| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1Division of Kinesiology, 2Department of Psychology, and 3Departments of Biomedical Engineering and Radiology, University of Michigan, Ann Arbor, Michigan
Submitted 24 October 2007; accepted in final form 10 February 2008
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Brashers-Krug et al. 1996; Krakauer et al. 2005; McCracken and Stelmach 1977; Seidler 2004, 2005; Zanone and Kelso 1997). Interference (negative transfer) can also occur, particularly when transfer is to a task with opposing requirements (cf. Bock et al. 2001; Caithness et al. 2004; Krakauer et al. 1999; Tong et al. 2002; Wigmore et al. 2002).
There has been an explosion of interest in skill acquisition in recent years, leading to the development of comprehensive theories regarding the neural bases of different forms of skill learning (cf. Ashe et al. 2006; Doyon and Benali 2005). Investigations of negative transfer have also attracted much interest, contributing to theories of motor memory consolidation (for recent reviews see Krakauer and Shadmehr 2006; Roberston et al. 2004). However, despite the ubiquitous nature of positive transfer in everyday life and the wealth of behavioral studies investigating this phenomenon (Bock et al. 2001; Brashers-Krug et al. 1996; Krakauer et al. 2005; McCracken and Stelmach 1977; Seidler 2004, 2005; Zanone and Kelso 1997), the neural bases of motor transfer have yet to be determined. We have recently shown that, although advancing age leads to declines in motor acquisition, the amount of savings that older adults show at transfer is equivalent to that of young adults (Seidler 2007), suggesting that the underlying mechanisms of the two processes differ.
One major difference between acquisition and transfer is that transfer should require retrieval of a previously acquired motor memory, whereas naïve learning does not. It is not clear whether starting with this template completely alters the neural bases of learning or merely allows the learner to move more quickly through the stages of learning. That is, it is not known whether transfer is a form of accelerated learning, with brain activation that overlaps with that seen in late learning. Previous investigations of savings would suggest that this is the case (Kojima et al. 2004; Medina et al. 2001; Smith et al. 2006). These studies demonstrate that a memory of prior learning remains, which can be relied on for faster subsequent learning. Smith et al. (2006) developed a computer model of short-term motor adaptation consisting of two processes: a fast learning module that was also fast to forget and a slow learning module that was slow to forget. The model successfully re-created several aspects of adaptation, including savings. Presumably, brain regions more active during the fast, early stages of learning would not be reengaged at savings because their memory for the learning decays quickly.
The purpose of the current investigation was to use functional magnetic resonance imaging (fMRI) to study the neural bases of transfer through the use of a visuomotor adaptation task. Participants adapted to three different rotations of the visual feedback display. Similar to previous investigations of savings (Kojima et al. 2004; Medina et al. 2001; Smith et al. 2006), participants underwent a washout session between each adaptive experience. This ensured that subjects initiated each adaptive experience from the same baseline performance level. We hypothesized that brain activation would be altered by prior learning history. Specifically, we predicted that activation at transfer (learning following prior task experience) would reveal evidence of retrieval of an internal model, representing the newly acquired mapping between visual and motor space. This would be reflected as activation in the cerebellum surrounding the posterior superior fissure at transfer of learning (Graydon et al. 2005; Imamizu et al. 2000, 2003). Furthermore, we predicted that regions typically involved with the early phases of adaptation would show a reduction in activity at transfer compared with that at acquisition, including the bilateral basal ganglia, and the right-lateralized prefrontal, premotor, and parietal cortex (Clower et al. 1996; Ghilardi et al. 2000; Inoue et al. 1997, 2000; Krakauer et al. 2004; Seidler et al. 2006).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Twenty-six right-handed young adults participated in this study. They were compensated for their participation, which took an average of 2.0 h. All participants signed a consent form approved by the Medical Institutional Review Board of the University of Michigan and filled out a health history questionnaire prior to participation. They were randomly assigned to one of two groups, with each following a different task order (see following text for details). Group 1 consisted of six women and eight men, with a mean age of 23.4 yr (SD = 3.9). Group 2 consisted of seven women and five men, with a mean age of 24.3 yr (SD = 5.0).
Experimental setup and procedure
Participants lay supine in a 3.0-Tesla magnet. Head motion was prevented by placement of padding around the head and the use of a soft headband. Participants moved a dual axis, potentiometer joystick with the right hand to hit targets presented on a liquid crystal display screen. Real-time feedback of the joystick position was provided as a small cursor moving on the screen, using a scaling factor of 1.0. The cursor was visible to participants throughout the entire movement. Movements were always initiated from a central home position (0.8-cm-diameter target) on the display screen. A target (0.8-cm diameter) appeared every 4 s, either 4.8 cm to the right, left, above, or below the centrally located home position. Participants were instructed to move the cursor representing the joystick position into the target as quickly as possible on target appearance and to hold the cursor within the target until it disappeared (2 s following its appearance). Participants were instructed to release their grip on the joystick handle at this point, allowing the elastic spring-loaded device to recenter for the next trial. The subsequent trial began 2 s later, resulting in an intertrial interval (from one target presentation to the next) of 4 s. Movements were performed in blocks of 24 trials, with a 30-s visual fixation rest period at the beginning and end of each block. Subjects visually fixated the start position during this control period. Data for each block were collected as separate fMRI "runs" of rest-task-rest, with a brief break between each run.
Participants adapted to three visuomotor rotations, with a return to normal display conditions between each adaptive experience. The rotation magnitudes were 15, 30, and 45° clockwise rotations of the cursor position about the central start location (cf. Cunningham and Welch 1994). That is, to achieve a target in the 45° rotation condition, subjects would need to aim 45° counterclockwise of the target. Group 1 subjects acquired the visuomotor rotations in the following order: 30, 15, 45°, whereas Group 2 subjects acquired them in the order: 45, 15, 30° (see Table 1 for details of trial presentation). The term learning refers to performance during the first adaptive experience (Table 1, blocks 3-5; 30° for Group 1, 45° for Group 2), whereas the term transfer refers to performance during the last adaptive experience (Table 1, blocks 13-15; 45° for Group 1, 30° for Group 2). This counterbalanced design allowed us to assess transfer of learning for subjects that learned the 45° rotation last compared with those that learned it first and for subjects that learned the 30° rotation last compared with those that acquired it first. A between-subjects design was necessary in this case to ensure that we were comparing subjects acquiring the same rotation magnitude (i.e., 30° first vs. 30° last) because we have previously shown that brain activation differs for participants adapting to either a 45 or a 30° rotation (Seidler et al. 2006). We exposed the participants to three rotations in an effort to maximize the amount of transfer that subjects might show (cf. Roller et al. 2001). Participants were not informed in advance as to whether the upcoming block was an adaptation or control block, nor were they provided with any information about the applied rotations. They were instructed to hit the target as rapidly as possible and to attempt to minimize both reaction time and movement time.
|
). The field of view (FOV) was 220 x 220 mm, voxel size was 3.4 x 3.4 x 4 mm, TR (repeat time to accomplish a full volume) was 2 s, and TE (echo time) was 25 ms. Thirty-two contiguous axial slices were acquired, encompassing the whole brain (including cerebellum). Structural images for data presentation and coregistration were acquired in the same slice locations using T1-weighted gradient echo pulse sequence (TE/TR/FA = 5 ms/300 ms/90°, FOV = 220 x 220 mm, voxel size = 0.86 x 0.86 x 4 mm). We also recorded the x and y positions of the joystick at a rate of 100 Hz for subsequent data analyses. Behavioral data processing
We analyzed the joystick data off-line, using custom data analysis routines. We first filtered the data with a dual low-pass Butterworth digital filter (cf. Winter 1990), using a cutoff frequency of 10 Hz. We computed the resultant joystick path by taking the square root of the sum of the squared x and y coordinate data at each time point. The tangential velocity profile was then computed via differentiation. Movement onset and offset were calculated by applying the optimal algorithm of Teasdale et al. (1993) to this velocity profile for each movement. We computed direction error (DE) to assess performance during adaptation. This is the angle between the line connecting the start and target positions (in joystick coordinates) and the line connecting the start with the spatial location of the joystick at the time of peak velocity (see Fig. 1). Since this measure was calculated at the time of peak velocity, it was not contaminated by on-line error corrections. Although DE was the primary variable that we focused on, we also computed the movement time of the initial ballistic movement toward the target (Sub1MT) and the reaction time (RT) for each trial to allow for comparison with a previous control study (Seidler et al. 2004). To determine the Sub1MT, we first divided movements into their initial ballistic movement toward the target and any subsequent submovement corrections (Meyer et al. 1988). The algorithm that we used searches for a period of acceleration following a period of deceleration or a change in the sign of the velocity. Thus the initial ballistic movement has "ended" when there is either a change in movement direction or an additional propulsive action is made. The Sub1MT was computed as the duration between movement onset and the end of the initial movement. RT is the time between appearance of the target and movement onset.
|
) was evaluated to determine whether the RM data met the assumption of sphericity (
>0.75). In cases where sphericity was met, the univariate tests were used to maintain power. Otherwise the P values were interpreted using the Huynh-Feldt adjusted degrees of freedom. fMRI data processing
We discarded the first two volumes per run to allow the MRI signal to reach its steady state. Movement correction was performed using the Automated Image Registration package (Woods et al. 1998). The amounts of translation and rotation about each of the axes required to achieve this were examined to ensure that subjects did not move their head >3 mm during the experiment. Mean and SD of the head translation motion estimates (in mm) were 0.59 (0.27) X, 0.35 (0.15) Y, and 0.33 (0.17) Z. Structural images were skull-stripped using FSL's Brain Extraction Tool (http://www.fmrib.ox.ac.uk/fsl). We used SPM99 for subsequent data analyses (Wellcome Department of Cognitive Neurology, London, UK; Friston et al. 1995) in the Matlab environment (The MathWorks, Sherborn, MA). A mean functional image was computed for each subject. The structural image was then coregistered to this mean image and spatially normalized to the Montreal Neurological Institute template (Evans et al. 1994). The obtained normalization parameters were applied to the subject's functional images. These functional images were spatially smoothed with a Gaussian smoothing kernel with a full width at half-maximum of 8 mm.
We created boxcar models, time-locked to the effect of interest and convolved with an estimate of the hemodynamic response function, for statistical analysis. We used high-pass filtering during analysis to remove low-frequency drift. Analyses were first performed at the single-subject level (fixed effects) across runs, comparing the task and visual fixation rest periods. The brain regions associated with the first learning experience have been reported in Seidler et al. (2006). To determine how the neural correlates of learning are altered by prior experience, we compared the initial learning experience in one group to the final learning experience in the other group. That is, we compared activation in block 3 of Group 1 subjects with that in block 13 of Group 2 subjects. Likewise, we compared activation in block 3 of Group 2 subjects with that in block 13 of Group 1 subjects. A similar comparison was performed for blocks 4 and 14 and blocks 5 and 15. Since the results for transfer to the 30 and 45° conditions did not differ qualitatively, we collapsed data across the 30 and 45° conditions to provide more power. In this way, we compared the initial and final learning experiences of each group.
We searched for regions that had either greater or less activation at transfer (last three adaptation blocks) than they did at acquisition (first three adaptation blocks). These single-subject results were then combined into a random-effects group analysis, which treats intersubject variability as a random effect, allowing statistical inference at the population level (Penny et al. 2003). These group analyses were evaluated using a threshold of P 
0.001 (Z 
2.98) and a minimum cluster size of 10 voxels. With these contrasts, it is not possible to determine whether significant results arise due to less activation at transfer of learning or more deactivation at acquisition. To answer this question, we extracted the time course from the voxel showing the highest Z score for each significant cluster of activation and computed the percentage signal change for each block. The percentage signal change was determined relative to the average of the control periods at the beginning and end of each respective block. That is, percentage signal change for block 1 was computed using the visual fixation rest periods for block 1, and so on.
Significant areas of activation were localized using the Talairach atlas (Talairach and Tournoux 1988), with medial motor areas identified as in Picard and Strick (1996) and cerebellar regions as in Schmahmann et al. (2000).
Correlation analyses
The relatively large number of subjects participating in this study allows for 1) an assessment of individual differences in acquisition and transfer and 2) their underlying brain activation patterns. For each subject, we quantified their success with acquisition by fitting a regression line to the DE across trials within the first adaptation block. We quantified success with transfer by fitting a regression line to the DE across trials within the first block of the final learning experience. Performance curves are typically well fit by power functions across the time course of learning. However, for these initial trials, we found that a linear regression line provided the best fit for individual subjects. We then determined which brain regions were correlated across subjects with the slopes of these lines at acquisition and at transfer by entering each subject's slope into a regression analysis within the statistical parametric mapping (SPM) program. That is, one analysis was performed in which we tested for brain regions that exhibited activation levels during the first block of acquisition that were correlated with the DE slope across trials. The second analysis searched for brain regions that showed activation during the first block of transfer that were correlated with the DE slope at transfer. These analyses searched for regions showing activation versus rest during the first block of transfer (or learning) that was scaled across subjects according to DE slope. In both cases, we evaluated the contrast for positive correlations only [greater activation associated with a steeper (absolute value) slope across DE, indicating faster learning]. These analyses were evaluated using a threshold of P 0.001 (Z 2.98) and a minimum cluster size of 10 voxels. We then performed a conjunction analysis within SPM across these two analyses to determine whether there was any brain activation in common between the two correlation analyses.
Evaluating the DE slope across learning and transfer trials allows us to examine brain regions correlated with the rate of learning at acquisition and transfer. Moreover, it has the advantage that the rate of change of a variable can be dissociated from the average amplitude of the variable, potentially allowing for a separation of learning and transfer effects from performance effects. Since previous studies have shown that brain activation varies with movement time, distance of movement, and speed of movement, we performed additional correlation analyses between the DE slope measure and these other performance variables: movement time, total distance, peak speed, and the time spent making corrective adjustments.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Figure 1A shows trajectories from the first two movements made by a representative participant under the 45° rotation condition. It is clear that these trajectories are displaced from the target path by about 45°, requiring the participant to make large, hook-like corrections late in the movement. Figure 1B shows the last two trials from the same individual (dark gray lines). Evidence of adaptation is seen in the straight trajectories toward the targets. The brain regions associated with adaptation for the two groups have been previously reported (Seidler et al. 2006). We found activation in the cerebellum and sensory and motor cortices, as previously reported for this type of learning (Clower et al. 1996; Ghilardi et al. 2000; Imamizu et al. 2000; Inoue et al. 1997, 2000; Krakauer et al. 2004). We also found bilateral basal ganglia activation during the first block of adaptation (Seidler et al. 2006). We observed activation in the right globus pallidus and putamen, along with the right prefrontal, parietal, and premotor cortices. We proposed that this right lateralized activation is associated with the early, spatial cognitive aspects of adaptation (cf. Eversheim and Bock 2001). We also observed activation in the left globus pallidus and caudate nucleus, along with left premotor and supplementary motor cortices, which may support the sensorimotor processes of adaptation (Seidler et al. 2006).
Transfer of a visuomotor mapping
BEHAVIORAL DATA. Figure 1B shows two movement trajectories made by a subject performing under the 45° rotation condition, after having already adapted to 30 and 15° rotations (light gray lines). These trajectories are clearly more directed toward the target than those presented in Fig. 1A, reflecting a savings in the rate of adaptation due to prior learning history.
Before assessing the degree of savings shown, it is important to first verify whether the washout blocks were successful at restoring performance to baseline (BL) levels. This was achieved by comparing performance from the second pretest block with block 2 of the subsequent two washout periods (i.e., the second blocks of all pairs of baseline blocks were compared for differences). Specifically, we tested whether the mean of the last three trials from each of these blocks differed significantly from each other for each of the groups. These values did not differ for either of the groups for DE (pairwise comparisons of these values within each group, P > 0.10 in all cases), indicating that each adaptive experience was initiated from the same level. The amount of original learning that took place for each group was also quantified because this could potentially influence the amount of subsequent savings that participants would show. The maximum possible amount of learning was estimated as the difference in performance from the mean of the last three pretest trials and the mean of the first three adaptation trials. The amount of learning that occurred was then estimated as the performance level at the end of adaptation (mean of the last three trials of A3) as a percentage of the maximum possible amount of learning. Thus a value of 100% would mean that learning was "complete" (i.e., performance returned to baseline levels) and a value of 0% would mean that no learning had taken place (i.e., performance for the last adaptation trials was no different from the initial adaptation trials). The mean value for Group 1 (30° adaptation) was 69%, whereas that for Group 2 (45° adaptation) was 59%. The two groups were not significantly different (P > 0.10).
Figure 2 presents DE values for the 15, 30, and 45° adaptation experiences for Groups 1 and 2. The top panel of Fig. 2 shows DE for the groups adapting to the 15° rotation, which was the second adaptive experience for both groups. Assessing whether there is a difference in rate of adaptation for this condition allows us to determine whether the effects of "far" transfer (adapting to 15° after learning 45°) differ from those of "near" transfer (adapting to 15° after learning 30°). There was, however, no group difference in adaptation to the 15° rotation. Initial acquisition of and transfer to the 30° rotation condition are shown in the middle panel of Fig. 2. The two groups have equivalent performance for the two pretest blocks. Once the adaptive stimulus (rotation of feedback display) is introduced, errors increase for both groups (A1 always shows significantly higher errors than the preceding BL block, repeated contrasts on block; P < 0.01). At first glance, it may appear that errors did not increase "enough" to match the magnitude of the rotation (i.e., A1 errors might be expected to be –30°), but averaging across the entire block brings down the mean value. That is, errors averaged about –30° for the first couple of trials within A1, but then were rapidly reduced with practice, across both trials and blocks [block x trial interaction across the adaptation blocks, F(22,482) = 5.1, P < 0.01; trial main effect, F(23,506) = 2.9, P < 0.01; block main effect, F(2,44) = 6.3, P < 0.01]. Savings at transfer is reflected in the fact that the performance improvement with practice proceeds more quickly across trials for the group adapting to the 30° rotation after having already experienced the 45 and 15° rotations (depicted with square symbols in the middle panel), in comparison to the group experiencing 30° as their first rotation [depicted with circle symbols in the middle panel; group x block x trial interaction across the adaptation blocks, F(46,1012) = 2.8, P < 0.01]. The experienced participants are also faster at readapting performance to the baseline visual display [group x block x trial interaction across the aftereffect blocks, F(23,506) = 1.8, P = 0.01]. A similar performance savings at transfer is observed in the bottom panel of Fig. 2, where DE is plotted for participants either acquiring or transferring to the 45° rotation condition [group x block x trial interaction across the adaptation blocks, F(46,1012) = 2.8, P < 0.01]. The top panel of Fig. 3 presents the group average across trials for the first block of 45°, at both learning and transfer, with participants showing a faster rate of change across trials at transfer than at learning.
|
|
NEUROIMAGING DATA.
Table 2 and Fig. 4 show brain regions that exhibited reduced activation at the first block of transfer compared with the first block of acquisition. These areas included the right primary motor cortex, the right inferior frontal gyrus, the cingulate motor area (CCZ in the nomenclature of Picard and Strick 1996), the right inferior temporal gyrus, and the cerebellum (see Fig. 5 for signal change values). Similarly, at the second block of adaptation, the right inferior frontal gyrus and inferior temporal gyrus again had less activation for motor transfer compared with that for acquisition. At the third block, the right inferior frontal gyrus, cerebellum, and bilateral thalamus had less activity at transfer of learning. All of these effects were due to less activation at transfer, as opposed to greater deactivation at acquisition (see Fig. 5). There were no areas that significantly increased their activation at transfer compared with the level shown at acquisition.
|
|
|
). To verify that these results reflect rate of learning and transfer, as opposed to simply reflecting kinematic performance levels, we computed the correlation between the learning/transfer slopes and movement time, total distance, peak speed, and the time spent making corrective adjustments (see Table 4). Only one of these relationships was significant (and even this would not survive a P value adjusted for multiple comparisons): there was a correlation between the slope of DE across trials at transfer to the 45° condition and the total distance traveled on average during this block (r = 0.47, P < 0.05). On the whole, however, there was no relationship between the slope measures and the kinematic performance variables. Additionally, individual differences in slope measures were not influenced by the level of performance on the preceding washout block. For transfer to the 30° condition, the correlation between slope across DE at transfer and DE at the end of the preceding washout block (average of the last five trials) was r = 0.08 (P = 0.42). For transfer to the 45° condition, the correlation between slope across DE at transfer and DE at the end of the preceding washout block (average of the last five trials) was r = 0.04 (P = 0.51).
|
|
|
).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Brashers-Krug et al. 1996; Krakauer et al. 2005; McCracken and Stelmach 1977; Seidler 2004, 2005; Zanone and Kelso 1997), we found behavioral evidence for positive transfer of learning across the subsequent adaptation sessions. Even though the washout blocks brought performance back to baseline levels, participants still demonstrated faster adaptation based on their prior experience with the task. In the current experiment, we sought to determine whether the brain regions associated with transfer of learning (learning following previous experience) differ from those associated with acquisition of a new visuomotor mapping, reflecting evidence of prior learning history. Our previous work demonstrated an age-related dissociation between acquisition and transfer of sensorimotor adaptation, raising the possibility that these two aspects of learning differ in terms of their underlying mechanisms (Seidler 2007). In the current investigation, we found that savings at transfer of learning was associated with a reduction in activity of brain regions typically recruited early in the adaptation process, including the right inferior frontal gyrus, primary motor cortex, inferior temporal gyrus, and the cerebellum (medial HIII). Although the right inferior frontal gyrus, right inferior temporal gyrus, and cerebellum HIII were consistently depressed at transfer compared with learning, there were other areas that showed less consistent effects (Table 2). These findings suggest that not only does the extent of involvement of brain regions differ from learning to transfer, but the time course of contribution is also altered.
The rate of adaptation at transfer was correlated with activity in the right cingulate gyrus, left superior parietal lobule, right inferior parietal lobule, left middle occipital gyrus, and bilaterally in the cerebellum, in regions surrounding the posterior superior fissure, which is thought to be the site of storage of internal models (Bursztyn et al. 2006; Imamizu et al. 2000, 2003; although for a contrasting view see Pasalar et al. 2006). Thus it seems that the neural correlates associated with transfer of learning look more like late learning activation (cf. Graydon et al. 2005; Imamizu et al. 2000), including cerebellar activity that is likely associated with retrieval of the previously formed internal model. This model would then need to be modified to meet the new task demands.
It is known that brain activation changes with movement rate (Jancke et al. 1998; Schlaug et al. 1996; Turner et al. 1998), force (Ashe 1997), and size of errors (Kitazawa et al. 1998), raising the possibility that performance differences between acquisition and transfer may have resulted in the activation effects that we observed. To address this issue for the current study, we conducted a control experiment in the context of the task used here (Seidler et al. 2004). We induced performance changes in the absence of learning in the control experiment by having subjects move a joystick to hit targets of differing sizes in a counterbalanced fashion. The target-to-target distance was the same amplitude as that in the current investigation. The variations in target size induced changes in movement time, velocity, and magnitude of corrective actions, as occurred during adaptation in the current study. The differences in these values from learning to transfer in the current study are well within the range of differences in these values across the four target sizes in the 2004 study (see Table 6). These similarities in kinematic performance across the two experiments validate the use of the control study to account for performance effects. In the current study, we found no areas that had greater activation at transfer than at acquisition. However, the reduction in activation in the right primary motor cortex, the cingulate motor area, and the medial cerebellum at transfer may be due to performance differences between the two groups since these sites overlap (based on qualitative comparison) with those seen in the control study.
In our previous work, we did not find that the lateral cerebellum exhibited changing activation for different levels of motor performance (Seidler et al. 2004). Further indication that the lateral cerebellar activation we observed at transfer of learning is not simply related to differences in error magnitude between acquisition and transfer is that different brain regions were correlated with error magnitude during acquisition compared with transfer and the DE slope at transfer was not correlated with the absolute level of DE. We found that cerebellar activation surrounding the posterior superior fissure was correlated with the slope of DE across trials at transfer of learning but not during initial acquisition. This region is thought to be the site of plasticity for acquired internal models (Imamizu et al. 2000, 2003). Although the studies by Imamizu and colleagues do not present the coordinates of cerebellar activation, they depict activity in the area surrounding the posterior, superior fissure of the cerebellum for adaptation of hand movements to altered presentation of position and velocity feedback. Moreover, the three-dimensional grid of cerebellar activation presented in Fig. 4 of Imamizu et al. (2003) shows coordinates that overlap with those observed in the current study (see Table 3). We suggest that successful retrieval of the previously acquired internal model allows adaptation to proceed more quickly at transfer.
We have previously proposed that the right lateralized activation seen early in the visuomotor adaptation process, including in the right prefrontal, premotor, and parietal cortices and the basal ganglia, is associated with the more cognitive aspects of adaptation (Seidler et al. 2006), including processes such as spatial attention and spatial working memory (Eversheim and Bock 2001). In the current study, we found consistent decreases in activation in the right inferior frontal gyrus, corresponding with BA 45 and 47, at transfer. This finding implies that transfer of adaptation is less cognitively demanding than acquisition, and that previous experience with similar tasks may allow a learner to move more quickly through the cognitive, early stages of learning. The regions that we found to be correlated with individual differences in the amount of savings at transfer included the right cingulate gyrus, left superior parietal lobule, right inferior parietal lobule, left middle occipital gyrus, bilateral dorsal premotor cortex, and bilaterally in the cerebellum, in regions surrounding the posterior superior fissure. These findings suggest that transfer may involve a fine-tuning of the visuospatial aspects of the preexisting internal model, once it has been retrieved from the cerebellum. Indeed, previous work has shown that outputs from internal models are sent to premotor regions after learning has occurred (Imamizu et al. 2007; Tamada et al. 1999). Another alternative is that the internal model may be represented in a distributed network encompassing cerebellar, parietal, occipital, and premotor regions.
One potential limitation of the current study was that adaptation was not "complete." That is, performance did not return to baseline levels by the end of each set of three adaptation blocks. However, performance did recover between 60 and 70% (with 100% reflecting a complete return to baseline levels) and, moreover, the amount of learning that occurred was enough to allow for faster adaptation at transfer (i.e., positive transfer). It is not clear how the neuroimaging results would differ if participants had more practice trials for each adaptive experience, but we speculate that activation at transfer would look even more like late learning in this case. It is difficult to speculate whether similar results might be found for other conditions of motor transfer, for example to another task or for movements made with another effector. We would predict, however, that the activation would be similar to that observed here, with evidence of retrieval of previous learning, combined with evidence of faster subsequent learning.
It is likely that both interference and facilitation are simultaneously occurring in our task, with both positive and negative transfer taking place between the variants of the adaptation task. We cannot dissociate between the two but, based on the behavioral data, it is clear that the net overall effect is savings (positive transfer). Moreover, it has been shown that washout blocks between multiple learning experiences, such as used in the current study, reduce interference (Krakauer et al. 2005). We have shown that this transfer of learning is associated with activation in the lateral cerebellum surrounding the posterior superior fissure, which likely supports retrieval of a preexisting internal model. The existence of this model appears to allow subjects to move more quickly through the earliest phases of learning and proceed directly to later learning processes.
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: R. D. Seidler, 401 Washtenaw Ave., Ann Arbor, MI 48109-2214 (E-mail: rseidler{at}umich.edu)
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Ashe J, Lungu OV, Basford AT, Lu X. Cortical control of motor sequences. Curr Opin Neurobiol 16: 213-221, 2006.[CrossRef][Web of Science][Medline]
Bock O, Schneider S, Bloomberg J. Conditions for interference versus facilitation during sequential sensorimotor adaptation. Exp Brain Res 138: 359-365, 2001.[CrossRef][Web of Science][Medline]
Brashers-Krug T, Shadmehr R, Bizzi E. Consolidation in human motor memory. Nature 382: 252-255, 1996.[CrossRef][Medline]
Bursztyn LL, Ganesh G, Imamizu H, Kawato M, Flanagan JR. Neural correlates of internal-model loading. Curr Biol 16: 2440-2445, 2006.[CrossRef][Web of Science][Medline]
Caithness G, Osu R, Bays P, Chase H, Klassen J, Kawato M, Wolpert DM, Flanagan JR. Failure to consolidate the consolidation theory of learning for sensorimotor adaptation tasks. J Neurosci 24: 8662-8671, 2004.
Clower DM, Hoffman JM, Votaw JR, Faber TL, Woods RP, Alexander GE. Role of posterior parietal cortex in the recalibration of visually guided reaching. Nature 383: 618-621, 1996.[CrossRef][Medline]
Cunningham HA, Welch RB. Multiple concurrent visual-motor mappings: implications for models of adaptation. J Exp Psych Hum Percept Perform 20: 987-999, 1994.[CrossRef]
Doyon J, Benali H. Reorganization and plasticity in the adult brain during learning of motor skills. Curr Opin Neurobiol 15: 161-167, 2005.[CrossRef][Web of Science][Medline]
Evans AC, Kamber M, Collins DL, Macdonald D. An MRI-based probabilistic atlas of neuroanatomy. In: Magnetic Resonance Scanning and Epilepsy (vol. 264 of NATO Advanced Study Institute, Series A, Life Sciences), edited by Shorvon S, Fish D, Andermann F, Bydder GM, Stefan H. New York: Plenum Press, 1994, p. 263-274.
Eversheim U, Bock O. Evidence for processing stages in skill acquisition: a dual-task study. Learn Mem 8: 183-189, 2001.
Friston KJ, Holmes AP, Worsley KJ, Poline JP, Frith CD, Frackowiak RSJ. Statistical parametric maps in functional imaging: a general linear approach. Hum Brain Mapp 2: 189-210, 1995.[Medline]
Ghilardi M, Ghez C, Dhawan V, Moeller J, Mentis M, Nakamura T, Antonini A, Eidelberg D. Patterns of regional brain activation associated with different forms of motor learning. Brain Res 871: 127-145, 2000.[CrossRef][Web of Science][Medline]
Graydon FX, Friston KJ, Thomas CG, Brooks VB, Menon RS. Learning-related fMRI activation associated with a rotational visuo-motor transformation. Cogn Brain Res 22: 373-383, 2005.[CrossRef][Medline]
Huynh H, Feldt LS. Conditions under which the mean square ratios in repeated measures designs have exact F-distributions. J Am Stat Assoc 65: 1582-1589, 1970.[CrossRef]
Imamizu H, Higuchi S, Toda A, Kawato M. Reorganization of brain activity for multiple internal models after short but intensive training. Cortex 43: 338-349, 2007.[CrossRef][Web of Science][Medline]
Imamizu H, Kuroda T, Miyauchi S, Yoshioka T, Kawato M. Modular organization of internal models of tools in the human cerebellum. Proc Natl Acad Sci USA 100: 5461-5466, 2003.
Imamizu H, Miyauchi S, Tamada T, Sasaki Y, Takino R, Putz B, Yoshioka T, Kawato M. Human cerebellar activity reflecting an acquired internal model of a new tool. Nature 403: 192-195, 2000.[CrossRef][Medline]
Inoue K, Kawashima R, Satoh K, Kinomura S, Goto R, Sugiura M, Ito M, Fukuda H. Activity in the parietal area during visuomotor learning with optical rotation. Neuroreport 18: 3979-3983, 1997.
Inoue K, Kawashima R, Satoh K, Kinomura S, Sugiura M, Goto R, Ito M, Fukuda H. A PET study of visuomotor learning under optical rotation. Neuroimage 11: 505-516, 2000.[CrossRef][Web of Science][Medline]
Jancke L, Specht K, Mirzazade S, Loose R, Himmelbach M, Lutz K, Shah NJ. A parametric analysis of the "rate effect" in the sensorimotor cortex: a functional magnetic resonance imaging analysis in human subjects. Neurosci Lett 252: 37-40, 1998.[CrossRef][Web of Science][Medline]
Kitazawa S, Kimura T, Yin PB. Cerebellar complex spikes encode both destinations and errors in arm movements. Nature 392: 494-497, 1998.[CrossRef][Medline]
Kojima Y, Iwamoto Y, Yoshida K. Memory of learning facilitates saccadic adaptation in the monkey. J Neurosci 24: 7531-7539, 2004.
Krakauer JW, Ghez C, Ghilardi MF. Adaptation to visuomotor transformations: consolidation, interference, and forgetting. J Neurosci 25: 473-478, 2005.
Krakauer JW, Ghilardi MF, Ghez C. Independent learning of internal models for kinematic and dynamic control of reaching. Nat Neurosci 2: 1026-1031, 1999.[CrossRef][Web of Science][Medline]
Krakauer JW, Ghilardi MF, Mentis M, Barnes A, Veytsman M, Eidelberg D, Ghez C. Differential cortical and subcortical activations in learning rotations and gains for reaching: a PET study. J Neurophysiol 91: 924-933, 2004.
Krakauer JW, Shadmehr R. Consolidation of motor memory. Trends Neurosci 29: 58-64, 2006.[CrossRef][Web of Science][Medline]
McCracken HD, Stelmach GE. A test of the schema theory of discrete motor learning. J Mot Behav 9: 193-201, 1977.[Web of Science]
Medina JF, Garcia KS, Mauk MD. A mechanism for savings in the cerebellum. J Neurosci 21: 4081-4089, 2001.
Meyer DE, Abrams RA, Kornblum S, Wright CE, Smith JEK. Optimality in human motor performance: ideal control of rapid aimed movements. Psychol Rev 95: 340-370, 1988.[CrossRef][Web of Science][Medline]
Noll DC, Cohen JD, Meyer CH, Schneider W. Spiral K-space MR imaging of cortical activation. J Magn Reson Imaging 5: 49-56, 1995.[Web of Science][Medline]
Pasalar S, Roitman AV, Durfee WK, Ebner TJ. Force field effects on cerebellar Purkinje cell discharge with implications for internal models. Nat Neurosci 9: 1404-1411, 2006.[CrossRef][Web of Science][Medline]
Penny WD, Holmes AP, Friston KJ. Random effect analysis. In: Human Brain Function (2nd ed.), edited by Frackowiak RSJ, Friston KJ, Frith C, Dolan R, Price CJ, Zeki S, Ashburner J, Penny WD. New York: Academic Press, 2003, p. 843-850.
Picard N, Strick PL. Motor areas of the medial wall: a review of their location and functional activation. Cereb Cortex 6: 342-353, 1996.
Robertson EM, Pascual-Leone A, Miall RC. Current concepts in procedural consolidation. Nat Rev Neurosci 5: 576-582, 2004.[CrossRef][Web of Science][Medline]
Roller CA, Cohen HS, Kimball KT, Bloomberg JJ. Variable practice with lenses improves visuo-motor plasticity. Brain Res Cogn Brain Res 12: 341-352, 2001.[CrossRef][Medline]
Schlaug G, Sanes JN, Thangaraj V, Darby DG, Jancke L, Edelman RR, Warach S. Cerebral activation covaries with movement rate. Neuroreport 7: 879-883, 1996.[Web of Science][Medline]
Schmahmann JD, Doyon J, Toga A, Evans A, Petrides M. MRI Atlas of the Human Cerebellum. San Diego, CA: Academic Press, 2000.
Schmidt RA, Young DE. Transfer of motor control in motor skill learning. In: Transfer of Learning: Contemporary Research and Applications, edited by Cormier SM, Hagman JD. Orlando, FL: Academic Press, 1987, p. 47-79.
Seidler RD. Multiple motor learning experiences enhance motor adaptability. J Cogn Neurosci 16: 65-73, 2004.[CrossRef][Web of Science][Medline]
Seidler RD. Differential transfer processes in incremental visuomotor adaptation. Motor Control 9: 40-58, 2005.[Web of Science][Medline]
Seidler RD. Aging dissociates motor learning and generalization of learning. Learn Mem 14: 17-21, 2007.
Seidler RD, Noll DC, Chintalapati P. Basal ganglia activation associated with sensorimotor adaptation. Exp Brain Res 175: 544-555, 2006.[CrossRef][Web of Science][Medline]
Seidler RD, Noll DC, Thiers G. Feedforward and feedback processes in motor control. Neuroimage 22: 1775-1783, 2004.[CrossRef][Web of Science][Medline]
Smith MA, Ghazizadeh A, Shadmehr R. Interacting adaptive processes with different timescales underlie short-term motor learning. PLoS Biol 4: 1035-1043, 2006.[Web of Science]
Talairach J, Tournoux P. Co-Planar Stereotaxic Atlas of the Human Brain. 3-Dimensional Proportional System: An Approach to Cerebral Imaging, translated by Rayport M. New York: Thieme Medical Publishers, 1988.
Tamada T, Miyauchi S, Imamizu H, Yoshioka T, Kawato M. Cerebro-cerebellar functional connectivity revealed by the laterality index in tool-use learning. Neuroreport 10: 325-331, 1999.[Web of Science][Medline]
Teasdale N, Bard C, Fleury M, Young D, Proteau L. Determining movement onsets from temporal series. J Mot Behav 25: 97-106, 1993.[Web of Science][Medline]
Tong C, Wolpert DM, Flanagan JR. Kinematics and dynamics are not represented independently in motor working memory: evidence from an interference study. J Neurosci 22: 1108-1113, 2002.
Turner RS, Grafton ST, Votaw JR, Delong MR, Hoffman JM. Motor subcircuits mediating the control of movement velocity: a PET study. J Neurophysiol 80: 2162-2176, 1998.
Wigmore V, Tong C, Flanagan JR. Visuomotor rotations of varying size and direction compete for a single internal model in motor working memory. J Exp Psychol Hum Percept Perform 28: 447-457, 2002.[CrossRef][Web of Science][Medline]
Winter DA. Biomechanics and Motor Control of Human Movement (2nd ed.). New York: Wiley, 1990.
Woods RP, Grafton ST, Holmes CJ, Cherry SR, Mazziotta JC. Automated image registration. I. General methods and intrasubject registration. J Comput Assist Tomogr 22: 139-152, 1998.[CrossRef][Web of Science][Medline]
Zanone PG, Kelso JAS. Coordination dynamics of learning and transfer: collective and component levels. J Exp Psychol Hum Percept Perform 23: 1454-1480, 1997.[CrossRef][Web of Science][Medline]
This article has been cited by other articles:
|
|
 |
|
  M. A. Perez, S. Tanaka, S. P. Wise, D. T. Willingham, and L. G. Cohen Time-Specific Contribution of the Supplementary Motor Area to Intermanual Transfer of Procedural Knowledge J. Neurosci., September 24, 2008; 28(39): 9664 - 9669. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
