Cortical kinematic processing of executed and observed goal-directed hand actions

    Cortical kinematic processing of executed and observed goal-directed hand actions Brice Marty, Mathieu Bourguignon, Veikko Jousm¨aki, Vincent Wens, Marc Op de Beeck, Patrick Van Bogaert, Serge Goldman, Riitta Hari, Xavier De Ti`ege PII: DOI: Reference:

S1053-8119(15)00569-8 doi: 10.1016/j.neuroimage.2015.06.064 YNIMG 12361

To appear in:

NeuroImage

Received date: Accepted date:

28 January 2015 23 June 2015

Please cite this article as: Marty, Brice, Bourguignon, Mathieu, Jousm¨ aki, Veikko, Wens, Vincent, Op de Beeck, Marc, Van Bogaert, Patrick, Goldman, Serge, Hari, Riitta, De Ti`ege, Xavier, Cortical kinematic processing of executed and observed goal-directed hand actions, NeuroImage (2015), doi: 10.1016/j.neuroimage.2015.06.064

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Title: Cortical kinematic processing of executed and observed goal-directed hand

PT

actions

SC RI

Authors: Brice Marty1*, Mathieu Bourguignon1,2*, Veikko Jousmäki1,2, Vincent Wens1, Marc Op de Beeck1, Patrick Van Bogaert1, Serge Goldman1, Riitta Hari2 and

Laboratoire de Cartographie fonctionnelle du Cerveau, UNI – ULB Neuroscience

MA

1

NU

Xavier De Tiège1

Institute, Université libre de Bruxelles (ULB), B-1070 Brussels, Belgium. Brain Research Unit, Department of Neuroscience and Biomedical Engineering,

ED

2

PT

Aalto University School of Science, FI-00076 AALTO, Espoo, Finland.

CE

*These authors equally contributed to this study.

AC

Corresponding author: Brice Marty, Laboratoire de Cartographie fonctionnelle du Cerveau, UNI – ULB Neuroscience Institute, 808 route de Lennik, 1070 Brussels, Belgium. Telephone: +3225556602, Fax: +3225556631, E-mail: [email protected]

ACCEPTED MANUSCRIPT Abstract Motor information conveyed by viewing the kinematics of an agent’s action

is

processed

in

the

brain

remains

to

be

PT

helps to predict how the action will unfold. Still, how observed movement kinematics clarified.

Here,

we

used

SC RI

magnetoencephalography (MEG) to determine at which frequency and where in the brain, the neural activity is coupled with the kinematics of executed and observed motor actions.

NU

Whole-scalp MEG signals were recorded from 11 right-handed healthy adults

MA

while they were executing (Self) or observing (Other) similar goal-directed hand actions performed by an actor placed in front of them. Actions consisted of pinching

ED

with the right hand green foam-made pieces mixed in a heap with pieces of other colors placed on a table, and put them in a plastic pot on the right side of the heap.

PT

Subjects’ and actor’s forefinger movements were monitored with an accelerometer. The coherence between movement acceleration and MEG signals was computed at

CE

the sensor level. Then, cortical sources coherent with movement acceleration were

AC

identified with Dynamic Imaging of Coherent Sources. Statistically significant sensor-level coherence peaked at the movement frequency (F0) and its first harmonic (F1) in both movement conditions. Apart from visual cortices, statistically significant local maxima of coherence were observed in the right posterior superior temporal gyrus (F0), bilateral superior parietal lobule (F0 or F1) and primary sensorimotor cortex (F0 or F1) in both movement conditions. These results suggest that observing others’ actions engages the viewer’s brain in a similar kinematics-related manner as during own action execution. These findings bring new insights into how human brain activity covaries with essential features of observed movements of others.

ACCEPTED MANUSCRIPT Key

words:

cortico-kinematic

coherence,

magnetoencephalography,

AC

CE

PT

ED

MA

NU

SC RI

PT

execution, action observation, mirror neurons.

action

ACCEPTED MANUSCRIPT 1. Introduction During social interaction, people can rely on subtle kinematic cues to infer the

PT

future course of other persons’ motor acts and behavior, for example grasping of a small or a large target, or aiming for social vs. individual, or competitive vs.

SC RI

cooperative behavior (Becchio et al., 2008, 2012; Manera et al., 2010, 2011, 2012; Sartori et al., 2011; Stapel et al., 2012). Also, the dynamics of a motor action conveys crucial information that enables the observer to understand the cognitive or emotional

NU

state of the agent performing the observed action (Rochat et al., 2013). Unraveling

MA

brain areas sensitive to the kinematics of others’ action is therefore of utmost importance to better understand the neural processes involved during action

ED

observation (for a review, see e.g., Kilner, 2011). Functional magnetic resonance imaging (fMRI) has demonstrated that premotor

PT

and parietal areas are tuned to observed motor actions complying to the “two-thirds power law” of motion (i.e. angular velocity is proportional to the two-third power of

CE

path curvature during natural movement), which is a ubiquitous feature of human

AC

motor behavior (Dayan et al., 2007; Casile et al., 2010; Di Dio et al., 2013). These data therefore suggest that these brain areas process some kinematic features of observed motor actions to enable people to understand how actions are actually performed (Di Dio et al., 2013). Unfortunately, the low temporal resolution of fMRI cannot reveal the neural dynamics associated with the fleeting changes in kinematics characterizing human motor actions. We

have

previously

used

magnetoencephalography

(MEG)

and

corticokinematic coherence (CKC) (Bourguignon et al., 2011) to search for significant coupling between the viewer's MEG signals and the kinematics of repetitive non-goal directed hand movements (right hand flexion-extension at about 3

ACCEPTED MANUSCRIPT Hz) performed by an actor placed in front of the viewer (Bourguignon et al., 2012b). In this setting, the viewer’s primary sensorimotor (SM1) cortices were significantly

PT

coupled to kinematics at the actor’s movement frequency (Bourguignon et al., 2012b). This finding supports the existence of a time-sensitive coupling between the activity

SC RI

of these brain areas and the kinematics of observed repetitive non-goal-directed hand movements (Bourguignon et al., 2012b). Although movement observation and execution were not directly compared in that study, the results obtained during action

NU

observation suggest that observing others’ actions engages the viewer’s SM1 cortex in

MA

a similar kinematics-related manner as during own action execution. Indeed, we have previously showed that, during similar movement execution, the bilateral SM1 cortex

ED

is similarly coupled to movement kinematics at movement frequency and its first harmonics (Bourguignon et al., 2011; Bourguignon et al., 2012a).

PT

Since the reactivity of the mirror-neuron system is considered to be strongest when the observed movement is goal-directed (for a review, see Rizzolatti and

CE

Sinigaglia, 2010), we here use time sensitive-MEG recordings to determine, without

AC

any a priori, at which frequency and where in the brain the neural activity is coupled with the kinematics of repetitive goal-directed hand actions, both during movement execution and observation. To that aim, we developed an ecologically valid experimental paradigm where we recorded MEG signals from subjects who were executing or observing similar goal-directed hand actions performed by an actor placed in front of them. We then searched for frequencies and brain areas of statistically significant coupling between movement acceleration and cortical MEG signals.

ACCEPTED MANUSCRIPT 2. Materials and methods 2.1. Participants

PT

Eleven healthy subjects (mean age 31.1 yrs; range 24–40 yrs; 5 males, 6 females) without any history of neuropsychiatric disease or movement disorders were

SC RI

studied. All subjects were right-handed (mean score 94, range 87–100) according to the Edinburgh handedness inventory (Oldfield, 1971). They participated after written informed consent. The study had prior approval by the ULB-Hôpital Erasme Ethics

2.2. Experimental paradigm

MA

NU

Committee and was performed in accordance with the Declaration of Helsinki.

ED

The MEG experiment comprised three conditions (Self, Other, and Rest), each lasting 5 min. The order of the conditions was counterbalanced across subjects.

PT

Figure 1 illustrates the two movement conditions. In the Self condition, a heap of foam-made pieces with different colors (~60

CE

green mixed among ~30 red, purple, orange or yellow; thickness ~4 mm, surface area

AC

~1 cm2) was placed on the table in front of the subject. The subject was asked to pinch the green pieces one-by-one with the right hand, and to move them into a plastic pot placed on the right of the heap. Subjects were asked to perform the repetitive right hand movement at ~1 Hz and to repeat the procedure until the pot was full; the whole process being repeated successively ~4 times during the 5-min condition. The left hand was positioned on the left thigh. To minimize eye movements, subjects were instructed to keep the gaze at the heap during the hand movements. The subjects’ hand actions were monitored with a 3-axis accelerometer (Acc) attached to the nail of the right forefinger. In the Other condition, subjects watched an actor performing the same task. The

ACCEPTED MANUSCRIPT actor sat 1.5 m in front of the subject inside the magnetically shielded room (MSR), hidden by a screen that left visible only the experimenter’s right hand and the heap of

PT

pieces placed on the table next to the pot, just below the screen. The actor was positioned in a face-to-face interaction with the subject. The actor’s right forefinger

SC RI

movements were monitored with an Acc. As in the Self condition, subjects were instructed to gaze at the heap. They were also asked to relax and not to move. During the Rest condition, subjects were instructed to relax, not to move, and to

NU

gaze at a point on the opposite wall of the MSR.

MA

In all conditions, subjects wore earplugs to minimize movement-related auditory contamination during movement conditions.

PT

ED

— Place Figure 1 about here—

2.3. Data acquisition

signals

CE

Neuromagnetic

were

recorded

with

a

whole-scalp-covering

AC

neuromagnetometer placed in a light-weight magnetically shielded room (Vectorview & MaxshieldTM; Elekta Oy, Helsinki, Finland) installed at the ULB-Hôpital Erasme. MEG signals were sampled at 1 kHz and bandpass-filtered at 0.1–330 Hz. Four headtracking coils monitored subjects’ head position inside the MEG helmet. The locations of the coils and at least 150 head-surface points (on scalp, nose, and face) with respect to anatomical fiducials were determined with an electromagnetic tracker (Fastrak, Polhemus, Colchester, VT, USA). Right-forefinger movements (subjects’ forefinger during the Self, actor’s forefinger during the Other) were monitored with an accelerometer (ADXL335 iMEMS Accelerometer, Analog Devices, Inc., Norwood, MA, USA). Surface electromyograms (EMGs) were recorded (with bipolar

ACCEPTED MANUSCRIPT derivations) from the first dorsal interosseous (FDI; ~10–mm inter-electrode distance) and extensor digitorum communis (EDC; ~20–mm inter-electrode distance) muscles

PT

bilaterally as we previously found CKC in SM1 areas of both hemispheres during execution or observation of unimanual movements (Bourguignon et al., 2012b;

SC RI

Bourguignon et al., 2012a). Eye movements and blinks were monitored with vertical and horizontal electrooculograms (EOGs). Electrocardiogam (ECG) was monitored using bipolar electrodes placed below the clavicles. The recording bandpass was 0.1–

NU

330 Hz for EMG, EOG and ECG, and 0–330 Hz for Acc signals; all signals were

MA

sampled at 1 kHz and recorded time-locked to MEG signals using the corresponding built-in amplifiers of the MEG device. High-resolution 3D-T1 cerebral magnetic

ED

resonance images (MRIs) were acquired on a 1.5 T MRI scanner (Intera, Philips, The

PT

Netherlands) at the ULB-Hôpital Erasme.

2.4. Data preprocessing

CE

Continuous MEG data were first preprocessed off-line using signal space

AC

separation method (Taulu et al., 2005) to suppress external interferences and to correct for head movements. Independent component analysis (ICA) was applied to MEG signals filtered through 1–25 Hz, and 2 to 4 components corresponding to eyeblink and heartbeat artifacts were identified using correlation between independent components and EOG and ECG signals, respectively. The corresponding components were subsequently subtracted from raw MEG signals. Whenever the amplitude of MEG signals filtered through 0.1–145 Hz exceeded the threshold set to 3 pT (magnetometers) or 0.7 pT/cm (gradiometers), the corresponding epochs were marked as artifact-contaminated and rejected from further analysis. We have used such artifact rejection approach has been successfully in several previous CKC studies

ACCEPTED MANUSCRIPT (Bourguignon et al. 2011, 2012a, 2012b; Piitulainen et al., 2013a, 2013b). Acceleration was computed at every sample as the Euclidian norm of the three

PT

band-passed accelerometer channels. Finally, MEG and acceleration signals were split into 3-s epochs with 1.5-s overlap, leading to frequency resolution of 0.33 Hz (Bortel

SC RI

and Sovka, 2007).

EMG signals from FDI and EDC muscles were bandpass-filtered at 20–195 Hz

NU

and subsequently rectified.

MA

2.5. Coherence analyses between accelerometer and MEG signals in sensor space The coherence is an extension of Pearson correlation coefficient to the

ED

frequency domain. It quantifies the degree of coupling between two signals x(t) and y(t) by providing a number between 0 (no linear dependency) and 1 (perfect linear

PT

dependency) for each frequency (Halliday et al., 1995). If Xk(f) and Yk(f) are the

K

CE

Fourier transform of the kth segment of x(t) and y(t), by defining power spectra as:

1 å X k ( f )X k* ( f ) K k=1

Pyy ( f ) =

1 åYk ( f )Yk* ( f ) K k=1

AC

Pxx ( f ) =

(1)

K

(2)

and cross-spectrum as

K

1 Pxy ( f ) = å X k ( f )Yk* ( f ) , K k=1

K being the number of artifact-free epochs, the coherence can be written as

(3)

ACCEPTED MANUSCRIPT

2

(4)

Pxx ( f )Pyy ( f )

PT

Cohxy ( f ) =

Pxy ( f )

SC RI

For each movement condition (Self and Other), coherence was computed using formula (4) between Acc and MEG signals. This method allowed for the identification of the frequencies of significant coupling between the signals, without

NU

any a priori hypothesis.

MA

For each movement condition (Self and Other), the frequencies that showed consistent coherence across subjects in sensor space were identified and defined as

ED

the frequencies of interest for subsequent source-level coherence analyses.

PT

2.6. Source level coherence analyses

To perform coherence analysis in source space, individual MRIs were first

CE

segmented using Freesurfer software (Martinos Center for Biomedical Imaging,

AC

Massachusetts, USA). MEG and MRI coordinate systems were then coregistered using the three anatomical fiducial points for initial estimation and the head-surface points to manually refine the surface coregistration. Then, the MEG forward model based on a one-shell boundary element model of the intracranial space was computed for three orthogonal current dipoles placed into a homogeneous 5-mm-grid source space that covered the whole brain (MNE suite; Martinos Center for Biomedical Imaging, Massachusetts, USA). For each source, the forward model was then reduced to its two principal components of highest singular value, which closely correspond to sources tangential to the skull. Coherence maps obtained for each movement condition (Self and Other) and frequency of interest were finally produced using the

ACCEPTED MANUSCRIPT Dynamic Imaging of Coherent Sources (DICS) approach (Gross et al., 2001) with minimum-variance Beamformer derived from broadband (0.1–45 Hz) MEG signals

PT

during movement period (Van Veen et al., 1997). This computationally efficient approach is equivalent to reconstructing virtual-sensor broad-band time-series with

SC RI

beamforming followed by estimating the coherence between these time series and Acc signals at the frequency of interest. Data from planar gradiometers and magnetometers were simultaneously used for source estimation. To do so, sensor

NU

signals (and the corresponding forward model coefficients) were normalized by their

MA

noise root mean square (rms) prior to Beamforming. The noise rms was estimated

ED

from the continuous MEG signals during Rest.

2.7. Group-level analyses in source space

PT

A non-linear transformation from individual MRIs to the standard Montreal Neurological Institute (MNI) brain was first computed using the spatial normalization

CE

algorithm implemented in Statistical Parametric Mapping (SPM8, Wellcome

AC

Department of Cognitive Neurology, London, UK) and then applied to individual MRIs and every coherence map. This procedure generated a normalized coherence map in the MNI space for each subject, movement condition (Self and Other), and frequency of interest. Coherence maps at the group level were then produced with the generalized f-mean of individual normalized maps, according to f (×) = atanh(

(×)) ,

namely the Fisher z-transform of the square root. This procedure transforms the noise on the coherence into an approximately normally distributed noise (Rosenberg et al., 1989). Thus, the computed coherence is an unbiased estimation of the mean coherence at the group level. In addition, this averaging procedure decreases the relative contribution of subjects characterized by high coherence values to the group

ACCEPTED MANUSCRIPT analysis (Bourguignon et al., 2012a).

PT

2.8. Statistical analyses 2.8.1. Coherence at the sensor level

SC RI

We used surrogate-data-based statistics to assess the statistical significance of coherence level in the 0.33–5 Hz frequency range, separately for each subject and each movement condition (Self and Other). This statistical procedure intrinsically

NU

corrects for multiple comparisons across sensors and frequencies. For each subject

MA

and condition, 1000 surrogate coherence spectra were obtained by computing coherence between real MEG signals and Fourier-transform surrogate Acc signals;

ED

the Fourier-transform surrogate imposes power spectrum to remain the same as in the original signal but it replaces the phase of Fourier coefficients by random numbers in

PT

the range [−π, π] (Faes et al., 2004). Then, a single maximum coherence value across all gradiometers in the 0.33–5.0 Hz frequency range was extracted for each surrogate

CE

coherence spectrum. Finally, the 95-percentile of this null distribution for maximum

AC

coherence yielded the coherence threshold at p < 0.05 corrected for multiple comparisons across sensors and frequencies.

2.8.2. Coherence at the group level Statistical significance of the local coherence maxima, identified in each grouplevel coherence map, was assessed with a non-parametric permutation test (Nichols and Holmes, 2002), following the procedure described in (Bourguignon et al., 2012b). This statistical procedure intrinsically corrects for multiple comparisons across all sources.

ACCEPTED MANUSCRIPT In practice, group-level difference maps were obtained by subtracting ftransformed Self or Other with Rest group-level coherence maps. Under the null

PT

hypothesis that coherence maps are the same whatever the experimental condition, the labeling Self or Other and Rest are exchangeable prior to difference map computation

SC RI

(Nichols and Holmes, 2002). To reject this hypothesis and to compute a threshold of statistical significance for the correctly-labeled difference map, the sample distribution of the maximum of the difference map’s absolute value was computed

NU

from the exhaustive permutation set. The 95-percentile of this sample distribution

MA

yielded the threshold at p < 0.05 corrected for multiple comparisons across all sources (Nichols and Holmes, 2002). All suprathreshold local coherence maxima were

ED

interpreted as indicative of brain regions showing statistically significant coherence

PT

with the kinematics of the executed or observed hand movements.

2.8.3. Statistical differences in movement frequency and coherence levels

CE

Differences in movement frequency between conditions (Self and Other), and

AC

the effects of condition and frequency of interest on sensor level coherence were assessed using the Friedman test (non-parametric repeated measures 1-way ANOVA). The effect of condition on the coherence level of brain areas showing significant coherence in both Self and Other was also assessed using the Friedman test. Results with p < 0.05 were considered statistically significant.

2.9. Coherence between accelerometer and EMG signals To control for the viewer’s unintended subtle movements, synchronized with observed movement kinematics during the Other condition, coherence was computed between subjects’ rectified EMG signals (left and right FDI and EDC muscles) and

ACCEPTED MANUSCRIPT subjects’ (Self) or actor’s (Other) Acc signals. Significant EMG–Acc coherence was expected in the Self condition since muscle activity is the driving force of movement

PT

kinematics, whereas no significant EMG–Acc coherence was expected in the Other condition, provided that the subjects did not move themselves while observing the

SC RI

actor’s movements.

The statistical significance of EMG–Acc coherence in the Self and Other conditions was assessed by comparing it with EMG–Acc coherence computed with

NU

original Acc signals and EMG in the Rest condition. We used the Wilcoxon signed

MA

rank test (a non-parametric paired difference test) to check the null hypothesis of absence of difference between original and surrogate EMG–Acc coherence. Results

AC

CE

PT

ED

with p < 0.05 were considered statistically significant.

ACCEPTED MANUSCRIPT 3. Results 3.1. Coherence analyses at the sensor level

spectra obtained in Self and Other in a typical subject.

PT

Figure 1 illustrates the raw accelerometer signals and the corresponding power

SC RI

The mean ± SD movement frequency (F0) across subjects, including the picking and placing of an item, was 0.79 ± 0.22 Hz for Self and 1.00 ± 0.15 Hz for Other; with a trend towards a difference in F0 between the conditions (p = 0.06).

NU

Figure 2 shows the individual coherence spectra between Acc and MEG signals

MA

superimposed for all subjects. In both conditions, statistically significant coherence peaked in 11/11 subjects at the frequency bin closest to F0 and in 10/11 subjects at the

ED

frequency bin closest to its first harmonic F1 (i.e., twice F0). Since coherence systematically peaked at F0 and F1 in line with previous CKC studies (Bourguignon

PT

et al. 2011, 2012a, 2012b, 2015; Piitulainen et al. 2013a, 2013b), these frequencies were defined as the frequencies of interest and used in further analyses. Note that F0

CE

(and F1) might correspond to a different frequency bins in different subjects, but only

AC

values at individual F0 (and F1) were used in further analyses. In the Self condition, the coherence was maximal at central sensors contralateral to hand movements (maximal coherence levels: 0.28 ± 0.13, F0; 0.16 ± 0.09, F1). In the Other condition, the coherence was maximal at parieto-occipital sensors (0.25 ± 0.13, F0; 0.14 ± 0.08, F1). Maximal coherence values did not differ between the Self and Other conditions at F0 (p = 0.51) nor at F1 (p = 0.17) but they were stronger at F0 than F1 in both conditions (Self, p = 0.0031; Other, p = 0.038).

—Place Figure 2 about here—

ACCEPTED MANUSCRIPT

3.2. Coherence analyses at the source level

PT

To identify the neuronal networks underlying CKC observed in the sensor space, similar coherence analyses were performed at F0 and F1 in the source space.

SC RI

Figure 3 illustrates cortical areas that display significant coherence at F0 and F1.

During the Self condition, apart from visual cortices, significant F0 coherence

NU

occurred in the right posterior superior temporal gyrus (pSTG; MNI coordinates of

MA

the local coherence maximum [65 –41 24] mm, level of local maximum coherence 0.08), bilateral SM1 cortex at the anatomical hand area with maximal amplitude over the hemisphere contralateral to hand movements (left SM1, [–54 –16 59], coherence

ED

0.12; right SM1, [45 –16 64], coherence 0.09), and the superior parietal lobules (SPL,

PT

area 7A, [–29 –61 69], coherence: 0.1). Significant F1 coherence was only found in bilateral SM1 cortex at the anatomical hand area, with maximum amplitude in the

CE

hemisphere contralateral to hand movements (left SM1, [–44 –21 64], coherence:

AC

0.13; right SM1, [45 –21 59], coherence: 0.08). During the Other condition, apart from visual cortices, significant F0 coherence was found in the right pSTG ([65 –36 19], coherence: 0.09), right SM1 cortex at the anatomical hand area ([45 –21 64], coherence: 0.10), and SPL bilaterally (left SPL, area 7A, [–29 –66 59], coherence: 0.11; right SPL, area 7A, [15 –66 74], coherence: 0.11). At F1, significant coherence was observed in left SM1 cortex at the anatomical hand area ([–49 –31 64], coherence: 0.05), right medial SM1 cortex ([25 –16 69], coherence: 0.06), and in the right SPL (area 7A, [20 –46 69], coherence: 0.08).

ACCEPTED MANUSCRIPT Coherence level at the pSTG, SPL, and SM1 cortex did not significantly differ between the Self and Other conditions except for the left SM1 cortex at F1 (p =

SC RI

— Place Figure 3 about here—

PT

0.008).

Figure 4 illustrates the brain areas showing significant coherence during both

NU

the Self and Other conditions at F0 or F1. F0 and F1 coherence maps were combined

MA

as the functional significance of the coupling at double movement frequency is still unclear (see Discussion). Practically, for each group-level coherence map, a

ED

significance mask was computed, with value 1 for voxels of significant coherence and 0 otherwise. F0 and F1 masks were combined using a logical OR, and the ensuring

PT

Self and Other masks were combined using a logical AND. This mask revealed that activity within the right pSTG, bilateral SPL and SM1 cortex was coupled with

AC

CE

movement kinematics during both movement execution and observation.

—Place Figure 4 about here—

3.3. Coherence between EMG and accelerometer signals Figure 5 illustrates the EMG–Acc coherence levels obtained for EMG signals in the Self and Other conditions. EMG–Acc coherence was statistically significant during the Self condition (p < 0.01, rank = 0 for all muscles) and non-significant during the Other condition (p > 0.10, rank ≥ 9 for all muscles). Of notice, in one subject, the quality of EMG signals

ACCEPTED MANUSCRIPT from the left EDC muscle was suboptimal during the Other condition and the

PT

corresponding data was not included in the analysis.

AC

CE

PT

ED

MA

NU

SC RI

— Place Figure 5 about here—

ACCEPTED MANUSCRIPT 4. Discussion Using an ecological experimental paradigm, our study demonstrates modulation

PT

of cortical activity within the right pSTG, bilateral SPL, and SM1 cortex, driven by action kinematics during execution and observation of repetitive goal-directed hand

SC RI

movements. The findings suggest that observing others’ action kinematics engages, in the viewer’s brain, neural mechanisms similar to those involved in execution of the same hand actions.

NU

We found significant coupling between the right pSTG activity and the

MA

kinematics of executed and observed goal-directed hand movements at F0. The posterior superior temporal sulcus (pSTS) region is known to be involved in the

ED

processing of observed biological motion, including grasping hand actions (for a review, see Allison et al., 2000). Indeed, single-cell recordings in macaque monkey

PT

pSTS have demonstrated the existence of visual cells sensitive to hand and body actions seen by the monkey; a major part of these cells also react when the monkey is

CE

seeing his own hand during similar self-produced actions (Perrett et al., 1989). These

AC

cells were rather selective to different actions and they responded most strongly for goal-centered actions (Perrett et al., 1989). The role of pSTS in the visual processing of goal-directed hand actions has been later confirmed in humans using functional neuroimaging (Allison et al., 2000). In particular, viewing reaching-to-grasp hand movements induced strong fMRI signals in the lower-bank of the right pSTG, in a cluster located close to the local right pSTG coherence maximum found in Self and Other (Pelphrey et al., 2004). Another fMRI study suggested this area to be involved in a detailed analysis of the observed action, especially its kinematics (Jastorff and Orban, 2009). Of notice, in a previous MEG study investigating the coupling between brain activity and repetitive non-goal directed hand movement kinematics

ACCEPTED MANUSCRIPT (flexion/extension of right-hand fingers), we did not find any significant coherence at pSTS or pSTG (Bourguignon et al., 2012b). Taken together, these data suggest that,

PT

in humans, the right pSTG is involved in time-sensitive encoding of the kinematics of goal-directed actions during both visual guidance of self-executed actions and

SC RI

observation of others’ actions. This “mirroring” of the observed hand kinematics at the right pSTG might represent the initial processing stage of observed kinematics that may be subsequently processed by higher-order areas (Gazzola and Keysers,

NU

2009; Rizzolatti and Sinigaglia, 2010) such as the SPL. Indeed, the SPL that

MA

displayed significant coherence in the Self and Other conditions is known to play a key role in sensorimotor integration (Hyvärinen, 1982; Filimon, 2010).

ED

During the Self condition, we found significant coherence at F0 and F1 between bilateral SM1 cortex activity and hand kinematics. As expected, the strongest

PT

coherence occurred in the hemisphere contralateral to hand movements. Coherence involved both the anterior and the posterior banks of the central sulcus, with no clear

CE

dissociation between the respective contribution of M1 and S1 cortices. These data

AC

are in line with our previous studies using fast (~3 Hz), repetitive, non-goal directed hand/finger movements (Bourguignon et al., 2011, 2012a; Piitulainen et al., 2013a,b). However, they represent an important extension of the previous results to goaldirected hand actions performed at slower pace. Importantly, in those previous studies, subjects were not visually attending to their moving hand during action execution. These findings implicate that the SM1 coherence observed during our Self condition is not driven by the processing of observed kinematics information only (as subjects were seeing their own hand during action execution) but rather by the processing of movement kinematics per se. In this study, the EMG–Acc coherence was statistically significant during Self compared with Rest for all investigated

ACCEPTED MANUSCRIPT muscles, including those of the upper limb contralateral to the moving hand. This finding suggests that ipsilateral SM1 coherence during right hand movements is in

PT

fact related to tiny involuntary rhythmic movements of the left upper limb at the frequency of the right-hand movements.

SC RI

We found rather similar coupling during Self and Other conditions between the SM1 cortex activity bilaterally (F0, right hemisphere; F1, left and right hemisphere) and movement kinematics. However, as a noticeable difference, Self coherence was

NU

stronger in the left hemisphere and Other coherence was stronger in the right

MA

hemisphere. Since movements were performed with the right hand in both Self and Other conditions but observed in a face-to-face perspective in the Other condition,

ED

this discrepancy in right/left hemisphere dominance between conditions might be related to the preference of adult subjects for mirror-image imitation (e.g., the

PT

subject’s left side corresponds to the actor’s right side) over anatomical imitation (e.g., the subject’s right side corresponds to the actor’s right side) during face-to-face

CE

interactions (Avikainen et al., 2003; Chiavarino et al., 2007). Further studies are

AC

required to confirm this interpretation of the present results. A second difference was that the coherence level at the left SM1 cortices was lower when movements were observed rather than executed. This finding was, however, expected since similar weaker recruitment has been previously demonstrated for the mu rhythm modulation (Hari et al., 1998; Caetano et al., 2007) and readiness-potentials-like activity evoked when the timing of observed actions is highly predictable (Kilner et al., 2004). Finally, since the EMG–Acc coherence level did not differ between the Other and Rest conditions in any of the investigated muscles, the coupling identified in Other do not arise from subtle movements performed by the subjects during action observation. Inhibition of some corticospinal

ACCEPTED MANUSCRIPT neurons during movement observation could account for this absence of automatic imitation of the observed action during the Other condition (Vigneswaran et al., 2013;

PT

Hari et al., 2014). We have thus demonstrated that SM1 cortex activity is phasically and

SC RI

bilaterally modulated by the kinematics of both self-executed and observed goaldirected hand actions and that this modulation is seen both in the F0 or F1 phase of SM1 cortex activity. Coupling at frequencies double the movement frequency has

NU

been repeatedly observed in previous movement-execution studies (Pollok et al.,

MA

2004, 2005; Bourguignon et al., 2011, 2012a; Piitulainen et al., 2013a,b) but its functional relevance is still unsettled. In the context of the present study, one may

ED

hypothesize that brain areas showing significant coherence only at F1 during Other are sensitive to particular aspects of observed movement kinematics that depend on

PT

the spectral features of the actor’s hand acceleration (Bourguignon et al., 2012a). On the other hand, the finding could be just an epiphenomenon related to the fact that the

CE

brain rhythms in these brain areas are less sinusoidal compared with the other

AC

coherent brain regions (Pollok et al., 2004; Bourguignon et al., 2012a). For this reason, F0 and F1 were logically combined into a common mask to identify brain areas of significant coupling during both executed and observed goal-directed movements. Further studies should however be designed to shed light on to relative role of the coupling at F0 and F1. Other critical issues triggered by these findings include the underlying neurophysiological mechanisms and the potential similarities or differences between action execution and action observation. With the MEG recordings used in this study, we cannot argue for e.g., activation vs. inhibition of neurons in the involved cortical areas as single-neuron recordings performed in humans and monkeys have shown that

ACCEPTED MANUSCRIPT action observation can activate or inhibit neurons that are active during action execution in, e.g., the M1 and supplementary motor cortex (Mukamel et al., 2010;

PT

Vigneswaran et al., 2013). Still, our findings suggest that some aspects of the processing of movement kinematics taking place in these brain areas during action

SC RI

execution are mirrored during the observation of similar action.

Still one related issue is to determine whether the observed coherence phenomena would pertain to motor or somatosensory mirroring. This question is

NU

particularly relevant since the comparison of active and passive finger movements

MA

suggests that the coupling observed at SM1 during movement execution is in fact mainly driven by somatosensory, primarily proprioceptive, afferent input to SM1

ED

cortex (Piitulainen et al., 2013a; Bouguignon et al., 2015). Primary somatosensory (S1) cortex is also activated when we perceive other people performing an action (for

PT

a review, see Keysers et al., 2010) and the thalamus sends proprioceptive input to both S1 and M1 cortices (Friel et al., 2005). Based on these observations, we may

CE

indeed hypothesize that part of the coupling phenomena observed in the Other

AC

condition are contributing to determining the somatosensory (i.e., proprioceptive) consequences of others’ motor actions (Gazzola and Keysers, 2009). Further studies are needed to clarify this important issue. Finally, we did not find any significant coherence, even at low threshold, in the premotor cortex and in the ventral parietal lobule, two key nodes of the mirror neuron system (for a review, see e.g., (Rizzolatti and Sinigaglia, 2010)). So, our present MEG study does not reproduce the fMRI finding of activity modulation in these brain areas by the observed kinematics of others’ motor actions (Dayan et al., 2007; Casile et al., 2010; Di Dio et al., 2013). Considering that MEG is particularly sensitive to neocortical currents (Goldenholz et al., 2009), the discrepancy between the findings

ACCEPTED MANUSCRIPT obtained with fMRI and with the time-sensitive MEG suggests that the processing of observed action kinematics by these brain areas is not strictly time-locked (Grafton

PT

and Hamilton, 2007; Kilner et al., 2007). In conclusion, we have demonstrated that the observation of others’ hand action

SC RI

kinematics phasically changes the activity of the viewer’s brain in the right pSTG and bilateral SPL and SM1 cortex. These changes mirror those observed during similar action execution. This phasic mirroring driven by action kinematics might represent a

NU

prerequisite for human brain exploitation of visual kinematics of others’ motor action

AC

CE

PT

ED

MA

to understand how observed actions are actually performed.

ACCEPTED MANUSCRIPT 5. Acknowledgments Xavier De Tiège is Postdoctorate Clinical Master Specialist at the FRS-FNRS,

PT

Belgium. This work was supported by a "Brains Back to Brussels" grant to Veikko Jousmäki from the Institut d’Encouragement de la Recherche Scientifique et de

SC RI

l'Innovation de Bruxelles (Brussels, Belgium), European Research Council (Advanced Grant #232946 to Riitta Hari.), the Fonds de la Recherche Scientifique (FRS-FNRS, Belgium, Research Credits: J009713), and the Academy of Finland.

NU

We thank Helge Kainulainen and Ronny Schreiber at the Brain Research Unit

MA

(Aalto University, Finland) for technical support. We also thank Professor Stéphane

AC

CE

PT

ED

Swillens at the Université libre de Bruxelles (ULB) for his support.

ACCEPTED MANUSCRIPT 6. References Allison T., Puce A., McCarthy G. 2000 Social perception from visual cues: role of the

PT

STS region. Trends Cogn Sci 4:267-278. Avikainen S., Wohlschlager A., Liuhanen S., Hanninen R., Hari R. 2003 Impaired

SC RI

mirror-image imitation in Asperger and high-functioning autistic subjects. Curr Biol 13:339-341.

Becchio C., Sartori L., Bulgheroni M., Castiello U. 2008 The case of Dr. Jekyll and

NU

Mr. Hyde: a kinematic study on social intention. Conscious Cogn 17:557-564.

MA

Becchio C., Manera V., Sartori L., Cavallo A., Castiello U. 2012 Grasping intentions: from thought experiments to empirical evidence. Front Hum Neurosci 6:117.

ED

Bortel R., Sovka P. 2007 Approximation of statistical distribution of magnitude squared coherence estimated with segment overlapping. Signal Processing

PT

87:1100-1117.

Bourguignon M., De Tiège X., Op de Beeck M., Pirotte B., Van Bogaert P., Goldman

CE

S., Hari R., Jousmäki V. 2011 Functional motor-cortex mapping using

AC

corticokinematic coherence. Neuroimage 55:1475-1479. Bourguignon M., Jousmäki V., Op de Beeck M., Van Bogaert P., Goldman S., De Tiège X. 2012a Neuronal network coherent with hand kinematics during fast repetitive hand movements. Neuroimage 59:1684-1691. Bourguignon M., De Tiège X., Op de Beeck M., Van Bogaert P., Goldman S., Jousmäki V., Hari R. 2012b Primary motor cortex and cerebellum are coupled with the kinematics of observed hand movements. Neuroimage 66C:500-507. Bourguignon M., De Tiège X., Op de Beeck M., Ligot N., Paquier P., Van Bogaert P., Goldman S., Hari R., Jousmäki V. 2013 The pace of prosodic phrasing

ACCEPTED MANUSCRIPT couples the listener's cortex to the reader's voice. Hum Brain Mapp 34:314326.

PT

Bourguignon M., Piitulainen H., De Tiège X., Jousmäki V., Hari R. 2015

feedback. Neuroimage 106:382-90.

SC RI

Corticokinematic coherence mainly reflects movement-induced proprioceptive

Caetano G., Jousmäki V., Hari R. 2007 Actor's and observer's primary motor cortices stabilize similarly after seen or heard motor actions. Proc Natl Acad Sci U S A

NU

104:9058-9062.

MA

Casile A., Dayan E., Caggiano V., Hendler T., Flash T., Giese M.A. 2010 Neuronal encoding of human kinematic invariants during action observation. Cereb

ED

Cortex 20:1647-1655.

Chiavarino C., Apperly I.A., Humphreys G.W. 2007 Exploring the functional and

PT

anatomical bases of mirror-image and anatomical imitation: the role of the frontal lobes. Neuropsychologia 45:784-795.

CE

Dayan E., Casile A., Levit-Binnun N., Giese M.A., Hendler T., Flash T. 2007 Neural

AC

representations of kinematic laws of motion: evidence for action-perception coupling. Proc Natl Acad Sci U S A 104:20582-20587.

Di Dio C., Di Cesare G., Higuchi S., Roberts N., Vogt S., Rizzolatti G. 2013 The neural correlates of velocity processing during the observation of a biological effector in the parietal and premotor cortex. Neuroimage 64:425-436. Faes L., Pinna G.D., Porta A., Maestri R., Nollo G. 2004 Surrogate data analysis for assessing the significance of the coherence function. IEEE Trans Biomed Eng 51:1156-1166. Filimon F. 2010 Human cortical control of hand movements: parietofrontal networks for reaching, grasping, and pointing. Neuroscientist 16:388-407.

ACCEPTED MANUSCRIPT Friel K.M., Barbay S., Frost S.B., Plautz E.J., Hutchinson D.M., Stowe A.M., Dancause N., Zoubina E.V., Quaney B.M., Nudo R.J. 2005 Dissociation of

PT

sensorimotor deficits after rostral versus caudal lesions in the primary motor cortex hand representation. J Neurophysiol 94:1312-1324.

SC RI

Gazzola V., Keysers C. 2009 The observation and execution of actions share motor and somatosensory voxels in all tested subjects: single-subject analyses of unsmoothed fMRI data. Cereb Cortex 19:1239-1255.

NU

Goldenholz D.M., Ahlfors S.P., Hämälaïnen M.S., Sharon D., Ishitobi M., Vaina

MA

L.M., Stufflebeam S.M. 2009 Mapping the signal-to-noise-ratios of cortical sources in magnetoencephalography and electroencephalography. Hum Brain

ED

Mapp 30:1077-1086.

Grafton S.T., Hamilton A.F. 2007 Evidence for a distributed hierarchy of action

PT

representation in the brain. Hum Mov Sci 26:590-616. Gross, J., Kujala, J., Hämäläinen, M., Timmermann, L., Schnitzler, A., Salmelin, R.,

CE

2001. Dynamic imaging of coherent sources: Studying neural interactions in

AC

the human brain. Proc Natl Acad Sci U S A 98, 694-699. Halliday D.M., Rosenberg J.R., Amjad A.M., Breeze P., Conway B.A., Farmer S.F. 1995 A framework for the analysis of mixed time series/point process data-theory and application to the study of physiological tremor, single motor unit discharges and electromyograms. Prog Biophys Mol Biol 64:237-278. Hari R., Forss N., Avikainen S., Kirveskari E., Salenius S., Rizzolatti G. 1998 Activation of human primary motor cortex during action observation: a neuromagnetic study. Proc Natl Acad Sci U S A 95:15061-15065. Hari R., Bourguignon M., Piitulainen H., Smeds E., De Tiège X., Jousmäki V. 2014 Human primary motor cortex is both activated and stabilized during

ACCEPTED MANUSCRIPT observation of other person's phasic motor actions. Philos Trans R Soc Lond B Biol Sci 369:20130171.

PT

Hyvärinen J. 1982 The parietal cortex of monkey and man. Berlin Heidelberg Springer-Verlag.

SC RI

Jastorff J., Orban G.A. 2009 Human functional magnetic resonance imaging reveals separation and integration of shape and motion cues in biological motion processing. J Neurosci 29:7315-7329.

NU

Keysers C., Kaas J.H., Gazzola V. 2010 Somatosensation in social perception. Nat

MA

Rev Neurosci 11:417-428.

Kilner J.M., Vargas C., Duval S., Blakemore S-J., Sirigu A. 2004 Motor activation

ED

prior to observation of a predicted movement. Nature Neurosci 7, 1299-1301. Kilner J.M., Friston K.J., Frith C.D. 2007 The mirror-neuron system: a Bayesian

PT

perspective. Neuroreport 18:619-623. Kilner J.M. 2011 More than one pathway to action understanding. Trends Cogn Sci

CE

15:352-357.

AC

Manera V., Schouten B., Becchio C., Bara B.G., Verfaillie K. 2010 Inferring intentions from biological motion: a stimulus set of point-light communicative interactions. Behav Res Methods 42:168-178.

Manera V., Becchio C., Cavallo A., Sartori L., Castiello U. 2011 Cooperation or competition? Discriminating between social intentions by observing prehensile movements. Exp Brain Res 211:547-556. Manera V., Cavallo A., Chiavarino C., Schouten B., Verfaillie K., Becchio C. 2012 Are you approaching me? Motor execution influences perceived action orientation. PLoS One 7:e37514.

ACCEPTED MANUSCRIPT Mukamel R., Ekstrom A.D., Kaplan J., Iacoboni M., Fried I. 2010 Single-neuron responses in humans during execution and observation of actions. Curr Biol

PT

20:750-756. Nichols T.E., Holmes A.P. 2002 Nonparametric permutation tests for functional

SC RI

neuroimaging: a primer with examples. Hum Brain Mapp 15:1-25. Oldfield R.C. 1971 The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9:97-113.

NU

Pelphrey K.A., Morris J.P., McCarthy G. 2004 Grasping the intentions of others: the

MA

perceived intentionality of an action influences activity in the superior temporal sulcus during social perception. J Cogn Neurosci 16:1706-1716.

ED

Perrett D.I., Harries M.H., Bevan R., Thomas S., Benson P.J., Mistlin A.J., Chitty A.J., Hietanen J.K., Ortega J.E. 1989 Frameworks of analysis for the neural

PT

representation of animate objects and actions. J Exp Biol 146:87-113. Piitulainen H., Bourguignon M., De Tiège X., Hari R., Jousmäki V. 2013a

CE

Corticokinematic coherence during active and passive finger movements.

AC

Neuroscience 238:361-370. Piitulainen H., Bourguignon M., De Tiège X., Hari R., Jousmäki V. 2013b Coherence between magnetoencephalography and hand-action-related acceleration, force, pressure, and electromyogram. Neuroimage 72:83-90. Pollok B., Gross J., Dirks M., Timmermann L., Schnitzler A. 2004 The cerebral oscillatory network of voluntary tremor. J Physiol 554:871-878. Pollok B., Sudmeyer M., Gross J., Schnitzler A. 2005 The oscillatory network of simple repetitive bimanual movements. Brain Res Cogn Brain Res 25:300311.

ACCEPTED MANUSCRIPT Rizzolatti G., Sinigaglia C. 2010 The functional role of the parieto-frontal mirror circuit: interpretations and misinterpretations. Nat Rev Neurosci 11:264-274.

PT

Rochat M.J., Veroni V., Bruschweiler-Stern N., Pieraccini C., Bonnet-Brilhault F., Barthelemy C., Malvy J., Sinigaglia C., Stern D.N., Rizzolatti G. 2013

SC RI

Impaired vitality form recognition in autism. Neuropsychologia 51:1918-1924. Rosenberg J.R., Amjad A.M., Breeze P., Brillinger D.R., Halliday D.M. 1989 The Fourier approach to the identification of functional coupling between neuronal

NU

spike trains. Prog Biophys Mol Biol 53:1-31.

MA

Sartori L., Becchio C., Castiello U. 2011 Cues to intention: the role of movement information. Cognition 119:242-252.

ED

Stapel J.C., Hunnius S., Bekkering H. 2012 Online prediction of others' actions: the contribution of the target object, action context and movement kinematics.

PT

Psychol Res 76:434-445.

Taulu S., Simola J., Kajola M. 2005 Applications of the Signal Space Separation

CE

Method. IEEE Trans Sign Proc 53:3359-3372.

AC

Van Veen B.D., van Drongelen W., Yuchtman M., Suzuki A. 1997 Localization of brain electrical activity via linearly constrained minimum variance spatial filtering. IEEE Trans Biomed Eng 44:867-880.

Vigneswaran G., Philipp R., Lemon R.N., Kraskov A. 2013 M1 corticospinal mirror neurons and their role in movement suppression during action observation. Curr Biol 23:236-243.

ACCEPTED MANUSCRIPT 7. Legends of the figures Figure 1

PT

Experimental paradigm. Top, left: Self condition. Subject’s neuromagnetic MEG activity is recorded while he is pinching the green pieces, one-by-one, with the right

SC RI

forefinger and thumb, and moving them into a plastic pot placed on the right of the heap. Top, right: Other condition from subject’s point of view. The experimenter performs the same repetitive movement as in the Self condition. Middle: One

NU

movement cycle (reach, grasp, reach, and drop), identical in the Self and Other

MA

conditions. Movement acceleration was recorded with a three-axis accelerometer attached to the right forefinger nail of the subject (Self) or experimenter (Other).

ED

Bottom, left: Five-second sample of acceleration signal from a typical subject and each condition. Bottom, right: Corresponding acceleration power spectra from the

CE

Figure 2

PT

same subject.

AC

Individual coherence spectra for each subject and condition. Each trace represents the coherence between MEG and accelerometer signals for a single subject. For each frequency bin, the coherence value displayed is the maximum coherence across all MEG sensors. Frequencies are expressed in F0 units (i.e., 1 corresponds to the individual movement frequency, 2 to its first harmonics, etc).

Figure 3 Group-level coherence maps superimposed on the surface rendering of the brain. All maps are thresholded at the statistically significant coherence level (lower bound of the color scale, permutation based-statistics) and the brain is viewed from the right

ACCEPTED MANUSCRIPT and top. Left: Self condition at the movement frequency F0 (top) and its first harmonics F1 (bottom). At F0, apart from visual areas, the maximum coherence

PT

occurred at the right pSTG, left SPL, and SM1 bilaterally. At F1, the maximum coherence occurred only at SM1 bilaterally. Right: Other condition at F0 (top) and

SC RI

F1 (bottom). At F0, apart from visual areas, the maximum coherence occurred at the right pSTG, right SM1, and SPL bilaterally. At F1, apart from visual areas, the maximum coherence occurred at the left SM1, right medial SM1 cortex, and right

MA

NU

SPL.

Figure 4

ED

Brain areas showing significant coherence during both Self and Other conditions. Activities within the right pSTG, bilateral SPL, and SM1 are coupled with movement

PT

kinematics during both movement execution and observation. The brain is viewed

AC

Figure 5

CE

from the right and the top.

Mean and standard deviation (across subjects) of coherence between rectified surface EMG and hand acceleration (Acc). f-transformed coherence was computed between rectified EMG and Acc signal for Self and Other at F0. These values were compared to surrogate EMG–Acc coherence obtained from Rest EMG and Self or Other Acc signals. Statistically significant coherence was detected in all muscles in Self only. For illustrative purpose only, Rest EMG-Acc coherence values are here subtracted from Self and Other EMG-Acc coherence. Stars represent significant values of coherence.

ED

MA

NU

SC RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT

Figure 1

SC RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

NU

Figure 2

NU

SC RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

Figure 3

NU

SC RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

Figure 4

NU

SC RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

Figure 5

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

NU

SC RI

PT

Highlights Mirroring of neural kinematic processing in executed and observed goaldirected action. Mirroring occurs in the right posterior superior temporal gyrus. Mirroring occurs in bilateral superior parietal lobule. Mirroring occurs in bilateral primary sensori-motor cortex. This mirroring might help humans to understand how observed actions