eeg mu rhythm and imitation impairments in individuals with autism spectrum disorder

www.elsevier.com/locate/b&c

Brain and Cognition 64 (2007) 228–237

EEG mu rhythm and imitation impairments in individualswith autism spectrum disorder

R. Bernier a,b,*, G. Dawson a,b,c,d, S. Webb b,c,d, M. Murias b,c,d

a Department of Psychology, University of Washington, USAb UW Autism Center, University of Washington, USA

c Center on Human Development and Disabilities, University of Washington, USAd Department of Psychiatry and Behavioral Sciences, University of Washington, USA

Accepted 18 March 2007Available online 23 April 2007

Abstract

Imitation ability has consistently been shown to be impaired in individuals with autism. A dysfunctional execution/observationmatching system has been proposed to account for this impairment. The EEG mu rhythm is believed to reflect an underlying execu-tion/observation matching system. This study investigated evidence of differential mu rhythm attenuation during the observation, exe-cution, and imitation of movements and examined its relation to behaviorally assessed imitation abilities. Fourteen high-functioningadults with autism spectrum disorder (ASD) and 15 IQ- and age-matched typical adults participated. On the behavioral imitation task,adults with ASD demonstrated significantly poorer performance compared to typical adults in all domains of imitation ability. On theEEG task, both groups demonstrated significant attenuation of the mu rhythm when executing an action. However, when observingmovement, the individuals with ASD showed significantly reduced attenuation of the mu wave. Behaviorally assessed imitation skillswere correlated with degree of mu wave attenuation during observation of movement. These findings suggest that there is execution/observation matching system dysfunction in individuals with autism and that this matching system is related to degree of impairmentin imitation abilities.� 2007 Elsevier Inc. All rights reserved.

Keywords: Autism; Imitation; EEG; Mu rhythm; Mirror neurons

1. Introduction

Imitative deficits in individuals with autism have consis-tently been observed (Williams, Whiten, & Singh, 2004). Infact, several researchers have suggested that imitation def-icits are one of the core impairments of autism spectrumdisorders (ASD; Dawson & Adams, 1984; Dawson &Lewy, 1989; Rogers & Pennington, 1991; Williams,Whiten, Suddendorf, & Perrett, 2001). Imitation impair-ments in autism were first reported by DeMyer and col-

0278-2626/$ - see front matter � 2007 Elsevier Inc. All rights reserved.

doi:10.1016/j.bandc.2007.03.004

* Corresponding author. Address: Box 357920, Center on HumanDevelopment and Disability, University of Washington, Seattle WA98195, USA. Fax: +1 206 543 5771.

E-mail address: [email protected] (R. Bernier).

leagues (DeMeyer et al., 1972) and in the thirty yearssince this initial report, over twenty studies have examinedvarying aspects of imitation in ASD. Most of these studieshave focused on imitation impairments in children withASD. Children with autism under 6 years of age showimpaired imitation skills compared to children with mentalretardation, developmental delay, and communication dis-orders as well as compared to typically developing children(Aldridge, Stone, Sweeney, & Bower, 2000; Charman et al.,1997; Charman et al., 1998; Dawson, Meltzoff, Osterling, &Rinaldi, 1998; Rogers, Hepburn, Stackhouse, & Wehner,2003; Sigman & Ungerer, 1984; Stone, Lemanek, Fishel,Fernandez, & Altemeier, 1990; Stone, Ousley, & Littleford,1997). School age children with autism also show imitationdeficits (Green et al., 2002; Hammes & Langdell, 1981;

mailto:[email protected]

R. Bernier et al. / Brain and Cognition 64 (2007) 228–237 229

Jones & Prior, 1985; Ohta, 1987; Smith & Bryson, 1998).Such impairments persist into adolescence and adulthoodalthough there are fewer studies that have examined thisage group (Avikainen, Wohlschlager, Liuhanen, Hanninen,& Hari, 2003; Hobson & Lee, 1999; Rogers, Bennetto,McEvoy, & Pennington, 1996). Hobson and Lee (1999)found that teens with autism were able to imitate goaldirected actions but failed to imitate the style in whichthe examiner performed the task with greater frequencythan the control group. Further, during the imitative acts,the subjects with autism made considerable reversal errors,errors in which the imitator replicates the gestures but inthe perspective of how the gesture was observed. Similarly,Avikainen et al. (2003) examined imitation in adultsthrough a task in which participants imitated the experi-menter’s actions using either the same hand or oppositehand. They found that while typical adults used the mirrorimage of the experimenter’s actions to increase responsespeed and accuracy, the individuals with Asperger syn-drome and high functioning autism failed to capitalize onthe mirror image.

It has been proposed that a deficit in self other mappingis the most parsimonious explanation for the imitationimpairments found in ASD (Williams et al., 2004). Theself-other mapping deficit hypothesis, which posits that abiological dysfunction of cross modal processes preventsthe individual with autism from forming and coordinatingrepresentations of the self and others, was earlier presentedby Meltzoff and Gopnik (1993) and Rogers and Penning-ton (1991). Evidence from infant imitation work suggeststhat cross modal processes necessitate an execution/obser-vation matching system (Meltzoff & Decety, 2003; Meltzoff& Moore, 1997). That is, in order for an infant to imitatefaces, the infant must cross modally map an observed facialexpression with his or her own motor execution of thatexpression. But the infant cannot see his or her face andso must have a supramodal internal representation thatenables a match between the observed and executedexpression.

1.1. Mu activity and the execution/observation system

The first neurophysiological evidence of an execution/observation matching system in humans comes fromEEG work conducted in the 1950s which focused on themu rhythm band oscillations. The EEG mu rhythm bandfalls between 8 and 13 Hz and is recorded from scalp elec-trodes over the sensorimotor cortex. Attenuation of restingEEG mu rhythm reflects desynchronization of the underly-ing cell assemblies, which suggests an increased load onthose cells (Pfurtscheller, Neuper, Andrew, & Edlinger,1997). Attenuation of the resting EEG mu rhythm wasobserved in adult subjects in response to the execution ofactions, both passive or reflex movements (Chatrian, Peter-sen, & Lazarte, 1959). Attenuation was also observed inadults who were watching others’ movements (boxing) onfilm (Gastaut & Bert, 1954). More recently, the finding of

mu wave attenuation when observing and executing actionshas been replicated in both adults (Arroyo et al., 1993;Babiloni et al., 2002; Babiloni et al., 1999; Babiloni et al.,2003; Cochin, Barthelemy, Lejeune, Roux, & Martineau,1998; Cochin, Barthelemy, Roux, & Martineau, 1999;Muthukumaraswamy, Johnson, & McNair, 2004; Muthu-kumaraswamy & Johnson, 2004; Pfurtscheller et al.,1997) and children (Cochin, Barthelemy, Rous, & Marti-neau, 2001; Lepage & Theoret, 2006; Martineau & Cochin,2003; Stroganova, Orekhova, & Posikera, 1999). Mu waveattenuation during observation of action is strictly a centralphenomenon. That is, movement in the body does notaccount for the mu wave attenuation when observingother’s actions. In two studies of typical adults, Muthu-kumaraswamy and colleagues determined that subtle activ-ity in the hand did not account for the reduced attenuationwhen observing human movement (Muthukumaraswamy& Johnson, 2004; Muthukumaraswamy et al., 2004). Fur-ther, imagining movement is sufficient to attenuate themu rhythm (Pfurtscheller, Brunner, Schlogl, & Lopes deSilva, 2006). In adults mu wave attenuation was found dur-ing the imagining of left and right hand motor movements.These consistent findings of attenuation of the mu waveduring the observation or imagining of human movement,as well as during the execution of movement, suggest thatthe mu wave reflects an execution/observation matchingsystem (Pineda, 2005).

In a study of 10 individuals with autism and typicalindividuals (6–47 years of age), Oberman and colleagues(2005) reported attenuation of the mu rhythm in the 8–13 Hz band during the execution and observation condi-tions for the typical group, whereas the autism groupshowed attenuation only during the execution condition.They concluded that the results were consistent with exe-cution/observation system dysfunction in individuals withautism.

Although individuals with autism appear to fail to showattenuation of mu activation during observation andclearly show behavioral impairments in imitation, the rela-tion between these to findings is unclear. This study aims tofurther explore the imitation deficits in a sample of highfunctioning adults with autism including the relationbetween behaviorally assessed imitative skills and theEEG index of the execution/observation matching system.It is hypothesized that adults with autism will demonstrateimitation impairments as well as reduced mu wave attenu-ation when observing biological movement. We furtherpropose that degree of EEG mu wave attenuation will becorrelated with level of imitative ability.

2. Methods

2.1. Participants

The original sample of participants included 17 adultmales with an idiopathic autism spectrum disorder (ASD)and 16 neuropsychiatrically and medically healthy male

230 R. Bernier et al. / Brain and Cognition 64 (2007) 228–237

adults, all with full scale IQ above 80. Groups werematched on IQ, chronological age, and handedness.Informed consent was obtained for all participants.

Participants with autism were diagnosed using the Aut-ism Diagnostic Interview-Revised (ADI-R; Lord, Rutter,& LeCouteur, 1994), the Autism Diagnostic ObservationSchedule-Generic (ADOS-G; Lord et al., 2000), and expe-rienced clinician judgment using DSM-IV criteria (Ameri-can Psychiatric Association, 2000). Intellectual ability wasassessed in both groups using the Wechsler Adult Intelli-gence Scale-3rd Edition (WAIS-III; Wechsler, 1997).

Of the original sample of 17 males with ASD and 16 typ-ical adults, three participants with ASD and one typicalparticipant were excluded prior to analysis due to excessivemovement artifacts that resulted in insufficient EEG datain one or more conditions. As a result, the final sample con-sisted of 14 males with ASD and 15 typical participants.Table 1 presents demographic information for bothgroups, including chronological age, IQ, ethnicity, andhandedness. Groups did not significantly differ on any ofthese factors.

The ASD group was comprised of 6 individuals withAutistic disorder, 5 individuals with Asperger syndrome,and 3 individuals with pervasive developmental disorder— not otherwise specified.

2.2. Behavioral assessments

2.2.1. Mature Imitation Task

The behavioral imitation assessment battery used forthis study was developed by Rogers and colleagues foruse with high functioning individuals with autism (Rogers,Cook, & Greiss-Hess, 2005). Participants were instructedto observe a videotaped model producing a variety of singleand sequenced hand gestures, single and sequenced facialexpressions, hand gestures requiring both hands, shortmeaningless hand movements, and actions on objects.The videotaped model was presented by computer monitorat a distance of 36 inches from the participant. Followingthe single trial of each gesture, expression, movement oraction, the screen became blank and participants imitatedthe observed behavior. The sequences differed in lengthof presentation based on modeled action. For example,the shortest presentation was a single step hand gesturelasting 2.5 s while the longest modeled action (rubbing alintbrush along one arm three times) lasted 20 s. Partici-pants were provided with as much time as needed toattempt to imitate the model.

Table 1Participant descriptive characteristics

Group N Ethnicity Handedness Age (y

Min.

ASD 14 13 White, 1 Other 12 Right, 2 Left 19.4Typical 15 13 White, 2 Other 11 Right, 4 Left 18.6

Participant responses were videotaped and later codedby a trained undergraduate who was blind to the diagnosticstatus of the participant. Imitations were coded followingprocedures outlined by Rogers and colleagues using theMature Imitation Task Manual (Rogers et al., 2005). Ges-tures, expressions, movements, and actions on objects werecoded on scales specific to each task. For example, handgestures were coded on two scales: finger placement andorientation and were scored with one point for correctimitation and zero for failure while the meaninglessmovements were scored on four scales: start position,movement, posture change, and end position, with a totalof two points earned on the start and end position scalesand one point for the movement and posture change scales.Errors, such as groping and overshooting, were coded andtallied separately. Groping errors consisted of additional orincorrect movements that resulted in an improved responseand included visual inspection of the gesture while over-shooting errors were coded when the position of the bodywas in the correct area but the hand or arm extendedbeyond that demonstrated in the model. Style of action(harsh or gentle) on objects was also coded. The coderwas initially trained to reliability, and inter-rater reliabilitywas maintained between the coder and first author on 18%of the protocols. For rater discrepancies of more than onepoint for an item, consensus scores were determined andutilized. Inter-rater reliability was calculated usingintraclass correlations with a mean correlation of .92 andrange of .68–.99 across the 7 scales as well as the errorand style rating scales.

Summary scores were calculated. An overall total scorewas calculated by summing the scores for each imitationtask. An overall error score was calculated by summingthe errors across all imitation tasks. Both single andsequenced hand tasks were combined for a ‘hand total’score and the same was done for the facial expression tasks(‘face total’). The single face and single hand tasks werecombined to yield a ‘single total’ score and the same wasdone for the sequenced tasks (‘sequenced total’). A styletotal was calculated by summing the number of style pointsearned during the actions on objects task.

2.3. EEG assessment

2.3.1. EEG recording

Electrical brain activity was recorded using a dense-array EEG system (EGI Inc., Eugene, OR). EEG wasrecorded from a 128 channel Geodesic sensor net and

ears) IQ: mean (SD)

Max. M (SD) FSIQ VIQ PIQ

37.5 23.6 (4.9) 114.0 (14.2) 117 (14.1) 107.6 (15.4)43.7 26.7 (8.7) 108.9 (15.6) 110.1 (14.7) 105.4 (18.2)


impedances were below 50 kX. EEG signals were analog fil-tered (0.1 Hz high-pass, 100 Hz elliptical low-pass) ampli-fied, and digitized at a 250 Hz. The participant satapproximately 75 cm from a video monitor that deliveredthe stimuli. The video stimuli were presented on the moni-tor at a size of 17.5 cm by 26.25 cm and subtended a visualangle of 13.3 by 19.8. All 128 channels were recorded con-tinuously and stimulus onset and cessation and eventmarkers were registered with the physiological record foroff-line segmentation of the data. The vertex electrodewas used as a reference, and the data were later re-refer-enced to an average reference. Data were post processedusing software provided by Electrical Geodesics, Inc.

2.3.2. EEG imitation paradigm

Following the paradigm developed by Muthukumar-aswamy and Johnson (2004) the experiment consisted offour conditions: resting, observe, execute, and imitate. Allstimuli were presented on videotape to ensure standardiza-tion of administration of the hand movements. In the rest-ing condition, the participant was asked to sit quietly withhands in lap. In the observe condition participants viewedan adult gripping a manipulandum (Muthukumaraswamyet al., 2004) with thumb and index finger as shown inFig. 1.

In the observe condition, participants sat with hands inlap and observed the action. In the execute condition, par-ticipants were verbally instructed to grip the manipulan-dum identical to that shown in the videotape in the samemanner as shown in the videotape. The video monitorshowed only the manipulandum at this time and no move-ment. The manipulandum was placed on the table 50 cmdirectly in front of the participant. In the imitate condition,the participant was instructed to imitate the experimentergripping the manipulandum as shown on the videotape.Each trial lasted approximately 3 s and there was a 7-s intertrial interval between each trial. Twenty trials in each con-dition were presented with condition order randomized.

Fig. 1. Hand grasping the manipulandum.

A synchronization pulse was sent to the EEG recordingequipment at the moment the participant grasped themanipulandum. These pulses served as the event markersfor the execute and imitate conditions. During the observecondition an event marker was sent to the EEG recordingwhen the model on the videotaped hand grasped themanipulandum. These event markers, recorded duringacquisition, were sent directly to the EEG record for usein EEG segmentation.

2.3.3. EEG analysis

Following procedures from Muthukumaraswamy et al.(2004), the EEG recording was segmented into unitsextending 2 s before and after the event markers. Segmentswith movement artifact were identified and removedthrough visual inspection as well as through NetStation’s(Electrical Geodesic) automated artifact detectionalgorithm. Following the parameters of the automatedalgorithm, channels were identified as bad if fast averageamplitude exceeded 200 lVs, if the differential averageamplitude exceeded 100 lVs, or if the channel had zerovariance. Channels were marked bad for all trials if theywere marked bad in 20% of all trials and trials werediscarded if they contained more than 10 bad channels.The mean trial rejection rate was 32.28% across both groups.There were no differences between groups in rate of trialrejection (Typical M = 32.27%; Autism M = 32.29%).

Fast Fourier transforms (FFTs) were performed in Mat-lab (Natick, MA) on two second segments of clean EEGdata. Power spectra were averaged across trials for eachcondition. As the mu rhythm, distinct from other EEGbands, has been proposed to reflect an execution/observa-tion matching system, the study focused on mu rhythm

Fig. 2. Cluster of eight electrodes surrounding C3 and C4 used forstatistical analyses.


activity only. The mu rhythm was defined as the frequencyband ranging from 8–13 Hz that is topographically cen-tered on the standard C3 and C4 positions (Muthukumar-aswamy et al., 2004; Oberman et al., 2005; Pineda, 2005).As shown in Fig. 2, following Muthukumaraswamy et al.(2004), a cluster of eight electrodes on each hemisphere sur-rounding the standard C3 and C4 positions were used forstatistical analyses.

Two measures of mu attenuation were used in thisstudy. The first measure was a ratio of the 8–13 Hz powerduring each of the execute, imitate, and observe conditionsrelative to 8–13 Hz power in the resting condition. Thisratio was used to account for individual variability in over-all EEG power in this frequency range. Due to the inherentnon-normality of ratio data, a log transform was calculatedfor each ratio. The log transform results in a negative valuerepresenting attenuation, a value of zero indicating noattenuation, and a positive value representing mu rhythmaugmentation. This measure was used to examine attenua-tion differences between groups and conditions.

The second measure of mu attenuation used was the dif-ference between an individual’s resting condition and thecondition of interest (i.e. imitate or observe). For this mea-sure of attenuation, greater values equal greater attenua-tion. Similar to the log ratio, this measure controls forindividual differences in baseline EEG power. This measureof attenuation was used to examine relations between EEGand behaviorally assessed imitation skills.

3. Results

3.1. Behavioral imitation skills

As shown in Table 2, on the Mature Imitation Task, theASD group scored significantly worse than the typicaladults. Typical adults were able to more accurately imitatethe model’s expressions and gestures and demonstratedfewer errors than the adults with autism.

3.2. Mu wave attenuation

Fig. 3 presents the log ratio of mu rhythm attenuationfor each condition for both groups. Significant attenuationin mu from baseline was found for both groups for each

Table 2Results for mature imitation task

Summary scale Mean (SD)

Typical group

Total score 185.2 (13.1)Face summary score 57.7 (4.3)Hand summary score 45.9 (7.8)Single task summary 28.8 (2.4)Sequenced task summary 74.9 (8.8)Style total 9.7 (1.9)Number of errors 12.4 (5.2)

condition. For the typical group: execute t(14) = �5.798,p < .001; imitate t(14) = �6.361, p < .001; observe t(14) =�6.211, p < .001. For the autism group: executet(13) = �4.445, p < .001; imitate t(13) = �4.443, p < .001;observe t(13) = �2.747, p < .017. A 2 (group) by 3 (condi-tion) ANOVA for repeated measures yielded a main effectof condition F(2,54) = 12.929, p < .001 but no main effectfor group or interaction effects.

Analysis for the typical group of the log ratio showed nodifferences in degree of mu rhythm attenuation between theexecute and observe conditions, t(14) = 1.803, p < ns. Sim-ilarly, there were no differences between degree of attenua-tion of the execute and imitate conditions, t(14) = .781,p < ns. However, for the typical group, greater attenuationwas found for the imitate condition relative to the observecondition, t(14) = 2.275, p < .05.

Analyses for the autism group revealed significant differ-ences in degree of attenuation between the execute, imitate,and observe conditions. The observe condition showed sig-nificantly less attenuation than the execute and imitate con-ditions, t(13) = 3.291, p < .01, t(13) = 3.653, p < .005,respectively. There were no differences between the imitateand execute conditions, t(13) = 1.106, p < ns.

Given that different patterns were found in EEG powerattenuation between the execute and observe conditions forthe autism group versus typical group, further analyseswere conducted between groups using the log ratio. Whilethe groups did not differ in mu power ratio in the executecondition (t(27) = 1.230, p = ns) or the imitate condition(t(27) = 1.201, p = ns), the groups did significantly differin mu power ratio for the observe condition, with the aut-ism group showing significantly lower mu wave power ratioas compared to the typical group (t(27) = 2.447, p < .05).

To assure that the effects of suppression were specific tothe mu rhythm and not the result of other activity, such asthe more powerful, posterior alpha activity, lead clusterswere selected from other scalp regions. Six leads wereselected from the frontal region (three around AF3 in theleft hemisphere and three around AF4 in the right hemi-sphere) and six leads were selected from the occipital pari-etal region (three around O1 in the left and three aroundO2 in the right). Across the conditions no consistent pat-tern of suppression was observed in the frequency bandunder investigation at these other electrode clusters. This

t-test Sig. p<

Autism group

155.1 (19.4) 4.92 .00146.1 (7.7) 5.076 .00135.1 (12.4) 2.832 .0124.3 (3.5) 4.047 .00156.9 (14.7) 4.037 .0014.9 (2.9) 5.299 .001

19.9 (10.1) 2.544 .05

Fig. 3. Mu power as a function of condition.


indicates that the observed suppression was specific to thecentral electrodes and not the result of other activity, suchas alpha desynchronization.

Given the heterogeneity in individuals with diagnoses onthe autism spectrum, power values were calculated for eachparticipant with autism for each of the four experimentalconditions and are presented in Table 3.

Intercorrelations of EEG power between each electrodefor both hemisphere clusters were calculated. Of the 168comparisons across the three conditions of execute,observe, and imitate, 144 were significant at the .05 level(138 at the .01 level). This did not differ by condition.

The latency to grasp the manipulandum during the imi-tation condition was calculated for a random subset of thesample (11 participants from each group). This was calcu-lated to check participants’ attention to the imitation task.The average times from onset of trial to grasp for the aut-ism and typical groups were 1.47 and 1.59 s, respectively.

Table 3Mu wave power values for each autism participant (in lV2)

Subject No. Execute Condition: m

Observe

1 .303940 (.125) .685585 (.32 .155272 (.064) .143876 (.03 .203481 (.081) .127112 (.04 .255224 (.070) .302710 (.05 .846268 (.478) 2.36863 (1.76 .549620 (.222) .880590 (.47 .183697 (.031) .248708 (.08 .424599 (.205) .666416 (.39 .571214 (.372) .842045 (.4

10 .706779 (.365) 2.00470 (.9511 .532764 (.175) 1.01521 (.3812 .123849 (.027) .133719 (.013 .211769 (.040) .358013 (.114 .175790 (.069) .150766 (.0

There were no differences between groups in latency tograsp the manipulandum, t(21) = .816, p = ns.

3.3. Relation between imitation and mu attenuation

Correlational analyses revealed relations between muwave attenuation and behavioral imitation skills. To cap-ture attenuation of mu power, a difference score was calcu-lated by subtracting the condition of interest (observe orimitate) from the resting condition for each individual.

Mu wave attenuation for the observe condition was cor-related with the face imitation summary scale for bothgroups combined r(29) = .567, p < .001 and also for bothgroups when examined separately: typical group:r(15) = .551, p < .05; autism group: r(14) = .536, p < .05(Fig. 4). Mu wave attenuation for the observe conditionwas also correlated with the total imitation score(r(29) = .509, p < .01), hand summary score (r(29) = .388,

ean (SD)

Imitate Resting

02) .336325 (.132) 1.08213 (.438)40) .165551 (.058) .407575 (.146)74) .158918 (.050) .214257 (.072)93) .175809 (.092) .783554 (.350)2) 1.22539 (.895) 3.45046 (1.75)34) .508607 (.221) .853169 (.750)60) .303731 (.103) 1.62923 (.417)08) .359777 (.110) .546059 (.137)64) .301757 (.149) 2.12702 (.423)7) .572928 (.167) 3.09828 (.587)8) .515492 (.190) 3.65624 (1.32)56) .106281 (.038) .152444 (.086)34) .203514 (.058) .285538 (.138)91) .226580 (.071) .117707 (.058)

Fig. 4. Relation between mu wave attenuation during observation andbehaviorally assessed facial imitation task summary score. Attenuation iscalculated as difference between power during observation and powerduring resting.


p < .05), sequenced task summary score (r(29) = .522,p < .01), and single task summary score (r(29) = .401,p < .05).

For mu wave attenuation in the imitate condition withboth diagnostic groups combined, correlations were foundwith total score (r(29) = .450, p < .05), face summary score(r(29) = .433, p < .05), hand summary score (r(29) = .422,p < .05), and sequence summary score (r(29) = .507,p < .01). When these relations were examined for the typi-cal group separately, only the correlation between imita-tion mu wave attenuation and the face summary scaleremained, r(15) = .571, p < .05. No correlation was foundbetween these two variables for the autism group alone.However, for the autism group, correlations were foundfor the total sample between imitation mu attenuationand total score (r(14) = .574, p < .05), hand summary score(r(14) = .537, p < .05), and sequence summary score(r(14) = .625, p < .05).

4. Discussion

The primary goals of this study were to clarify the nat-ure of the imitation deficits in autism and explore a possibleneurophysiological correlate of these deficits. Both adultswith ASD and typical adults showed significant mu waveattenuation during the execute condition as compared tothe resting condition. However, during the observationcondition, adults with autism showed reduced mu waveattenuation, as compared to the typical adults. To theextent to which mu wave attenuation is a sensitive indexof the functioning of the execution/observation matchingsystem, these results suggest that this neural system may

be disrupted in autism. A class of neurons, called mirrorneurons, first identified in monkeys in the prefrontal cor-tex, provides a possible neurological mechanism for an exe-cution/observation matching system. Indirect evidencefrom imaging and psychophysiological studies suggeststhere is a mirror neuron system in humans. This systemin humans consists of the posterior inferior frontal gyrusand inferior parietal cortex, and activates both when anaction is observed and when it is executed (Rizzolatti &Craighero, 2004). While it is unclear what the relationshipis between the mu rhythm and mirror neurons, it has beenproposed that the mu rhythm might reflect ‘‘downstreammodulation of motor neurons by cells in the premotor cor-tex, some of which are mirror neurons’’ (Muthukumarasw-amy et al., 2004; Pineda, 2005, p. 57).

This study found support for continued impairments inimitation skills for adults with ASD who have intellectualabilities in the average range. While there have beennumerous studies examining imitation skills in autism,the vast majority of these studies have been conducted withchildren. Two studies conducted with teens found contin-ued evidence of imitation impairments (Loveland et al.,1994; Rogers et al., 1996) and one other study of adultswith ASD found significantly more errors and no benefitsin using a mirror image to enhance performance (Avikai-nen et al., 2003). The current study found continuedimpairments in single and multiple step acts that involvedboth hand and face imitation, imitative acts that requiredboth hands, and actions that involved meaningless move-ments. The imitative deficits observed in childhood in pre-vious studies clearly persist into adulthood.

Moreover, the degree of mu wave attenuation whenobserving an action was found to be correlated with behav-iorally assessed imitation skills. Lower levels of mu attenu-ation were associated with poorer imitation skills, and thiswas most robust for facial imitation skills. This result isconsistent with the hypothesis that functioning of theexecution/observation matching system is associated withlevel of imitation skill, and more specifically, that the imi-tation impairments found in individuals with autism mightbe related to dysfunction of the execution/observationmatching system.

The more robust correlation with facial imitation ascompared to hand or stylistic imitation may be due to dif-ferences in the subject of imitation. The imitation of facialexpressions relies more heavily on cross modal communica-tion than imitation of styles or hand gestures because thereis no visual feedback. That is, when imitating a hand ges-ture, it is entirely possible for the imitator to look downat his or her hand to help with the imitation. This was, infact, observed numerous times during the behavioral task.As such, facial imitation tasks may require more activationof the execution/observation matching system, whichunderlies cross modal communication, than imitation ofhand gestures or action styles. This greater recruitment ofthe underlying execution/observation system may thentranslate into greater desynchronization of the underlying


neural assemblies resulting in greater mu wave attenuation.A dysfunctional execution/observation system, therefore,could result in greater impairments in facial imitation thanin other aspects of imitation.

While the typical adults demonstrated the same level ofattenuation of the mu rhythm when both observing andexecuting an action, the autism group failed to show thesame degree of attenuation when observing the action.However, the autism group did demonstrate attenuationwhen executing an action, suggesting intact sensorimotorsystems underlying self initiated actions but not thoseunderlying the observation of biologically relevant move-ment. A disruption in the execution/observation matchingsystem could contribute to deficits in social behavior as thissystem allows for translation of others’ observed actionsinto understanding through motor representation of thataction. A disruption of the neural processes that allowfor understanding others actions would then lead to diffi-culties in social learning, communication, and interaction.

Williams et al. (2001) first proposed that dysfunction ofan execution/observation matching system, in the form ofmirror neurons, underlies the imitation impairments in aut-ism. Recently, several researchers have conducted studiesrelevant to this execution/observation matching system inindividuals with autism (Avikainen, Kulomaki, & Hari,1999; Hadjikhani, Joseph, Snyder, & Tager-Flusberg,2005; Oberman et al., 2005; Theoret et al., 2005; Williamset al., 2006). Differences between children and teens withautism and controls have been found in activation of themirror neuron system in humans during functional MRI(Dapretto et al., 2006; Williams et al., 2006). Structuralbrain differences in mirror neuron regions have also beenobserved in adults and correlate with autism symptomolo-gy (Hadjikhani et al., 2005).

There are alternative explanations for our findings ofreduced mu attenuation in individuals with autism duringthe observation of actions other than a faulty execution/observation matching system. First, it is possible that theautism group was performing some other mental task whenwatching the hand motion. It is also possible that the aut-ism group was attending to different aspects of the actionon the video screen. That is, their eye fixations could havefocused on irrelevant parts of the screen, such as themanipulandum or a portion of the blank, gray back-ground. If this were the case, during the imitation condi-tion, the autism group might fail to imitate the action orshow a significant delay when performing the imitation.First, attention was carefully monitored during the taskand all subjects maintained fixation on the monitor duringstimulus presentation. Second, average latency to grasp themanipulandum during the imitation condition was assessedand no differences were found between groups. None of theparticipants in the autism group failed to imitate the grasp-ing of the manipulandum for any trials; none showed a sig-nificant delay when imitating the hand grasping thewooden block. Thus, it is unlikely that attention factorscan account for these findings.

Another possibility is that disruption of the visual pro-cessing system or a disruption to other input systems maydisrupt the circuitry in the execution/observation matchingsystem. While first order motion processing appears to beintact in individuals with autism (Dakin & Frith, 2005),there is evidence to suggest that there is a disruption inthe processing of biological motion and specifically pro-cessing in the superior temporal sulcus (Boddaert et al.,2004; Pelphrey, Morris, & McCarthy, 2005; Waiter et al.,2004). Further research is necessary to clarify the role thatthe processing of biological motion plays in the execution/observation matching system, imitation, and the social def-icits in autism. The most parsimonious explanation is thatdisruptions to multiple circuits acting in concert result inthe pattern of social impairments seen in autism.

However, the execution/observation system cannot beentirely dysfunctional, as individuals with autism are ableto imitate, albeit more poorly than individuals with typicaldevelopment. Likewise, there is much variability in theoryof mind abilities, empathy, and other social cognition def-icits in individuals with autism. Therefore, it could bedegrees of dysfunction of this system, acting in concert withdegrees of disruptions in other neural systems, such as thesuperior temporal sulcus, the limbic system, the cerebellum,and the ventromedial prefrontal cortex, that potentiallylead to the range and variety of deficits observed in autism.

In summary, the execution/observation matching sys-tem is believed to serve as the underlying neural circuitryfor imitation. This system allows the observer to translateobserved actions into an understanding of the actionthrough the activation of the observer’s motor system. Dis-ruptions in the circuitry that translates visual input intomotor understanding can significantly impact social inter-action. In light of the fact that deficits in social cognitionare a core feature of autism, it is logical to examine this cir-cuitry as one of the many sources of impairment in autism.This study provides further evidence of a disruption in theexecution/observation system in autism and the relationbetween this circuitry and imitation skills.

References

Aldridge, M., Stone, K., Sweeney, M., & Bower, T. (2000). Preverbalchildren with autism understand the intentions of others. Developmen-

tal Science, 3, 294–301.American Psychiatric Association (2000). Diagnostic and statistical manual

of mental disorders (4th ed., Text Revision). Washington, DC: APA.Arroyo, S., Lesser, R., Gordon, B., Uematsu, S., Jackson, D., & Webber,

R. (1993). Functional significance of the mu rhythm of human cortex:An electrophysiologic study with subdural electrodes. Electroenceph-

alography Clinical Neurophysiology, 87, 76–87.Avikainen, S., Kulomaki, T., & Hari, R. (1999). Normal movement

reading in Asperger subjects. Neuroreport, 10, 3467–3470.Avikainen, S., Wohlschlager, A., Liuhanen, S., Hanninen, R., & Hari, R.

(2003). Impaired mirror-image imitation in Asperger and high-func-tioning autistic subjects. Current Biology, 13, 339–341.

Babiloni, C., Babiloni, F., Carducci, F., Cincotti, F., Cocozza, G., DelPercio, C., et al. (2002). Human cortical electroencephalography(EEG) rhythms during the observation of simple aimless movements:A high-resolution EEG study. Neuroimage, 17, 559–572.


Babiloni, C., Carducci, F., Cincotti, F., Rossini, P., Neuper, C.,Pfurtscheller, G., et al. (1999). Human movement-related potentialsvs desynchronization of EEG alpha rhythm: A high-resolution EEGstudy. Neuroimage, 10, 658–665.

Babiloni, C., Del Percio, C., Babiloni, F., Carducci, F., Cincotti, F.,Moretti, D., et al. (2003). Transient human cortical responses duringthe observation of simple finger movements: A high-resolution EEGstudy. Human Brain Mapping, 20, 148–157.

Boddaert, N., Chabane, N., Gervaise, H., Good, C., Bourgeois, M.,Plumet, M., et al. (2004). Superior temporal sulcus anatomicalabnormalities in childhood autism: A voxel-based morphometryMRI study. Neuroimage, 23, 364–369.

Charman, T., Swettenham, J., Baron-Cohen, S., Cox, A., Baird, G., &Drew, A. (1997). Infants with autism: An investigation of empathy,pretend play, joint attention, and imitation. Developmental Psychology,

33, 781–789.Charman, T., Swettenham, J., Baron-Cohen, S., Cox, A., Baird, G., &

Drew, A. (1998). An experimental investigation of social-cognitiveabilities in infants with autism: Clinical implications. Infant Mental

Health Journal, 19, 260–275.Chatrian, G., Petersen, M., & Lazarte, J. (1959). The blocking of the

rolandic wicket rhythm and some central changes related to move-ment. Electroencephalography Clinical Neurophysiology Supplement,

11, 497–510.Cochin, S., Barthelemy, C., Lejeune, B., Roux, S., & Martineau, J. (1998).

Perception of motion and qEEG activity in human adults. Electroen-

cehalography and Clinical Neurophysiology, 107, 287–295.Cochin, S., Barthelemy, C., Roux, S., & Martineau, J. (1999). Observation

and execution of movement: Similarities demonstrated by quantifiedelectroencephalography. European Journal of Neuroscience, 11,1839–1842.

Cochin, S., Barthelemy, S., Rous, S., & Martineau, J. (2001). Electroen-cephalographic activity during perception of motion in childhood.European Journal of Neuroscience, 13, 1791–1796.

Dakin, S., & Frith, U. (2005). Vagaries of visual perception in autism.Neuron, 48, 507–597.

Dapretto, M., Davies, M., Pfeifer, J., Scott, A., Sigman, M., Bookheimer,S., et al. (2006). Understanding emotions in others: Mirror neurondysfunction in children with autism spectrum disorders. Nature

Neuroscience, 9, 28–30.Dawson, G., & Adams, A. (1984). Imitation and social responsiveness in

autistic children. Journal of Abnormal Child Psychology, 12, 209–226.Dawson, G., & Lewy, A. (1989). Arousal, attention, and socioemotional

impairments of individuals with autism. In G. Dawson (Ed.), Autism,

Nature, diagnosis and treatment (pp. 49–74). New York: Guilford.Dawson, G., Meltzoff, A., Osterling, J., & Rinaldi, J. (1998). Neuropsy-

chological correlates of early symptoms of autism. Child Development,

69, 1276–1285.DeMeyer, M., Alpern, G., Barons, S., DeMyer, W., Churchill, D.,

Hingtgen, J., et al. (1972). Imitation in autistic, early schizophrenic,and non-psychotic subnormal children. Journal of Autism and Child

Schizophrenia, 2, 264–287.Gastaut, H., & Bert, J. (1954). EEG changes during cinematographic

presentation; moving picture activation of the EEG. Electroen-

cephalography Clinical Neurophysiology Supplement, 6,433–444.

Green, D., Baird, G., Barnett, A., Henderson, L., Huber, J., & Henderson,S. (2002). The severity and nature of motor impairment in Asperger’ssyndrome: A comparison with specific developmental disorder ofmotor function. Journal of Child Psychology and Psychiatry, 43,655–668.

Hadjikhani, N., Joseph, R., Snyder, J., & Tager-Flusberg, H. (2005).Anatomical differences in the mirror neuron system and socialcognition network in autism. Cerebral Cortex, November 23, [epubahead of print].

Hammes, J., & Langdell, T. (1981). Precursors of symbol formation andchildhood autism. Journal of Autism and Developmental Disorders, 11,331–346.

Hobson, R., & Lee, A. (1999). Imitation and identification in autism.Journal of Child Psychology and Psychiatry, 40, 649–659.

Jones, V., & Prior, M. (1985). Motor imitation abilities and neurologicalsigns in autistic children. Journal of Autism and Developmental

Disorders, 15, 37–46.Lepage, J., & Theoret, H. (2006). EEG evidence for the presence of an

action observation- execution system in children. European Journal of

Neuroscience, 23, 2505–2510.Lord, C., Risi, S., Lambrecht, L., Cook, E., Leventhal, B., DiLavore, P.,

et al. (2000). The autism diagnostic observation schedule-generic: Astandard measure of social and communication deficits associated withthe spectrum of autism. Journal of Autism and Developmental

Disorders, 30, 205–223.Lord, C., Rutter, M., & LeCouteur, A. (1994). Autism Diagnostic

Interview-Revised: A revised versions of a diagnostic interview forcaregivers of individuals with possible pervasive developmentaldisorders. Journal of Autism and Developmental Disorders, 24,659–685.

Loveland, K., Tunali Kotoski, B., Pearson, D., Brelsford, K., Ortegon, J.,& Chen, R. (1994). Imitation and expression of facial affect in autism.Development and Psychopathology, 6, 433–444.

Martineau, J., & Cochin, S. (2003). Visual perception in children: Human,animal and virtual movement activates different cortical areas.International Journal of Psychophysiology, 51, 37–44.

Meltzoff, A., & Decety, J. (2003). What imitation tells us about socialcognition: A rapprochement between developmental psychology andcognitive neuroscience. Philosophical Transactions of the Royal Society

of London, Series B, Biological Sciences, 358, 491–500.Meltzoff, A., & Gopnik, A. (1993). The role of imitation in understanding

persons and developing theories of mind. In S. Baron-Cohen & H.Tager-Flusberg (Eds.), Understanding other minds: Perspectives from

autism. Oxford: Oxford University Press.Meltzoff, A., & Moore, M. (1997). Explaining facial imitation: A

theoretical model. Early Development & Parenting, 6, 179–192.Muthukumaraswamy, S., & Johnson, B. (2004). Changes in rolandic mu

rhythm during observation of a precision grip. Psychophysiology, 41,152–156.

Muthukumaraswamy, S., Johnson, B., & McNair, N. (2004). Mu rhythmmodulation during observation of an object-directed grasp. Cognitive

Brain Research, 19, 195–201.Oberman, L., Hubbard, E., McCleery, J., Altschuler, E., Ramachandran,

V., & Pineda, J. (2005). EEG evidence for mirror neuron dysfunctionin autism spectrum disorders. Cognitive Brain Research, 24, 190–198.

Ohta, M. (1987). Cognitive disorders of infantile autism: A studyemploying the WISC, spatial relationships, conceptualization, andgestural imitation. Journal of Autism and Developmental Disorders, 17,45–62.

Pelphrey, J., Morris, J., & McCarthy, G. (2005). Neural basis of eye gazeprocessing deficits in autism. Brain, 128, 1038–1048.

Pfurtscheller, G., Brunner, C., Schlogl, A., & Lopes da Silva, F. (2006).Mu rhythm (de)synchronization and EEG single-trial classification ofdifferent motor imagery tasks. Neuroimage, 31, 153–159.

Pfurtscheller, G., Neuper, C., Andrew, C., & Edlinger, A. (1997). Footand hand area mu rhythms. International Journal of Psychophysiology,

26, 121–135.Pineda, J. (2005). The functional significance of mu rhythms: Translating

‘‘seeing’’ and ‘‘hearing’’ into ‘‘doing’’. Brain Research Reviews, 50,57–68.

Rizzolatti, G., & Craighero, L. (2004). The mirror-neuron system. Annual

Review of Neuroscience, 27, 169–192.Rogers, S., & Pennington, B. (1991). A theoretical approach to the deficits

in infantile autism. Development and Psychopathology, 3, 137–162.Rogers, S., Bennetto, L., McEvoy, R., & Pennington, B. (1996). Imitation

and pantomime in high-functioning adolescents with autism spectrumdisorders. Child Development, 67, 2060–2073.

Rogers, S., Cook, I. & Greiss-Hess, L. (2005). Mature Imitation Task.Unpublished coding manual. M.I.N.D. Institute, University of Cali-fornia, Davis.


Rogers, S., Hepburn, S., Stackhouse, T., & Wehner, E. (2003). Imitationperformance in toddlers with autism and those with other developmentaldisorders. Journal of Child Psychology and Psychiatry, 44, 763–781.

Sigman, M., & Ungerer, J. (1984). Attachment behaviors in autistic children.Journal of Autism and Developmental Disorders, 14(3), 231–244.

Smith, I., & Bryson, S. (1998). Gesture imitation in autism I: Nonsymbolicpostures and sequences. Cognitive Neuropsychology, 15, 747–770.

Stone, W., Lemanek, K., Fishel, P., Fernandez, M., & Altemeier, W.(1990). Play and imitation skills in the diagnosis of autism in youngchildren. Pediatrics, 64, 1688–1705.

Stone, W., Ousley, O., & Littleford, C. (1997). Motor imitation in youngchildren with autism: What’s the object? Journal of Abnormal Child

Psychology, 25, 475–485.Stroganova, T., Orekhova, E., & Posikera, I. (1999). EEG alpha rhythm in

infants. Clinical Neurophysiology, 110, 997–1012.Theoret, H., Halligan, E., Kobayashi, M., Fregni, F., Tager-Flusberg, H.,

& Pascual-Leone, A. (2005). Impaired motor facilitation during action

observation in individuals with autism spectrum disorder. Current

Biology, 15, R84–R85.Waiter, G., Williams, J., Murray, A., Gilchrist, A., Perrett, D., & Whiten,

A. (2004). A voxel-based investigation of brain structure in maleadolescents with autistic spectrum disorder. Neuroimage, 22, 619–625.

Wechsler, D. (1997). Wechsler adult intelligence scale-third edition (WAIS-

III). San Antonio, TX: Psych. Corp.Williams, J., Waiter, G., Gilchrist, A., Perrett, D., Murray, A., & Whiten,

A. (2006). Neural mechanisms of imitation and ‘mirror neuron’functioning in autism spectrum disorder. Neuropsychologia, 44,610–621.

Williams, J., Whiten, A., & Singh, T. (2004). A systematic review of actionimitation in autistic spectrum disorder. Journal of Autism and

Developmental Disorders, 34(3), 285–299.Williams, J., Whiten, A., Suddendorf, T., & Perrett, D. (2001). Imitation,

mirror neurons and autism. Neuroscience and Biobehavioral Reviews,

25, 287–295.

eeg mu rhythm and imitation impairments in individuals with autism spectrum disorder

Documents