Frontal Lobe Contributions to Theory of Mind

Valerie E StoneUniversity of California Davis

Simon Baron-CohenUniversity of Cambridge

Robert T KnightUniversity of California Davis

Abstract

n ldquoTheory of mindrdquo the ability to make inferences aboutothersrsquo mental states seems to be a modular cognitive capacitythat underlies humansrsquo ability to engage in complex socialinteraction It develops in several distinct stages which can bemeasured with social reasoning tests of increasing difcultyIndividuals with Aspergerrsquos syndrome a mild form of autismperform well on simpler theory of mind tests but show decitson more developmentally advanced theory of mind tests Wetested patients with bilateral damage to orbito-frontal cortex(n = 5) and unilateral damage in left dorsolateral prefrontal

cortex (n = 5) on a series of theory of mind tasks varying indifculty Bilateral orbito-frontal lesion patients performed simi-larly to individuals with Aspergerrsquos syndrome performing wellon simpler tests and showing decits on tasks requiring moresubtle social reasoning such as the ability to recognize a fauxpas In contrast no specic theory of mind decits were evi-dent in the unilateral dorsolateral frontal lesion patients Thedorsolateral lesion patients had difculty only on versions ofthe tasks that placed demands on working memory n

INTRODUCTION

Humans like many other species use a variety of cues(facial expression body posture tone of voice) to predictothersrsquo behavior An animal that recognizes another ani-malrsquos threatening body posture for example might pro-duce a defensive response in anticipation of a possibleattack However humans do not simply respond to oth-ersrsquo behavior We also explicitly model and respond toother peoplersquos mental states their knowledge intentionsbeliefs and desires This ability to make inferences aboutothersrsquo mental states has been termed theory of mind(Premack amp Woodruff 1978 Wellman 1990) Theory ofmind is the term most widely used in the literature andis the term we will use here There has been debate overwhether the ability to infer othersrsquo mental states is a trueimplicit ldquotheoryrdquo or the result of more general inferentialabilities (Astington amp Gopnik 1991 Gopnik amp Wellman1992) or whether it is best characterized as taking theldquointentional stancerdquo (Dennett 1987) We will not be ad-dressing these controversies here Rather our concern iswhether particular brain regions may subserve the abil-ity to make mentalistic inferences

Theory of mind shows evidence of modularity in the

copy 1998 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 105 pp 640ndash656

same sense that language does (1) Theory of mind canbe selectively impaired in the developmental disorder ofautism while other aspects of cognition are relativelyspared (Baron-Cohen Leslie amp Frith 1985 Baron-Cohen1989b Baron-Cohen 1995) (2) Theory of mind can beselectively spared while other cognitive functions areimpaired as in Downrsquos syndrome and Williams syndrome(Karmiloff-Smith Klima Bellugi Grant amp Baron-Cohen1995) (3) Use of theory of mind is also rapid (4) auto-matic requiring no effortful attention (Heider amp Simmel1944) and (5) universal as far as is known (Avis amp Harris1991) (6) Finally theory of mind has a particular stereo-typed developmental sequence

The structure of the theory of mind mechanism canbe elucidated by examining what happens at each devel-opmental stage and what happens when there is a break-down at particular developmental stages Theory of mindrst manifests itself in joint attention and protodeclara-tive pointing (Baron-Cohen 1989a 1995) at about 18months In joint attention the child is able to understandnot only what another individual is looking at but thatthe child and some other person are looking at the sameobject Before 18 months infants may be able to under-stand the fact that ldquoMommy sees the toyrdquo but around 18

months the child begins to understand ldquoMommy seesthe toy that I seerdquo In protodeclarative pointing the childuses pointing to call adultsrsquo attention to objects that thechild wants them to attend to Many children withautism do not show either joint attention or protodecla-rative pointing (Baron-Cohen 1989a 1995) They do notlook at what other people look at nor do they usepointing to draw adultsrsquo attention to things The nextstage in the development of theory of mind is pretendplay in which children are able to decouple pretendfrom reality Between 18 and 24 months children beginto understand the mental state of ldquopretendrdquo (Leslie1987) Also by age 2 children seem to have a rm graspof the mental state of desire for example ldquoJohn wants ahamburgerrdquo (Wellman amp Woolley 1990) Childrenrsquos un-derstanding of desire precedes their understanding ofbelief

Between ages 3 and 4 children develop the ability tounderstand false belief (Gopnik amp Astington 1988Johnson amp Wellman 1980 Wellman 1990 Wimmer ampPerner 1983) Prior to this age a child does not under-stand that other people can hold beliefs about the worldthat differ from the childrsquos own Thus children assumethat other people know the same things they knowBetween ages 3 and 4 however children begin to under-stand that other people may not know all the things thatthey know and therefore that others may hold falsebeliefs Tests of false belief measure the ability of chil-dren to understand that another person can hold a beliefthat is mistaken These tests demonstrate that childrenare representing othersrsquo mental states othersrsquo beliefsrather then the physical state of the world or their ownstate of knowledge (Dennett 1978)

Between ages 6 and 7 children begin to understandthat other people can also represent mental states At thisage children begin to be able to understand second-order false belief ldquobelief about beliefrdquo (Perner amp Wimmer1985) In a typical second-order false belief task theproblem might run something like this A man and awoman are in a room The woman puts something some-where such as putting a book on a shelf She then leavesthe room The man hides the book in another locationUnbeknownst to him the woman is peeking backthrough a keyhole or a window and sees him movingthe book The subject is asked ldquoWhen the woman comesback in where will the man think that she thinks thebook isrdquo To solve this problem the child needs to beable to represent not only each personrsquos belief stateabout the location of the object but also the manrsquosmistaken belief about the womanrsquos belief state

Theory of mind can break down at certain of thesedevelopmental stages Children with autism who do notshow joint attention may never develop these theory ofmind abilities Children with autism are impaired in pre-tend play (Baron-Cohen 1995) Most children withautism cannot solve false belief tasks or second-orderfalse belief tasks (Baron-Cohen et al 1985 Perner

Leekam amp Wimmer 1987) Some higher-functioning in-dividuals with autism can eventually pass rst-order falsebelief tasks but will fail second-order false belief tasks(Baron-Cohen 1989b Happeacute 1993) In general autisticsrsquodifculties are with epistemic mental states concerningknowledge or belief They do seem to understand themental state of desire (Baron-Cohen Leslie amp Frith 1986Tager-Flusberg 1989 1993)

It should be noted that children with autism andyoung children do not simply lack the ability to dometa-representation They can pass what is called a falsephotograph test (Zaitchik 1990) In this test a Polaroidpicture is taken of a toy placed on a table The toy isthen moved and then the photo comes out and is devel-oped Before the child sees what is in the photographthe experimenter asks ldquoWhat will the picture showrdquoYoung children and children with autism are not fooledinto thinking that the photograph will show a table withnothing on it even though the toy has been moved Thusthey can understand physical representations such asphotographs and pictures but not mental repre-sentations (Charman amp Baron-Cohen 1992 1995Leekam amp Perner 1991 Leslie amp Thaiss 1992)

Later between ages 9 and 11 children develop furthertheory of mind abilities such as the ability to understandand recognize faux pas A faux pas occurs when some-one says something they should have not have said notknowing or not realizing that they should not say it Tounderstand that a faux pas has occurred one has torepresent two mental states that the person saying itdoes not know that they should not say it and that theperson hearing it would feel insulted or hurt Thus thereis both a cognitive component and empathic affectivecomponent On our new test of faux pas detectionBaron-Cohen OrsquoRiordan Stone Jones and Plaisted(1997) found that girls could perform well on this testby age 9 boys by age 11 Boys and girls of ages 7 or 8although they could pass rst- and second-order falsebelief tasks did not perform well on the faux pas task

To validate that the faux pas task is indeed a theoryof mind test Baron-Cohen et al (1997) also tested it onindividuals with Aspergerrsquos syndrome a mild form ofautism In these individuals language develops at a nor-mal time pace and their IQ is often normal Howeverthey still have many subtle social decits On our newtest of faux pas detection Baron-Cohen et al (1997)found that like 7 to 8 year-olds individuals with Asper-gerrsquos syndrome could pass rst- and second-order falsebelief tasks but were impaired on the faux pas task Theirtheory of mind performance was comparable to that of7- to 9-year-old children The faux pas task is a thus agood measure of subtle theory of mind decits Perfor-mance on the most developmentally advanced theory ofmind tasks is an index of how severe a personrsquos theoryof mind decit is Subtle theory of mind decits can onlybe picked up with the most developmentally advancedtasks

Stone et al 641

Little is known about the neurological basis for theoryof mind Such a complex cognitive ability does not seema likely candidate for localizationmdasha neural network orcircuit is more plausible Recent neuroimaging studieshave reported that regions of the frontal lobes appear tobe active during theory of mind tasks suggesting thatthese may be part of a theory of mind circuit Baron-Cohen Ring Moriarty Schmitz Costa and Ell (1994)found orbito-frontal activation during a simple theory ofmind task requiring recognition of mental state termsFletcher et al (1995) found activation in Brodmannrsquosareas 8 and 9 in the left medial frontal cortex during amore complex theory of mind task involving deceptionand belief attribution Goel Grafman Sadato and Hallett(1995) also found activation in the left medial frontalcortex during a task requiring mental state inferences

Baron-Cohen et al (1994) used a task requiring sub-jects to judge whether each word on a list of words hadto do with the mind or was something the mind coulddo and compared it to a task requiring a judgment ofwhether each word on another list had to do with thebody or was something the body could do Children withautism performed poorly on this mental state terms taskbut not on the body terms indicating that this task wasmeasuring theory of mind Using single photon emissioncomputerized tomograph (SPECT) imaging on a sampleof developmentally normal control subjects males aged20 to 30 Baron-Cohen et al found that right orbito-fron-tal cortex (OFC) was signicantly more active during themental state term recognition task than during the con-trol task relative to frontal polar cortex and posteriorregions One limitation of their study however is thatthey only carried out hypothesis-led region-of-interest(ROI) analyses and did not measure activation in themedial or dorsolateral frontal cortex so it is unknownhow active those areas were during their task The taskalso has no inference component

Fletcher et al (1995) and Goel et al (1995) used morecomplex tasks requiring subtle inferences about mentalstates In Goel et alrsquos study subjects were asked to makeinferences about objects that required either a visualdescription of the object memory retrieval an inferencefrom the objectrsquos form to its function or an inferencethat required modeling another personrsquos mental stateThey found selective activation for the task requiringmentalistic inferences in left medial frontal cortex andthe left temporal lobe In Fletcher et alrsquos study subjectsread a story and then answered a question about thestory that required a mentalistic inference They an-swered the question silently to themselves without mak-ing any overt response There were two control tasks forcomparison stories involving subtle physical inferencesbut not mentalistic inferences and paragraphs consistingof unrelated sentences Subjects were told which typeof story they were about to read and then were scannedboth while reading and while answering the question

Using positron emission tomography (PET) imagingwhen pixels active during the physical inference storieswere subtracted from pixels active during the theory ofmind stories Brodmannrsquos areas 8 and 9 and the anteriorcingulate showed up as active only during the theory ofmind task OFC was not specically active during thetheory of mind task They conclude that their data ldquopin-pointed the medial dorsal region of the left frontal cor-tex as being critically involved in mentalisingrdquo (Fletcheret al 1995 p 121)

Results from lesion patients thus far have not providedany conclusive evidence about which areas might becritical for theory of mind computations Patients withdamage to orbito-frontal cortex and with ventromedialdamage that is damage that includes both orbital andmedial frontal cortex typically have severe decits insocial functioning (Blumer amp Benson 1975 DamasioTranel amp Damasio 1990 Eslinger amp Damasio 1985Kaczmarek 1984 Mattson amp Levin 1990 Saver ampDamasio 1991) These patients are able to correctly ana-lyze social situations in the abstract but when theyrespond to similar situations in real life they chooseinappropriate courses of action (Eslinger amp Damasio1985 Saver amp Damasio 1991) These patients can oftensay what the correct response is but have difcultychanging their behavior to respond appropriately to thesocial situation or to changing reinforcements in theenvironment (Rolls 1996) Orbito-frontal patients oftensay inappropriate things and appear disinhibited(Mattson amp Levin 1990) Their conversation typicallydoes not respond to signals of whether the other personis interested in what they are saying or whether they areon topic (Kaczmarek 1984) Based on some similaritiesbetween OFC patients and patients with autismmdashim-paired social judgment increased indifference anddecits in the pragmatics of conversationmdashBaron-Cohenand Ring (1994) have suggested that OFC is part of aneural circuit for mindreading and that the social impair-ment following OFC damage occurs because part of thetheory of mind module is damaged However as far aswe are aware no direct test of theory of mind such asa test of false belief attribution has been reported withOFC patients

Only patients with damage in dorsolateral frontal cor-tex (DFC) have been directly tested on any kind oftheory of mind task Price Daffner Stowe and Mesulam(1990) report two adult patients with bilateral DFC dam-age early in life who had difculties with empathy andfailed a perspective-taking task The task was one devel-oped by Flavell Botkin Fry Wright and Jarvis (1968) andalthough not explicitly designed as a theory of mind testdoes require a theory of mind The subject is given a mapof a town and told that someone at a certain location onthe map needs to get to a particular house on the mapand is lost The experimenter then reads a set of direc-tions for getting from where the lost person is to the

642 Journal of Cognitive Neuroscience Volume 10 Number 5

house The directions contain four different ambiguitiessuch that a person could make mistakes and end up atthe wrong house After reading the directions the experi-menter asks the subject to identify which parts of thedirections were ambiguous and could have led the lostperson to make a mistake This task requires perspective-taking and understanding false belief However the taskalso places considerable demands on working memorybecause the subject has to keep all of the directions inmemory to answer the question Because DFC patientstypically have difculty with working memory (StussEskes amp Foster 1994) these patients could have failedon this task because of working memory limitationsrather than because their theory of mind was impaired

We undertook to test a series of developmentallygraded theory of mind tasks in frontal lobe patients todetermine if any subtle theory of mind decits could bepicked up in patients with lesions in the frontal lobe Wetested patients with damage in orbito-frontal cortex be-cause they clearly have decits in social behavior andbecause Baron-Cohen et al (1994) found OFC activationwith a theory of mind task We also tested patients withdamage to dorsolateral frontal cortex to compare theirtheory of mind performance to that of the patientstested by Price et al (1990) We used tasks in which wecould control for the working memory demands of thetask

SUBJECTS

We tested ve patients with damage to the left lateralfrontal cortex including both dorsal regions of the lat-eral frontal cortex and more ventrolateral regions Wewill refer to these patients as having DFC damage Thisdoes not imply that their damage is restricted to dorso-lateral regions of the lateral frontal cortex only that allve patients have DFC damage Figure 1 shows comput-erized axial tomograph (CT) reconstructions for individ-ual patients and Figure 2 shows the degree of overlapof the patientsrsquo lesions in different areas (see Table 1 forpatient characteristics) Of the ve DFC patients fourhad damage to the lateral portion of Brodmannrsquos area 8and three of these also had damage to lateral area 9These patients had middle cerebral artery infarcts sotheir damage included only the lateral part of areas 8 and9 sparing the medial portion All ve patients had dam-age in the middle frontal gyrus and middle frontal sulcusBrodmannrsquos area 46 Our map of area 46 is based onrecent quantitative analysis of cytoarchitectonic featuresof cells in different areas of the frontal cortex(Rajkowska amp Goldman-Rakic 1995a 1995b) Area 46includes ldquocentral portions of one or more convolutionsof the middle frontal gyrus and extending to the depthof the middle frontal sulcusrdquo (Rajkowska amp Goldman-Rakic 1995b p 328) Slices 4 and 5 in our lesion overlapgure Figure 2 show that all ve DFC patients have

damage in the middle frontal gyrus and the depth of themiddle frontal sulcus These patients all had unilaterallesions from middle cerebral artery infarcts Bilaterallesions in DFC due to stroke are rare

Three of the DFC patients were aphasic JC WE andRT All three had difculty producing speech A motorspeech evaluation characterized these patients as havingapraxia and disarthria WErsquos score on the Western Apha-sia Battery was 963 JCrsquos was 919 WE scored 55 onthe Boston Naming Test JC scored 45 Both WE and JCare characterized as having anomic aphasia For compari-son OA who is not aphasic scored 996 on the WABand 58 on the Boston Naming Test

We also tested ve patients with bilateral damage toOFC from head trauma See Table 1 and Figures 3 and 4for summaries of the areas damaged in each patient andCT and magnetic resonance image (MRI) scans showingthe extent of the damage DH RV RM and RB allhad extensive bilateral damage in area 11 and no damageto either the lateral frontal cortex or the basal forebrainarea The damage in RV and RB also included the polarparts of area 10 bilaterally with minimal damage to area38 on the left side the very tip of the left temporal lobeIn addition RB had more extensive damage to the lefttemporal lobe including areas 38 28 21 and 20 andsparing the left amygdala RMrsquos damage also includedthe polar part of area 10 on the left and about 1 cm ofthe right anterior temporal lobe area 38 with the rightamygdala spared RM also had extensive damage to theleft temporal lobe extending back approximately 5 cmincluding areas 38 27 28 21 and 20 and the leftamygdala MR was the most unilateral of the OFC lesionpatients in our sample with extensive damage to area 11on the right and partial damage to area 11 on the leftMR also had some lateral frontal damage on the rightboth dorsolateral and ventrolateral frontal areas 47 45and 9 some damage to the polar portion of area 10 andsome medial frontal damage on the right to the anteriorcingulate area 33 His basal forebrain area was intact onthe left and damaged on the right Strictly unilateraldamage to the OFC is rare because it is not a region inwhich strokes occur

In addition we tested one patient with damage re-stricted to the anterior temporal cortex patient BG Hewas included in the sample to control for the bilateraltemporal damage of RV RB and RM All patients wereat least 6 months post-lesion

We also tested ve non-brain-damaged age-matchedcontrol subjects matched for education with the least-educated patients that is having only a high schooleducation

Tasks

The tasks we used were developmentally graded rangingin difculty from tasks that normal 4-year-old children

Stone et al 643

can do to tasks that children cannot do until ages 9 to11 We used three tasks altogether

1 First-order false belief tasks which develop around3 to 4 years

2 Second-order false belief tasks which developaround 6 to 7 years

3 Comprehension of social faux pas which developsaround 9 to 11 years (see ldquoMethodsrdquo)

The logic for this methodology was that decits in the-ory of mind should be more evident in the tasks thatdevelop later therefore the severity of theory of mindimpairment could be estimated by looking at whichtasks patients had difculty with Other work on decitsin dorsolateral frontal patients has used this kind ofmethodology to pick up subtle decits For instanceGoldstein Bernard Fenwick Burgess and McNeil (1993)tested a patient with a left frontal lobectomy who per-formed in the normal range on the Wisconsin Card

Sorting Task (WCST) Using tasks that were similar to theWCST but required more complex planning these re-searchers were able to document subtle executive func-tion decits

RESULTS

False Belief Tasks

Based on the rst- and second-order false belief tasksneither patient group shows a pronounced theory ofmind decit such as that seen in autism Some patientsmade errors on false belief tasks when they had toremember the stories However when the story was infront of the subjects so that there was no memory loadpatients made almost no errors (see Tables 2 and 3 andFigures 5 and 6)1 Even in the condition with a memoryload it was rare for patients to make errors on the falsebelief question alone without also making errors on thecontrol questions The difference in the proportion of

Figure 1 Extent of patientsrsquo lesions in the dorsolateral frontal cortex reconstructed from CT scans

644 Journal of Cognitive Neuroscience Volume 10 Number 5

problems correct between these two conditions on therst-order false belief tasks was statistically signicant forfour out of ve of the DFC patients LS (z = 292 p lt001) RT (z = 459 p lt 000001) JC (z = 266 p lt 001and WE (z = 211 p lt 005) The difference betweenConditions 1 and 2 was signicant for LS (z = 234 p lt001) and RT (z = 188 p lt 005) on the second-orderfalse belief tasks It was also signicant for RV an OFC

patient (z = 346 p lt 0001) on the rst-order falsebelief task After completing 10 rst-order false beliefproblems in Condition 1 RV requested not to be testedin that condition any more because he found it sodifcult

Differences in performance between the false belieftask and the true belief task were not signicant for anyof the frontal patients except JC (410 vs 1010 z =

Figure 2 Extent of overlapof DFC patientsrsquo lesions recon-structed from CT scans Thearea of 80 to 100 overlap isin area 46 the middle frontalgyrus and the depth of themiddle frontal sulcus

Table 1 Patient Characteristics

Patient AgeTime Since

Lesion Onset Location of Lesion Brodmannrsquos Areas Damaged

JC 72 9 years Left lateral frontal including DFCsuperior temporal

Lateral areas 8 9 areas 44 45 46 areas 4 6

OA 64 12 years Left lateral frontal including DFC Lateral areas 8 9 area 46

WE 67 6 months Left lateral frontal including DFC Lateral area 8 areas 44 45 46

LS 68 22 years Left lateral frontal including DFC Lateral areas 8 9 area 46

RT 80 11 years Left lateral frontal including DFC Areas 45 46

DH 34 16 years Bilateral OFC Extensive bilateral area 11

MR 42 18 years Bilateral OFC Extensive right area 11 partial left area 11right areas 47 45 9 10 and 33 rightbasal forebrain area

RB 51 22 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial left area 38

RM 46 20 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial right area 38 left areas38 28 21 and 20 amygdala spared

RV 45 5 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 left polar area10 partial right area 38 left area 37 2821 and 20 left amygdala

BG 53 4 years Anterior temporal Bilateral area 38

Stone et al 645

293 p lt 0002) this was true only in the condition witha memory load

Because three of the dorsolateral frontal patients wereaphasic and had difculty producing speech we did notask subjects to justify their responses on the false belieftasks However LS spontaneously offered mentalisticjustications on several problems Even on two problemsshe got wrong she provided mental state explanationsthat made sense of her ldquoerrorsrdquo for example ldquoHersquoll prob-ably think itrsquos on the shelf because he wonrsquot see it onthe desk where he left it so that would be the logicalplace to lookrdquo

Order Effects Practice Effects

There were no effects of prior exposure to a particularstory and no practice effects Subjectsrsquo performance didnot improve over the course of testing Subjects were notmore likely to give correct answers for false belief tasksthey were given in the second session than tasks theywere given in the rst session regardless of condition

Because subjects did not perform signicantly betterin the second testing session than in the rst testingsession familiarity with the stories being used did notseem to help them answer correctly Only memory loadpredicts their performance indicating that workingmemory limitations may provide the best explanation oftheir results

In the condition without a memory load there was noevidence for a theory of mind decit at the level of 4-to 7-year-old children in either the orbito-frontal or thedorsolateral frontal patient group

Faux Pas Task

Control subjects and DFC patients correctly detected allof the faux pas and correctly answered who committedthe faux pas In contrast almost all of the OFC patientsmade errors detecting faux pas answering that nothingawkward had been said in the story They sometimesanswered No to the question ldquoDid someone say some-thing they shouldnrsquot have saidrdquo The difference between

Figure 3 Three patientswith bilateral damage toorbito-frontal cortex CT scansare shown for patient DHand MRI scans are shown forpatients MR and RV The im-ages in columns 1 and 2show the damage to orbito-frontal cortex column 3shows that dorsolateral frontalcortex is spared in patientsDH and RV with some rightdorsolateral damage in patientMR Medial frontal cortex isspared in all patients

646 Journal of Cognitive Neuroscience Volume 10 Number 5

the OFC patients and the control subjects (who were atceiling) in the detection of faux pas is signicant (t8 =2828 p lt 002) All of the OFC patients answered thecontrol questions correctly Thus it appears that theyunderstood the stories yet didnrsquot realize that somethinginappropriate had been said (see Table 4)

In contrast the DFC patients got confused about thedetails of the story and as a consequence all four an-swered some of the control questions incorrectly Forexample WE mixed up who the surprise party was forand therefore made errors on both the control questionand the question of who committed the faux pas be-cause he had mixed up the two charactersrsquo names RTmade a similar mistake on this problem Although heanswered the rst two questions correctly he got mixedup on the later questions and made mistakes OA statedon two of the control questions that he didnrsquot rememberwhat the answer was All of these errors were made even

though the story was right in front of the patients onthe desk and the experimenter told them that they couldlook back at the stories They did not always do so evenwhen they got confused

All of the stories contained a faux pas that is someonesaying something awkward that should not have beensaid There are two possible explanations for why sub-jects might answer yes to the question of whether some-thing awkward had been said on all 10 stories Oneexplanation is that they correctly recognized all of thefaux pas Another possible explanation is that theymerely had a Yes bias We can control for the fact thatthe DFC patients always answered Yes on this questionby looking at their responses to the question of who saidsomething they should not have said If they did notunderstand that a faux pas had been committed andwere merely saying Yes because of a Yes bias they wouldhave not answered the questions about who committed

Figure 4 The top two pa-tients RM and RB have bilat-eral damage to theorbito-frontal cortex and addi-tional bilateral damage to theanterior temporal cortex withmore extensive damage onthe left RMrsquos anterior tempo-ral damage is more extensivethan RBrsquos and includes oneamygdala The patient in thelast row BG has restricted an-terior temporal damage Col-umn 1 shows damage to thetemporal lobes columns 2and 3 show damage (or inthe case of BG lack of dam-age) to orbito-frontal cortexAll images are T2 weightedaxial MRI slices

Stone et al 647

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

months the child begins to understand ldquoMommy seesthe toy that I seerdquo In protodeclarative pointing the childuses pointing to call adultsrsquo attention to objects that thechild wants them to attend to Many children withautism do not show either joint attention or protodecla-rative pointing (Baron-Cohen 1989a 1995) They do notlook at what other people look at nor do they usepointing to draw adultsrsquo attention to things The nextstage in the development of theory of mind is pretendplay in which children are able to decouple pretendfrom reality Between 18 and 24 months children beginto understand the mental state of ldquopretendrdquo (Leslie1987) Also by age 2 children seem to have a rm graspof the mental state of desire for example ldquoJohn wants ahamburgerrdquo (Wellman amp Woolley 1990) Childrenrsquos un-derstanding of desire precedes their understanding ofbelief

Between ages 3 and 4 children develop the ability tounderstand false belief (Gopnik amp Astington 1988Johnson amp Wellman 1980 Wellman 1990 Wimmer ampPerner 1983) Prior to this age a child does not under-stand that other people can hold beliefs about the worldthat differ from the childrsquos own Thus children assumethat other people know the same things they knowBetween ages 3 and 4 however children begin to under-stand that other people may not know all the things thatthey know and therefore that others may hold falsebeliefs Tests of false belief measure the ability of chil-dren to understand that another person can hold a beliefthat is mistaken These tests demonstrate that childrenare representing othersrsquo mental states othersrsquo beliefsrather then the physical state of the world or their ownstate of knowledge (Dennett 1978)

Between ages 6 and 7 children begin to understandthat other people can also represent mental states At thisage children begin to be able to understand second-order false belief ldquobelief about beliefrdquo (Perner amp Wimmer1985) In a typical second-order false belief task theproblem might run something like this A man and awoman are in a room The woman puts something some-where such as putting a book on a shelf She then leavesthe room The man hides the book in another locationUnbeknownst to him the woman is peeking backthrough a keyhole or a window and sees him movingthe book The subject is asked ldquoWhen the woman comesback in where will the man think that she thinks thebook isrdquo To solve this problem the child needs to beable to represent not only each personrsquos belief stateabout the location of the object but also the manrsquosmistaken belief about the womanrsquos belief state

Theory of mind can break down at certain of thesedevelopmental stages Children with autism who do notshow joint attention may never develop these theory ofmind abilities Children with autism are impaired in pre-tend play (Baron-Cohen 1995) Most children withautism cannot solve false belief tasks or second-orderfalse belief tasks (Baron-Cohen et al 1985 Perner

Leekam amp Wimmer 1987) Some higher-functioning in-dividuals with autism can eventually pass rst-order falsebelief tasks but will fail second-order false belief tasks(Baron-Cohen 1989b Happeacute 1993) In general autisticsrsquodifculties are with epistemic mental states concerningknowledge or belief They do seem to understand themental state of desire (Baron-Cohen Leslie amp Frith 1986Tager-Flusberg 1989 1993)

It should be noted that children with autism andyoung children do not simply lack the ability to dometa-representation They can pass what is called a falsephotograph test (Zaitchik 1990) In this test a Polaroidpicture is taken of a toy placed on a table The toy isthen moved and then the photo comes out and is devel-oped Before the child sees what is in the photographthe experimenter asks ldquoWhat will the picture showrdquoYoung children and children with autism are not fooledinto thinking that the photograph will show a table withnothing on it even though the toy has been moved Thusthey can understand physical representations such asphotographs and pictures but not mental repre-sentations (Charman amp Baron-Cohen 1992 1995Leekam amp Perner 1991 Leslie amp Thaiss 1992)

Later between ages 9 and 11 children develop furthertheory of mind abilities such as the ability to understandand recognize faux pas A faux pas occurs when some-one says something they should have not have said notknowing or not realizing that they should not say it Tounderstand that a faux pas has occurred one has torepresent two mental states that the person saying itdoes not know that they should not say it and that theperson hearing it would feel insulted or hurt Thus thereis both a cognitive component and empathic affectivecomponent On our new test of faux pas detectionBaron-Cohen OrsquoRiordan Stone Jones and Plaisted(1997) found that girls could perform well on this testby age 9 boys by age 11 Boys and girls of ages 7 or 8although they could pass rst- and second-order falsebelief tasks did not perform well on the faux pas task

To validate that the faux pas task is indeed a theoryof mind test Baron-Cohen et al (1997) also tested it onindividuals with Aspergerrsquos syndrome a mild form ofautism In these individuals language develops at a nor-mal time pace and their IQ is often normal Howeverthey still have many subtle social decits On our newtest of faux pas detection Baron-Cohen et al (1997)found that like 7 to 8 year-olds individuals with Asper-gerrsquos syndrome could pass rst- and second-order falsebelief tasks but were impaired on the faux pas task Theirtheory of mind performance was comparable to that of7- to 9-year-old children The faux pas task is a thus agood measure of subtle theory of mind decits Perfor-mance on the most developmentally advanced theory ofmind tasks is an index of how severe a personrsquos theoryof mind decit is Subtle theory of mind decits can onlybe picked up with the most developmentally advancedtasks

Stone et al 641

Little is known about the neurological basis for theoryof mind Such a complex cognitive ability does not seema likely candidate for localizationmdasha neural network orcircuit is more plausible Recent neuroimaging studieshave reported that regions of the frontal lobes appear tobe active during theory of mind tasks suggesting thatthese may be part of a theory of mind circuit Baron-Cohen Ring Moriarty Schmitz Costa and Ell (1994)found orbito-frontal activation during a simple theory ofmind task requiring recognition of mental state termsFletcher et al (1995) found activation in Brodmannrsquosareas 8 and 9 in the left medial frontal cortex during amore complex theory of mind task involving deceptionand belief attribution Goel Grafman Sadato and Hallett(1995) also found activation in the left medial frontalcortex during a task requiring mental state inferences

Baron-Cohen et al (1994) used a task requiring sub-jects to judge whether each word on a list of words hadto do with the mind or was something the mind coulddo and compared it to a task requiring a judgment ofwhether each word on another list had to do with thebody or was something the body could do Children withautism performed poorly on this mental state terms taskbut not on the body terms indicating that this task wasmeasuring theory of mind Using single photon emissioncomputerized tomograph (SPECT) imaging on a sampleof developmentally normal control subjects males aged20 to 30 Baron-Cohen et al found that right orbito-fron-tal cortex (OFC) was signicantly more active during themental state term recognition task than during the con-trol task relative to frontal polar cortex and posteriorregions One limitation of their study however is thatthey only carried out hypothesis-led region-of-interest(ROI) analyses and did not measure activation in themedial or dorsolateral frontal cortex so it is unknownhow active those areas were during their task The taskalso has no inference component

Fletcher et al (1995) and Goel et al (1995) used morecomplex tasks requiring subtle inferences about mentalstates In Goel et alrsquos study subjects were asked to makeinferences about objects that required either a visualdescription of the object memory retrieval an inferencefrom the objectrsquos form to its function or an inferencethat required modeling another personrsquos mental stateThey found selective activation for the task requiringmentalistic inferences in left medial frontal cortex andthe left temporal lobe In Fletcher et alrsquos study subjectsread a story and then answered a question about thestory that required a mentalistic inference They an-swered the question silently to themselves without mak-ing any overt response There were two control tasks forcomparison stories involving subtle physical inferencesbut not mentalistic inferences and paragraphs consistingof unrelated sentences Subjects were told which typeof story they were about to read and then were scannedboth while reading and while answering the question

Using positron emission tomography (PET) imagingwhen pixels active during the physical inference storieswere subtracted from pixels active during the theory ofmind stories Brodmannrsquos areas 8 and 9 and the anteriorcingulate showed up as active only during the theory ofmind task OFC was not specically active during thetheory of mind task They conclude that their data ldquopin-pointed the medial dorsal region of the left frontal cor-tex as being critically involved in mentalisingrdquo (Fletcheret al 1995 p 121)

Results from lesion patients thus far have not providedany conclusive evidence about which areas might becritical for theory of mind computations Patients withdamage to orbito-frontal cortex and with ventromedialdamage that is damage that includes both orbital andmedial frontal cortex typically have severe decits insocial functioning (Blumer amp Benson 1975 DamasioTranel amp Damasio 1990 Eslinger amp Damasio 1985Kaczmarek 1984 Mattson amp Levin 1990 Saver ampDamasio 1991) These patients are able to correctly ana-lyze social situations in the abstract but when theyrespond to similar situations in real life they chooseinappropriate courses of action (Eslinger amp Damasio1985 Saver amp Damasio 1991) These patients can oftensay what the correct response is but have difcultychanging their behavior to respond appropriately to thesocial situation or to changing reinforcements in theenvironment (Rolls 1996) Orbito-frontal patients oftensay inappropriate things and appear disinhibited(Mattson amp Levin 1990) Their conversation typicallydoes not respond to signals of whether the other personis interested in what they are saying or whether they areon topic (Kaczmarek 1984) Based on some similaritiesbetween OFC patients and patients with autismmdashim-paired social judgment increased indifference anddecits in the pragmatics of conversationmdashBaron-Cohenand Ring (1994) have suggested that OFC is part of aneural circuit for mindreading and that the social impair-ment following OFC damage occurs because part of thetheory of mind module is damaged However as far aswe are aware no direct test of theory of mind such asa test of false belief attribution has been reported withOFC patients

Only patients with damage in dorsolateral frontal cor-tex (DFC) have been directly tested on any kind oftheory of mind task Price Daffner Stowe and Mesulam(1990) report two adult patients with bilateral DFC dam-age early in life who had difculties with empathy andfailed a perspective-taking task The task was one devel-oped by Flavell Botkin Fry Wright and Jarvis (1968) andalthough not explicitly designed as a theory of mind testdoes require a theory of mind The subject is given a mapof a town and told that someone at a certain location onthe map needs to get to a particular house on the mapand is lost The experimenter then reads a set of direc-tions for getting from where the lost person is to the

642 Journal of Cognitive Neuroscience Volume 10 Number 5

house The directions contain four different ambiguitiessuch that a person could make mistakes and end up atthe wrong house After reading the directions the experi-menter asks the subject to identify which parts of thedirections were ambiguous and could have led the lostperson to make a mistake This task requires perspective-taking and understanding false belief However the taskalso places considerable demands on working memorybecause the subject has to keep all of the directions inmemory to answer the question Because DFC patientstypically have difculty with working memory (StussEskes amp Foster 1994) these patients could have failedon this task because of working memory limitationsrather than because their theory of mind was impaired

We undertook to test a series of developmentallygraded theory of mind tasks in frontal lobe patients todetermine if any subtle theory of mind decits could bepicked up in patients with lesions in the frontal lobe Wetested patients with damage in orbito-frontal cortex be-cause they clearly have decits in social behavior andbecause Baron-Cohen et al (1994) found OFC activationwith a theory of mind task We also tested patients withdamage to dorsolateral frontal cortex to compare theirtheory of mind performance to that of the patientstested by Price et al (1990) We used tasks in which wecould control for the working memory demands of thetask

SUBJECTS

We tested ve patients with damage to the left lateralfrontal cortex including both dorsal regions of the lat-eral frontal cortex and more ventrolateral regions Wewill refer to these patients as having DFC damage Thisdoes not imply that their damage is restricted to dorso-lateral regions of the lateral frontal cortex only that allve patients have DFC damage Figure 1 shows comput-erized axial tomograph (CT) reconstructions for individ-ual patients and Figure 2 shows the degree of overlapof the patientsrsquo lesions in different areas (see Table 1 forpatient characteristics) Of the ve DFC patients fourhad damage to the lateral portion of Brodmannrsquos area 8and three of these also had damage to lateral area 9These patients had middle cerebral artery infarcts sotheir damage included only the lateral part of areas 8 and9 sparing the medial portion All ve patients had dam-age in the middle frontal gyrus and middle frontal sulcusBrodmannrsquos area 46 Our map of area 46 is based onrecent quantitative analysis of cytoarchitectonic featuresof cells in different areas of the frontal cortex(Rajkowska amp Goldman-Rakic 1995a 1995b) Area 46includes ldquocentral portions of one or more convolutionsof the middle frontal gyrus and extending to the depthof the middle frontal sulcusrdquo (Rajkowska amp Goldman-Rakic 1995b p 328) Slices 4 and 5 in our lesion overlapgure Figure 2 show that all ve DFC patients have

damage in the middle frontal gyrus and the depth of themiddle frontal sulcus These patients all had unilaterallesions from middle cerebral artery infarcts Bilaterallesions in DFC due to stroke are rare

Three of the DFC patients were aphasic JC WE andRT All three had difculty producing speech A motorspeech evaluation characterized these patients as havingapraxia and disarthria WErsquos score on the Western Apha-sia Battery was 963 JCrsquos was 919 WE scored 55 onthe Boston Naming Test JC scored 45 Both WE and JCare characterized as having anomic aphasia For compari-son OA who is not aphasic scored 996 on the WABand 58 on the Boston Naming Test

We also tested ve patients with bilateral damage toOFC from head trauma See Table 1 and Figures 3 and 4for summaries of the areas damaged in each patient andCT and magnetic resonance image (MRI) scans showingthe extent of the damage DH RV RM and RB allhad extensive bilateral damage in area 11 and no damageto either the lateral frontal cortex or the basal forebrainarea The damage in RV and RB also included the polarparts of area 10 bilaterally with minimal damage to area38 on the left side the very tip of the left temporal lobeIn addition RB had more extensive damage to the lefttemporal lobe including areas 38 28 21 and 20 andsparing the left amygdala RMrsquos damage also includedthe polar part of area 10 on the left and about 1 cm ofthe right anterior temporal lobe area 38 with the rightamygdala spared RM also had extensive damage to theleft temporal lobe extending back approximately 5 cmincluding areas 38 27 28 21 and 20 and the leftamygdala MR was the most unilateral of the OFC lesionpatients in our sample with extensive damage to area 11on the right and partial damage to area 11 on the leftMR also had some lateral frontal damage on the rightboth dorsolateral and ventrolateral frontal areas 47 45and 9 some damage to the polar portion of area 10 andsome medial frontal damage on the right to the anteriorcingulate area 33 His basal forebrain area was intact onthe left and damaged on the right Strictly unilateraldamage to the OFC is rare because it is not a region inwhich strokes occur

In addition we tested one patient with damage re-stricted to the anterior temporal cortex patient BG Hewas included in the sample to control for the bilateraltemporal damage of RV RB and RM All patients wereat least 6 months post-lesion

We also tested ve non-brain-damaged age-matchedcontrol subjects matched for education with the least-educated patients that is having only a high schooleducation

Tasks

The tasks we used were developmentally graded rangingin difculty from tasks that normal 4-year-old children

Stone et al 643

can do to tasks that children cannot do until ages 9 to11 We used three tasks altogether

1 First-order false belief tasks which develop around3 to 4 years

2 Second-order false belief tasks which developaround 6 to 7 years

3 Comprehension of social faux pas which developsaround 9 to 11 years (see ldquoMethodsrdquo)

The logic for this methodology was that decits in the-ory of mind should be more evident in the tasks thatdevelop later therefore the severity of theory of mindimpairment could be estimated by looking at whichtasks patients had difculty with Other work on decitsin dorsolateral frontal patients has used this kind ofmethodology to pick up subtle decits For instanceGoldstein Bernard Fenwick Burgess and McNeil (1993)tested a patient with a left frontal lobectomy who per-formed in the normal range on the Wisconsin Card

Sorting Task (WCST) Using tasks that were similar to theWCST but required more complex planning these re-searchers were able to document subtle executive func-tion decits

RESULTS

False Belief Tasks

Based on the rst- and second-order false belief tasksneither patient group shows a pronounced theory ofmind decit such as that seen in autism Some patientsmade errors on false belief tasks when they had toremember the stories However when the story was infront of the subjects so that there was no memory loadpatients made almost no errors (see Tables 2 and 3 andFigures 5 and 6)1 Even in the condition with a memoryload it was rare for patients to make errors on the falsebelief question alone without also making errors on thecontrol questions The difference in the proportion of

Figure 1 Extent of patientsrsquo lesions in the dorsolateral frontal cortex reconstructed from CT scans

644 Journal of Cognitive Neuroscience Volume 10 Number 5

problems correct between these two conditions on therst-order false belief tasks was statistically signicant forfour out of ve of the DFC patients LS (z = 292 p lt001) RT (z = 459 p lt 000001) JC (z = 266 p lt 001and WE (z = 211 p lt 005) The difference betweenConditions 1 and 2 was signicant for LS (z = 234 p lt001) and RT (z = 188 p lt 005) on the second-orderfalse belief tasks It was also signicant for RV an OFC

patient (z = 346 p lt 0001) on the rst-order falsebelief task After completing 10 rst-order false beliefproblems in Condition 1 RV requested not to be testedin that condition any more because he found it sodifcult

Differences in performance between the false belieftask and the true belief task were not signicant for anyof the frontal patients except JC (410 vs 1010 z =

Figure 2 Extent of overlapof DFC patientsrsquo lesions recon-structed from CT scans Thearea of 80 to 100 overlap isin area 46 the middle frontalgyrus and the depth of themiddle frontal sulcus

Table 1 Patient Characteristics

Patient AgeTime Since

Lesion Onset Location of Lesion Brodmannrsquos Areas Damaged

JC 72 9 years Left lateral frontal including DFCsuperior temporal

Lateral areas 8 9 areas 44 45 46 areas 4 6

OA 64 12 years Left lateral frontal including DFC Lateral areas 8 9 area 46

WE 67 6 months Left lateral frontal including DFC Lateral area 8 areas 44 45 46

LS 68 22 years Left lateral frontal including DFC Lateral areas 8 9 area 46

RT 80 11 years Left lateral frontal including DFC Areas 45 46

DH 34 16 years Bilateral OFC Extensive bilateral area 11

MR 42 18 years Bilateral OFC Extensive right area 11 partial left area 11right areas 47 45 9 10 and 33 rightbasal forebrain area

RB 51 22 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial left area 38

RM 46 20 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial right area 38 left areas38 28 21 and 20 amygdala spared

RV 45 5 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 left polar area10 partial right area 38 left area 37 2821 and 20 left amygdala

BG 53 4 years Anterior temporal Bilateral area 38

Stone et al 645

293 p lt 0002) this was true only in the condition witha memory load

Because three of the dorsolateral frontal patients wereaphasic and had difculty producing speech we did notask subjects to justify their responses on the false belieftasks However LS spontaneously offered mentalisticjustications on several problems Even on two problemsshe got wrong she provided mental state explanationsthat made sense of her ldquoerrorsrdquo for example ldquoHersquoll prob-ably think itrsquos on the shelf because he wonrsquot see it onthe desk where he left it so that would be the logicalplace to lookrdquo

Order Effects Practice Effects

There were no effects of prior exposure to a particularstory and no practice effects Subjectsrsquo performance didnot improve over the course of testing Subjects were notmore likely to give correct answers for false belief tasksthey were given in the second session than tasks theywere given in the rst session regardless of condition

Because subjects did not perform signicantly betterin the second testing session than in the rst testingsession familiarity with the stories being used did notseem to help them answer correctly Only memory loadpredicts their performance indicating that workingmemory limitations may provide the best explanation oftheir results

In the condition without a memory load there was noevidence for a theory of mind decit at the level of 4-to 7-year-old children in either the orbito-frontal or thedorsolateral frontal patient group

Faux Pas Task

Control subjects and DFC patients correctly detected allof the faux pas and correctly answered who committedthe faux pas In contrast almost all of the OFC patientsmade errors detecting faux pas answering that nothingawkward had been said in the story They sometimesanswered No to the question ldquoDid someone say some-thing they shouldnrsquot have saidrdquo The difference between

Figure 3 Three patientswith bilateral damage toorbito-frontal cortex CT scansare shown for patient DHand MRI scans are shown forpatients MR and RV The im-ages in columns 1 and 2show the damage to orbito-frontal cortex column 3shows that dorsolateral frontalcortex is spared in patientsDH and RV with some rightdorsolateral damage in patientMR Medial frontal cortex isspared in all patients

646 Journal of Cognitive Neuroscience Volume 10 Number 5

the OFC patients and the control subjects (who were atceiling) in the detection of faux pas is signicant (t8 =2828 p lt 002) All of the OFC patients answered thecontrol questions correctly Thus it appears that theyunderstood the stories yet didnrsquot realize that somethinginappropriate had been said (see Table 4)

In contrast the DFC patients got confused about thedetails of the story and as a consequence all four an-swered some of the control questions incorrectly Forexample WE mixed up who the surprise party was forand therefore made errors on both the control questionand the question of who committed the faux pas be-cause he had mixed up the two charactersrsquo names RTmade a similar mistake on this problem Although heanswered the rst two questions correctly he got mixedup on the later questions and made mistakes OA statedon two of the control questions that he didnrsquot rememberwhat the answer was All of these errors were made even

though the story was right in front of the patients onthe desk and the experimenter told them that they couldlook back at the stories They did not always do so evenwhen they got confused

All of the stories contained a faux pas that is someonesaying something awkward that should not have beensaid There are two possible explanations for why sub-jects might answer yes to the question of whether some-thing awkward had been said on all 10 stories Oneexplanation is that they correctly recognized all of thefaux pas Another possible explanation is that theymerely had a Yes bias We can control for the fact thatthe DFC patients always answered Yes on this questionby looking at their responses to the question of who saidsomething they should not have said If they did notunderstand that a faux pas had been committed andwere merely saying Yes because of a Yes bias they wouldhave not answered the questions about who committed

Figure 4 The top two pa-tients RM and RB have bilat-eral damage to theorbito-frontal cortex and addi-tional bilateral damage to theanterior temporal cortex withmore extensive damage onthe left RMrsquos anterior tempo-ral damage is more extensivethan RBrsquos and includes oneamygdala The patient in thelast row BG has restricted an-terior temporal damage Col-umn 1 shows damage to thetemporal lobes columns 2and 3 show damage (or inthe case of BG lack of dam-age) to orbito-frontal cortexAll images are T2 weightedaxial MRI slices

Stone et al 647

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

Little is known about the neurological basis for theoryof mind Such a complex cognitive ability does not seema likely candidate for localizationmdasha neural network orcircuit is more plausible Recent neuroimaging studieshave reported that regions of the frontal lobes appear tobe active during theory of mind tasks suggesting thatthese may be part of a theory of mind circuit Baron-Cohen Ring Moriarty Schmitz Costa and Ell (1994)found orbito-frontal activation during a simple theory ofmind task requiring recognition of mental state termsFletcher et al (1995) found activation in Brodmannrsquosareas 8 and 9 in the left medial frontal cortex during amore complex theory of mind task involving deceptionand belief attribution Goel Grafman Sadato and Hallett(1995) also found activation in the left medial frontalcortex during a task requiring mental state inferences

Baron-Cohen et al (1994) used a task requiring sub-jects to judge whether each word on a list of words hadto do with the mind or was something the mind coulddo and compared it to a task requiring a judgment ofwhether each word on another list had to do with thebody or was something the body could do Children withautism performed poorly on this mental state terms taskbut not on the body terms indicating that this task wasmeasuring theory of mind Using single photon emissioncomputerized tomograph (SPECT) imaging on a sampleof developmentally normal control subjects males aged20 to 30 Baron-Cohen et al found that right orbito-fron-tal cortex (OFC) was signicantly more active during themental state term recognition task than during the con-trol task relative to frontal polar cortex and posteriorregions One limitation of their study however is thatthey only carried out hypothesis-led region-of-interest(ROI) analyses and did not measure activation in themedial or dorsolateral frontal cortex so it is unknownhow active those areas were during their task The taskalso has no inference component

Fletcher et al (1995) and Goel et al (1995) used morecomplex tasks requiring subtle inferences about mentalstates In Goel et alrsquos study subjects were asked to makeinferences about objects that required either a visualdescription of the object memory retrieval an inferencefrom the objectrsquos form to its function or an inferencethat required modeling another personrsquos mental stateThey found selective activation for the task requiringmentalistic inferences in left medial frontal cortex andthe left temporal lobe In Fletcher et alrsquos study subjectsread a story and then answered a question about thestory that required a mentalistic inference They an-swered the question silently to themselves without mak-ing any overt response There were two control tasks forcomparison stories involving subtle physical inferencesbut not mentalistic inferences and paragraphs consistingof unrelated sentences Subjects were told which typeof story they were about to read and then were scannedboth while reading and while answering the question

Using positron emission tomography (PET) imagingwhen pixels active during the physical inference storieswere subtracted from pixels active during the theory ofmind stories Brodmannrsquos areas 8 and 9 and the anteriorcingulate showed up as active only during the theory ofmind task OFC was not specically active during thetheory of mind task They conclude that their data ldquopin-pointed the medial dorsal region of the left frontal cor-tex as being critically involved in mentalisingrdquo (Fletcheret al 1995 p 121)

Results from lesion patients thus far have not providedany conclusive evidence about which areas might becritical for theory of mind computations Patients withdamage to orbito-frontal cortex and with ventromedialdamage that is damage that includes both orbital andmedial frontal cortex typically have severe decits insocial functioning (Blumer amp Benson 1975 DamasioTranel amp Damasio 1990 Eslinger amp Damasio 1985Kaczmarek 1984 Mattson amp Levin 1990 Saver ampDamasio 1991) These patients are able to correctly ana-lyze social situations in the abstract but when theyrespond to similar situations in real life they chooseinappropriate courses of action (Eslinger amp Damasio1985 Saver amp Damasio 1991) These patients can oftensay what the correct response is but have difcultychanging their behavior to respond appropriately to thesocial situation or to changing reinforcements in theenvironment (Rolls 1996) Orbito-frontal patients oftensay inappropriate things and appear disinhibited(Mattson amp Levin 1990) Their conversation typicallydoes not respond to signals of whether the other personis interested in what they are saying or whether they areon topic (Kaczmarek 1984) Based on some similaritiesbetween OFC patients and patients with autismmdashim-paired social judgment increased indifference anddecits in the pragmatics of conversationmdashBaron-Cohenand Ring (1994) have suggested that OFC is part of aneural circuit for mindreading and that the social impair-ment following OFC damage occurs because part of thetheory of mind module is damaged However as far aswe are aware no direct test of theory of mind such asa test of false belief attribution has been reported withOFC patients

Only patients with damage in dorsolateral frontal cor-tex (DFC) have been directly tested on any kind oftheory of mind task Price Daffner Stowe and Mesulam(1990) report two adult patients with bilateral DFC dam-age early in life who had difculties with empathy andfailed a perspective-taking task The task was one devel-oped by Flavell Botkin Fry Wright and Jarvis (1968) andalthough not explicitly designed as a theory of mind testdoes require a theory of mind The subject is given a mapof a town and told that someone at a certain location onthe map needs to get to a particular house on the mapand is lost The experimenter then reads a set of direc-tions for getting from where the lost person is to the

642 Journal of Cognitive Neuroscience Volume 10 Number 5

house The directions contain four different ambiguitiessuch that a person could make mistakes and end up atthe wrong house After reading the directions the experi-menter asks the subject to identify which parts of thedirections were ambiguous and could have led the lostperson to make a mistake This task requires perspective-taking and understanding false belief However the taskalso places considerable demands on working memorybecause the subject has to keep all of the directions inmemory to answer the question Because DFC patientstypically have difculty with working memory (StussEskes amp Foster 1994) these patients could have failedon this task because of working memory limitationsrather than because their theory of mind was impaired

We undertook to test a series of developmentallygraded theory of mind tasks in frontal lobe patients todetermine if any subtle theory of mind decits could bepicked up in patients with lesions in the frontal lobe Wetested patients with damage in orbito-frontal cortex be-cause they clearly have decits in social behavior andbecause Baron-Cohen et al (1994) found OFC activationwith a theory of mind task We also tested patients withdamage to dorsolateral frontal cortex to compare theirtheory of mind performance to that of the patientstested by Price et al (1990) We used tasks in which wecould control for the working memory demands of thetask

SUBJECTS

We tested ve patients with damage to the left lateralfrontal cortex including both dorsal regions of the lat-eral frontal cortex and more ventrolateral regions Wewill refer to these patients as having DFC damage Thisdoes not imply that their damage is restricted to dorso-lateral regions of the lateral frontal cortex only that allve patients have DFC damage Figure 1 shows comput-erized axial tomograph (CT) reconstructions for individ-ual patients and Figure 2 shows the degree of overlapof the patientsrsquo lesions in different areas (see Table 1 forpatient characteristics) Of the ve DFC patients fourhad damage to the lateral portion of Brodmannrsquos area 8and three of these also had damage to lateral area 9These patients had middle cerebral artery infarcts sotheir damage included only the lateral part of areas 8 and9 sparing the medial portion All ve patients had dam-age in the middle frontal gyrus and middle frontal sulcusBrodmannrsquos area 46 Our map of area 46 is based onrecent quantitative analysis of cytoarchitectonic featuresof cells in different areas of the frontal cortex(Rajkowska amp Goldman-Rakic 1995a 1995b) Area 46includes ldquocentral portions of one or more convolutionsof the middle frontal gyrus and extending to the depthof the middle frontal sulcusrdquo (Rajkowska amp Goldman-Rakic 1995b p 328) Slices 4 and 5 in our lesion overlapgure Figure 2 show that all ve DFC patients have

damage in the middle frontal gyrus and the depth of themiddle frontal sulcus These patients all had unilaterallesions from middle cerebral artery infarcts Bilaterallesions in DFC due to stroke are rare

Three of the DFC patients were aphasic JC WE andRT All three had difculty producing speech A motorspeech evaluation characterized these patients as havingapraxia and disarthria WErsquos score on the Western Apha-sia Battery was 963 JCrsquos was 919 WE scored 55 onthe Boston Naming Test JC scored 45 Both WE and JCare characterized as having anomic aphasia For compari-son OA who is not aphasic scored 996 on the WABand 58 on the Boston Naming Test

We also tested ve patients with bilateral damage toOFC from head trauma See Table 1 and Figures 3 and 4for summaries of the areas damaged in each patient andCT and magnetic resonance image (MRI) scans showingthe extent of the damage DH RV RM and RB allhad extensive bilateral damage in area 11 and no damageto either the lateral frontal cortex or the basal forebrainarea The damage in RV and RB also included the polarparts of area 10 bilaterally with minimal damage to area38 on the left side the very tip of the left temporal lobeIn addition RB had more extensive damage to the lefttemporal lobe including areas 38 28 21 and 20 andsparing the left amygdala RMrsquos damage also includedthe polar part of area 10 on the left and about 1 cm ofthe right anterior temporal lobe area 38 with the rightamygdala spared RM also had extensive damage to theleft temporal lobe extending back approximately 5 cmincluding areas 38 27 28 21 and 20 and the leftamygdala MR was the most unilateral of the OFC lesionpatients in our sample with extensive damage to area 11on the right and partial damage to area 11 on the leftMR also had some lateral frontal damage on the rightboth dorsolateral and ventrolateral frontal areas 47 45and 9 some damage to the polar portion of area 10 andsome medial frontal damage on the right to the anteriorcingulate area 33 His basal forebrain area was intact onthe left and damaged on the right Strictly unilateraldamage to the OFC is rare because it is not a region inwhich strokes occur

In addition we tested one patient with damage re-stricted to the anterior temporal cortex patient BG Hewas included in the sample to control for the bilateraltemporal damage of RV RB and RM All patients wereat least 6 months post-lesion

We also tested ve non-brain-damaged age-matchedcontrol subjects matched for education with the least-educated patients that is having only a high schooleducation

Tasks

The tasks we used were developmentally graded rangingin difculty from tasks that normal 4-year-old children

Stone et al 643

can do to tasks that children cannot do until ages 9 to11 We used three tasks altogether

1 First-order false belief tasks which develop around3 to 4 years

2 Second-order false belief tasks which developaround 6 to 7 years

3 Comprehension of social faux pas which developsaround 9 to 11 years (see ldquoMethodsrdquo)

The logic for this methodology was that decits in the-ory of mind should be more evident in the tasks thatdevelop later therefore the severity of theory of mindimpairment could be estimated by looking at whichtasks patients had difculty with Other work on decitsin dorsolateral frontal patients has used this kind ofmethodology to pick up subtle decits For instanceGoldstein Bernard Fenwick Burgess and McNeil (1993)tested a patient with a left frontal lobectomy who per-formed in the normal range on the Wisconsin Card

Sorting Task (WCST) Using tasks that were similar to theWCST but required more complex planning these re-searchers were able to document subtle executive func-tion decits

RESULTS

False Belief Tasks

Based on the rst- and second-order false belief tasksneither patient group shows a pronounced theory ofmind decit such as that seen in autism Some patientsmade errors on false belief tasks when they had toremember the stories However when the story was infront of the subjects so that there was no memory loadpatients made almost no errors (see Tables 2 and 3 andFigures 5 and 6)1 Even in the condition with a memoryload it was rare for patients to make errors on the falsebelief question alone without also making errors on thecontrol questions The difference in the proportion of

Figure 1 Extent of patientsrsquo lesions in the dorsolateral frontal cortex reconstructed from CT scans

644 Journal of Cognitive Neuroscience Volume 10 Number 5

problems correct between these two conditions on therst-order false belief tasks was statistically signicant forfour out of ve of the DFC patients LS (z = 292 p lt001) RT (z = 459 p lt 000001) JC (z = 266 p lt 001and WE (z = 211 p lt 005) The difference betweenConditions 1 and 2 was signicant for LS (z = 234 p lt001) and RT (z = 188 p lt 005) on the second-orderfalse belief tasks It was also signicant for RV an OFC

patient (z = 346 p lt 0001) on the rst-order falsebelief task After completing 10 rst-order false beliefproblems in Condition 1 RV requested not to be testedin that condition any more because he found it sodifcult

Differences in performance between the false belieftask and the true belief task were not signicant for anyof the frontal patients except JC (410 vs 1010 z =

Figure 2 Extent of overlapof DFC patientsrsquo lesions recon-structed from CT scans Thearea of 80 to 100 overlap isin area 46 the middle frontalgyrus and the depth of themiddle frontal sulcus

Table 1 Patient Characteristics

Patient AgeTime Since

Lesion Onset Location of Lesion Brodmannrsquos Areas Damaged

JC 72 9 years Left lateral frontal including DFCsuperior temporal

Lateral areas 8 9 areas 44 45 46 areas 4 6

OA 64 12 years Left lateral frontal including DFC Lateral areas 8 9 area 46

WE 67 6 months Left lateral frontal including DFC Lateral area 8 areas 44 45 46

LS 68 22 years Left lateral frontal including DFC Lateral areas 8 9 area 46

RT 80 11 years Left lateral frontal including DFC Areas 45 46

DH 34 16 years Bilateral OFC Extensive bilateral area 11

MR 42 18 years Bilateral OFC Extensive right area 11 partial left area 11right areas 47 45 9 10 and 33 rightbasal forebrain area

RB 51 22 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial left area 38

RM 46 20 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial right area 38 left areas38 28 21 and 20 amygdala spared

RV 45 5 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 left polar area10 partial right area 38 left area 37 2821 and 20 left amygdala

BG 53 4 years Anterior temporal Bilateral area 38

Stone et al 645

293 p lt 0002) this was true only in the condition witha memory load

Because three of the dorsolateral frontal patients wereaphasic and had difculty producing speech we did notask subjects to justify their responses on the false belieftasks However LS spontaneously offered mentalisticjustications on several problems Even on two problemsshe got wrong she provided mental state explanationsthat made sense of her ldquoerrorsrdquo for example ldquoHersquoll prob-ably think itrsquos on the shelf because he wonrsquot see it onthe desk where he left it so that would be the logicalplace to lookrdquo

Order Effects Practice Effects

There were no effects of prior exposure to a particularstory and no practice effects Subjectsrsquo performance didnot improve over the course of testing Subjects were notmore likely to give correct answers for false belief tasksthey were given in the second session than tasks theywere given in the rst session regardless of condition

Because subjects did not perform signicantly betterin the second testing session than in the rst testingsession familiarity with the stories being used did notseem to help them answer correctly Only memory loadpredicts their performance indicating that workingmemory limitations may provide the best explanation oftheir results

In the condition without a memory load there was noevidence for a theory of mind decit at the level of 4-to 7-year-old children in either the orbito-frontal or thedorsolateral frontal patient group

Faux Pas Task

Control subjects and DFC patients correctly detected allof the faux pas and correctly answered who committedthe faux pas In contrast almost all of the OFC patientsmade errors detecting faux pas answering that nothingawkward had been said in the story They sometimesanswered No to the question ldquoDid someone say some-thing they shouldnrsquot have saidrdquo The difference between

Figure 3 Three patientswith bilateral damage toorbito-frontal cortex CT scansare shown for patient DHand MRI scans are shown forpatients MR and RV The im-ages in columns 1 and 2show the damage to orbito-frontal cortex column 3shows that dorsolateral frontalcortex is spared in patientsDH and RV with some rightdorsolateral damage in patientMR Medial frontal cortex isspared in all patients

646 Journal of Cognitive Neuroscience Volume 10 Number 5

the OFC patients and the control subjects (who were atceiling) in the detection of faux pas is signicant (t8 =2828 p lt 002) All of the OFC patients answered thecontrol questions correctly Thus it appears that theyunderstood the stories yet didnrsquot realize that somethinginappropriate had been said (see Table 4)

In contrast the DFC patients got confused about thedetails of the story and as a consequence all four an-swered some of the control questions incorrectly Forexample WE mixed up who the surprise party was forand therefore made errors on both the control questionand the question of who committed the faux pas be-cause he had mixed up the two charactersrsquo names RTmade a similar mistake on this problem Although heanswered the rst two questions correctly he got mixedup on the later questions and made mistakes OA statedon two of the control questions that he didnrsquot rememberwhat the answer was All of these errors were made even

though the story was right in front of the patients onthe desk and the experimenter told them that they couldlook back at the stories They did not always do so evenwhen they got confused

All of the stories contained a faux pas that is someonesaying something awkward that should not have beensaid There are two possible explanations for why sub-jects might answer yes to the question of whether some-thing awkward had been said on all 10 stories Oneexplanation is that they correctly recognized all of thefaux pas Another possible explanation is that theymerely had a Yes bias We can control for the fact thatthe DFC patients always answered Yes on this questionby looking at their responses to the question of who saidsomething they should not have said If they did notunderstand that a faux pas had been committed andwere merely saying Yes because of a Yes bias they wouldhave not answered the questions about who committed

Figure 4 The top two pa-tients RM and RB have bilat-eral damage to theorbito-frontal cortex and addi-tional bilateral damage to theanterior temporal cortex withmore extensive damage onthe left RMrsquos anterior tempo-ral damage is more extensivethan RBrsquos and includes oneamygdala The patient in thelast row BG has restricted an-terior temporal damage Col-umn 1 shows damage to thetemporal lobes columns 2and 3 show damage (or inthe case of BG lack of dam-age) to orbito-frontal cortexAll images are T2 weightedaxial MRI slices

Stone et al 647

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

house The directions contain four different ambiguitiessuch that a person could make mistakes and end up atthe wrong house After reading the directions the experi-menter asks the subject to identify which parts of thedirections were ambiguous and could have led the lostperson to make a mistake This task requires perspective-taking and understanding false belief However the taskalso places considerable demands on working memorybecause the subject has to keep all of the directions inmemory to answer the question Because DFC patientstypically have difculty with working memory (StussEskes amp Foster 1994) these patients could have failedon this task because of working memory limitationsrather than because their theory of mind was impaired

We undertook to test a series of developmentallygraded theory of mind tasks in frontal lobe patients todetermine if any subtle theory of mind decits could bepicked up in patients with lesions in the frontal lobe Wetested patients with damage in orbito-frontal cortex be-cause they clearly have decits in social behavior andbecause Baron-Cohen et al (1994) found OFC activationwith a theory of mind task We also tested patients withdamage to dorsolateral frontal cortex to compare theirtheory of mind performance to that of the patientstested by Price et al (1990) We used tasks in which wecould control for the working memory demands of thetask

SUBJECTS

We tested ve patients with damage to the left lateralfrontal cortex including both dorsal regions of the lat-eral frontal cortex and more ventrolateral regions Wewill refer to these patients as having DFC damage Thisdoes not imply that their damage is restricted to dorso-lateral regions of the lateral frontal cortex only that allve patients have DFC damage Figure 1 shows comput-erized axial tomograph (CT) reconstructions for individ-ual patients and Figure 2 shows the degree of overlapof the patientsrsquo lesions in different areas (see Table 1 forpatient characteristics) Of the ve DFC patients fourhad damage to the lateral portion of Brodmannrsquos area 8and three of these also had damage to lateral area 9These patients had middle cerebral artery infarcts sotheir damage included only the lateral part of areas 8 and9 sparing the medial portion All ve patients had dam-age in the middle frontal gyrus and middle frontal sulcusBrodmannrsquos area 46 Our map of area 46 is based onrecent quantitative analysis of cytoarchitectonic featuresof cells in different areas of the frontal cortex(Rajkowska amp Goldman-Rakic 1995a 1995b) Area 46includes ldquocentral portions of one or more convolutionsof the middle frontal gyrus and extending to the depthof the middle frontal sulcusrdquo (Rajkowska amp Goldman-Rakic 1995b p 328) Slices 4 and 5 in our lesion overlapgure Figure 2 show that all ve DFC patients have

damage in the middle frontal gyrus and the depth of themiddle frontal sulcus These patients all had unilaterallesions from middle cerebral artery infarcts Bilaterallesions in DFC due to stroke are rare

Three of the DFC patients were aphasic JC WE andRT All three had difculty producing speech A motorspeech evaluation characterized these patients as havingapraxia and disarthria WErsquos score on the Western Apha-sia Battery was 963 JCrsquos was 919 WE scored 55 onthe Boston Naming Test JC scored 45 Both WE and JCare characterized as having anomic aphasia For compari-son OA who is not aphasic scored 996 on the WABand 58 on the Boston Naming Test

We also tested ve patients with bilateral damage toOFC from head trauma See Table 1 and Figures 3 and 4for summaries of the areas damaged in each patient andCT and magnetic resonance image (MRI) scans showingthe extent of the damage DH RV RM and RB allhad extensive bilateral damage in area 11 and no damageto either the lateral frontal cortex or the basal forebrainarea The damage in RV and RB also included the polarparts of area 10 bilaterally with minimal damage to area38 on the left side the very tip of the left temporal lobeIn addition RB had more extensive damage to the lefttemporal lobe including areas 38 28 21 and 20 andsparing the left amygdala RMrsquos damage also includedthe polar part of area 10 on the left and about 1 cm ofthe right anterior temporal lobe area 38 with the rightamygdala spared RM also had extensive damage to theleft temporal lobe extending back approximately 5 cmincluding areas 38 27 28 21 and 20 and the leftamygdala MR was the most unilateral of the OFC lesionpatients in our sample with extensive damage to area 11on the right and partial damage to area 11 on the leftMR also had some lateral frontal damage on the rightboth dorsolateral and ventrolateral frontal areas 47 45and 9 some damage to the polar portion of area 10 andsome medial frontal damage on the right to the anteriorcingulate area 33 His basal forebrain area was intact onthe left and damaged on the right Strictly unilateraldamage to the OFC is rare because it is not a region inwhich strokes occur

In addition we tested one patient with damage re-stricted to the anterior temporal cortex patient BG Hewas included in the sample to control for the bilateraltemporal damage of RV RB and RM All patients wereat least 6 months post-lesion

We also tested ve non-brain-damaged age-matchedcontrol subjects matched for education with the least-educated patients that is having only a high schooleducation

Tasks

The tasks we used were developmentally graded rangingin difculty from tasks that normal 4-year-old children

Stone et al 643

can do to tasks that children cannot do until ages 9 to11 We used three tasks altogether

1 First-order false belief tasks which develop around3 to 4 years

2 Second-order false belief tasks which developaround 6 to 7 years

3 Comprehension of social faux pas which developsaround 9 to 11 years (see ldquoMethodsrdquo)

The logic for this methodology was that decits in the-ory of mind should be more evident in the tasks thatdevelop later therefore the severity of theory of mindimpairment could be estimated by looking at whichtasks patients had difculty with Other work on decitsin dorsolateral frontal patients has used this kind ofmethodology to pick up subtle decits For instanceGoldstein Bernard Fenwick Burgess and McNeil (1993)tested a patient with a left frontal lobectomy who per-formed in the normal range on the Wisconsin Card

Sorting Task (WCST) Using tasks that were similar to theWCST but required more complex planning these re-searchers were able to document subtle executive func-tion decits

RESULTS

False Belief Tasks

Based on the rst- and second-order false belief tasksneither patient group shows a pronounced theory ofmind decit such as that seen in autism Some patientsmade errors on false belief tasks when they had toremember the stories However when the story was infront of the subjects so that there was no memory loadpatients made almost no errors (see Tables 2 and 3 andFigures 5 and 6)1 Even in the condition with a memoryload it was rare for patients to make errors on the falsebelief question alone without also making errors on thecontrol questions The difference in the proportion of

Figure 1 Extent of patientsrsquo lesions in the dorsolateral frontal cortex reconstructed from CT scans

644 Journal of Cognitive Neuroscience Volume 10 Number 5

problems correct between these two conditions on therst-order false belief tasks was statistically signicant forfour out of ve of the DFC patients LS (z = 292 p lt001) RT (z = 459 p lt 000001) JC (z = 266 p lt 001and WE (z = 211 p lt 005) The difference betweenConditions 1 and 2 was signicant for LS (z = 234 p lt001) and RT (z = 188 p lt 005) on the second-orderfalse belief tasks It was also signicant for RV an OFC

patient (z = 346 p lt 0001) on the rst-order falsebelief task After completing 10 rst-order false beliefproblems in Condition 1 RV requested not to be testedin that condition any more because he found it sodifcult

Differences in performance between the false belieftask and the true belief task were not signicant for anyof the frontal patients except JC (410 vs 1010 z =

Figure 2 Extent of overlapof DFC patientsrsquo lesions recon-structed from CT scans Thearea of 80 to 100 overlap isin area 46 the middle frontalgyrus and the depth of themiddle frontal sulcus

Table 1 Patient Characteristics

Patient AgeTime Since

Lesion Onset Location of Lesion Brodmannrsquos Areas Damaged

JC 72 9 years Left lateral frontal including DFCsuperior temporal

Lateral areas 8 9 areas 44 45 46 areas 4 6

OA 64 12 years Left lateral frontal including DFC Lateral areas 8 9 area 46

WE 67 6 months Left lateral frontal including DFC Lateral area 8 areas 44 45 46

LS 68 22 years Left lateral frontal including DFC Lateral areas 8 9 area 46

RT 80 11 years Left lateral frontal including DFC Areas 45 46

DH 34 16 years Bilateral OFC Extensive bilateral area 11

MR 42 18 years Bilateral OFC Extensive right area 11 partial left area 11right areas 47 45 9 10 and 33 rightbasal forebrain area

RB 51 22 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial left area 38

RM 46 20 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial right area 38 left areas38 28 21 and 20 amygdala spared

RV 45 5 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 left polar area10 partial right area 38 left area 37 2821 and 20 left amygdala

BG 53 4 years Anterior temporal Bilateral area 38

Stone et al 645

293 p lt 0002) this was true only in the condition witha memory load

Because three of the dorsolateral frontal patients wereaphasic and had difculty producing speech we did notask subjects to justify their responses on the false belieftasks However LS spontaneously offered mentalisticjustications on several problems Even on two problemsshe got wrong she provided mental state explanationsthat made sense of her ldquoerrorsrdquo for example ldquoHersquoll prob-ably think itrsquos on the shelf because he wonrsquot see it onthe desk where he left it so that would be the logicalplace to lookrdquo

Order Effects Practice Effects

There were no effects of prior exposure to a particularstory and no practice effects Subjectsrsquo performance didnot improve over the course of testing Subjects were notmore likely to give correct answers for false belief tasksthey were given in the second session than tasks theywere given in the rst session regardless of condition

Because subjects did not perform signicantly betterin the second testing session than in the rst testingsession familiarity with the stories being used did notseem to help them answer correctly Only memory loadpredicts their performance indicating that workingmemory limitations may provide the best explanation oftheir results

In the condition without a memory load there was noevidence for a theory of mind decit at the level of 4-to 7-year-old children in either the orbito-frontal or thedorsolateral frontal patient group

Faux Pas Task

Control subjects and DFC patients correctly detected allof the faux pas and correctly answered who committedthe faux pas In contrast almost all of the OFC patientsmade errors detecting faux pas answering that nothingawkward had been said in the story They sometimesanswered No to the question ldquoDid someone say some-thing they shouldnrsquot have saidrdquo The difference between

Figure 3 Three patientswith bilateral damage toorbito-frontal cortex CT scansare shown for patient DHand MRI scans are shown forpatients MR and RV The im-ages in columns 1 and 2show the damage to orbito-frontal cortex column 3shows that dorsolateral frontalcortex is spared in patientsDH and RV with some rightdorsolateral damage in patientMR Medial frontal cortex isspared in all patients

646 Journal of Cognitive Neuroscience Volume 10 Number 5

the OFC patients and the control subjects (who were atceiling) in the detection of faux pas is signicant (t8 =2828 p lt 002) All of the OFC patients answered thecontrol questions correctly Thus it appears that theyunderstood the stories yet didnrsquot realize that somethinginappropriate had been said (see Table 4)

In contrast the DFC patients got confused about thedetails of the story and as a consequence all four an-swered some of the control questions incorrectly Forexample WE mixed up who the surprise party was forand therefore made errors on both the control questionand the question of who committed the faux pas be-cause he had mixed up the two charactersrsquo names RTmade a similar mistake on this problem Although heanswered the rst two questions correctly he got mixedup on the later questions and made mistakes OA statedon two of the control questions that he didnrsquot rememberwhat the answer was All of these errors were made even

though the story was right in front of the patients onthe desk and the experimenter told them that they couldlook back at the stories They did not always do so evenwhen they got confused

All of the stories contained a faux pas that is someonesaying something awkward that should not have beensaid There are two possible explanations for why sub-jects might answer yes to the question of whether some-thing awkward had been said on all 10 stories Oneexplanation is that they correctly recognized all of thefaux pas Another possible explanation is that theymerely had a Yes bias We can control for the fact thatthe DFC patients always answered Yes on this questionby looking at their responses to the question of who saidsomething they should not have said If they did notunderstand that a faux pas had been committed andwere merely saying Yes because of a Yes bias they wouldhave not answered the questions about who committed

Figure 4 The top two pa-tients RM and RB have bilat-eral damage to theorbito-frontal cortex and addi-tional bilateral damage to theanterior temporal cortex withmore extensive damage onthe left RMrsquos anterior tempo-ral damage is more extensivethan RBrsquos and includes oneamygdala The patient in thelast row BG has restricted an-terior temporal damage Col-umn 1 shows damage to thetemporal lobes columns 2and 3 show damage (or inthe case of BG lack of dam-age) to orbito-frontal cortexAll images are T2 weightedaxial MRI slices

Stone et al 647

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

can do to tasks that children cannot do until ages 9 to11 We used three tasks altogether

1 First-order false belief tasks which develop around3 to 4 years

2 Second-order false belief tasks which developaround 6 to 7 years

3 Comprehension of social faux pas which developsaround 9 to 11 years (see ldquoMethodsrdquo)

The logic for this methodology was that decits in the-ory of mind should be more evident in the tasks thatdevelop later therefore the severity of theory of mindimpairment could be estimated by looking at whichtasks patients had difculty with Other work on decitsin dorsolateral frontal patients has used this kind ofmethodology to pick up subtle decits For instanceGoldstein Bernard Fenwick Burgess and McNeil (1993)tested a patient with a left frontal lobectomy who per-formed in the normal range on the Wisconsin Card

Sorting Task (WCST) Using tasks that were similar to theWCST but required more complex planning these re-searchers were able to document subtle executive func-tion decits

RESULTS

False Belief Tasks

Based on the rst- and second-order false belief tasksneither patient group shows a pronounced theory ofmind decit such as that seen in autism Some patientsmade errors on false belief tasks when they had toremember the stories However when the story was infront of the subjects so that there was no memory loadpatients made almost no errors (see Tables 2 and 3 andFigures 5 and 6)1 Even in the condition with a memoryload it was rare for patients to make errors on the falsebelief question alone without also making errors on thecontrol questions The difference in the proportion of

Figure 1 Extent of patientsrsquo lesions in the dorsolateral frontal cortex reconstructed from CT scans

644 Journal of Cognitive Neuroscience Volume 10 Number 5

problems correct between these two conditions on therst-order false belief tasks was statistically signicant forfour out of ve of the DFC patients LS (z = 292 p lt001) RT (z = 459 p lt 000001) JC (z = 266 p lt 001and WE (z = 211 p lt 005) The difference betweenConditions 1 and 2 was signicant for LS (z = 234 p lt001) and RT (z = 188 p lt 005) on the second-orderfalse belief tasks It was also signicant for RV an OFC

patient (z = 346 p lt 0001) on the rst-order falsebelief task After completing 10 rst-order false beliefproblems in Condition 1 RV requested not to be testedin that condition any more because he found it sodifcult

Differences in performance between the false belieftask and the true belief task were not signicant for anyof the frontal patients except JC (410 vs 1010 z =

Figure 2 Extent of overlapof DFC patientsrsquo lesions recon-structed from CT scans Thearea of 80 to 100 overlap isin area 46 the middle frontalgyrus and the depth of themiddle frontal sulcus

Table 1 Patient Characteristics

Patient AgeTime Since

Lesion Onset Location of Lesion Brodmannrsquos Areas Damaged

JC 72 9 years Left lateral frontal including DFCsuperior temporal

Lateral areas 8 9 areas 44 45 46 areas 4 6

OA 64 12 years Left lateral frontal including DFC Lateral areas 8 9 area 46

WE 67 6 months Left lateral frontal including DFC Lateral area 8 areas 44 45 46

LS 68 22 years Left lateral frontal including DFC Lateral areas 8 9 area 46

RT 80 11 years Left lateral frontal including DFC Areas 45 46

DH 34 16 years Bilateral OFC Extensive bilateral area 11

MR 42 18 years Bilateral OFC Extensive right area 11 partial left area 11right areas 47 45 9 10 and 33 rightbasal forebrain area

RB 51 22 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial left area 38

RM 46 20 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial right area 38 left areas38 28 21 and 20 amygdala spared

RV 45 5 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 left polar area10 partial right area 38 left area 37 2821 and 20 left amygdala

BG 53 4 years Anterior temporal Bilateral area 38

Stone et al 645

293 p lt 0002) this was true only in the condition witha memory load

Because three of the dorsolateral frontal patients wereaphasic and had difculty producing speech we did notask subjects to justify their responses on the false belieftasks However LS spontaneously offered mentalisticjustications on several problems Even on two problemsshe got wrong she provided mental state explanationsthat made sense of her ldquoerrorsrdquo for example ldquoHersquoll prob-ably think itrsquos on the shelf because he wonrsquot see it onthe desk where he left it so that would be the logicalplace to lookrdquo

Order Effects Practice Effects

There were no effects of prior exposure to a particularstory and no practice effects Subjectsrsquo performance didnot improve over the course of testing Subjects were notmore likely to give correct answers for false belief tasksthey were given in the second session than tasks theywere given in the rst session regardless of condition

Because subjects did not perform signicantly betterin the second testing session than in the rst testingsession familiarity with the stories being used did notseem to help them answer correctly Only memory loadpredicts their performance indicating that workingmemory limitations may provide the best explanation oftheir results

In the condition without a memory load there was noevidence for a theory of mind decit at the level of 4-to 7-year-old children in either the orbito-frontal or thedorsolateral frontal patient group

Faux Pas Task

Control subjects and DFC patients correctly detected allof the faux pas and correctly answered who committedthe faux pas In contrast almost all of the OFC patientsmade errors detecting faux pas answering that nothingawkward had been said in the story They sometimesanswered No to the question ldquoDid someone say some-thing they shouldnrsquot have saidrdquo The difference between

Figure 3 Three patientswith bilateral damage toorbito-frontal cortex CT scansare shown for patient DHand MRI scans are shown forpatients MR and RV The im-ages in columns 1 and 2show the damage to orbito-frontal cortex column 3shows that dorsolateral frontalcortex is spared in patientsDH and RV with some rightdorsolateral damage in patientMR Medial frontal cortex isspared in all patients

646 Journal of Cognitive Neuroscience Volume 10 Number 5

the OFC patients and the control subjects (who were atceiling) in the detection of faux pas is signicant (t8 =2828 p lt 002) All of the OFC patients answered thecontrol questions correctly Thus it appears that theyunderstood the stories yet didnrsquot realize that somethinginappropriate had been said (see Table 4)

In contrast the DFC patients got confused about thedetails of the story and as a consequence all four an-swered some of the control questions incorrectly Forexample WE mixed up who the surprise party was forand therefore made errors on both the control questionand the question of who committed the faux pas be-cause he had mixed up the two charactersrsquo names RTmade a similar mistake on this problem Although heanswered the rst two questions correctly he got mixedup on the later questions and made mistakes OA statedon two of the control questions that he didnrsquot rememberwhat the answer was All of these errors were made even

though the story was right in front of the patients onthe desk and the experimenter told them that they couldlook back at the stories They did not always do so evenwhen they got confused

All of the stories contained a faux pas that is someonesaying something awkward that should not have beensaid There are two possible explanations for why sub-jects might answer yes to the question of whether some-thing awkward had been said on all 10 stories Oneexplanation is that they correctly recognized all of thefaux pas Another possible explanation is that theymerely had a Yes bias We can control for the fact thatthe DFC patients always answered Yes on this questionby looking at their responses to the question of who saidsomething they should not have said If they did notunderstand that a faux pas had been committed andwere merely saying Yes because of a Yes bias they wouldhave not answered the questions about who committed

Figure 4 The top two pa-tients RM and RB have bilat-eral damage to theorbito-frontal cortex and addi-tional bilateral damage to theanterior temporal cortex withmore extensive damage onthe left RMrsquos anterior tempo-ral damage is more extensivethan RBrsquos and includes oneamygdala The patient in thelast row BG has restricted an-terior temporal damage Col-umn 1 shows damage to thetemporal lobes columns 2and 3 show damage (or inthe case of BG lack of dam-age) to orbito-frontal cortexAll images are T2 weightedaxial MRI slices

Stone et al 647

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

problems correct between these two conditions on therst-order false belief tasks was statistically signicant forfour out of ve of the DFC patients LS (z = 292 p lt001) RT (z = 459 p lt 000001) JC (z = 266 p lt 001and WE (z = 211 p lt 005) The difference betweenConditions 1 and 2 was signicant for LS (z = 234 p lt001) and RT (z = 188 p lt 005) on the second-orderfalse belief tasks It was also signicant for RV an OFC

patient (z = 346 p lt 0001) on the rst-order falsebelief task After completing 10 rst-order false beliefproblems in Condition 1 RV requested not to be testedin that condition any more because he found it sodifcult

Differences in performance between the false belieftask and the true belief task were not signicant for anyof the frontal patients except JC (410 vs 1010 z =

Figure 2 Extent of overlapof DFC patientsrsquo lesions recon-structed from CT scans Thearea of 80 to 100 overlap isin area 46 the middle frontalgyrus and the depth of themiddle frontal sulcus

Table 1 Patient Characteristics

Patient AgeTime Since

Lesion Onset Location of Lesion Brodmannrsquos Areas Damaged

JC 72 9 years Left lateral frontal including DFCsuperior temporal

Lateral areas 8 9 areas 44 45 46 areas 4 6

OA 64 12 years Left lateral frontal including DFC Lateral areas 8 9 area 46

WE 67 6 months Left lateral frontal including DFC Lateral area 8 areas 44 45 46

LS 68 22 years Left lateral frontal including DFC Lateral areas 8 9 area 46

RT 80 11 years Left lateral frontal including DFC Areas 45 46

DH 34 16 years Bilateral OFC Extensive bilateral area 11

MR 42 18 years Bilateral OFC Extensive right area 11 partial left area 11right areas 47 45 9 10 and 33 rightbasal forebrain area

RB 51 22 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial left area 38

RM 46 20 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 polar area 10bilaterally partial right area 38 left areas38 28 21 and 20 amygdala spared

RV 45 5 years Bilateral OFC amp anterior temporal Extensive bilateral area 11 left polar area10 partial right area 38 left area 37 2821 and 20 left amygdala

BG 53 4 years Anterior temporal Bilateral area 38

Stone et al 645

293 p lt 0002) this was true only in the condition witha memory load

Because three of the dorsolateral frontal patients wereaphasic and had difculty producing speech we did notask subjects to justify their responses on the false belieftasks However LS spontaneously offered mentalisticjustications on several problems Even on two problemsshe got wrong she provided mental state explanationsthat made sense of her ldquoerrorsrdquo for example ldquoHersquoll prob-ably think itrsquos on the shelf because he wonrsquot see it onthe desk where he left it so that would be the logicalplace to lookrdquo

Order Effects Practice Effects

There were no effects of prior exposure to a particularstory and no practice effects Subjectsrsquo performance didnot improve over the course of testing Subjects were notmore likely to give correct answers for false belief tasksthey were given in the second session than tasks theywere given in the rst session regardless of condition

Because subjects did not perform signicantly betterin the second testing session than in the rst testingsession familiarity with the stories being used did notseem to help them answer correctly Only memory loadpredicts their performance indicating that workingmemory limitations may provide the best explanation oftheir results

In the condition without a memory load there was noevidence for a theory of mind decit at the level of 4-to 7-year-old children in either the orbito-frontal or thedorsolateral frontal patient group

Faux Pas Task

Control subjects and DFC patients correctly detected allof the faux pas and correctly answered who committedthe faux pas In contrast almost all of the OFC patientsmade errors detecting faux pas answering that nothingawkward had been said in the story They sometimesanswered No to the question ldquoDid someone say some-thing they shouldnrsquot have saidrdquo The difference between

Figure 3 Three patientswith bilateral damage toorbito-frontal cortex CT scansare shown for patient DHand MRI scans are shown forpatients MR and RV The im-ages in columns 1 and 2show the damage to orbito-frontal cortex column 3shows that dorsolateral frontalcortex is spared in patientsDH and RV with some rightdorsolateral damage in patientMR Medial frontal cortex isspared in all patients

646 Journal of Cognitive Neuroscience Volume 10 Number 5

the OFC patients and the control subjects (who were atceiling) in the detection of faux pas is signicant (t8 =2828 p lt 002) All of the OFC patients answered thecontrol questions correctly Thus it appears that theyunderstood the stories yet didnrsquot realize that somethinginappropriate had been said (see Table 4)

In contrast the DFC patients got confused about thedetails of the story and as a consequence all four an-swered some of the control questions incorrectly Forexample WE mixed up who the surprise party was forand therefore made errors on both the control questionand the question of who committed the faux pas be-cause he had mixed up the two charactersrsquo names RTmade a similar mistake on this problem Although heanswered the rst two questions correctly he got mixedup on the later questions and made mistakes OA statedon two of the control questions that he didnrsquot rememberwhat the answer was All of these errors were made even

though the story was right in front of the patients onthe desk and the experimenter told them that they couldlook back at the stories They did not always do so evenwhen they got confused

All of the stories contained a faux pas that is someonesaying something awkward that should not have beensaid There are two possible explanations for why sub-jects might answer yes to the question of whether some-thing awkward had been said on all 10 stories Oneexplanation is that they correctly recognized all of thefaux pas Another possible explanation is that theymerely had a Yes bias We can control for the fact thatthe DFC patients always answered Yes on this questionby looking at their responses to the question of who saidsomething they should not have said If they did notunderstand that a faux pas had been committed andwere merely saying Yes because of a Yes bias they wouldhave not answered the questions about who committed

Figure 4 The top two pa-tients RM and RB have bilat-eral damage to theorbito-frontal cortex and addi-tional bilateral damage to theanterior temporal cortex withmore extensive damage onthe left RMrsquos anterior tempo-ral damage is more extensivethan RBrsquos and includes oneamygdala The patient in thelast row BG has restricted an-terior temporal damage Col-umn 1 shows damage to thetemporal lobes columns 2and 3 show damage (or inthe case of BG lack of dam-age) to orbito-frontal cortexAll images are T2 weightedaxial MRI slices

Stone et al 647

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

293 p lt 0002) this was true only in the condition witha memory load

Because three of the dorsolateral frontal patients wereaphasic and had difculty producing speech we did notask subjects to justify their responses on the false belieftasks However LS spontaneously offered mentalisticjustications on several problems Even on two problemsshe got wrong she provided mental state explanationsthat made sense of her ldquoerrorsrdquo for example ldquoHersquoll prob-ably think itrsquos on the shelf because he wonrsquot see it onthe desk where he left it so that would be the logicalplace to lookrdquo

Order Effects Practice Effects

There were no effects of prior exposure to a particularstory and no practice effects Subjectsrsquo performance didnot improve over the course of testing Subjects were notmore likely to give correct answers for false belief tasksthey were given in the second session than tasks theywere given in the rst session regardless of condition

Because subjects did not perform signicantly betterin the second testing session than in the rst testingsession familiarity with the stories being used did notseem to help them answer correctly Only memory loadpredicts their performance indicating that workingmemory limitations may provide the best explanation oftheir results

In the condition without a memory load there was noevidence for a theory of mind decit at the level of 4-to 7-year-old children in either the orbito-frontal or thedorsolateral frontal patient group

Faux Pas Task

Control subjects and DFC patients correctly detected allof the faux pas and correctly answered who committedthe faux pas In contrast almost all of the OFC patientsmade errors detecting faux pas answering that nothingawkward had been said in the story They sometimesanswered No to the question ldquoDid someone say some-thing they shouldnrsquot have saidrdquo The difference between

Figure 3 Three patientswith bilateral damage toorbito-frontal cortex CT scansare shown for patient DHand MRI scans are shown forpatients MR and RV The im-ages in columns 1 and 2show the damage to orbito-frontal cortex column 3shows that dorsolateral frontalcortex is spared in patientsDH and RV with some rightdorsolateral damage in patientMR Medial frontal cortex isspared in all patients

646 Journal of Cognitive Neuroscience Volume 10 Number 5

the OFC patients and the control subjects (who were atceiling) in the detection of faux pas is signicant (t8 =2828 p lt 002) All of the OFC patients answered thecontrol questions correctly Thus it appears that theyunderstood the stories yet didnrsquot realize that somethinginappropriate had been said (see Table 4)

In contrast the DFC patients got confused about thedetails of the story and as a consequence all four an-swered some of the control questions incorrectly Forexample WE mixed up who the surprise party was forand therefore made errors on both the control questionand the question of who committed the faux pas be-cause he had mixed up the two charactersrsquo names RTmade a similar mistake on this problem Although heanswered the rst two questions correctly he got mixedup on the later questions and made mistakes OA statedon two of the control questions that he didnrsquot rememberwhat the answer was All of these errors were made even

though the story was right in front of the patients onthe desk and the experimenter told them that they couldlook back at the stories They did not always do so evenwhen they got confused

All of the stories contained a faux pas that is someonesaying something awkward that should not have beensaid There are two possible explanations for why sub-jects might answer yes to the question of whether some-thing awkward had been said on all 10 stories Oneexplanation is that they correctly recognized all of thefaux pas Another possible explanation is that theymerely had a Yes bias We can control for the fact thatthe DFC patients always answered Yes on this questionby looking at their responses to the question of who saidsomething they should not have said If they did notunderstand that a faux pas had been committed andwere merely saying Yes because of a Yes bias they wouldhave not answered the questions about who committed

Figure 4 The top two pa-tients RM and RB have bilat-eral damage to theorbito-frontal cortex and addi-tional bilateral damage to theanterior temporal cortex withmore extensive damage onthe left RMrsquos anterior tempo-ral damage is more extensivethan RBrsquos and includes oneamygdala The patient in thelast row BG has restricted an-terior temporal damage Col-umn 1 shows damage to thetemporal lobes columns 2and 3 show damage (or inthe case of BG lack of dam-age) to orbito-frontal cortexAll images are T2 weightedaxial MRI slices

Stone et al 647

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

the OFC patients and the control subjects (who were atceiling) in the detection of faux pas is signicant (t8 =2828 p lt 002) All of the OFC patients answered thecontrol questions correctly Thus it appears that theyunderstood the stories yet didnrsquot realize that somethinginappropriate had been said (see Table 4)

In contrast the DFC patients got confused about thedetails of the story and as a consequence all four an-swered some of the control questions incorrectly Forexample WE mixed up who the surprise party was forand therefore made errors on both the control questionand the question of who committed the faux pas be-cause he had mixed up the two charactersrsquo names RTmade a similar mistake on this problem Although heanswered the rst two questions correctly he got mixedup on the later questions and made mistakes OA statedon two of the control questions that he didnrsquot rememberwhat the answer was All of these errors were made even

though the story was right in front of the patients onthe desk and the experimenter told them that they couldlook back at the stories They did not always do so evenwhen they got confused

All of the stories contained a faux pas that is someonesaying something awkward that should not have beensaid There are two possible explanations for why sub-jects might answer yes to the question of whether some-thing awkward had been said on all 10 stories Oneexplanation is that they correctly recognized all of thefaux pas Another possible explanation is that theymerely had a Yes bias We can control for the fact thatthe DFC patients always answered Yes on this questionby looking at their responses to the question of who saidsomething they should not have said If they did notunderstand that a faux pas had been committed andwere merely saying Yes because of a Yes bias they wouldhave not answered the questions about who committed

Figure 4 The top two pa-tients RM and RB have bilat-eral damage to theorbito-frontal cortex and addi-tional bilateral damage to theanterior temporal cortex withmore extensive damage onthe left RMrsquos anterior tempo-ral damage is more extensivethan RBrsquos and includes oneamygdala The patient in thelast row BG has restricted an-terior temporal damage Col-umn 1 shows damage to thetemporal lobes columns 2and 3 show damage (or inthe case of BG lack of dam-age) to orbito-frontal cortexAll images are T2 weightedaxial MRI slices

Stone et al 647

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

the faux pas correctly However they all answered thesequestions correctly if they answered the control ques-tions correctly They made errors on who committed thefaux pas only when they mixed up the two charactersand got confused about the story overall We concludethat the DFC patients did not answer Yes merely becausethey had a Yes bias but rather were correctly recognizingthe faux pas contained in the stories

For the empathic understanding question asked in alater session (eg ldquoHow do you think Jill feltrdquo) all sub-jects gave similar answers and all subjects demonstratedappropriate empathic understanding For example in thestory about Jill and the curtains all subjects indicatedthat Jill would have felt hurt or angry Even OFC patientswho had failed to detect faux pas on particular storiesmade appropriate empathic inferences about what thecharacters in those stories would have felt Their answersdid not differ from those of control subjects

DISCUSSION

The performance of the bilateral OFC patients on thesetasks is parallel to what has been found for individualswith Aspergerrsquos syndrome They had no difculty under-standing the stories as indexed by their performance onthe control questions but they failed to recognize thatsome faux pas had been committed Their performanceon this task is consistent with their behavior in everydaylife in which they frequently say inappropriate thingsand inappropriately analyze social situations Like chil-dren who are 7 to 8 years old and individuals with As-pergerrsquos syndrome they can pass rst- and second-ordertheory of mind tasks but make errors on the more dif-cult faux pas task Further research with unilateral OFCpatients would provide more insight into the effect of bi-lateral vs unilateral OFC lesions However the one pa-tient MR in our OFC sample whose lesions were mostlyunilateral was the most impaired on the faux pas task

Table 2 Proportion of Problems Correct and Types of Errors on False Belief Problems with and without Memory Load(n = 20 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested Correct

FBErrors

FB +Control

ControlErrors Correct

FBErrors

FB +Control

ControlErrors

DFC Patients

LS 55 3 4 2 95 0 1 0

RT 25 4 7 4 95 0 1 0

OA 100 0 0 0 100 0 0 0

JC 70 0 4 2 100 0 0 0

WE 80 0 2 2 100 0 0 0

Mean 66 14 34 2 98 0 04 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV 50 0 510 0 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 05 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on false belief questions onlyFB + Control Number of problems on which subject made errors on both false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

648 Journal of Cognitive Neuroscience Volume 10 Number 5

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

Two things are necessary for someone to detect a fauxpas One must understand that one person has knowl-edge that the other person is unaware of or that oneperson has a mistaken belief and one must have theempathic understanding about what kinds of thingssomeone would nd upsetting or insulting Baron-Cohen(1991) found that people with autism could understandothersrsquo emotions if those emotions were caused by situ-ations or desires However they did not understand emo-tions that were caused by belief Their decit wasspecically in integrating empathy with mental stateattribution The OFC patients in this study and the Asper-gerrsquos subjects in Baron-Cohen et alrsquos (1997) study maybe exhibiting a more subtle version of this same type ofdecit Because they can pass rst- and second-orderfalse belief tasks we infer that their errors are not dueto cognitive limitations in understanding the mentalstates of the characters in the stories that is under-standing the levels of false and mistaken belief in the

faux pas stories Because the OFC patients got all theldquoempathyrdquo questions right in the faux pas task perform-ing in the same way as controls we conclude that theirempathic understanding of what another person wouldnd upsetting is intact Rather their errors may be dueto problems connecting their theory of mind inferenceswith an understanding of emotion This interpretation isconsistent with Brothers and Ringrsquos (1992) idea that theamygdala and orbito-frontal cortex are essential brainstructures for the ldquohotrdquo aspects of theory of mind thatis for interpreting the valence and signicance of othersrsquoactions and intentions Many other authors although notinvestigating theory of mind have reported the impor-tance of OFC and the amygdala for understanding thesignicance of othersrsquo actions (Adolphs Tranel Damasioamp Damasio 1994 Cahill et al 1996 Damasio et al 1990Franzen amp Myers 1973 Hornak Rolls amp Wade 1996Kling amp Steklis 1976 McGaugh 1990 Morris et al 1996Rolls 1996 Young et al 1995)

Table 3 Proportion of Problems Correct and Types of Errors on Second-Order False Belief Problems with and withoutMemory Load (n = 10 problems in each condition)

Condition 1 Memory Load Condition 2 No Memory Load

Grouptested

Correct

FBErrors

FB +Control

ControlErrors

Correct

FBErrors

FB +Control

Control Errors

DFC Patients

LS 40 3 2 1 90 0 1 0

RT 70 1 2 0 100 0 0 0

OA 100 0 0 0 100 0 0 0

JC 100 0 0 0 100 0 0 0

WE 100 0 0 0 100 0 0 0

Mean 82 08 08 02 99 0 02 0

OFC patients

DH 100 0 0 0 100 0 0 0

MR 100 0 0 0 100 0 0 0

RV not tested 100 0 0 0

RM 100 0 0 0 not tested

RB 100 0 0 0 100 0 0 0

Mean 100 0 0 0 100 0 0 0

Anterior Temporal Control

BG 100 0 0 0 100 0 0 0

Normal Controls

Mean 100 0 0 0 100 0 0 0

Key to column headingsFB Errors Number of problems on which subject made errors on second-order false belief questions onlyFB + Control Number of problems on which subject made errors on both second-order false belief and control questionsControl Errors Number of problems on which subject made errors on control questions only

Stone et al 649

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

Saver and Damasio (1991) found that ventromedialpatients had no difculty with ldquoabstract social knowl-edgerdquo that is these patients could gure out solutionsfor interpersonal problems between other people quitewell It was only when they had to make social decisionsin their own lives that they showed a decit In our studyalthough the faux pas stories were all about other peo-ple the OFC patients could not always apply abstractsocial knowledge to tell that something was said thatwould be awkward or should have not been said This isconsistent with the idea that recognizing a faux pas evenwhen committed by someone else requires some affec-tive understanding in addition to abstract mental stateattribution

We note that we found no evidence that the patientswith anterior temporal damage had any difculty withthe short narratives we used Patients BG with onlyanterior temporal damage and RB with OFC and ante-rior temporal damage made no errors on any task RMmade only faux pas errors Mazoyer et al (1993) andFletcher et al (1995) noted that the anterior temporalregion was active during tasks that involved narrative butnot when subjects were reading unconnected sentencesIt appears that the damaged areas in the anterior tempo-

ral cortex in the patients we tested were not crucial forunderstanding these simple narratives

The PET imaging results of Fletcher et al (1995) andGoel et al (1995) point to Brodmannrsquos areas 8 and 9 inthe left medial frontal cortex as being ldquocritically involvedin mentalisingrdquo (Fletcher et al 1995 p 121) Our studysupplements theirs by investigating the role of otherfrontal regions OFC and DFC in theory of mind infer-ences We found that OFC patientsrsquo performance on thefaux pas task in combination with being able to passsimple rst- and second-order false belief tasks is consis-tent with their having a subtle theory of mind decitlike Aspergerrsquos subjects We found no similar evidence fortheory of mind decits in unilateral DFC patients Ourpatients did not have damage in the areas activated inFletcher et alrsquos study so we cannot comment onwhether these medial areas are critical for theory ofmind inferences

Theory of mind is a complex high-level cognitive abil-ity developing over the course of many years and com-pleting its development relatively late Baron-Cohen(1995) has discussed several components of theory ofmind suggesting that it is not a unitary module but acollection of inferential abilities Such a complex multi-

Figure 5 Patientsrsquo perfor-mance ( correct) on 20 rst-order false belief problems

650 Journal of Cognitive Neuroscience Volume 10 Number 5

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

component cognitive ability is unlikely to be localizedin a small region of the cortex We propose instead thattheory of mind is a distributed circuit involving manyregions of the cortex in addition to the limbic systemBased on our results here with bilateral OFC patients weconclude that OFC is part of this circuit perhaps particu-larly involved in theory of mind tasks with an affectivecomponent From previous neuroimaging results the leftmedial frontal cortex would appear to be part of thiscircuit also The left DFC patients in our study did notshow decits on any of our tasks so we conclude thatthe left DFC is not crucial to this theory of mind circuitBased on our sample we cannot rule out the possibilitythat bilateral damage to the DFC would produce theoryof mind decits Price et al (1990) did nd decits on aperspective-taking task in patients with bilateral DFCdamage although they did not control for the workingmemory demands of the task Further studies investigat-ing the question of how bilateral DFC patients performon theory of mind tasks are needed particularly studiesthat control for working memory The DFC may be in-volved in the operation of theory of mind in real time

in social interaction because of the rapid changes ofattention required in order to keep up with social inter-action The system may be redundant or plastic so thatfocal damage in only one region of the theory of mindcircuit does not produce strong impairments in the abil-ity to make inferences about another personrsquos mentalstates To make an inference as complex as what anotherperson may be thinking many areas of the brain mustwork together

METHODS

First-Order False Belief Tasks

These tasks were designed to test subjectsrsquo ability toinfer that someone can have a mistaken belief that isdifferent from their own true belief They were based onWimmer and Perner (1983) and Baron-Cohen et al(1985) False belief tasks typically involve one personputting an object somewhere in the presence of anotherperson and then leaving the room The second personmoves the object to another location while the rst

Figure 6 Patientsrsquo perfor-mance ( correct) on 10second-order false beliefproblems

Stone et al 651

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

person is away The rst person returns and the subjectis asked three questions the ldquobelief questionrdquo whichasks where the rst person thinks the object is andrequires an understanding of othersrsquo mental states theldquoreality questionrdquo which asks where the object really isand the ldquomemory questionrdquo which asks where the ob-ject was in the beginning The control questions ensurethat the subject knows the real current location of theobject and has an accurate memory of where it wasbefore it was moved Subjects who get these questionswrong in addition to the belief question are assumed tohave problems with memory or comprehension notfalse belief

We gave subjects a total of 20 rst-order false beliefproblems with a series of control conditions to test forworking memory problems

Physical Inference vs Mentalistic Inference

One possible confound in the standard false belief taskis that the belief question is both the only question thatasks about mental states and the only question thatrequires an inference rather than just memory Thus weconstructed 10 false belief tasks that allowed us to askanother control question that required a physical infer-ence rather than a mentalistic inference The subject wasread a story and shown photographs depicting the ac-

tion in the story Any reasonable answer was accepted ascorrect for the physical inference

For example one of the stories involved Bill and Jimstanding in their ofce talking First we checked that thesubjects could correctly identify Bill and Jim in thepictures Jim puts an open bottle of ink on his desk Ashe is doing so some ink spills The picture of this scenewas a drawing of an ink bottle with ink splashing outand did not show where the ink spilled None of thesubsequent pictures showed the surface of the desk Jimthen leaves the ofce Bill moves the ink bottle to acabinet and closes the cabinet He goes back over byJimrsquos desk and Jim comes back in There is a wall be-tween the door through which Jim comes back in andthe desk so that Jim cannot see the desk as he reentersThe questions asked were

Belief question When Jim comes back in where will hethink the ink bottle is (correct answer on the desk)

Reality question Where is the ink bottle (in the cabinet)Memory question Where was the ink bottle in the

beginning (desk or Jimrsquos hand)Inference question Where would there be an ink stain

(On the desk or on the oor next to the desk We alsoaccepted as correct answers such as ldquoon Billrsquos handrdquoor ldquoon the shelfrdquo although no subjects gave suchanswers without also saying ldquoon the deskrdquo)

Table 4 Types of Errors Made on Faux Pas Task (n = 10 problems)

GroupTested

DetectedFaux Pas

ControlQuestions

Correctly NamedWho Committed

Faux Pas

Correct AnswerldquoWhy Shouldnrsquot

Have Saidrdquo

Correct AnswerldquoWhy Did They

Say Itrdquo

Correct AnswerldquoHow Would X

Feelrdquo

DFC patients

LS 10 8 8 7 8

RT 10 8 10 9 9

OA 10 8 8 10 9

WE 10 9 9 7 9

Mean 10 825 875 825 875

OFC patients

DH 9 10 9 9 10 10

MR 6 10 6 6 10 10

RV 7 10 7 7 10 10

RM 8 10 8 8 10 10

RB 10 10 10 10 10 10

Mean 8 10 8 8 10 10

Anterior Temporal Control

BG 10 10 10 10 10 10

Normal Controls

Mean 10 10 10 10 10 10

652 Journal of Cognitive Neuroscience Volume 10 Number 5

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

An error is typically scored if the subject says thatwhen the rst person comes back in (Jim in the aboveexample) he thinks the object is in the location it wasmoved to (the cabinet in the above example)

False Belief vs True Belief

Another possible confound is the working memory de-mands of a false belief task The number of story ele-ments that must be held in mind to answer the falsebelief question may be difcult for dorsolateral frontalpatients Thus we constructed 10 more false belief tasksthat could be compared to ldquotrue beliefrdquo tasks which tellthe same story but in which the rst person is presentand watching while the second person moves the objectto another location Thus the story elements are thesame the only difference is that the ldquotrue beliefrdquo tasksdo not require attribution of a false belief Mixing truebelief problems in with false belief problems also pre-vents patients from using a strategy of always answeringthe belief question with the rst location of the objectBoth false belief and true belief stories were videotapedwith actors portraying the story In this task rather thanhaving to give a verbal answer subjects were providedwith pictures of the two locations and asked to point toone of the two in response to the belief questions andcontrol questions

For example one story went as follows in the falsebelief and true belief versions

False Belief

Tony puts some Coke in the cabinet Then he leaves theroom Maria comes along and moves the Coke fromthe cabinet into the refrigerator to chill Later Tonycomes back in

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

True Belief

Tony puts some Coke in the cabinet Maria comes alongand says ldquoWe shouldnrsquot put it there we should put itover hererdquo She moves the Coke from the cabinet tothe refrigerator to chill while Tony watches

Where does Tony think the Coke isWhere is the CokeWhere did Tony put the Coke in the beginning

Controlling for Memory Load

In order to further control for the working memorydemands of these tasks subjects were tested on each ofthe above tasks under two conditions one in which theyhad to remember the story and one in which they didnot If subjects truly had difculty with false belief and

mentalistic inferences memory load should not haveaffected their performance

Condition 1 Memory Load

The subject was read the story while being shown eitherphotographs or a videotape depicting the story actionIf the subject was being shown photos the photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions The pictures were either the same photo-graphs as in Condition 1 or prints of video stills fromthe videotape used in Condition 1

All 20 false belief problems were tested twice oncein Condition 1 and once in Condition 2 counterbalancedbetween two testing sessions For example if a subjectsaw a particular story in Condition 2 in the rst sessionhe or she would see that story in Condition 1 in thesecond session For example some subjects got thevideo version (Condition 1) of the false belieftrue belieftasks in the rst session and others got the video stills(Condition 2) in the rst session The two testing ses-sions were separated by at least month

RM one of the OFC patients completed only Condi-tion 1 the memory load condition on the false belieftasks because he moved to another state during thestudy This was the more difcult condition He com-pleted Condition 1 on the rst- and second-order falsebelief tasks and answered all of the questions on the fauxpas task

Second-Order False Belief Tasks

These tasks were designed to test the ability to under-stand what someone else thinks about what anotherperson thinks Second-order false belief is a more subtletest of theory of mind impairment than rst-order falsebelief but because the task was styled after tasks usedon 6 and 7 year olds it is still not a very complex theoryof mind task

The second-order false belief tasks were adapted fromthe rst-order false belief tasks used above that requiredsubjects to make both mentalistic and physical infer-ences In each story Person 1 puts an object somewhereand leaves the room Person 2 moves the object WhilePerson 1 is out of the room he or she peeks back in andsees the object being moved but Person 2 does notknow that Person 1 has seen this The subject is askedldquoWhen Person 1 comes back in where will Person 2think that Person 1 thinks the object isrdquo Control ques-

Stone et al 653

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

tions ask where the object really is and where the objectwas in the beginning and ask a question requiring aphysical inference Again the subject was read the storyand shown a series of photographs depicting the actiondescribed in the story

As an example two characters Martha and Oliver aresitting in the kitchen talking Oliver is eating cookiesFirst we checked that the subjects could correctly iden-tify Martha and Oliver in the pictures Oliver gets up andleaves the room Martha closes up the box of cookiesand puts them away in a cabinet While he is outside ofthe room Oliver looks back through the keyhole andsees Martha moving the cookies Martha goes back andsits down Then Oliver opens the door

Belief question Where does Martha think that Oliverthinks the cookies are (correct answer on the table)

Reality question Where are the cookies (in the cabinet)Memory question Where were the cookies in the begin-

ning (on the table)Inference question Where would there be cookie

crumbs (on the table on the oor)

Subjects were given 10 such problems Subjects weretested on each problem twice once in a condition inwhich they had to remember the story and one in whichthey did not Subjects were tested in these two condi-tions in two separate sessions counterbalancing whichcondition was tested in which session

Condition 1 Memory Load

The subject was read the story while being shown pho-tographs depicting the story action The photos wereshown one at a time and covered up by the next oneThe subject had to remember the entire story in orderto answer the questions

Condition 2 No Memory Load

The subject was read the story while being shown pic-tures depicting the story action All the pictures re-mained in front of the subject while he or she answeredthe questions

Recognition of Faux Pas Task

Subjects were read a story that told about the occur-rence of a faux pas So that subjects did not have toremember the stories the page with the story on it wasplaced in front of the subject while it was being readand while questions were being asked afterward Forexample two of the stories follow

Jeanette bought her friend Anne a crystal bowl for awedding gift Anne had a big wedding and therewere a lot of presents to keep track of About a yearlater Jeanette was over one night at Annersquos for din-

ner Jeanette dropped a wine bottle by accident onthe crystal bowl and the bowl shattered ldquoIrsquom reallysorry Irsquove broken the bowlrdquo said Jeanette ldquoDonrsquotworry rdquo said Anne ldquoI never liked it anyway Some-one gave it to me for my weddingrdquo

Helenrsquos husband was throwing a surprise party forher birthday He invited Sarah a friend of Helenrsquosand said ldquoDonrsquot tell anyone especially Helenrdquo Theday before the party Helen was over at Sarahrsquos andSarah spilled some coffee on a new dress that washanging over her chair ldquoOhrdquo said Sarah ldquoI was go-ing to wear this to your partyrdquo ldquoWhat partyrdquo saidHelen ldquoCome onrdquo said Sarah ldquoLetrsquos go see if we canget the stain outrdquo

All of the stories contained a faux pas someone sayingsomething awkward In three of the stories the faux paswas the last thing said in the story and in seven of themit was not the last thing said

After the story subjects were asked a series of ques-tions

1 Did someone say something they shouldnrsquot havesaid (Tests for detection of faux pas)

2 Who said something they shouldnrsquot have said(Tests for understanding of faux pas)

3 Why shouldnrsquot they have said it (Requires under-standing mental state of listener)

4 Why did they say it (Requires understanding men-tal state of speaker)

5 An example is What had Jeanette given Anne forher wedding (Control question that asks about somedetail of the story)

Questions 2 through 4 were only asked if the subjectdetected the faux pas that is answered Yes to Question1 If the subject answered No to Question 1 the experi-menter skipped to 5 the control question Subjects weregiven 10 such stories

Understanding a faux pas requires understanding botha mental state of belief or knowledge and having someempathic understanding of how the person in the storywould feel We wanted to test empathic understandingseparately so in a later session subjects were given thesame faux pas stories and asked how they thought thecharacters in the stories would feel For example onestory involved a woman Lisa criticizing some curtainsin her friend Jillrsquos apartment not realizing that Jill hadjust bought them Subjects were read the story andasked ldquoHow do you think Jill feltrdquo

Acknowledgments

This research was supported by NINDS grant F32 NS09977 toValerie E Stone NINDS grant P01 NS17778 to Robert T Knightand Michael S Gazzaniga NINDS grant NS21135 to Robert TKnight and the McDonnell-Pew Foundation The authors wouldlike to thank Donatella Scabini for recruiting patients for re-search Alison Gopnik and Rich Ivry for helpful methodological

654 Journal of Cognitive Neuroscience Volume 10 Number 5

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

suggestions Clint Glaze for help in preparing the stimulusmaterials and Kathy Baynes Robin Dolan Jim Eliassen JeffHutsler Donna Lemongello Helmi Lutsep Oliver Miller LauraNisenson Mado Proverbio Tony Ro Josh Rubinstein DougWalters and Martha Whitman for acting in the stimulus mate-rials

Reprint requests should be sent to Valerie Stone Departmentof Psychology University of Denver 2155 South Race StreetDenver CO 80208-2478

Notes

1 After several problems Patient RV asked not to be testedin Condition 1 with a memory load He said it was too hardand he found he forgot the stories Thus he was not tested onCondition 1 in all problems on the false belieftrue belief taskor in the second-order false belief task

REFERENCES

Adolphs R Tranel D Damasio H amp Damasio A (1994) Im-paired recognition of emotion in facial expressions follow-ing bilateral damage to the human amygdala Nature 372669ndash672

Astington J W amp Gopnik A (1991) Theoretical explanationsof childrenrsquos understanding of the mind British Journalof Developmental Psychology 9 7ndash31

Avis J amp Harris P (1991) Belief-desire reasoning amongBaka children Evidence for a universal conception ofmind Child Development 62 460ndash467

Baron-Cohen S (1989a) Perceptual role-taking and protode-clarative pointing in autism British Journal of Develop-mental Psychology 7 113ndash127

Baron-Cohen S (1989b) The autistic childrsquos theory of mindA case of specic developmental delay Journal of ChildPsychology and Psychiatry 30 285ndash297

Baron-Cohen S (1991) Do people with autism understandwhat causes emotion Child Development 62 385ndash395

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

Baron-Cohen S Leslie A amp Frith U (1985) Does the autis-tic child have a ldquotheory of mindrdquo Cognition 21 37ndash46

Baron-Cohen S Leslie A amp Frith U (1986) Mechanical be-havioral and intentional understanding of picture storiesin autistic children British Journal of Developmental Psy-chology 4 113ndash125

Baron-Cohen S OrsquoRiordan M Stone V E Jones R ampPlaisted K (1997) Recognition of faux pas by normallydeveloping children and children with Asperger syndromeUnpublished manuscript University of Cambridge

Baron-Cohen S amp Ring H (1994) The relationship betweenEDD and ToMM Neuropsychological and neurobiologicalperspectives In P Mitchell amp C Lewis (Eds) Origins ofan understanding of mind (pp 183ndash207) Hillsdale NJErlbaum

Baron-Cohen S Ring H Moriarty J Schmitz B Costa D ampEll P (1994) Recognition of mental state terms Clinicalndings in children with autism and a functional neuroi-maging study of normal adults British Journal of Psychia-try 165 640ndash649

Blumer D amp Benson D F (1975) Personality changes withfrontal and temporal lesions In D F Benson amp D Blumer(Eds) Psychiatric aspects of neurological disease(pp 151ndash169) New York Grune amp Stratton

Brothers L amp Ring B (1992) A neuroethological framework

for the representation of minds Journal of Cognitive Neu-roscience 4 107ndash118

Cahill L Haier R J Fallon J Alkire M T Tang C KeatorD Wu J amp McGaugh J L (1996) Amygdala activity at en-coding correlated with long-term free recall of emotionalinformation Proceedings of the National Academy of Sci-ences 93 8016ndash8021

Charman T amp Baron-Cohen S (1992) Understanding beliefsand drawings A further test of the metarepresentation the-ory of autism Journal of Child Psychology and Psychia-try 33 1105ndash1112

Charman T amp Baron-Cohen S (1995) Understanding pho-tos models and beliefs A test of the modularity thesis oftheory of mind Cognitive Development 10 287ndash298

Damasio A R Tranel D amp Damasio H (1990) Individualswith sociopathic behavior caused by frontal damage failto respond autonomically to social stimuli BehavioralBrain Research 41 81ndash94

Dennett D (1978) Beliefs about beliefs Behavioral amp BrainSciences 4 568ndash570

Dennett D (1987) The intentional stance Cambridge MAMIT Press

Eslinger P J amp Damasio A R (1985) Severe disturbanceof higher cognition after bilateral frontal lobe ablationPatient EVR Neurology 35 1731ndash1741

Flavell J H Botkin P Fry C Wright T amp Jarvis P (1968)The development of role-taking and communication inchildren New York Wiley

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

Franzen E A amp Myers R E (1973) Neural control of socialbehavior Prefrontal and anterior temporal cortex Neuro-psychologia 11 141ndash157

Goel V Grafman J Sadato N amp Hallett M (1995) Model-ing other minds Neuroreport 6 1741ndash1746

Goldstein L H Bernard S Fenwick P B C Burgess P Wamp McNeil J (1993) Unilateral frontal lobectomy can pro-duce strategy application disorder Journal of NeurologyNeurosurgery and Psychiatry 56 274ndash276

Gopnik A amp Astington J W (1988) Childrenrsquos under-standing of representational change and its relation to theunderstanding of false belief and the appearance-reality dis-tinction Child Development 59 26ndash37

Gopnik A amp Wellman H M (1992) Why the childrsquos theoryof mind really is a theory Mind amp Language 7 145ndash171

Happeacute F G E (1993) Communicative competence and the-ory of mind in autism A test of Relevance Theory Cogni-tion 48 101ndash119

Heider F amp Simmel M (1944) An experimental study of ap-parent behavior American Journal of Psychology 57243ndash259

Hornak J Rolls E T amp Wade D (1996) Face and voice ex-pression identication in patients with emotional and be-havioral changes following ventral frontal lobe damageNeuropsychologia 34 247ndash261

Johnson C N amp Wellman H M (1980) Childrenrsquos develop-ing understanding of mental verbs Remember know andguess Child Development 51 1095ndash1102

Kaczmarek B (1984) Analysis of verbal utterances in pa-tients with focal lesions of frontal lobes Brain and Lan-guage 21 52ndash58

Karmiloff-Smith A Klima E Bellugi U Grant J amp Baron-Cohen S (1995) Is there a social module Language faceprocessing and theory of mind in individuals with Wil-liams syndrome Journal of Cognitive Neuroscience 7196ndash208

Stone et al 655

Kling A amp Steklis H D (1976) A neural substrate for aflia-tive behavior in nonhuman primates Brain Behavior ampEvolution 13 216ndash238

Leekam S amp Perner J (1991) Does the autistic child have aldquometarepresentationalrdquo decit Cognition 40 203ndash218

Leslie A M (1987) Pretense and representation in infancyThe origins of ldquotheory of mindrdquo Psychological Review94 412ndash426

Leslie A M amp Thaiss L (1992) Domain specicity in con-ceptual development Neuropsychological evidence fromautism Cognition 43 225ndash251

Mattson A J amp Levin H S (1990) Frontal lobe dysfunctionfollowing closed head injury Journal of Nervous andMental Disease 178 282ndash291

Mazoyer B M Tzourio N Frak V Syrota A Murayama NLevrier O Salamon G Dehaene S Cohen L amp Mehler J(1993) The cortical representation of speech Journal ofCognitive Neuroscience 5 467ndash479

McGaugh J L (1990) Signicance and remembrance Therole of neuromodulatory systems Psychological Science1 15ndash25

Morris J S Frith C D Perrett D I Rowland D YoungA W Calder A J amp Dolan R J (1996) A differential neu-ral response in the human amygdala to fearful and happyfacial expressions Nature 383 812ndash815

Perner J Leekam S R amp Wimmer H (1987) Three-year-oldrsquos difculty with false belief The case for a conceptualdecit British Journal of Developmental Psychology 5125ndash137

Perner J amp Wimmer H (1985) ldquoJohn thinks that Marythinks that rdquo Attribution of second-order false beliefsby 5- to 10-year-old children Journal of ExperimentalChild Psychology 39 437ndash471

Premack D amp Woodruff G (1978) Does the chimpanzeehave a theory of mind Behavioral and Brain Sciences 1515ndash526

Price B Daffner K Stowe R amp Mesulam M (1990) Thecomportmental learning disabilities of early frontal lobedamage Brain 113 1383ndash1393

Rajkowska G amp Goldman-Rakic P S (1995a) Cytoarchitec-tonic denition of prefrontal areas in the normal human

cortex I Remapping of areas 9 and 46 using quantitativecriteria Cerebral Cortex 5 307ndash322

Rajkowska G amp Goldman-Rakic P S (1995b) Cytoarchitec-tonic denition of prefrontal areas in the normal humancortex II Variability in locations of areas 9 and 46 and rela-tionship to the Talairach coordinate system Cerebral Cor-tex 5 323ndash337

Rolls E T (1996) The orbitofrontal cortex PhilosophicalTransactions of the Royal Society 351 1433ndash1443

Saver J L amp Damasio A R (1991) Preserved access andprocessing of social knowledge in a patient with acquiredsociopathy due to ventromedial frontal damage Neuropsy-chologia 29 1241ndash1249

Stuss D T Eskes G A amp Foster J K (1994) Experimentalneuropsychological studies of frontal lobe functions InJ Grafman amp F Boller (Eds) Handbook of neuropsychol-ogy Vol 9 (pp 149ndash185) Amsterdam Elsevier

Tager-Flusberg H (1989) A psycholinguistic perspective onlanguage development in the autistic child In G Dawson(Ed) Autism Nature diagnosis and treatment (pp 92ndash115) New York Guilford Press

Tager-Flusberg H (1993) What language reveals about theunderstanding of minds in children with autism In SBaron-Cohen H Tager-Flusberg amp D J Cohen (Eds) Un-derstanding other minds Perspectives from autism(pp 138ndash157) Oxford Oxford University Press

Wellman H M (1990) The childrsquos theory of mind Cam-bridge MA MIT Press

Wellman H M amp Woolley J D (1990) From simple desiresto ordinary beliefs The early development of everyday psy-chology Cognition 35 245ndash275

Wimmer H amp Perner J (1983) Beliefs about beliefs Repre-sentation and constraining function of wrong beliefs inyoung childrenrsquos understanding of deception Cognition13 103ndash128

Young A W Aggleton J P Hellawell D J Jahnson MBroks P amp Hanley J R (1995) Face processing impair-ments after amygdalotomy Brain 118 15ndash24

Zaitchik D (1990) When representations conict with real-ity The preschoolerrsquos problem with false beliefs andldquofalserdquo photographs Cognition 35 41ndash68

656 Journal of Cognitive Neuroscience Volume 10 Number 5