Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Oct 14, 2003; 100(21): 12492–12497.
Published online Oct 6, 2003. doi:  10.1073/pnas.2035019100
PMCID: PMC218785

Multiple shifts in the representation of a motor sequence during the acquisition of skilled performance


When do learning-related changes in performance occur? Here we show that the knowledge of a sequence of movements evolves through several distinctive phases that depend on two critical factors: the amount of practice as well as the passage of time. Our results show the following. (i) Within a given session, large performance gains constituted a signature for motor novelty. Such gains occurred only for newly introduced conditions irrespective of the absolute level of performance. (ii) A single training session resulted in both immediate but also time-dependent, latent learning hours after the termination of practice. Time in sleep determined the time of expression of these delayed gains. Moreover, the delayed gains were sequence-specific, indicating a qualitative change in the representation of the task within 24 h posttraining. (iii) Prolonged training resulted in additional between-session gains that, unlike the effects of a single training session, were confined to the trained hand. Thus, the effects of multisession training were qualitatively different than the immediate and time-dependent effects of a single session. Altogether, our results indicate multiple time-dependent shifts in the representation of motor experience during the acquisition of skilled performance.

Several lines of evidence support the notion that the brain exhibits a significant degree of experience-dependent functional plasticity even in adulthood (110). This plasticity may underlie the acquisition and long-term retention of skills (procedural memory) (1113). There is growing evidence, however, indicating that different brain areas are involved in the initial, compared with subsequent, phases of learning after practice in a given motor as well as nonmotor task (1422). These practice-related changes in the set of brain areas engaged by a repeating task were reported to occur mostly within a single session. However, recent behavioral data have shown that the acquisition of skilled performance occurs on a time scale of hours, days, and weeks (2, 5, 7, 1113, 18, 2124). A leading notion, the “power law of practice,” suggests that the evolution of skilled performance is determined solely by the number of task repetitions (15, 7, 1113, 25, 26); there are, nevertheless, compelling indications that the passage of time is also an important factor in the acquisition of skills (2, 24, 2729).

Based on behavioral and imaging studies, two distinct stages in skill acquisition were proposed: early, relating to within-session improvement, and late, slow changes in performance that can be observed across several (daily) sessions of practice (1, 2, 5, 7, 10, 12, 13, 28). To account for robust delayed gains in performance that emerged after a latent period of more than a few hours after a single-session training, the notion of an intermediate phase corresponding to the posttraining hours has been proposed (2, 7, 13, 2731). The conjecture is that a process of memory consolidation requiring time to become effective (in terms of performance) can be triggered by the training experience under certain conditions and requires time to become effective (in terms of performance) in both perceptual and motor tasks [but also in the acquisition of cognitive skills (12)]. Recent studies further suggest that sleep may contribute significantly to the development of the delayed gains in this type of learning (28, 30, 32).

It is not clear, however, whether the effects of a single training session, with ample time given for the evolution of delayed gains, can be conceptualized as the unit of skill acquisition, i.e., that multisession training gain constitutes but the sum of incremental gains of a number of single sessions. An earlier study (2) showed that all gains in speed of performance after completing long-term training on a sequence of movements were highly restricted by the physical parameters of the training experience, with no transfer to the untrained hand or to different arrangements of the trained movement components comprising the sequence. Here we show that this remarkable specificity evolves in a stepwise manner both within and between sessions. Within a given session, large performance gains occurred only for newly introduced conditions irrespective of the absolute level of performance. Although after a single session qualitative and quantitative changes in performance emerged in time, further qualitative rather than just quantitative changes occurred during multisession training.


Participants. Eighty-three young adults [mean, 23 (1734) years] were paid to train on a finger-opposition task. Participants were right-handed, had no medical conditions that could impair fine motor performance, reported >6 h of sleep per night, and had no sleep–wake-cycle disruptions. Musicians, professional typists, and evening-type persons [Circadian Clock Questionnaire, based on Horne and Ostberg (33)] were excluded. Informed consent was obtained before the experiment.

The Task. The participants performed the instructed movements while lying supine with the head immobilized in a head holder. The hand was positioned on the subject's chest with the elbow flexed, in direct view (palm-facing) of a video camera, to allow recording of all digit movements. Visual feedback was not afforded. The motor task was as described (1). Subjects were instructed to oppose the fingers of the left (nondominant) hand to the thumb in a given sequence (Fig. 1). In the initial session each participant underwent a pretraining performance test, a training session, and an immediate posttraining performance test. Each training session consisted of 160 repetitions of a randomly assigned sequence (A or B) that were divided into 10 training blocks. During training, the initiation of each sequence was cued by an auditory signal at a rate of 0.4 Hz (2.5 sec per sequence). For performance testing, participants were provided only with an initial cue and were instructed to continuously tap the sequence as rapidly and accurately as possible until given a stop signal (test block, 30 sec). Each test session consisted of four blocks. The intervals between test blocks were kept constant (50 sec). No feedback was provided. Participants were instructed that occasional errors should not be corrected and to continue with the task without pause as smoothly as possible.

Fig. 1.
The finger-to-thumb opposition task. The two sequences were matched for number of movements per digit and mirror-reversed in relation to each other (in terms of order).

Three transfer conditions were tested: performance of the trained sequence with the untrained hand and of a reversed sequence of identical component movements with both the trained and untrained hand. The trained condition was always tested first, then the reversed sequence with the trained hand, and then, randomly ordered, the two sequences with the untrained hand.

Design and Procedure. Experiment 1. The performance of 36 participants was assessed before a single practice session, immediately after the session, and 24 and 48 h posttraining (experiment 1a). After completing this phase, 16 participants were randomly assigned to continue training for five additional sessions, spaced 1–3 days apart, on the same sequence as in the initial session (experiment 1b). On average this phase took 15 ± 4 days. Forty-eight hours after the initial training session and again after the completion of prolonged training, subjects were tested in the three transfer conditions. Two participants who did not participate in the prolonged training phase were retested >5 months after the single training session. As a separate control group, eight additional participants were trained for a single session and retested 5 h posttraining.

Experiment 2. To assess the transfer capability by the end of a single training session, eight subjects were given training as described for experiment 1 and tested on the three transfer conditions immediately after training.

Experiment 3. To test the relative contributions of time and time in sleep to the posttraining (between-session) gains in performance, we measured the speed and accuracy of performance in another group of 15 subjects using the same task at two sessions spaced 12 h apart from each other. An over-day (OD) (n = 7) group was tested and trained in the morning and retested in the evening with no sleep allowed during the day, whereas the participants of the overnight (ON) (n = 8) group were first tested and trained in the evening and retested the next morning after normal sleep. Both groups were retested 24 h posttraining.

Experiment 4. Control experiments with 16 additional subjects were used to compare the baseline differences in the performance of the sequences between the two hands. Each subject performed the chosen sequence by both hands, with the order of the hands balanced across subjects. Statistical analysis revealed no hand (right vs. left, P = 0.12) or sequence (A or B, P = 0.47) differences in baseline (untrained) performance. Speed and accuracy data were analyzed by using a general linear model. The statistical analysis of block-by-block accuracy gains was influenced by the fact that, throughout training, participants committed very few errors.


Training on a given sequence of movements resulted in early (within-session) and delayed (posttraining, time-dependent) gains in performance triggered by a single training session in terms of both speed and accuracy (experiment 1a; Fig. 2). Immediate posttraining testing showed significant gains in speed with no loss in accuracy compared with baseline. The average gain in speed was 5.6 sequences (equal to 28 additional opposition movements) executed within the 30-sec duration of the test period. However, 24 h posttraining, additional robust gains in speed were observed concurrently with a significant reduction in the absolute number of errors. The average gain in speed 24 h posttraining, compared with the immediate posttraining performance, was 3.5 sequences (>17 additional opposition movements), and the average number of errors was reduced from 1.5 to 0.7 per test block. The improvement in performance throughout this limited time window was stepwise with no improvement evident during the first 5 h afterthetermination of training. Five hours posttraining, both speed and accuracy were maintained at the level achieved immediately after training [F(1) = 0.257, P = 0.628]. This process of latent improvement seems to near asymptote by 24 h, although 48 h posttraining, small but significant additional gains in speed were observed [approximately three additional opposition movements; F(1) = 7.286, P = 0.011]. Analysis of the differences in performance speed during the testing occasions demonstrated that the critical period for the posttraining improvement were the first 24 h after the training session (t = 7.074, P < 0.001).

Fig. 2.
Fast (within-session) and slow (posttraining) gains (experiment 1a). Two indices of performance are plotted for the first 48 h after a single training session: speed, mean number of correct sequences (Upper), and accuracy, mean number of errors (Lower ...

To probe the nature of the internal representations presumably subserving the large gains in performance triggered by a single training session, the ability of participants to generalize their experience under different task conditions was measured 48 h posttraining. There were significant practice-related gains for all transfer conditions relative to the baseline performance of the trained sequence (Fig. 2). The effects of training were transferred almost completely to the right hand both in terms of speed and accuracy. Transfer to the untrained hand was not affected by the order of tested conditions. Although performance in the first block of the transfer condition was significantly slower [F(1) = 8.49, P = 0.006], the difference between the transfer and the trained conditions disappeared by the third test block [F(1) = 2.76, P = 0.11]. Accuracy gains transferred completely to the nontrained hand [F(1) = 1.75, P = 0.21]. In contrast, there was only limited transfer of the time-dependent gains to the performance of the reversed sequence by either hand: None of the delayed gains were transferable to the nontrained sequence. Only the within-session gains were transferred [P, nonsignificant, trained condition (LtT) immediately after training vs. left hand-reversed (LtR) and right hand-reversed (RtR) sequences 48 h posttraining].

A comparison of the transfer pattern 48 h posttraining to the immediate posttraining transfer tests (experiment 2) suggested a strengthening of a sequence-specific representation of the task with the passage of time (Fig. 3). Immediately after training the participants showed a clear advantage for the performance of the trained sequence in the trained hand [F(1) = 24.73, P = 0.002] but no difference in speed between the trained and control sequence performance in the untrained hand. {A trend for reduction [F(1) = 5.40, P = 0.05] in the number of errors in the trained sequence is obtained.} By 48 h posttraining, however, the delayed gains in both hands were specific for the trained sequence (both speed and errors; Fig. 2). These findings suggest that the delayed gains 48 h after the completion of a single training session were sequence-specific but not effector-dependent. Thus, our results indicate an early setting up and time-dependent strengthening of a motor routine specific for the execution of the trained sequence of movements at an effector-independent level. Throughout training, subjects made very few errors. No speed–accuracy tradeoff was evident in all tested conditions. Moreover, the correlation between performance rate and errors was negative (r = -0.84, P < 0.001), indicating concurrent gains in both speed and accuracy.

Fig. 3.
Immediate posttraining performance (experiment 2). Baseline and immediate posttraining scores for LtT and the three transfer conditions (LtR, RtR, and RtT). Arrow, the training interval. Speed (Upper) and accuracy (Lower) are shown as described for Fig. ...

To test the relative contributions of time and time in sleep to the posttraining (between-session) gains in performance, we measured the speed and accuracy of performance in another group of individuals using the same task at two sessions spaced 12 h apart from each other (experiment 3, Fig. 4). An OD group was tested and trained in the morning and retested in the evening with no sleep allowed during the day, whereas the participants of the ON group were first tested and trained in the evening and retested the next morning after normal sleep. All individuals of the ON group showed a significant improvement in speed during the 12-h period [F(1) = 46.53, P < 0.001]. In contrast, participants from the OD group, without sleep, did not improve during the 12-h period [F(1) = 1.29, P = 0.29]. Both groups, however, achieved significant delayed gains by 48 h after the training session [OD: F(1) = 82.25, P < 0.001; ON: F(1) = 73.09, P < 0.001]. These findings indicate that the effects of training, although not expressed as gains during the 12 h of the awaking state (OD group), became effective after a subsequent night sleep and were as large as the gains of the ON group with sleep closely after the training experience. This suggests that the delayed effects of training were maintained in some latent form during the waking state for at least 12 h.

Fig. 4.
Contributions of time and time in sleep to motor performance during the first 24 h posttraining (experiment 3). (Upper) ON group. (Lower) OD group. Baseline, immediately, and 12- and 48-h posttraining scores for the trained condition. Arrow, the training ...

A previous study showed effective retention of the skill after prolonged (5 weeks) practice 1 year after training was stopped (2). To study the long-term effects of a single session of training, two subjects were retested after an interval of 5 and 8 months in the trained and transfer conditions (experiment 1a, Fig. 5). The performance gains attained 48 h after a single session of training were fully retained. Moreover, the pattern of transfer at 5 and 8 months posttraining was similar to that found at 48 h posttraining, indicating the preservation of the effector independent representation of the trained sequence for many months after a single learning experience.

Fig. 5.
Long-term retention after a single training session (experiment 1a). Shown is performance 48 h posttraining and the retention (Re) test. The retention intervals were 5 and 8 months for participants HD and DI, respectively. T, trained; LR, left reversed; ...

There were additional, significant, incremental gains in speed with no loss in accuracy as a function of prolonged training (experiment 1b). Performance almost leveled off by the sixth session with a difference of less than one sequence gained between the fifth and sixth sessions. The performance 48 h after the initial training session, compared with that after the completion of the five additional training sessions, is shown in Fig. 6. Performance speed for the trained sequence in the trained hand robustly improved [F(5) = 33.7, P < 0.001], with accuracy maintained at the level attained 48 h posttraining (P, nonsignificant).

Fig. 6.
Single versus multisession training (experiment 1b). Performance 48 h after a single training session compared with performance after five additional training sessions. Arrows, training sessions. Speed (Upper) and accuracy (Lower) are shown as described ...

Fig. 6 shows that the large performance gains achieved after multisession training were much less susceptible to transfer. Whereas the gains 48 h after training were found to be specific for the movement sequence but effector-invariant, the long-term training gains were both sequence-specific and effector-dependent. Switching from the left to the right (untrained) hand after long-term training resulted in a significantly slower performance of the trained sequence (F = 32.05, P < 0.001). Fig. 6 Lower shows that, in parallel to the lack of transfer of the additional gains in speed, the absolute number of errors significantly increased both when switching to the alternate movement sequence and to the different effector after long-term training. Thus, the results of the transfer tests after multisession training indicated a critical shift in the nature of the representation of the trained sequence compared with the outcome of a single session of training immediately, 48 h, or months posttraining.

Although robust changes occur on a time scale of hours and days, our results also demonstrated that important changes in performance occurred on a much shorter time scale. Whenever participants were introduced to a new task condition, they performed differentially as a function of two independent aspects of experience. The initial performance level highly depended on previous experience with the task (including experience with a different hand or a different sequence of movements). However, the actual experience (familiarity) with any specific task condition was reflected in the changes in performance. Our data show that whenever (early as well as late in practice) novel conditions were introduced, there were very rapid block-by-block gains (Figs. (Figs.22 and and6).6). To test this hypothesis, we calculated for all individual subjects the slopes of linear regression lines that were fitted to the four test blocks of each test condition (Fig. 7). The comparison between the slopes of all the novel conditions with the trained conditions demonstrated significant differences between the two groups (t = 6.22, P < 0.001, two-sample t test). The novel conditions included all conditions in which no specific training was given: LtT at baseline, LtR, RtR, and RtT 48 h after the first training session, and LtR, RtR, and RtT after prolonged training. The trained conditions included all testing occasions with the trained sequence after a single training session immediately and 24 and 48 h posttraining as well as after multisession training. When performing a nontrained, novel condition, all participants showed a robust block-by-block improvement reflected in the positively inclined slopes. The limited experience (four test blocks) with an untrained condition during the first transfer testing (after the initial training session) did not affect the rate of improvement significantly in the second (after prolonged training) transfer test. In contrast, the trained conditions were characterized by minimal changes in performance on consecutive test blocks, resulting in rather flat regression lines. The magnitude of this novelty effect, however, was not related to the absolute level of performance. Thus, although participants showed robust transfer in terms of speed and accuracy of the training gains to the untrained hand 48 h posttraining, the slopes of the fitted linear regression lines clearly indicated that performing the sequence with the untrained hand constituted a novel experience.

Fig. 7.
Block-by-block changes in performance (novelty effect). Data points are the mean slopes of linear regression lines fitted to the four blocks of each test condition (experiment 1b). Novel, all conditions in which no specific training was given (from left ...


Altogether, our results indicate that the knowledge gained from a training experience undergoes a number of important qualitative in addition to quantitative changes. Experience exerts its effects on performance in different time windows: on the scale of 1 min (block-by-block gains) within a single training session and also hours after the termination of the single training experience. Moreover, training beyond a single session could result in a subsequent shift in the representation of a trained movement sequence (to a less abstract level) and thus narrow the transfer of the acquired knowledge to a nontrained motor effector.

The very fast gains accrued during the testing of new (untrained) conditions were evident on a time scale of minutes independent of the level of performance and effector. These rapid gains, as indicated by the slopes of the regression lines fitted to the four successive test blocks, occurred only at the very beginning of training and were not found on any of the subsequent measurements of performance in the trained condition. In the transfer conditions, in which only very limited experience was given (constituting the test itself) these rapid gains were consistently found both 48 h posttraining and after the long-term training condition. We propose that these rapid gains can be considered a marker for the novelty of the experience. Thus, rapid performance gains within a given session may represent the tuning of a previously established motor routine after repeated iterations of the task under novel task conditions [as suggested to occur in perceptual learning (29)]. This phase in motor learning can be conceptualized therefore as the motor system analogue of perceptual or cognitive repetition priming (12). From this perspective, this “novelty effect” may reflect the ability of the motor system to differentiate familiar from unfamiliar experiences. Moreover, our results show that this differentiation is achieved regardless of the absolute performance level and thus may depend on amount of experience (e.g., the number of task iterations) rather than parameters of actual performance.

A leveling off (saturation) of block-by-block performance gains was shown recently to be a critical requirement for the induction of delayed gains in a perceptual (counting) task (12). The emergence of delayed gains depended on the saturation of the block-by-block performance gains within the practice session, indicating persistent adaptation. Our results indicate that practice to a point below the saturation of the rapid gains, as in the transfer conditions, was insufficient to switch off the novelty effect. Rapid gains thus can be related to the notion of motor adaptation as a setting up of a motor routine in a given novel setting (5, 21, 22, 24). Thus, our findings suggest the possibility that the saturation of the block-by-block gains within a given practice session may constitute an important factor in triggering long-term motor learning. How much experience is needed to saturate the novelty effect is an open question. Four successive (30 sec long) test blocks were not sufficient for turning off the novelty effect; after prolonged training, the transfer conditions (previously experienced 48 h posttraining) were still novel. In the present experimental design, because all the conditions were tested within a few minutes of each other, mutual interference effects (e.g., refs. 12, 24, 34) could not be ruled out as a factor limiting performance. However, relative to the main effect of learning, the amount of interference between conditions was presumably minimal, because there was no elevation in error rates in all the posttraining conditions.

Delayed gains in performance, requiring time (hours) and time in sleep to evolve, were reported in perceptual learning in human adults and in animals (7, 12, 13, 27, 28, 3032). The more commonly recognized aspect of consolidation, most often relating to nonprocedural memory, designates a process that modifies the memory of a given experience into a more robust and enduring form (35, 36). Such immunity to interference was redemonstrated recently for motor learning by showing that two conflicting motor tasks can be learned and retained but only if the training on each task was separated in time by a critical interval of 4–5 h (5, 21, 24). In line with previous conjectures concerning perceptual learning (7, 12, 13), we propose the parsimonious notion that the evolution of delayed gains and of immunity to interference may be different aspects of time-dependent processes that are triggered by a critical amount of experience. Thus, similar neural mechanisms (consolidation processes) may subserve the two delayed effects of training in motor learning (2, 11).

The first 24 h posttraining constitute a time window during which a stepwise gain in performance occurred. Our results show that during the 48 h after a single (first) training session, the delayed gains were on the order of the gains attained during (within) the session. Five hours and, in another group of subjects, even 12 h posttraining no changes relative to the immediate posttraining measurements were found. By 24 h posttraining the process of latent improvement seems to near asymptote. Although 48 h posttraining small but significant additional gains in speed were observed with no loss in accuracy, it is not clear whether these additional gains indicate that the process of consolidation, triggered by the single training session, can continue beyond the first 24 h posttraining. One cannot rule out that these gains were induced by the additional training inherent to the measurement of performance at 24 h. In line with indications from animal studies (35), a recent study has suggested that (procedural memory) consolidation processes in a perceptual task indeed may extend beyond the first 24-h posttraining interval (12).

Several mechanisms were proposed to underlie memory formation during the consolidation period, the leading notion being that the process is based on experience-triggered de novo protein synthesis, leading to long-lasting changes in synaptic efficacy (36, 37). The underlying assumption is that the processes subserving the consolidation of memory occur within a selected group of excitatory neurons that have been active in the training experience, i.e., are restricted to the neuronal populations that were engaged directly in the performance of the trained task (38). However, recent behavioral and physiological studies suggest that this classical Hebbian idea needs to be extended to accommodate the evidence that different neuronal populations are involved in early and late stages of learning, i.e., that intensive practice and consolidation processes may result in the recruitment of a different set of brain areas other than that involved early on in practice (2, 6, 17, 21).

Recent studies (28, 30) suggest that time in sleep is considerably more effective than time spent in the awake state for the establishment of delayed gains. The question of whether delayed improvement is just a function of time or whether sleep is especially important for consolidation of recently acquired procedural memory is still an important open question (39). Total and partial sleep deprivations were shown to alter subsequent delayed gains in performance in both explicit and implicit memory tasks (27, 28, 40, 41). Our results provide further support for the contribution of time in sleep to the learning of a motor task. However, we also show that sleep may be critical to the timing of the (behavioral) expression of the delayed gains in motor learning. The delayed effects of training were as large 24 h posttraining in both the OD and ON groups, suggesting that the effects of training were maintained in some latent form during the waking state for at least 12 h. Indeed, similar findings were reported by Walker et al. (figure 2 B and C in ref. 30). The presence or absence of sleep thus may determine the time point in which time-dependent processes become effective.

One cannot rule out that circadian effects by themselves may influence the neuronal activity subserving consolidation in the motor system. However, some indirect indication against circadian effects comes from the finding that within-session motor learning and baseline performance were similar for OD and ON subjects. Furthermore, Fischer et al. (28) showed that performance improvement was observed equally after daytime and nighttime sleep. Altogether, the results suggest that the manipulation of time of training relatively to the time of sleep may contribute significantly to the optimization of various training protocols.

A key finding of our study is that the representation of the task after multisession training (Fig. 6) was qualitatively different from the representation 48 h after a single training session (Fig. 5), i.e., after allowing ample time for consolidation. This difference could not be ascribed to the passage of time per se, because the pattern of transfer several months after a single training session was found to be similar to the one measured 48 h posttraining, indicating an effector-independent representation of the trained sequence of movements. In contrast, the transfer pattern after a single session as compared with the results of multisession training in the same individuals indicates a shift to both a sequence-specific and effector-dependent representation of the task. Thus, the final long-term representation of a task may be either effector-dependent or effector-independent according to the history of training and, specifically, to the parsing of training into multiple sessions. The effects of multisession training, therefore, turned out to be not simply incremental (additive, as a classical learning curve would predict) but rather resulted in a qualitative shift in the knowledge gained from the experience.

These findings are in line with a suggestion by Hikosaka et al. (6), based on trial-and-error learning in monkeys, that a sequential motor procedure is acquired independently by two cortical systems, one using spatial and the other motor coordinates. The former is rapid and flexible, relying on attention and working memory. The latter is slowly evolving and dependent on the availability of prolonged practice. Our results show, however, that consolidated, long-term memory of the task can be maintained at a non-effector-dependent level of representation (i.e., accessible to the untrained hand) if only a single training session (with practice to asymptotic performance) is afforded. Thus, long-term motor memory does not necessarily depend on effector-specific representations. Our results suggest that the switch to a hand-specific representation may occur as a result of more extensive practice and/or the parsing of training experience into multiple sessions.

Contrary to the notion of skills as corresponding to more general performance routines (42), our results, in agreement with others (6, 7, 12, 17, 43), clearly show that more training resulted in more specific gains. Thus, practice makes transfer imperfect. Altogether, our results raise critical challenges to a leading model of learning, the power law of practice: both time after practice and the parsing of experience should be treated as important factors in models of procedural memory acquisition.


This research was partially supported by the Dominic–Einhorn Foundation and the Israeli Ministry of Health (Office of the Chief Scientist).


Abbreviations: OD, over-day; ON, overnight; LtT, trained condition; LtR, left hand-reversed; RtR, right hand-reversed; RtT, right hand-trained.


1. Karni, A., Meyer, G., Jezzard, P., Adams, M. M., Turner, R. & Ungerleider, L. G. (1995) Nature 377, 155-158. [PubMed]
2. Karni, A., Meyer, G., Rey-Hipolito, C., Jezzard, P., Adams, M. M., Turner, R. & Ungerleider, L. G. (1998) Proc. Natl. Acad. Sci. USA 95, 861-868. [PMC free article] [PubMed]
3. Sanes, J. N. & Donoghue, J. P. (2000) Annu. Rev. Neurosci. 23, 393-415. [PubMed]
4. Gilbert, C. D., Sigman, M. & Crist, R. E. (2001) Neuron 31, 681-697. [PubMed]
5. Brashers-Krug, T., Shadmehr, R. & Bizzi, E. (1996) Nature 382, 252-255. [PubMed]
6. Hikosaka, O., Nakahara, H., Rand, M. K., Saka, I. K., Lu, X., Nakamura, K., Miyachi, S. & Doya, K. (1999) Trends Neurosci. 22, 464-471. [PubMed]
7. Karni, A. & Sagi, D. (1993) Nature 365, 250-252. [PubMed]
8. Singer, W., Tretter, F. & Yinon, U. (1982) J. Physiol. (London) 324, 239-248. [PMC free article] [PubMed]
9. Recanzone, G. H., Schreiner, C. E. & Merzenich, M. M. (1993) J. Neurosci. 13, 87-103. [PubMed]
10. Nudo, R. J., Milliken, G. W., Jenkins, W. M. & Merzenich, M. M. (1996) J. Neurosci. 16, 785-807. [PubMed]
11. Salmon, D. P. & Butters, N. (1995) Curr. Opin. Neurobiol. 5, 184-190. [PubMed]
12. Hauptmann, B. & Karni, A. (2002) Brain Res. Cognit. Brain Res. 13, 313-322. [PubMed]
13. Karni, A. & Bertini, G. (1997) Curr. Opin. Neurobiol. 7, 530-535. [PubMed]
14. Rand, M. K., Hikosaka, O., Miyachi, S., Lu, X., Nakamura, K., Kitaguchi, K. & Shimo, Y. (2000) Exp. Brain Res. 131, 293-304. [PubMed]
15. Raichle, M. E., Fiez, J. A., Videen, T. O., MacLeod, A. M., Pardo, J. V., Fox, P. T. & Petersen, S. E. (1994) Cereb. Cortex 4, 8-26. [PubMed]
16. Toni, I., Krams, M., Turner, R. & Passingham, R. E. (1998) NeuroImage 8, 50-61. [PubMed]
17. Tracy, J. I., Faro, S. S., Mohammed, F., Pinus, A., Christensen, H. & Burkland, D. (2001) Brain Res. Cognit. Brain Res. 10, 303-316. [PubMed]
18. Hikosaka, O., Rand, M. K., Miyachi, S. & Miyashita, K. (1995) J. Neurophysiol. 74, 1652-1661. [PubMed]
19. Grafton, S. T., Mazziotta, J. C., Presty, S., Friston, K. J., Frackowiak, R. S. & Phelps, M. E. (1992) J. Neurosci. 12, 2542-2548. [PubMed]
20. Gordon, A. M., Lee, J. H., Flament, D., Ugurbil, K. & Ebner, T. J. (1998) Exp. Brain Res. 121, 153-166. [PubMed]
21. Shadmehr, R. & Holcomb, H. H. (1997) Science 277, 821-825. [PubMed]
22. Doyon, J., Penhune, V. & Ungerleider, L. G. (2003) Neuropsychologia 41, 252-262. [PubMed]
23. Keele, S. W., Jennings, P., Jones, S., Caulton, D. & Cohen, A. (1995) J. Mot. Behav. 27, 17-30.
24. Shadmehr, R. & Brashers-Krug, T. (1997) J. Neurosci. 17, 409-419. [PubMed]
25. Kirsner, K., Speelman, C. & Schofield, P. (1993) in Implicit Memory: New Developments, eds. Graf, P. & Masson, M. (Lawrence Erlbaum, Hillsdale, NJ).
26. Newell, K. M., Liu, Y. T. & Mayer-Kress, G. (2001) Psychol. Rev. 108, 57-82. [PubMed]
27. Stickgold, R., James, L. & Hobson, J. A. (2000) Nat. Neurosci. 3, 1237-1238. [PubMed]
28. Fischer, S., Hallschmid, M., Elsner, A. L. & Born, J. (2002) Proc. Natl. Acad. Sci. USA. 99, 11987-11991. [PMC free article] [PubMed]
29. Karni, A. (1996) Brain Res. Cognit. Brain Res. 5, 39-48. [PubMed]
30. Walker, M. P., Brakefield, T., Morgan, A., Hobson, J. A. & Stickgold, R. (2002) Neuron 35, 205-211. [PubMed]
31. Karni, A., Tanne, D., Rubenstein, B. S., Askenasy, J. J. & Sagi, D. (1994) Science 265, 679-682. [PubMed]
32. Maquet, P., Schwartz, S., Passingham, R. & Frith, C. (2003) J. Neurosci. 23, 1432-1440. [PubMed]
33. Horne, J. A. & Ostberg, O. (1976) Int. J. Chronobiol. 4, 97-110. [PubMed]
34. Goedert, K. M. & Willingham, D. B. (2002) Learn. Mem. 9, 279-292. [PMC free article] [PubMed]
35. McGaugh, J. L. (2000) Science 287, 248-251. [PubMed]
36. Dudai, Y. (2002) Curr. Opin. Neurobiol. 12, 211-216. [PubMed]
37. Zipser, D., Kehoe, B., Littlewort, G. & Fuster, J. (1993) J. Neurosci. 13, 3406-3420. [PubMed]
38. Hebb, D. O. (1949) The Organization of Behavior (Wiley, New York).
39. Laureys, S., Peigneux, P., Perrin, F. & Maquet, P. (2002) Neuron 35, 5-7. [PubMed]
40. Karni, A. (1995) Lancet 345, 395. [PubMed]
41. Peigneux, P., Laureys, S., Delbeuck, X. & Maquet, P. (2001) NeuroReport 12, A111-A124. [PubMed]
42. Poldrack, R. A. & Gabrieli, J. D. (2001) Brain 124, 67-82. [PubMed]
43. Rand, M. K., Hikosaka, O., Miyachi, S., Lu, X. & Miyashita, K. (1998) Exp. Brain Res. 118, 293-297. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...