Neural Correlates of Detecting Pretense: Automatic ... · Neural Correlates of Detecting Pretense: Automatic Engagement of the Intentional Stance under Covert Conditions ... magnetic

Neural Correlates of Detecting Pretense AutomaticEngagement of the Intentional Stance under

Covert Conditions

Tim P German Jeffrey L Niehaus Meghan P RoartyBarry Giesbrecht and Michael B Miller

Abstract

amp Typically developing children begin to produce andunderstand pretend play between 18 and 24 months of ageand early pretense has been argued to be a candidate lsquolsquocorersquorsquocapacity central to the deployment of representations of otherpeoplesrsquo mental statesmdashlsquolsquotheory of mindrsquorsquo In a functionalmagnetic resonance imaging study 16 healthy adult volun-teers were imaged while watching short (5 sec) clips of actorswho either performed simple everyday actions or pretendedto perform a similar set of actions under covert conditions(eg participants were not directed to attend to actorsrsquo mentalstates) There was increased activity in the medial prefrontalareas (Brodmannrsquos areas [BA] 9632 9 and 10) inferior

frontal gyrus bilaterally (BA 44 47) temporo-parietal regions(BA 21 and 22) and parahippocampal areas including theamygdala when subjects viewed pretend actions as com-pared with real actions This result suggests that at least someareas previously implicated in making explicit mental statejudgments are also strongly activated in response to actionsthat call for mental state interpretation (eg pretense) evenwhen there is no explicit instruction for lsquolsquomind readingrsquorsquo Thisoutcome is discussed in terms of accounts that proposelsquolsquotheory of mindrsquorsquo to be underwritten by automatic specializedmechanisms for the interpretation of the behavior of socialagents amp

INTRODUCTION

Social behavior rests on a foundational capacity perhapsuniquely human to understand the actions reactionsand interactions of other social agents in terms of theirmental states This capacity is variously known as lsquolsquothe-ory of mindrsquorsquo commonsense psychology beliefndashdesirereasoning mind reading and as what Dennett (1987)termed lsquolsquotaking the intentional stancersquorsquo Theory of mindappears to operate automatically without effort and inthe typical case develops rapidly over the first few yearsof life without formal teaching and largely indepen-dently of IQ (Leslie 1994a) These facts suggest to manycognitive scientists that its acquisition and functioning isbased on a reliably developing and perhaps modularcognitive mechanism (Scholl amp Leslie 1999 Leslie1994b 2000b) Support for this view is found in thecase of childhood autism where there appears to be aspecific cognitive deficit in mental state reasoning whichresults in social imaginative and communicative defi-cits alongside relatively spared abilities in other do-mains (Baron-Cohen 1995 Leslie amp Thaiss 1992Baron-Cohen Leslie amp Frith 1985)

Attention in cognitive neuroscience has more recentlyfocused on identifying the brain areas that underwritethe capacity to deploy a theory of mind utilizing meth-ods of functional imaging including positron emissiontomography (PET Ruby amp Decety 2003 Gallagher JackRoepstorff amp Frith 2002 Brunet Sarfarti Hardy-Bayleamp Decety 2000 Castelli Happe Frith amp Frith 2000Fletcher et al 1995 Goel Grafman Sadato amp Hallet1995) functional magnetic resonance imaging (fMRIMason Banfield amp Macrae 2004 Ramnani amp Miall2004 Saxe amp Kanwisher 2003 Ferstl amp von Cramon2002 McCabe Houser Ryan Smith amp Trouard 2001Vogeley et al 2001 Gallagher et al 2000 Baron-Cohenet al 1999) and event-related potentials (ERP Sabbaghamp Taylor 2000) as well as investigations based onpatients with acquired or developmental brain lesions(Fine Lumsden amp Blair 2001 Lough Gregory ampHodges 2001 Rowe Bullock Polkey amp Morris 2001Stuss Gallup amp Alexander 2001 Channon amp Crawford2000 Stone Baron-Cohen amp Knight 1998)

This body of work although deploying a range oftasks populations and methods has converged on theidentification of several brain regions that are thoughtto underlie theory of mind either because they are com-monly activated when subjects are required to engagein explicit reasoning involving the mental states ofUniversity of California

D 2004 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 1610 pp 1805ndash1817

other people (eg see Saxe Carey amp Kanwisher 2004Gallagher amp Frith 2003 for reviews) or because damageto those areas results in deficits in social cognitivefunction (eg Rowe et al 2001 Stuss et al 2001Stone et al 1998) The areas most commonly implicat-ed include areas of the medial prefrontal cortex (Brod-mannrsquos areas [BA] 932) areas of the posterior superiortemporal gyrus at the junction with the parietal cor-tex and anterior lateral temporal regions (eg tem-poral poles) Areas also implicated although lessconsistently include the orbito-frontal cortex (Baron-Cohen Ring Moriarty amp Schmitz 1994) and the amyg-dala (Grezes Frith amp Passingham 2004 Stone Baron-Cohen Calder Keane amp Young 2003 Fine et al 2001Baron-Cohen et al 1999)

The activations seen in imaging studies of theoryof mind (some of which are summarized in Table 1)indicate candidate brain areas that are involved inovert mental state judgment In these studies theoryof mind activity is typically isolated by subtracting ac-

tivity from control tasks designed to parallel mentalstate reasoning tasks in all aspects save the criticalrequirement to consider the mental states of othersIdeally these tasks form a lsquolsquominimal pairrsquorsquo differing onlyin mental state versus control content For exampleFletcher et al (1995) explicitly instructed subjects tosolve story-based tasks by reasoning about the mentalstates of the protagonists and compared activations inthat task to activations in control tasks where subjectswere also explicitly directed to avoid considering men-tal states In the study by Castelli et al (2000) subjectswere explicitly informed of three different types ofanimation they would see including information thatsome would seem lsquolsquoas if [the triangles] were takinginto account their reciprocal feelings and thoughtsrsquorsquo(Castelli et al 2000 p 322) before being asked toview the animations

Importantly activations in some of the areas associat-ed with theory of mind content are also seen in caseswhere the only difference between conditions is the

Table 1 Representative Medial Frontal Activations in Mental State Reasoning Tasks

Coordinates

Task Study BA x y z

Theory of mind story comprehension(vs control stories)

Fletcher et al (1995) 8 12 42 40

Judging others knowledge about artifactfunction (vs simple inference about function)

Goel et al (1995) 9 12 38 32

Judging mental states from eye region(vs judging sex)

Baron-Cohen et al (1999) 2432

80

3644

014

Theory of mind story and cartoon comprehension(vs control storiescartoons)

Gallagher et al (2000) 10 10 48 12

Explaining complex motion patterns of movingtriangles (vs random motion)

Castelli et al (2000) 9 4 60 32

Picture sequence completion (intention sequences vsphysical causality sequences)

Brunet et al (2000) 8 4 56 44

Theory of mind story comprehension(self and other content) vs control stories

Vogeley et al (2001) 10 6 56 2

On-line cooperation with human (vs computer opponent) McCabe et al (2001) 10 5 52 10

On-line strategic game with human (vs computer opponent) Gallagher et al (2002) 910

108

5054

3012

Judging coherence in theory of mind sentence pairs(vs language control task)

Ferstl and von Cramon (2002) 9 19 49 30

Semantic judgments about persons vs about objects Mitchell et al (2002) 03

5439

210

First- vs third-person judgments Ruby and Decety (2003) 0 20 70

Judging actions as performable by humans vs by dogs Mason et al (2004) 89 10 48 34

Predicting action in person vs computer Ramnami and Miall (2004) 8 56 24

1806 Journal of Cognitive Neuroscience Volume 16 Number 10

need to take an explicit lsquolsquointentional stancersquorsquo (Gallagheret al 2002) In this PET study participants played astrategic game (paper scissors stone) either againstwhat they thought was a human opponent or againstwhat they thought was a computer following a range ofknown strategies The human computer subtractionwas conducted to calculate the difference in activity inresponse to the same series of random trials insertedinto the human or computer sequence thus isolatingthe intentional stance (ie the requirement to reasonexplicitly about mental states) from associated featuresof mental state reasoning (eg detecting social agentsbiological motion mental state language etc) Gallagheret al (2002) identified only the anterior paragingulategyrus area as active when subjects thought they wereplaying a human versus a computer opponent leadingthem to characterize this area (BA 932) as uniquelyinvolved when participants take the intentional stancelsquolsquoeven when there are no additional verbal or visual cuesto assist mentalizingrsquorsquo (2002 p 819) This is in contrastto the other putative areas identified in off-line tasks(eg areas of the temporo-parietal cortex the temporalpoles) that are activated by lsquolsquosignals important for men-talizing even when the subject is not adopting anintentional stancersquorsquo (ibid p 819)

In the current study we address the question ofwhether areas associated with the deployment of mentalstate reasoning are activated in a task where no explicitjudgments about mental states are made Instead wepresent adult subjects with stimuli in which we hypoth-esize that the intentional stance will be automaticallyengagedmdashcases where an actor performs an act ofpretense The detection and interpretation of pretenseis one of the earliest documented theory of mindcapacities emerging in typical development between 18and 24 months (Leslie 1987 1994b 2000a) and becausechildren this age are too young to have much explicitknowledge about pretense nor require any kind ofinstruction to reason about mental states it is assumedto be based on mechanisms that automatically createmental state representations for the child from the inputof motherrsquos behavior (German amp Leslie 2000 2001Leslie 2000b) Thus we predict that pretend acts willprovoke the engagement of the intentional stance in theabsence of overt instructions to consider the mentalstates of the actor

Note that while a number of the prior studies haveavoided explicitly instructing or cuing the participantsto think about mental states (eg Mason et al 2004Saxe amp Kanwisher 2003 Brunet et al 2000) these tasksdo nevertheless typically involve an explicit judgmentabout human action For example Saxe and Kanwisher(2003) required participants to judge the content ofa characterrsquos belief while Mason et al (2004) askedparticipants to decide if particular actions could beperformed by humans or not Our aim here is to assesswhether brain areas associated with mental state reas-

oning can be activated where the task does not involveany explicit judgment about peoplesrsquo actions at allIndeed in the current task participants are directedtoward making judgments about entirely different as-pects of the stimuli

Interestingly studies have been reported that reportmedial prefrontal cortex activations associated with taskswhere arguably no explicit action or mental state judg-ments were required Mitchell Heatherton and Macrae(2002) showed activations in medial prefrontal cortexin a task requiring individuals to judge whether certainadjectives applied to people versus objects Althoughno explicit action judgment tasks were made makingthis arguably a covert task the task still required theexplicit consideration of mental states especially inthe cases of certain adjectives (eg devious anxious)Another candidate lsquolsquocovertrsquorsquo task was reported by Calderet al (2002) who identified medial prefrontal activa-tions associated with the processing of eye-directioninformation in a task where no explicit instructions toconsider mental states were given Finally Singer et al(2004) showed that areas of the anterior cingulate areactivated during an lsquolsquoempathyrsquorsquo task in which femaleparticipants witnessed their partner receiving a painfulstimulus under no instructions to consider their beliefsor feelings

In the current study 16 healthy volunteers wereimaged via fMRI as they watched short video clips ofactors performing simple actions in short blocks (egFigure 1) Half the actions involved the actors reallyperforming the act in question (eg pouring tea into acup reaching a book from a shelf etc) The other halfinvolved the actors pretending to perform such actionsParticipants were initially asked to perform a cover taskto ensure attention to the entire clips which consistedin deciding whether the film clip had been prematurelyedited To allow for this half the real and half thepretend clips were filmed in such a way that the actionwas interrupted by blue screen before the conclusion ofthe event After each clip the subjects were asked torespond via a button press as to whether the video clipwas complete To our knowledge no other study haspreviously attempted to determine whether areas asso-ciated with mental state reasoning (such as medialprefrontal cortex) are activated in response to displaysinvolving specific actions of social agents under covertconditions (ie conditions that call for no explicitmental state or action judgment)

RESULTS

Behavioral Data

All 16 subjects were appropriately engaged in the task(ie lsquolsquowas the film completersquorsquo) The average correctresponse across subjects for the task was 8831 witha low of 67 and a high of 98 It should be noted that

German et al 1807

errors were distributed among the clips across the scanand were not confined to the first clips in a block Someclips were somewhat ambiguous and thus errors wereexpected and it was unlikely that the sequences of threeclips of a given type were detectable by subjects

Brain Activity Associated with Pretense

The critical comparison in this study was between thosebrain regions that were more active for viewing pretendactions than for viewing real actions in a covert condi-tion Table 2 and Figure 2 illustrate significant activationsacross the 16 subjects using a random effects model anda statistical threshold of p lt 005 (uncorrected) and avoxel extent of 30 (see Methods) We chose to use athreshold that would reveal any regions that might besensitive to the covertly pretense condition while stillmaintaining a statistical threshold and voxel extent thatwould survive a corrected probability criterion for clus-ters (Forman et al 1995 Xiong Gao Lancaster amp Fox1995) Three regions in the medial prefrontal cortexwere significantly active including an anterior region inthe superior frontal gyrus (BA 9) an area posterior tothat encompassing the anterior cingulate and the medialfrontal gyri (BA 3269) and an area inferior to that inthe medial frontal gyrus (BA 10) Several regions werealso significantly active in the right and left ventrolateralprefrontal cortex including the right and left inferiorfrontal gyrus or operculum (BA 47) and more superiorregions in the right and left inferior frontal gyrus (BA44) Two regions in the temporo-parietal region wereactive the left middle temporal gyrus (BA 21) and theright superior temporal gyrus (BA 22) The right fusi-

form gyrus (BA 37) the left amygdala and the rightparahippocampal gyrus were also active These regionshave often been implicated in previous studies involvinglsquolsquotheory of mindrsquorsquo tasks that have relied on explicitmeasures

An activation in one of the critical regions mentionedabove the medial frontal gyrus (BA 10) represented thedifference between two deactivations relative to base-line Figure 3 illustrates an analysis of the percent signalchange from baseline at the local maxima In this casethe average signal intensity across the block for both thepretend and real conditions was less than the signalintensity during the baseline state (ie viewing the bluescreen) In contrast the average signal intensity in theanterior cingulate for both the pretend and real con-ditions was greater than the baseline state Yet in bothcases the signal intensity associated with the pretendcondition was significantly greater than the signal inten-sity associated with the real condition Although little isknown about the difference between two activationsversus the difference between two deactivations it hasbeen suggested that in the case of two deactivations thecondition closer to the baseline state may representmental operations that overlap with the default process-ing mode of the brain (Gusnard Akbudak Shulmanamp Raichle 2001) Several researchers have linked vari-ous cognitive processes to this overlap including per-son knowledge (Mitchell et al 2002) self-referencing(Kelley et al 2002) and coherence processing (Ferstl ampvon Cramon 2002) It should be noted however thatother regions in the medial prefrontal cortex that areassociated with pretense in our study do not reflect thedifference between two deactivations

Figure 1 Example of actions viewed by subjects in the experiment The figure shows the real action (putting a book on a shelf ) and the

corresponding pretend version The dark line indicates the approximate point at which incomplete clips would end Note that subjects would seeonly one example of each action being performed by a given actor

1808 Journal of Cognitive Neuroscience Volume 16 Number 10

The opposite contrast comparing brain regions thatwere more active for viewing real actions than forviewing pretend actions in the covert condition pro-duced just one significant activation in the group anal-ysis (using a random effects model and a statisticalthreshold of p lt 005 uncorrected and a voxel extentof 30 as illustrated in Figure 2) That activation was inthe right precentral gyrus (BA 6)

Brain Activity Associated with Pretend Actionswith Real Objects in the Covert Condition

We were concerned that activations associated withviewing pretend actions may only appear if the pretendaction included an object that was missing from thescene So half of the pretend actions included the objectused in the real action while the other half of the

pretend actions did not include this object As illustratedin Figure 4 analyzing only those pretend actions thatincluded a real object in comparison with the real ac-tions revealed significant activations in the group anal-ysis ( p lt 05 uncorrected and a voxel extent of 10) inall the same regions implicated when collapsing acrossconditions with objects and those without with theexception of the activations in the medial temporal lobe(left amygdala and right parahippocampus)

DISCUSSION

Areas of the brain typically associated with mental statereasoning under explicit and overt instructions includethe (1) medial prefrontal cortex (2) areas of the poste-rior superior temporal gyrus and (3) the temporal poles(Gallagher amp Frith 2003) In the current study a coverttask was employed to determine whether theory ofmind areas are responsive to theory of mind contentin the absence of any instruction to reason about themental states of social agents or any requirement tomake a judgment about the actions of social agents inthe scene The results showed increased activations inareas of medial prefrontal and inferior prefrontal cortexand in areas of the temporo-parietal junction in re-sponse to pretend versus real actions consistent withprevious imaging studies There was no evidence in thisstudy of any anterior lateral temporal lobe (temporalpole) activations In addition there were further notableactivations in bilateral medial temporal lobe regions (leftamygdala and right parahippocampus) We discuss thislatter activation pattern first before moving to discussthe areas more typically seen in theory of mind neuro-imaging studies

The role of the amygdala in mental state reasoninghas been a matter of some debate Damage to the amyg-dala in various primates can cause disturbance of socialbehavior (Kling 1972) and humans with bilateral amyg-dala damage have problems in emotion recognitionfrom facial expressions and social judgments (AdolphsTranel amp Damasio 1998) as well as problems in ex-plicit theory of mind tasks involving judging the mentalstates of other people (Stone et al 2003 Fine et al2001) Further the amygdala may be a critical struc-ture that is damaged in autism (Abell et al 1999Aylward et al 1999) resulting in social cognition failuresforming the core of the disorder (Baron-Cohen et al1999)

However other theorists suggest that the amygdalamay play a limited role in social cognition citing datafrom mature and neonatal rhesus monkeys in whichthe amygdala was selectively and completely lesioned(Amaral et al 2003) Despite some social and emotionaldisruptions in these animals in the evaluation of threatsin the environment the mature animals demonstratednormal to increased levels of social interaction withconspecifics and neonatal lesioned animals engaged in

Table 2 Regions Based on Peak Activations of IncreasedActivity Associated with Viewing Pretend Actions Comparedwith Viewing Real Actions in the Covert Condition

BA x y z Voxelsz

Score

Anterior cingulatemedial frontal gyri

3269 6 31 32 102 387

Superior frontal gyrus 9 9 57 25 140 358

Medial frontal gyrus 10 3 52 3 51 387

R inferior frontal gyrus 47 33 20 11 335 426

R inferior frontal gyrus 44 56 18 10 54 363

L inferior frontal gyrus 47 39 14 13 238 365

L inferior frontal gyrus 44 48 10 16 31 318

L middle frontal gyrus 10 30 53 6 39 344

L middle frontal gyrus 11 45 40 12 32 336

L middle temporalgyrus

21 50 52 3 63 356

R superior temporalgyrus

22 59 58 14 56 366

R inferior temporalgyrus

19 48 70 2 130 356

R fusiform gyrus 37 42 53 12 33 335

L amygdala 18 4 12 33 335

R parahippocampalgyrus

21 10 22 36 371

R thalamus 21 11 12 131 392

Caudate 3 9 9 30 309

Cerebellum 12 60 40 195 383

R = right L = left BA = Brodmannrsquos area x y and z are in Talairachcoordinates all activations that survived a statistical threshold of p lt005 (uncorrected) and a voxel extent of 30

German et al 1809

typical social behaviors (eg facial expressions groom-ing and play) during development

Increased activity in the amygdala has only rarely beenreported in neuroimaging studies of mental state reas-oning (Gallagher amp Frith 2003) although one possiblereason might be the relative difficulty in demonstrating

activations in medial temporal lobe structures as hasbeen discussed in the case of hippocampus activationsin memory encoding and retrieval (Schacter amp Wagner1999) Note that two neuroimaging studies that havereported amygdala activation both involve the explicitcalculation of affective mental states from the eye regionof faces (Wicker Perrett Baron-Cohen amp Decety 2003Baron-Cohen et al 1999) rendering uncertain the rolethe amygdala plays beyond mediating responses to theemotional content of the stimuli This interpretation isalso offered to explain amygdala activation demonstrat-ed in a recent study looking at brain regions sensitive tocalculating the deceptive intent of an actor (Grezes et al2004) Note also that the amygdala and parahippocam-pal gyrus activity in the current study did not dependon processing information from facial expressions noron processing any other emotion information theactors in both real and pretend video clips in the currentstudy maintained neutral expressions throughout theactions Thus the current results provide qualified sup-port for the possible role of the amygdala in supportingmental state calculations (Brothers 1990)

The current study also identified areas of the brainmore typically associated with theory of mind contentincluding the medial prefrontal cortex (BA 3269 9 and10) inferior frontal cortex (BA 44 and 47) and regions

Figure 2 Four axial sections for two separate contrasts show significant activations that exceeded the statistical threshold Functional imagesare superimposed on one subjectrsquos high-resolution anatomical image with the left side of the image corresponding to the left side of the brain

Adjacent to the axial sections are the glass brain representations revealing all the significant activations for that particular contrast Viewing pretend

actions relative to viewing real actions significantly increased activations in the medial prefrontal cortex ventrolateral prefrontal cortex medial

temporal lobe and in the temporo-parietal junction whereas viewing real actions relative to pretend actions only produced a single activation in theright precentral gyrus

Figure 3 Peak signal change relative to the resting baseline at thelocal maxima of three key regions (ACC = anterior cingulate MeFG =

medial frontal gyrus STG = superior temporal gyrus) for the pretend

and real conditions

1810 Journal of Cognitive Neuroscience Volume 16 Number 10

of the temporo-parietal junction (BA 21 22) Howevermost theory of mind neuroimaging studies employexplicit theory of mind tasks where judgments aboutthe actions of social agents are confounded with instruc-tions to think about mental states in executing the taskbe it comprehension of verbal stories (Vogeley et al2001 Gallagher et al 2000 Fletcher et al 1995)understanding mental state-based humorous cartoons(Gallagher et al 2000) completing picture sequencesbased on agentsrsquo intentions (Brunet et al 2000) judg-ing the likely knowledge of another person (Goel et al1995) or judging the intentions of or explaining themotion patterns of triangles engaging in complex men-tal state-based interactions (Castelli et al 2000)

In one study the explicit instruction to take theintentional stance was isolated from content differencein the signal between the experimental (theory of mind)and control conditions (Gallagher et al 2002) In thisstudy subjects played a strategic game (paper scissorsstone) against either a human opponent (theory of mindcondition) or against a computer following one ofseveral known strategies (control condition) for severaltrials before being confronted with a series of the exactsame sequence of trials in each condition The onlydifference between conditions therefore was that theparticipants in the theory of mind condition were in-

structed to think about the beliefs of their opponentwhile those in the computer control condition were notall the properties of the signal being processed wereotherwise identical This study found increased activityin the paracingulate gyrus (BA 329) as reported forother explicit theory of mind tasks but interestinglythere were no activations seen in other areas typicallyassociated with theory of mind tasks (eg temporo-parietal areas or the temporal poles)

Gallagher et al (2002) argued that this circumscribedregion of medial prefrontal cortex is critically involvedin explicit mental state reasoningmdashtaking the inten-tional stancemdasheven in the absence of other behavioralcues associated with social agents such as mental statelanguage (Fletcher et al 1995) visual depictions ofhumans (Gallagher et al 2000 Baron-Cohen et al1999) or cues to social agency such as biologicalmotion (Castelli et al 2000) which have been shownin other studies to activate areas of superior temporalcortex (Grossman amp Blake 2001 see Allison Puce ampMcCarthy 2000 for a review) According to this analysisalthough superior temporal and temporal parietal areasmight be involved in some aspects of mental statereasoning (such as the detection and interpretation ofgoal directed actions) it is in medial prefrontal cortexregions where representations of beliefs and other

Figure 4 Pretend actions

with real objects compared to

real actions (middle panels

using a more liberal threshold)produced similar activations

as pretend actions with both

real objects and missing

objects

Pretend Actions vsReal Actions

Pretend (w real objects)Actions vs Real Actions

German et al 1811

knowledge-based mental states (eg pretends expec-tation) are calculated (Gallagher amp Frith 2003) Thispicture is supported by evidence suggesting that lesionsto frontal areas seem to cause disruptions in overtmental state reasoning (eg Rowe et al 2001 Loughet al 2001 Stuss et al 2001 Channon amp Crawford2000 Stone et al 1998)

The current study demonstrates that it is not neces-sary to explicitly instruct participants to take the inten-tional stance as did Gallagher et al (2002) or indeedmake any kind of overt judgment about mental states(eg Saxe amp Kanwisher 2003 Brunet et al 2000) to ac-tivate these medial frontal areas they can also be en-gaged by mental state content in situations where thebehavior of a social agent is such that a mental stateinterpretation is called formdasha phenomenon we dub theautomatic engagement of the intentional stance Partialsupport for this idea is provided by a previous studythat demonstrated medial prefrontal cortex activationsassociated with a covert task and theory of mind relatedcontent (Calder et al 2002) In that study increasedregional cerebral blood flow was observed in medialprefrontal regions (ie BA 89 coordinates 2 42 and 36)in response to increasing proportions of horizontallyaverted gaze The authors attributed this result to theautomatic engagement of processing of the possibletargets of the pictured personrsquos mental states in caseswhere the goal or focus of the actor was not obvious(eg the face was not looking at the subject herself seealso Singer et al 2004 Mitchell et al 2002 for cases oftheory of mind areas active in tasks with no explicitjudgments about othersrsquo mental states) Note also thatthe medial prefrontal and temporo-parietal areas areamong those identified as having high levels of resting-state metabolic activity (Gusnard et al 2001) andthis is consistent with the notion that spontaneousongoing automatic mental state calculation may occurin these regions

Although theory of mind is sometimes characterizedas a lsquolsquocomplex higher cognitive function and a ubiqui-tous element of social reasoning and problem solvingrsquorsquo(Gallagher et al 2002 p 814) there is a long traditionin cognitive developmental research that has assumedthat theory of mind knowledge is initially acquired bycognitive mechanisms that operate automatically overthe behavior of social agents to calculate representa-tions of the underlying mental states (see eg Leslie1987 1994a 2000a 2000b see also German amp Leslie2000 2001) A motivating consideration for this view isthe fact that very young children are capable of inter-preting even nonliteral behavior as such from early indevelopmentmdashunderstanding pretense emerges some-time between 18 and 24 months in the typical casemdashlong before any explicit knowledge about pretense hasbeen acquired and long before success is achieved onthe basic beliefndashdesire reasoning problems that areargued to mark mature theory of mind skills (eg the

false-belief task Wellman Cross amp Watson 2001 seealso Bloom amp German 2000)

In a pretend situation such as when mother talks into abanana as if it were a telephone the child is not con-fused nor does she learn that bananas are an additionalclass of object that can be labeled by the word lsquolsquotele-phonersquorsquo Instead she is capable of calculating motherrsquosunderlying mental state (Leslie 1987) most likely aidedby particular signals of pretense that the mother pro-vides which differ from those involved in real action(Lillard amp Witherington 2004) Because 18-month-oldchildren cannot plausibly know very much about men-tal states explicitly (see eg German amp Leslie 2001Lillard 1993 for discussion) Leslie (1987) proposed acomputational model in which the behavior of socialagents is taken as input to a specialized mechanismthat automatically calculates from that behavioral de-scription a mental state representation that makes ex-plicit information about the event including the agent(mother) and a mental state (eg pretense) that is takento a proposition (lsquolsquois a telephonersquorsquo) in respect to someobject anchored in the real world (of the banana Leslie1987 1994a 2000b) The proposition is lsquolsquodecoupledrsquorsquofrom typical inputndashoutput relations so as to avoid confu-sion with lsquolsquoprimaryrsquorsquo representations of the real world(see Leslie 1994b for further discussion)

This mechanism the lsquolsquotheory of mind mechanismrsquorsquo isproposed to underwrite the capacity for the later devel-opment of theory of mind knowledge Early theory ofmind abilities can be plausibly thought of as comprisingmechanisms that allow for mental state information tobe attended to even in very young children withoutmuch explicit knowledge about mental states and eventhough mental states cannot be directly seen heard orfelt and without the benefit of any formal instructionsto do so Being able to attend to mental states is a pre-requisite for learning about them (German amp Leslie2000 Leslie 2000a) It is our contention that the currentstudy provides support for the continued operation ofreliably developing mechanisms for the detection andrepresentations of mental statesmdashsystems that are en-gaged automatically in response to certain streams ofperceived behavior

One possible alternative characterization of the resultsreported here might be in terms of whether the pretendfilms cause activations that are in response to lsquolsquonoveltyrsquorsquoArguably participants might have seen pretend actionsas more novel than real actions There are two ways toread this alternative One interpretation is that the pat-tern of brain responses identified here might be simplya result of some general reaction to novel stimuli orstimuli that signal a novel response type are requiredIt has been reported previously that prefrontal ERPresponses (eg the P3a occurring 300ndash400 msec post-stimulus) are seen when participants receive cues indi-cating they should change rules in the Wisconsin CardSorting Task (eg Barcelo Periaoez amp Knight 2002)

1812 Journal of Cognitive Neuroscience Volume 16 Number 10

and lesions to the lateral prefrontal cortex result in anattenuated novelty response (Knight 1984) Hippocam-pal regions are also implicated given that patients withposterior hippocampal lesions show attenuated ERPresponses to novelty (Knight 1996) Functional imagingdata also suggest that a testndashretest-reliable lateral tem-poral signal in response to novel auditory stimuli canoccur in both bilateral superior temporal and bilateralinferior frontal gyri (Kiehl amp Liddle 2003)

Although the current study cannot rule out that ac-tivations observed to pretense were partially caused bynovelty there are reasons to suspect that the patternof responses is not entirely explicable that way Firstnovelty has been associated with lateral rather thanmedial activations in frontal cortex (Kiehl amp Liddle2003 Kirino Belger Goldman-Rakic amp McCarthy2000 Knight 1984) Second Ferstl and von Cramon(2002) showed that in a task requiring participants tojudge the coherence of sentences it was coherent andnot incoherent sentences that resulted in medial fron-tal activations Plausibly incoherent sentences wouldbe more likely to be construed as novel than coherentsentences

Alternatively perhaps the mental state regions areengaged not because of novelty in general but ratherbecause of novelty within the domain of human actionPerhaps a range of atypical human actions including butnot limited to pretense result in the automatic engage-ment of mental state calculations It is exactly accordingto this sense of novelty that the current predictions weremade Pretense was selected as a scenario likely to be animportant behavioral cue to the calculation of mentalstates requiring representations decoupled from realitygiven its important early emergence in infancy as amarker of mental state representation It is not claimedhere that the activations seen in this study should bethought of as caused if and only if pretense is detectedbut rather that pretend actions should be one casewhere the intentional stance is reliably and automaticallyengaged with no need for any prior instruction to attendto mental states

It follows from this perspective that other classes ofhuman action might also result in the engagementof mental state reasoning areas Mitchell et al (2002)showed theory of mind-type activations in response tojudgments made about people as opposed to objectsand Singer et al (2004) showed that anterior cingulateactivations were associated with empathy toward aclose friendrsquos pain in both cases there was no explicitdirection to consider the mental states of others Wespeculate that other cases where activations might beexpected will involve cases where the goal that theagent pursues is atypical (indeed pretense might bethought of as a special case within this class involvinga lsquolsquoknowingrsquorsquo violation of a typical goal) Other kindsof actions that violate the actorsrsquo goals (eg wheresomeone fails to gain an object because they are

mistaken about the location of that object) might alsoresult in theory of mind areas to be activated automat-ically (ie with no instruction to calculate the mentalstate content)

A final issue concerns the extent to which activationsin mental state reasoning tasks such as that reported inthe current study and indeed more broadly should bethought of as resulting from processes that have to dowith the deployment of representations of mental statesthemselves (eg conceptual competence) or as result-ing from processes (perhaps that have a more domaingeneral penetrable character) associated with per-forming calculations over those special representations(eg maintenance of representations in working mem-ory inhibition of alternative mental state contentsselection among those alternatives etc)

Distinguishing theory of mind-specific activationsfrom associated task performance activations requiresthe adoption of lsquolsquominimal pairrsquorsquo designs For exampleSaxe and Kanwisher (2003) showed that activations intemporo-parietal areas are specific to belief reasoning bycontrolling for task structure using a nonmental analog(eg the lsquolsquofalse-photorsquorsquo task Zaitchik 1990) This ap-proach will be important in deciding whether otherlsquolsquotheory of mindrsquorsquo areas are activated because they per-form (relatively) domain general processing resourcesshared by belief reasoning and other tasks with similarcomputational structure (see also Saxe et al 2004 Ferstlamp Von Cramon 2002)

Whereas the cognitive developmental literature hasvery clearly articulated the competencendashperformancedistinction in thinking about developing performanceon the false-belief task (see eg Bloom amp German 2000for a review) the possible separable contributions to theactivation patterns seen in theory of mind neuroimagingstudies of (1) neural areas representing mental stateconcepts themselves (eg mechanisms producing men-tal state representations) and (2) neural areas responsi-ble for maintaining manipulating and selecting betweencandidate representations with different content has notyet been clearly delineated (see Leslie et al in press forspecific models of beliefndashdesire reasoning that articulatethis distinction)

Conclusions

In summary the current study shows that a set ofneural mechanisms for the interpretation of the actionsof social agents are automatically engaged when aviewer is confronted by instances of social behaviorthat require a decoupled mental state representationto be calculatedmdashin this case an act of pretense Thesemechanisms appear to comprise circuits in temporo-parietal regions possibly associated with the detectionof goal directed social action (eg Frith amp Frith 1999)and areas of the prefrontal cortex that have been as-sumed to reflect the explicit adoption of the lsquolsquointentional

German et al 1813

stancersquorsquo including the calculation of representationsdecoupled from reality (eg Gallagher et al 2002) Itis our view that research on further cases where episte-mic mental state calculations might be engaged by theactions of social agents without explicit experimental in-struction will be an important addition to research effortsbased on tasks involving explicit and overt tasks of socialreasoning Via this dual route the functional and neuralorganization of this fundamental social capacity may bemore rapidly elucidated

METHODS

Subjects

Sixteen subjects between the ages of 18 and 29 wererecruited from the Dartmouth College community(8 men and 8 women) No subject reported any abnor-mal neurological history and all had normal or cor-rected-to-normal visual acuity Subjects were paid fortheir participation All subjects gave informed consent inaccordance with the guidelines set by the Committee forthe Protection of Human Subjects at Dartmouth Collegeand by the Human Subjects Committee at the Universityof California Santa Barbara CA

Materials and Design

We created a series of video clips of actors performingsimple acts (eg putting a hat on a hook pouring a cupof tea) There were 12 different actions (see Appendix)each one performed by one of four different actors(2 men and 2 women) Each clip was filmed (on a CanonZR 45 DV Camcorder) for each actor both as a realaction (the actor really performed the act) and as apretend action (the actor pretended to perform theaction) For half of the pretend actions one or moreobjects for the act was missing (eg no teapot) whilefor the other half all objects were present and thepretense consisted of miming the act with the objectsinstead of really performing it Actors were instructednot to make overt pretense lsquolsquodisplaysrsquorsquo to the camera(eg significant facial affect overt exaggeration to thecamera) because as indicated we were interested inisolating a minimal difference in content signal betweenpretense and real acts as far as possible

Four sets of clips were then created from this rawmaterial by editing the actions such that half the timethe clip ended before the act was finished (incompleteclips) and the rest of the time the act finished beforethe clip ended Thus there were four kinds of clips inthe study (PC = pretend complete PI = pretendincomplete RC = real complete and RI = real incom-plete) Each clip was approximately 6 sec long andthese were edited in iMovie (Apple Computer Cuperti-no CA) to create trials that comprised a short sectionof blue screen (1 sec) followed by a video clip of one

of the four types followed by a further short sectionof blue screen (1 sec) before a white response screen(2 sec) reminding the participant to respond and in-dicating the relevant options on the appropriate side(eg left = complete right = incomplete) Each trialthus lasted 10 sec The clips varied slightly in lengtharound a mean length of 6 sec with time added orsubtracted to the blue screen such that the trial lengthwas constant

These clips were arranged into blocks of three clipsfor a total of 30 sec per block Each block contained justone of the four kinds of clips Selection was constrainedsuch that each action performed by each actor appearedin only one possible film type and therefore subjectswould not see the same actor performing the same clipas both pretend and real or as both complete andincomplete The 16 blocks of clips along with 4 restblocks consisting entirely of blue screen were thenarranged into a random sequence and edited togetherbefore being exported to DVD (iDVD Apple) The totallength of each DVD film was 10 min and 10 sec (eg20 blocks of 30 sec each plus 10 sec of blue screen at theoutset of each film)

Behavioral Procedure

Subjects were instructed that they would be seeing asequence of short video clips depicting peopleperforming simple everyday actions They were told towatch the clips to the end because in some cases theclips have been edited such that they would finishbefore the action was complete but in other cases theclips would finish only after the action was completeThey were asked to make responses with button presseson a box held in the left hand in response to completeclips and in the right hand in responses to incompleteclips only after they saw the white response screen aftereach clip Responses were recorded to determinewhether participants responded appropriately

Functional Magnetic ResonanceImaging Parameters

Imaging was performed at Dartmouth College on a15-T whole-body scanner (General Electric MedicalSystems Signa Milwaukee WI) with a standard headcoil Visual stimuli were presented via a standard DVDplayer and projected to a screen positioned at the headend of the bore by an Epson (model ELP-7000) LCDprojector Subjects viewed the screen through a mirrormounted on top of the head coil A fiber-optic light-sensitive key-press that interfaced with the PsyScopeButton Box (New Micros Dallas TX) was used to recordsubjectsrsquo responses Cushions were used to minimizehead movement

Anatomical images were acquired using a high-resolution 3-D spoiled gradient recovery sequence

1814 Journal of Cognitive Neuroscience Volume 16 Number 10

(124 sagittal slices TE = 6 msec TR = 2500 msec flipangle = 258 voxel size = 1 1 12 mm) Each ses-sion included two functional runs Functional imageswere collected using a gradient spin-echo echo-planarsequence sensitive to BOLD contrast (T2 240 repeti-tions TR = 2500 msec TE = 35 msec flip angle = 908375 375 mm in-plane resolution 25 axial slices at45-mm slice thickness and 1-mm skip between slicesallowed for imaging of the whole brain) The first fourscans of each functional run were dropped to allow forT1 saturation to stabilize Signal loss was observed insmall portions of the orbito-frontal cortex due to thesinus cavities

Functional Magnetic Resonance ImagingData Analyses

fMRI data were analyzed using Statistical ParametricMapping software (SPM2 Wellcome Department ofCognitive Neurology London UK Friston et al 1995)Functional data were first corrected for differences inslice acquisition timing by resampling all slices in timeto match the first slice then realigned across the tworuns to correct for head movement coregistered witheach participantrsquos anatomical data and then trans-formed into a standard anatomical space based onthe ICBM 152 brain template (Montreal NeurologicalInstitute) which approximates Talairach and Tour-nouxrsquos (1988) atlas space Images were resampled into3-mm cubic voxels and then spatially smoothed usingan 8-mm full width half maximum isotropic Gaussiankernel

For each participant a general linear model incorpo-rating task effects (modeled as a box-car function con-volved with the canonical hemodynamic responsefunction) and a linear trend was used to compute t con-trast images (weighted parameter estimates) for eachcomparison at each voxel These individual contrastimages were then submitted to a second-level random-effects analysis to create group images The resultingstatistical maps were thresholded at p lt 005 (uncor-rected) To reduce the rate of false positives a spatialextent threshold of 30 contiguous voxels was also ap-plied (Poline Worsley Evans amp Friston 1997 Formanet al 1995 Xiong et al 1995 Friston Worsley Frack-owiak Mazziotta amp Evans 1994)

A peak signal change analysis was performed at thelocal maxima of the key regions of interest This analysisentailed computing the percent signal intensity duringthe pretend real and resting-baseline epochs on asubject-by-subject basis starting at 75 sec after the startof the epoch (ie allowing the hemodynamic responseto peak) and extending to the end of the epoch Peaksignal relative to the resting baseline was then derived inthe pretend and real conditions by subtracting theresting baseline activity

APPENDIX

Acknowledgments

This research was supported by the Center for Brain ImagingUniversity of California Santa Barbara CA We thank AldoCimino Elaine German Max Krasnow Jayne Niehaus AlexSchwartz and Danielle Truxaw for help with stimuli prepara-tion Tammy Laroche for help with subject recruitment andJack van Horn for technical assistance We also thank ToddHeatherton and two anonymous reviewers for helpful com-ments on a previous draft

Reprint requests should be sent to Tim P German Departmentof Psychology University of California Santa Barbara CA 93106-9660 or via e-mail germanpsychucsbedu

The data reported in this experiment have been deposited inthe fMRI Data Center (httpwwwfmridcorg) The accessionnumber is 2-2004-1173P

REFERENCES

Abell F Krams M Ashburner J Passingham R Friston KFrackowiak R Happe F Frith C amp Frith U (1999)The neuroanatomy of autism A voxel-based whole brainanalysis of structural scans NeuroReport 10 1647ndash1651

Adolphs R Tranel D amp Damasio A R (1998) The humanamygdala in social judgment Nature 393 470ndash474

Allison T Puce A amp McCarthy G (2000) Social perceptionfrom visual cues Role of the STS region Trends in CognitiveSciences 4 267ndash278

Amaral D G Bauman M D Capitanio J P Lavenex PMason W A Mauldin-Jourdain M L amp Mendoza S P(2003) The amygdala Is it an essential component of theneural network for social cognition Neuropsychologia 41517ndash522

Aylward E H Minshew N J Goldstein G Honeycutt N AAugustine A M Yates K O Barta P E ampPearlson G D (1999) MRI volumes of amygdala andhippocampus in non-mentally retarded autisticadolescents and adults Neurology 53 2145ndash2150

Barcelo F Perianez J A amp Knight R T (2002) Think

Actions for covert task

Throw paper ball into trash

Opening a jar

Putting on shoe

Getting a cup from a shelf

Eat some yogurt with a spoon

Driving a screw into wall

Blowing out a candle

Crack an egg into a pan

Peeling a potato

Lighting a match

Unscrewing a light bulb

Slice a potato

German et al 1815

differently A brain orienting response to task noveltyNeuroReport 13 1887ndash1892

Baron-Cohen S (1995) Mindblindness An essay on autismand theory of mind Cambridge MIT Press

Baron-Cohen S Leslie A M amp Frith U (1985) Does theautistic child have a theory of mind Cognition 2137ndash46

Baron-Cohen S Ring H Moriarty J amp Schmitz B (1994)Recognition of mental state terms Clinical findings inchildren with autism and a functional neuroimaging studyof normal adults British Journal of Psychiatry 165640ndash649

Baron-Cohen S Ring H A Wheelwright S Bullmore E TBrammer M J Simmons A amp Williams S C (1999) Socialintelligence in the normal and autistic brain An fMRI studyEuropean Journal of Neuroscience 11 1891ndash1898

Bloom P amp German T P (2000) Two reasons to abandonthe false belief task as a test of theory of mind Cognition77 B25ndashB31

Brothers L (1990) The social brain A project for integratingprimate behavior and neurophysiology in a new domainConcepts in Neuroscience 1 27ndash51

Brunet E Sarfati Y Hardy-Bayle M amp Decety J (2000) APET investigation of the attribution of intentions with anonverbal task Neuroimage 11 157ndash166

Calder A J Lawrence A D Keane J Scott S KOwen A M Christoffels I amp Young A W (2002) Readingthe mind from eye gaze Neuropsychologia 40 1129ndash1138

Castelli F Happe F Frith U amp Frith C (2000) Movementand mind A functional imaging study of perception andinterpretation of complex intentional movement patternsNeuroimage 12 314ndash325

Channon S amp Crawford S (2000) The effects of anteriorlesions on performance on a story comprehension testLeft anterior impairment on a theory of mind-type taskNeuropsychologia 38 1006ndash1017

Dennett D C (1987) The intentional stance CambridgeMIT Press

Ferstl E C amp von Cramon D Y (2002) What does thefrontomedian cortex contribute to language processingCoherence or theory of mind Neuroimage 17 1599ndash1612

Fine C Lumsden J amp Blair R J R (2001) Dissociationbetween lsquolsquotheory of mindrsquorsquo and executive functions in apatient with early left amygdala damage Brain 124287ndash298

Fletcher P C Happe F Frith U Baker S C Dolan R JFrackowiak R S J amp Frith C D (1995) Other minds in thebrain A functional imaging study of lsquolsquotheory of mindrsquorsquo instory comprehension Cognition 57 109ndash128

Forman S D Cohen J D Fitzgerald M Eddy W FMintun M A amp Noll D C (1995) Improved assessmentof significant activation in functional magnetic resonanceimaging (fMRI) Use of a cluster-size threshold MagneticResonance in Medicine 33 636ndash647

Friston K J Holmes A P Worsley K J Poline J BFrith C D amp Frackowiak R S J (1995) Statisticalparametric maps in functional imaging A general linearapproach Human Brain Mapping 2 189ndash210

Friston K J Worsley K J Frackowiak R S J Mazziotta J Camp Evans A C (1994) Assessing the significance of focalactivations using their spatial extent Human BrainMapping 1 214ndash220

Frith C D amp Frith U (1999) Interacting mindsmdashA biologicalbasis Science 286 1692ndash1695

Gallagher H L amp Frith C D (2003) Functional imaging oflsquolsquotheory of mindrsquorsquo Trends in Cognitive Sciences 7 77ndash83

Gallagher H L Happe F Brunswick N Fletcher P CFrith U amp Frith C D (2000) Reading the mind in

cartoons and stories An fMRI study on lsquolsquotheory of mindrsquorsquoin verbal and nonverbal tasks Neuropsychologia 3811ndash21

Gallagher H L Jack A I Roepstorff A amp Frith C D (2002)Imaging the intentional stance in a competitive gameNeuroimage 16 814ndash821

German T P amp Leslie A M (2000) Attending to and learningabout mental states In P Mitchell amp K Riggs (Eds)Childrenrsquos reasoning and the mind (pp 229ndash252)Hove Psychology Press

German T P amp Leslie A M (2001) Childrenrsquos inferencesfrom knowing to pretending and believing British Journalof Developmental Psychology 19 59ndash83

Goel V Grafman J Sadato N amp Hallett M (1995)Modelling other minds NeuroReport 6 1741ndash1746

Grezes J Frith C D amp Passingham R (2004) Brainmechanisms for inferring deceit in the actions of othersJournal of Neuroscience 24 5500ndash5505

Grossman E D amp Blake R (2001) Brain activity evoked byinverted and imagined biological motion Vision Research41 1475ndash1482

Gusnard D A Akbudak E Shulman G L amp Raichle M E(2001) Medial prefrontal cortex and self-referential mentalactivity Relation to a default mode of brain functionProceedings of the National Academy of Sciences USA98 4259ndash4264

Kelley W M Macrea C N Wyland C L Caglar S Inati Samp Heatherton T F (2002) Finding the self An event-related fMRI study Journal of Cognitive Neuroscience14 785ndash794

Kiehl K A amp Liddle P F (2003) Reproducibility of thehemodynamic response to auditory oddball stimuli Asix-week testndashretest study Human Brain Mapping 1842ndash52

Kirino E Belger A Goldman-Rakic P amp McCarthy G(2000) Prefrontal activation evoked by infrequenttarget and novel stimuli in a visual detection taskAn event related functional magnetic resonanceimaging study The Journal of Neuroscience 206612ndash6618

Kling A (1972) Effects of amygdalectomy on socio-affectivebehavior in non-human primates In B E Eleftheriou(Ed) Neurobiology of the amygdala (pp 511ndash536)New York Plenum

Knight R T (1984) Decreased response to novel stimuliafter prefrontal lesions in man Electroencephalographyand Clinical Neurophysiology Evoked Potentials 599ndash20

Knight R T (1996) Contribution of human hippocampalregion to novelty detection Nature 383 256ndash259

Leslie A M (1987) Pretense and representation Theorigins of lsquolsquotheory of mindrsquorsquo Psychological Review 94412ndash426

Leslie A M (1994a) Pretending and believing Issues in thetheory of ToMM Cognition 50 211ndash238

Leslie A M (1994b) ToMM ToBy and agency Corearchitecture and domain specificity In L Hirschfeld ampS Gelman (Eds) Mapping the mind Domain specificityin cognition and culture (pp 119ndash148) New YorkCambridge University Press

Leslie A M (2000a) How to acquire a lsquolsquorepresentationaltheory of mindrsquorsquo In D Sperber (Ed) MetarepresentationsA multidisciplinary perspective (pp 197ndash223) Oxford UKOxford University Press

Leslie A M (2000b) lsquolsquoTheory of mindrsquorsquo as a mechanismof selective attention In M Gazzaniga (Ed) The newcognitive neurosciences (2nd ed pp 1235ndash1247)Cambridge MIT Press

1816 Journal of Cognitive Neuroscience Volume 16 Number 10

Leslie A M German T P amp Pollizi P (in press) Belief-desirereasoning as a process of selection Cognitive Psychology

Leslie A M amp Thaiss L (1992) Domain specificity inconceptual development Neuropsychological evidencefrom autism Cognition 43 225ndash251

Lillard A S (1993) Young childrenrsquos conceptualization ofpretense Action or mental representational state ChildDevelopment 64 372ndash386

Lillard A S amp Witherington D (2004) Mothersrsquo behaviormodifications during pretense snacks and their possiblesignal value for toddlers Developmental Psychology 4095ndash113

Lough S Gregory C amp Hodges J R (2001) Dissociationof social cognition and executive function in frontalvariant frontotemporal dementia Neurocase 7123ndash130

Mason M F Banfield J F amp Macrea C N (2004) Thinkingabout actions The neural substrates of person knowledgeCerebral Cortex 14 209ndash214

McCabe K Houser D Ryan L Smith V amp Trouard T(2001) A funcional imaging study of cooperation intwo-person reciprocal exchange Proceedings of theNational Academy of Sciences USA 9811832ndash11835

Mitchell J P Heatherton T F amp Macrea C N (2002)Distinct neural systems subserve person and objectknowledge Proceedings of the National Academy ofSciences USA 99 15238ndash15243

Poline J B Worsley K J Evans A C amp Friston K J(1997) Combining spatial extent and peak intensityto test for activations in functional imaging Neuroimage5 83ndash96

Ramnani N amp Miall R C (2004) A system in the human brainfor predicting the actions of others Nature Neuroscience 785ndash90

Rowe A D Bullock P R Polkey C E amp Morris R G (2001)lsquolsquoTheory of mindrsquorsquo impairments and their relationship toexecutive functioning following frontal lobe excisionsBrain 124 600ndash616

Ruby P amp Decety J (2003) What you believe versus what youthink they believe A neuroimaging study of conceptualperspective taking European Journal of Neuroscience17 2475ndash2480

Sabbagh M A amp Taylor M (2000) Neural correlates of thetheory-of-mind reasoning An event-related potential studyPsychological Science 11 46ndash50

Saxe R Carey S amp Kanwisher N (2004) Understandingother minds Linking developmental psychology andfunctional neuroimaging Annual Review of Psychology55 87ndash124

Saxe R amp Kanwisher N (2003) People thinking aboutpeople The role of the tempo-parietal junction in lsquolsquotheoryof mindrsquorsquo Neuroimage 19 1835ndash1842

Schacter D L amp Wagner A D (1999) Medial temporal lobeactivations in fMRI and PET studies of episodic encoding andretrieval Hippocampus 9 7ndash24

Scholl B J amp Leslie A M (1999) Modularity developmentand lsquolsquotheory of mindrsquorsquo Mind and Language 14 131ndash153

Singer T Seymour B OrsquoDoherty J Kaube HDolan R J amp Frith C D (2004) Empathy for paininvolves affective but not sensory components of painScience 303 1157ndash1162

Stone V E Baron-Cohen S Calder A Keane J amp Young A(2003) Acquired theory of mind impairments in individualswith bilateral amygdala lesions Neuropsychologia 41209ndash220

Stone V E Baron-Cohen S amp Knight R T (1998) Frontallobe contributions to theory of mind Journal of CognitiveNeuroscience 10 640ndash656

Stuss D T Gallop G G Jr amp Alexander M P (2001) Thefrontal lobes are necessary for lsquolsquotheory of mindrsquorsquo Brain 124279ndash286

Talairach J amp Tournoux P (1988) Co-planar stereotaxicatlas of the brain New York Thieme

Vogeley K Bussfeld P Newen A Herrmann FHappe F Falkai P Maier W Shah N J Fink G Ramp Zilles K (2001) Mind reading Neural mechanisms oftheory of mind and self-perspective Neuroimage 14170ndash181

Wellman H M Cross D amp Watson J (2001) Meta-analysis oftheory-of-mind development The truth about false beliefChild Development 72 655ndash684

Wicker B Perret D I Baron-Cohen S amp Decety J (2003)Being the target of anotherrsquos emotion A PET studyNeuropsychologia 41 139ndash146

Xiong J Gao J Lancaster J L amp Fox P T (1995)Clustered pixels analysis for functional MRI activationstudies of the human brain Human Brain Mapping 3287ndash301

Zaitchik D (1990) When representations conflict with realityThe preschoolerrsquos problem with false beliefs and lsquofalsersquophotographs Cognition 35 41ndash68

German et al 1817