can we simulate an action that we temporarily cannot perform?
TRANSCRIPT
Neurophysiologie Clinique/Clinical Neurophysiology (2014) 44, 433—445
Disponible en ligne sur
ScienceDirectwww.sciencedirect.com
ORIGINAL ARTICLE/ ARTICLE ORIGINAL
Can we simulate an action that wetemporarily cannot perform?Pouvons-nous simuler une action que nous sommestemporairement dans l’incapacité d’exécuter ?
C. Calmelsa,∗, S. Pichonb,c, J. Grèzesd,e
a Institut national du sport, de l’expertise et de la performance, département recherche, laboratoire SEP,Paris, Franceb Laboratory for Behavioral Neurology and Imaging of Cognition, Department of Neuroscience, MedicalSchool, University of Geneva, Geneva, Switzerlandc Swiss Center for Affective Sciences, University of Geneva, Geneva, Switzerlandd LNC, INSERM U960, IEC, École Normale Supérieure, Paris, Francee UMR-S975, Inserm U975, CNRS UMR7225, Centre de Neuroimagerie de Recherche — CENIR, Centre deRecherche de l’Institut du Cerveau et de la Moelle épinière, Université Pierre et Marie Curie Paris 6,Groupe Hospitalier Pitié-Salpêtrière, Paris, France
Received 23 September 2013; accepted 6 August 2014Available online 23 August 2014
KEYWORDSAction observationnetwork;Motor injury;Motor expertperformers;fMRI
SummaryAims of the study. — The scope of individuals’ motor repertoire and expertise influences theway they perceive the actions of others. When observing skilled actions, experts recruit thecortical network subserving action perception (action observation network, AON) to a greaterextent than non-experts. However, it remains unknown whether and how a temporary motorinjury affects activation within the AON.Materials and methods. — To investigate this issue, brain hemodynamic activity was recordedtwice in thirteen national female gymnasts suffering from a lower extremity injury at the onsetof the experiment. The gymnasts were scanned one month after the injury and were showngymnastics routines they were able and temporarily unable to perform. Six months later, aftercomplete recovery, they were scanned again and shown the same routines they were now ableto practice.
Results. — Results showed: first, that the level of activity within the inferior parietal lobule andMT/V5/EBA (extrastriate body area), areas constitutive of the AON, was independent of thegymnasts’ physical condition. Second, when gymnasts were hurt (vs. when recovered), higheractivity in the cerebellum was detected.∗ Corresponding author. Service recherche, Institut National du Sport, de l’Expertise et de la Performance, 11, avenue du Tremblay,75012 Paris, France. Tel.: +33 1 41 74 43 73; fax: +33 1 41 74 45 35.
E-mail address: [email protected] (C. Calmels).
http://dx.doi.org/10.1016/j.neucli.2014.08.0040987-7053/© 2014 Elsevier Masson SAS. All rights reserved.
434 C. Calmels et al.
Conclusion. — The equal contribution of MT/V5/EBA and inferior parietal lobule during theobservation of movements the gymnasts were able or unable to practice suggests respectivelythat physical provisional incapacity does not interfere with the perceptual processing of bodyshape and motion information, and that motor expertise may prevent the decay of sensorimotorrepresentations. Higher activations in the cerebellum may suggest that this structure plays arole in dissociating perceived physically feasible movements from those that are provisionallyunfeasible.© 2014 Elsevier Masson SAS. All rights reserved.
MOTS CLÉSRéseau del’observationd’actions ;Blessure ;Experts en motricité ;IRMf
RésuméButs de l’étude. — Le niveau d’expertise motrice des individus influence la manière de percevoirles actions d’autrui. Des individus observant des actions motrices qu’ils maîtrisent parfaitementrecruteront plus intensément le réseau neuronal de l’observation d’actions que des sujets nepossédant pas le bagage moteur leur permettant de réaliser ces actions. Cependant, on ne saitsi l’incapacité temporaire de réaliser certaines actions affecte l’activation de ce réseau et sitel est le cas, de quelle facon.Matériel et méthode. — Pour tenter de répondre à ce questionnement, treize gymnastesexpertes, blessées au membre inférieur au début de l’expérimentation, ont été scannées àdeux reprises. Le premier examen IRM a eu lieu lorsqu’elles étaient blessées. Ces gymnastesont observé des mouvements qu’elles étaient en mesure de réaliser et des mouvements qu’ellesétaient incapables temporairement d’exécuter. Lors du second examen IRM, six mois plus tard,les sujets étaient aptes à effectuer tous les mouvements visionnés.Résultats. — Premièrement, le niveau d’activation dans une partie des régions du réseau del’observation (i.e., lobe pariétal inférieur, MT/V5/EBA) ne dépendait pas de la conditionphysique des gymnastes. Deuxièmement, lorsque les gymnastes étaient blessées (vs. gymnastesrétablies), une activation plus importante au sein du cervelet a été révélée.Conclusion. — L’activation de MT/V5/EBA et du lobe pariétal inférieur, que les gymnastes soientdans l’incapacité ou non de réaliser les mouvements observés, suggère respectivement quecette incapacité n’interfère pas dans le traitement perceptif des informations relatives aumouvement et à la configuration du corps et, d’autre part, que l’expertise motrice préviendraitla dégradation des représentations sensorimotrices. Des activations importantes au sein ducervelet laisseraient à penser que celui-ci serait en mesure de discerner les mouvementsréalisables physiquement des mouvements temporairement infaisables.
. Tou
I
TtgpNeasi[tpi(t
apraT
lventaoeiobpta
[batf
© 2014 Elsevier Masson SAS
ntroduction
hroughout the last decade, it has been establishedhat action perception entails covert motor activity sug-esting that observers use their own motor system toerceive the actions of others (see [41,64,68] for reviews).otably, neuroimaging studies in humans have providedxtensive evidence that a circuit composed of premotornd parietal cortices, primary and secondary somatosen-ory cortices, cerebellum and occipito-temporal areas isnvolved in action observation [9,10,12—14,16,18,19,24,27,]33—35,39,40,50,51,55,69,80,85,87]. This circuit is labeledhe action observation network (AON). Within this circuit,arietal and premotor areas are recognized to be involvedn the prediction of others’ actions before their realizationi.e., prediction of the outcome of others’ actions duringhe viewing of their actions) [1,3,4,32,70—72,86,92].
While the recruitment of motor representations duringction perception has been investigated in both healthyarticipants and experts, it remains unclear whether these
epresentations are altered when observers suffer fromtemporary motor inability to perform specific actions.he goal of the present study was thus to shed some
mio
s droits réservés.
ight on the effects of transient sensory and motor depri-ation on action perception in a population of injuredxpert gymnasts. From the existing literature, two alter-atives can be proposed. On the one hand, it is knownhat motor memory is stable over time and remains rel-tively preserved from aging as compared to other formsf memories [49,77,91]. It is also recognized that expertsngage the AON to a greater extent than non-experts dur-ng action observation [11,12,18,63,66]. From these results,ne can suggest that motor expertise might prevent a possi-le decay of sensorimotor representations during an injuryeriod when an athlete has ceased training. Consequently,emporary injury should have negligible impact on AONctivity.
On the other hand, deficits in peripheral sensations8] and lesions within the motor system [60,61,73] haveeen shown to affect action perception and impinge onction recognition performance. Impact on action recogni-ion may be due to the lack of peripheral somatosensoryeedback, which is essential for accessing, updating, and
aintaining internal motor representations [8]. In a sim-lar way to the structural and functional reorganizationf sensorimotor maps following amputation [75], limb
tcisohfaGkdsmtAimiwugoe
whhtfwPH0
Motor injury and AON
immobilization also affects sensorimotor representationsstructurally: decreased cortical thickness has been reportedin the deprived sensorimotor cortex after only two weekspost-injury [46]. From this perspective, one can suggestthat a prolonged immobilization will compromise accessibil-ity to sensorimotor representations and as a consequence,reduced activity in AON should be detected while athletesobserve movements they are temporarily unable to perform.
Investigating whether motor representations are alteredwhen expert athletes suffer from a transient motor inabilityto perform specific actions is relevant to (sport) rehabili-tation and sport psychology. If alterations are established,interventions using motor imagery or action observationmight be tested with the aim of maintaining motor repre-sentations and compensating for the lack of sensorimotorinput during the period of injury [5,52].
In this study, we set out to experimentally addresswhether and how a temporary motor injury has impact onexpert gymnasts’ activation within the AON. Using fMRI,we recorded brain hemodynamic activity in expert gym-nasts with a lower limb injury while they watched videosof gymnastic routines involving either lower or upper limbs(see Fig. 1). Subjects were scanned at two time points: onemonth after a lower limb injury (hurt session) that causeda temporary inability to perform movements requiring thelower limbs, but not training routines involving the upperlimbs; six months later, when they had fully recovered (healsession).
This sport was chosen because it allows injured gymnaststo continue performing the routines that involve the non-injured body part.
Materials and methods
Subjects
Thirteen expert female gymnasts aged 16—26 years (meanage = 19.9, SD = 3.4) participated. Inclusion criteria were
Figure 1 Snapshots of floor and bar routine stimuli that wereused in the experiment. The red square indicates the maincondition of interest: gymnasts suffering from a lower extrem-ity injury observed floor routines they were temporarily unableto perform.
S
Sniaesjttlp
ptw(owsoptvew
435
hreefold. First, participants took part in the French nationalhampionship each year; second, they performed exercisesnvolving mainly either lower limbs (exercises on floor:tretched backward and forward saltos with one twist)r upper limbs (exercises on uneven bars: kip, cast toandstand, clear hip circle to handstand, and giant swingorwards backwards to handstand); third, they suffered from
lower extremity injury at the onset of the experiment.ymnasts reported injuries in the foot (n = 2), ankle (n = 7) ornee (n = 4) region. The injuries were due to sprain (n = 10),islocation (n = 1), or fracture (n = 2). All injuries were clas-ified as moderate; i.e., they required the two to threeonths rehabilitation before the gymnast was authorized by
he medical staff to gradually resume regular floor routines.pproximately three months later (seven months after the
njury onset), subjects regained their full range of move-ent: they were able to tumble as they did before being
njured and without any discomfort. As a comparison group,e initially planned to recruit a second gymnast group withpper extremity injuries. Nevertheless, only two expertymnasts complied with our inclusion criteria after 2.5 yearsf recruitment procedure. We were therefore constrained toxclude this part of the experimental design.
All subjects had normal vision (i.e., they could seeithout correction) and no past neurological or psychiatricistory. All gave written informed consent and subjects whoad reached the age of majority were paid for their par-icipation. Separate parental consents were also obtainedor the subjects who were under the age of 18. The studyas approved by the local ethics committee (Comité derotection des Personnes d’Ile de France VI, CPP) and theealth Ministry (Direction Générale de la Santé, DGS2006-065).
timuli
ix uneven bar and six floor routines were executed by threeational left-sided gymnasts (see below) who did not partic-pate in the fMRI study. These routines were recorded using
digital camera. The bar and floor routines possess differ-nt characteristics: bar routines are composed of kips, giantwings whereas floor routines involve performing straightumps followed by acrobatic skills at the trampoline; i.e.he former employs predominantly the upper limbs, the lat-er the lower limbs. The performers wore similar dark blueeotards and performed their routines in front of a white andale brown background.
Videos were computer edited and fragments (25 frameser second) were selected from initial materials. The dura-ion of the floor video-clips ranged between 3040—5640 ms,hereas for the uneven bars the range was 3520 to 7240 ms
see Table 1). Gymnasts generally have a preferred rotatingrientation when performing backward and forward saltosith one twist. In the present participants, ten were left-
ided performing backward and forward twists to the left,ne was right-sided, and two were ‘‘left and right-sided’’erforming backward twists to the left and forward twists
o the right. Using the Adobe After-effect software, all theideos were mirrored to match the preferred rotating ori-ntation of each gymnast. As a result, four kinds of packageere edited:
436 C. Calmels et al.
Table 1 Characteristics of the stimuli used in the experimental procedure.
Stimuli of interest Stimuli of non-interest
Dynamic stimuli Static stimuli Null events Oddballs
Bar Floor Bar Floor
36 36 36 36 72 32
Duration Duration Duration Duration
4671 msSD = 1098
4440 msSD = 583
4671 msSD = 1098
4440 msSD = 583
4500 msSD = 0
4490 msSD = 652
ically
•
•
•
•
pbm
irtts
P
TSiv0
•
•
P
Tp(tt[ti
ebsaptp(eTbng
t
•••••
•
icul
rnpa(jd
f
F
SD: standard deviation. Durations of bar routines were not statist
a left package (performing backward and forward twiststo the left);a right package (performing backward and forward twiststo the right);a first mixed package (performing backward twist to theleft and forward twist to the right);a second mixed package (performing backward twist tothe right and forward twist to the left).
Static images depicting the beginning, middle, and endarts of each routine were edited from those floor andar videos and presented with the same duration as theiratched video-clips.Finally, six routines, different from those used in exper-
mental conditions, served as oddball stimuli (durationange: 3440 to 5640 ms). A red target point (1000 ms dura-ion) was added at the center of each frame 1000 ms afterhe onset of the video. Subjects were asked to attend alltimuli and detect the occurrence of the oddball target.
rocedure for stimulus validation
en female competitive gymnasts (mean age = 15.4,D = 0.9), who did not participate in the fMRI study, werenvited to carefully observe the above-mentioned editedideo-clips and evaluate on a graduated scale ranging fromto 10:
whether the action involved the upper (scale 1) or thelower (scale 2) limb and to which extent;whether a lower (scale 3) or an upper (scale 4) extremityinjury would interfere with performing this action and towhich extent.
rocedure for the fMRI study
he experiment was divided into 2 independent sessionser subject within a six-month period. The first sessioni.e., Hurt) took place one month after participants had sus-ained an injury. A one-month period was chosen to ensure
hat cortical changes had already occurred. Langer et al.46] described structural changes (i.e., decreased corticalhickness within sensorimotor cortex) after two weeks ofmmobilization. As the gymnasts were suffering from lowerCpwF
different to those of floor routines; t(34) = .78, P = .44.
xtremity injuries, they were still able to practice on thears, yet they were unable to execute floor routines. Sub-equently, regular verbal assessments with the participantsnd their coach were conducted to monitor the recoveryrocess. When an injured gymnast returned to full activity,he investigator was warned that this gymnast was able toerform her floor routines without pain. The second sessioni.e., Heal) thus took place when gymnasts had fully recov-red i.e. when they were able to return to full floor activity.o sum up, during the first scanning session, gymnasts viewedar routines they could physically perform, and floor routi-es they could not perform. In the second scanning session,ymnasts were able to execute all the routines.
Each session included 124 stimuli that were presentedwice in a random order:
18 dynamic floor routines (6 move × 3 gymnasts);18 dynamic bar routines (6 moves × 3 gymnasts);18 static floor postures;18 static bar postures;16 dynamic floor and bar routine oddballs (4 moves× 3 gymnasts + 2 moves × 2 gymnasts);36 null events (see Table 1).
Null events (black screen) of 4500 ms duration were alsoncluded to provide better estimation efficiency in order toharacterize the shape of the BOLD response at short stim-lus onset asynchrony [36]. Presentation of a black screenasting 1 s followed each stimulus.
To avoid motor contamination by stimulus-relatedesponses and to insure that the subjects attend toon-target stimuli, we used an oddball paradigm wherearticipants observe stimuli of interest without respondingnd react only to sporadic incidences of target stimulimovies of non-interest with a red dot overlaid). The sub-ects conformed to the task since all the oddball stimuli wereetected.
MRI acquisition
unctional images were acquired on a 3T TRIO TIM (Siemens,
ENIR, Pitié-Salpêtrière, France) using a T2*-weighted echo-lanar sequence at 32 interleaved 3 mm-thick axial slicesith 0.99 mm gap (TR = 2080 ms, TE = 30 ms, flip angle = 90◦,OV = 192 mm, 3 × 3 × 3 mm voxels). A total of 680 functional
ttsaiSD
tcd(Atra1s[
R
S
Ritittrtswite8
f
1otH(H[
nH—Ohiw
Motor injury and AON
volumes were acquired for each subject. For each par-ticipant, we then acquired a high-resolution structuralT1-weighted anatomical image (TR = 2300 ms, TE = 4.18 ms,flip angle = 9◦, FOV = 256 mm, 1 × 1 × 1 mm voxels) paral-lel to the AC-PC plane (Anterior Commissure — PosteriorCommissure), covering the whole brain.
fMRI data preprocessing
The data were preprocessed using SPM2 (http://www.fil.ion.ucl.ac.uk/spm). The first 5 volumes were discarded toallow for T1 equilibration effects and the last one wasdropped. The remaining 674 functional images were reori-ented, slice-time corrected to the middle slice and spatiallyrealigned to the first volume. These images were normal-ized (MNI template) and smoothed with an isotropic 8 mmfull-width-half-maximum (FWHM) Gaussian kernel.
fMRI statistical analyses
We performed standard analyses using the general lin-ear model (GLM) as implemented in SPM2. A two-stagegeneral linear model was used. At the subject level,condition-specific effects were modeled separately for eachscanning session (i.e., Hurt and Heal). The following sixconditions were specified: two conditions where partic-ipants perceived dynamic bar and floor routines (BarsDyn, Floor Dyn), two conditions during which static barand floor postures were presented (Bars Stat, Floor Stat),one where participants perceived oddball stimuli, and onecorresponding to motor responses. Null events were implic-itly modeled. Each stimulus onset was convolved with acanonical HRF and the corresponding stimulus duration(3040—7240 ms). Motor responses were also modeled ina separate regressor. Finally, covariates capturing resid-ual movement-related artifacts were included (3 rigid-bodytranslations and 3 rotations determined from initial imageco-registration) as well as a single covariate represent-ing the mean (constant) BOLD signal over the sessions.The model included a high-pass filter with a cut-off periodat 128 s.
We also estimated the efficiency of the model for eachcondition/contrast and subject using the following for-mula: trace (contrast*inv(X’*X)*contrast’)-1 (from http://imaging.mrc-cbu.cam.ac.uk/imaging/DesignEfficiency).Average efficiency ± SD were entered in a three wayrepeated measures ANOVA with the factor ‘‘session’’ (Hurt,Heal), ‘‘characteristic’’ (Dyn, Stat) and ‘‘routine’’ (Bar,Floor). Efficiency did not differ between sessions (P = 0.41)and more importantly there was no significant interactionbetween the factors session and characteristic (P = 0.11)nor routine: (P = 0.38).
Then, at the second level (i.e., group-level), a ran-dom effect model was estimated in the whole brain,implementing a 2 × 2 × 2 repeated measures ANOVA withthe factors ‘‘session’’ (Hurt, Heal), ‘‘characteristic’’ (Dyn,Stat), and ‘‘routine’’ (Bars, Floor). In this way, the variance
estimates at the group-level incorporated appropriatelyweighted within-subject and between-subject varianceeffects. A non-sphericity correction was applied for variancedifferences across conditions.b(pm
437
To constrain the search of changes in activity withinhe AON, an anatomical inclusion mask, which con-ains all the regions of the AON, was specified whenetting up the statistical model. Main effects and inter-ctions were either presented unmasked or maskednclusively by the simple main effect of (Dyn vs.tat) at P = .001 ([(Bars + Floor) Hurt + (Bars + Floor) Heal]yn — [(Bars + Floor) Hurt + (Bars + Floor) Heal] Stat).
We reported activations that survived a voxel-wisehreshold of T > 3.19 (P < .001, uncorrected) with a minimumluster size of 10 voxels [48]. All peaks that survived falseiscovery rate (FDR) and family-wise error (FWE) correctionsP < .05) at the voxel-level were indicated in the table(s).ll statistical maps were overlaid on sections of the Colinemplate brain available in SPM. Surface rendering was car-ied out using Caret [88] and the PALS-B12 atlas [89], anverage brain atlas derived from structural MRI volumes of2 normal young adults that were adjusted to the ICBM-152pace. Brain regions were labeled using the atlas of Duvernoy26].
esults
timulus validation
esults show that floor and bar routines were correctlydentified by control subjects. For the floor routines,he mean score (scale range 0—10) assessing lower limbnvolvement (9.36 ± SEM .02) was significantly higher thanhe one assessing upper limb participation (6.95 ± 0.13;(9) = —8.85, P < .001). The reverse was found for the baroutines (upper limb: 9.56 ± .02 vs. lower limb: 6.26 ± 0.1;(9) = 16.91, P < .001) (see Fig. 2). Moreover, subjects con-idered (interference scores) that lower extremity injuryould substantially compromise performance of floor rout-
nes (0.49 ± .03) as compared to bar routines (7.89 ± 0.13;(9) = —24.63, P < .001). The contrary was observed for upperxtremity injury (bar routines: 0.35 ± .01 vs. floor routines:.66 ± 0.16; t(9) = 25.2, P < .001) (see Fig. 2).
MRI Results
. First, to test whether a temporary injury had impactn the level of activity within AON irrespective ofhe routines, we estimated the main effect Heal vs.urt: [Bars (Dyn + Stat) + Floor (Dyn + Stat)] Heal vs. [Bars
Dyn + Stat) + Floor (Dyn + Stat)] Hurt and the inverse effecturt vs. Heal: [Bars (Dyn + Stat) + Floor (Dyn + Stat)] Hurt vs.
Bars (Dyn + Stat) + Floor (Dyn + Stat)] Heal.While the masked main effect Heal vs. Hurt revealed
o suprathreshold clusters, the unmasked contrast Heal vs.urt disclosed one cluster in the left paracentral lobe (—828 76) that survived voxel-level FWE correction (P < .05).n the other hand, the masked main effect Hurt vs. Healighlighted brain areas displaying a higher activation dur-ng the perception of gymnastics routines when gymnastsere hurt: the right SMA and cerebellum (crus 1, VI) on
oth hemispheres whose peaks survived FDR correctionP < .05) (see Table 2). We ensured that localisation of theseeaks was not accompanied with potential dropouts in theask in neighbouring regions. Examination of the parameter
438 C. Calmels et al.
Figure 2 Stimulus validation. A. Mean scores assessing perceived involvement of upper and lower limbs for floor and bar routines.B. Mean scores assessing perceived interference of lower and upper injuries on floor and bar routines (a lower score reflects inter-f AsteE
6wostnTct
(c
iw
erence whereas a higher score reflect a lack of interference).rror bars correspond to ± 1 SEM.
estimates revealed a greater activation when gymnastsere injured compared to when they recovered irrespectivef the observed routine (see Fig. 3A—D). Interestingly, theame areas of the cerebellum were found to be activated inhe contrast (Floor Dyn Hurt vs. Floor Dyn Heal), but they did
ot survive voxel-level FDR or FWE correction (P < .05) (seeable 3). The unmasked contrast Hurt vs. Heal revealed sixlusters that survived voxel-level FWE correction (P < .05) inhe left postcentral gyrus (BA3), left inferior frontal gyrussa((
Table 2 Brain areas activated during the perception of dynamic sversus when they fully recovered.
Anatomical areas MNI Z
x y z
Hurt vs Heal masked by Dyn vs Stat at P = .001Right SMA (BA6) 10 6 70 3Left cerebellum (crus 1) *a —42a —56a —34a 6Left cerebellum (VI) *a —28a —70a —20a 4Right cerebellum (crus 1) 44 —58 —28 3Right cerebellum (VI) 34 —66 —24 4
SMA: supplementary motor area. P < .001 (uncorrected), cluster size > 1at the voxel-level (P < .05). Asterisk (*) indicates peaks which survivewith ↓.
a Indicate peaks which survived family-wise error (FWE) correction a
Table 3 Brain areas activated during the perception of dynamicwhen they fully recovered.
Anatomical areas MNI Z s
x y z
Floor Dyn Hurt vs. Floor Dyn HealLeft cerebellum (crus 1) —44 —54 —34 4.0Right cerebellum (crus 1) 38 —72 —34 3.4Right cerebellum (VI) 30 —70 —22 4.0Right cerebellum (VII) 32 —68 —44 3.4
P < .001 (uncorrected), cluster size > 10 voxels. None of the above peakcorrection at the voxel-level. Subpeaks in clusters marked with ↓.
risk (*) indicates statistically significant differences (*P < .001).
BA44), cerebellum (left crus 1, right IV—V, right VI) anderebellar vermis (see Table 4).
2. Second, to test whether a temporary injury onlympacted on the perception of the routines the gymnastsere unable to perform (i.e., floor routines), we computed
ession-by-routine interaction. The first interaction aimedt revealing brain areas with higher activity when recoveredvs. when injured): [Floor (Dyn + Stat) Heal — FloorDyn + Stat) Hurt] — [Bars (Dyn + Stat) Heal — Bars (Dyn + Stat)
timuli of bar and floor routines when the gymnasts were hurt
score P valueFWE cor P valueFDR cor Cluster
.83 .258 .006 23
.50a < .001a < .001a 223a
.62a .014a .001a ↓223a
.76 .316 .007 13
.11 .104 .003 10
0 voxels. All peaks survived false discovery rate (FDR) correctiond correction at the cluster level. Subpeaks in clusters marked
t the voxel-level.
stimuli of floor routines when the gymnasts were hurt versus
core P valueFWE cor
P valueFDR cor
Cluster
7 .120 .278 344 .645 .278 ↓627 .120 .278 247 .613 .278 62
s survived false discovery rate (FDR) or family-wise error (FWE)
Motor injury and AON 439
Figure 3 Statistical maps showing brain areas activated during the perception of dynamic stimuli of bar and floor routines whenthe gymnasts were hurt versus when they had fully recovered. Mean group parameter estimates are plotted for dynamic conditionsin the cerebellum (crus 1) (A), the cerebellum (VI) (B), and the supplementary motor area (SMA) (C). (D) Differences of activitybetween the hurt and heal sessions during the observation of bar and floor routines (displayed for each subject and for the group).
in thhurt
icytt(l1
D
Tattrectlt
Parameter estimates were extracted from the corrected peaksAlmost all of the subjects showed increased activity during the
Hurt]. The second interaction aimed at revealing brain areaswith higher activity when injured (vs. when recovered):[Floor (Dyn + Stat) Hurt — Floor (Dyn + Stat) Heal] — [Bars(Dyn + Stat) Hurt — Bars (Dyn + Stat) Heal]. No suprathresh-old clusters were detected whether the contrasts weremasked or not.
3. Third, we only concentrated on (floor) routinesthat gymnasts were unable to perform. We computedthe interaction [(Floor Dyn Heal — Floor Stat Heal) — (FloorDyn Hurt — Floor Stat Hurt)] and the inverse interaction[(Floor Dyn Hurt — Floor Stat Hurt) — (Floor Dyn Heal — FloorStat Heal)]. These interactions investigated whether anybrain areas displayed a greater response for dynamic overstatic stimuli based on the gymnasts’ physical condition.These interactions were unmasked and masked inclusivelyby [(Floor Hurt + Floor Heal) Dyn — (Floor Hurt + Floor Heal)Stat] at P = .001. No suprathreshold clusters were detectedwhether the contrasts were masked or not.
4. Finally, we performed a conjunction analysis (nullhypothesis) to reveal areas that were commonly recruitedduring the observation of bar and floor routines both when
the gymnasts were injured and when they had recovered:[(Floor Dyn Hurt — Floor Stat Hurt) ∩ (Floor Dyn Heal — FloorStat Heal) ∩ (Bars Dyn Hurt — Bars Stat Hurt) ∩ (Bars DynHeal — Bars Stat Heal)]. Note that this type of conjunctioncggo
e left cerebellum crus 1 (—42 —56 —34) and VI (—28 —70 —20).period. S stands for Subject. Error bars correspond to ± 1 SEM.
s equivalent to a logical AND operation as it requires allontrasts to be individually significant [54]. This analysisielded bilateral activations in the supramarginal gyrus andhe visual area MT/V5. Activations were also detected inhe right hemisphere within the supplementary motor areaBA6) and inferior parietal lobule (PF, PGp). Finally, activityocated in the left hemisphere within the cerebellum (crus) was revealed (see Table 5 and Fig. 4).
iscussion
he aim of this study was to determine whether and howtemporary motor inability to perform an action impacted
he level of BOLD activity within AON while observing actionshat physical condition enabled to perform or not. Onlyesults stemming from contrasts masked by the simple mainffect of (Dynamic vs. Static) and surviving voxel-level FWEorrection at P < .05 are discussed. Our results showed thathe activity within motion area MT/V5 and inferior parietalobule was independent of the gymnasts’ physical condi-ion and ability to perform the viewed actions and that the
erebellum displayed higher levels of activity whenymnasts were hurt (vs. when recovered). These results sug-est partial support of the hypothesis of a negligible impactf injury on AON activity.
440 C. Calmels et al.
Table 4 Brain areas activated during the perception of dynamic and static stimuli of bar and floor routines when the gymnastswere hurt versus when they had fully recovered.
Anatomical areas MNI Z score P valueFWE cor
P valueFDR cor
Cluster
x y z
Hurt vs HealRight SMA (BA6) 10 6 70 3.83 .258 .006 28Left precentral gyrus (BA6) —22 —4 62 3.47 .609 .013 10Right precentral gyrus (BA6) 20 —10 66 3.69 .375 .008 126Left postcentral gyrus (BA3a)a —40a —24a 36a 4.48a .026a .001a 35a
Left inferior frontal gyrus (parsopercularis, BA44)a
—56a 14a 18a 4.35a .044a .001a 126a
Right inferior frontal gyrus (parsopercularis, BA44)
62 18 14 4.14 .093 .003 45
Left inferior parietal lobule (BA2) —52 —26 48 3.88 .220 .005 22Left inferior parietal lobule (PGp) 40 —66 30 3.48 .602 .013 27Left cerebellum (crus 1) *a —42a —56a —34a 6.50a < .001a < .001a 377a
Left cerebellum (VI) * —28 —70 —20 4.52 .233 .005 ↓377Left cerebellum (IV—V) —10 —56 —14 3.47 .605 .013 13Right cerebellum (crus 1) *a 38a —70a —32a 4.45a .029a .001a ↓344a
Right cerebellum (VI) *a 30a —70a —22a 5.81a < .001a < .001a 344a
Right cerebellum (VI) 28 —56 —36 3.73 .335 .007 21Right cerebellum (IV—V)a 24a —48a —22a 5.02a .003a < .001a 136a
Cerebellar vermis (4/5)a 2a —50a 2a 4.71a .010a < .001a 35a
SMA: supplementary motor area. P < .001 (uncorrected), cluster size > 10 voxels. All peaks survived false discovery rate (FDR) correctionat the voxel-level (P < .05). Asterisk (*) indicates peaks which survived correction at the cluster level. Subpeaks in clusters markedwith ↓.
ion a
A
WcaI
st
a Indicate peaks which survived family-wise error (FWE) correct
ON activation when injured and when recovered
e confirmed previous findings reporting that the per-eption of actions elicited activity within the MT/V5rea and inferior parietal lobule [13,15,29,34,35,50,85].mportantly, our results extend previous findings by
mo
v
Table 5 Common activations during the observation of bar and flactivations were determined by a conjunction analysis between.
Anatomical areas MNI
x y z
Right supplementary motor area (BA6) 8 —4 74Left supramarginal gyrus —50 —38 28Right supramarginal gyrus*a 58a —28a 26a
Right inferior parietal lobule (PF) *a 62a —40a 22a
Right inferior parietal lobule (PGp) *a 52a —68a 12a
Left middle occipital gyrus (MT/V5)a -42a —72a 8a
Right middle temporal gyrus (MT/V5)a 52a —66a 8a
Left cerebellum (Crus 1) —36 —52 —32
[(Floor Dyn Hurt — Floor Stat Hurt) ∩ (Floor Dyn Heal — Floor Stat HeStat Heal)]. P < .001 (uncorrected), cluster size > 10 voxels. All peaks s(P < .05). Asterisk (*) indicates peaks which survived correction at the c
a Indicate peaks which survived family-wise error (FWE) correction a
t the voxel-level.
howing that these areas are responsive to the percep-ion of human movements regardless of whether these
ovements are physically feasible or unfeasible by thebserver.MT/V5 has traditionally been involved in processing
isual motion [2]. Nevertheless, the recent literature has
oor routines irrespective of subjects’ state of health. These
Z score P valueFWE corFFDR cor
P valueFDR cor.
Cluster
3.73 .307 .007 323.75 .475 .010 864.45a .029a .001a ↓786a
5.41a < .001a < .001a ↓786a
6.17a < .001a < .001a 786a
6.98a < .001a < .001a 67a
6.53a < .001a < .001a 91a
3.77 .307 .007 17
al) ∩ (Bars Dyn Hurt — Bars Stat Hurt) ∩ (Bars Dyn Heal — Barsurvived false discovery rate (FDR) correction at the voxel-levelluster level. Subpeaks in clusters marked with ↓.
t the voxel-level.
Motor injury and AON 441
Figure 4 Statistical maps showing common brain areas recruited both during the hurt and heal sessions when observing floorand bar routines (P < .001, uncorrected). Mean group parameter estimates are shown in all experimental conditions. Activations
us (M
atliem
ecttasmoplbslawFeoaAd
I
were found in (from left to right) the left middle temporal gyrsupramarginal gyrus. Error bars correspond to ± 1 SEM.
consistently pointed toward an overlap between MT/V5 andan area involved in the processing of body parts and shape,the extrastriate body area (EBA). It is now acknowledgedthat MT/V5 is composed of at least 4 functionally distinctregions: the ventral part of the medial superior tempo-ral area (MSTv), the fundus of the superior temporal area(FST), and the V4 transitional zone (V4t) [28,45]. Surpris-ingly, it has been recently shown that body selective voxelsin MT/V5 are more sensitive to body shape informationthan to motion stimuli, whereas body selective voxels inMSTv and FST are both sensitive to body shape and motion[28]. The present coordinates of MT/V5/EBA are close tothose of MT/V5 and MSTv defined by Kolster et al. [45] sug-gesting that the provisional incapacity of the gymnasts toperform the observed movements does not interfere withthe perceptual processing of body shape and motion infor-mation.
The inferior parietal lobule was also found to be activatedduring both sessions (i.e., Heal and Hurt). Electrophysiolog-ical recordings in the inferior parietal cortex of monkeysshowed that this area is involved in action preparation[42] and in extracting potential motor representations inresponse to visual stimuli or ‘‘motor intentions’’ [78,79].Preparatory activity has also been observed in the ante-rior and the posterior inferior parietal lobule of the humanbrain: this activity is related to motor intention since itoccurs even in trials in which no movement is executed [83].This finding is in line with Desmurget et al. [21,22] propo-sition that the posterior parietal cortex (i.e., angular andsupramarginal gyri; BA39, BA40) is involved in the generation
of conscious motor intentions in which processing may berelated to sensory predictions. Involvement of the parietalcortex is also suggested in encoding the kinematics and ori-entation of movements within the body space [17,43,57,67]m
Wo
T/V5), the right inferior parietal lobule (PGp), and the right
nd storing kinesthetic representations to map them ontohe premotor and motor areas [76]. The present activationies in area PF [90], which probably corresponds to area 7bn the macaque monkey [58]. It is in this region that Fogassit al. [31] reported the presence of mirror neurons in theacaque brain.The present result showing equal contribution of the pari-
tal cortex during the observation of movements that onean or cannot perform, appears to be in contradiction withhe hypothesis that sensory feedback is a prerequisite forhe access and maintenance of motor representations andssociated predictions. In the absence of such peripheralensations, motor representations are supposed to fade anday be no longer reachable [8,20,53]. An extended amount
f practice in sport (here, more than 10 years of intensiveractice) may lead to more stable representations that areess subject to decay [23]. For example, an investigationy Filippi et al. [30] provides some evidence that long-termtructural changes were occurred within the inferior parietalobule three months after ceasing extensive training of handctions. However, additional experiments are needed to testhether motor expertise does indeed prevent such decay.or instance, inclusion of a non-expert group in which non-xpert subjects suffer from a lower extremity injury at thenset of the experiment and who recover six months laternd subsequently comparing the level of activity within theON over time with that of the expert gymnasts could beone.
ncreased activity associated with temporary
otor injuryhile MT/V5/EBA and inferior parietal areas constitutivef the AON were activated both during the Hurt and Heal
4
sawt
dtbci(rtrqtWbictImcisptmcTbtamiec
lomrstead[atic[
tppps[p
L
TaTycploto‘ddsganctsstesaius
C
Iipbrtailgps
D
Tc
A
TFtc
42
essions, activity within the cerebellum (crus 1 and VI)ppears to be higher when gymnasts observed routines theyere temporarily unable to perform (as compared to when
hey had recovered).One possible interpretation for such increased activity
uring the Hurt session arises from the hierarchical predic-ive coding framework pioneered by Rao and Ballard [65] andased on Bayesian principles. The main idea is that eachortical level of this model forecasts the response at thenferior level through feedback connections. The mismatchi.e., estimated error) between prediction and incomingesponse is carried back through feedforward connectiono the superior level where the prediction of the incomingesponse is corrected/updated. A new prediction is subse-uently generated. If there is a match, the structure ofhe cortical level that sends the prediction is reinforced.ithin a hierarchical predictive coding framework, the cere-
ellum, which is part of the AON [12,18,34,50,81,84,94],s known to play a key role in the prediction of sensoryonsequences of actions [6,7,93] and in updating predic-ions about the visual consequences of one’s behavior [82].t is also engaged in making predictions about the move-ent of a visual target [59] or the velocity of a visual
ursor that has been occluded [56]. Hence, while activ-ty within parietal and MT/V5 areas did not differ acrossessions, we speculate that when injured the process ofredicting the outcome of others’ actions when viewingheir actions may be affected. As a consequence, the esti-ated error between predicted outcomes and incoming
onsequences of the viewed movement may be higher.his suggestion is in accordance with the model proposedy Shadmehr and Krakauer [74], stating that the func-ion of the cerebellum is system identification. It formsn internal model that predicts sensory consequences ofotor commands and corrects these commands through
nternal feedback whether there is a discrepancy betweenxpected and actual sensory consequences of the motorommands.
Another possible interpretation for increases in cerebel-ar activity (crus 1, VI) in response to observed movementsne cannot provisionally perform is related to motor auto-aticity. Although the gymnasts had regained their full
ange of performance when they underwent the secondcanning session, they may no longer used to going throughheir routines at an ‘‘automatic’’ level. The observedffect might therefore be seen as reduced cerebellarctivity due to the lack of repetitive practice sessionsuring the period of injury. Interestingly, Lehericy et al.47] showed that activity within cerebellar lobules crus 1nd VI decreased as motor learning progressed. Similarly,he scientific literature tends to agree that cerebellums involved in early stages of motor learning and that itsontribution is negligible when automaticity is reached25,62].
Further studies are nevertheless warranted to inves-igate the two possibilities. Exploration of whether arovisional incapacity to execute motor actions is cou-led with difficulties in predicting and recognizing other
eople’s actions, as shown for deficits in peripheralensations [8] and lesions within the motor system60,61,73], would allow further testing of the first pro-osal.CpfC
C. Calmels et al.
imitations of the study
he first limitation of the present study arises from thebsence of a second group with injuries in the upper limbs.his inclusion was part of the initial design of our study,et, only two expert gymnasts with upper extremity injuriesomplied with our inclusion criteria during the recruitmenthase (2.5 years). We therefore focused on the group withower limb injuries. The second limit concerns the absencef a behavioral measure that would have been useful to testhe hypothesis that a temporary motor injury has impactn action recognition. A third limit is the absence of a‘control’’ group i.e. gymnasts who did not incur injuryuring a period of 6 months. Indeed, a between-subjectsesign could have been used in addition to the within-ubjects design. This option was considered but high-levelymnasts who do not get injured during such a long periodre rare. Gymnastics at this level is a demanding, tech-ical, and risky sport, as the slightest performance erroran be disastrous. It can result in minor or serious injurieshat partially or fully interrupt training and occasionallytop a gymnast’s career. A fourth limitation is the smallample size. Statistical power may be invoked to explainhe absence of positive results in the AON. Although smallffects may remain undetected due to our small sampleize, it should be noted that comparisons between the injurynd recovery periods yielded significant (FWE) differencesn the cerebellum, suggesting that the experiment was notnderpowered for detecting strong effects between the twoessions.
onclusion
n conclusion, the present study revealed two main find-ngs. First, part of the AON (i.e., MT/V5/EBA and inferiorarietal lobule), involved in the perceptual processing ofody movements and shape and associated with actionepresentations, responded to observed actions irrespec-ive of the gymnasts’ physical condition and thus of theirbility to perform the viewed action. The second results that the higher levels of activity within cerebellarobules crus 1 and VI when the gymnasts were hurt sug-ests that the cerebellum dissociates perceived movementshysically feasible from those which are provisionally unfea-ible.
isclosure of interest
he authors declare that they have no conflicts of interestoncerning this article.
cknowledgements
his study was supported by a grant (no 07-012) from therench Ministry of Health and Sport. The authors are gratefulo Kevin Nigaud and Eric Bertasi for their skilful techni-al assistance (CENIR, ICM, CRICM, UPMC/INSERM UMR 975
NRS 7225, hôpital Pitié-Salpêtrière). Part of this paper wasresented at the 17th Annual Meeting of the Organizationor Human Brain Mapping, June 26—30, 2011, Quebec City,anada.
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
Motor injury and AON
References
[1] Aglioti SM, Pazzaglia M. Sounds and scents in (social) action.Trends Cogn Sci 2011;15:47—55.
[2] Albright TD. Direction and orientation selectivity of neu-rons in visual area MT of the macaque. J Neurophysiol1984;52:1106—30.
[3] Avenanti A, Urgesi C. Understanding ‘‘what’’ others do: mirrormechanisms play a crucial role in action perception. Soc CognAffect Neurosci 2011;6:257—9.
[4] Avenanti A, Annella L, Candidi M, Urgesi C, Aglioti SM. Com-pensatory plasticity in the action observation network: virtuallesions of STS enhance anticipatory simulation of seen actions.Cereb Cortex 2013;23:570—80.
[5] Bassolino M, Campanella M, Bove M, Pozzo T, Fadiga L. Train-ing the motor cortex by observing the actions of others duringimmobilization. Cereb Cortex 2013 [ahead of print].
[6] Blakemore SJ, Frith CD, Wolpert DM. The cerebellum isinvolved in predicting the sensory consequences of action. Neu-roreport 2001;12:1879—84.
[7] Blakemore SJ, Sirigu A. Action prediction in the cerebellum andin parietal lobe. Exp Brain Res 2003;153:239—45.
[8] Bosbach S, Cole J, Prinz W, Knoblich G. Inferring another’sexpectation from action: the role of peripheral sensation. NatNeurosci 2005;8:1295—7.
[9] Buccino G, Binkofski F, Fink GR, Fadiga L, Fogassi L, GalleseV, et al. Action observation activates premotor and parietalareas in a somatotopic manner: an fMRI study. Eur J Neurosci2001;13:400—4.
[10] Buccino G, Vogt S, Ritzl A, Fink GR, Zilles K, FreundHJ, et al. Neural circuits underlying imitation learning ofhand actions: an event-related fMRI study. Neuron 2004;42:323—34.
[11] Calvo-Merino B, Glaser DE, Grèzes J, Passingham RE, HaggardP. Action observation and acquired motor skills: an fMRI studywith expert dancers. Cereb Cortex 2005;15:1243—9.
[12] Calvo-Merino B, Grèzes J, Glaser DE, Passingham RE, HaggardP. Seeing or doing? Influence of visual and motor familiarity inaction observation. Curr Biol 2006;16:1905—10.
[13] Casper S, Zilles K, Laird AR, Eickhoff SB. ALE meta-analysis ofaction observation and imitation in the human brain. Neuroim-age 2010;50:1148—67.
[14] Cattaneo L, Barchiesi G, Tabarelli D, Arfeller C, Sato M,Glenberg AM. One’s motor performance predictability modu-lates the understanding of others’ actions through adaptationof premotor visuo-motor neurons. Soc Cogn Affect Neurosci2011;6:301—10.
[15] Chong TTJ, Cunnington R, Williams MA, Kanwisher N, Mat-tingley JB. fMRI adaptation reveals mirror neurons in humaninferior parietal cortex. Curr Biol 2008;18:1576—80.
[16] Cochin S, Barthelemy C, Roux S, Martineau J. Observation andexecution of movement: similarities demonstrated by quanti-fied electroencephalography. Eur J Neurosci 1999;11:1839—42.
[17] Colby CL, Goldberg ME. Space and attention in parietal cortex.Ann Rev Neurosci 1999;22:319—49.
[18] Cross ES, Hamilton AFC, Grafton ST. Building a motor simula-tion de novo: observation of dance by dancers. Neuroimage2006;31:1257—67.
[19] Cross ES, Hamilton AC, Kraemer DJM, Kelley WM, Grafton ST.Dissociable substrates for body motion and physical experi-ence in the human action observation network. Eur J Neurosci2009;30:1383—92.
[20] De Jong BM, Leenders KL, Paans AMJ. Right parieto-premotoractivation related to limb-independent antiphase movement.
Cereb Cortex 2002;12:1213—7.[21] Desmurget M, Sirigu A. A parietal-premotor network formovement intention and motor awareness. Trends Cogn Sci2009;13:411—9.
[
443
22] Desmurget M, Reilly KT, Richard N, Szathmari A, Mottolese C,Sirigu A. Movement intention after parietal cortex stimulationin humans. Science 2009;324:811—3.
23] Diersch N, Cross ES, Stadler W, Schutz-Bosbach S, Rieger M.Representing others’ actions: the role of expertise in the agingmind. Psychol Res 2012;76:525—41.
24] Dinstein I, Hasson U, Rubin N, Heeger DJ. Brain areas selectivefor both observed and executed movements. J Neurophysiol2007;98:1415—27.
25] Doyon J, Bellec P, Amsel R, Penhume V, Monchi O, CarrierJ, et al. Contribution of the basal ganglia and functionallyrelated brain structure to motor learning. Behav Brain Res2009;199:61—75.
26] Duvernoy HM. The human brain. Surface, three-dimensionalsectional anatomy with MRI, and blood supply. Vienna:Springer-Verlag; 1999.
27] Fadiga L, Fogassi L, Pavesi G, Rizzolatti G. Motor facilitationduring action observation: a magnetic stimulation study. J Neu-rophysiol 1995;73:2608—11.
28] Ferri S, Kolster J, Jastorff J, Orban GA. The overlap of the EBAand the MT/V5 cluster. Neuroimage 2013;66:412—25.
29] Filimon F, Nelson JD, Hagler DJ, Sereno MI. Humancortical representations for reaching: mirror neurons forexecution, observation, and imagery. Neuroimage 2007;37:1315—28.
30] Filippi M, Ceccarelli A, Pagani E, Gatti R, Rossi A, Stefanelli,et al. Motor learning in healthy humans is associated to graymatter changes: A tensor-based morphometry study. PLoS ONE2010;5:e10198.
31] Fogassi L, Gallese V, Fadiga L, Rizzolatti G. Neurons respon-ding to the sight of goal directed hand/arm actions in theparietal areas PF (7b) of the macaque monkey. Soc Neurosci1998;24:257.5 [Abstr.].
32] Friston K, Mattout J, Kilner J. Action understanding and activeinference. Biol Cybern 2011;104:137—60.
33] Gangitano M, Mottaghy FM, Pascual-Leone A. Phase-specificmodulation of cortical motor output during movement obser-vation. Neuroreport 2001;12:1489—92.
34] Gazzola V, Keysers C. The observation and execution of actionsshare motor and somatosensory voxels in all tested subjects:single-subject analyses of unsmoothed fMRI data. Cereb Cortex2009;19:1239—55.
35] Grèzes J, Armony JL, Rowe J, Passingham RE. Activationsrelated to ‘‘mirror’’ and ‘‘canonical’’ neurones in humanbrain: an fMRI study. Neuroimage 2003;18:928—37.
36] Henson RNA. Analysis of fMRI time series: linear time-invariantmodels, event-related fMRI and optimal experimental design.In: Frackowiak RSJ, Friston KJ, Frith C, Dolan R, Price C,editors. Human brain function. London: Elsevier; 2003. p.793—822.
39] Iacoboni M, Wood RP, Brass M, Bekkering H, Mazziotta JC, Riz-zolatti G. Cortical mechanisms of human imitation. Science1999;286:2526—8.
40] Iacoboni M, Molnar-Szakacs I, Gallese V, Buccino G, MazziottaJC, Rizzolatti G. Grasping the intentions of others with one’sown mirror neuron system. PLoS Biol 2005;3:e79.
41] Jeannerod M. The representing brain: neural correlates ofmotor intention imagery. Behav Brain Sci 1994;17:187—245.
42] Kalaska JF, Crammond DJ. Deciding not to GO: neuronalcorrelates of response selection in a GO/NOGO task in pri-mate premotor and parietal cortex. Cereb Cortex 1995;5:410—28.
43] Kheradmand A, Lasker A, Zee DS. Transcranial magneticstimulation (TMS) of the supramarginal gyrus: a window toperception of upright. Cereb Cortex [Ahead of print].
45] Kolster H, Peeters R, Orban GA. The retinotopic organizationof the human middle temporal area MT/V5 and its corticalneighbors. J Neurosci 2010;30:9801—20.
4
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
44
46] Langer N, Hanggi J, Muller NA, Simmen HP, Jancke L.Effects of limb immobilization on brain plasticity. Neurology2012;78:182—8.
47] Lehericy S, Benali H, van de Moortele P-F, Pélégrini-Issac M,Waechter T, Ugurbil K, et al. Distinct basal ganglia territoriesare engaged in early and advanced motor sequence learning.Proc Natl Acad Sci 1995;102:12556—71.
48] Lieberman MD, Cunningham WA. Tools of the trade. Type I andtype II error concerns in fMRI research: re-balancing the scale.Soc Cogn Affect Neurosci 2009;4:423—8.
49] Mitchell DB, Brown AS, Murphy DR. Dissociations betweenprocedural and episodic memory: effects of time and aging.Psychol Aging 1990;5:264—76.
50] Molenberghs P, Cunnington R, Mattingley JB. Brain regions withmirror properties: a meta-analysis of 125 human fMRI studies.Neurosci Biobehav Rev 2012;36:341—9.
51] Molnar-Szakacks I, Kaplan J, Greefield PM, Iacoboni M.Observing complex action sequences: the role of fronto-parietal mirror neuron system. Neuroimage 2006;33:923—35.
52] Naito E, Kochiyama T, Kitada R, Nakamura S, Matsumura M,Yonekura Y, et al. Internally simulated movement sensationsduring motor imagery activate cortical motor areas and thecerebellum. J Neurosci 2002;22:3683—91.
53] Naito E, Roland PE, Grefkes C, Choi HJ, Eickhoff S, GeyerS, et al. Dominance of the right hemisphere and roleof area 2 in human kinaesthesia. J Neurophysiol 2005;93:1020—34.
54] Nichols T, Brett M, Andersson J, Wager T, Poline JB. Valid con-junction inference with the minimum statistic. Neuroimage2005;25:653—60.
55] Nishitani N, Hari R. Temporal dynamics of cortical repre-sentation for action. Proc Natl Acad Sci U S A 2000;97:913—8.
56] O’Reilly JX, Mesulam MM, Nobre AC. The cerebellum pre-dicts the timing of perceptual events. J Neurosci 2008;28:2252—60.
57] Oztop E, Imamizu H, Cheng G, Kawato M. A computationmodel of anterior intraparietal (AIP) neurons. Neurocomputing2006;69:1354—61.
58] Passingham RE. The specializations of the human neocortex.In: Milner AD, editor. Comparative neuropsychology. New York:Oxford University Press; 1998. p. 271—98.
59] Paulin MG. The role of cerebellum in motor control and per-ception. Brain Behav Evol 1993;41:39—50.
60] Pavlova M, Staudt M, Sokolov A, Birbaumer N, Krageloh-Mann I. Perception and production of biological movementin patients with early periventricular brain lesions. Brain2003;126:692—701.
61] Pazzaglia M, Smania N, Corato E, Aglioti SM. Neural underpin-nings of gesture discrimination in patients with limb apraxia. JNeurosci 2008;28:3030—41.
62] Penhune VB, Steele CJ. Parallel contributions of cerebellar,striatal and M1 mechanisms to motor sequence learning. BehavBrain Res 2012;226:579—91.
63] Pilgramm S, Lorey B, Stark R, Munzert J, Zentgraf K. The role ofown-body representations in action observation: a functionalMRI study. Neuroreport 2009;20:997—1001.
64] Prinz W. Perception and action planning. Eur J Cogn Psychol1997;9:129—54.
65] Rao RPN, Ballard DH. Predictive coding in the visual cortex:a functional interpretation of some extra-classical receptive-field effects. Nat Neurosci 1999;2:79—87.
66] Reid VM, Belsky J, Johnson MH. Infant perception of humanaction: toward a developmental cognitive neuroscience of indi-
vidual differences. Cogn Brain Behav 2005;IX:193—210.67] Rizzolatti G, Fogassi L, Gallese V. Parietal cortex: from sight toaction. Curr Opin Neurobiol 1997;7:562—7.
[
C. Calmels et al.
68] Rizzolatti G, Fogassi L, Gallese V. Neurophysiological mecha-nisms underlying the understanding and imitation of action.Nat Rev Neurosci 2001;2:661—70.
69] Ricciardi E, Bonino D, Sani L, Vecchi T, Guazzelli M, Haxby JV,et al. Do we really need vision? How blind people ‘‘see’’ theactions of others. J Neurosci 2009;29:9719—24.
70] Schippers MB, Keysers C. Mapping the flow of informationwithin the putative mirror neuron system during gesture obser-vation. Neuroimage 2011;57:37—44.
71] Schubotz RI. Prediction of external events with our motorsystem: Towards a new framework. Trends Cogn Sci2007;11:211—8.
72] Schütz-Bosbach S, Prinz W. Perceptual resonance: action-induced modulation of perception. Trends Cogn Sci2007;11:349—55.
73] Serino A, De Filippo L, Casavecchia C, Coccia M, Shiffrar M,Ladavas E. Lesions to the motor system affect action percep-tion. J Cogn Neurosci 2010;22:413—26.
74] Shadmehr R, Krakauer JW. A computational neuroanatomy formotor control. Exp Brain Res 2008;185:359—81.
75] Simoes EL, Bramati I, Rodrigues E, Franzoi A, Moll J, Lent R,et al. Functional expansion of sensorimotor representation andstructural reorganization of callosal connections in lower limbamputees. J Neurosci 2012;29:3211—20.
76] Sirigu A, Duhamel JR, Cohen L, Pillon B, Dubois B, Agid Y. Themental representation of hand movements after parietal cor-tex damage. Science 1996;273:1564—8.
77] Smith CD, Walton A, Loveland AD, Umberger GH, Kryscio RJ,Gash DM. Memories that last in old age: motor skill learningand memory preservation. Neurobiol Aging 2005;26:883—90.
78] Snyder L, Batista AP, Andersen RA. Coding of intention in theposterior parietal cortex. Nature 1997;386:167—70.
79] Snyder L, Batista AP, Andersen RA. Intention-related activ-ity in the posterior parietal cortex: a review. Vision Res2000;40:1433—41.
80] Strafella AP, Paus T. Modulation of cortical excitability duringaction observation: a transcranial magnetic stimulation study.Neuroreport 2000;11:2289—92.
81] Suchan B, Melde C, Herzog H, Homberg V, Seitz RJ. Activationdifferences in observation of hand movements for imitation orvelocity judgement. Behav Brain Res 2008;188:78—83.
82] Synofzik M, Lindner A, Their P. The cerebellum updates pre-dictions about the visual consequences of one’s behavior. CurrBiol 2008;18:814—8.
83] Toni I, Thoenissen D, Zilles K. Movement preparation and motorintention. Neuroimage 2001;14:S110—7.
84] Torriero S, Oliveri M, Koch G, Caltagirone C, Petrosini L. Thewhat and how of observational learning. J Cogn Neurosci2007;19:1656—63.
85] Turella L, Erb M, Grood W, Castiello U. Visual features of anobserved agent do not modulate human brain activity duringaction observation. Neuroimage 2009;46:844—53.
86] Umilta MA, Kohler E, Gallese V, Fogassi L, Fadiga L, Keysers C,et al. I know what you are doing: a neurophysiological study.Neuron 2001;31:155—65.
87] Urgesi C, Candidi M, Fabbro F, Romani M, Aglioti S. Motor facil-itation during action observation: topographic mapping of thetarget muscle and influence of the onlooker’s posture. Eur JNeurosci 2006;23:2522—30.
88] Van Essen DC, Drury HA, Dickson J, Harwell J, Hanlon D,Anderson CH. An integrated software suited for surface-basedanalyses of cerebral cortex. J Am Med Inform Assoc 2001;8:443.
89] Van Essen DC, population-average A. landmark-and surface-based (PALS) atlas of human cerebral cortex. Neuroimage
2005;28:635—62.90] Von Economo C, Koskinas GN. The cytoarchitectonics of thehuman cerebral cortex. London: Oxford University Press; 1929.
[
Motor injury and AON
[91] Walton A, Scheib JL, McLean S, Zhang Z, Grondin R.Motor memory preservation in aged monkeys mirrors that
of aged humans on a similar task. Neurobiol Aging 2008;29:1556—62.[92] Wilson M, Knoblich G. The case for motor involvement in per-ceiving conspecifics. Psychol Bull 2005;131:460—73.
[
445
93] Wolpert DM, Miall RC, Kawato M. Internal models in the cere-bellum. Trends Cogn Sci 1998;2:338—47.
94] Zentgraf K, Stark R, Reiser M, Kunzell S, Schienle A, KirschP, et al. Differential activation of pre-SMA and SMA properduring action observation: effects of instructions. Neuroimage2005;26:662—72.