Abstract

The objective of this study was to investigate brain areas involved in distinguishing sensory events caused by self-generated movements from similar sensory events caused by externally generated movements using functional magnetic resonance imaging. Subjects performed 4 types of movements: 1) self-generated voluntary movement with visual feedback, 2) externally generated movement with visual feedback, 3) self-generated voluntary movement without visual feedback, and 4) externally generated movement without visual feedback, this design. This factorial design makes it possible to study which brain areas are activated during self-generated ankle movements guided by visual feedback as compared with externally generated movements under similar visual and proprioceptive conditions. We found a distinct network, comprising the posterior parietal cortex and lateral cerebellar hemispheres, which showed increased activation during visually guided self-generated ankle movements. Furthermore, we found differential activation in the cerebellum depending on the different main effects, that is, whether movements were self- or externally generated regardless of visual feedback, presence or absence of visual feedback, and activation related to proprioceptive input.

Introduction

Generation of movement requires integration of motor commands with sensory feedback in order to achieve behavioral goals such as reaching toward a target or grasping an object. One important mean by which a motor act is accomplished is through the ability to distinguish between the sensory consequences caused by one's own movements and the sensory effects imposed by external events. In order to make this distinction, the brain must be able to differentiate between the sensory effects caused by a self-generated movement and similar sensory effects caused by external perturbations of the limbs. At present, the exact neural mechanisms that compute such information are not known, although the integration of proprioceptive, visual, and other sensory stimuli with the motor commands is thought to be of critical importance. The purpose of this study was to use functional magnetic resonance imaging (fMRI) to study which parts of the brain display selective activation when visual feedback is caused by a self-generated movement compared with situations where the visual feedback is caused by externally generated movements.

The posterior parietal cortex (PPC), together with other motor-related structures such as primary (M1) and premotor cortices (PMCs), supplementary motor areas (SMAs), the basal ganglia, and cerebellum, is a key structure involved in planning and control of movements (Iacoboni 2006). An important role of the PPC is online visual guidance of movements (Jeannerod 1988; Stein and Glickstein 1992; Clower et al. 1996; Desmurget et al. 1999), which require integration of efferent motor commands, and afferent proprioceptive and visual feedback. Most studies on humans have focused on upper limb movements where visual control plays an important role in both reaching and grasping movements (Prablanc et al. 2003; Culham et al. 2003). However, visual guidance of lower limb movements also plays an important role for obstacle avoidance and accurate foot placement during everyday locomotion (Reynolds and Day 2005). Until now, only limited effort has been put into the study of accurate visual guidance of lower limb movements using functional imaging techniques. There are experimental benefits of studying lower limb movements during fMRI: If we want to record accurate kinematical measures of the movements used for online guidance, the positioning of recording devices such as Electromyographic (EMG) electrodes and goniometers outside the scanner bore reduce the induced gradient artifact noise as compared with a positioning inside the scanner bore. We have also studied ankle movements in order to understand whether visual guidance of simple lower limb flexion-extension movements require cortical control at a level that is comparable with more complicated reaching and grasping movements and whether relatively simple lower limb movements require involvement of rather complex control mechanisms involved in forward models used for movement control. There is evidence available showing cortical involvement in human locomotion (see Nielsen 2003 for review), and we believe that the underlying anatomical networks responsible for visual guidance of lower limb movements are similar to those used in upper limb visual guidance during reaching movements. We therefore hypothesize that self-generated movement using visual feedback guidance will lead to increased activation in posterior regions of the superior parietal lobule, because these regions have shown increased activation during reaching movements (Iacoboni 2006).

Efference copies (ECs) (von Holst and Mittelstaedt 1950) or corollary discharges (CDs) (Sperry 1950) are important neural mechanisms involved in feedforward movement control and learning, and we believe these signals are important components in sensorimotor integration. EC and CD are copies of efferent motor commands or outcome measures after a comparison between the expected and actual sensory consequences of a movement. It has been proposed that ECs are generated in SMA (Haggard and Whitford 2004) or PMC (Chronicle and Glover 2003; Ellaway et al. 2004), whereas the comparison between the expected and the actual feedback takes place in PPC and the cerebellum (Blakemore and Sirigu 2003). The main computational module contains a predictor, that is, a forward model, and a controller, that is, an inverse model, plus a context-dependent signal, known as the responsibility signal. The responsibility signal and the forward model, together determine the degree to which a particular module should participate in the motor control process. The inverse model is involved in the learning process and determines whether the desired goal is achieved (Wolpert and Kawato 1998; Haruno et al. 2001). The EC is the input to the forward model, and the CD is used in the comparison with the actual sensory consequence of a voluntary movement and the outcome of this comparison is used for correction and learning. Neuroimaging data have shown that these modules exist in the cerebellum and are responsible for controlling motor commands in different contextual settings in particular switches from one module to another (Imamizu et al. 2003, 2004). It has also been confirmed indirectly in behavioral studies that ECs are involved in movement error correction (Angel 1976; Desmurget and Grafton 2000). Another noteworthy feature of ECs and CDs is that they may influence the perceived sensory consequences of self-generated movements by letting the activation in the lateral anterior cerebellum reflect whether sensory stimuli are self- or externally generated (Blakemore et al. 1998). In this study, we believe that the cerebellum will be involved in the selection of the appropriate internal (forward/inverse) models that depends on the features characterizing the different movement types used in the present experiment. In the main interaction effect that we will study, possible cerebellar candidate regions of interest may be the lateral cerebellar hemispheres because they are connected with the parietal cortex, which we hypothesize will be involved in visuomotor integration (Clower et al. 2001).

In order to study effects of self-generated visual feedback, we developed an experimental design that allowed us to separate signals involved in the visual guidance of self-generated ankle movements from signals that originate from the proprioceptive signals when the ankle joint is externally moved by an experimenter. An essential aspect of this design is the fact that the self- and externally generated movements are closely matched with respect to kinematics and visual feedback provided to the subject. In order to accomplish such similarities, previous fMRI studies have relied on electrophysiological measurements outside the scanner environment and afterward assumed that similar performance was obtained during scanning (Ehrsson et al. 2001; MacIntosh et al. 2004; Ciccarelli et al. 2005). In the present study, we recorded EMG activity throughout the scanning period and were able to ensure absence of muscle activity during externally generated movements. We furthermore provided visual biofeedback from a goniometer placed on the subject's ankle joint throughout the scanning to ensure that the self- and externally generated movements were comparable.

Materials and Methods

Subjects

Eighteen subjects were scanned and used for the analysis (10 females, 8 males, and mean age 27.3 years, range 18–42 years). The experiments were approved by the local ethics committee (KF 01-131/03) and followed the regulations expressed in the Declaration of Helsinki (1964).

Magnetic Resonance Scanning

All scans were acquired using a 3 Tesla magnetic resonance (MR) scanner (Siemens Magnetom Trio, Erlangen, Germany). For fMRI blood oxygen level–dependent (BOLD) weighted scans, a whole-brain gradient-echo echo planar imaging (EPI) sequence with a repetition time (TR) = 2400 ms, echo time (TE) = 30 ms, and flip angle = 90° was employed, using a 64 × 64 matrix with an in-plane resolution of 3 × 3 mm2. Each volume consisted of 40 slices, with slice thickness 3 mm. The slices were obtained in an interleaved fashion beginning with the bottom slice. Each functional experiment consisted of 350 EPI volumes, lasting 840 s in all. This scan was performed twice, once for each of the 2 tasks with and without visual feedback. Structural scans were also acquired using a magnetization-prepared–rapid gradient-echo sequence with TR = 1540 ms, TE = 3.93 ms, flip angle = 9°, and 192 × 256 × 256 acquired matrix, with a resolution of 1 × 1 × 1 mm3.

Physiological Monitoring

The subjects' pulse and respiration were measured using an MR-compatible pulse oximeter and a respiration belt. Pulse and respiration were sampled at 50 Hz and recorded using scanner software. EMG activity was obtained from muscle tibialis anterior (TA) in the left leg using a bipolar surface electrode configuration. Silver/silver chloride (Ag/AgCl) electrodes (Blue Sensor, Ambu Inc., Ølstykke, Denmark) were placed over the belly of the muscle with an interelectrode distance of 2 cm. The EMG signal was amplified (2000×), using a custom-built EMG amplifier, filtered (band pass, 25 Hz to 1 kHz), sampled at 2 kHz, and stored on a PC for off-line analysis (CED 1401+ with Spike 2.5 software, Cambridge Electronic Design Ltd, Cambridge, UK).

The position of the ankle joint was measured using a twin axis SG110A ankle goniometer (Biometrics Ltd, Gwent, UK). Goniometer signals were calibrated, amplified, and sampled in Spike 2.5. During one part of the experimental session, goniometer signals were displayed to the subject as visual feedback of movement using a Canon LV 7545 LCD projector with a refresh rate of 60 Hz. The feedback was projected onto a screen and viewed via mirrors placed on the head coil above the subject's head. In the other part of the experiment, the same visual feedback was provided but without the signal from the goniometer.

Both the EMG and goniometer signal were influenced by gradient artifacts from the scanner; however, with appropriate placement of the recording electrodes and wires, it was possible to obtain almost gradient artifact–free recordings as displayed in Figure 1, showing amplified and band-pass filtered EMG as described above without any further filtering.

Figure 1.

Kinematical data and feedback. Kinematical data from (a) self-generated movement and (b) externally generated movement. The upper curve in each display shows the EMG activity and the lower curve the goniometer signal. Notice that none of the curves are filtered after they are recorded. (c) Shows the visual feedback during a self- or externally generated movement of the ankle. The feedback was made similar in all subjects such that they had to move the goniometer signal up to the upper horizontal line (target line) for each contraction. This window was shown for 2 s. The line below the goniometer signal indicates the pace (pace signal). When the line traversed from left to right, one full contraction and relaxation had to be made. In the upper center, the cue signal was displayed for 2 s in the respective color (red, green, or blue) before the block lasting 12 s. During movement, the goniometer had to be moved up to the target line, preferably, such that it just touched the target line in the center of the screen. The feedback window has been color inverted for display reasons and converted into gray scale.

Figure 1.

Kinematical data and feedback. Kinematical data from (a) self-generated movement and (b) externally generated movement. The upper curve in each display shows the EMG activity and the lower curve the goniometer signal. Notice that none of the curves are filtered after they are recorded. (c) Shows the visual feedback during a self- or externally generated movement of the ankle. The feedback was made similar in all subjects such that they had to move the goniometer signal up to the upper horizontal line (target line) for each contraction. This window was shown for 2 s. The line below the goniometer signal indicates the pace (pace signal). When the line traversed from left to right, one full contraction and relaxation had to be made. In the upper center, the cue signal was displayed for 2 s in the respective color (red, green, or blue) before the block lasting 12 s. During movement, the goniometer had to be moved up to the target line, preferably, such that it just touched the target line in the center of the screen. The feedback window has been color inverted for display reasons and converted into gray scale.

Task

The primary task of the subjects was to perform ankle dorsiflexion movements with the left foot. Movements were either self-generated by the subject or externally generated by an experimenter. Subjects were scanned twice. In both scans, the task alternated between self-generated movements and externally generated movements separated by periods of rest. One scan was performed with a visual feedback indicating ankle displacement from the goniometer and a pace signal generated by the computer. This was presented in the Spike 2.5 software package. The other scan was performed without the visual feedback of the movement only with the pace signal present. Half of the subjects were first scanned with visual feedback, the other half without visual feedback first. Subjects were placed supine within the scanner bore. The subject's feet were placed in a custom-built wooden box with pedals restricting sideways movements. The rotational axis was about the malleolus. The weight of the pedal meant that one movement cycle consisted of dorsiflexion followed by plantarflexion of the ankle joint during concentric, respectively, eccentric contraction of the ankle dorsiflexors. An experimenter placed inside the scanner room produced the externally generated movements by moving a pedal and matched the amplitude of the movements with the movements performed by the subjects. In addition, the experimenter viewed the same visual feedback as the subject thereby making the same movement frequency as during the movements performed by the subject.

Subjects alternated between 12 s of rest, 12 s of self-generated movement, 12 s of rest, and 12 s of externally generated movement. Before each of these periods, a 2-s color-coded cue indicated the kind of task that was about to be performed. Red indicated rest, green indicated self-generated movement, and blue indicated externally generated movement. The nuances of the colors were matched with respect to luminance. The pace signal was a line moving from the left to the right of the screen. During this period, one ankle movement cycle should be performed by the subject or was made by the experimenter. The pace frequency was 0.5 Hz, meaning that the subject performed 6 full dorsilflexion movements during the periods of movement.

Experimental Design

To summarize the 4 different movement and feedback combinations, subjects performed or were exposed to self-generated movement with visual feedback (SF), externally generated movement with visual feedback (EF), self-generated movement without visual feedback (SN), and externally generated movement without visual feedback (EN). Thereby, it was a 2 × 2 factorial design, where the 2 factors were movement type and visual feedback type, that is, (self-generated movement, externally generated movement) × (with visual feedback, without visual feedback). Having this design, it is possible to study the interaction between movement type and feedback type.

Kinematical Analysis

The goniometer signal, which was used for postprocessing scanning analysis, was low-pass filtered with a cutoff frequency at 2.5 Hz, in order to remove gradient noise from the scanner. The onset time of each individual movement was calculated and used for the analysis of the brain activity. Peak values in the goniometer signal were calculated in order to compare the amplitude of the movements in the self- and externally generated movements. This was done separately for movements with and without visual feedback as an analysis of variance (ANOVA). The EMG activity was recorded in order to clarify whether the subject had performed any contractions of TA during the externally generated movements.

fMRI Analysis

All fMRI analyses were performed in SPM2 (http://www.fil.ion.ucl.ac.uk/spm/).

Preprocessing

All scans were realigned to the first volume, then normalized to a standard EPI template in Montreal Neurological Institute (MNI) space, and spatial smoothing was applied with a (8-mm full-width half maximum) Gaussian smoothing kernel.

Single Subject Analysis

The analyses were performed using a general linear model (GLM). For each subject, a GLM was constructed with 2 separate sessions: one for the experiment with visual feedback and one for the experiment without. We used the onset times for each contraction during the periods of self-generated movements and modeled the period containing 6 contractions as 6 events. The displacement of the ankle joint during the 6 externally generated movements was modeled in a similar way. Two regressors were constructed within each session of the GLM: one for all contractions during self-generated movements and one for all displacements during externally generated movements. These regressors were a collection of delta functions at the time points for the individual movements, as found in the kinematical analysis. Both of the regressors were convolved with a canonical hemodynamic response function (HRF). We believe this approach will reflect the actual kinematical behavior better than using 12 s blocks, that is, a boxcar regressor, at the epochs where we expected the subjects had moved. However, given the temporal smoothing abilities of the HRF, a 12-s block and 6 delta functions separated by 2 s do not give significant differences with respect to the final regressor used in the estimation of the GLM.

In order to correct for the structured noise induced by respiration and cardiac pulsation, we included RETROICOR (RETROspective Image–based CORrection method) nuisance covariates in the design matrix (Glover et al. 2000). These regressors are a Fourier expansion of the aliased cardiac and respiratory oscillations. Six regressors for respiration and 10 regressors for cardiac pulsation were included. We also included 24 regressors that remove residual movement artefacts with spin-history effects, which have been shown to remain even after image realignment (Friston et al. 1996). At time t, these can be expressed as (x(t), y(t), z(t), Rx(t), Ry(t), Rz(t), x(t − 1), y(t − 1), z(t − 1), Rx(t − 1), Ry(t − 1), Rz(t − 1), x2(t), y2(t), z2(t), Rx2(t), Ry2(t), Rz2(t), x2(t − 1), y2(t − 1), z2(t − 1), Rx2(t − 1), Ry2(t − 1), Rz2(t − 1)), where t − 1 corresponds to movement in the previous scan, that is, the spin-history effect. This set of nuisance regressors has also been shown to reduce inter- and intrasubject variation significantly (Lund et al. 2005). Having all 4 types of nuisance regressors in the design improves the assumption of independently and identically distributed errors (Lund et al. 2006). For the analysis, we also applied a high-pass filter with a cutoff frequency at 1/128 Hz.

Group Analysis

Second-level group analysis was made as random effects analysis. The estimated beta images for the self-generated movement and externally generated movement regressors from each of the single subjects from their separate GLMs were used in an ANOVA, that is, we had a two-by-two factorial design with the factors (self-generated movement, externally generated movement) × (with visual feedback, without visual feedback), which can be separated out in the 4 groups: SF, EF, SN, and EN. We performed nonsphericity correction at the second level (Glaser and Friston 2004). For the purpose of thresholding the obtained statistical images, we correct for multiple comparisons using the false discovery rate (FDR) approach at q < 0.05 (Genovese et al. 2002).

Interaction Effects

In order to investigate where the activation in the brain was larger when visual feedback was generated by the subject as part of a self-generated movement as compared with the activation evoked by self-generated movement without feedback or visual feedback without self-generated movement, we made the comparison (SF > SN) > (EF > EN). This means that we subtracted the activation during self-generated movement without feedback from activation during self-generated movement with feedback. From this, we subtracted the difference between externally generated movements with and without feedback. Thereby, it was possible to determine activation from visual feedback when it is self-generated where the effect of proprioceptive feedback was subtracted under the assumption that proprioceptive feedback is the same during self- and externally generated movements. Expressed as a t-contrast, this question is equivalent to the comparison [1 0 −1 0] > [0 1 0 −1] = [1 0 −1 0] − [0 1 0 −1]. This expression can be simplified to the expression [1 −1 −1 1] by vector subtraction.

We also tested where externally generated visual feedback gave rise to larger brain activation than self-generated visual feedback; again we subtracted the effect of proprioceptive feedback, that is, we performed the following comparison: (EF > SF) > (EN > SN). Again this interaction can be expressed as an equivalent comparison of t-contrasts [−1 1 0 0] > [0 0 −1 1] = [−1 1 0 0] − [0 0 −1 1], which is equivalent to the t-contrast [−1 1 1 −1] by vector subtraction.

Main Effects

The results from the 4 main effects are reported, that is, activation during SF, EF, SN, and EN all versus the implicit rest condition. For display reasons, we also applied a threshold using Gaussian random field (GRF) theory correcting for multiple comparisons (GRF P < 0.05) to the main effects (Fig. 3a–d). The adaptive FDR correction method gave, for these particular main effects and conjunction analyses (Fig. 3e), activation in large parts of the brain, which made it hard to see the location of the peak activation, when the maps were displayed as maximum intensity projections (MIPs).

Figure 2.

Self-generated visual feedback. (a) Shows the interaction (SF > EF > (SN > EN) presented as an overlayed activation map onto a template structural MRI and as a MIP. This map shows areas where there is an extra effect on the visual feedback due to the self-generated movement. (b) Shows areas more activated during self-generated than during externally generated movements, regardless of whether feedback was present or absent. (c) Shows areas where there was more activation with than without feedback. Activation displayed in (a) show the integration of the activation in (b) and (c). (d) Shows the common activation for self-generated movements and subpanel (e) the common activation for externally generated movements. Activation shown in (b) can then be considered as the difference in activation between the maps in subpanels (d) and (e). (f) Shows common activation for movements with feedback. (g) Shows common activation for movements without feedback. Subpanel (c) can then be considered as the difference between the maps presented in subpanels (f) and (g). All maps are corrected for multiple comparisons using FDR and thresholded at q < 0.05.

Figure 2.

Self-generated visual feedback. (a) Shows the interaction (SF > EF > (SN > EN) presented as an overlayed activation map onto a template structural MRI and as a MIP. This map shows areas where there is an extra effect on the visual feedback due to the self-generated movement. (b) Shows areas more activated during self-generated than during externally generated movements, regardless of whether feedback was present or absent. (c) Shows areas where there was more activation with than without feedback. Activation displayed in (a) show the integration of the activation in (b) and (c). (d) Shows the common activation for self-generated movements and subpanel (e) the common activation for externally generated movements. Activation shown in (b) can then be considered as the difference in activation between the maps in subpanels (d) and (e). (f) Shows common activation for movements with feedback. (g) Shows common activation for movements without feedback. Subpanel (c) can then be considered as the difference between the maps presented in subpanels (f) and (g). All maps are corrected for multiple comparisons using FDR and thresholded at q < 0.05.

Contrast and Conjunction Activations

Contrast activations are also reported, these include the conjunction of SF > EF and SN > EN, that is, SF > EF ∧ (logical AND operator) SN > EN, showing where self-generated movements give rise to more activation than externally generated movements (with and without visual feedback), that is, where self-generated movements give rise to increased activation compared with externally generated movements regardless of visual feedback. The individual contrasts SF > EF and SN > EN are reported in the supplementary material. We also performed a conjunction analysis of 2 contrast effects, SF > SN ∧ EF > EN, which reveal areas more activated with than without visual feedback, in order to display the effect of visual feedback.

We also performed the conjunction, EF > SF ∧ EN > SN, where we find areas showing activation during externally generated movements compared with self-generated movements regardless of visual feedback. This is also reported in the supplementary material in details.

In order to find brain areas that were activated in common for the main effects versus implicit rest condition, we performed the following conjunction analyses: SF ∧ SN to show activation in common for self-generated movements with and without visual feedback versus rest, EF ∧ EN to show activation in common for externally generated movements with and without visual feedback versus rest, SF ∧ EF to show what is in common for self- and externally generated movements with visual feedback versus rest, and SN ∧ EN to show what is in common for self- and externally generated movements without visual feedback versus rest.

The conjunction procedure is performed using the framework described in Friston et al. (2005), as suggested by Nichols et al. (2005). This conjunction method, as implemented in SPM2, can be performed within the ANOVA analysis and can be made in order to determine common activations between 2 or more contrasts, whereas a t-contrast over multiple regressors only looks for the mean. Thereby, the t-contrast can potentially show activation that is only significant in one condition, but if this activation is very strong, the mean of the 2 conditions is significant, but not the individual contrasts. The contrast [1 1 −1 −1] would in principle also show the same as the conjunction of SF > SN ∧ EF > EN, but because the contrast approach displays the mean of a linear combination of the regressors, whereas the conjunction exclusively show what is significant for SF > SN and (meaning logical AND) significant for EF > EN separately, we have chosen the conjunction.

Contrast Estimates

We have displayed contrast estimates for the group analysis from some of the regions showing significant activation for the main effects in Figure 3, and in Figure 4, we also display contrast estimates from cerebellar regions showing significant effects in some of the contrasts and conjunction analyses. The units of the contrast estimates correspond to percentages of whole-brain mean activation, for example, a value of 0.5 indicates an activation of 0.5% of the whole-brain mean from a particular voxel.

Figure 3.

Main effects and conjunctions of main effects. (a)–(d) Show maps of the main effects SF, EF, SN, and EN against the implicit rest condition thresholded at q < 0.05, using FDR correction in the first column and using GRF theory (p < 0.05, GRF) in the second column, in order to show a more nuanced maps, which is difficult using FDR, in particular, for EF and EN. (e) Shows the common activation for all 4 main effects. (f) Shows contrast estimates from 4 regions (M1-S1, SMA, frontal operculum, and SII) found activated in all 4 conditions. M1-S1 appears to show the same level of activation in all 4 conditions, SMA and frontal operculum appear to show slightly more activation during self-generated movements. However, the difference is not statistical significant. SII appears to show more activation during externally generated movements. This effect is significant confirmed by the contrast EF > SF ∧ EN > SN displayed in Supplementary Figure 1b and Supplementary Table 2.

Figure 3.

Main effects and conjunctions of main effects. (a)–(d) Show maps of the main effects SF, EF, SN, and EN against the implicit rest condition thresholded at q < 0.05, using FDR correction in the first column and using GRF theory (p < 0.05, GRF) in the second column, in order to show a more nuanced maps, which is difficult using FDR, in particular, for EF and EN. (e) Shows the common activation for all 4 main effects. (f) Shows contrast estimates from 4 regions (M1-S1, SMA, frontal operculum, and SII) found activated in all 4 conditions. M1-S1 appears to show the same level of activation in all 4 conditions, SMA and frontal operculum appear to show slightly more activation during self-generated movements. However, the difference is not statistical significant. SII appears to show more activation during externally generated movements. This effect is significant confirmed by the contrast EF > SF ∧ EN > SN displayed in Supplementary Figure 1b and Supplementary Table 2.

Functional Localization

Functional activations are reported based on their localization in MNI space. Peak activations within activated clusters were identified and superimposed on a single subject T1-weighted structural MRI (SPM2 template image). The localization was then compared with the atlas provided by Duvernoy (1999) and confirmed in at least 2 different slice orientations, that is, both in an axial and a coronal slice, before the location, and MNI coordinate was assigned a macroscopic anatomical location. With respect to anatomical localization within the cerebellum, which is not included in the Duvernoy atlas, we used the SPM Anatomy toolbox (Eickhoff et al. 2005) with its accompanying template image, which provide gross anatomical information about regions localized in MNI space and probabilistic information about some but not all Brodmann areas. This approach should be a valid approach in order to determine macroscopic anatomical localization (S. Eickhoff, personal communication).

Results

Kinematics

We found no significant differences (P > 0.05 ANOVA) for any subjects between the amplitudes of the self- and externally generated movements, nor when we compared movements with and without visual feedback. The EMG recordings revealed that there were no signs of muscle activity during passive movements.

Self-Generated Movements with Visual Feedback

The increased BOLD signal caused by visual feedback during self-generated movements (SF > SN) > (EF > EN) was found in bilateral middle occipital gyrus (MOG), superior occipital gyrus, superior parietal gyrus, and right middle temporal gyrus (MTG). Right cerebellum (VI) and left cerebellum crus 1 were also significant. Activation was significant at q < 0.05 (FDR corrected for multiple comparisons). Table 1 together with Figure 2 summarizes these results. The cerebellar activation with an additional contrast estimate (see “Main Effects” in “Results” for an explanation) is displayed in Figure 4 in yellow.

Figure 4.

Cerebellar activations. The top display shows activations in the cerebellum from 6 different contrasts, interactions, or conjunctions in different colors. Yellow—(SF > SN) > (EF > EN) found in left cerebellum crus 1 and right cerebellum (VI). Red—(EF > SF) > (EN > SN) found in left crus 1 and crus 2. Green—SF > EF ∧ SN > EN found in cerebellar vermis (9) and left cerebellar hemisphere (VI). Dark blue—SF ∧ EF ∧ SN ∧ EN found in left medial cerebellar peduncle. Light blue (cyan)—SF > SN ∧ EF > EN found in left cerebellum (VIII). Magenta—EF > SF ∧ EN > SN found in right cerebellum crus 1 and crus 2. The same threshold is used in all maps (q < 0.05, FDR). The bottom display shows contrast estimates of the 4 regressors for all subjects from the different cerebellar regions. The colors correspond to those used in the activation maps.

Figure 4.

Cerebellar activations. The top display shows activations in the cerebellum from 6 different contrasts, interactions, or conjunctions in different colors. Yellow—(SF > SN) > (EF > EN) found in left cerebellum crus 1 and right cerebellum (VI). Red—(EF > SF) > (EN > SN) found in left crus 1 and crus 2. Green—SF > EF ∧ SN > EN found in cerebellar vermis (9) and left cerebellar hemisphere (VI). Dark blue—SF ∧ EF ∧ SN ∧ EN found in left medial cerebellar peduncle. Light blue (cyan)—SF > SN ∧ EF > EN found in left cerebellum (VIII). Magenta—EF > SF ∧ EN > SN found in right cerebellum crus 1 and crus 2. The same threshold is used in all maps (q < 0.05, FDR). The bottom display shows contrast estimates of the 4 regressors for all subjects from the different cerebellar regions. The colors correspond to those used in the activation maps.

Table 1

Self-generated modulation of visual feedback

Movement type Anatomical region MNI coordinates Z-score 
(SF > SN) > (EF > EN) Superior parietal gyrus −21, −63, 63 5.19 
  −12, −78, 54 4.02 
  18, −66, 60 4.47 
  15, −57, 69 3.82 
  24, −60, 69 3.74 
 Superior occipital gyrus −15, −84, 42 3.50 
 MOG −39, −87, 24 4.93 
  42, −84, 24 5.18 
 Angular gyrus −30, −90, 33 4.39 
  27, −87, 36 3.65 
 MTG 48, −78, 15 4.03 
  −45, −87, 6 3.71 
 Cerebellum (crus 1) −39, −45, −33 3.37 
 Cerebellum (VI) 33, −39, −30 3.92 
SF > EF ∧ SN > EN Cerebellar vermis (9) 3, −63, −33 4.62 
 Cerebellum (VI) −21, −66, −21 4.47 
SF > SN ∧ EF > EN Short insular gyrus −36, 21, −12 3.12 
 SFG −21, −15, 63 3.04 
 Superior frontal sulcus 24, −12, 57 3.26 
 Precentral gyrus −57, 3, 39 3.46 
  54, 6, 36 3.15 
  −54, −21, 36 3.27 
  63, −18, 39 5.16 
 Fusiform gyrus 36, −66, −18 4.16 
 Intraparietal sulcus −27, −48, 54 5.38 
 Angular gyrus 30, −51, 57 5.24 
  −21, −75, 36 3.79 
  27, −75, 36 4.99 
 MOG −18, −84, 36 3.42 
  21, −90, 27 5.88 
  48, −75, 0 6.71 
  −42, −81, 3 5.77 
 Lateral occipital sulcus −42, −66, 6 4.17 
  51, −63, 3 5.92 
 Fourth occipital gyrus −18, −90, −15 4.21 
  15, −81, −12 4.72 
 Cerebellum (VIII) −12, −69, −48 3.41 
 Superior colliculus −3, −27, −9 2.86 
Movement type Anatomical region MNI coordinates Z-score 
(SF > SN) > (EF > EN) Superior parietal gyrus −21, −63, 63 5.19 
  −12, −78, 54 4.02 
  18, −66, 60 4.47 
  15, −57, 69 3.82 
  24, −60, 69 3.74 
 Superior occipital gyrus −15, −84, 42 3.50 
 MOG −39, −87, 24 4.93 
  42, −84, 24 5.18 
 Angular gyrus −30, −90, 33 4.39 
  27, −87, 36 3.65 
 MTG 48, −78, 15 4.03 
  −45, −87, 6 3.71 
 Cerebellum (crus 1) −39, −45, −33 3.37 
 Cerebellum (VI) 33, −39, −30 3.92 
SF > EF ∧ SN > EN Cerebellar vermis (9) 3, −63, −33 4.62 
 Cerebellum (VI) −21, −66, −21 4.47 
SF > SN ∧ EF > EN Short insular gyrus −36, 21, −12 3.12 
 SFG −21, −15, 63 3.04 
 Superior frontal sulcus 24, −12, 57 3.26 
 Precentral gyrus −57, 3, 39 3.46 
  54, 6, 36 3.15 
  −54, −21, 36 3.27 
  63, −18, 39 5.16 
 Fusiform gyrus 36, −66, −18 4.16 
 Intraparietal sulcus −27, −48, 54 5.38 
 Angular gyrus 30, −51, 57 5.24 
  −21, −75, 36 3.79 
  27, −75, 36 4.99 
 MOG −18, −84, 36 3.42 
  21, −90, 27 5.88 
  48, −75, 0 6.71 
  −42, −81, 3 5.77 
 Lateral occipital sulcus −42, −66, 6 4.17 
  51, −63, 3 5.92 
 Fourth occipital gyrus −18, −90, −15 4.21 
  15, −81, −12 4.72 
 Cerebellum (VIII) −12, −69, −48 3.41 
 Superior colliculus −3, −27, −9 2.86 

Note: The table display anatomical regions with accompanying peak coordinates in MNI space where self-generated visual feedback give rise to increased activation over and above effects caused by the externally generated visual feedback and proprioceptive influence ([SF > SN] > [EF > EN]). Then, we have regions showing more activation during self-generated movements, regardless of visual feedback (SF > EF ∧ SN > EN). Finally, regions showing more activation with feedback (SF > SN ∧ EF > EN), regardless of movement type, are reported. All regions were found activated above a corrected thresholds level (q < 0.05, FDR).

Self-Generated > Externally Generated Movements

The conjunction between the 2 contrasts SF > EF ∧ SN > EN was also tested and shown in Figures 2b and 4 in green and reported in Table 1. Significant activation was found in cerebellum (V1) and the cerebellar vermis (9). The conjunction was performed based on the 2 individual contrasts SF > EF and SN > EN, which showed significant (FDR q < 0.05) differences in the cerebellum (vermis and VI, VIII) (see Table 2 and Supplementary Fig. 2). With visual feedback, increased activation was present during self-generated movements compared with externally generated movements in both cerebellar hemispheres (VI) and in the vermis (Supplementary Fig. 2a). Without visual feedback, activation was only present in the left cerebellar hemisphere and vermis (Supplementary Fig. 2b).

Table 2

Conjunction of main effects

Movement type Anatomical region MNI coordinates Z-score 
SF ∧ SN SFG 0, −6, 54 4.86 
 Frontal operculum 54, 12, 0 4.95 
 Precentral gyrus 6, −30, 69 7.29 
 Postcentral gyrus 12, −36, 78 7.11 
  66, −21, 27 5.03 
 Lateral fissure–parietal operculum −45, −33, 21 4.74 
 Calcarine sulcus 15, −78, 3 4.77 
 Cerebellum (middle cerebellar peduncle) −15, −33, −30 7.30 
 Cerebellar vermis (1/2) 0, −45, −18 6.12 
 Cerebellum (V1) −30, −66, −24 6.00 
 Cerebellum (IX) −6, −63, −48 4.93 
EF ∧ EN Precentral gyrus 6, −30, 66 8.00 
  −48, 0, −3 5.37 
  54, 9, 0 5.20 
 Postcentral gyrus 12, −39, 72 7.94 
 Cingulate gyrus 6, −9, 48 6.74 
 Lateral fissure–parietal operculum −45, −33, 21 7.64 
  42, −27, 18 7.56 
 Putamen −24, 0, 9 5.06 
  27, 0, 6 5.87 
 Cerebellum (middle cerebellar peduncle) −15, −33, −30 7.12 
SF ∧ EF SFG 0, −6, 54 4.86 
 Frontal operculum 54, 12, 0 4.95 
 Precentral gyrus 6, −30, 69 7.29 
  −57, 3, 39 4.51 
 Postcentral gyrus 12, −36, 78 7.10 
 Supramarginal gyrus 63, −21, 42 5.95 
  45, −30, 36 4.91 
 Lateral fissure–parietal operculum −45, −33, 21 4.74 
 MOG −45, −81, 0 5.00 
  48, −75, 0 5.98 
 Cerebellum (middle cerebellar peduncle) −15, −33, −30 7.28 
 Cerebellum (VI) −27, −69, −24 3.84 
  30, −75, −18 2.99 
 Cerebellum (VIII) −12, −69, −48 4.10 
SN ∧ EN Frontal operculum 54, 9, 3 5.08 
 Precentral gyrus 6, −30, 69 7.50 
  6, −12, 72 6.67 
 Postcentral gyrus 12, −39, 72 7.63 
 Short insular gyrus −45, 3, 0 4.90 
  45, 6, 3 4.85 
 Lateral fissure–parietal operculum −45, −33, 21 5.86 
  45, −27, 21 5.75 
 Lateral fissure 60, −30, 24 5.57 
 Cerebellum (middle cerebellar peduncle) −15, −33, −30 7.12 
Movement type Anatomical region MNI coordinates Z-score 
SF ∧ SN SFG 0, −6, 54 4.86 
 Frontal operculum 54, 12, 0 4.95 
 Precentral gyrus 6, −30, 69 7.29 
 Postcentral gyrus 12, −36, 78 7.11 
  66, −21, 27 5.03 
 Lateral fissure–parietal operculum −45, −33, 21 4.74 
 Calcarine sulcus 15, −78, 3 4.77 
 Cerebellum (middle cerebellar peduncle) −15, −33, −30 7.30 
 Cerebellar vermis (1/2) 0, −45, −18 6.12 
 Cerebellum (V1) −30, −66, −24 6.00 
 Cerebellum (IX) −6, −63, −48 4.93 
EF ∧ EN Precentral gyrus 6, −30, 66 8.00 
  −48, 0, −3 5.37 
  54, 9, 0 5.20 
 Postcentral gyrus 12, −39, 72 7.94 
 Cingulate gyrus 6, −9, 48 6.74 
 Lateral fissure–parietal operculum −45, −33, 21 7.64 
  42, −27, 18 7.56 
 Putamen −24, 0, 9 5.06 
  27, 0, 6 5.87 
 Cerebellum (middle cerebellar peduncle) −15, −33, −30 7.12 
SF ∧ EF SFG 0, −6, 54 4.86 
 Frontal operculum 54, 12, 0 4.95 
 Precentral gyrus 6, −30, 69 7.29 
  −57, 3, 39 4.51 
 Postcentral gyrus 12, −36, 78 7.10 
 Supramarginal gyrus 63, −21, 42 5.95 
  45, −30, 36 4.91 
 Lateral fissure–parietal operculum −45, −33, 21 4.74 
 MOG −45, −81, 0 5.00 
  48, −75, 0 5.98 
 Cerebellum (middle cerebellar peduncle) −15, −33, −30 7.28 
 Cerebellum (VI) −27, −69, −24 3.84 
  30, −75, −18 2.99 
 Cerebellum (VIII) −12, −69, −48 4.10 
SN ∧ EN Frontal operculum 54, 9, 3 5.08 
 Precentral gyrus 6, −30, 69 7.50 
  6, −12, 72 6.67 
 Postcentral gyrus 12, −39, 72 7.63 
 Short insular gyrus −45, 3, 0 4.90 
  45, 6, 3 4.85 
 Lateral fissure–parietal operculum −45, −33, 21 5.86 
  45, −27, 21 5.75 
 Lateral fissure 60, −30, 24 5.57 
 Cerebellum (middle cerebellar peduncle) −15, −33, −30 7.12 

Note: The 4 different conjunctions show what is in common for self-generated movement (regardless of feedback), externally generated movements (regardless of feedback), movements with visual feedback (regardless of movement type), and movements without visual feedback (regardless of movement type). All statistical maps are thresholded at q < 0.05 (corrected using FDR).

With > Without Visual Feedback

The conjunction SF > SN ∧ EF > EN showed increased BOLD signal within large parts of the dorsal visual pathway (superior parietal lobule), V1 and V5, and surprisingly also within the precentral gyrus (premotor regions) and cerebellum (VIII). The cerebellar activation is additionally displayed in Figure 4 in cyan. The contrast estimate showed a larger BOLD response with than without visual feedback, but the estimate of the BOLD response was also larger for the self-generated movements.

The results are summarized in Figure 2c. The activations from the individual contrast, of which the conjunction was constructed, are displayed in Supplementary Figure 2c,d.

Main Effects of Movements versus Rest

Activations for the 4 main effects are reported in Supplementary Table 1 and displayed in Figure 3 as MIPs at FDR- and GRF-corrected significance levels.

Figure 3f shows contrast estimates for 4 different regions that showed increased activation for the main effects. The contrast estimates were made from a voxel in precentral/postcentral gyri (M1-S1), lateral fissure/parietal operculum/secondary somatosensory (SII), superior frontal gyrus (SFG) (SMA), and frontal operculum. All main effects showed large activations within the right sensorimotor (M1-S1) regions of the foot area, and left cerebellum near the medial cerebellar peduncle. The tasks with visual feedback also showed activation within V5 and calcarine sulcus (V1). Both self- and externally generated movements showed activation within SII regions, but it was more pronounced for the externally generated movements. The frontal operculum along with SMA was also found activated both for self- and externally generated movements, but here, the activation was more pronounced for the self-generated movements. A conjunction analysis of all the movement types SF, SN, EF, and EN versus rest (SF ∧ EF ∧ SN ∧ EN) is shown in Figure 3e, and the cerebellar activation is also displayed in Figure 4 in blue.

Conjunctions of Main Effects

The conjunctions of the main effects show activation in common for self- (Fig. 2d) and externally generated movements (Fig. 2e). Common activation with feedback is displayed in Figure 2f and without feedback in Figure 2g. All conjunctions showed increased activation in pre- and postcentral gyrus, lateral fissure (SII), frontal operculum, and middle cerebellar peduncle. In common for the externally generated movements with and without feedback (EF ∧ EN) was in addition activation of the putamen. In common for self- and externally generated movements with feedback was in addition activation in MOG. For the self-generated movements, there was also activation in the SFG. The results of the pairwise conjunctions displayed in Figure 3c,d are also reported in Table 2.

Externally Generated Movements with Visual Feedback

Increased BOLD signal caused by passive viewing of the visual feedback during externally generated movements (EF > SF) > (EN > SN) was found in lateral orbital gyrus, gyrus rectus (GR), SFG, middle frontal gyrus, inferior frontal gyrus, superior temporal gyrus, MTG, inferior and superior temporal sulci, parahippocampal gyrus, amygdala, precuneus, cuneus, and cerebellum (crus 1 and 2). Activation was significant at q < 0.05 (FDR corrected for multiple comparisons). These results are summarized in Table 1 and displayed in Supplementary Figure 1a. The cerebellar activation with additional contrast estimates is displayed in Figure 4 in red.

Externally Generated > Self-Generated Movements

Activation in the conjunction EF > SF ∧ EN > SN was found bilaterally in precuneus, middle temporal gyri, amygdale, parahippocampal gyri, GR, and in the both cerebellar hemispheres for EF > SF and in the right cerebellar hemisphere only for EN > SN. The results of the conjunction are displayed in Supplementary Figure 1b. In Figure 4, the contrast estimates for the cerebellar region for the conjunction (EF > SF ∧ EN > SN) is displayed along with the cerebellar activations. Given the contrast estimates, there is a BOLD signal increase during the externally generated movements and a decrease during the self-generated movements.

Discussion

The main findings of the study are that the PPC showed increases in the BOLD signal, when visual feedback occurred during a self-generated movement compared with an externally generated movement, and the cerebellum exhibited different patterns of activation depending on movement type, feedback, and interactions between movement signals and feedback.

Brain Activity during Voluntary Movement with Self-Generated Visual Feedback

We interpret the PPC activation observed for the self-generated visual feedback as a reflection of the integration of visual and proprioceptive feedback with the efferent motor command. The location corresponds well with our prediction of activation in superior and posterior regions of the parietal lobule (Iacoboni 2006). This is similar to reaching movements, whereas grasping movements engage more inferior and anterior parietal regions. As sketched in Figure 2, we believe the PPC integrates the activation-related self-generated movement found in the cerebellar vermis and left hemisphere of the cerebellum (Fig. 2b) with the activation in the dorsal visual stream including PPC and frontal motor regions like PMC related to visual feedback (Fig. 2c). Furthermore, the activation in the lateral hemispheres of the cerebellum has an anatomical plausible location (Clower et al. 2001; Imamizu et al. 2003) and its functional role may reflect the fact that these regions subserve a module specific for visual movement guidance.

The fact that we found PPC activation in visuomotor integration is not surprising, because the PPC has been shown to be involved in multiple tasks that require information from both the sensory system and the motor system. The PPC codes vision and central motor signals into a common frame of reference used in guidance of movement (Cohen and Andersen 2002). It has been associated with visually guided reaching (Kertzman et al. 1997; Desmurget et al. 1999; Eskandar and Assad 1999) and has been found to be involved in multisensory integration for movement guidance (reviewed by Fogassi and Lupino 2005). Furthermore, the PPC plays a role in visuomotor coordinate transformation in general (Grefkes et al. 2004) and, specifically, during reaching (Della-Maggiore et al. 2004). Several studies indicate that the PPC is also involved in the transformation of visual representations into an egocentric framework (Andersen et al. 1985; Stein 1989).

The fact that PPC is activated during relatively simple ankle movements suggests that not only do upper and lower limb movements share common networks for visual guidance, but it also suggests that lower limb locomotion guided by vision may be controlled by highly developed association regions in the parietal lobe. Our results are also parallel to the findings by Jueptner et al. (1996), who compared activation when subjects copied prespecified line drawings with activation related to the generation of line drawings themselves. This comparison demonstrated bilateral PPC activation and activation of the cerebellum (vermis, nuclei, and the left lateral hemisphere).

The distinct activation of PPC found in the present study during self-generated movements with visual feedback compared with externally generated movements might be explained by a difference in the level of attention to the movement during the 2 types of movement. However, the intraparietal regions, in which increased activation has previously been found when subjects actively direct their attention toward a motor task, are located more anterior than the region, which we have identified (Rowe et al. 2002). Furthermore, intraparietal activation in attention-demanding tasks was accompanied by frontal and insular activation, which was absent in our data. Finally, we did not instruct the subjects to change their level of attention, but if such changes nevertheless took place, they are unlikely to have happened systematically throughout the experiment in all subjects. Based on these considerations, we do not think that our results are caused by changes in the level of attention.

Of particular interest, it has been shown that the PPC is engaged in the detection of self-generated movements (MacDonald and Paus 2003); indeed, patients with parietal lesions exhibit a reduced ability to discriminate between their own movements and that of others (Sirigu et al. 1999). This suggests that the PPC may not only be involved in visual guidance but also in the interpretation of whether a movement is self-generated or externally generated and thus partly explain the larger activation in the PPC during the self-generated movements in our study.

Activation in the Cerebellum

As seen in Figure 4, we have demonstrated different patterns of activation in the cerebellum, depending on the type of movement and visual feedback.

Incoming proprioceptive signals, common for self- and externally generated movements, gave rise to activation in the cerebellum near the medial cerebellar peduncle (depicted in blue, Fig. 4). Both the present study and other findings support a role of this area in the processing of proprioceptive feedback. First, we found activation during both self- and externally generated movements when we compared it with the rest condition as a conjunction. Second, 2 of the participating subjects, who received electrical stimulation of the tibial nerve, showed activation in the exact same region (Christensen MS and Nielsen JB, unpublished data). Finally, the present result corresponds to previous findings from Ciccarelli et al. (2005) of a similar conjunction analysis of active and passive foot movements.

Areas of the cerebellum, which showed significantly greater activation during self-generated movements compared with externally generated movements, SF > EF ∧ SN > EN, (depicted in green, Fig. 4), corresponded well to previous monkey studies (reviewed by Stein and Glickstein 1992), which have shown activation of the vermis when movements were self-generated. Furthermore, the separation of cerebellar functions done by Imamizu et al. (2003) into a cognitive part in the lateral cerebellum and a sensorimotor part in the medial anterior part corresponds well with our finding of the engagement during self-generated movements.

Providing visual feedback during movement compared with absent feedback (SF > SN ∧ EF > EN, depicted in cyan, Fig. 4) gave rise to activation in cerebellum (VIII) in an inferior region but still not in the lateral portions, again supporting the medial-lateral sensorimotor versus cognitive separation of cerebellar functions (Imamizu et al. 2003). Passive displacement of the ankle gave rise to activation in right posterior lateral cerebellar hemisphere (EF > SF ∧ EN > SN, depicted in magenta, Fig. 4), and whether this activation can be considered lateral or medial is debatable and hence, potentially, may diverge from the medial-lateral cerebellar function separation.

Both interaction analyses, that is, self- and externally generated movements with feedback ([SF > EF] > [SN > EN] depicted in yellow and [EF > SF] > [EN > SN] depicted in red, Fig. 4), gave rise to lateralized activation in anterior (self-generated) and posterior (externally generated) activation. This suggested that visuomotor integration is a higher order cognitive process given the location of the cerebellar activation.

From these results, we draw 2 conclusions. First, we suggest that the anterior parts of the cerebellum are responsible for voluntary guidance of movements, whereas the posterior regions are engaged when subjects passively are exposed to visual or proprioceptive feedback. Second, there seems to be a distinction between the numbers of activated regions within the cerebellum depending on whether the investigated contrast just represent activation related to passive sensory processing (like EF > SF ∧ EN > SN and SF > SN ∧ EF > EN), whereas contrasts investigating the interaction between self-generated movements and sensory feedback all give rise to activation in 2 or more regions ([SF > EF] > [SN > EN], [EF > SF] > [EN > SN], and SF > EF ∧ SN > EN). Thirdly, our results seem to correspond with previous separations of cerebellar function into sensorimotor versus cognitive based on an anatomical medial versus lateral distinction. We interpret these findings in the context of recent developments in cerebellar functions related to prediction of sensory consequences during self-generated movements and how multiple modules of forward predictions are coupled with inverse controller models and responsibility signals (Wolpert and Kawato 1998; Haruno et al. 2001). Each of the different contrasts show activation corresponding to its responsible module and when integration of visual, proprioceptive, or motor command signals occurs, we find activation in multiple modules corresponding to the specific task.

The cerebellum has previously been found to attenuate the sensory consequences of self-generated movements (Blakemore et al. 1998). We have been able to replicate this finding showing a significant decrease in cerebellar activation during self-generated movements (EF > SF ∧ EN > SN), and this activation seems to follow the activation in SII, where there is a significant decrease in activation during self-generated movements compared with externally generated movements (See contrast estimates for SII in Fig. 3f) and for cerebellum in Figure 4 (depicted in magenta). But whether the cerebellum alone is responsible for this attenuation remains unknown because other structures also show significant differences when this contrast is tested, which is shown in Supplementary Figure 1b.

Activation in Sensorimotor Cortices

We found that self- and externally generated movements with and without visual feedback shared primary activation sites within the sensorimotor strip in the pre- and postcentral gyri (Fig. 3, Supplementary Table 1). The fact that we did not find any significant differences when we compared self- with externally generated movements (SF > EF and SN > EN) suggests that the main contribution to this primary sensorimotor activation is of a sensory nature rather than due to the efferent motor drive. Similar to us, Ciccarelli et al. (2005) did not find any significant differences between the activation in the contralateral hemisphere of the moving foot in the primary sensorimotor areas when self- and externally generated movements were compared (SF > EF and SN > EN). But our results are in contrast to previous PET studies (Christensen et al. 2000; Johannsen et al. 2001) and to the general opinion that self-generated movements give rise to more M1 activation than externally generated movements.

Studies of self- and externally generated finger movements (Mima and others, 1999; Reddy and others, 2001) showed a tendency toward increased bilateral activation of M1 during self-generated movement, which was only strictly contralateral in the externally generated movement. Some of the differences in the results between our study and previous studies can be traced back to differences in the movement type, the investigated body part used, and the activation paradigms (i.e., pseudorandomized blocks [Ciccarelli et al. 2005] in comparison to predictable blocks [our study], duration of blocks, movement frequency, and even scanner differences). Further studies are needed to clarify these issues. Such studies would require accurate measures of kinematics in order to ensure that movements are comparable across subjects. Our findings, together with previous studies (Mima et al. 1999; Reddy et al. 2001; Ciccarelli et al. 2005), suggest that the general notion of strict unilateral activation in primary motor areas during movements needs revision.

The self-generated movements showed an activation peak within an area in the SFG [0, −6, 54] (SF ∧ SN), whereas the externally generated movements show a distinct peak in the cingulate gyrus [6, −9, 48] (EF ∧ EN), both within the large cluster of sensorimotor activation, but when compared, these differences were not statistically significant. The increased activation in the SFG (SMA) was expected during self-generated movement (Kornhuber and Deecke 1965) and may reflect activation related to the generation of an EC (Haggard and Whitford 2004), but the increased activation in the cingulate gyrus (cingulate motor area) during the externally generated movements was surprising. These differences are observed as locations of maximum changes in the BOLD signal within a cluster of activated voxels, but they are in contrast to a study by Dobkin et al. (2004), where the opposite was found with a block paradigm of longer block durations.

Methodological Considerations

Accurate interpretations of brain activation data obtained with fMRI studies of motor behavior require that experimental parameters, such as movement frequency and amplitude, are fixed, and only the variable of interest is changed (i.e., visual feedback). Otherwise, the results may be influenced by systematic differences that were not monitored or accounted for. Recording of EMG has previously been shown to be a feasible technique for reliable measurement of muscle activity during fMRI scanning, when the signals are appropriately filtered off-line (van Duinen et al. 2005). The present study has confirmed this and demonstrated that behavioral information of the performed movement during the scanning may also be obtained and presented online for the subject as biofeedback with the use of commercially available goniometers. This is an important technical advancement as compared with other studies, which have been forced to use measurements outside the scanner environment prior to scanning (Ciccarelli et al. 2005), which is clearly suboptimal. Online measurement of the performed movements also provides a way of ensuring that the subject performs the movements as requested and with minimal variability from trial to trial. Biofeedback may also be necessary for simulation of real-life settings inside the scanner environment.

As this and other studies have shown sensory feedback generated by movement makes an important contribution to the brain activation observed in imaging studies, and proper interpretation of the brain activation observed in motor control paradigms requires that such feedback signals are distinguished from those originating from central motor processes. We therefore suggest that passive (externally generated) movements are used as control in such motor control studies and that accurate measurement of the performance of the movement and the muscle activity is obtained in order to ensure the comparability between self- and externally generated movements.

The Functional Network underlying Sensorimotor Integration

We have demonstrated that PPC and the cerebellum are engaged in visuomotor integration during self-generated movements with visual feedback. Furthermore, we have shown that different regions of the cerebellum are responsible for neural computations underlying both voluntary executions of movements, integration of proprioceptive signals per se and during passive displacement of the ankle joint with passive “viewing” of visual feedback.

Based on our results and knowledge from previous studies, we believe that PPC and the lateral cerebellar hemispheres integrate the proprioceptive signals with the visual feedback and the efferent motor command. The neural counterpart of the proprioceptive feedback from the ankle movement is pronounced both in the cerebellar medial peduncle and SI, the neural counterpart for the visual feedback is pronounced along the dorsal visual stream, and the motor command is represented in M1 or PMC/SMA. Although we did not find significant differences in M1/PMC/SMA when we compared self-with externally generated movements, previous studies suggest that these regions are important in motor integration.

The role of the cerebellum should probably be seen in the context of forward models and their involvement in comparisons between predictions of sensory consequences of movements, actual sensory consequences of movements, and desired states. In addition to a medial/lateral separation of cerebellar motor/cognitive functions, as suggested by Imamizu et al. (2003), we propose an anterior/posterior separation into an active motor engagement/passive sensory processing of the cerebellum. When multiple sensorimotor modalities are engaged in a task, like integrating vision with movements, multiple kinds of modules containing forward, inverse, and desired models of the present state are involved, whereas when only a single modality is investigated, like the main effect of visual feedback, only a single kind of module is involved.

Conclusion

We have been able to distinguish areas within the PPC and cerebellum that show increased activation during self-generated ankle dorsiflexion movements with visual feedback. The activation of these areas was augmented by the influence of the self-generated movement rather than proprioceptive feedback. The cerebellum, on the other hand, showed specific activation for visual feedback caused by externally generated movements under similar proprioceptive conditions. We speculate that these areas may play a role in the perceptual experience of self-generated body movements.

Supplementary Material

Supplementary material can be found at:http://www.cercor.oxfordjournals.org/.

We would like to thank Torben E. Lund for helpful comments regarding the analysis. M.S.C was supported by a 3-year PhD grant from the Faculty of Science, University of Copenhagen. S.S.G was supported by a grant from the research priority area “Body and Mind,” University of Copenhagen. The MR scanner was donated by the Simon Spies Foundation. Conflict of Interest: None declared.

References

Andersen
RA
Essick
GK
Siegel
RM
Encoding of spatial location by posterior parietal neurons
Science
 , 
1985
, vol. 
230
 
4724
(pg. 
456
-
458
)
Angel
RW
Efference copy in the control of movement
Neurology
 , 
1976
, vol. 
26
 
12
(pg. 
1164
-
1168
)
Blakemore
S-J
Sirigu
A
Action prediction in the cerebellum and in the parietal lobe
Exp Brain Res
 , 
2003
, vol. 
153
 
2
(pg. 
239
-
245
)
Blakemore
S-J
Wolpert
DM
Frith
CD
Central cancellation of self-produced tickle sensation
Nat Neurosci
 , 
1998
, vol. 
1
 
7
(pg. 
635
-
640
)
Christensen
LO
Johannsen
P
Sinkjær
T
Petersen
N
Pyndt
HS
Nielsen
JB
Cerebral activation during bicycle movements in man
Exp Brain Res
 , 
2000
, vol. 
135
 
1
(pg. 
66
-
72
)
Chronicle
EP
Glover
J
A ticklish question: Does magnetic stimulation of the primary motor cortex give rise to an “efference copy.” Cortex
3
 , 
2003
, vol. 
9
 (pg. 
105
-
110
)
Ciccarelli
O
Toosy
AT
Marsden
JF
Wheeler-Kingshott
CM
Sahyoun
C
Matthews
PM
Miller
DH
Thompson
AJ
Identifying brain regions for integrative sensorimotor processing with ankle movements
Exp Brain Res
 , 
2005
, vol. 
166
 
1
(pg. 
31
-
41
)
Clower
DM
Hoffman
JM
Votaw
JR
Faber
TL
Woods
RP
Alexander
GE
Role of posterior parietal cortex in the recalibration of visually guided reaching
Nature
 , 
1996
, vol. 
383
 
6601
(pg. 
618
-
621
)
Clower
DM
West
RA
Lynch
JC
Strick
PL
The inferior parietal lobule is the target of output from the superior colliculus, hippocampus, and cerebellum
J Neurosci
 , 
2001
, vol. 
21
 
16
(pg. 
6283
-
6291
)
Cohen
YE
Andersen
RA
A common refrence frame for movement plans in the posterior parietal cortex
Nat Rev Neurosci
 , 
2002
, vol. 
3
 
7
(pg. 
553
-
562
)
Culham
JC
Danckert
SL
DeSouza
JF
Gati
JS
Menon
RS
Goodale
MA
Visually guided grasping produces fMRI activation in dorsal but not ventral stream brain areas
Exp Brain Res
 , 
2003
, vol. 
153
 
2
(pg. 
180
-
189
)
Della-Maggiore
V
Malfait
N
Ostry
DJ
Paus
T
Stimulation of the posterior parietal cortex interferes with arm trajectory adjustments during the learning of new dynamics
J Neurosci
 , 
2004
, vol. 
24
 
44
(pg. 
9971
-
9976
)
Desmurget
M
Epstein
CM
Turner
RS
Prablanc
C
Alexander
GE
Grafton
ST
Role of the posterior parietal cortex in updating reaching movements to a visual target
Nat Neurosci
 , 
1999
, vol. 
2
 
6
(pg. 
563
-
567
)
Desmurget
M
Grafton
S
Forward modeling allows feedback control for fast reaching movements
Trends Cogn Sci
 , 
2000
, vol. 
4
 
11
(pg. 
423
-
431
)
Dobkin
BH
Firestine
A
West
M
Saremi
K
Woods
R
Ankle dorsiflexion as an fMRI paradigm to assay motor control for walking during rehabilitation
Neuroimage
 , 
2004
, vol. 
23
 
1
(pg. 
370
-
381
)
Duvernoy
HM
The human brain: surface, three-dimensional sectional anatomy with MRI, and blood supply
1999
Berlin (Germany)
Springer Verlag
Ehrsson
HH
Fagergren
E
Forssberg
H
Differential fronto-parietal activation depending on force used in a precision grip task: an fMRI study
J Neurophysiol
 , 
2001
, vol. 
85
 
6
(pg. 
2613
-
2623
)
Eickhoff
SB
Stephan
KE
Mohlberg
H
Grefkes
C
Fink
GR
Amunts
K
Zilles
K
A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data
Neuroimage
 , 
2005
, vol. 
25
 (pg. 
1325
-
1335
)
Ellaway
PH
Prochazka
A
Chan
M
Gauthier
MJ
The sense of movement elicited by transcranial magnetic stimulation in humans is due to sensory feedback
J Physiol
 , 
2004
, vol. 
556
 
2
(pg. 
651
-
660
)
Eskandar
EN
Assad
JA
Dissociation of visual, motor and predictive signals in parietal cortex during visual guidance
Nat Neurosci
 , 
1999
, vol. 
2
 
1
(pg. 
88
-
93
)
Fogassi
L
Luppino
G
Motor functions of the parietal lobe
Curr Opin Neurobiol
 , 
2005
, vol. 
15
 
6
(pg. 
626
-
631
)
Friston
KJ
Penny
WD
Glaser
DE
Conjunction revisited
Neuroimage
 , 
2005
, vol. 
25
 
3
(pg. 
661
-
667
)
Friston
KJ
Williams
S
Howard
R
Frackowiak
RS
Turner
R
Movement-related effects in fMRI time-series
Magn Reson Med
 , 
1996
, vol. 
35
 
3
(pg. 
346
-
355
)
Genovese
CR
Lazar
NA
Nichols
T
Thresholding of statistical maps in functional neuroimaging using the false discovery rate
Neuroimage
 , 
2002
, vol. 
15
 
4
(pg. 
870
-
878
)
Glaser
D
Friston
K
Frackowiak
RSJ
Friston
KJ
Frith
CD
Dolan
RJ
Price
CJ
Zeki
S
Ashburner
J
Penny
W
Chapter 39: variance components
Human brain function
 , 
2004
2nd ed
San Diego (CA)
Elsevier
(pg. 
781
-
791
)
Glover
GH
Li
T-Q
Ress
D
Image-based method for retrospective correction of physiological motion effects in fMRI: RETROICOR
Magn Reson Med
 , 
2000
, vol. 
44
 
1
(pg. 
162
-
167
)
Grefkes
C
Ritzl
A
Zilles
K
Fink
GR
Human medial intraparietal cortex subserves visuomotor transformation
Neuroimage
 , 
2004
, vol. 
23
 
4
(pg. 
1494
-
1506
)
Haggard
P
Whitford
B
Supplementary motor area provides an efferent signal for sensory suppression
Brain Res Cogn Brain Res
 , 
2004
, vol. 
19
 
1
(pg. 
52
-
58
)
Haruno
M
Wolpert
DM
Kawato
M
MOSAIC Model for Sensorimotor Learning and Control
Neural Comp
 , 
2001
, vol. 
13
 (pg. 
2201
-
2220
)
Iacoboni
M
Forthcoming
Visuo-motor integration and control in the human posterior parietal cortex: evidence from TMS and fMRI
Neuropsychologia
 , 
2006
Imamizu
H
Kuroda
T
Mizauchi
S
Yoshioka
T
Kawato
M
Modular organization of internal models of tools in the human cerebellum
Proc Natl Acad Sci USA
 , 
2003
, vol. 
100
 (pg. 
5461
-
5466
)
Imamizu
H
Kuroda
T
Yoshioka
T
Kawato
M
Functional magnetic resonance imaging examination of two modular architectures for switching multiple internal models
J Neurosci
 , 
2004
, vol. 
24
 
5
(pg. 
1173
-
1181
)
Jeannerod
M
The neural and behavioural organization of goal-directed movements
1988
Oxford (UK)
Oxford University Press
Johannsen
P
Christensen
LO
Sinkjær
T
Nielsen
JB
Cerebral functional anatomy of voluntary contractions of ankle muscles in man
J Physiol
 , 
2001
, vol. 
535
 
2
(pg. 
397
-
406
)
Jueptner
M
Jenkins
IH
Brooks
DJ
Frackowiak
RS
Passingham
RE
The sensory guidance of movement: a comparison of the cerebellum and basal ganglia
Exp Brain Res
 , 
1996
, vol. 
112
 
3
(pg. 
462
-
474
)
Kertzman
C
Schwarz
U
Zeffiro
TA
Hallett
M
The role of posterior parietal cortex in visually guided movements in humans
Exp Brain Res
 , 
1997
, vol. 
114
 
1
(pg. 
170
-
183
)
Kornhuber
HH
Deecke
L
Hirnpotentialänderungen bei Willkürbewegungen und passiven Bewegungen des Menschen: Bereitschaftspotential und reafferente Potentiale
Pflügers Arch Gesamte Physiol Menschen Tiere
 , 
1965
, vol. 
284
 (pg. 
1
-
17
)
Lund
TE
Madsen
KH
Sidaros
K
Luo
W-L
Nichols
TE
Non-white noise in fMRI: does modelling have an impact?
Neuroimage
 , 
2006
, vol. 
29
 
1
(pg. 
54
-
66
)
Lund
TE
Nørgaard
MD
Rostrup
E
Rowe
JB
Paulson
OB
Motion or activity: their role in intra- and inter-subject variation in fMRI
Neuroimage
 , 
2005
, vol. 
26
 
3
(pg. 
960
-
964
)
MacDonald
PA
Paus
T
The role of parietal cortex in awareness of self-generated movements: a transcranial magnetic stimulation study
Cereb Cortex
 , 
2003
, vol. 
13
 
9
(pg. 
962
-
967
)
MacIntosh
BJ
Mraz
R
Baker
N
Tam
F
Staines
WR
Graham
SJ
Optimizing the experimental design for ankle dorsiflexion fMRI
Neuroimage
 , 
2004
, vol. 
22
 
4
(pg. 
1619
-
1627
)
Mima
T
Sadato
N
Yazawa
S
Hanakawa
T
Fukuyama
H
Yonekura
Y
Shibasaki
H
Brain structures related to active and passive finger movements in man
Brain
 , 
1999
, vol. 
122
 
10
(pg. 
1989
-
1997
)
Nichols
T
Brett
M
Anderson
J
Wager
T
Poline
J-B
Valid conjunction inference with the minimum statistic
Neuroimage
 , 
2005
, vol. 
25
 
3
(pg. 
653
-
660
)
Nielsen
JB
How we walk: central control of muscle activity during human walking
Neuroscientist
 , 
2003
, vol. 
9
 
3
(pg. 
195
-
204
)
Prablanc
C
Desmurget
M
Grea
H
Neural control of on-line guidance of hand reaching movements
Prog Brain Res
 , 
2003
, vol. 
142
 (pg. 
155
-
170
)
Reddy
H
Floyer
A
Donaghy
M
Matthews
PM
Altered cortical activation with finger movement after peripheral denervation: comparison of active and passive tasks
Exp Brain Res
 , 
2001
, vol. 
138
 
4
(pg. 
484
-
491
)
Reynolds
RF
Day
BL
Visual guidance of the human foot during a step
J Physiol
 , 
2005
, vol. 
569
 (pg. 
677
-
684
Pt 2
Rowe
J
Friston
K
Frackowiak
R
Passingham
R
Attention to action: specific modulation of corticocortical interactions in humans
Neuroimage
 , 
2002
, vol. 
17
 
2
(pg. 
988
-
998
)
Sirigu
A
Daprati
E
Pradat-Diehl
P
Franck
N
Jeannerod
M
Perception of self-generated movement following left parietal lesion
Brain
 , 
1999
, vol. 
122
 
10
(pg. 
1867
-
1874
)
Sperry
R
Neural basis of the spontaneous optokinetic response produced by visual inversion
J Comp Physiol Psych
 , 
1950
, vol. 
43
 
6
(pg. 
482
-
489
)
Stein
JF
Representation of egocentric space in the posterior parietal cortex
Q J Exp Physiol
 , 
1989
, vol. 
74
 
5
(pg. 
583
-
606
)
Stein
JF
Glickstein
M
Role of the cerebellum in visual guidance of movement
Physiol Rev
 , 
1992
, vol. 
72
 
4
(pg. 
967
-
1017
)
van Duinen
H
Zijdewind
I
Hoogudin
H
Maurits
N
Surface EMG measurements during fMRI at 3T: accurate EMG recordings after artifact correction
Neuroimage
 , 
2005
, vol. 
27
 
1
(pg. 
240
-
246
)
von Holst
E
Mittelstaedt
H
Das Reafferenzprincip. (Wechselwirkungen zwischen Zentralnervensystem und Peripherie)
Naturwissenschaften
 , 
1950
, vol. 
37
 (pg. 
464
-
476
)
Wolpert
DM
Kawato
M
Multiple paired forward and inverse models for motor control
Neural Netw
 , 
1998
, vol. 
11
 (pg. 
1026
-
1034
)