Cerebral Cortex August 2007;17:1906--1917
doi:10.1093/cercor/bhl101
Advance Access publication October 23, 2006
Watching Your Foot Move—An fMRI Study
of Visuomotor Interactions during Foot
Movement
Mark Schram Christensen1,2, Jesper Lundbye-Jensen1,3,
Nicolas Petersen1,3, Svend Sparre Geertsen1,3, Olaf B. Paulson2
and Jens Bo Nielsen1,3
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.
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
Keywords: cerebellum, externally generated, posterior parietal cortex,
self-generated
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
Ó The Author 2006. Published by Oxford University Press. All rights reserved.
For permissions, please e-mail: [email protected]
1
Institute of Exercise and Sport Sciences, University of
Copenhagen, Nørre Allé 51, DK-2200 København N, Denmark,
2
Danish Research Centre for Magnetic Resonance,
Copenhagen University Hospital Hvidovre, Kettegård Allé 30,
DK-2650 Hvidovre, Denmark and 3Department of Medical
Physiology, University of Copenhagen, Faculty of Health
Sciences, Blegdamsvej 3, DK-2200 København N, Denmark
model, and a controller, that is, an inverse model, plus a contextdependent 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 selfand 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 3 64 matrix with an in-plane resolution of 3 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 3 256 3 256
acquired matrix, with a resolution of 1 3 1 3 1 mm3.
Physiological Monitoring
The subjects’ pulse and respiration were measured using an MRcompatible 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 (20003), 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.
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
Cerebral Cortex August 2007, V 17 N 8 1907
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 3 2 factorial design, where the 2 factors were movement type and visual feedback type, that is, (self-generated movement,
externally generated movement) 3 (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.
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.
1908 Visuomotor Interactions during Foot Movements
d
Christensen et al.
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), x 2(t ), y 2(t ), z 2(t ), Rx 2(t ), Ry 2(t ), Rz 2(t ), x 2(t – 1),
y 2(t – 1), z 2(t – 1), Rx 2(t – 1), Ry 2(t – 1), Rz 2(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 twoby-two factorial design with the factors (self-generated movement,
externally generated movement) 3 (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 selfgenerated movement as compared with the activation evoked by selfgenerated 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).
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.
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
Cerebral Cortex August 2007, V 17 N 8 1909
Table 1
Self-generated modulation of visual feedback
Movement type
Anatomical region
MNI coordinates
Z-score
(SF [ SN) [ (EF [ EN)
Superior parietal gyrus
21,
12,
18,
15,
24,
15,
39,
42,
30,
27,
48,
45,
39,
33,
3,
21,
36,
21,
24,
57,
54,
54,
63,
36,
27,
30,
21,
27,
18,
21,
48,
42,
42,
51,
18,
15,
12,
3,
5.19
4.02
4.47
3.82
3.74
3.50
4.93
5.18
4.39
3.65
4.03
3.71
3.37
3.92
4.62
4.47
3.12
3.04
3.26
3.46
3.15
3.27
5.16
4.16
5.38
5.24
3.79
4.99
3.42
5.88
6.71
5.77
4.17
5.92
4.21
4.72
3.41
2.86
Superior occipital gyrus
MOG
Angular gyrus
MTG
SF [ EF ^ SN [ EN
SF [ SN ^ EF [ EN
Cerebellum (crus 1)
Cerebellum (VI)
Cerebellar vermis (9)
Cerebellum (VI)
Short insular gyrus
SFG
Superior frontal sulcus
Precentral gyrus
Fusiform gyrus
Intraparietal sulcus
Angular gyrus
MOG
Lateral occipital sulcus
Fourth occipital gyrus
Cerebellum (VIII)
Superior colliculus
63, 63
78, 54
66, 60
57, 69
60, 69
84, 42
87, 24
84, 24
90, 33
87, 36
78, 15
87,
6
45, 33
39, 30
63, 33
66, 21
21, 12
15, 63
12, 57
3, 39
6, 36
21, 36
18, 39
66, 18
48, 54
51, 57
75, 36
75, 36
84, 36
90, 27
75,
0
81,
3
66,
6
63,
3
90, 15
81, 12
69, 48
27, 9
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).
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.
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).
With > Without Visual Feedback
The conjunction SF > SN ^ EF > EN showed increased BOLD
signal within large parts of the dorsal visual pathway (superior
1910 Visuomotor Interactions during Foot Movements
d
Christensen et al.
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 GRFcorrected 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.
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.
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 selfgenerated movements.
Discussion
The main findings of the study are that the PPC showed
increases in the BOLD signal, when visual feedback occurred
Cerebral Cortex August 2007, V 17 N 8 1911
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 selfgenerated 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.
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
1912 Visuomotor Interactions during Foot Movements
d
Christensen et al.
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
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.
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
Cerebral Cortex August 2007, V 17 N 8 1913
Table 2
Conjunction of main effects
Movement type
Anatomical region
MNI coordinates
SF ^ SN
SFG
Frontal operculum
Precentral gyrus
Postcentral gyrus
0,
54,
6,
12,
66,
45,
15,
15,
0,
30,
6,
6,
48,
54,
12,
6,
45,
42,
24,
27,
15,
0,
54,
6,
57,
12,
63,
45,
45,
45,
48,
15,
27,
30,
12,
54,
6,
6,
12,
45,
45,
45,
45,
60,
15,
EF ^ EN
Lateral fissure--parietal operculum
Calcarine sulcus
Cerebellum (middle cerebellar peduncle)
Cerebellar vermis (1/2)
Cerebellum (V1)
Cerebellum (IX)
Precentral gyrus
Postcentral gyrus
Cingulate gyrus
Lateral fissure--parietal operculum
Putamen
SF ^ EF
Cerebellum (middle cerebellar peduncle)
SFG
Frontal operculum
Precentral gyrus
Postcentral gyrus
Supramarginal gyrus
Lateral fissure--parietal operculum
MOG
Cerebellum (middle cerebellar peduncle)
Cerebellum (VI)
SN ^ EN
Cerebellum (VIII)
Frontal operculum
Precentral gyrus
Postcentral gyrus
Short insular gyrus
Lateral fissure--parietal operculum
Lateral fissure
Cerebellum (middle cerebellar peduncle)
6, 54
12,
0
30, 69
36, 78
21, 27
33, 21
78,
3
33, 30
45, 18
66, 24
63, 48
30, 66
0, 3
9,
0
39, 72
9, 48
33, 21
27, 18
0,
9
0,
6
33, 30
6, 54
12,
0
30, 69
3, 39
36, 78
21, 42
30, 36
33, 21
81,
0
75,
0
33, 30
69, 24
75, 18
69, 48
9,
3
30, 69
12, 72
39, 72
3,
0
6,
3
33, 21
27, 21
30, 24
33, 30
Z-score
4.86
4.95
7.29
7.11
5.03
4.74
4.77
7.30
6.12
6.00
4.93
8.00
5.37
5.20
7.94
6.74
7.64
7.56
5.06
5.87
7.12
4.86
4.95
7.29
4.51
7.10
5.95
4.91
4.74
5.00
5.98
7.28
3.84
2.99
4.10
5.08
7.50
6.67
7.63
4.90
4.85
5.86
5.75
5.57
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).
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.
1914 Visuomotor Interactions during Foot Movements
d
Christensen et al.
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 mediallateral 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 selfgenerated 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 offline (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
Cerebral Cortex August 2007, V 17 N 8 1915
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 selfgenerated 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 selfgenerated body movements.
Supplementary Material
Supplementary material
oxfordjournals.org/.
can
be
found
at:http://www.cercor.
Notes
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.
Address correspondence to email: [email protected].
References
Andersen RA, Essick GK, Siegel RM. 1985. Encoding of spatial location by
posterior parietal neurons. Science. 230(4724):456--458.
Angel RW. 1976. Efference copy in the control of movement. Neurology.
26(12):1164--1168.
Blakemore S-J, Sirigu A. 2003. Action prediction in the cerebellum and in
the parietal lobe. Exp Brain Res. 153(2):239--245.
Blakemore S-J, Wolpert DM, Frith CD. 1998. Central cancellation of selfproduced tickle sensation. Nat Neurosci. 1(7):635--640.
1916 Visuomotor Interactions during Foot Movements
d
Christensen et al.
Christensen LO, Johannsen P, Sinkjær T, Petersen N, Pyndt HS, Nielsen
JB. 2000. Cerebral activation during bicycle movements in man. Exp
Brain Res. 135(1):66--72.
Chronicle EP, Glover J. 2003. A ticklish question: Does magnetic
stimulation of the primary motor cortex give rise to an ‘‘efference
copy.’’ Cortex. 39:105--110.
Ciccarelli O, Toosy AT, Marsden JF, Wheeler-Kingshott CM, Sahyoun C,
Matthews PM, Miller DH, Thompson AJ. 2005. Identifying brain
regions for integrative sensorimotor processing with ankle movements. Exp Brain Res. 166(1):31--41.
Clower DM, Hoffman JM, Votaw JR, Faber TL, Woods RP, Alexander GE.
1996. Role of posterior parietal cortex in the recalibration of visually
guided reaching. Nature. 383(6601):618--621.
Clower DM, West RA, Lynch JC, Strick PL. 2001. The inferior parietal
lobule is the target of output from the superior colliculus, hippocampus, and cerebellum. J Neurosci. 21(16):6283--6291.
Cohen YE, Andersen RA. 2002. A common refrence frame for movement
plans in the posterior parietal cortex. Nat Rev Neurosci. 3(7):553--562.
Culham JC, Danckert SL, DeSouza JF, Gati JS, Menon RS, Goodale MA.
2003. Visually guided grasping produces fMRI activation in dorsal but
not ventral stream brain areas. Exp Brain Res. 153(2):180--189.
Della-Maggiore V, Malfait N, Ostry DJ, Paus T. 2004. Stimulation of the
posterior parietal cortex interferes with arm trajectory adjustments
during the learning of new dynamics. J Neurosci. 24(44):9971--9976.
Desmurget M, Epstein CM, Turner RS, Prablanc C, Alexander GE, Grafton
ST. 1999. Role of the posterior parietal cortex in updating reaching
movements to a visual target. Nat Neurosci. 2(6):563--567.
Desmurget M, Grafton S. 2000. Forward modeling allows feedback control for fast reaching movements. Trends Cogn Sci. 4(11):423--431.
Dobkin BH, Firestine A, West M, Saremi K, Woods R. 2004. Ankle
dorsiflexion as an fMRI paradigm to assay motor control for walking
during rehabilitation. Neuroimage. 23(1):370--381.
Duvernoy HM. 1999. The human brain: surface, three-dimensional
sectional anatomy with MRI, and blood supply. Berlin (Germany):
Springer Verlag.
Ehrsson HH, Fagergren E, Forssberg H. 2001. Differential fronto-parietal
activation depending on force used in a precision grip task: an fMRI
study. J Neurophysiol. 85(6):2613--2623.
Eickhoff SB, Stephan KE, Mohlberg H, Grefkes C, Fink GR, Amunts K,
Zilles K. 2005. A new SPM toolbox for combining probabilistic
cytoarchitectonic maps and functional imaging data. Neuroimage.
25:1325--1335.
Ellaway PH, Prochazka A, Chan M, Gauthier MJ. 2004. The sense of
movement elicited by transcranial magnetic stimulation in humans is
due to sensory feedback. J Physiol. 556(2):651--660.
Eskandar EN, Assad JA. 1999. Dissociation of visual, motor and predictive
signals in parietal cortex during visual guidance. Nat Neurosci.
2(1):88--93.
Fogassi L, Luppino G. 2005. Motor functions of the parietal lobe. Curr
Opin Neurobiol. 15(6):626--631.
Friston KJ, Penny WD, Glaser DE. 2005. Conjunction revisited. Neuroimage. 25(3):661--667.
Friston KJ, Williams S, Howard R, Frackowiak RS, Turner R. 1996.
Movement-related effects in fMRI time-series. Magn Reson Med.
35(3):346--355.
Genovese CR, Lazar NA, Nichols T. 2002. Thresholding of statistical
maps in functional neuroimaging using the false discovery rate.
Neuroimage. 15(4):870--878.
Glaser D, Friston K. 2004. Chapter 39: variance components. In:
Frackowiak RSJ, Friston KJ, Frith CD, Dolan RJ, Price CJ, Zeki S,
Ashburner J, Penny W, editors. Human brain function. 2nd ed. San
Diego (CA): Elsevier. p. 781--791.
Glover GH, Li T-Q, Ress D. 2000. Image-based method for retrospective
correction of physiological motion effects in fMRI: RETROICOR.
Magn Reson Med. 44(1):162--167.
Grefkes C, Ritzl A, Zilles K, Fink GR. 2004. Human medial intraparietal
cortex subserves visuomotor transformation. Neuroimage. 23(4):
1494--1506.
Haggard P, Whitford B. 2004. Supplementary motor area provides an
efferent signal for sensory suppression. Brain Res Cogn Brain Res.
19(1):52--58.
Haruno M, Wolpert DM, Kawato M. 2001. MOSAIC Model for Sensorimotor Learning and Control. Neural Comp. 13:2201--2220.
Iacoboni M. Forthcoming 2006. Visuo-motor integration and control in
the human posterior parietal cortex: evidence from TMS and fMRI.
Neuropsychologia.
Imamizu H, Kuroda T, Mizauchi S, Yoshioka T, Kawato M. 2003. Modular
organization of internal models of tools in the human cerebellum.
Proc Natl Acad Sci USA. 100:5461--5466.
Imamizu H, Kuroda T, Yoshioka T, Kawato M. 2004. Functional magnetic
resonance imaging examination of two modular architectures
for switching multiple internal models. J Neurosci. 24(5):1173--1181.
Jeannerod M. 1988. The neural and behavioural organization of goaldirected movements. Oxford (UK): Oxford University Press.
Johannsen P, Christensen LO, Sinkjær T, Nielsen JB. 2001. Cerebral
functional anatomy of voluntary contractions of ankle muscles in
man. J Physiol. 535(2):397--406.
Jueptner M, Jenkins IH, Brooks DJ, Frackowiak RS, Passingham RE. 1996.
The sensory guidance of movement: a comparison of the cerebellum
and basal ganglia. Exp Brain Res. 112(3):462--474.
Kertzman C, Schwarz U, Zeffiro TA, Hallett M. 1997. The role of
posterior parietal cortex in visually guided movements in humans.
Exp Brain Res. 114(1):170--183.
Kornhuber HH, Deecke L. 1965. Hirnpotentialänderungen bei
Willkürbewegungen und passiven Bewegungen des Menschen:
Bereitschaftspotential und reafferente Potentiale. Pflügers Arch
Gesamte Physiol Menschen Tiere. 284:1--17.
Lund TE, Madsen KH, Sidaros K, Luo W-L, Nichols TE. 2006. Non-white
noise in fMRI: does modelling have an impact? Neuroimage. 29(1):
54--66.
Lund TE, Nørgaard MD, Rostrup E, Rowe JB, Paulson OB. 2005. Motion or
activity: their role in intra- and inter-subject variation in fMRI.
Neuroimage. 26(3):960--964.
MacDonald PA, Paus T. 2003. The role of parietal cortex in awareness of
self-generated movements: a transcranial magnetic stimulation
study. Cereb Cortex. 13(9):962--967.
MacIntosh BJ, Mraz R, Baker N, Tam F, Staines WR, Graham SJ. 2004.
Optimizing the experimental design for ankle dorsiflexion fMRI.
Neuroimage. 22(4):1619--1627.
Mima T, Sadato N, Yazawa S, Hanakawa T, Fukuyama H, Yonekura Y,
Shibasaki H. 1999. Brain structures related to active and passive
finger movements in man. Brain. 122(10):1989--1997.
Nichols T, Brett M, Anderson J, Wager T, Poline J-B. 2005. Valid
conjunction inference with the minimum statistic. Neuroimage.
25(3):653--660.
Nielsen JB. 2003. How we walk: central control of muscle activity during
human walking. Neuroscientist. 9(3):195--204.
Prablanc C, Desmurget M, Grea H. 2003. Neural control of on-line
guidance of hand reaching movements. Prog Brain Res. 142:155-170.
Reddy H, Floyer A, Donaghy M, Matthews PM. 2001. Altered cortical
activation with finger movement after peripheral denervation:
comparison of active and passive tasks. Exp Brain Res. 138(4):
484--491.
Reynolds RF, Day BL. 2005. Visual guidance of the human foot during
a step. J Physiol. 569(Pt 2):677--684.
Rowe J, Friston K, Frackowiak R, Passingham R. 2002. Attention to
action: specific modulation of corticocortical interactions in humans. Neuroimage. 17(2):988--998.
Sirigu A, Daprati E, Pradat-Diehl P, Franck N, Jeannerod M. 1999.
Perception of self-generated movement following left parietal lesion.
Brain. 122(10):1867--1874.
Sperry R. 1950. Neural basis of the spontaneous optokinetic response produced by visual inversion. J Comp Physiol Psych. 43(6):
482--489.
Stein JF. 1989. Representation of egocentric space in the posterior
parietal cortex. Q J Exp Physiol. 74(5):583--606.
Stein JF, Glickstein M. 1992. Role of the cerebellum in visual guidance of
movement. Physiol Rev. 72(4):967--1017.
van Duinen H, Zijdewind I, Hoogudin H, Maurits N. 2005. Surface EMG
measurements during fMRI at 3T: accurate EMG recordings after
artifact correction. Neuroimage. 27(1):240--246.
von Holst E, Mittelstaedt H. 1950. Das Reafferenzprincip. (Wechselwirkungen zwischen Zentralnervensystem und Peripherie). Naturwissenschaften. 37:464--476.
Wolpert DM, Kawato M. 1998. Multiple paired forward and inverse
models for motor control. Neural Netw 11:1026--1034.
Cerebral Cortex August 2007, V 17 N 8 1917