NeuroImage 15, 373–385 (2002) doi:10.1006/nimg.2001.0976, available online at http://www.idealibrary.com on
The Preparation and Execution of Self-Initiated and ExternallyTriggered Movement: A Study of Event-Related fMRI R. Cunnington,* ,† C. Windischberger,‡ L. Deecke,† and E. Moser‡ ,§ *Department of Psychology, Monash University, Clayton, Australia; and ‡Institute for Medical Physics, §Department of Radiodiagnostics, and †Department of Clinical Neurology, University of Vienna, Vienna, Austria Received March 27, 2001
Studies of functional brain imaging in humans and single cell recordings in monkeys have generally shown preferential involvement of the medially located supplementary motor area (SMA) in self-initiated movement and the lateral premotor cortex in externally cued movement. Studies of event-related cortical potentials recorded during movement preparation, however, generally show increased cortical activity prior to selfinitiated movements but little activity at early stages prior to movements that are externally cued at unpredictable times. In this study, the spatial location and relative timing of activation for self-initiated and externally triggered movements were examined using rapid event-related functional MRI. Twelve healthy righthanded subjects were imaged while performing a brief finger sequence movement (three rapid alternating button presses: index–middle–index finger) made either in response to an unpredictably timed auditory cue (between 8 to 24 s after the previous movement) or at selfpaced irregular intervals. Both movement conditions involved similar strong activation of medial motor areas including the pre-SMA, SMA proper, and rostral cingulate cortex, as well as activation within contralateral primary motor, superior parietal, and insula cortex. Activation within the basal ganglia was found for self-initiated movements only, while externally triggered movements involved additional bilateral activation of primary auditory cortex. Although the level of SMA and cingulate cortex activation did not differ significantly between movement conditions, the timing of the hemodynamic response within the pre-SMA was significantly earlier for self-initiated compared with externally triggered movements. This clearly reflects involvement of the pre-SMA in early processes associated with the preparation for voluntary movement. © 2002 Elsevier Science
INTRODUCTION Several functionally distinct motor areas have been identified within the frontal cortex immediately anterior to the primary motor area. These motor areas are differentially involved in movements made under dif-
ferent conditions. Electrophysiological studies in monkeys have suggested that the medially located supplementary motor area (SMA) is more involved in selfinitiated movements whereas the lateral premotor area is more involved in externally triggered movement (Romo and Schultz, 1987; Thaler et al., 1988; Mushiake et al., 1991). Picard and Strick (1996) further distinguish between two subregions within the supplementary motor area: the anterior or pre-SMA, which projects mainly to primary motor cortex, appears to be more involved in higher-order aspects of movement planning and preparation, whereas the caudal SMA or SMA proper contains a higher proportion of direct spinal projections (Macpherson et al., 1982) and shows a higher degree of movement execution-related activity. Similar evidence in humans is not so clear. Electrophysiological studies have mainly examined changes in cortical activity which precede voluntary movement and are thought to reflect processes associated with the preparation for movement. Such motor preparatory activity may be inferred from the slow negative scalp potential which precedes voluntary movement, termed the Bereitschaftspotential or readiness potential (Deecke et al., 1969). This readiness potential is significantly greater preceding self-paced movement compared with externally triggered movement (Papa et al., 1991; Jahanshahi et al., 1995), particularly when externally triggered movements are cued at unpredictable times (Cunnington et al., 1995). Motor preparatory activity reflected in the readiness potential is thought to arise predominantly from the SMA (Deecke and Kornhuber, 1978), lending support to evidence from animal studies that the SMA plays a particular role in self-paced movement. The cortical source of scalp recorded electrical potentials, however, is difficult to accurately localize and the extent to which the SMA contributes to the readiness potential has been questioned (Neshige et al., 1988; Boetzel et al., 1993). More recent studies using high-density EEG recordings and current source density analysis techniques have provided strong evidence that early pre-
373 ©
1053-8119/02 $35.00 2002 Elsevier Science All rights reserved.
374
CUNNINGTON ET AL.
movement activity originates within midline cortical areas including the SMA (Kno¨sche et al., 1996; Ball et al., 1999; Cui et al., 1999). Such studies, however, have focused on self-initiated voluntary movements only and have not examined differences in the localization or timing of cortical activity for externally triggered movements. Studies of functional brain imaging, using PET and functional MRI, have generally shown greater activation of the SMA for self-initiated compared with externally triggered movement (Rao et al., 1993; Wessel et al., 1997; Deiber et al., 1999; Jenkins et al., 2000), although there have been contradictions (Remy et al., 1994). Functional imaging studies have also shown strong bilateral activation of lateral premotor areas for externally triggered movements (Van Oostende et al., 1997; Wessel et al., 1997; Catalan et al., 1998; Haslinger et al., 2001). Studies directly comparing externally triggered and self-initiated movements, however, have generally failed to show significant changes in the level of lateral premotor cortex activation contingent on external cuing (Wessel et al., 1997; Weeks et al., 2001) as might have been expected from electrophysiological studies of nonhuman primates. Several functional imaging studies have examined the relative timing of activation of SMA, lateral premotor, and primary motor areas over periods of movement preparation using event-related functional MRI. In a delayed cued movement task Richter et al. (1997a) showed that activity within both SMA and lateral premotor areas increases during the movement preparation period (the delay between a warning cue and the GO cue), while the primary motor cortex showed only weak activation during the preparation period and the greatest activation during movement execution. In a study of purely self-paced movement Wildgruber et al. (1997) similarly showed that the peak activation within SMA precedes that within primary motor cortex and therefore most likely reflects activation of the SMA during movement preparation preceding movement execution. Studies directly comparing self-initiated and externally triggered movements, however, have only examined levels of activation throughout prolonged periods of movement performance and have therefore lacked the necessary temporal resolution to examine the relative timing of activity between cortical areas. It seems likely that externally triggered movements, in which rapid movements are made immediately in response to an unpredictable external cue, would involve less early-stage movement preparation and activation of motor areas only immediately prior to movement execution. Such differences in the timing of involvement of different motor cortical areas have not been examined. In this study we use a rapid event-related fMRI acquisition to explore involvement of different cortical regions over different time intervals prior to and dur-
ing voluntary movement. With such fast fMRI sequences at high field strength (3 T), it is possible to acquire brain images at a sufficient rate to directly examine changes in blood oxygen level within different brain regions for single actions or events (Cunnington et al., 1999; Windischberger et al., 1999). Although the change in local blood deoxyhemoglobin density measured by fMRI is slow and delayed by 3–5 s following increases in neural activity, the hemodynamic response delay is relatively consistent between cortical areas (Richter et al., 1997b) and can resolve differences in the timing of activation of different areas (Friston et al., 1998; Menon et al., 1998). This method therefore allows us to examine, with high spatial resolution, the relative involvement of motor cortical areas in preparatory and execution-related stages of motor control processing and to compare the level and relative timing of activity patterns for self-paced and externally triggered movement. MATERIAL AND METHODS Image Acquisition Imaging was performed on a 3-T Medspec S300 widebore scanner equipped with a whole-body gradient system and birdcage head coil (Bruker Medical, Ettlingen, Germany), using a phase-corrected single-shot gradient-echo EPI sequence. Fifteen horizontal slices of 6 mm thickness and 64 2 matrix (3.0 ⫻ 3.0-mm resolution) were acquired covering most of the brain with a slice acquisition time of 75 ms and volume repetition time of 1125 ms (TE ⫽ 40 ms; RBW ⫽ 100 kHz). The slice acquisition time was always constant, including between the last slice of a volume and the first slice of the next volume. In this way the frequency of scanner noise was a constant 13.3 Hz, with no variation between volume acquisitions. Two trials of 280 volumes, plus 8 initial saturation volumes, were acquired for each subject. Head motion artifacts were minimized by using an individually fitted plaster helmet designed in our institute (Edward et al., 2000). A T1-weighted structural image of 256 2 matrix and 32 slices (1 ⫻ 1 ⫻ 3-mm resolution) was also acquired for each subject using a MDEFT sequence optimized for 3-T field strength. Subjects and Task Twelve healthy right-handed subjects (6 male, 6 female), aged between 23 and 40 years (mean 28.8 years), participated in the study. Subjects performed a brief sequential button-press task. Throughout the study subjects rested the index and middle fingers of the right hand on separate microswitches. The required movement involved three rapid alternating button presses with the index–middle–index finger. Re-
fMRI OF SELF-INITIATED AND EXTERNALLY-CUED MOVEMENT
sponse times were recorded from the microswitches via a separate PC. Two different cuing conditions were examined: Externally triggered movement. A 50-Hz pure tone (200 ms duration) was presented via headphones at unpredictable intervals and subjects performed the button-press movement as quickly as possible following the tone. The interval between tones was either 8, 10, 12, 14, 16, 18, 20, 22, or 24 s. Each of these intermovement intervals occurred twice at pseudo-random times throughout the sequence, providing 18 unpredictably cued movements with a mean intertrial interval of 16 s over the whole 280 volume sequence. Self-initiated movement. No cues were provided and subjects made movements at freely selected but irregular times. All subjects had previously received practice with the irregularly timed tones used for externally triggered movements and were carefully instructed to randomly vary the intervals between their self-initiated movements, sometimes long and sometimes short, similar to the irregular intervals of the tones. Since the slice acquisition time was always constant, including between volumes, and the scanner noise (13.3 Hz) was thereby constant, there were no additional external timing cues which subjects could use to pace their movements. Each condition was examined in a separate fMRI acquisition session of 280 brain volumes (approximately 5 min duration). The temporal jitter between volume repetition times (1.125 s) and intertrial intervals for cued conditions (all multiples of 2 s) ensured that different time points of the task-related BOLD response were sampled over repeated trials, thus providing a higher effective sampling rate (or higher temporal resolution) than reflected in the volume repetition time alone. Image Preprocessing All data preprocessing and analysis was performed using Statistical Parametric Mapping (SPM99; London, UK; Friston et al., 1995). Functional image volumes were first corrected for differences in the timing of acquisition between slices by using a sinc interpolation in time to shift the signal in each slice relative to the time of acquisition of the middle slice (Henson et al., 1999; Josephs and Henson, 1999). To correct for subjects’ head motion over time, functional images were spatially realigned to the first image in the time series using a six-parameter rigid body transformation, and a mean functional image was computed for each time series. Each mean functional image was spatially normalized to the reference system of Talaraich and Tournoux (1988) using a standard EPI template based on the reference brain of the Montreal Neurological Institute (Cocosco et al., 1997). Func-
375
tional images were then resliced to 2 ⫻ 2 ⫻ 2-mm voxels according to the resulting spatial realignment and normalization parameters using sinc interpolation in space. Finally, in order to conform with assumptions on which SPM is based, functional images were spatially smoothed with a 4-mm full-width half-maximum isotropic Gaussian kernel and temporally smoothed with a Gaussian kernel of 4-s full-width half-maximum. Structural T1-weighted images for each subject were segmented to extract just the brain from the overall image. These brain-only images were then also spatially normalized into the reference space of Talairach and Tournoux (1988) using a T1 template based on the same reference brain of the Montreal Neurological Institution and resliced to 1 ⫻ 1 ⫻ 1-mm voxels. A group mean structural brain image was then calculated and used for displaying the anatomical location of group functional results. Image Analysis Two sessions per subject for the 12 subjects were analyzed in a single design using the general linear model implementation in SPM99. Hemodynamic responses to the movement task in each session were modeled with a canonical hemodynamic response function (HRF) and its first-order temporal derivative (Josephs et al., 1997). Inclusion of the temporal derivative as a covariate in the general linear model allows deviations in the latency of hemodynamic response onset to be accommodated in the model and allows statistical inferences based on latency differences between conditions to be made (Friston et al., 1998). The timing of modeled events for each subject was specified as the mean reaction time prior to movement onset (between 460 and 887 ms prior to movement according to subject reaction times for externally triggered movements; see Results). This modeled time point represented on average the time of presentation of the auditory cue for externally triggered movements and was always the identical time point in relation to movement onset for every trial for both externally triggered and self-initiated movements. This point in time immediately prior to movement onset was selected for two main reasons. First, with the inclusion of the temporal derivative of the canonical HRF, differences in the timing of actual hemodynamic responses relative to the canonical HRF can be accommodated. The model should therefore be sensitive to actual hemodynamic responses with onsets both earlier (during early movement preparation) and later (during movement execution) than the canonical HRF. Second, the major aim of the study was to examine differences in motor preparatory activity which may begin up to 2 s prior to movement onset for selfinitiated movements (Deecke and Kornhuber, 1978; Cunnington et al., 1995) and perhaps only at the time
376
CUNNINGTON ET AL.
FIG. 1. Areas of activation (P uncorrected ⬍ 0.0001) along the mesial hemisphere wall for externally triggered (a) and self-initiated (b) movements overlaid on the midline sagittal slice of the group mean structural MRI. The white line marks the vertical axis through the anterior commissure, showing the approximate location of the border between pre-SMA (anterior) and SMA-proper (posterior).
of movement execution for externally triggered movements. We therefore chose a time relatively late in movement preparation (according to electrophysiological studies of self-initiated movement) which should be most sensitive to detect onset latency differences in preparatory components between conditions. The magnitude of responses in each condition (the parameter estimates for the height of the canonical HRF in each session) and differences in the magnitude of responses between conditions (positive and negative contrasts between the HRF height parameter estimates for each of the two conditions) were tested across all 12 subjects in a conjunction analysis (Friston et al., 1999). Differences in response latencies between conditions (contrasts between the parameter estimates of the HRF temporal derivative for each condition; Friston et al., 1998) were also tested in a conjunction analysis. Contrasts based on parameter estimates of the temporal derivative can reveal differences in actual hemodynamic response latencies because adding and subtracting the temporal derivative and the canonical HRF basically shifts the canonical HRF in time by an amount dependent on the weighting assigned to the temporal derivative. Parameter estimates of the height of the temporal derivative therefore basically reflect the amount of the temporal shift in the canonical HRF to best fit the actual hemodynamic response. This is based on the assumption that the shape of the actual hemodynamic response does not differ between conditions, but just varies in time. Fitted hemodynamic re-
sponses were therefore plotted over time for voxels that showed significant differences based on contrasts of the temporal derivatives in order to ensure that such significant contrasts were not biased by differences in hemodynamic response shape but reflected significant differences in hemodynamic response latencies. Voxels in which the peak activation in the conjunction analysis over 12 subjects exceeded the corrected statistical threshold P corrected ⬍ 0.05 were considered significantly activated. When overlaying regions with significant peak activations on the mean structural brain image, a more lenient statistical threshold of P uncorrected ⬍ 0.0001 was used. This allowed better visualization both of the extent and of direction of spread of activation in areas with a significant peak, as well as areas which may have shown a trend toward increased activation although not reaching the more conservative corrected statistical threshold. RESULTS Task Performance Mean reaction times for externally triggered movements varied from 460 to 887 ms between subjects (mean ⫾ SD across subjects 679 ⫾ 142 ms). The trialby-trial variability in reaction times within subjects was generally less than that between subjects (mean SD of trial-by-trial reaction times 99 ms). The mean interbutton movement time for the three button
fMRI OF SELF-INITIATED AND EXTERNALLY-CUED MOVEMENT
377
FIG. 2. Areas of activation (P uncorrected ⬍ 0.0001) within left primary motor and superior parietal areas (62– 64 mm above the AC–PC plane) for externally triggered (a) and self-initiated (b) movements.
presses for externally triggered movements (334 ⫾ 126 ms) and self-paced movements (351 ⫾ 140 ms) were not significantly different (t(11) ⫽ 0.94, p ⬎ 0.05). Subjects performed self-initiated movements with similar mean rate and intertrial variability as that for externally triggered movements. The mean interresponse intervals for self-initiated movements (15.7 ⫾ 4.2 s across subjects) compared with externally triggered movements (16 s for all subjects) were not significantly different (t(11) ⫽ 0.28, P ⬎ 0.05). The standard deviations of trial-by-trial interresponse intervals for self-initiated movements (6.85 s across subjects) compared with externally triggered movements (5.31 s
for all subjects) were also not significantly different (t(11) ⫽ 1.58, P ⬎ 0.05). This relatively high standard deviation of intertrial response intervals indicates that subjects did indeed perform self-initiated movements at irregularly timed intervals similar to that for externally triggered movements. Functional Imaging Results—Amplitudes of Hemodynamic Responses Areas of significant activation for both externally triggered and self-initiated movement, as well as areas showing significant differences between conditions, are
378
CUNNINGTON ET AL.
TABLE 1 (a) Areas Showing Significant Activation for Externally Triggered and Self-Initiated Movements Externally triggered movements Area SMA Cingulate Left primary motor Left superior parietal Left insula (middle) Left insula (posterior) Left red nucleus Right auditory cortex Left auditory cortex Right putamen (Left putamen
Self-initiated movements
BA
Peak coordinates
Z score
P corrected
Peak coordinates
Z score
P corrected
6 32 4 4 5 13 13
0, ⫺6, 50 2, 16, 40 ⫺36, ⫺24, 64 ⫺44, ⫺10, 56 ⫺38, ⫺44, 60 ⫺42, 4, ⫺4 ⫺42, ⫺16, ⫺6 ⫺10, ⫺22, ⫺2 66, ⫺26, 14 ⫺44, ⫺22, 8
Inf 7.62 7.31 6.51 5.96 5.55 5.73 6.35 5.80 5.78
⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 0.002 0.015 0.006 ⬍0.001 0.004 0.004
2, ⫺4, 52 2, 8, 42 ⫺38, ⫺24, 64 ⫺46, ⫺10, 56 ⫺34, ⫺36, 64 ⫺42, 2, ⫺4
Inf Inf Inf 6.01 6.33 5.88
⬍0.001 ⬍0.001 ⬍0.001 0.001 ⬍0.001 0.003
⫺8, ⫺24, ⫺2
7.29
⬍0.001
24, ⫺2, ⫺2 ⫺22, 2, 0
6.62 5.14
⬍0.001 0.105)
41, 42 41, 42
(b) Areas Showing Significantly Greater Activation for Externally Triggered Movements Compared with Self-Initiated Movements Area
BA
Peak coordinates
Z score
P corrected
Right auditory cortex (Left auditory cortex
41 41
42, ⫺30, 8 ⫺38, ⫺34, 8
5.31 4.14
0.049 ⬍0.001 uncorrected)
(c) Areas Showing Significantly Earlier Activation for Self-Initiated Compared with Externally Triggered Movements Area
BA
Peak coordinates
Z score
P corrected
SMA (Cingulate
6 32
2, 0, 54 2, 12, 44
5.62 5.13
0.011 0.110)
shown in Table 1. The strongest activation for both movement conditions was found in the supplementary motor area and cingulate cortex (Fig. 1). This area of activation extended from SMA proper (4 to 6 mm posterior to the vertical axis through the anterior commissure (AC)) to pre-SMA (12 to 16 mm anterior to the vertical AC axis) and deep into the cingulate cortex at its anterior extent for both self-initiated and externally triggered movements. The second most significant activation was found within the left primary motor cortex, in a highly consistent location for both movement conditions (Fig. 2). This area extended anteriorly and inferiorly, with a second peak of significant activation at Talairach coordinates close to the border between primary motor (Brodmann’s area 4) and lateral premotor cortex (Brodmann’s area 6). This second peak, however, was the most anterior extent of the left motor cortex activation. There were, therefore, no significantly activated voxels that were located more anteriorly clearly within lateral premotor cortex even at the more lenient uncorrected probability threshold (P uncorrected ⬍ 0.0001). There were also no ipsilateral areas of activation, even at the uncorrected threshold, within right-side primary motor or lateral premotor areas. Other areas of consistent activation for both movement conditions were found in the left superior parietal
cortex (Brodmann’s area 5; Fig. 2) and left insula cortex (Brodmann’s area 13), extending posteriorly in the insula cortex for externally triggered movements. Interestingly, a strong area of activation was also found on the left side in the midbrain with a peak at Talairach coordinates corresponding with the red nucleus for both movement conditions (Fig. 3). For externally triggered movements only, activation was also found bilaterally in primary auditory and lateral temporal cortex (Brodmann’s areas 41 and 42). Although the position of the peak activation differed on left and right sides, at the uncorrected statistical threshold (P uncorrected ⬍ 0.0001) the area of activation on both sides clearly extended into previously reported probabilistic boundaries of primary auditory cortex (Penhune et al., 1996; Rademacher et al., 2001). For self-initiated movements only, basal ganglia activation was also found in the lentiform nucleus, along the border between the putamen and external segment of the globus pallidus of the right hemisphere (Fig. 3). Activation in the equivalent area of the basal ganglia of the left hemisphere just failed to reach significance at the corrected probability threshold (Z ⫽ 5.14, P corrected ⫽ 0.105). For externally triggered movements, there was no evidence of increased activation within the basal ganglia, even at the more lenient uncorrected probability threshold (P uncorrected ⬍ 0.0001).
fMRI OF SELF-INITIATED AND EXTERNALLY-CUED MOVEMENT
379
FIG. 3. Areas of subcortical activation (P uncorrected ⬍ 0.0001) located within the left midbrain (0 –2 mm below the AC–PC plane) for both externally triggered (a) and self-initiated (b) movements and bilaterally within the lentiform nucleus of the basal ganglia for self-initiated movements only.
380
CUNNINGTON ET AL.
FIG. 4. Areas showing significantly earlier hemodynamic response latencies (P uncorrected ⬍ 0.0001) for self-initiated compared with externally triggered movements along the anterior mesial hemisphere wall. The white line marks the vertical axis through the anterior commissure, showing the approximate location of the border between pre-SMA and SMA proper.
Differential comparisons between conditions showed only significantly greater activation within the right primary auditory cortex for externally triggered movements compared with self-initiated movements (Table 1b). Activation in the equivalent area of the left primary auditory cortex failed to reach significance at the corrected level (Z ⫽ 4.14, P uncorrected ⬍ 0.0001). No areas showed significantly greater activation for self-initiated movement compared with externally triggered movements. The areas of strongest activation within supplementary motor area, cingulate cortex, and primary motor cortex, therefore, did not differ significantly in their levels of activation for externally triggered and self-initiated movements. Functional Imaging Results—Timing of Hemodynamic Responses The comparison of hemodynamic response latencies between conditions (the differential contrast of parameter estimates for the temporal derivatives) showed a significantly earlier hemodynamic response onset for self-initiated compared with externally triggered movement within the SMA (Table 1c). Within cingulate cortex, the difference in hemodynamic response latency just failed to reach signif-
icance at the corrected level (Z ⫽ 5.13, P corrected ⫽ 0.110). At the less conservative level (P uncorrected ⬍ 0.0001), the area showing an earlier hemodynamic response for self-initiated movement includes preSMA (on the vertical axis through the AC) and cingulate cortex, but extends only marginally posterior to the vertical AC axis into SMA proper (Fig. 4). No areas showed significantly earlier hemodynamic response onsets for externally triggered compared with self-initiated movements. The voxel within the SMA showing the maximally significant difference in hemodynamic response onsets (Talairach coordinates 2, 0, 54) was examined in more detail. Fitted hemodynamic response functions for this SMA voxel were calculated for each condition and contrasted with mean fitted hemodynamic response functions for peak voxels within left primary sensorimotor cortex (Fig. 5). As can be seen, there was little difference in the shape of the fitted hemodynamic response between conditions, but clearly an earlier response latency within the SMA for self-initiated compared with externally triggered movements (mean difference in peak latency 1.48 ⫾ 0.85 s across subjects). When comparing across cortical areas, the peak latency of the fitted hemodynamic response within the SMA preceded
fMRI OF SELF-INITIATED AND EXTERNALLY-CUED MOVEMENT
381
primary motor, supplementary motor, superior parietal, and insula cortex during voluntary upper-limb movement (Colebatch et al., 1991; Jahanshahi et al., 1995; Boecker et al., 1998). Interestingly, we also found significant subcortical activation within the midbrain for both movement conditions and in the basal ganglia for self-initiated movements. Subcortical Activation: The Midbrain and Basal Ganglia
FIG. 5. Mean fitted hemodynamic response curves for self-initiated (thick line) and externally triggered (thin line) movements: (a) for the peak pre-SMA voxel (Talairach coordinates 2, 0, 54: Table 1c); and (b) the average of peak voxels within left primary sensorimotor cortex (Table 1a). The error bars mark the mean position and standard error of peak amplitudes and peak latencies measured from fitted hemodynamic response curves of individual subjects.
that within left sensorimotor cortex by 1.20 s for selfinitiated movements and by only 0.25 s for externally triggered movements. DISCUSSION Both self-initiated and externally triggered movements activated a common network of areas including supplementary motor area, cingulate cortex, primary motor, superior parietal, and insula cortex. The magnitude of the hemodynamic response in these areas did not differ significantly between movement conditions; however, the timing of the hemodynamic response within the supplementary motor area was significantly earlier for self-initiated compared with externally triggered movements. Whole-brain functional imaging studies generally show activation in a similar network of areas including
In the midbrain, significant activation was found at Talairach coordinates corresponding to the red nucleus of the left hemisphere. The red nucleus is the origin of the rubrospinal tract which descends contralaterally to the lateral spinal cord and, together with direct corticospinal projections, influences motor neurons innervating distal limb muscles. The red nucleus receives input from ipsilateral primary motor cortex, supplementary motor area, and cingulate cortex, as well as significant input from the cerebellum (Burman et al., 2000). Connections from the cerebellum to the rubrospinal tract, via the red nucleus, provide a relatively direct link by which the cerebellum can influence corticospinal output. Previously, the most consistent activation within the red nucleus in humans has been shown in conjunction with strong cerebellar activity during involuntary tremor in essential tremor patients (Wills et al., 1994; Bucher et al., 1997). Most of the cerebellum was not included in the 15 slices acquired in this study; therefore, it is not known whether red nucleus activation was linked with cerebellar activity or whether it perhaps reflected motor cortical influences on the rubrospinal tract. Self-initiated movements were also associated with activation of the lentiform nucleus within the basal ganglia. The basal ganglia are a complex set of subcortical structures which receive input from much of the cortex and project via the thalamus back to prefrontal and motor cortical areas (Cummings, 1993). Through their role in corticosubcortical motor circuits linking supplementary motor and lateral premotor areas (Hoover and Strick, 1993), the basal ganglia are thought to be heavily involved in the internal control and higher order planning of complex movement (Marsden, 1987; Boecker et al., 1998). Interestingly, functional imaging studies often show activation within the basal ganglia during voluntary movement, but fail to find any difference in the level of activation between different movement tasks (see Brooks, 1995, for review), including between internally generated and externally triggered movements (Jahanshahi et al., 1995; Jenkins et al., 2000). Similarly, we found no significant difference in the level of activity within lentiform nucleus when directly comparing externally triggered and self-initiated movements; however, when contrasted with the resting baseline
382
CUNNINGTON ET AL.
level, significant basal ganglia activation was found for self-initiated movements but not for externally triggered movements. SMA and Cingulate Motor Area Activation The most significant area of activation was found along the motor areas of the mesial hemisphere wall, including the pre-SMA, SMA proper, and cingulate cortex. Activation in these areas did not differ in magnitude between movement conditions, but began earlier for self-initiated movements than for externally triggered movements. The SMA is known to play an important role in the internal preparation and control of complex and sequential movement (Cunnington et al., 1996). The pre-SMA in particular shows greater activation when movements are more complex (Boecker et al., 1998) and internally generated (Deiber et al., 1999) and is active early during movement preparation (Lee et al., 1999). The SMA proper appears more to play a motor executive role (Stephan et al., 1995; Boecker et al., 1998) and is active later in conjunction with primary motor cortex activity (Lee et al., 1999). In the cingulate cortex, we found activation predominantly in the posterior end of the rostral cingulate motor area as defined by Picard and Strick (1996), an area which is most involved in motor tasks which are complex or which require some aspect of internal movement selection (Picard and Strick, 1996). As with the pre-SMA, this rostral cingulate motor area shows greater activation during internally generated movements (Wessel et al., 1997; Deiber et al., 1999) and is more involved in early processes of movement preparation (Ball et al., 1999) and in the internal representation or imagination of movement (Stephan et al., 1995). Both the rostral cingulate motor area and the pre-SMA therefore appear to be commonly involved in the internal preparation and control of complex movement. The lack of any activation within lateral premotor areas for both externally triggered and self-initiated movements and the lack of any difference in the magnitude of SMA activation between conditions were unexpected. Previous functional imaging studies involving externally cued movements have generally examined tasks in which single submovements are always made in response to a timing cue at a relatively slow rate. Submovements may be part of an overall sequence performed from memory (Deiber et al., 1991; Van Oostende et al., 1997; Catalan et al., 1998; Deiber et al., 1999) or may be freely selected or internally determined (Deiber et al., 1991; Van Oostende et al., 1997; Haslinger et al., 2001); however, the timing of every submovement is always cued externally at a relatively slow rate. These tasks therefore rely little on internal control mechanisms to prepare and coordinate
the timing of submovement initiation normally involved in ongoing complex motor sequence performance, but require close sensory monitoring to enable the rapid initiation of single submovements in response to external cues. These cued tasks involve strong bilateral activation of lateral premotor areas (Van Oostende et al., 1997; Wessel et al., 1997; Catalan et al., 1998; Haslinger et al., 2001) and when contrasted with movements which are purely self-generated significant differences in activation of the supplementary motor area are found (Wessel et al., 1997; Deiber et al., 1999). In our movement task, only the timing of the initiation of the overall motor sequence was externally cued. Although the sequence was short (three button presses), it still required some internal coordination of the timing of submovement initiation, particularly as the sequence involved alternating movements of two fingers such that the submovements could never be performed simultaneously nor specified without reference to their temporal order. Also, it was stressed to subjects that they should perform the movement task as quickly as possible, emphasizing the need for overall speed in performing the complete movement sequence. Movements would therefore have required a significant degree of preparation and/or preprogramming. Significant activation of supplementary and cingulate motor areas, rather than lateral premotor areas, may therefore reflect involvement of such internal motor preparation and control processes both when the initial onset of the sequence was externally cued and when self-determined. The Timing of Pre-SMA Activation Although the magnitude of activation within SMA and cingulate motor areas did not differ significantly between movement conditions, the onset of activation for self-initiated compared with externally triggered movements was significantly earlier within the preSMA and showed a trend toward earlier activation within the posterior section of the rostral cingulate motor area. Activation within the SMA was also clearly earlier than that within left sensorimotor cortex for self-initiated movements, whereas for externally triggered movements the timing of activation was similar in both cortical areas. There is much evidence that the SMA is normally active early during motor preparatory stages prior to self-initiated voluntary movement onset. Electrophysiological studies have long shown increased cortical activity centered on the scalp approximately above the region of the SMA beginning around 1 to 2 s prior to self-initiated movement (Deecke et al., 1969; Deecke and Kornhuber, 1978). Studies using high-density EEG recordings and current source density analysis have more clearly localized this early preparatory ac-
fMRI OF SELF-INITIATED AND EXTERNALLY-CUED MOVEMENT
tivity to the SMA (Ball et al., 1999; Cui et al., 1999). The study of Ball et al. (1999). in particular, used current density analysis constrained to the cortical surface based on realistic head models obtained from each subject’s own structural MR image to examine the source and temporal pattern of premovement cortical activity in detail. They clearly showed activity beginning in the anterior cingulate motor areas approximately 2.5 s prior to movement and then spreading through pre-SMA and an intermediate area between pre-SMA and SMA proper approximately 2.2 s prior to movement onset. Functional imaging studies have also shown increased activation of the SMA during early stages of movement preparation prior to movement onset (Richter et al., 1997a; Wildgruber et al., 1997; Lee et al., 1999). The study of Lee et al. (1999) further showed different temporal profiles of activity over subregions within the SMA, with the pre-SMA showing the greatest activity during early preparation and the SMA proper showing the greatest activity later during movement execution. We have similarly shown early preparatory activation within the pre-SMA, clearly beginning before movement-related activation of primary sensorimotor areas for self-initiated movements. Although we have not directly measured the onset latency of this early pre-SMA activation, the peak latency of the hemodynamic response was on average 1.48 s earlier for selfinitiated than for externally triggered movements. These results are in good agreement with electrophysiological studies showing increased motor preparatory activity beginning 1 to 2 s prior to self-initiated movement, but little if any early preparatory activity prior to movements which are externally cued at unpredictable times (Cunnington et al., 1995; Jahanshahi et al., 1995). SUMMARY AND CONCLUSIONS Both self-initiated and externally triggered movements involved activation throughout a common network of areas typically found to be involved in the organization and control of voluntary movement of the upper limb. Most significantly, activation of similar magnitude for both movement conditions was found along mesial wall motor areas including the pre-SMA, SMA proper, and posterior end of the rostral cingulate motor cortex. This perhaps reflects a significant degree of internal representation and control which may have allowed optimal performance given the goal of the task (to perform the complete motor sequence as quickly as possible with no additional external cues to guide submovements) when the timing of initial onset of the movement sequence was both externally-triggered and
383
self-determined. There was, however, a significant difference in the timing of activation of the pre-SMA between movement conditions. Pre-SMA activity began significantly earlier for self-initiated movements, with a mean peak latency 1.48 s prior to that for externally triggered movements. This reflects involvement of the pre-SMA in early stage movement preparation processes which precede the onset of voluntary movements. ACKNOWLEDGMENTS This study was supported by a Jubila¨umsfond grant of the Austrian National Bank (P8184). R.C. was also supported by a Neil Hamilton Fairley Fellowship of the National Health and Medical Research Council of Australia (987046).
REFERENCES Ball, T., Schreiber, A., Feige, B., Wagner, M., Lu¨cking, C. H., and Kristeva-Feige, R. 1999. The role of higher-order motor areas in voluntary movement as revealed by high-resolution EEG and fMRI. NeuroImage 10: 682– 694. Boecker, H., Dagher, A., Ceballos-Baumann, A. O., Passingham, R. E., Samuel, M., Friston, K. J., Poline, J.-P., Dettmers, C., Conrad, B., and Brooks, D. J. 1998. Role of the human rostral supplementary motor area and the basal ganglia in motor sequence control: Investigations with H 215O PET. J. Neurophysiol. 79: 1070 –1080. Boetzel, K., Plendl, H., Paulus, W., and Scherg, M. 1993. Bereitschaftspotential: Is there a contribution of the supplementary motor area? Electroencephalogr. Clin. Neurophysiol. 89: 187–196. Brooks, D. J. 1995. The role of the basal ganglia in motor control: Contributions from PET. J. Neurol. Sci. 128: 1–13. Bucher, S. F., Seelos, K. C., Dodel, R. C., Reiser, M., and Oertel, W. H. 1997. Activation mapping in essential tremor with functional magnetic resonance imaging. Ann. Neurol. 41: 32– 40. Burman, K., Darian-Smith, C., and Darian-Smith, I. 2000. Macaque red nucleus: Origins of spinal and olivary projections and terminations of cortical inputs. J. Comp. Neurol. 423: 179 –196. Catalan, M. J., Honda, M., Weeks, R. A., Cohen, L. G., and Hallett, M. 1998. The functional neuroanatomy of simple and complex sequential finger movements: A PET study. Brain 121: 253–264. Cocosco, C. A., Kollokian, V., Kwan, R. K. S., and Evans, A. C. 1997. Brainweb: Online interface to a 3D MRI simulated brain database. NeuroImage 5: 425. Colebatch, J. G., Deiber, M.-P., Passingham, R. E., Friston, K. J., and Frackowiak, R. S. J. 1991. Regional cerebral blood flow during voluntary arm and hand movements in human subjects. J. Neurophysiol. 65: 1392–1401. Cui, R. Q., Huter, D., Lang, W., and Deecke, L. 1999. Neuroimage of voluntary movement: Topography of the Bereitschaftspotential, a 64-channel DC current source density study. NeuroImage 9: 124 – 134. Cummings, J. L. 1993. Fronto-subcortical circuits and human behavior. Arch. Neurol. 50: 873– 879. Cunnington, R., Bradshaw, J. L., and Iansek, R. 1996. The role of the supplementary motor area in the control of voluntary movement. Hum. Mov. Sci. 15: 627– 647. Cunnington, R., Iansek, R., Bradshaw, J. L., and Phillips, J. G. 1995.
384
CUNNINGTON ET AL.
Movement-related potentials in Parkinson’s disease: Presence and predictability of temporal and spatial cues. Brain 118: 935–950. Cunnington, R., Windischberger, C., Deecke, L., and Moser, E. 1999. The use of single event fMRI and fuzzy clustering analysis to examine haemodynamic response timecourses in supplementary motor and primary motor cortical areas. Biomedizinische Technik 44: 116 –119. Deecke, L., and Kornhuber, H. H. 1978. An electrical sign of participation of the mesial ‘supplementary’ motor cortex in human voluntary finger movement. Brain Res. 159: 473– 476. Deecke, L., Scheid, P., and Kornhuber, H. H. 1969. Distribution of readiness potential, pre-motion positivity, and motor potential of the human cerebral cortex preceding voluntary finger movements. Exp. Brain Res. 7: 158 –168. Deiber, M.-P., Honda, M., Ibanez, V., Sadato, N., and Hallett, M. 1999. Mesial motor areas in self-initiated versus externally triggered movements examined with fMRI: Effect of movement type and rate. J. Neurophysiol. 81: 3065–3077. Deiber, M.-P., Passingham, R. E., Colebatch, J. G., Friston, K. J., Nixon, P. D., and Frackowiak, R. S. J. 1991. Cortical areas and the selection of movement: A study with positron emission tomography. Exp. Brain Res. 84: 393– 402. Edward, V., Windischberger, C., Cunnington, R., Erdler, M., Lanzenberger, R., Mayer, D., Endl, W., and Beisteiner, R. 2000. Quantification of fMRI artifact reduction by a novel plaster cast head holder. Hum. Brain Mapp. 11: 207–213. Friston, K. J., Fletcher, P., Josephs, O., Holmes, A., Rugg, M., and Turner, R. 1998. Event-related fMRI: Characterizing differential responses. NeuroImage 7: 30 – 40. Friston, K. J., Holmes, A. P., Price, C. J., Buchel, C., and Worsley, K. J. 1999. Mulisubject fMRI studies and conjunction analyses. NeuroImage 10: 385–396. Friston, K. J., Holmes, A. P., Worsley, K. J., Poline, J.-P., Frith, C. D., and Frackowiak, R. S. J. 1995. Statistical parametric maps in functional imaging: A general linear approach. Hum. Brain Mapp. 2: 189 –210. Haslinger, B., Erhard, P., Ka¨mpfe, N., Boecker, H., Rummeny, E., Schwaiger, M., Conrad, B., and Ceballos-Baumann, A. O. 2001. Event-related functional magnetic resonance imaging in Parkinson’s disease before and after levodopa. Brain 124: 558 –570. Henson, R., Buchel, C., Josephs, O., and Friston, K. J. 1999. The slice-timing problem in event-related fMRI. NeuroImage 6: S125. Hoover, J. E., and Strick, P. L. 1993. Multiple output channels in the basal ganglia. Science 259: 819 – 821. Jahanshahi, M., Jenkins, I. H., Brown, R. G., Marsden, C. D., Passingham, R. E., and Brooks, D. J. 1995. Self-initiated versus externally triggered movements. I. An investigation using measurement of regional cerebral blood flow with PET and movementrelated potentials in normal and Parkinson’s disease subjects. Brain 118: 913–933. Jenkins, I. H., Jahanshahi, M., Jueptner, M., Passingham, R. E., and Brooks, D. J. 2000. Self-initiated versus externally triggered movements. II. The effect of movement predictability on regional cerebral blood flow. Brain 123: 1216 –1228. Josephs, O., and Henson, R. 1999. Event-related functional magnetic resonance imaging: Modelling, inference and optimization. Phil. Trans. R. Soc. Lond. B Biol. Sci. 354: 1215–1228. Josephs, O., Turner, R., and Friston, K. J. 1997. Event-related fMRI. Hum. Brain Mapp. 5: 243–248. Kno¨sche, T., Praamstra, P., Stegeman, D., and Peters, M. 1996. Linear estimation discriminates midline sources and a motor cortex contribution to the readiness potential. Electroencephalogr. Clin. Neurophysiol. 99: 183–190.
Lee, K.-M., Chang, K.-H., and Roh, J.-K. 1999. Subregions within the supplementary motor area activated at different stages of movement preparation and execution. NeuroImage 9: 117– 123. Macpherson, J., Wiesendanger, M., Marangoz, C., and Miles, T. S. 1982. Corticospinal neurones of the supplementary motor area of monkeys; A single unit study. Exp. Brain Res. 48: 81– 88. Marsden, C. D. 1987. What do the basal ganglia tell premotor cortical areas. In Motor Areas of the Cerebral cortex: Ciba Foundation Symposium 132 (R. Porter, Eds.), pp. 282–300. Wiley, Chichester. Menon, R. S., Luknowsky, D. C., and Gati, J. S. 1998. Mental chronometry using latency-resolved functional MRI. Proc. Natl. Acad. Sci. USA 95: 10902–10907. Mushiake, H., Inase, M., and Tanji, J. 1991. Neuronal activity in the primate premotor, supplementary, and precentral motor cortex during visually guided and internally determined sequential movements. J. Neurophysiol. 66: 705–718. Neshige, R., Lu¨ders, H. O., and Shibasaki, H. 1988. Recording of movement-related potentials from scalp and cortex in man. Brain 111: 719 –736. Papa, S. M., Artieda, J., and Obeso, J. A. 1991. Cortical activity preceding self-initiated and externally triggered voluntary movement. Mov. Disord. 6: 217–224. Penhune, V. B., Zatorre, R. J., MacDonald, J. D., and Evans, A. C. 1996. Interhemispheric anatomical differences in human primary auditory cortex: Probabilistic mapping and volume measurement from magnetic resonance scans. Cereb. Cortex 6: 661– 672. Picard, N., and Strick, P. L. 1996. Motor areas of the medial wall: A review of their location and functional activation. Cereb. Cortex 6: 342–353. Rademacher, J., Morosan, P., Schormann, T., Schleicher, A., Werner, C., Freund, H.-J., and Zilles, K. 2001. Probabilistic mapping and volume measurement of human primary auditory cortex. NeuroImage 13: 669 – 683. Rao, S. M., Binder, J. R., Bandettini, P. A., Hammeke, T. A., Yetkin, F. Z., Jesmanowicz, A., Lisk, L. M., Morris, G. L., Mueller, W. M., Estkowski, L. D., Wong, E. C., Haughton, V. M., and Hyde, J. S. 1993. Functional magnetic resonance imaging of complex human movements. Neurology 43: 2311–2318. Remy, P., Zilbovicius, M., Leroy-Willig, A., Syrota, A., and Samson, Y. 1994. Movement- and task-related activations of motor cortical areas: A positron emission tomographic study. Ann. Neurol. 36: 19 –26. Richter, W., Andersen, P. M., Georgopoulos, A. P., and Kim, S.-G. 1997a. Sequential activity in human motor areas during a delayed cued finger movement task studied by time-resolved fMRI. NeuroReport 8: 1257–1261. Richter, W., Ugurbil, K., Georgopoulos, A. P., and Kim, S.-G. 1997b. Time-resolved fMRI of mental rotation. NeuroReport 8: 3697–3702. Romo, R., and Schultz, W. 1987. Neuronal activity preceding selfinitiated or externally timed arm movements in area 6 of monkey cortex. Exp. Brain Res. 67: 656 – 662. Stephan, K. M., Fink, G. R., Passingham, R. E., Silbersweig, D., Ceballos-Baumann, A. O., Frith, C. D., and Frackowiak, R. S. J. 1995. Functional anatomy of the mental representation of upper extremity movements in healthy subjects. J. Neurophysiol. 73: 373–386. Talairach, J., and Tournoux, P. 1988. Co-planar Stereotaxic Atlas of the Human Brain. Thieme, Stuttgart. Thaler, D., Rolls, E. T., and Passingham, R. E. 1988. Neuronal activity of the supplementary motor area (SMA) during internally
fMRI OF SELF-INITIATED AND EXTERNALLY-CUED MOVEMENT and externally triggered wrist movements. Neurosci. Lett. 93: 264 –269. Van Oostende, S., Van Hecke, P., Sunaert, S., Nuttin, B., and Marchal, G. 1997. FMRI studies of the supplementary motor area and the premotor cortex. NeuroImage 6: 181–190. Weeks, R. A., Honda, M., Catalan, M. J., and Hallett, M. 2001. Comparison of auditory, somatosensory, and visually instructed and internally generated finger movements: A PET study. NeuroImage 14: 219 –230. Wessel, K., Zeffiro, T., Toro, C., and Hallett, M. 1997. Self-paced versus metronome-paced finger movements: A positron emission tomography study. J. Neuroimage 7: 145–151.
385
Wildgruber, D., Erb, M., Klose, U., and Grodd, W. 1997. Sequential activation of supplementary motor area and primary motor cortex during self-paced finger movement in human evaluated by functional MRI. Neurosci. Lett. 227: 161–164. Wills, A. J., Jenkins, I. H., Thompson, P. D., Findley, L. J., and Brooks, D. J. 1994. Red nuclear and cerebellar but no olivary activation associated with essential tremor: A positron emission tomography study. Ann. Neurol. 36: 636 – 642. Windischberger, C., Barth, M., Beisteiner, R., Cunnington, R., Edward, V., Erdler, M., Kaindl, T., and Moser, E. 1999. Fuzzy clustering as a model-free analysis method in single-event fMRI. NeuroImage 9: S561.
