| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Human Motor Control Section and 2 Cognitive Neuroscience Section, Medical Neurology Branch, National Institute of Neurological Disorders and Stroke, Bethesda, 20892-1428; 3 Laboratory of Systems Neuroscience, National Institute of Mental Health, Poolesville, Maryland 20837; and 4 Institut National de la Santé et de la Recherche Médicale, Centre d'Exploration et de Recherche Medicales par Emission de Positrons, 69003 Lyon, France
 |
ABSTRACT |
|---|
 Abstract
 Introduction
Methods
Results
Discussion
References
|
|---|
Deiber, M.-P., S. P. Wise, M. Honda, M. J. Catalan, J. Grafman, and M. Hallett. Frontal and parietal networks for conditional motor learning: a positron emission tomography study. J. Neurophysiol. 78: 977-991, 1997. Studies on nonhuman primates show that the premotor (PM) and prefrontal (PF) areas are necessary for the arbitrary mapping of a set of stimuli onto a set of responses. However, positron emission tomography (PET) measurements of regional cerebral blood flow (rCBF) in human subjects have failed to reveal the predicted rCBF changes during such behavior. We therefore studied rCBF while subjects learned two arbitrary mapping tasks. In the conditional motor task, visual stimuli instructed which of four directions to move a joystick (with the right, dominant hand). In the evaluation task, subjects moved the joystick in a predetermined direction to report whether an arrow pointed in the direction associated with a given stimulus. For both tasks there were three rules: for the nonspatial rule, the pattern within each stimulus determined the correct direction; for the spatial rule, the location of the stimulus did so; and for the fixed-response rule, movement direction was constant regardless of the pattern or its location. For the nonspatial rule, performance of the evaluation task led to a learning-related increase in rCBF in a caudal and ventral part of the premotor cortex (PMvc, area 6), bilaterally, as well as in the putamen and a cingulate motor area (CM, area 24) of the left hemisphere. Decreases in rCBF were observed in several areas: the left ventro-orbital prefrontal cortex (PFv, area 47/12), the left lateral cerebellar hemisphere, and, in the right hemisphere, a dorsal and rostral aspect of PM (PMdr, area 6), dorsal PF (PFd, area 9), and the posterior parietal cortex (area 39/40). During performance of the conditional motor task, there was only a decrease in the parietal area. For the spatial rule, no rCBF change reached significance for the evaluation task, but in the conditional motor task, a ventral and rostral premotor region (PMvr, area 6), the dorsolateral prefrontal cortex (PFdl, area 46), and the posterior parietal cortex (area 39/40) showed decreasing rCBF during learning, all in the right hemisphere. These data confirm the predicted rCBF changes in premotor and prefrontal areas during arbitrary mapping tasks and suggest that a broad frontoparietal network may show decreased synaptic activity as arbitrary rules become more familiar.
A burgeoning brain-imaging literature has focused to a large extent on higher brain functions restricted to our species, such as speech, semantic information processing, and other aspects of language (Buckner et al. 1995 Subjects
We studied seven normal volunteers, five males and two females, aged 22-50 yr (mean, 31 ± 11 yr). One female subject was eliminated from the analysis because she made a large number of errors and showed no improvement in performance over the scanning sessions. All subjects were right-handed as measured by the Edinburgh Inventory (Oldfield 1971 Experimental design
For each subject, positron emission tomography (PET) scans of rCBF were performed sequentially using H215O as the tracer. During scanning, the subject moved a joystick, constrained by a grooved template that restricted joystick movement to four directions from center. Visual cues were presented on a video screen 58 cm from the subject. The display was masked to a 9 × 9° square (visual angle). The joystick moved 6.5 cm to the endpoint of the groove, and the movement was recorded as completed and the response time (RT) measured at the midpoint of its trajectory.
Data acquisition
Subjects lay in a supine position in a dimly lit, sound-attenuated room. The subject's head was immobilized with an individually fitted thermoplastic face mask. A small plastic catheter was placed in the left cubital vein for injection of radioisotope. PET of the brain was performed using a GE Advance system (General Electric, Schenectady, NY). Data were acquired in three-dimensional mode and reconstructed into 35 contiguous transaxial planes separated by 4.25 mm (center-to-center), which covers the both the vertex and the entire cerebellum. In-plane and axial resolution are 5.2 and 4.6 mm full-width half-maximum, respectively. Emission scans were attenuation corrected with a transmission scan collected before each session during the exposure of a 68Ge/68Ga external rotating source. After a 10-mCi bolus injection of H2 15O, scanning was started when the brain radioactive count rate reached a threshold value, and continued for 60 s. Integrated radioactivity accumulated in the 60 s of scanning was used as an index of rCBF. Ten minutes elapsed between each injection. No arterial blood sampling was performed, and thus the images collected are those of tissue activity. Subjects began performing the task at the time of the injection, i.e., 15-20 s before the beginning of the scan and continued for a total of ~105 s.
Data analysis
Calculations and image matrix manipulations were performed in PROMATLAB (Mathworks, Sherborn, MA) on a SPARC 20 computer (Sun Microsystems, Mountain View, CA) with software for image analysis (SPM, MRC Cyclotron Unit, London, UK). Statistical parametric maps are spatially extended statistical processes that are used to characterize regionally specific effects in imaging data (Friston et al. 1991
Designation of anatomic structures
As described above, the procedure used for group analysis of the PET data was based on the resizing of the PET scans to a standard anatomic space (Talairach and Tournoux 1988 Behavior
There were 54 responses (in 102 s) for each behavioral block in the conditional motor task; 34 (in 108 s) in the evaluation task. The number of responses made per scan was not highly variable, but there was sometimes a one-response difference from scan-to-scan. As shown in Fig. 3, A and B, RT decreased over the four consecutive scans for both tasks and both conditional rules. For each task-rule combination, a separate repeated-measures analysis of variance (ANOVA) was performed with RT over scanning blocks as a within-subject factor (Greenhouse-Geisser corrected). RT was always significantly shorter in scan 4 than in scan 1 (conditional motor task, nonspatial rule: F = 5.24, P < 0.05; spatial rule: F = 31.03, P < 0.002; evaluation task, nonspatial rule: F = 125.44, P < 0.001; spatial rule:F = 7.99, P < 0.04). Three-way ANOVA showed a significant interaction between task and scan number (F = 5.736, P < 0.05), showing a greater learning effect in the evaluation task.
Neuroimaging
PRINCIPAL COMPONENT ANALYSIS.
Figure 4 shows the loadings for each of the 10 scans per session on the first (Fig. 4, A and B) and second (Fig. 4, C and D) principal components, divided by task. Note that the order of scanning blocks 2-5 (as a group) and 6-9 (also as a group) were counterbalanced among subjects. Accordingly, the placement of the data for the nonspatial rule before those for the spatial rule in Fig. 4 is arbitrary. Together, the first and second principal components (PC1 and PC2) accounted for ~67-70% of the variance in the data. PC1 distinguishes between all scanning blocks in which the subjects had to attend to the visual stimuli and use that information to select a conditional response (spatial or nonspatial rules) versus scans of the other, unconditional (fixed-response) rule. The trend for PC2 loading corresponds, inversely, with changes in RT (Figs. 3, A and B) over the four consecutive scanning blocks for each rule. Both PC1 and PC2 also reveal a smaller within-session trend, as indicated by the differential loadings of the first and last scanning block for the fixed-response rule. To distinguish overall within-session time trends from learning-related trends, PC2 was recalculated for the 10 scans in the order they occurred. This analysis confirmed that the loading inverted between blocks 5 and 6, i.e., at the transition between the two rules (Fig. 4, C and D), regardless of which rule was presented first.
CONDITIONAL RULES VERSUS FIXED-RESPONSE RULE.
Several brain regions showed greater rCBF for the two conditional rules than for the fixed-response rule (Fig. 5, Table 2). These regions correspond closely with the positively loaded voxels for PC1 (not illustrated) and included a number of cerebellar and visual areas. Few frontal areas showed significant contrasts, although a left ventrocaudal premotor area (PMvc, area 6) did so in both tasks. Right PMvc showed similar rCBF increases but only in the evaluation task.
NONSPATIAL RULE: EVALUATION TASK.
Several areas showed a decrease in rCBF as subjects improved performance on the nonspatial rule in the evaluation task (Fig. 6; Table 3), a trend that corresponds to PC2 (see above). These decreases paralleled the improvement in RT (Fig. 3B) and the decrease in error rate (Fig. 3D). The most dramatic of these decreases was seen in the ventral part of the left prefrontal cortex (Fig. 6A), extending onto the orbital surface at its most lateral extent (PFv, area 47/12). In the right hemisphere, a rostral aspect of the dorsal premotor cortex (PMdr, area 6) showed a less dramatic, but significant rCBF decrease, extending rostrally into area 8, with all or nearly all suprathreshold voxels rostral to the VAC line (Fig. 6C). A subsidiary contrast peak was found in the PFd (area 9), also on the right side (Fig. 6D). In the posterior parietal cortex (areas 39/40), a similar rCBF decrease was found (Fig. 6B). When using a monotonic linear model only (but not when contrasting the first and last scans of the 4-scan sequence), a broad area of the left, lateral cerebellar hemisphere showed a significant decrease in rCBF (Table 3; not illustrated in Fig. 6).
NONSPATIAL RULE: CONDITIONAL MOTOR TASK.
Significant contrasts for the nonspatial rule of the conditional motor task were much less extensive than for the evaluation task (Fig. 8A; Table 3) with no significant contrasts occurring in the frontal cortex. A posterior parietal region (area 39/40) showed a rCBF decrease as performance improved and overlapped to a large extent with that showing a similar change in the evaluation task (Fig. 6B).
SPATIAL RULE: EVALUATION TASK.
There were no significant rCBF contrasts in either parietal or frontal cortex for the evaluation tasks' spatial rule, although there were increases in the left putamen and insula (Table 4, not illustrated).
SPATIAL RULE: CONDITIONAL MOTOR TASK.
The right PFdl (area 46) and the caudally adjacent right PMvr (area 6) showed significant rCBF decreases as performance improved (Fig. 9, B and C). A posterior parietal region (area 39/40) did so, as well (Fig. 9A), also on the right side, and it overlapped with the parietal region described above for the nonspatial rule (Figs. 6B and 8A). Contiguous voxels above threshold extended ventrocaudally into the occipital cortex.
Comparisons among tasks and rules
Several motor regions, including PMvc (area 6) and anterior cingulate (area 24) areas, as well as the cerebellum, showed greater rCBF for conditional-rule scans (spatial and nonspatial rules, combined) than for fixed-response scans (Fig. 5; Table 2). It is possible that these regions are especially important in the guidance of behavior by concrete visuomotor rules. Alternatively, these contrasts might be attributed to the selection of an action on each trial (Deiber et al. 1991 Learning effects
Early blocks for a given conditional rule were associated with significantly different rCBF rates than later ones. Before concluding that these differences are learning effects, it is important to consider overall within-session trends that may undermine that interpretation. Principal component analysis shows that such within-session effects did, indeed, occur (Fig. 4). However, the second principal component (PC2) reveals the same inversion of loading at the rule transition (from scanning block 5 to 6) regardless of which conditional rule was presented first. This finding indicates that PC2 and the related SPM contrasts (Figs. 6-9; Tables 3 and 4) reflect some aspect of the time within a given rule, presumably learning. Nevertheless, an influence of general rCBF trends over the course of the 10-scan sessions cannot be ruled out.
Prefrontal cortex
The prefrontal decreases in rCBF described here resemble those observed for motor sequence learning (Jenkins et al. 1994 PRESENT FINDINGS IN THE CONTEXT OF THEORIES OF THE PREFRONTAL LOBE.
Among the many, diverse theories of frontal lobe function in primates, two general classes can be discerned. One class, which focuses on the temporary storage of information, has been termed the working-memory theory of the frontal lobe (Goldman-Rakic 1988 DIFFERENCES AMONG PREFRONTAL AREAS.
The most dramatic decreases in rCBF observed in the present study were those in the left PFv, a region Petrides and Pandya (1994) Premotor cortex
In both right and left hemispheres, a ventrocaudal aspect of the lateral premotor cortex (PMvc) showed an rCBF increase during conditional motor learning in the evaluation task. Except for small regions in the insula and putamen (see below), it was the only region in any rule or task to do so. This result differs from that of a previous study of conditional motor learning by Deiber et al. (1991) Parietal cortex
A large, right posterior parietal region (area 39/40) also showed decreasing rCBF during most tasks and rules. A similarly located, predominant right parietal activation was described by Jenkins et al. (1994) Cerebellum
Left lateral cerebellar hemisphere showed a decrease during learning (in the nonspatial rule, evaluation task). However, this decrease only reached statistical significance when all four scans were considered and contrasted with a monotonic linear contrast model. The role of the cerebellum, especially the lateral cerebellar hemisphere, in sensory-sensory conditional learning has been supported in the clinical literature (Bracke-Tolkmitt et al. 1989 Basal ganglia
The left putamen showed time-dependent increases in rCBF for both the spatial and nonspatial rules, in the conditional motor and evaluation tasks, respectively. Besides the premotor cortex (PMvc, area 6), this was the principal brain region to show such increases during learning. In monkeys, neurons in the striatum show learning-related activity changes (Tremblay et al. 1994 Comparison with neuronal activity in monkeys
As monkeys learn novel, nonspatial conditional motor rules, some neurons in the supplementary eye field decrease discharge rates as learning progresses. These cells often become inactive when a mapping rule becomes highly familiar (Chen, L., and Wise 1995a,b). Some of these neurons, termed learning-selective by Chen and Wise, show a monotonic decline in discharge rate, but most have an initial increase in activity followed by a steep decline as performance improves. The decreases in rCBF observed in PMdr (area 6), PFv (area 47/12), PFd (area 9), posterior parietal cortex (areas 39/40), and elsewhere during learning appear more similar to the monotonic pattern. It is important to emphasize, however, that we cannot rule out a rCBF increase during the pre-PET training session. If the learning and processing of explicit knowledge about the task, which occurred before the beginning of the scans, were reflected in a blood flow increase, then the overall pattern would be an initial increase, followed by a decline, as for most of the single-unit data.
Conclusions
Three main findings emerge from the present study: as subjects become increasingly familiar with conditional rules, a broad frontal-parietal network shows decreased rates of blood flow; the posterior parietal part of this network remains relatively constant from rule-to-rule and task-to-task, but the prefrontal components vary; and, rostral and dorsal premotor areas show learning-related decreases in rCBF, much like prefrontal and parietal cortex, whereas more caudal and ventral premotor areas, along with the putamen and a cingulate motor area, show rCBF increases. The learning-related changes in cortical blood flow resemble the evolution of neuronal activity as monkeys learn conditional motor mappings.
INTRODUCTION
Abstract
Introduction
Methods
Results
Discussion
References
; Demonet et al. 1994; Kapur et al. 1994a,b; Paulesu et al. 1993; Petersen et al. 1988, 1989; Tulving et al. 1994a,b; Wise et al. 1991). Fewer efforts have been directed toward those advanced behavioral capabilities that we share with other primates. Conditional motor learning typifies one such faculty; one allowing the flexibility to map any stimulus onto any motor response. This form of motor learning depends on the integrity of premotor and prefrontal areas in both humans (Halsband and Freund 1990; Petrides 1987, 1990) and monkeys (Gaffan and Parker 1997; Halsband and Passingham 1982, 1985; Murray and Wise 1997; Passingham 1985; Petrides 1982, 1985). Moreover, learning-related evolution in single-cell activity has been observed in the monkey premotor cortex during conditional motor learning (Chen and Wise 1995a,b, 1996; Mitz et al. 1991).
). In view of the finding by Chen and Wise (1995a) that many premotor cortical cells are only active during conditional motor learning (as opposed to stable performance), we reexamined the prediction by studying rCBF as subjects improved their performance according to such rules. Because both spatial and symbolic forms of information are important in conditional motor learning, we used two different conditional rules, one relying on visuospatial information and the other on nonspatial visual information. Further, to accentuate the explicit aspects of conditional information processing, we required subjects to recognize and discriminate potential movement directions specified by a cue, as well as to generate directional movements as instructed by the same cues. These data have been reported previously in abstract form (Deiber et al. 1996).
METHODS
Abstract
Introduction
Methods
Results
Discussion
References
). The protocol was approved by the Institutional Review Board, and all subjects gave written informed consent.
View larger version (14K):
[in a new window]
FIG. 1.
Conditional motor task, nonspatial rule. Largest squares represent 9 × 9° masked video screen, with 0.5° center fixation spot ( 
), not to scale. Small squares represent the 2 × 2° patterned stimuli. For simplicity, each pattern is illustrated as appearing in upper left corner, but could appear near any of 4 corners. On each trial, a pattern is displayed for 200 ms. Then pattern disappeared (leaving the central spot). Subject had a limit of 1.7 s from stimulus offset to move joystick (bottom, 
) in 1 of 4 directions, corresponding to diagonals of square display. Direction of joystick movement instructed by each pattern is represented by . For conditional motor task's spatial rule, same stimuli were presented but were associated with joystick movement responses as listed in Table 1.
View this table:
TABLE 1.
Conditional associations learned by subjects
View larger version (20K):
[in a new window]
FIG. 2.
Evaluation task, nonspatial rule. On each trial, the following sequence of screens was presented: pattern plus fixation spot (250 ms), fixation spot only (500 ms, not illustrated), an arrow, which pointed to 1 of 4 corners (250 ms), and a YES/NO screen, in which words no and yes were displayed in lower left and upper right corner, respectively (1,700 ms). Subject had a response limit of 1,700 ms to move joystick to upper right to report that arrow matched the direction associated with pattern or to lower left to report that it did not. For simplicity of illustration, each pattern is displayed in upper left corner and only a selection of the possible combinations are shown. Patterned stimuli could appear near any of 4 corners and were followed by a "correct" or "incorrect" arrow (representing a potential response) in a balanced sequence. For evaluation task's spatial rule, same sequence of screens was presented, but subjects evaluated whether arrow pointed in direction associated with location of stimulus as listed in Table 1.
, 1994; Worsley et al. 1992). The scans from each session of each subject were realigned using the first scan as a reference. The six parameters of this rigid body transformation were estimated using a least squares approach (Friston et al. 1995a). After realignment, all images were transformed into the standard space of a brain atlas (Talairach and Tournoux 1988). The spatial normalization involves linear and nonlinear three-dimensional transformations to match each scan to a reference image that already conforms to the standard brain space (Friston et al. 1995a). Images then were smoothed with an isotropic Gaussian kernel (15 mm full-width half-maximum). The effect of global differences in rCBF between scans was removed by scaling activity in each pixel proportional to the global activity so as to adjust the mean global activity of each scan to 50 ml·100 g
1·min1.
), which removes a systematic difference among subjects as a confounding effect. To test hypotheses about the specific regional effects of the condition, the estimates were compared using linear contrasts. The resulting set of voxel values for each contrast constitutes a statistical parametric map of the t statistic (SPM{t}). The SPM{t} were transformed to the unit normal distribution (SPM{Z}) and threshold was set at 2.33. The resulting foci then were characterized in terms of peak activation height and spatial extent. The significance of rCBF changes in each region was estimated using the probability that the peak activation observed could have occurred by chance and/or that the observed number of contiguous voxels could have occurred by chance over the entire volume analyzed (Friston et al. 1994). A corrected P value of 0.05 was used as a final threshold for significance.
). The data were decomposed into two sets of orthogonal vectors using singular value decomposition (SVD) as follows
M = U*S*VT; where M is the original data matrix with 10 rows (1 for each block) and one column for each voxel, U and V are unitary orthogonal matrices denoting pattern across conditions and in space, respectively, T denotes transposition, and S is a diagonal matrix of decreasing singular value. Each column of V and U can be interpreted as a spatially distributed pattern and the corresponding profile over different conditions, respectively, associated with each principal component.
![[<IT>U</IT><IT>S</IT><IT>V</IT>] = SVD[<IT>M</IT>] so that:](/content/vol78/issue2/fulltext/977/img001.gif)
View larger version (46K):
[in a new window]
FIG. 5.
Contrast analysis for rule-guided (spatial and nonspatial) versus fixed-response scanning blocks. Left: conditional motor task; right: evaluation task. Voxels having Z values exceeding significance threshold of 2.33 with Bonferroni correction for multiple comparisons (P < 0.05) are displayed on a gray scale, with lower Z scores represented in light gray and higher ones in dark gray. SPMs are displayed in anatomic space of (Talairach and Tournoux 1988 ) as a maximum intensity projection viewed from right side (sagittal view), back (coronal view), and top (transverse view) of brain.
). This procedure allowed us to relate coordinates to cytoarchitectonic labels as depicted in that atlas. In recognition of the limitations of this technique, we also have described the localized contrasts in rCBF in terms of general regions of the frontal lobe, taking into account the general organizational schemes of Passingham (1993), Petrides and Pandya (1994), and others (He et al. 1995; Wise et al. 1996b). In so doing, we have taken into account both the primary and subsidiary contrast peaks as detected through SPM and illustrated the contiguous voxels that exceed a Z statistic of 2.33 for regions showing significant rCBF contrasts. In the nomenclature used in this report, area 46 is equated with the dorsolateral prefrontal cortex (PFdl), area 9 with the dorsal prefrontal cortex (PFd), and area 47 (area 47/12 of Petrides and Pandya) with the ventro-orbital prefrontal cortex (PFv). Subdivisions of premotor cortex (PM) are recognized on the basis of their relative locations within area 6 (Wise et al. 1996b). PM be divided into rostral (PMr) and caudal zones (PMc) based on the vertical anterior commissure (VAC) frontal plane. We also made distinctions between dorsal (PMd) and ventral (PMv) zones, based on the range of horizontal coordinates (z-axis) reported for the frontal eye field (Paus 1996): any z < 40 was taken as PMv, whereas any z >50 was taken as PMd. Coordinates between those levels were not designated as either PMd or PMv.
RESULTS
Abstract
Introduction
Methods
Results
Discussion
References
View larger version (27K):
[in a new window]
FIG. 3.
Movement performance over 6 subjects. A and B: mean response time and standard deviation for each block of behavior. Nonspatial and spatial rules labeled with bars over data points. C and D: total number of errors in each scan summed over all subjects. Response time shortens and number of errors drops from scan 1 to scan 4 in each rule. Note that because order of scanning blocks 2-5 and 6-9 (both as a group) were counterbalanced among subjects, placement of data for nonspatial rule before those for spatial rule is arbitrary. Fixed-response rules (F) were always first and last (10th) scans of session. Note also that total number of trials per behavioral block differed for each task (i.e., 54 and 34 trials for conditional motor and evaluation tasks, respectively).
View larger version (41K):
[in a new window]
FIG. 4.
Principal component (PC) analysis. A and B: first principal component (PC1). C and D: second principal component (PC2). 
, fixed-response rule; 
, nonspatial rule; 
, spatial rule. F, fixed-response rule. Note that, as in Fig. 3, order of spatial versus nonspatial rules in arbitrary for this figure. Percentage of total variance accounted for by each PC is indicated.
View this table:
TABLE 2.
Nonspatial and spatial rules versus fixed-response rule
View larger version (28K):
[in a new window]
FIG. 6.
Evaluation task, nonspatial rule: voxels with rCBF decreases from 1st to 4th scan. VAC, vertical line passing through the anterior commissure; VPC, vertical line passing through the posterior commissure. A-D: cerebral areas with significant rCBF changes are labeled, and for each of them, mean adjusted rCBF and standard deviations are plotted at voxel of maximum Z score in each of 4 scans.
View this table:
TABLE 3.
Learning dependent decreases in rCBF
View larger version (33K):
[in a new window]
FIG. 7.
Evaluation task, nonspatial rule: voxels with rCBF increases from 1st to 4th scan. Format as in Fig. 6.
View this table:
TABLE 4.
Learning dependent increases in rCBF
View larger version (27K):
[in a new window]
FIG. 8.
Conditional motor task, nonspatial rule: voxels with rCBF decreases from first to fourth scan. Only 1 cerebral area has significant rCBF changes, right parietal cortex (A). Format as in Fig. 6.
View larger version (38K):
[in a new window]
FIG. 9.
Conditional motor task, spatial rule: voxels with rCBF decreases from first to fourth scan. Format as in Fig. 6.
DISCUSSION
Abstract
Introduction
Methods
Results
Discussion
References
) rather than guidance by conditional rules per se. Further, because the fixed-response rule did not require subjects to attend to either the visual patterns or their locations, rCBF differences might reflect the increased attentional demands in the conditional-rule scans. Indeed, visual areas in both hemispheres show rCBF enhancements during the conditional-rule scans (Fig. 5, Table 2). Such increases may result from the orientation of attention to those stimuli, the importance of frontal-visual cortex interactions in conditional motor learning (Gaffan and Parker 1997), or both.
) and increasing familiarity with a noun-verb generation task (Raichle et al. 1994). Similar results have been obtained during skill learning tasks, including mirror drawing (Imamura et al. 1996) and adaptation to externally imposed force fields (Shadmehr et al. 1996). During certain kinds of sequence learning, by contrast, rCBF tends to increase during learning, especially as explicit knowledge is gained (Doyon et al. 1996; Grafton et al. 1992, 1994, 1995; Hazeltine et al. 1996; Hikosaka et al. 1996; Honda et al. 1996). One interpretation of a decrease in rCBF, especially as applied to prefrontal cortex, is that it reflects a disengagement of that area as a behavior becomes routine and the task demands become less ambiguous.
, 1995; Owen et al. 1996a,b; Pascual-Leone et al. 1996). That theory emphasizes the role of the frontal cortex in short-term maintenance of information in a variety of sensory domains, including the encoding and retrieval of information about events (Kapur et al. 1994a,b; Tulving et al. 1994a) and language (Fiez et al. 1995; Petersen et al. 1988, 1990; Wise et al. 1991). A large number of brain-imaging studies have been directed toward verifying the role of frontal cortex in sensory working memory (Buckner and Peterson 1996; D'Esposito et al. 1995; Hugdahl et al. 1995; Kapur et al. 1995; McCarthy et al. 1996; Owen et al. 1996a,b; Petrides 1996).
, 1995), which is used to organize behavior thematically in accordance with long-term goals. This idea corresponds to some extent with the view, mainly derived from studies of nonhuman primates, that the frontal lobe functions in learning what response to select based on context (Passingham 1993). The present view emphasizes this central executive aspect of frontal lobe function (see also Frith et al. 1991; Jenkins et al. 1994; Raichle et al. 1994). From our perspective, the present data are most consistent with the hypothesis that localized frontal areas, as well as the parietal networks associated with them, show potentiated synaptic activity when routine rules need to be rejected and new ones adopted (Wise et al. 1996b). Thus the decrease in rCBF observed in prefrontal (and rostral premotor) areas may reflect a relaxation of the cortical network subserving a behavior as that behavior becomes routine.
have termed area 47/12. They argued that it is homologous to part of area 12 in macaque monkeys. The left PFv has been previously reported to show activation during conditional tasks, although the results have not always been discussed in those terms. These tasks include extrinsically cued word finding (Frith et al. 1991), a newly learned (reversal) of a sensorially cued eye movement (Paus et al. 1993), word selection contrary to previously learned associations (Paus et al. 1993), generation of the uses of terms presented visually, i.e., semantic association (Petersen et al. 1988), and both visual and auditory association tasks involving semantic processing (Petersen et al. 1989). Passingham (1993) has postulated that the ventral prefrontal areas, area 47/12 among them, are the principal areas engaged in learning behavioral responses based on context. In accord with this idea, ventro-orbital PF cortex (Murray and Wise 1997) and its intrahemispheric interactions with visual areas (Gaffan and Parker 1997) recently have been shown to be essential for conditional (but not unconditional) learning in macaque monkeys. Alternatively, it has long been held, on the grounds that it receives direct projections from inferotemporal cortex, that PFv is specialized for processing or selecting nonspatial rather than spatial visual information (Jones and Powell 1970; Webster et al. 1994). Brain imaging (Courtney et al. 1996; Haxby et al. 1991; Ungerleider and Haxby 1994) and neurophysiological (Goldman-Rakic 1995; Wilson et al. 1993) data have been presented as supporting this view, although those interpretations of the neuroimaging data have been challenged (Rushworth et al. 1997). The present data are consistent with both hypotheses.
; McCarthy et al. 1996). Accordingly, the finding of right hemisphere decreases in rCBF during spatial rule learning, seen in the conditional motor task, is consistent with prevailing views on PF organization. However, we have no compelling explanation for the failure to observe comparable changes for spatial rule learning in the evaluation task.
). However, we have no compelling explanation for the failure to observe rCBF changes in PFd (or, for that matter, PFv) during nonspatial rule learning in the conditional motor task. One possibility, for both PFd and PFv, is that the lack of a need for explicit, evaluative information processing in the conditional motor task resulted in less overall engagement of PF in the task, regardless of the subjects' previous exposure to the arbitrary mappings.
. They contrasted rCBF during performance of nonspatially instructed movements with that during the performance of a fixed response and found no premotor or prefrontal areas with significant rCBF increases. Similarly, Paus et al. (1993) did not report significant activation of the lateral premotor areas during a variety of conditional motor tasks. The present result confirms the predictions made on the basis of neuropsychological and neurophysiological studies in monkeys (Halsband and Passingham 1982, 1985; Passingham 1985; Petrides 1982, 1985) and humans (Halsband and Freund 1990; Petrides 1987, 1990) pointing to the lateral premotor cortex as an important site for conditional visuomotor learning. This region of premotor cortex with increasing rCBF during learning appears to be near that described by Grafton et al. (1996) as activated in imagined grasping. They identified this zone with area 44 (coordinates 
43, 0, 30), which appears rostral to what we term PMvc (48, 8, 32, and 42, 10, 32). Note, however, that the activated voxels in our data abut the VAC line (Fig. 7). Because the subjects may have been imagining the instructed movement during the 500-ms delay period, we cannot rule out a role for motor imagery in the present results. However, such imagery would have to change over the consecutive scans within each rule to affect our principal results, and this appears unlikely.
found a similarly located site activated in a conditional oculomotor task, but interpreted it as the frontal eye field. Because our subjects were instructed to fixate the central spot during the scan and reported that they were able to do so, we do not believe that our results reflect overt eye movements.
; Ghosh and Gattera 1995; He et al. 1993; Lu et al. 1994; Wise et al. 1996a, 1997). A similar distinction is reflected in PET data for both the medial (Colebatch et al. 1991; Deiber et al. 1991; Matelli et al. 1993; Picard and Strick 1996; Stephan et al. 1995) and lateral (Deiber et al. 1991) premotor areas.
, also showed an increase in rCBF during learning. Although data from nonhuman primates indicates that cingulate areas are not necessary for performance of conditional motor tasks (Chen, Y., et al. 1995), the cingulate motor areas may be important in acquisition of such behavior.
when comparing newly learned with previously learned motor sequences. These authors attribute this finding to spatial attention, as suggested by other PET studies (Corbetta et al. 1993; Petersen et al. 1994). It is possible that subjects attend to stimuli less intensely as the mappings become more familiar (Coull et al. 1996; Paus et al. 1996). However, there are other possibilities. In all of the conditional rules and tasks used here, a conversion of visual information into the spatial and/or motor domain was required. The rCBF changes in posterior parietal cortex therefore may reflect the mapping of visual information into a different coordinate scheme (Andersen 1995). As these coordinate transformations become routine and relatively automatic, the importance of the posterior parietal cortex may diminish.
; Canavan et al. 1994), including for tasks involving nonspatial visual stimuli as instructions.
) that resemble those in premotor areas (see below), and interrupting pallido-thalamocortical connections at the thalamic level causes dramatic deficits in conditional motor performance (Canavan et al. 1989).
). In the supplementary eye field, neurons with learning-dependent and learning-selective activity are intermixed. However, on the present data and the assumption that rCBF increases reflect enhanced excitation, we suggest that learning-selective activity may predominate in prefrontal areas, whereas learning-dependent neurons may be most important in the caudal premotor cortex.
; Wise 1996), but here PMvc shows rCBF increases during learning. Further, the measure of conditional motor learning in the monkey experiments cited above was the proportion of correct responses rather than improving RT, as in the present report. It is possible that different or completely distinct neuronal mechanisms underlie these two learning measures, but we find no compelling reason to assume so.
| |
ACKNOWLEDGEMENTS |
|---|
The authors thank Dr. Ilsun M. White for comments on an earlier version of this manuscript.
| |
FOOTNOTES |
|---|
Address for reprint requests: S. P. Wise, Laboratory of Systems Neuroscience, National Institute of Mental Health, PO Box 608, Poolesville, MD 20837. E-mail: spw{at}codon.nih.gov
Received 30 December 1996; accepted in final form 1 May 1997.
| |
REFERENCES |
|---|
|
Abstract
Introduction
Methods
Results
Discussion
References
|
|---|
M, K.,
OKURA, M.,
LÅNGSTROM, B.
Regional activation of human cerebral cortex upon an adaptation in mirror drawing.
Neurosci. Lett.
209: 185-188, 1996.[Medline]
basal ganglia system in primates.
Crit Rev Neurobiol
10: 317-356, 1996b.[Medline]This article has been cited by other articles:
|
|
 |
|
  H. Ruge and U. Wolfensteller Rapid Formation of Pragmatic Rule Representations in the Human Brain during Instruction-Based Learning Cereb Cortex, November 4, 2009; (2009) bhp228v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Nakayama, T. Yamagata, J. Tanji, and E. Hoshi Transformation of a Virtual Action Plan into a Motor Plan in the Premotor Cortex J. Neurosci., October 8, 2008; 28(41): 10287 - 10297. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Wu, P. Chan, and M. Hallett Modifications of the interactions in the motor networks when a movement becomes automatic J. Physiol., September 1, 2008; 586(17): 4295 - 4304. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Brovelli, N. Laksiri, B. Nazarian, M. Meunier, and D. Boussaoud Understanding the Neural Computations of Arbitrary Visuomotor Learning through fMRI and Associative Learning Theory Cereb Cortex, July 1, 2008; 18(7): 1485 - 1495. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. G. Richardson, G. Lassi-Tucci, C. Padoa-Schioppa, and E. Bizzi Neuronal Activity in the Cingulate Motor Areas During Adaptation to a New Dynamic Environment J Neurophysiol, March 1, 2008; 99(3): 1253 - 1266. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. J. Suminski, S. M. Rao, K. M. Mosier, and R. A. Scheidt Neural and Electromyographic Correlates of Wrist Posture Control J Neurophysiol, February 1, 2007; 97(2): 1527 - 1545. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. B. Postuma and A. Dagher Basal Ganglia Functional Connectivity Based on a Meta-Analysis of 126 Positron Emission Tomography and Functional Magnetic Resonance Imaging Publications Cereb Cortex, October 1, 2006; 16(10): 1508 - 1521. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Aramaki, M. Honda, T. Okada, and N. Sadato Neural Correlates of the Spontaneous Phase Transition during Bimanual Coordination Cereb Cortex, September 1, 2006; 16(9): 1338 - 1348. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Amiez, P. Kostopoulos, A.-S. Champod, and M. Petrides Local morphology predicts functional organization of the dorsal premotor region in the human brain. J. Neurosci., March 8, 2006; 26(10): 2724 - 2731. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Grol, F. P. de Lange, F. A. J. Verstraten, R. E. Passingham, and I. Toni Cerebral Changes during Performance of Overlearned Arbitrary Visuomotor Associations J. Neurosci., January 4, 2006; 26(1): 117 - 125. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Wu and M. Hallett A functional MRI study of automatic movements in patients with Parkinson's disease Brain, October 1, 2005; 128(10): 2250 - 2259. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Lehericy, H. Benali, P.-F. Van de Moortele, M. Pelegrini-Issac, T. Waechter, K. Ugurbil, and J. Doyon Distinct basal ganglia territories are engaged in early and advanced motor sequence learning PNAS, August 30, 2005; 102(35): 12566 - 12571. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Bermudez and R. J. Zatorre Conditional Associative Memory for Musical Stimuli in Nonmusicians: Implications for Absolute Pitch J. Neurosci., August 24, 2005; 25(34): 7718 - 7723. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. A. Boettiger and M. D'Esposito Frontal Networks for Learning and Executing Arbitrary Stimulus-Response Associations J. Neurosci., March 9, 2005; 25(10): 2723 - 2732. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Tanaka, M. Honda, and N. Sadato Modality-Specific Cognitive Function of Medial and Lateral Human Brodmann Area 6 J. Neurosci., January 12, 2005; 25(2): 496 - 501. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Harada, D. N. Saito, K.-I. Kashikura, T. Sato, Y. Yonekura, M. Honda, and N. Sadato Asymmetrical Neural Substrates of Tactile Discrimination in Humans: A Functional Magnetic Resonance Imaging Study J. Neurosci., August 25, 2004; 24(34): 7524 - 7530. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. A. Fridman, T. Hanakawa, M. Chung, F. Hummel, R. C. Leiguarda, and L. G. Cohen Reorganization of the human ipsilesional premotor cortex after stroke Brain, April 1, 2004; 127(4): 747 - 758. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Eliassen, T. Souza, and J. N. Sanes Experience-Dependent Activation Patterns in Human Brain during Visual-Motor Associative Learning J. Neurosci., November 19, 2003; 23(33): 10540 - 10547. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. A. Bunge, I. Kahn, J. D. Wallis, E. K. Miller, and A. D. Wagner Neural Circuits Subserving the Retrieval and Maintenance of Abstract Rules J Neurophysiol, November 1, 2003; 90(5): 3419 - 3428. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R.-A. Muller, N. Kleinhans, N. Kemmotsu, K. Pierce, and E. Courchesne Abnormal Variability and Distribution of Functional Maps in Autism: An fMRI Study of Visuomotor Learning Am J Psychiatry, October 1, 2003; 160(10): 1847 - 1862. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. A. Chouinard, Y. D. Van Der Werf, G. Leonard, and T. Paus Modulating Neural Networks With Transcranial Magnetic Stimulation Applied Over the Dorsal Premotor and Primary Motor Cortices J Neurophysiol, August 1, 2003; 90(2): 1071 - 1083. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Hanakawa, M. Honda, N. Sawamoto, T. Okada, Y. Yonekura, H. Fukuyama, and H. Shibasaki The Role of Rostral Brodmann Area 6 in Mental-operation Tasks: an Integrative Neuroimaging Approach Cereb Cortex, November 1, 2002; 12(11): 1157 - 1170. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. R. Simon, M. Meunier, L. Piettre, A. M. Berardi, C. M. Segebarth, and D. Boussaoud Spatial Attention and Memory Versus Motor Preparation: Premotor Cortex Involvement as Revealed by fMRI J Neurophysiol, October 1, 2002; 88(4): 2047 - 2057. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Toni, J. Rowe, K. E. Stephan, and R. E. Passingham Changes of Cortico-striatal Effective Connectivity during Visuomotor Learning Cereb Cortex, October 1, 2002; 12(10): 1040 - 1047. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Brass and D. Y. von Cramon The Role of the Frontal Cortex in Task Preparation Cereb Cortex, September 1, 2002; 12(9): 908 - 914. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. F. S. Rushworth, T. Paus, and P. K. Sipila Attention Systems and the Organization of the Human Parietal Cortex J. Neurosci., July 15, 2001; 21(14): 5262 - 5271. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Kurata, T. Tsuji, S. Naraki, M. Seino, and Y. Abe Activation of the Dorsal Premotor Cortex and Pre-Supplementary Motor Area of Humans During an Auditory Conditional Motor Task J Neurophysiol, September 1, 2000; 84(3): 1667 - 1672. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Toni, N. D. Schluter, O. Josephs, K. Friston, and R. E. Passingham Signal-, Set- and Movement-related Activity in the Human Brain: An Event-related fMRI Study Cereb Cortex, January 1, 1999; 9(1): 35 - 49. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Shima and J. Tanji Both Supplementary and Presupplementary Motor Areas Are Crucial for the Temporal Organization of Multiple Movements J Neurophysiol, December 1, 1998; 80(6): 3247 - 3260. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. T. Grafton, A. H. Fagg, and M. A. Arbib Dorsal Premotor Cortex and Conditional Movement Selection: A PET Functional Mapping Study J Neurophysiol, February 1, 1998; 79(2): 1092 - 1097. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
