurements of reaction times with millisecond accuracy. They were projected onto a screen located in front of the subject and were visualized through the reflection of mirrors embedded within the head coil. In this task,

the subjects were asked to press as quickly as possible one of four buttons orresponding to the location of a visual stimulus that appeared, either randomly or in a repeating sequence that they knew explicitly prior to scanning (see text for more details). The subjects were scanned over three separate sessions with intervening

periods of practice administered just before the second and third scan sessions. (B) Bar graph illustrating the subjects’ mean reaction time in both Random and Learning conditions across the three scanning sessions. A repeated analysis of variance revealed that subjects improved their ability to execute the finger sequence in

both the Learning and the Random conditions ( p _ .01), but that their level of improvement in reaction time across sessions was greater in the Learning than in the Random condition ( p _ .05). (C) Z-score maps showing the dynamic changes in activity in the striatum and cerebellum from Session 1 to Session 3. Significant activity in the striatum was observed in Session 3 only, when subjects attained high levels of performance on this task. By sharp contrast, a decrease in fMRI signal was seen in the cerebellar cortex from Session 1 to Session 3. second and third scan session. Importantly, during scanning, we used a head coil rather

than a surface coil, as used in our first study (Karni et al., 1995), so that changes in patterns of activation across the whole brain could be observed. The behavioral results demonstrated that, as a group, the subjects showed consistent improvement in executing

the sequence of finger movements across scanning sessions (Fig. 3B) and attained the slow learning phase, as their level of performance became stable in Session 3. Analysis 560 UNGERLEIDER, DOYON, AND KARNI

of the fMRI data in the early phase of learning (i.e., Session 1) revealed activations in the cerebellum (Fig. 3C), as well as in the right anterior cingulate, dorsal premotor, and inferior parietal regions. At the end of Session 3, however, these cerebellar and cortical regions showed significant reductions in activity. By contrast, right hemispheric activations were now observed in the striatum (see Fig. 3C), as well as in the supplementary motor area, ventrolateral prefrontal cortex, precuneus, and inferior parietal area. The latter findings suggest that when a sequence of movements is well learned and its execution has become “automatic,” a distributed neural system composed of the striatum and related motor cortical regions, but not the cerebellum, may be sufficient to express and retain the learned behavior (Doyon, 1997; Doyon & Ungerleider, 2002). Prior imaging studies of motor sequence learning are consistent with these findings. Such studies have revealed that the cerebellum is active during the fast learning phase when the initial motor routine is being established (e.g., Jenkins et al., 1994; Doyon et

al., 1996), but that this activity decreases with practice and may become undetectable when the sequential movements are well learned (Grafton et al., 1994; for review, see Doyon & Ungerleider, 2002). Some investigators have also reported striatal activations in the early acquisition phase of motor sequence learning, when subjects have to rely more strongly on the use of cognitive strategies and working memory (Jenkins et al.,

1994). However, the results of other studies have shown that the striatum is significantly more activated when subjects have reached asymptotic performance on the task than when they are at the beginning of the acquisition process (e.g., Grafton et al., 1994; Doyon et al., 1996). Furthermore, unlike the pattern of activity changes in the cerebellum, no

decrease in striatal activity is observed with extended practice. The latter findings suggest that the striatum (in concert with the motor cortical areas discussed below) may be critical for the long-term storage of well-learned sequences of movements (Doyon & Ungerleider, 2002).

At the level of the cortex, the evidence thus far indicates that significant experiencerelated functional reorganization develops within specific motor and associative areas (e.g., Karni et al., 1995, 1998; Classen et al., 1998). Indeed, the creation of a long-term representation of this skill appears to necessitate, at the very least, the contribution of both the SMA and M1 (Karni et al., 1995, 1998; Classen et al., 1998; Gordon et al.,

1998). The presence of M1 changes in our first fMRI study, but not in our second, likely reflects the degree of practice subjects received; that is, activity-related learning in M1 may be apparent only when subjects are overlearned on a task.

HOW ARE MOTOR EXPERIENCES CONSOLIDATED?

As first demonstrated during perceptual skill learning, some performance gains become apparent only after a period of time has elapsed following the end of practice. For example, on a simple visual detection task, Karni and Sagi (1993) have described gains in performance emerging after a minimum of 6–8 h following training on the task. This finding has suggested that there may be a process of consolidation, which is initiated by the practice session and spans several hours before becoming evident at the behavioral level. Evidence for a similar process of consolidation in the motor system comes, in part, from a physiological study by Wise et al. (1998) who recorded neuronal activity while monkeys adapted to visuomotor transforms between a joystick and the cursor on a screen

MOTOR SKILL LEARNING 561 it controlled. Not only did task-related activity change in M1 and premotor cortex during motor adaptation, but changes in activity continued for dozens of trials after performance

reached a plateau. Furthermore, there was no clear asymptote of this activity change up to the 30-min limit of the monitoring time, hence suggesting that the continued change of activity in motor cortex reflected the early stages of consolidation. Based on such findings, we looked for behavioral evidence of consolidation by examining whether delayed gains in performance might occur after training on a motor sequence task.

Subjects were trained on the same finger-to-thumb opposition sequential movement task used in our original imaging study of M1 (Fig. 2A). Performance was tested before, immediately after, and then 24 h after a single training session on one of the two sequences. During testing, subjects were required to tap each sequence as accurately and rapidly as

possible over a test interval of 30 s. In the training session, one of the sequences, randomly chosen, was tapped at a rate of 2 Hz, paced by a metronome, in six short training intervals of 40 s each, separated by 2–3 min of rest. Motor performance for the two sequences before, immediately after, and on the day following training is shown in Fig. 2C. Initial

performance of the two sequences did not differ in terms of either speed or accuracy. Training, however, induced a significant gain in speed for the trained sequence, with no change in accuracy. Little improvement was found for the untrained sequence. On the following day, with no additional training, a significant gain in speed, compared to the

immediate posttraining performance level, was found for the trained sequence only (Fig. 2C). These findings demonstrate that not all learning in a sequential finger-opposition task is concurrent with practice. A limited amount of paced opposition movements was sufficient, not only to improve performance during the session, but also to initiate significant

additional gains that affected performance by the next day. These results, together with the physiological findings in monkeys described above (Wise et al., 1998), support the idea that some gains require time to become effective and continue to develop after motor practice has ended.

Additional psychophysical evidence for the existence of consolidation for motor learning comes from a motor adaptation study by Brashers-Krug et al. (1996). These investigators trained subjects to move a manipulandum against a force field and then, at varying intervals

after training was terminated, introduced a second task. If the second task was introduced 4–6 h after the first, interference was produced. By contrast, beyond a time window of about 6 h, introduction of the second task had no effect. These results, and those of Karni and Sagi (1993), thus point to a window of about 6 h for consolidation to occur. We have repeated this interference experiment using a motor sequence instead of a motor adaptation task and obtained essentially the same results (Rey-Hipolito et al., 1997). Thus, altogether, the behavioral data indicate that human motor memory continues to evolve after the training session and, with the passage of time, is transformed into a long-term trace.

NEURAL CORRELATES OF FUNCTIONAL REORGANIZATION

In the motor system, the neural correlates of functional reorganization have been most extensively studied in M1. Animal work by Donoghue and colleagues, as well as several other investigators, has demonstrated that M1 plasticity depends upon the integrity of the horizontal connections that span the entire region and that this cortical substrate works 562 UNGERLEIDER, DOYON, AND KARNI

by associating M1 neurons into functional assemblies, which are then involved in constructing new motor maps (Donoghue, 1995; Donoghue et al.,1996; Huntley, 1997; Sanes& Donoghue, 2000). Evidence supporting the role of the horizontal connections as a substrate for cortical plasticity comes from several sources. First, intracellular analyses and field-potential recordings of in vitro slice preparations have revealed the existence of horizontal connections throughout M1, especially in superficial layers II and III and in deep layer V (Donoghue et al., 1996; see Sanes & Donoghue, 2000, for review). Second, Jacobs and Donoghue (1991) have shown that local blockade of GABAergic inhibition in one part of M1 unmasks existing horizontal connections that then reveal the existence of hidden representations of limb movements in other parts of M1. Third, Huntley (1997) has demonstrated that the extent of early M1 reorganization is related to the distribution of the intrinsic horizontal connections within M1. This investigator showed that cutting the

motor nerve of rats’ whiskers produced significant expansions of the forelimb area into the former vibrissa territory; however, this reorganization was apparent only in areas that contained strong horizontal connections between the whiskers and the forelimb areas.

Importantly, no change in movement representation was observed in regions in which these connections were either absent or sparse. It has been proposed that rapid cortical plasticity can occur through synaptic modification of the local horizontal circuitry, which is modulated via N-methyl-D-aspartate receptors (Nudo et al., 1997; Sanes & Donoghue, 2000). However, persistent changes in the efficacy of this horizontal circuit have been proposed to result from synaptic modifications through

mechanisms like long-term potentiation and long-term depression. These mechanisms are

believed to up- or down-regulate the strength of the horizontal connections depending on the pattern of activity (context) (Hess et al., 1996; Hess & Donoghue, 1996a,b). Along with these physiological changes in the horizontal circuit, neurobiological alterations

have also been reported following motor learning. Indeed, behavioral training is known to induce changes in gene expression (Kleim et al., 1996), protein synthesis (Hyden & Lange, 1983), and neuron morphology, as measured by dendritic density (Greenough et al., 1985; Kolb et al., 1994; Withers & Greenough, 1989). Interestingly, Kleim et al. (2002) have demonstrated not only that motor learning produces changes in

synaptogenesis, but also that these structural changes are colocalized to the region within which alterations in learning-dependent representational maps are observed following extended practice. These researchers have demonstrated that, compared to control animals, rats trained to reach and grasp food pellets through a slot have significantly more synapses

per neuron within layer V, specifically in the caudal forelimb area. Such findings indicate, for the first time, that both functional and structural plasticity occurs simultaneously within the same cortical region and thus provide strong evidence that morphological changes contribute to the slow learning phase of a skilled motor behavior.

FUTURE DIRECTIONS

A key unresolved issue still facing neuroscientists concerns the neural site (or sites) mediating the consolidation of new sequences of movements. Results from Shadmehr and Holcomb (1997) have shown that the cerebellum is critical for consolidating motor adaptation learning, and those from Imamizu et al. (2000) have demonstrated that this MOTOR SKILL LEARNING 563 structure constitutes a storage site for this form of motor memory. One might therefore expect that the striatum (Doyon & Ungerleider, 2002) and possibly the motor cortex (Wise et al., 1998) would play an equally important role in the consolidation of movement

sequences, as these structures contribute to the development and maintenance of the final representation of this type of motor skill. At present, however, this remains a working hypothesis, awaiting experimental investigation.

REFERENCES

Brashers-Krug, T., Shadmehr, R., & Bizzi, E. (1996). Consolidation in human motor memory. Nature, 382, 252–255.

Classen, J., Liepert, J., Wise, S. P., Hallett, M., & Cohen, L. G. (1998). Rapid plasticity of human cortical movement representation induced by practice. Journal of Neurophysiology, 79, 1117–1123.

Donoghue, J. P. (1995). Plasticity of adult sensorimotor representations. Current Opinion in Neurobiology, 5, 749–754.

Donoghue, J. P., Hess, G., & Sanes, J. N. (1996). Motor cortical substrates and mechanisms for learning. In J.

R. Bloedel, T. J. Ebner, & S. P. Wise (Eds.), Acquisition of motor behavior in vertebrates (pp. 363–386).

Cambridge, MA: MIT Press.

Doyon, J. (1997). Skill learning. In J. D. Schmahmann (Ed.), The cerebellum and cognition (pp. 273–294). San

Diego: Academic Press.

Doyon, J., Gaudreau, D., Laforce, R. J., Castonguay, M., Bedard, P. J., Bedard, F., & Bouchard, J. P. (1997).

Role of the striatum, cerebellum, and frontal lobes in the learning of a visuomotor sequence. Brain and Cognition, 34, 218–245.

Doyon, J., Owen, A. M., Petrides, M., Sziklas, V., & Evans, A. C. (1996). Functional anatomy of visuomotor skill learning in human subjects examined with positron emission tomography. European Journal of

Neuroscience, 8, 637–648.

Doyon, J., & Ungerleider, L. G. (2002). Functional anatomy of motor skill learning. In L. R. Squire and D. L.

Schacter (Eds.), Neuropsychology of memory, 3rd ed. New York: Guilford.

Gordon, A. M., Lee, J. H., Flament, D., Ugurbil, K., & Ebner, T. J. (1998). Functional magnetic resonance imaging of motor, sensory, and posterior parietal cortical areas during performance of sequential typing

movements. Experimental Brain Research, 121, 153–166.

Grafton, S. T., Woods, R. P., & Mike, T. (1994). Functional imaging of procedural motor learning: Relating cerebral blood flow with individual subject performance. Human Brain Mapping, 1, 221–234.

Greenough, W. T., Larson, J. R., & Withers, G. S. (1985). Effects of unilateral and bilateral training in a reaching task on dendritic branching of neurons in the rat motor-sensory forelimb cortex. Behavioral and Neural

Biology, 44, 301–314.

Hess, G., Aizenman, C. D., & Donoghue, J. P. (1996). Conditions for the induction of long-term potentiation in layer II/III horizontal connections of the rat motor cortex. Journal of Neurophysiology, 75, 1765–1778.

Hess, G., & Donoghue, J. P. (1996a). Long-term depression of horizontal connections in rat motor cortex. European Journal of Neuroscience, 8, 658–665.

Hess, G., & Donoghue, J. P. (1996b). Long-term potentiation and long-term depression of horizontal connections in rat motor cortex. Acta Neurobiological Experimentalis (Warsaw), 56, 397–405.

Huntley, G. W. (1997). Correlation between patterns of horizontal connectivity and the extent of short-term representational plasticity in rat motor cortex. Cerebral Cortex, 7, 143–156.

Hyden, H., & Lange, P.W. (1983). Modification of membrane-bound proteins of the hippocampus and entorhinal cortex by change in behavior in rats. Journal of Neuroscience Research, 9, 37–46.

Imamizu, H., Miyauchi, S., Tamada, T., Sasaki, Y., Takino, R., Putz, B., Yoshioka, T., & Kawato, M. (2000). Human cerebellar activity reflecting an acquired internal model of a new tool. Nature, 403, 192–195.

Jacobs, K. M., & Donoghue, J. P. (1991). Reshaping the cortical motor map by unmasking latent intracortical connections. Science, 251, 944–947.

564 UNGERLEIDER, DOYON, AND KARNI

Jenkins, I. H., Brooks, D. J., Nixon, P. D., Frackowiak, R. S., & Passingham, R. E. (1994). Motor sequence learning: A study with positron emission tomography. Journal of Neuroscience, 14, 3775–3790.

Karni, A. (1996). The acquisition of perceptual and motor skills: A memory system in the adult human cortex. Cognitive Brain Research, 5, 39–48.

Karni, A., Meyer, G., Jezzard, P., Adams, M. M., Turner, R., & Ungerleider, L. G. (1995). Functional MRI evidence for adult motor cortex plasticity during motor skill learning. Nature, 377, 155–158.

Karni, A., Meyer, G., Rey-Hipolito, C., Jezzard, P., Adams, M. M., Turner, R., & Ungerleider, L. G. (1998). The acquisition of skilled motor performance: Fast and slow experience-driven changes in primary motor cortex. Proceedings of the National Academy of Science of the United States of America, 95, 861–868.

Karni, A., & Sagi, D. (1993). The time course of learning a visual skill. Nature, 365, 250–252.

Kleim, J. A., Lussnig, E., Schwarz, E. R., Comery, T. A., & Greenough, W. T. (1996). Synaptogenesis and

Fos expression in the motor cortex of the adult rat after motor skill learning. Journal of Neuroscience, 16, 4529–4535.

Kleim, J. A., Barbay, S., Cooper, N. R., Hogg, T. M., Reidel, C. N., Remple, M. S., & Nudo, R. J. (2002). Motor learning-dependent synaptogenesis is localized to functionally reorganized motor cortex. Neurobiological Learning and Memory, 77, 63–77.

Kolb, B., Buhrmann, K., McDonald, R., & Sutherland, R. J. (1994). Dissociation of the medial prefrontal, posterior parietal, and posterior temporal cortex for spatial navigation and recognition memory in the rat.

Cerebral Cortex, 4, 664–680.

Merzenich, M. M., & Sameshima, K. (1993). Cortical plasticity and memory. Current Opinion in Neurobiology, 3, 187–196.

Middleton, F. A., & Strick, P. L. (1997). Cerebellar output channels. International Review of Neurobiology, 41, 61–82.

Nudo, R. J., Milliken, G. W., Jenkins, W. M., & Merzenich, M. M. (1996). Use-dependent alterations of movement representations in primary motor cortex of adult squirrel monkeys. Journal of Neuroscience, 16, 785–807.

Nudo, R. J., Plautz, E. J., & Milliken, G. W. (1997). Adaptive plasticity in primate motor cortex as a consequence

of behavioral experience and neuronal injury. Seminars in Neuroscience, 9, 13–23.

Rey-Hipolito, C., Adams, M. M., Ungerleider, L. G., & Karni, A. (1997). When practice makes perfect: Time dependent evolution of skilled motor performance. Society for Neuroscience Abstracts, 23, 1052. Sanes, J. N., & Donoghue, J. P. (2000). Plasticity and primary motor cortex. Annual Review of Neuroscience, 23, 393–415.

Schieber, M. H. (1995). Muscular production of individuated finger movements: The role of extrinsic finger muscles. Journal of Neuroscience, 15, 284–297.

Shadmehr, R., & Holcomb, H. H. (1997). Neural correlates of motor memory consolidation. Science, 277, 821–825.

Wise, S. P., Moody, S. L., Blomstrom, K. J., & Mitz, A. R. (1998). Changes in motor cortical activity during visuomotor adaptation. Experimental Brain Research, 121, 285–299.

Withers, G. S.,&Greenough,W. T. (1989). Reach training selectively alters dendritic branching in subpopulations of layer II–III pyramids in rat motor–somatosensory forelimb cortex. Neuropsychologia, 27, 61–69.