Conscious and Unconscious Representations of Observed Actions in the Human Motor System.

Conscious and Unconscious Representations of ObservedActions in the Human Motor System

Alan D. A. Mattiassi1*, Sonia Mele1*, Luca F. Ticini2,and Cosimo Urgesi1,3

Abstract

■ Action observation activates the observerʼs motor system.These motor resonance responses are automatic and triggeredeven when the action is only implied in static snapshots. How-ever, it is largely unknown whether an action needs to be con-sciously perceived to trigger motor resonance. In this study, weused single-pulse TMS to study the facilitation of corticospinalexcitability (a measure of motor resonance) during supraliminaland subliminal presentations of implied action images. We useda forward and backward dynamic masking procedure that suc-cessfully prevented the conscious perception of prime stimulidepicting a still hand or an implied abduction movement of theindex or little finger. The prime was followed by the supraliminalpresentation of a still or implied action probe hand. Our resultsrevealed a muscle-specific increase of motor facilitation following

observation of the probe hand actions that were consciouslyperceived as compared with observation of a still hand. Crucially,unconscious perception of prime hand actions presented beforeprobe still hands did not increase motor facilitation as comparedwith observation of a still hand, suggesting that motor resonancerequires perceptual awareness. However, the presentation of amasked prime depicting an action that was incongruent withthe probe hand action suppressed motor resonance to the probeaction such that comparable motor facilitation was recordedduring observation of implied action and still hand probes. Thissuppression of motor resonance may reflect the processing ofaction conflicts in areas upstream of the motor cortex and maysubserve a basic mechanism for dealing with the multiple andpossibly incongruent actions of other individuals. ■

INTRODUCTION

In everyday life, we simultaneously deal with the actionsof numerous agents that we must flexibly imitate, com-plement, or react to (Sartori, Bucchioni, & Castiello,2013). Optimal interactions in such a crowded social worldrequire fast, accurate, and dynamic representations ofothersʼ actions that involve motor resonance responsesin the observersʼ brains (Rizzolatti & Craighero, 2004).Single-pulse TMS experiments have shown that actionobservation triggers a strictly congruent, muscle-specificfacilitation of corticospinal excitability that reflects motorresonance (e.g., Fadiga, Craighero, & Olivier, 2005). Al-though mirror-like motor facilitation during action obser-vation was initially thought to imply covert imitation, ithas been demonstrated that it takes place independentlyof explicit instruction to rehearse the actions (Fadiga,Fogassi, Pavesi, & Rizzolatti, 1995) and to be modulatedby subtle kinematic differences in the movements evenwhen these differences cannot be recognized by the ob-

servers (Sartori, Bucchioni, & Castiello, 2012). Moreover,it even occurs during passive viewing of static images thatimply body actions (Avenanti, Annella, Candidi, Urgesi, &Aglioti, 2013; Urgesi et al., 2010; Urgesi, Moro, Candidi, &Aglioti, 2006).In a similar vein, behavioral studies have shown that

viewing incongruent versus congruent dynamic actionsequences (Kilner, Paulignan, & Blakemore, 2003; Brass,Bekkering, & Prinz, 2001; Brass, Bekkering, Wohlschläger,& Prinz, 2000; Stürmer, Aschersleben, & Prinz, 2000) orsingle frames that imply actions (Vogt, Taylor, & Hopkins,2003; Craighero, Bello, Fadiga, & Rizzolatti, 2002; Brasset al., 2000; Craighero, Fadiga, & Rizzolatti, 1999; Craighero,Fadiga, Umiltà, & Rizzolatti, 1996) may affect the correctexecution of the observerʼs movements. These visuomotorinterference effects have been demonstrated to occureven if the observer ignores the action stimulus while per-forming a different task-relevant action (Vainio, Tucker, &Ellis, 2007). Furthermore, they are not suppressed byattentional modulation of the predictability of the prime–cue association obtained by varying the proportion of con-gruent and incongruent trials (Hogeveen & Obhi, 2013).It is worth noting that there is evidence that both motorfacilitation, at the neurophysiological level, and visuomotorinterference effects, at the behavioral level, are modulated

1Università di Udine, 2University of Manchester, UK, 3Istituto diRicovero e Cura a Carattere Scientifico “E. Medea,” Polo FriuliVenezia Giulia, San Vito al Tagliamento (Pordenone), Italy*These authors contributed equally to the study.

© 2014 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 26:9, pp. 2028–2041doi:10.1162/jocn_a_00619

by attention. For example, the electrophysiological indicesof motor activation during action observation (i.e., sup-pression of beta band recorded from central sites) are am-plified by the increasing task demands of active imitationor counting task with respect to passive observation, butthey are present in all conditions (Muthukumaraswamy &Singh, 2008). In a similar vein, visuomotor interferenceeffects are attenuated by the activation of self-related pro-cessing by mirror self-observation or judgment of evaluativepersonal statements, but are still present in all conditions(Spengler, von Cramon, & Brass, 2010).Together, these studies suggest that motor resonance

is elicited automatically, that is, in every instance in whicha biological movement is perceived independently fromthe presence or absence of executive control, cognitiveeffort, intentionality, strategies, or focus of attention ofthe observer (Bargh, Schwader, Hailey, Dyer, & Boothby,2012). Although these top–down processes are not neces-sary for motor resonance to occur, they can modulateits intensity, increasing or attenuating its manifestations.Currently, however, it is still unclear whether motor reso-nance requires observersʼ awareness of the perception ofthe movement or whether it also occurs in response toactions that are not consciously perceived. Varying the vis-ibility of a stimulus has been widely used to manipulateperceptual awareness, as indexed by observersʼ abilityto report the perception when they are prompted to doso (Van den Bussche, Van den Noortgate, & Reynvoet,2009). It has been argued that phenomenal experience isnot a direct reflection of perceptual processes and thatthe effect of consciously perceived stimuli may be differ-ent than that of unconsciously perceived ones (e.g., seeMerikle, Smilek, & Eastwood, 2001; Merikle & Daneman,1998, for reviews). For instance, masked visual presen-tations of prime stimuli that are associated with a latera-lized motor response, including hand postures (Vainio &Mustonen, 2011), may affect the execution of incongruentversus congruent responses and modulate motor cortexactivity (DʼOstilio & Garraux, 2012; Théoret, Halligan,Kobayashi, Merabet, & Pascual-Leone, 2004; Dehaene,Naccache, Le ClecʼH, Koechlin, &Mueller, 1998). However,these masked visuomotor prime effects strictly dependon the selection of the motor response concurrently re-quired by the explicit task (Eimer & Schlaghecken, 1998,2003) and may reflect the influence of the masked primeson action execution processes rather than motor reso-nance representations.On the other hand, automatic motor responses during

passive observation of emotional face and body expres-sions have been reported even when stimuli are presentedto the blind hemifield of hemianopic patients (Van denStock et al., 2011; Tamietto et al., 2009) or to healthy in-dividuals with masking procedures (Dimberg, Thunberg,& Elmehed, 2000). However, such automatic motor re-sponses may be specific for stimuli with emotional valenceand may not be involved in nonemotional action percep-tion (Tamietto & De Gelder, 2010). In fact, mirror-like,

muscle-specific facilitation of corticospinal excitabilityduring observation of grasping actions was attenuated bypresentation of the moving finger in partially shadowedillumination conditions (Sartori & Castiello, 2013); thegrasping act, however, was still visible, and mirror-likemotor facilitation was attenuated but not extinguished,thus leaving unquestioned whether motor resonancerequires perceptual awareness.

In this study, to investigate the impact of the uncon-scious perception of nonemotional actions on motorresonance responses, we tested the modulation of cor-ticospinal activity of individuals that passively observedmasked hand action stimuli.

METHODS

Participants

Twenty-two healthy volunteers (11 women and 11 men)aged 19–34 years (mean = 23 years, SD = 4 years) par-ticipated to the experiment. Participants had normal orcorrected-to-normal visual acuity and were all right-handedaccording to a standard handedness inventory (Briggs &Nebes, 1975). All participants were naive to the purposeof the experiment but received detailed information aboutthe procedures. Participants gave written informed consentand received course credit for their participation in thestudy. The procedures were approved by the ethics com-mittee of the Scientific Institute E. Medea (Bosisio Parini,Como, Italy) and were in accordance with the ethical stan-dards of the 1964 Declaration of Helsinki. None of theparticipants had neurological, psychiatric, or other medicalproblems or any contraindications to TMS (Wasserman,1998). There were no reports or observations of anydiscomfort or adverse effects during TMS.

Stimuli

Stimuli were color pictures taken with a digital cameraduring the execution of abduction movements of the rightindex or little finger. Pictures from eight models (fourwomen and four men, aged 20–23 years) were used asstimuli to minimize habituation and loss of attention. Foreach model, three pictures of the hand were taken: (a) apicture in a resting position, (b) a picture with the indexfinger abducted (a movement that requires contractionof the first dorsal interosseous or first dorsal interosseusmuscle [FDI] muscle), and (c) a picture with the littlefinger abducted (a movement that requires the contrac-tion of the abductor digiti minimi [ADM] muscle). Topreserve the appearance of naturalistic movement, eachpicture was taken while the model was moving the finger.Pictures were taken from an egocentric perspective in lightof evidence of higher mirror-like motor facilitation inresponse to hand movements viewed from egocentricthan allocentric perspective (Alaerts, Heremans, Swinnen,

Mattiassi et al. 2029

& Wenderoth, 2009; Maeda, Kleiner-Fisman, & Pascual-Leone, 2002). Light conditions were kept constant acrossthe three images of each model and the remaining lumi-nance differences were manually corrected using CorelPaint Shop Pro X (Corel, Inc., Mountain View, CA). Arotating mask was prepared by overlapping two identicalstar-like geometrical figures that were textured with ascrambled version of the hand pictures using the Bryce3-D software (DAZ Productions, Inc., Salt Lake City, UT).To mask the displacement of the fingers to the left orthe right side of the screen, the two star-like figures rotatedin opposite directions at 0.63 Hz (i.e., 3° every 13.33 msecrefresh event), with a velocity almost comparable to thatof the abduction/adduction displacement of the finger(2.8–4.3° every 13.33 msec refresh event). This way, wewere able to mask not only the perception of the formcues (hand shape) but also any motion cues induced bythe apparent displacement of the index or little finger.We prepared two different versions of the mask that werepresented randomly in different trials. In one version, theforeground figure rotated clockwise, whereas the back-ground figure rotated counterclockwise and vice versa forthe other version. Hand and mask stimuli were presentedon a uniform background and subtended a central 6° ×8.5° region.

Procedure

Stimuli were presented in a subliminal masked primingparadigm (Figure 1). Each trial started with a fixation crosspresented in the center of the screen for 500 msec andproceeded to the successive presentation of the three fol-lowing hand stimuli: a sample hand for 250 msec, a primehand for 53 msec, and a probe hand for 250 msec. Primeduration was selected to be just below the threshold ofconscious perception (i.e., subliminal) on the basis ofpreliminary behavioral data on a different group of indi-viduals that could not report the presence of hand primespresented in the same paradigm as in this study. Picturesshowing relaxed or abducted fingers were used insteadof video clips because the stimulus duration required bysubliminal masked presentations is too short for allow-ing even two-frame movement sequences at the typical25-Hz video frame rate. Nonetheless, previous studies(Avenanti, Annella, et al., 2013; Urgesi et al., 2010; Urgesi,Moro, et al., 2006) have provided evidence of muscle-specific, mirror-like facilitation in response to single framesdepicting an implied action, thus supporting that ourstimuli are adept to study motor resonance. The samplehand was always a still hand, whereas the prime and theprobe stimuli could depict a still hand or a hand implyingan abduction movement of the index or little finger. Primepresentation was forward and backward masked with therotating star-like figure. The mask preceding the prime(forward mask) could rotate for 40, 80, 120, or 160 msec;the duration of the rotation was randomly selected foreach trial and could not be predicted; random duration

of the rotating forward mask was aimed at limiting anticipa-tion of the time of prime presentation, which could beperceived by participants as a flickering of the mask. Onthe contrary, the mask following the prime (backwardmask) presented a rotation of 120 msec, thus inducingconstant backward masking effects. Notably, the rotationof the backward mask started from a point congruent tothat expected from a continuous rotation of the star-likefigure during the prime presentation time; this facilitatedthe perceptual fusion of the forward and backward maskrotation into a single, smooth movement, thus strengthen-ing the possible masking effect. After a 2000, 2500, 3000, or3500 msec delay following probe presentation (with delayduration randomly varying for each trial), a responsescreen was presented with the request to report whethera still or a moving probe (either index or little finger move-ment) was pictured in the preceding trial. In differenttrials, participants were asked: “Was the hand still ormoving?”, “Was the hand moving or still?”, “Was the handmoving?”, or “Was the hand still?” In no trials were theparticipants asked to report whether the probe hand dis-played movement of the index or little finger. The questionwas presented for 3000 msec on the left of the responsescreen, whereas two possible answers (still or moving;yes or no) were simultaneously presented on the right.Participants were instructed to report by means of a vocalresponse the position of the correct answer (up or down)based on the type of probe. The vertical position of thecorrect answer was randomly varied across trials. This pro-cedure was chosen to prevent spurious priming effectson the size of motor-evoked potentials (MEPs) such assublexical processes that are known to activate M1 (e.g.,McGettigan et al., 2012).A single TMS pulse was delivered at one of twomoments

during the trial, either 133 msec (early delay) or 307 msec(late delay) after the onset of the probe, correspondingto 307 msec after the onset of the prime in the early delayand 307 msec after the onset of the probe in the latedelay. A blank screen was presented before the nexttrial to create interpulse intervals ranging from 11,713 to11,833 msec (Brasil-Neto et al., 1992).Participants were tested in a single experimental session

lasting ∼75 min. They sat in a dimly lit room 57 cm awayfrom a 21-in. CRT monitor (resolution: 1024 × 768 pixels,refresh frequency: 75 Hz) with their head positioned ona chin rest. Participants were instructed to pay attentionto the sequence of hands presented on the screen butwere informed only of the presence of the sample, mask,and probe hands. Thus, they were naive regarding thepresence of the prime hand.The different conditions were presented in a ran-

domized order in eight blocks of 48 trials each. In halfof the trials a still probe was presented, and in the otherhalf a moving probe was presented. Half of the movingprobe trials displayed an index finger movement, andthe other half presented a movement of the little finger.Thus, the same numbers of moving and nonmoving trials

2030 Journal of Cognitive Neuroscience Volume 26, Number 9

Figure 1. Time line of the experimental trials. Trial structure: Each trial started with a fixation cross (500 msec). Subsequently, a still samplehand (250 msec), a prime hand (53 msec), and a probe hand (250 msec) were presented. The prime and the probe stimuli could depict eithera still hand or a hand implying the abduction movement of the index or little finger. Prime presentation was forward and backward maskedwith a rotating star-like figure that was presented before the prime for 40, 80, 120, or 160 msec and continued after the prime for 120 msec.The trials ended with a blank screen, and participants were prompted to report the motion of the probe stimulus. TMS pulse timeline: A singleTMS pulse was delivered at one of two moments during the trial, either 307 msec (early TMS delay) after prime onset (corresponding to133 msec after probe onset) or 307 msec (late TMS delay) after probe onset.


were presented to avoid biasing the participants in themotion detection task. The TMS delay was randomlyvaried in each trial. In total, 384 MEPs were recorded fromeach muscle, with 192 still probe trials (32 trials for eachof the 2 delays × 3 primes), 96 index finger movementprobes (16 trials for each of the 2 delays × 3 primes)and 96 little finger movement probes (16 trials for eachof the 2 delays × 3 primes). Two series of eight baselineMEPs were recorded at the beginning and at the endof the experimental session, during which participantswere required to look at the screen, but no stimuli werepresented.

After the experimental session, participants wereprompted to report any discrepancies between the instruc-tions given and their understanding of the trial structureduring the experiment. This allowed us to test for spon-taneous report of the prime presence. Following thisreport, the experimenter described the actual trial struc-ture, including the presence of the prime. Participants werethen asked for a confirmation of their previous report andwere required to perform a control, forced-choice taskto determine whether they could discriminate the typeof prime after being informed of its presence. In thisforced-choice task, participants were presented with theexperimental trials, and after each trial, they were askedto press one of three keys to report whether the primeshowed a still hand, a hand with the index finger abducted,or a hand with the little finger abducted. They wereinstructed to answer “by feeling” and to guess for everytrial in which they did not notice the presence of theprime. The trial structure used was identical to that usedduring TMS, with the exception that the response screenwas replaced by a request for a button press in the middleof the screen. This ensured that participants were facedwith the same conditions as in the main experiment. Notime limit was given for the participantsʼ responses.The next trial started immediately after the response. All384 trials of the experimental session were presentedand randomized in two blocks of 192 trials each. Thecontrol task lasted approximately 4 min.

Electromyography Recording and TMS

To check for muscle specificity and for any modulationrelated to the different observational conditions, bothFDI and ADM MEPs were simultaneously recorded. Sur-face Ag/AgCl disposable electrodes (1 cm diameter) wereplaced in a belly-tendonmontage for each muscle and con-nected to a Biopac MP-36 system (BIOPAC Systems, Inc.,Goleta, CA) for amplification, band-pass filtering (5 Hz to20 kHz), and digitization of the EMG signal (sampling rate:50 kHz).

A 70-mm figure-eight stimulation coil (Magstimpolyurethane-coated coil) connected to a Magstim 200Rapid (The Magstim Company, Carmarthenshire, Wales,UK) was used to perform focal TMS (maximum output =2 T at coil surface, pulse duration = 250 μsec, rise time =

60 μsec). The coil was placed tangentially on the scalp,with the handle pointing backward and approximately45° lateral from the midline such that coil was placedperpendicularly to the line of central sulcus (Di Lazzaroet al., 1998). The coil position was marked on the par-ticipantsʼ scalp. The coil was held on the scalp by a coilholder with an articulated arm, and the experimentercontinuously checked the position of the coil with re-spect to the marks and compensated for any small move-ments of the participantʼs head during data collection.The exact position of the coil varied for each participanton the basis of the optimal scalp position (OSP), whichwas defined as the position from which MEPs fromboth the FDI and ADM muscles with maximal amplitudewere recorded. The OSP was detected by moving the coilaround the motor hand area of the left motor cortex pro-jection on the scalp and delivering single TMS pulses ofconstant intensity. The resting motor threshold, definedas the lowest TMS intensity able to evoke MEPs withamplitudes of at least 50 μV after 5 of 10 stimulations inthe higher threshold muscle (ADM), was determined byholding the stimulation coil over the OSP. To record stableMEPs from both muscles, stimulation intensity during therecording session was 120% of the resting motor thresholdand ranged from 50% to 84% (mean = 67%, SD = 8.46%)of maximum stimulator output. Importantly, the chosenscalp positions and stimulation intensities allowed usto record clear and stable EMG signals (10 MEPs of 10 TMSpulses) from both recorded muscles in all participants.During MEP recordings, the background EMG signal wascontinuously monitored, and when voluntary contractionsof the recorded muscles were detected, participants wereencouraged to fully relax their muscles. The peak-to-peakMEP amplitudes (in millivolts) were collected and storedon a computer for offline analysis.

Data Handling

Five participants were excluded from the analysis becausethey reported the presence of the prime spontaneously orafter receiving a description of the actual trial structure.This procedure is weak to false positives (i.e., reports ofseeing the prime after the explanation of its presence byparticipants that were actually unaware of the prime) butnot to false negatives (i.e., participants that were awareof the prime failing to report its presence). The remainingparticipants were entirely unaware of the presence of theprime. The peak-to-peak amplitude of each MEP was cal-culated, and trials with background activities greater than50 μV, amplitudes less than or equal to 50 μV, or am-plitudes greater or less than ±2.5 SD from the mean werediscarded. After this procedure, mean values were ob-tained from an average of 92.52% (SD = 7.31%) of therecorded MEPs per condition. The number of recordedMEPs did not vary across the different experimentalconditions between the early and late delays (all Fs <2.6), but there was a significant effect of muscle in the early


delay MEPs; fewer MEPS from the little finger (mean =91.87%, SD = 1.09%) than from the index finger (mean =92.79%, SD = 1.02%) were used.For each participant and each condition, the mean MEP

amplitude for each muscle and for each TMS pulse delaywas expressed as percent change from the mean value ofthe baseline MEPs of that muscle (baselines were collapsedacross the MEPs recorded at the beginning and at the endof the experimental session). This procedure allowed usto obtain an MEP ratio index of motor facilitation, hereafterreferred to as the MEPratio, which takes into account inter-individual differences in baseline corticospinal excitabilityand allowed improving normal distribution of the variablesas checked with the Kolmogorov–Smirnov test for nor-mality. The levels of the prime and probe for each experi-mental trials were coded into the following categoriesbased on the correspondence between the muscle fromwhich MEPs were recorded and the muscle that drivesthe observed movement: (a) still:still hands for bothmuscles; (b) related:index finger movements for FDI MEPsand little finger movements for ADM MEPs; (c) unrelated:little finger movements for FDI MEPs and index fingermovements for ADM MEPs. It is worth noting that be-havioral studies of visuomotor priming effects (e.g., Kilneret al., 2003; Brass et al., 2000; Stürmer et al., 2000) tendto code the conditions according to the congruency be-tween the prime and the probe actions. However, becausethe critical measure in this study is the facilitation of thecorticospinal excitability of specific muscle, which actionis shown and not only whether prime and probe pairsare congruent versus incongruent is crucial for the effect.Thus, in keeping with previous single-pulse TMS studiesof visuomotor interaction (Catmur, Mars, Rushworth, &Heyes, 2011), we coded the prime and probe conditionsin terms of whether they showed a static hand or an im-plied action that was related versus unrelated to the motorrole of the recorded muscle. Furthermore, because theaction (i.e., index finger abduction) that is related to theFDI motor role is unrelated to the ADM motor role,and vice versa, expressing conditions in terms of related/unrelated actions, instead of index/little finger abduction,allowed us to directly test somatotopic motor facilitation(i.e., greater motor facilitation during observation ofactions whose execution requires the recorded musclevs. observation of static hand and of actions that do notinvolve the recorded muscle), independently of whichspecific action and muscle were involved.The MEPratios for each condition were entered into

two separate 2 × 3 × 3 repeated-measures ANOVAs,one for each TMS delay, with muscle (FDI vs. ADM),prime (still vs. related vs. unrelated), and probe (still vs.related vs. unrelated) as within-subject variables. Separateanalyses for the two TMS delays were performed to avoidthe spurious effect of the context in which the pulse wasdelivered: In the early condition, the pulse was deliveredwhile the probe was still on the screen, whereas in the latecondition, the pulse was delivered after the probe offset,

thus confounding possible comparisons between delays.Post hoc multiple, pairwise comparisons were performedusing the Duncan test. A significance threshold of p <.05 was set for all statistical analyses. Effects sizes wereestimated using the partial eta-square measure (ηp2). Dataare reported as the mean ± SEM.

We expected that, if motor facilitation is independent ofperceptual awareness, muscle-specific motor facilitationshould be obtained in response to both probe and primeaction stimuli (main effects of prime and probe). Conver-sely, if the motor representation of observed actions isdependent on conscious processing, we expected to ob-tain motor facilitation in response to the probe but notthe prime (main effect of probe). Finally, the maskedaction prime may not trigger motor facilitation per se,but it may modulate the response of the motor system toconsciously perceived actions. In this case, we wouldexpect an effect of the prime–probe congruence onlyfor implied action probes that call for motor represen-tations, that is, only when the probe shows a movementrelated to the recorded muscle (prime × probe inter-action). Thus, the presentation of congruent versus in-congruent prime–probe pairs should be specific fortrials in which the probe shows an action that is relatedto the motor role of the recorded muscle (i.e., relatedprime and related probe pairs vs. unrelated or stillprime and related probe pairs) but not when the probeshows an action that is not related to the motor role ofthe recorded muscle (i.e., related prime and unrelatedprobe pairs vs. unrelated or still prime and unrelatedprobe pairs). This would provide evidence that uncon-scious perception of the prime affected muscle-specific,mirror-like motor facilitation.

RESULTS

Behavioral Data

Analysis of the behavioral responses of participants inthe TMS session revealed that they paid attention to thestimuli and successfully discriminated whether the probewas moving or not (the mean accuracy in each conditionranged from 89.89 ± 2.66% to 96.69 ± 1.43%). No differ-ences were obtained between the different experimentalconditions (all Fs ≤ 2.22), suggesting that participantswere equally accurate in both early and late delay trials,regardless of the prime or probe. In a similar vein, a signaldetection analysis (Macmillan & Kaplan, 1985) showedthat participants had high sensitivity levels for detectingthe probe motion when it was preceded by both still hand(d0 = 4.07 ± 0.24; one-sample t test against 0: t(16) =17.014, p < .001) and implied action primes (d0 = 3.57 ±0.19; one-sample t test against 0: t(16) = 18.62, p < .001),with the difference between the two conditions not reach-ing the significance threshold, t(16) = 1.781, p = .094.On the other hand, no response bias was obtained whenthe probe was preceded by still (ln(β) = −0.21 ± 0.63;


one-sample t test against 0: t(16) = −0.34, p = .616) andimplied action primes (ln(β) = 0.71 ± 0.38; one-samplet test against 0: t(16) = 1.82, p = .087); no difference wasobtained between the two prime conditions, t(16) =−0.909, p = .151.

The 17 participants who were entered into the analysisreported that they were unaware of the presence of theprime in the TMS session. Their overall discriminationaccuracy in the post-TMS session was 39.22% (SD =10.03%), which was not significantly different from chance(two-tailed one-sample t test against 33.33%, t(16) =−1.895, p= .076, Cohenʼs d = 0.47). Although such over-all comparison showed that participantsʼ discriminationability tended to be higher than that expected by guess-ing, effect size was small. More importantly, inspection ofthe mean accuracy values for each prime–probe pair (Ta-ble 1) revealed that such an effect was driven by apparentlyaccurate responding when the prime and the probe ac-tions were congruent but not when the probe depictedan action incongruent with the prime or a neutral, stillhand. Indeed, a 3 × 3 repeated-measures ANOVA of theparticipantsʼ prime discrimination accuracy with Probe(still vs. index finger abduction vs. little finger abduction)and Prime (still vs. index finger abduction vs. littlefinger abduction) as within-subject variables showed thatparticipantsʼ response to the prime was modulated bythe type of probe: the main effect of Probe was significant,F(2, 32) = 3.82, p = .033, ηp2 = 0.1926, and was furtherqualified by the interaction between Probe and Prime, F(4,64) = 12.32, p< .001, ηp2 = 0.4351. Post hoc tests showedthat, in still probe trials, participants were more accuratein discriminating still primes than primes showing index( p = .006) or little finger abduction movements ( p <.001). The last two conditions did not differ from oneanother ( p= .347). Furthermore, in the trials with probesshowing an index or little finger abduction, participantswere more accurate in identifying primes showing a move-ment congruent with the probe than they were for primesshowing a still hand (index finger probes: p < .001; littlefinger probes: p = .013). Participants also tended to bemore accurate for congruent than incongruent probe-prime movements, but this difference reached signifi-cance only for the index ( p < .001) and not for the littlefinger probes ( p = .062). No difference was obtained be-tween incongruent movement and still hand primes (indexfinger probes: p = .369; little finger probes: p = .461). To-gether, these results suggest that, even when participantswere informed about the presence of the prime and wereactively pursuing its identification, they tended to reportthe action showed by the probe and not that shown bythe prime. This explains greater accuracy levels for con-gruent prime–probe pairs versus incongruent ones andmarginally above-chance overall accuracy levels. These re-sults corroborate the participantsʼ subjective reports show-ing that the masked action primes were not consciouslyperceived and could not be explicitly discriminated in aforced-choice task.

Corticospinal Excitability

TheMEPs recorded from each muscle at the beginning andat the end of the experimental session were entered into a2 × 2 repeated-measures ANOVA with Muscle (FDI vs.ADM) and Session (pre- vs. post-TMS) as within-subjectvariables. A main effect of Muscle was found, F(1, 16) =6.12, p < .001, ηp2 = 0.5705: MEPs recorded from theFDI (1.811 ± 0.473 mV) were greater than MEPs recordedfrom the ADM (0.883 ± 0.131 mV, p = .025). The maineffect of Session and the interaction between Muscle andSession were not significant (all Fs < 1, ηp2 < 0.01).Table 2 shows the raw MEP amplitudes (in millivolts)

in the different experimental conditions. Inspection ofTable 2 reveals that MEP amplitude at both TMS delayswas higher during the observational task than at baseline,independently of whether participants were looking ata still hand or a hand performing a movement that waseither related or unrelated to the motor role of the re-corded muscles. This was confirmed by dependent-samplet tests (one-tailed) comparing the average of raw MEPsduring all observation conditions at each TMS delay andfor each muscle. Indeed, FDI MEP amplitude was higherduring observation than at baseline when TMS was de-livered at either early (2.067 ± 0.511 mV; t(16) = 2.8,p = .006) or late (2.039 ± 0.513 mV; t(16) = 2.52, p =.011) delay from stimulus onset; the same comparisonswith baseline for the ADMmuscle were only marginally sig-nificant (early: 1.078 ± 0.195 mV; t(16) = 1.69, p = .055;late: 1.066 ± 0.192 mV; t(16) = 1.61, p= .064). In a similarvein, dependent-sample t tests (two-tailed) showed thatno difference was obtained between MEP amplitudes re-corded at early and late TMS delays during the observationconditions for either FDI, t(16) = 0.88, p = .39, or ADM,t(16) = 0.9, p = .38, muscle. In summary, during obser-vation of hand stimuli, we found an overall increase of cor-ticospinal excitability that was not specifically associated toimplied actions as it also occurred for still hand images.This is in keeping with previous studies showing thatbody part observation activates the motor cortex more

Table 1. Accuracy (Percent Correct Responses) inDiscriminating the Prime during the PostexperimentalControl Task

Accuracy in the Prime Discrimination Task (%)

Prime

Still Index Finger Little Finger

Probe Still 66.91 ± 1.62 36.76 ± 1.54 26.1 ± 1.21

Index finger 26.47 ± 1.56 66.91 ± 1.84 16.91 ± 0.9

Little finger 27.21 ± 1.11 34.56 ± 1.99 54.41 ± 1.8

Mean (SEM ) accuracy values for each combination of prime and probepresentation are shown. “Still” indicates the presentation of a still handstimulus; “Index finger” and “Little finger” indicate the presentation ofhand with the index or little finger abducted, respectively.


than baseline trials in which participants either keep theeyes closed or fixate at a blank screen (e.g., Raos, Kilintari,& Savaki, 2013; Borgomaneri, Gazzola, & Avenanti, 2012;Hodzic, Muckli, Singer, & Stirn, 2009; Schütz-Bosbach,Mancini, Aglioti, & Haggard, 2006).

Modulation of Motor Facilitation

Previous analysis of raw MEP amplitudes showed that,as compared with viewing a blank screen, observation ofhand images increased corticospinal excitability. Becausemirror-like motor facilitation is indexed by a greater activa-tion during observation of dynamic (or implied action)bodies or body parts as compared with still bodies or bodyparts (Urgesi, Candidi, Fabbro, Romani, & Aglioti, 2006;Urgesi, Moro, et al., 2006; Fadiga et al., 2005), in the suc-cessive analysis, we tested how corticospinal excitabilitywas modulated according to the hand image that waspresented. We, thus, expressed MEP amplitude of eachmuscle as percent change (MEPratio; Figure 2) from thatmuscle baseline, averaging the baseline MEPs collected atthe beginning and at the end of the experimental session.This gave a motor facilitation index of how much themotor cortex was facilitated during observation of handimages as compared with baseline. The MEPratio valuesfor the two TMS delays were entered into two separate2 × 3 × 3 (Muscle × Prime × Probe) repeated-measuresANOVAs to test whether motor facilitation was greaterduring observation of related actions than still hands andunrelated actions when depicted in either the probe orthe prime.The ANOVA for the early TMS delay condition yielded

no significant effect (all Fs < 1.1, p > .35), suggesting thatmotor facilitation at this early delay from stimulus onset

was not specific for the action depicted in the images.The ANOVA for the late TMS delay condition yielded sig-nificant main effect of Probe, F(2, 32) = 5.6776, p = .008,ηp2 = 0.2619. The main effect of Prime was not significant,F(2,32) = 1.5, p = .239, ηp2 = 0.086, but a significant in-teraction between Prime and Probe was found, F(4, 64) =3.23, p = .017, ηp2 = 0.168, suggesting that motor facilita-tion during observation of muscle-related probe actionswas modulated by masked prime actions. The three-wayinteraction was not significant, F(4, 64) < 1, ηp2 < 0.05,suggesting comparable patterns of results for the twomuscles when movements that were related versus un-related to their motor roles were observed. Simple effectanalysis of mirror-like motor facilitation in response to re-lated as compared with still and unrelated probes showedsignificant effects for still, F(1, 16) = 9.09, p = .008, andrelated, F(1, 16) = 11.76, p = .003, prime trials but not forunrelated prime trials, F(1, 16) < 1. Thus, the observationof a probe stimulus implying a finger movement induced asignificant facilitation of the corticospinal representationof the muscle involved in the execution of that move-ment when the probe was preceded by a still (neutral)prime and by a related (congruent) prime. In contrast,no mirror-like facilitation was obtained when the probewas preceded by unrelated primes, showing comparablemotor facilitation for related than still and unrelatedprobes. Pairwise post hoc analysis of the interaction be-tween the probe and prime hands revealed no differencesamong the three prime types in the unrelated and stillprobe conditions (all ps > .102). Conversely, when theprobe was related to the recorded muscle, MEPratios inthe unrelated prime trials (117.34 ± 10.89%) were lowerthan those in the related prime trials (130.9 ± 13.08%, p=.006) and marginally lower than the MEPratios in the still

Table 2. Mean ± SEM Raw Amplitudes (in mV) of MEPs Recorded from the FDI and ADM Muscles in Each Experimental Condition

Raw Amplitude of MEPs (mV)

Early Delay Late Delay

Probe Prime FDI ADM FDI ADM

Still Still 2.08 ± 0.51 1.07 ± 0.2 2.12 ± 0.52 1.06 ± 0.19

Still Related 2.03 ± 0.5 1.05 ± 0.18 1.99 ± 0.52 1.01 ± 0.18

Still Unrelated 1.95 ± 0.46 1.1 ± 0.2 1.99 ± 0.5 1.07 ± 0.2

Related Still 2.11 ± 0.52 1.13 ± 0.21 2.11 ± 0.54 1.14 ± 0.2

Related Related 2.11 ± 0.52 1.09 ± 0.2 2.18 ± 0.52 1.1 ± 0.2

Related Unrelated 2.12 ± 0.53 1.05 ± 0.18 2.01 ± 0.49 1.06 ± 0.19

Unrelated Still 2.04 ± 0.54 1.07 ± 0.21 2.01 ± 0.52 1.06 ± 0.2

Unrelated Related 2.05 ± 0.51 1.04 ± 0.19 2.01 ± 0.51 1.03 ± 0.18

Unrelated Unrelated 2.1 ± 0.53 1.11 ± 0.2 1.93 ± 0.52 1.07 ± 0.2

“Still” indicates the presentation of a still hand stimulus; “related” indicates the presentation of a stimulus that depicts an abduction of the fingercontrolled by the recorded muscle (shown in column); “unrelated” indicates the presentation of a stimulus that depicts an abduction of the finger notcontrolled by the recorded muscle.


prime trials (126.66 ± 11.77%, p = .059); the related andstill prime trials did not differ from each other ( p = .308).Indeed, MEPratios in the related probe-unrelated primecondition were not significantly different from MEPratiosin the still and unrelated probe conditions regardless ofthe prime type (all ps > .319). Because motor facilitationfor still hands and unrelated actions reflects nonspecificactivation in response to viewing body parts, no evidenceof mirror-like motor facilitation was obtained when relatedmovement probes were preceded by primes showing amovement that is unrelated to the motor role of the re-corded muscle. Thus, presentation of an unrelated primesuppressed motor facilitation to a level that is expectedin response to observation of body parts independentlyof movement information.

DISCUSSION

In this study, we provide evidence that motor resonance,as indexed by greater motor facilitation in response torelated than static hands and unrelated actions, is notelicited by unconscious action perception but requiresperceptual awareness. Nevertheless, its expression in re-sponse to consciously perceived actions is modulated byunconscious perception of incongruent actions. We testedcorticospinal excitability while participants passively ob-served static images of either still or moving hands pre-ceded by masked primes of either the same or differentmovements. Our results show the following: (i) observa-

tion of the action probe, which was consciously perceived,engendered a somatotopic activation of the cortical repre-sentation of the muscles actually involved in the sameaction when compared with observation of a still handor of unrelated actions; (ii) subliminal perception of theaction prime did not increase motor facilitation morethan observation of a still hand, suggesting that actionsneed to be consciously perceived to evoke mirror-likemotor facilitation (motor resonance); and (iii) unconsciouspriming of actions unrelated to the motor role of therecorded muscle interfered with mirror-like motor facili-tation (motor resonance) in response to consciouslyperceived probe actions that are related to the motor roleof the recorded muscle.

Conscious Perception of Implied Actions Activatesthe Motor System

Several studies have shown that passive observation ofdynamic displays (Tomeo, Cesari, Aglioti, & Urgesi, 2013;Avenanti & Urgesi, 2011; Avenanti, Bolognini, Maravita, &Aglioti, 2007; Urgesi, Candidi, et al., 2006; Fadiga et al.,2005) and static images of bodies in action (Avenanti,Annella, et al., 2013; Urgesi et al., 2010; Proverbio, Riva,& Zani, 2009; Urgesi, Moro, et al., 2006) triggers somatoto-pic activation of the motor system and, thus, facilitates thecorticospinal representation of those muscles involvedin the perceived actions; this effect is thought to reflectmotor resonance. Hence, motor resonance is involved in

Figure 2. Effects of conscious and unconscious observation of actions. MEPratio (% of baseline) of MEPs recorded from the FDI and ADM musclesin the different prime and probe combinations. MEPs recorded from FDI and ADM are collapsed because the same pattern of results was obtainedfor the two muscles (see text). The prime and probe types were coded according to whether the hand depicted a still hand (still), a finger abductionmovement that was related to the recorded muscle (related), or a finger abduction movement that was not related to the recorded muscle (unrelated).Error bars indicate the SEM, asterisks indicate significant comparisons, and @ indicates marginally significant comparisons.


the extrapolation of dynamic information from actual aswell as implied action stimuli. Although static images wereused in this study, these images were presented in asequence of three snapshots (i.e., sample, prime, andprobe) rather than in a single static frame. This sequencemay have contributed to the apparent motion perceptionif the mask served as a temporal occluding object and thesuccessive hand postures were amodally completedinto a continuous perception of movement (Shiffrar &Freyd, 1993). If true, this interpretation is in keeping withthe notion that the amodal completion of occluded ac-tions may involve automatic motor resonance processes(Avenanti & Urgesi, 2011; Orgs, Bestmann, Schuur, &Haggard, 2011; Urgesi et al., 2010).Importantly, the different delays of TMS stimulation

allowed us to test modulation of motor facilitation duringobservation of still, related, and unrelated actions after acomparable delay (307 msec) from prime (early delay)and probe (late delay) presentation. Thus, we could testmuscle-specific, mirror-like facilitation separately forunconsciously and consciously perceived stimuli. MEGstudies (Nishitani, Avikainen, & Hari, 2004; Nishitani &Hari, 2000) have found that the responses to action obser-vation peak first in the visual occipital areas (at approxi-mately 118 msec from the onset of the stimulus) beforesubsequent activation of the STS, inferior parietal lobule,inferior frontal areas, and, finally, M1 at approximately300 msec. Similarly, a recent TMS study (Barchiesi &Cattaneo, 2012) found that muscle-specific effects in M1were observed only 250 msec after action stimuli onset.Therefore, any direct motor response to unconsciousprime perception (i.e., main effect of prime) was likelyto occur in the early TMS delay, because in that case pro-cessing of the prime action, but not of the probe action,was likely to have reached the motor cortex. In contrast,we did not find any modulation of the corticospinal ex-citability for the unconsciously perceived prime at theearly TMS delay independently of the type of probe pre-sented nor did we find any effect of prime presentationwhen the probe depicted a still hand or an unrelatedaction, thus ruling out that unconscious prime action per-ception increased motor facilitation in a muscle-specificfashion more than observation of a still hand. Conversely,the late TMS delay, when the TMS pulse was deliveredafter 307 msec from probe onset, allowed us to show thatmuscle-specific motor responses are facilitated for theconsciously perceived probe. In summary, our finding thatmotor facilitation in response to consciously perceivedprobe actions was obtained at the late TMS delay corrobo-rates the timing of cortical activations reported in previousMEG and TMS studies and suggests a late involvement ofM1 in mapping observed (implied) actions.

Motor Resonance Requires Perceptual Awareness

The main aim of this study was to test whether perceptualawareness is an essential condition for the emergence of

motor resonance responses, measured as a facilitation ofcorticospinal excitability during observation of actionsthat are related to the motor role of the recorded muscleas compared with observation of a still hand or of anaction that is unrelated to the motor role of the recordedmuscle. The forward and backward masking procedureallowed us to present the prime for a relatively long time(53 msec) but nonetheless prevented conscious percep-tion by most participants (17/22 individuals). Indeed, evenwhen these participants were explicitly informed aboutthe presence of the prime and were required to discrimi-nate whether it depicted a still hand or an index or littlefinger abduction, their discrimination abilities were atchance. This suggests that manipulation of the voluntarycontrol of the attentional focus did not affect the percep-tion of the prime, because prime discrimination was notpossible even when participants were actively pursuingit. This does not exclude, however, that presentation ofthe probe stimulus may have contributed to prevent con-scious perception of the prime with automatic attentionalshift. Whether the locus of prime perception disruptionwas at a perceptual or attentional level, our paradigmensured that participants remained completely unawareof the prime action.

In striking contrast with the muscle-specific responseto the probe, no motor facilitation was measured inresponse to the presentation of the prime (independentlyof the probe type and TMS delay). The absence of motorresponses to masked action primes cannot be ascribed tothe interferential effect of probe movement presentationbecause M1 responses to the prime in the still hand probeconditions were also absent. Nor can this absence re-flect that the presentation of either static or implied actionprobes completely suppressed the early processing of themasked prime stimulus in the visual system, thus prevent-ing the result of this processing to be forwarded to themotor system. Indeed, we have evidence that the primewas processed to a certain extent and the result of thisprocessing affected the activation of the motor system,because the presentation of incongruent action primessuppressed the motor resonance to implied action probes(see below). Furthermore, it is also unlikely that theabsence of motor responses to masked action primeswas because of inappropriate timing of the TMS pulses;indeed, in both the early and late TMS delay conditions,corticospinal excitability was tested after the 250 msecthreshold for the establishment of muscle-specific motorresonance responses to the prime (Barchiesi & Cattaneo,2012). Given our experimental settings, however, we can-not exclude the possibility that masked action stimuli mayhave modulated motor cortex activity at earlier phases ofstimulus processing, for example, in the 60–90 msec timewindow in which a nonspecific activation of the motorcortex in response to action stimuli has been reported(Lepage, Tremblay, & Théoret, 2010). However, becausethis early activation lacks muscle specificity, it is unlikelyto reflect motor resonance. Together, our data suggest


that motor resonance requires awareness of the actionstimuli.

Action stimuli with an emotional valence are knownto activate the motor system in conditions of reducedvisibility or when participants are not aware of the stimuli.Emotional face and body expressions induce fast facialmimicry responses (Tamietto et al., 2009; i.e., EMG preac-tivation of the muscles involved in the same expressions)and motor and premotor cortical activations (Van denStock et al., 2011) even when presented in the blindhemifield of hemianopic patients or in masked presenta-tion conditions in healthy individuals (Dimberg et al.,2000). Such affective facial mimicry responses, however,are not specific for the observed body parts because theyalso occur in response to whole-body emotional expres-sions and lack the muscle-specific properties of motorresonance responses. Thus, these responses seem to bemediated by the activation of subcortical structures in re-sponse to emotionally valenced stimuli and are not involvedin nonemotional actions (Tamietto & De Gelder, 2010).

Unconscious Presentation of Incongruent ActionsInhibits Motor Resonance

Although the subliminal presentation of masked primeactions did not trigger motor facilitation per se, it affectedthe activation of the motor cortex in response to con-sciously perceived actions. In other words, masked primeactions that were unrelated to the motor role of therecorded muscle interfered with motor facilitation inresponse to probe actions that were related to the motorrole of the recorded muscle. This evidence of impliciteffects of unconscious action perception on motor reso-nance suggests the dissociation between visual aware-ness and the perceptual effects of masked action primes(Van den Bussche et al., 2009). Thus, in keeping with thefinding that unconscious prime perceptionmay affect motorexecution processes during visuomotor masked primingtasks (Eimer & Schlaghecken, 1998, 2003), we show evi-dence that masked primes can affect motor resonance.

The suppression of motor facilitation during actionobservation is unlikely to involve a specific exogenousinhibition of motor activation induced by the abrupt in-terruption of stimulus presentation, as reported in visuo-motor masked priming tasks (Eimer & Schlaghecken,1998, 2003). Indeed, motor facilitation was suppressedfor incongruent presentations of prime and probe move-ments, whereas visuomotor masked primes tend to in-hibit the execution of congruent actions. Because nofacilitatory or inhibitory responses to the prime wereobtained with neutral (i.e., still hand) probes, the inter-action between the probe and prime does not seem tobe because of the orienting of spatial attention to theone or to the other side of the space. Accordingly, pre-vious single-pulse TMS studies have shown that motorfacilitation during action observation is based on mappingthe specific muscle involved in the actual execution of the

action rather than on the spatial compatibility between themodelʼs and observerʼs body part (Alaerts, Van Aggelpoel,Swinnen, & Wenderoth, 2009; Urgesi, Candidi, et al.,2006). Furthermore, behavioral studies have shown thatvisuomotor interference effects occurred when the ob-served and executed movements involved different hands(i.e., left vs. right hand; Catmur & Heyes, 2011; Bertenthal,Longo, & Kosobud, 2006; Brass et al., 2000) or thesame hand viewed from different perspectives (egocentricvs. allocentric perspective; Bortoletto, Mattingley, &Cunnington, 2013) and, thus, occurred in the oppositesides of the space. Even when visuomotor interactionwas affected by the spatial compatibility between the exe-cuted and observedmovement, this resulted into inversionof the effect (i.e., facilitation for incongruent actions) ratherthan into its suppression (Vainio & Mustonen, 2011). Assuch, the suppression of mirror-like (i.e., muscle-specific)facilitation obtained in this study is unlikely to be becauseof the effects of spatial attention. In a similar vein, it seemsuntenable a pure attentional account for the suppres-sion of motor resonance to the probe, according to whichthe presentation of the incongruent prime may havediverted the focus of attention away from processing theprobe action. Indeed, on the one hand, the prime wasnot perceived consciously and could not be discriminatedeven when the focus of attention of the observer wasexplicitly focused on its presentation, thus ruling out a rolefor voluntary control of attention. On the other hand,although prime presentation may have influenced probeprocessing by causing automatic attentional shift, theresults of the online task suggest that participants couldeasily detect the probe action independently from theaction shownby the prime, thus ruling out that the primedisrupted conscious perception of the probe at an extentcapable of explaining the complete suppression of mirror-like motor facilitation. Accordingly, previous studiesshowed that attentional modulation of action percep-tion may modulate, but not suppress, motor resonance(Muthukumaraswamy & Singh, 2008). Thus, it is likelythe effects of unconscious perception of prime actions onmotor resonance to probe actions were at least partiallyindependent from attentional modulation processes.The present single-pulse TMS study is not informative

regarding the possible neural site of the interaction be-tween subliminal and supraliminal action processing. Wecan speculate that such interaction likely involves an areaor set of areas that have direct or indirect connections tothe primary motor cortex and that process both supra- andsubliminal presentations of actions, thus explaining the ef-fects of the masked action prime. Considering the patternof results, two possible scenarios can be hypothesized.On the one hand, the facilitation induced by supraliminalprobe actions and its suppression by subliminal incongru-ent prime actions may be because of separate mechanismsoperating within the motor system. A possible candidateis the ventral premotor cortex, which is strongly con-nected to M1 (Prabhu et al., 2009) and seems to be the


main source of the motor resonance facilitation responsesobserved in M1 (Avenanti, Annella, et al., 2013; Avenanti,Candidi, & Urgesi, 2013; Avenanti et al., 2007). There isevidence of pyramidal tract neurons in themacaque ventralpremotor cortex (Kraskov, Dancause, Quallo, Shepherd, &Lemon, 2009; but see a similar finding in M1; Vigneswaran,Philipp, Lemon, & Kraskov, 2013) that may exhibit one oftwo main patterns of activity: whereas one pattern consistsin facilitation of the hand motor neurons that mirrors theobserved action and corresponds to what is generallycalled motor resonance, the other consists in a numberof neurons being suppressed during action observation.This suppression is thought to prevent overt movementsduring action observation, but because the same neuronsshowed inhibitory responses in a variety of conditions(e.g., grip with object, grip without object, concealed grip)it may be that this inhibitory mechanism is less selectivewith regard to the specific action observed when comparedwith classical mirror facilitation. Importantly, facilitationand suppression neurons seem to be part of two parallelmechanisms in the ventral premotor cortex, and theirpossible differential sensitivity to perceptual awarenessand attentional modulation may explain the pattern ofresults obtained in this study in terms of facilitation ofcorticospinal excitability. Indeed, assuming that suppres-sion neurons may respond to observed actions indepen-dently of perceptual awareness, suppression of motoractivity in response to the prime and/or probe actionsmay remain unnoticed in a gross measure of corticospinalexcitability as done with TMS-evoked motor potentials.Nevertheless, the suppression of their activity after presen-tation of two incongruent actions, which is likely strongeras compared with presentation of a single action, maycontribute to suppress the expression of facilitation neu-ron activity in response to consciously perceived probeactions.On the other hand, the suppression of motor facilitation

for incongruent prime–probe actions may stem specificallyfrom the processing of the conflict between two incon-gruent action representations in areas beyond the motorcortex but directly or indirectly connected to it. The detec-tion of a conflict in these areas may generate a signal thatis sent to the motor system to suppress motor resonancein response to consciously perceived actions. Unconsciousprocessing of the conflict between different action-relatedcues may involve several cortical and subcortical regions(DʼOstilio & Garraux, 2012). For example, the pre-SMAis involved in exerting control over voluntary actions dur-ing response conflict (Nachev, Wydell, Oʼneill, Husain, &Kennard, 2007) and may also subserve the suppressionof incongruent motor resonance responses. Furthermore,controlling the automatic tendency to imitate otherʼs ac-tions, as tested in the visuomotor priming paradigm, in-volves activation in the ventromedial pFC and the TPJ(Spengler, Brass, Kühn, & Schütz-Bosbach, 2010; Brass,Ruby, & Spengler, 2009). These areas are implicated inhigher-level mentalizing abilities, such as the sense of

agency, perspective taking, and self-referential processing(Brass et al., 2009) and seem to be also involved in prevent-ing the coding of othersʼ movements to be translated intoself-movements. Finally, inhibitory afferents to M1 viacorticostriatal pathways play a role in masked visuomotorpriming effects (DʼOstilio & Garraux, 2012; Bowman,Schlaghecken, & Eimer, 2006) and may also be involvedduring action observation (Marceglia et al., 2009) and,especially, for suppressing unwanted motor resonanceresponses.

In conclusion, we show that, although motor resonancein response to nonemotional actions is automatic, it re-quires the perceptual awareness of actions and operatesat a conscious level of motor representation. Althoughthe presentations of subliminal actions do not activatethe motor cortex, they suppress motor resonance re-sponses to incongruent actions that are consciously per-ceived. We suggest that this suppression reflects theprocessing of action conflicts in areas upstream of themotor cortex and may subserve a basic mechanism fordealing with the multiple and possibly incongruent ac-tions of other individuals. Motor resonance responses,indeed, seem to endow the social brain with the abilityof simulating anticipatory representations of unfoldingactions ahead of their realization and in the absenceof complete perceptual information from the environ-ment (Avenanti, Candidi, et al., 2013; Avenanti & Urgesi,2011; Wilson & Knoblich, 2005). Such representations,however, need to be flexibly updated on the basis ofupcoming information, to take into account not onlyactual deployment of actions but also their future phasesand to be modulated by the social context toward imita-tive or complimentary responses (Sartori et al., 2013).The results of this study suggest that modulation ofmotor resonance according to the social context (in thiscase incongruent actions) may also occur at an implicitlevel and independently from the observerʼs perceptualawareness.

Acknowledgments

This research was supported by grants from Istituto Italiano diTecnologia SEED 2009 (Prot. No. 21538; to C. U.), from theMinistero Istruzione Università e Ricerca (Progetti di Ricerca diInteresse Nazionale, PRIN 2009; Prot. No. 2009A8FR3Z; FuturoIn Ricerca, FIR 2012, Prot. N. RBFR12F0BD; to C. U.), and fromIstituto di Ricovero e Cura a Carattere Scientifico “E. Medea”(Ricerca Corrente 2012, Ministero Italiano della Salute; to C. U.).

Reprint requests should be sent to Cosimo Urgesi, Departmentof Human Sciences, University of Udine, via Margreth, 3, I-33100Udine, Italy, or via e-mail: [email protected].

REFERENCES

Alaerts, K., Heremans, E., Swinnen, S. P., & Wenderoth, N.(2009). How are observed actions mapped to the observerʼsmotor system? Influence of posture and perspective.Neuropsychologia, 47, 415–422.


Alaerts, K., Van Aggelpoel, T., Swinnen, S. P., & Wenderoth, N.(2009). Observing shadow motions: Resonant activity withinthe observerʼs motor system? Neuroscience Letters, 461,240–244.

Avenanti, A., Annella, L., Candidi, M., Urgesi, C., & Aglioti,S. M. (2013). Compensatory plasticity in the actionobservation network: Virtual lesions of STS enhanceanticipatory simulation of seen actions. Cerebral Cortex,23, 570–580.

Avenanti, A., Bolognini, N., Maravita, A., & Aglioti, S. M. (2007).Somatic and motor components of action simulation.Current Biology, 17, 2129–2135.

Avenanti, A., Candidi, M., & Urgesi, C. (2013). Vicarious motoractivation during action perception: Beyond correlationalevidence. Frontiers in Human Neuroscience, 7, 185.

Avenanti, A., & Urgesi, C. (2011). Understanding “what” othersdo: Mirror mechanisms play a crucial role in action perception.Social Cognitive & Affective Neuroscience, 6, 257–259.

Barchiesi, G., & Cattaneo, L. (2012). Early and late motorresponses to action observation. Social Cognitive &Affective Neuroscience, 8, 711–719.

Bargh, J. A., Schwader, K. L., Hailey, S. E., Dyer, R. L., &Boothby, E. J. (2012). Automaticity in social-cognitiveprocesses. Trends in Cognitive Science, 16, 593–605.

Bertenthal, B. I., Longo, M. R., & Kosobud, A. (2006).Imitative response tendencies following observation ofintransitive actions. Journal of Experimental Psychology:Human Perception and Performance, 32, 210–225.

Borgomaneri, S., Gazzola, V., & Avenanti, A. (2012). Motormapping of implied actions during perception of emotionalbody language. Brain Stimulation, 5, 70–76.

Bortoletto, M., Mattingley, J. B., & Cunnington, R. (2013).Effects of context on visuomotor interference depends onthe perspective of observed actions. Plos One, 8, e53248.

Bowman, H., Schlaghecken, F., & Eimer, M. (2006). A neuralnetwork model of inhibitory processes in subliminal priming.Visual Cognition, 13, 401–480.

Brasil-Neto, J. P., Cohen, L. G., Panizza, M., Nilsson, J., Roth, B. J.,& Hallett, M. (1992). Optimal focal transcranial magneticactivation of the humanmotor cortex: Effects of coil orientation,shape of the induced current pulse, and stimulus intensity.Journal of Clinical Neurophysiology, 9, 132–136.

Brass, M., Bekkering, H., & Prinz, W. (2001). Movementobservation affects movement execution in a simpleresponse task. Acta Psychologica, 106, 3–22.

Brass, M., Bekkering, H., Wohlschläger, A., & Prinz, W.(2000). Compatibility between observed and executedfinger movements: Comparing symbolic, spatial, andimitative cues. Brain and Cognition, 44, 124–143.

Brass, M., Ruby, P., & Spengler, S. (2009). Inhibition of imitativebehaviour and social cognition. Philosophical Transactionsof the Royal Society, Series B, Biological Sciences, 364,2359–2367.

Briggs, G. G., & Nebes, R. D. (1975). Patterns of handpreference in a student population. Cortex, 11, 230–238.

Catmur, C., & Heyes, C. (2011). Time course analysesconfirm independence of imitative and spatial compatibility.Journal of Experimental Psychology: Human Perceptionand Performance, 37, 409–421.

Catmur, C., Mars, R. B., Rushworth, M. F., & Heyes, C. (2011).Making mirrors: Premotor cortex stimulation enhancesmirror and counter-mirror motor facilitation. Journal ofCognitive Neuroscience, 23, 2352–2362.

Craighero, L., Bello, A., Fadiga, L., & Rizzolatti, G. (2002).Hand action preparation influences the responses to handpictures. Neuropsychologia, 40, 492–502.

Craighero, L., Fadiga, L., & Rizzolatti, G. (1999). Action forperception: A motor-visual attentional affect. Journal

of Experimental Psychology: Human Perception andPerformance, 25, 1673–1692.

Craighero, L., Fadiga, L., Umiltà, C. A., & Rizzolatti, G. (1996).Evidence for visuomotor priming effect. NeuroReport, 8,347–349.

Dehaene, S., Naccache, L., Le ClecʼH, G., Koechlin, E.,& Mueller, M. (1998). Imaging unconscious semanticpriming. Nature, 395, 597–600.

Di Lazzaro, V., Oliviero, A., Profice, P., Saturno, E., Pilato, F.,Insola, A., et al. (1998). Comparison of descending volleysevoked by transcranial magnetic and electric stimulation inconscious humans. Electroencephalography and ClinicalNeurophysiology, 109, 397–401.

Dimberg, U., Thunberg, M., & Elmehed, K. (2000). Unconsciousfacial reactions to emotional facial expressions. PsychologicalScience, 11, 86–89.

DʼOstilio, K., & Garraux, G. (2012). Dissociation betweenunconscious motor response facilitation and conflict inmedial frontal areas. European Journal of Neuroscience,35, 332–340.

Eimer, M., & Schlaghecken, F. (1998). Effects of masked stimulion motor activation: Behavioral and electrophysiologicalevidence. Journal of Experimental Psychology: HumanPerception and Performance, 24, 1737–1747.

Eimer, M., & Schlaghecken, F. (2003). Response facilitationand inhibition in subliminal priming. Biological Psychology,64, 7–26.

Fadiga, L., Craighero, L., & Olivier, E. (2005). Human motorcortex excitability during the perception of othersʼ action.Current Opinion in Neurobiology, 15, 213–218.

Fadiga, L., Fogassi, L., Pavesi, G., & Rizzolatti, G. (1995). Motorfacilitation during action observation: A magnetic stimulationstudy. Journal of Neurophysiology, 73, 2608–2611.

Hodzic, A., Muckli, L., Singer, W., & Stirn, A. (2009). Corticalresponses to self and others. Human Brain Mapping, 30,951–962.

Hogeveen, J., & Obhi, S. S. (2013). Automatic imitation isautomatic, but less so for narcissists. Experimental BrainResearch, 224, 613–621.

Kilner, J. M., Paulignan, Y., & Blakemore, S. J. (2003). Aninterference effect of observed biological movement onaction. Current Biology, 13, 522–525.

Kraskov, A., Dancause, N., Quallo, M. M., Shepherd, S., &Lemon, R. N. (2009). Corticospinal neurons in macaqueventral premotor cortex with mirror properties: A potentialmechanism for action suppression? Neuron, 64, 922–930.

Lepage, J. F., Tremblay, S., & Théoret, H. (2010). Earlynon-specific modulation of corticospinal excitability duringaction observation. European Journal of Neuroscience,31, 931–937.

Macmillan, N. A., & Kaplan, H. L. (1985). Detection theoryanalysis of group data: Estimating sensitivity from averagehit and false-alarm rates. Psychological Bulletin, 98, 185–199.

Maeda, F., Kleiner-Fisman, G., & Pascual-Leone, A. (2002).Motor facilitation while observing hand actions: Specificityof the effect and role of observerʼs orientation. Journalof Neurophysiology, 87, 1329–1335.

Marceglia, S., Fiorio,M., Foffani, G.,Mrakic-Sposta, S., Tiriticco,M.,Locatelli, M., et al. (2009). Modulation of beta oscillationsin the subthalamic area during action observation inParkinsonʼs disease. Neuroscience, 161, 1027–1036.

McGettigan, C., Warren, J. E., Eisner, F., Marshall, C. R.,Shanmugalingam, P., & Scott, S. K. (2012). Neural correlatesof sublexical processing in phonological working memory.Journal of Cognitive Neuroscience, 23, 121–143.

Merikle, P. M., & Daneman, M. (1998). Psychologicalinvestigations of unconscious perception. Journal ofConsciousness Studies, 5, 5–18.


Merikle, P. M., Smilek, D., & Eastwood, J. D. (2001). Perceptionwithout awareness: Perspectives from cognitive psychology.Cognition, 79, 115–134.

Muthukumaraswamy, S. D., & Singh, K. D. (2008). Modulationof the human mirror neuron system during cognitiveactivity. Psychophysiology, 45, 896–905.

Nachev, P., Wydell, H., Oʼneill, K., Husain, M., & Kennard, C.(2007). The role of the pre-supplementary motor area inthe control of action. Neuroimage, 36, T155–T163.

Nishitani, N., Avikainen, S., & Hari, R. (2004). Abnormalimitation-related cortical activation sequences in Aspergerʼssyndrome. Annals of Neurology, 55, 558–562.

Nishitani, N., & Hari, R. (2000). Temporal dynamics of corticalrepresentation of action. Proceedings of the NationalAcademy of Sciences, U.S.A., 97, 913–918.

Orgs, G., Bestmann, S., Schuur, F., & Haggard, P. (2011). Frombody form to biological motion: The apparent velocity ofhuman movement biases subjective time. PsychologicalScience, 22, 712–717.

Prabhu, G., Shimazu, H., Cerri, G., Brochier, T., Spinks, R. L.,Maier, M. A., et al. (2009). Modulation of primary motorcortex outputs from ventral premotor cortex during visuallyguided grasp in the macaque monkey. The Journal ofPhysiology, 587, 1057–1069.

Proverbio, A. M., Riva, F., & Zani, A. (2009). Observation ofstatic pictures of dynamic actions enhances the activity ofmovement-related brain areas. PLoS One, 4, e5389.

Raos, V., Kilintari, M., & Savaki, H. E. (2013). Viewing a forelimbinduces widespread cortical activations. Neuroimage, 89C,122–142.

Rizzolatti, G., & Craighero, L. (2004). The mirror-neuronsystem. Annual Review of Neuroscience, 27, 169–192.

Sartori, L., Bucchioni, G., & Castiello, U. (2012). Motor cortexexcitability is tightly coupled to observed movements.Neuropsychologia, 50, 2341–2347.

Sartori, L., Bucchioni, G., & Castiello, U. (2013). Whenemulation becomes reciprocity. Social Cognitive &Affective Neuroscience, 8, 662–669.

Sartori, L., & Castiello, U. (2013). Shadows in the mirror.NeuroReport, 24, 63–67.

Schütz-Bosbach, S., Mancini, B., Aglioti, S. M., & Haggard, P.(2006). Self and other in the human motor system.Current Biology, 16, 1830–1834.

Shiffrar, M., & Freyd, J. J. (1993). Timing and apparent motionpath choice with human body photographs. PsychologicalScience, 4, 379–384.

Spengler, S., Brass, M., Kühn, S., & Schütz-Bosbach, S.(2010). Minimizing motor mimicry by myself: Self-focusenhances online action-control mechanisms during motorcontagion. Consciousness & Cognition, 19, 98–106.

Spengler, S., von Cramon, D. Y., & Brass, M. (2010). Resistingmotor mimicry: Control of imitation involves processescentral to social cognition in patients with frontal andtemporo-parietal lesions. Social Neuroscience, 5, 401–416.

Stürmer, B., Aschersleben, G., & Prinz, W. (2000).Correspondence effects with manual gestures andpostures: A study of imitation. Journal of ExperimentalPsychology: Human Perception and Performance, 26,1746–1759.

Tamietto, M., Castelli, L., Vighetti, S., Perozzo, P., Geminiani, G.,& Weiskrantz, L. (2009). Unseen facial and bodily expressionstrigger fast. Proceedings of the National Academy ofSciences, U.S.A., 106, 17661–17666.

Tamietto, M., & De Gelder, B. (2010). Neural bases of thenon-conscious perception of emotional signals. NatureReviews Neuroscience, 11, 697–709.

Théoret, H., Halligan, E., Kobayashi, M., Merabet, L., &Pascual-Leone, A. (2004). Unconscious modulation of motorcortex excitability revealed with transcranial magneticstimulation. Experimental Brain Research, 155, 261–264.

Tomeo, E., Cesari, P., Aglioti, S. M., & Urgesi, C. (2013).Fooling the kickers but not the goalkeepers: Behavioraland neurophysiological correlates of fake action detectionin soccer. Cerebral Cortex, 23, 2765–2778.

Urgesi, C., Candidi, M., Fabbro, F., Romani, M., & Aglioti,S. M. (2006). Motor facilitation during action observation:Topographic mapping of the target muscle and influenceof the onlookerʼs posture. European Journal ofNeuroscience, 23, 2522–2530.

Urgesi, C., Maieron, M., Avenanti, A., Tidoni, E., Fabbro, F.,& Aglioti, S. M. (2010). Simulating the future of actionsin the human corticospinal system. Cerebral Cortex, 20,2511–2521.

Urgesi, C., Moro, V., Candidi, M., & Aglioti, S. M. (2006).Mapping implied body actions in the human motorsystem. Journal of Neuroscience, 26, 7942–7949.

Vainio, L., & Mustonen, T. (2011). Mapping the identity of aviewed hand in the motor system: Evidence from stimulus-response compatibility. Journal of Experimental Psychology:Human Perception and Performance, 37, 207–221.

Vainio, L., Tucker, M., & Ellis, R. (2007). Precision and powergrip priming by observed grasping. Brain Cognition, 65,195–207.

Van den Bussche, E., Van den Noortgate, W., & Reynvoet, B.(2009). Mechanisms of masked priming: A meta-analysis.Psychological Bulletin, 135, 452–477.

Van den Stock, J., Tamietto, M., Sorger, B., Pichon, S.,Grézes, J., & De Gelder, B. (2011). Cortico-subcorticalvisual, somatosensory, and motor activations for perceivingdynamic whole-body emotional expressions with andwithout striate cortex (V1). Proceedings of the NationalAcademy of Sciences, U.S.A., 108, 16188–16193.

Vigneswaran, G., Philipp, R., Lemon, R. N., & Kraskov, A.(2013). M1 corticospinal mirror neurons and their rolein movement suppression during action observation.Current Biology, 23, 236–243.

Vogt, S., Taylor, P., & Hopkins, B. (2003). Visuomotorpriming by pictures of hand postures: Perspective matters.Neuropsychologia, 41, 941–951.

Wasserman, E. M. (1998). Risk and safety of repetitivetranscranial magnetic stimulation: Report and suggestedguidelines from the International Workshop on theSafety of Repetitive Transcranial Magnetic Stimulation,June 5–7, 1996. Electroencephalography and ClinicalNeurophysiology, 108, 1–16.

Wilson, M., & Knoblich, G. (2005). The case for motorinvolvement in perceiving conspecifics. PsychologicalBulletin, 131, 460–473.


Conscious and Unconscious Representations of Observed Actions in the Human Motor System.

Documents

Transcript of Conscious and Unconscious Representations of Observed Actions in the Human Motor System.