Brain Injury, July 2010; 24(7–8): 978–987

ORIGINAL ARTICLE

Theory of Mind in patients with ventromedial or dorsolateralprefrontal lesions following traumatic brain injury

ALESSANDRA GERACI1, LUCA SURIAN1,2, MARCO FERRARO3, &ANNA CANTAGALLO4

1Department of Cognitive Sciences, 2Center for Mind-Brain Studies (Cimec), University of Trento, Italy,3Rehabilitation Unit, General Hospital-University of Padua, Italy, and 4Department of Rehabilitation,

Hospital-University of Ferrara, Italy

(Received 27 July 2009; revised 24 May 2010; accepted 18 April 2010)

AbstractPrimary objective: Previous studies on patients with traumatic brain injuries (TBI) and diffuse brain damages have reportedselective deficits in mental states reasoning or ‘Theory of Mind’ (ToM). The goal of the current study is to investigate thefundamental role of the prefrontal cortex in two ToM components: inferential reasoning and social perception.Research design: Selective cognitive impairments following a TBI provide crucial evidence for assessing competing models ofspecific aspects of the cognitive system.Method and procedure: This study compared the performance of patients with predominantly focal lesions in theventromedial (n¼ 11) or dorsolateral prefrontal cortex (n¼ 7) with matched controls (n¼ 20). All subjects performed twoToM tasks: the Eyes Test and the Faux-pas Test.Results: It was found that both groups of patients performed equally poorly on the Eyes Test, but only patients withpredominantly lesions in the ventromedial cortex performed poorly on the Faux-pas test. The group effects on ToM taskscould not be reduced to differences in the global severity of brain injuries.Conclusions: These results provide evidence supporting some current models of the fractionation of the mindreading systemand support the claim that the ventromedial cortex plays a fundamental role in inferential reasoning.

Keywords: Theory of Mind, traumatic brain injury, prefrontal cortex, social cognition

Introduction

The effects of traumatic brain injury (TBI) oncommunication skills and social cognition havebeen investigated in a number of previous studies.Patients with TBI have been reported to show socialisolation [1], poor social skills [2], poor understand-ing of non-literal language [3–5] and difficulties inevaluating the communicative adequacy of utter-ances [6]. They have been described also as havingreduced expression of affection [7] and empathy [8],egocentric biases and inappropriate levels of socialinteraction [4].

One possible explanation for at least some of thesedifficulties in social interaction and communicationis that TBI may result in an acquired impairment inrepresenting and reasoning about mental states, forshort Theory of Mind (ToM). The investigations ofToM deficits in patients with TBI have oftenfocused on three related and different problems:the validity of ToM tasks, the fractionation of ToMcompetence and the localization of mental processesin the brain. The order in which these problemsare listed roughly corresponds to the amount ofempirical data currently available that address them.

Correspondence: Alessandra Geraci, Department of Cognitive Sciences, University of Trento, Corso Bettini 31, 38068 Rovereto (TN), Italy.Tel: þ39-0464-80-8669. E-mail: alessandra.geraci@unitn.it

ISSN 0269–9052 print/ISSN 1362–301X online � 2010 Informa Healthcare Ltd.DOI: 10.3109/02699052.2010.487477

Bibby and McDonald [9] assessed a group ofpatients with TBI on both first- and second-orderToM tasks. The former are meant to tap the abilityto use first-order meta-representations, whereas thelater require the attribution of more complex,second-order meta-representations. The clinicalgroup performed worse than healthy controls onboth ToM tasks and related and control tasks. Ageneralized weakness in inferential skills, combinedwith linguistic and working memory limitations, maythus explain the failures on ToM tasks, particularlyfailures on those requiring second order meta-representations. This conclusion was also supportedby Henry et al. [10], who found poor executivefunctioning in patients with TBI and a correlationbetween these deficits and scores in an advancedmindreading task (the ‘Eyes Test’) [11].

However, lesions following TBI may differ sub-stantially in size and site and thus different causalpathways may underlie poor ToM skills in patientswith TBI. According to Samson et al. [12], beliefreasoning errors of patients with prefrontal lesionsmay arise mainly from an executive functioningdeficit. In contrast, the errors of patients with lesionsin the temporo-parietal junction (TPJ) may bedomain-specific and concern the conceptual knowl-edge involved in ToM.

Havet-Thomassin et al. [13] employed four exec-utive functions tests (Tower of London, StroopColour Word Test, Modified Card Sorting Testsand Trail Making Test) and two ToM tasks (theEyes Test and the Character Intentions Task). Theyfound no relationship between the patients’ perfor-mance on executive functioning tests and theirscores on ToM tasks. These results dovetail withthose on patients that had undergone frontal lobeexcisions [14] and brain-damage from differentaetiologies [15–19] and suggest that failures ofToM tasks in patients are not merely an effect ofweak executive functioning. Stone et al. [20] studiedToM skills in five patients with bilateral damage inthe orbito-frontal cortex and five patients withunilateral left lesions in the dorsolateral frontalcortex. ToM was assessed with three kinds oftasks: first-order false belief tasks, second-orderfalse belief tasks and the Faux-pas test. Patientswith orbito-frontal damage showed good perfor-mance on first- and second- order false belief tasks,but performed poorly on the Faux-pas test, a testthat taps more subtle social reasoning required instory comprehension and failed to detect that some-thing awkward had been said in the stories. Incontrast patients with left dorsolateral lesions per-formed like healthy controls in the Faux-pas test andfailed only in the versions of the tasks that made highworking memory demands. This data suggests thatwhile bilateral lesions in the orbitofrontal areas may

result in domain-specific ToM deficits, lesions in theleft dorsolateral areas are associated with domaingeneral processing limitations.

All previous studies of ToM in patients with TBI,with the exception of Stone et al. [20], includedpatients with lesions extending in several corticalareas. Neurological and neurosurgical investigationsdistinguish between diffuse and focal lesions result-ing from TBI in the frontal, orbito-frontal andtemporal poles [21, 23, 26].

Numerous studies have shown the involvement ofprefrontal areas in mental states reasoning [19,27–30]. In a review of the relevant neuroimagingliterature, Gallagher and Frith [31] claim that thenetwork sub-serving ToM includes a distinctive keyregion located in the medial prefrontal cortex, whilethe superior temporal sulcus and the temporal polesbilaterally are not uniquely associated with ToM. Incontrast, Saxe and Kanwisher [32] claimed that thetemporo-parietal junction is a key region in reason-ing about the true or false contents of others’ beliefs.Studies on brain injuries with heterogeneous aetiol-ogies have also pointed out the role of the prefrontalcortex in ToM [14, 17–19, 33].

While a consensus is emerging about the claimthat different ToM components may have differentneural correlates, no agreement has been reached yeton how ToM skills should be functionally fraction-ated and which neural structures subserve thedifferent ToM components. According to somerecent proposals, ToM can be analysed into twomain components: a perceptual component thatdeals mainly with decoding social information fromfacial expressions and other social signals and aninferential, or conceptual, component, that isresponsible for mental state attribution in socialscenarios [34–38]. Another series of proposals triedto analyse the mindreading system in affective vscognitive components. In an fMRI study, Hyneset al. [39] demonstrated that the medial orbitofron-tal cortex was more involved in emotional ascompared to cognitive perspective-taking.McDonald and Flanagan [40] showed that inpatients with TBI emotion recognition and firstorder theory of mind judgements were not related toinferential reasoning skills, whereas second-ordertheory of mind judgements were related to socialperception. Shamay-Tsoory and Aharon-Peretz [41]studied patients with heterogeneous aetiologies andwith different prefrontal lesions (ventromedial, dor-solateral and mixed) or with posterior lesions. Thisstudy was aimed at examining whether ‘affective’ToM processing is, to some extent, distinct from‘cognitive’ aspects of ToM and depends in part onseparate neuroanatomical substrates. Patients withlesions in the ventromedial cortex had scores signif-icantly lower than patients with lesions in the

ToM following TBI 979

dorsolateral cortex in the affective trials of the storiestask, but not in the cognitive trials. Poor perfor-mance in ‘cognitive trials’ was found associated withextensive prefrontal damage. A recent study showedthe existence of two distinct neural systems forempathy in humans [42]. The findings showed thatinferior frontal gyrus (IFG) was critical for affectiveempathy on the emotion recognition task andventromedial prefrontal cortex was critical for cog-nitive empathy on the theory of mind task.

The main aim of the study reported here was toinvestigate whether affective-perceptual and cogni-tive aspects of ToM competence may be dispropor-tionally damaged by lesions in different prefrontalareas by assessing ToM in patients with TBI andpredominantly focal prefrontal lesions. While previ-ous studies have investigated the performance ofpatients with TBI on a number of ToM tasks, this isthe first study on patients with TBI aimed atevaluating whether different patterns of ToM diffi-culties follow from lesions in different prefrontalcortical sites.

Method

Participants

Eighteen patients with predominantly focal injury inthe prefrontal cortex following a traumatic braininjury (TBI) were recruited in two public healthservices for rehabilitation in Ferrara and Padova(Italy). The aetiology for all cases was head injurycaused by a domestic or road accident. None of thepatients reported a history of psychiatric disease or apre-morbid history of alcohol or drug addiction.Patients with more diffuse axonal injuries wereexcluded on the basis of MRI/CT and neurologicalexamination. Patients with comorbid neurologicalconditions (e.g. stroke or pre-morbid seizure disor-ders) or marked difficulties in executive functioningwere also excluded, as reported by a screening test,the Frontal Assessment Battery (FAB) [43].

All patients lived in their home. Testing wasconducted at the chronic phase of recovery. Allpatients were at least 11 months post-trauma orsurgery, except one that was 5 months post-trauma.The time after onset ranged from 5–288 months(M¼34.72, SD¼ 64.93) and the days of uncon-sciousness ranged from 1–60 (M¼ 14.72, SD¼ 13).The inclusion conditions for patients were: (a) aLevel of Cognitive Functioning (LCF) of 7(Automatic, Appropriate Response) or 8(Purposeful, Appropriate Response) [44]; and (b)an equivalent global score on the WCST equal orgreater than 2. Equivalent global scores below 2reveal weak executive functioning, whereas equiva-lent score equal or greater than 2 reveal executive

functioning within the normal range (see Proceduresection for more details). The GCS (Glasgow ComaScale) scores, measured at arrival to hospital, avail-able for 15 patients, ranged from 3–14 (for GCS� 7the TBI is severe: nine patients; for GCS 8–13 theTBI is moderate: six cases; for GCS 14–15 the TBIis mild: one case; Not Notified in two cases). PTA(post-traumatic amnesia) scores, available for allpatients, ranged from 1–30 weeks (M¼ 13.17,SD¼ 10.43).

Patients with TBI were divided into two groupsaccording to their lesion locations (see Table I andFigure 2), a group of 11 patients with predominantlyfocal injury in the ventromedial prefrontal cortex(VM group) and a group of seven with predomi-nantly focal injury in the dorsolateral prefrontalcortex (DL group). The mean non-verbal IQ of theVM and DL groups did not differ significantly fromthe mean IQ of the HC group, t(29)¼ 0.52 andt(25)¼ 1.4, both p’s> 0.10, respectively.

Twenty healthy participants served as controls(HC group). The HC group included six womenand 14 men, aged 21–50 (M¼36.0, SD¼9.27),with 8–13 years of education (M¼ 11.45,SD¼ 2.23). Healthy controls matched the patientson demographic variables, as shown by an analysis ofvariance (ANOVA) which found no significantdifference between the groups with regard to age(F(2, 35)¼1.61, p¼ 0.21) and years of education(F(2, 35)¼0.33, p¼ 0.72). All subjects completedthe Raven Progressive Matrices to obtain a measureof general intellectual functioning.

Anatomical classification

A neurologist and two neuropsychologists carriedout the anatomic examination and classification ofpatients’ lesions based on the neurological examina-tion and on the most recent CT or MRI scans. Forinclusion, lesions had to be predominantly localizedin the ventromedial or dorsolateral regions of theprefrontal cortex. Lesion sites were transcribed fromCT or MRI images to the appropriate slices of theMRIcro program (Rorden, www.sph.sc.edu/comd/rorden/mricro.html, and see Figure 1) and evaluatedusing standard atlases [45].

The 18 patients were divided in two groupsaccording to their lesion site: seven patients formedthe group with pre-dominantly focal injury in thedorsolateral prefrontal cortex (DL group) and 11 thegroup with pre-dominantly focal injury in the ven-tromedial prefrontal cortex (VM group, see Table Ifor information on patients). In the DL group, fivepatients showed focal lesions in the dorsolateralprefrontal cortex (Broadmann dorsolateral areas8, 9, 43, 44, 45); the remaining two patients(BOL and AND) also had a small parietal lesion.

980 A. Geraci et al.

AZZ BAC BAD BAN

BRE CRE DEP MIN

SIN SOL BOR

Figure 1. Maps of lesions of the patients with predominantly focal injury in ventromedial prefrontal cortex.

Table I. Clinical features of the TBI participants.

Subjects AgeEducation

(years) Lesion site Side IQaExecutive

functioningbTime post-injury

(months) GCSc

Duration ofunconsciousness

(days)PTAd

(weeks)

And 47 13 dorsolateral left 102 4 54 NN 1 4Bol 46 13 dorsolateral right 95 3 11 11 11 5Cle 31 8 dorsolateral left 90 3 19 6 19 5Brz 56 15 dorsolateral left 95 3 10 14 8 1Gia 46 8 dorsolateral right 90 3 12 5 26 4Raz 20 9 dorsolateral left 95 3 15 10 5 1Tam 43 8 dorsolateral left 90 3 14 12 7 2Azz 20 13 ventromedial left 95 2 30 3 16 24Bac 39 13 ventromedial bilateral 90 3 28 4 2 13Bad 32 13 ventromedial left 106 4 12 3 12 25Ban 22 13 ventromedial left 94 3 12 NN 20 30Bor 37 13 ventromedial left 93 4 12 <8 10 8Bre 46 8 ventromedial bilateral 90 3 16 11 60 25Cre 28 8 ventromedial bilateral 94 3 5 8 10 4Dep 46 13 ventromedial bilateral 90 2 60 3 15 24Min 30 8 ventromedial right 92 3 15 3 15 22Sin 38 8 ventromedial bilateral 95 2 288 NN 10 15Sol 23 13 ventromedial bilateral 99 4 12 4 18 25

a Raven’s Progressive Matrices (1938).b Wisconsin Card Sorting Test, equivalent of global score in italian calibration [49].c Glasgow Coma Scale (Teasdale and Jennet, 1974). NN = not notified, IQ = intellectual quotient.d PTA = Post-traumatic amnesia.

ToM following TBI 981

Five patients of the DL group had a left lesion, twopatients had a right lesion. In the VM group, ninepatients showed focal lesions in the ventromedialprefrontal cortex (Brodmann mesial areas 8, 9, 10,24, 32 and orbital areas 10, 11, 12 and 14), onepatient (SOL) also showed a small right parietallesion and another patient (DEP) showed two smalland circumscribed bilateral occipital lesions. In theVM group, five patients had a left lesion, one a rightlesion and six a bilateral lesion.

Considering both groups of patients, the estimatedvolume of the prefrontal lesions ranged from56–940 cm3 (M¼265.06 cm3, SD¼206.53 cm3).In the DL group the volume of the prefrontal lesionsranged from 186–289 (M¼ 234.14 cm3, SD¼33.51 cm3); in the VM group it ranged from56–940 (M¼ 284.73 cm3, SD¼ 265.99 cm3). Thesize of the prefrontal lesions did not differ signifi-cantly in the two groups, t(16)¼ 0.49, p¼0.62.

The DL group included seven men, aged 20–56years (M¼41.29, SD¼11.94), with 8–15 years ofeducation (M¼ 10.57, SD¼2.99), an average IQ(M¼92.14, SD¼ 2.67), Glasgow Coma Score(GCS) scores from 5–14 (M¼ 9.67, SD¼ 3.50;GCS scores were available for six patients of theDL group) and PTA scores from 1–5 weeks(M¼3.14, SD¼ 1.77). The VM group includedtwo women and nine men, aged 20–46 (M¼ 32.82,SD¼ 9.16) with 8–13 years of education(M¼11.18, SD¼ 2.52), a good IQ (M¼ 94.36,SD¼ 4.71), GCS scores from 3–11 (M¼ 5.22,SD¼ 2.99; available for nine patients) and PTAscores from 4–30 weeks (M¼ 19.55, SD¼ 8.26).The VM and DL groups did not differ significantlyon chronological age, t(16)¼1.7, education years,t(16)¼0.46, and IQ, t(16)¼ 1.12, all p’s> 0.10.The mean GCS of patients of the VM group wassignificantly lower than those of patients of the DLgroup, t(13)¼2.63, p< 0.05, whereas the PTAduration is longer in patients of the VM group,t(16)¼6.35, p< 0.001. Patients’ performances ofboth groups on the WCST did not differ signifi-cantly, t(16)¼ 0.45, p¼ 0.65.

Materials and procedures

Theory of mind–perception: Eyes Test. To test ‘per-ceptual’ aspects of ToM competence we used anItalian adaptation of the Eyes Test [46] which isbased on the original English version devised byBaron-Cohen et al. [11]. In order to perform well onthis test participants need to know a number ofmental state terms, match them to the ocular regionof a face and decide whether the terms describeadequately the face by choosing one adjective amongfour possible alternatives. The correct interpretationof visual cues is thus a crucial component of this

task, but another important aspect is lexical knowl-edge. Participants were presented 37 black and whitephotographs showing the ocular regions of male andfemale adults. On each trial, first a photograph waspresented and the participants were asked a controlquestions about the gender of the person in thephoto. Then, four adjectives describing complexemotions or other mental states (e.g. dispirited,bored, embarrassed, flirting) were shown below thepicture and participants were asked to choose theword that best described the photograph.Participants were asked to read all four wordsbefore making their choice and, if they felt thatmore than one word was applicable, to ‘choose justone word, the word which you consider to be mostsuitable’. The Experimenter asked: ‘Which wordbest describes what this person is feeling or think-ing?’ In the Italian version, the four terms presentedin each trial were written below the photo, ratherthan at the corners of the photo as in Baron-Cohenet al. [11]. Following the procedure used by Baron-Cohen et al. [11], participants were encouraged toconsult a glossary of all words used in the taskwhenever they felt they were not sure about theirmeaning. The norms for the Italian adult populationand all the verbal materials are reported in Serafinand Surian [46]. To minimize the negative effects ofimpulsive tendencies, patients were asked to look atthe stimuli for 30 seconds before responding. Themaximum score on test and control questionswas 36.

Theory of mind–reasoning: The Faux-pas Test.

Participants were presented with 10 stories thattold about the occurrence of a ‘social boathook’ orfaux pas. The stories used in the present study wereselected from the 20 faux pas items of the originaltest. In each story, someone said something awkward[47]. Each story involved two or three characters andat least two separate statements. The language usedwas simple. After hearing each story, subjects wereasked a series of questions. The first questionassessed whether a faux pas was present: ‘Didsomeone say something they shouldn’t have said?’Understanding a faux pas required understanding amental states of belief or knowledge and having anempathic understanding of how the person in thestory would feel. If subject answered ‘yes’ to the firstquestion then he/she was asked the other fourquestions. Subjects who answer ‘no’ to the firstquestion don’t get asked this question and score a 0for that item. In the second question the participantswere asked to identify who said the inappropriateutterance. The third questions asked why he/sheshould not have said the silly thing. The fourthquestion asked why the speaker said the silly thing.

982 A. Geraci et al.

The fifth question required an attribution of knowl-edge to the speaker and the sixth question was aboutthe feelings of the listener. Following the testquestions there were two control questions assessingwhether the participant was confused or forgot thedetails of the story. Each story was posted in front ofthe subject and the experimenter read it aloud. Aftera story had been read, participants answered the testand control questions while keeping the printed storyin front of them. This study used 10 of original 20faux pas stories reported in Baron-Cohen et al. [47].The maximum score was 60 on test questions and 20on control questions.

Executive functioning: The Wisconsin Card Sorting Test

(WCST). The WCST was administered usingstandard materials (128 cards) and instructions[48]. Performance was scored according the rulesset out by Laiacona et al. [49]. This task is widelyused for clinical and research purposes to assessreasoning abilities. It requires a number of cognitiveskills including attention shifting, working memoryand cognitive flexibility in problem-solving. Duringthe test a patient is required to match a seriesof cards according to their similarity with fourstimulus cards. At the beginning of the test, fourstimulus cards were placed in front of the participantdisplaying one red triangle, two green stars, threeyellow crosses and four blue circles (from left toright). Three possible sorting categories wereassumed: form, colour and number. The participantwas notified only whether the responses were right orwrong, without mentioning the underlying sortingcriterion. When subject had made 10 consecutivecorrect matches, the criterion was changed, withoutwarning. Performance was assessed using the ‘equiv-alent global score’ [49]. The global score is com-puted by subtracting from the total number ofadministered trials the number of categories com-pleted multiplied by 10. This scoring system provedeffective in identifying dysexecutive patients in acoincided and informative way [49].

Results

The performance of the three groups on the twoToM tasks and the executive functioning test issummarized in Table II. A significant group effectwas found on the test questions of the Eyes Test,F(2, 35)¼ 12.82, p< 0.001, �2

p = 0.42. Both clinicalgroups performed worse than healthy controls(Sheffe post-hoc comparisons, p< 0.01) and themean scores of the two clinical groups did not differsignificantly. All three groups performed at ceilingon the control questions, F(2,35)¼0.23, p¼ 0.79,�2

p = 0.013.A significant group effect for was also found on the

Faux-pas test questions, F(2,35)¼ 19.21, p< 0.001,�2

p = 0.52. The VM group performed significantlyworse than controls (Sheffe, p< 0.001) and the DLgroup (p< 0.02). In contrast, in the DL group scoresdid not differ significantly from controls (p¼0.18).All three groups performed well on the controlquestions of the Faux-pas test and did not differsignificantly from each other, F(2,35)¼ 1.23,p¼ 0.30, �2

p = 0.06.WCST scores correlated with the Eyes Test scores

in both clinical groups (VM group: r¼ 0.64; DLgroup: r¼ 0.80, both p’s< 0.05), but did not corre-late with the Faux-pas Test scores (VM group:r¼0.06; DL group: r¼�0.09, both p’s> 0.05).Also, the performance on the Faux-pas test ques-tions did not correlate with the performance on theEyes Test (DL group: r¼�0.15, p¼ 0.73; VMgroup: r¼ 0.54, p¼ 0.08; HC group: r¼ 0.28,p¼ 0.23).

GCS scores did not correlate with Eyes Test (VMgroup: r¼�0.19; DL groups: r¼�0.37, bothp’s> 0.05) and Faux-pas Test scores (VM group:r¼�0.19; DL group: r¼�0.32, both p’s>0.05).Similarly, PTA duration did not correlate with EyesTest scores (VM group: r¼ 0.26; DL group:r¼0.42, both p’s>0.05) and Faux-pas Test scores(VM group: r¼ 0.56; DL group: r¼�0.44, bothp’s> 0.05).

Altogether, these results show that the differencebetween the two groups of patients on the Faux-pas

Table II. Means and standard deviations on the Theory of Mind tasks (Eyes Test and Faux pas) and the Wisconsin Card Sorting Test(WCST) for the two groups of patients with ventromedial (VM) or dorsolateral (DL) lesions and for the healthy controls.

VM group DL group Control group

M SD M SD M SD

Eyes Test 22.6 4.3 19.8 3.8 27.7 4.1Eyes Test control 35.2 0.9 35.2 1.0 35.4 0.8Faux Pas 25.8 20.1 45.6 15.0 56.2 4.2Faux Pas control 19.6 0.7 19.8 0.4 19.9 0.3WCST 3.0 0.8 3.2 0.4

ToM following TBI 983

Test cannot be explained away with reference to apossible confounding with GCS scores and PTAduration, despite the fact that the patients withpredominantly ventromedial lesions did have a meanGCS significantly lower than patients with predom-inantly dorsolateral lesions. In accordance with this,this study compared VM and DL groups on Faux-pas test results with an ANOVA, which did notreveal a significant effect of group, F(1,16)¼ 4.62,p¼ 0.05, �2

p = 0.224, also after controlling for theeffect of GCS, F(1,12)¼ 3.15, p¼ 0.10, �2

p = 0.208and PTA duration, F(1,15)¼0.09, p¼ 0.76,�2

p = 0.006.This study also compared the VM and DL groups

on Eyes Test results with an ANOVA, that did notreveal a significant effect of group, F(1,16)¼ 2.84,p¼ 0.11, �2

p = 0.151, also after controlling for theeffect of GCS, F(1,12)¼ 1.64, p¼ 0.22, �2

p = 0.120and PTA duration, F(1,15)¼0.05, p¼ 0.81,�2

p = 0.004. From these results, the global severityof injury did not seem the determining factor ofpatients’ performances on both ToM tasks.

Discussion

Two tasks were used to assess social cognition inpatients with TBI, the Eyes Test and the Faux-pasTest. While the former requires the interpretation ofvisual social information, the success on the latterdepends on mental states reasoning skills. Patientswere divided into a dorsolateral and a ventromedialgroup, according to the site of their predominantlyfocal injury. It was found that both groups

performed equally poorly on the Eyes Test, butonly patients with ventromedial lesions performedpoorly on the Faux-pas Test. These findings have anumber of important theoretical implications thatsuggest that both dorsolateral and ventromedialprefrontal areas play an important role in interpret-ing social information encoded in the ocular regionof the face. In contrast, as previously found by Stoneet al. [20], patients with dorsolateral lesions per-formed similarly to healthy controls on the Faux-pasTest, suggesting that dorsolateral areas are not a corecomponent of the neural circuits involved in mentalstates reasoning. Moreover, in the Faux-pas Testpatients with ventromedial lesions performed poorlycompared to both the healthy controls and patientswith dorsolateral lesions [50]. This supports theclaim, put forward in a number of previous studies[18, 27], that the ventromedial prefrontal cortexplays a key role in the inferential components ofToM competence.

The present findings are also relevant for thecurrent debate on the role of executive deficits onpatient’s difficulties in ToM tasks. No significantrelation was found between the performance on theFaux-pas Test and performance on the executivefunctioning test. In the present study patients withsevere executive deficits were not included, hencethese conclusions may not be generalized to patientswith frontal damage that show substantial limitationsof executive functioning. Nevertheless, this finding isimportant because it provides support for the claimthat not all ToM difficulties in patients with TBI canbe ascribed to weak executive functioning [13, 20].

AND BOL CLE BRZ

GIA RAZ TAM

Figure 2. Maps of lesions of the patients with predominantly focal injury in dorsolateral prefrontal cortex.

984 A. Geraci et al.

This provides further evidence that executive func-tioning and ToM may be dissociated[15, 16, 38, 51]. These results suggest that somepatients with TBI present a selective deficit ininferential reasoning that is independent of executiveabilities. These acquired ToM deficit may sharemany features with the deficits found in other clinicalgroups like persons with autism [52], AspergerSyndrome [20] or schizophrenia [53].

The patients’ difficulties in the Eyes Test are moredifficult to interpret with regard to the domain-specific vs domain-general nature of their underlyingdeficit. The performance of both groups on this testwas poor and, given the patients’ overall good scoresof WCST, one could propose that lower scores onthe Eyes Test do not depend on domain-generalprocessing limitations. However, patients’ scoreswere positively correlated with the WCST scoresand this suggests that lower scores on the Eyes Testcan, at least in part, be the result of limitations inexecutive functioning. This would support the con-clusions proposed by Henry et al. [10] that foundthat verbal fluency, a sensitive measure of executivefunctioning, was significantly associated with EyesTest scores and proposed that patients’ lower scoreson that test depended on weak executive functions.The results of the present study cannot be taken asshowing clearly the involvement of both ventrome-dial and dorsolateral prefrontal areas in the percep-tual components of ToM competence, because theimpairment of either group of patients, or both,could be due to domain general executive dysfunc-tions or performance factors. While the associationof Eyes Test and WCST scores point to a meaning-ful role of domain general resources in the success ofsome ToM tasks in patients with TBI, as suggestedby a number of previous studies [6, 10, 51], patientswith good scores on WCST and poor performanceson the Eyes Test suggest that these difficultiesshould not be construed solely as a result of adomain general deficit related to the executivedemands of the Eyes test.

Theoretically interesting implications may bederived by the dissociation between the Eyes Testand the Faux-pas Test in the patients with damage inthe dorsolateral prefrontal cortex. This dissociationshows that perceptual and inferential aspects are twodifferent components involved in ToM reasoningtasks, as in previous studies [34, 35, 37, 41]. Tager-Flusberg [34] has suggested that theory of mindreasoning can be fractionated into two components:(1) detecting mental states from perceptual andobservable information; and (2) reasoning aboutthose mental states predicting others actions. Thisstudy extents Sabbagh’s [35] research and Tager-Flusbergh’s [34] hypotheses and reveals that in theprefrontal cortex dorsolateral areas are less involved

than ventromedial areas in the inferential compo-nents of theory of mind and also that both areas areinvolved in the attribution of mental states byprocessing of information from the ocular region ofthe face. The reported differences between patientswith ventromedial or dorsolateral lesions cannot beascribed to their demographic variables or globalseverity of injury since the two groups of patients didnot differ significantly on these variables. Theseresults are also coherent with Frith and Frith’s[54, 55] model of the neural bases of ToM.According to such a model, the mindreadingsystem can be analysed into three main components:the medial prefrontal cortex, the temporal poles andthe posterior superior sulcus [56, 57]. The medialprefrontal cortex is the basis of the decouplingmechanism that creates mental states representa-tions as distinct from representations of physicalstates.

A major limitation of the present study is that,given the small sample size and the heterogeneity oftheir severity, it does not exclude the possibility thatthe findings are only coincidental. Laterality provedto be critical in several previous studies [20] and thusthe conclusions reached need to be investigatedfurther in future studies that will be able to comparepatients with left and right lesions in the dorsolateraland ventromedial prefrontal cortex. Also, the controlquestions used for the ToM tasks yielded undesir-able ceiling effects that could be avoided in futurestudies by using more sensitive measures andimproved control tasks [58].

Acknowledgements

We would like to thank all the patients, the staff atthe Ferrara and Padua Hospitals and the healthyparticipants that took part in this study for their kindcollaboration. We are also grateful to RobertoCubelli and Carlo Umilta for their support andhelpful suggestions.

Declaration of interest: The authors report noconflicts of interest. The authors alone are respon-sible for the content and writing of the paper.

References

1. Lezak MD, Howieson DB, Loring DW. Neuropsychologicalassessment. 4th ed. New York: Oxford University Press; 2004.

2. Spatt J, Zebenholzer K, Oder W. Psychosocial long-termoutcome of severe head injury as perceived by patients,relatives, and professionals. Acta Neurologica Scandinavica1997;95:173–179.

3. Dennis M, Purvis K, Barnes MA, Wilkinson M, Winner E.Understanding of literal truth, ironic criticism, and deceptivepraise following childhood head injury. Brain and Language2001;78:1–16.

ToM following TBI 985

4. McDonald S, Pearce S. Clinical insights into pragmatictheory: Frontal lobe deficits and sarcasm. Brain andLanguage 1996;53:81–104.

5. Winner E, Brownell H, Happe F, Blum A, Pincus D.Distinguishing lies from jokes: Theory of mind deficits anddiscourse interpretation in right hemisphere brain-damagedpatients. Brain and Language 1998;62:89–106.

6. Surian L, Siegal M. Sources of performance on theory ofmind tasks in right hemisphere-damaged patients. Brain andLanguage 2001;78:224–232.

7. Santoro J, Spiers M. Social cognitive factors in braininjury—associated personality change. Brain Injury 1994;8:265–276.

8. Eslinger PJ. Neurological and neuropsychological bases ofempathy. European Neurology 1998;39:193–199.

9. Bibby H, McDonald S. Theory of mind after traumatic braininjury. Neuropsychologia 2005;43:99–114.

10. Henry JD, Philips LH, Crawford JR, Ietswaart B, Summer F.Theory of mind following traumatic brain injury: The role ofemotion recognition and executive dysfunction.Neuropsychologia 2006;44:1623–1628.

11. Baron-Cohen S, Wheelwright S, Hill J, Raste Y, Plumb I.The reading the mind in the eyes test revised version: A studywith normal adults, and adults with asperger syndrome andhigh functioning autism. Journal of Child Psychology andPsychiatry 2001;42:241–251.

12. Samson D, Apperly IA, Kathirgamanathan U,Humphreys GW. Seeing it my way: A case of a selectivedeficit in inhibiting self-perspective. Brain 2005;128:1102–1111.

13. Havet–Thomassin V, Allain P, Etcharry Bouyx F, Le Gall D.What about theory of mind after severe brain injury?Brain Injury 2006;20:83–91.

14. Rowe AD, Bullock PR, Polkey CE. ‘Theory of mind’impairments and their relationship to executive function-ing following frontal lobe excisions. Brain 2001;124:600–616.

15. Bach L, Happe F, Fleminger S, Powell J. Theory of mind:Independence of executive function and role of the frontalcortex in acquired brain injury. Cognitive Neuropsychiatry2000;5:175–192.

16. Varley R, Siegal M, Want SC. Severe impairment in grammardoes not preclude theory of mind. Neurocase 2001;7:489–493.

17. Amodio D, Frith C. Meeting of mind: The medial frontalcortex and social cognition. Nature Reviews 2006;7:268–277.

18. Siegal M, Varley R. Neural systems involved in ‘theory ofmind’. Nature Reviews in Neuroscience 2002;3:463–471.

19. Happaney K, Zelazo PD, Stuss DT. Development oforbitofrontal function: Current themes and future directions.Brain & Cognition 2004;55:1–10.

20. Stone VE, Baron-Cohen S, Knight RT. Frontal lobe contri-butions to theory of mind. Journal of Cognitive Neuroscience1998;10:640–656.

21. Bigler ED. The lesion(s) in traumatic brain injury:Implications for clinical neuropsychology. Archives ofClinical Neuropsychology 2001;16:95–131.

22. Gaetz M. The neurophysiology of brain injury. ClinicalNeurophysiology 2004;115:4–18.

23. Gennarelli TA. Mechanisms of brain injury. Journal ofEmergency Medicine 1993;11:5–11.

24. Gennarelli TA, Graham DI. Neuropathology of head inju-ries. Seminar in Clinical Neuropsychiatry 1998;3:160–175.

25. Adams JH, Doyle D, Graham DI, Lawrence AE,McLellan DR, Gennarelli TA, Pastuszko M, Sakamoto T.The contusion index: A reappraisal in human and experi-mental non-missile head injury. Neuropathology and AppliedNeurobiology 1985;1:299–308.

26. Levin HS, High WM, Goethe KE, Sisson RA, Overall JE,Rhoades HM, Eisenberg HM, Kalisky Z, Gary HE.The neurobehavioral rating scale: Assessment of the beha-vioural sequelae of head injury by the clinician. Journal ofNeurology, Neurosurgery and Psychiatry 1987;50:183–193.

27. Fletcher PC, Happe F, Frith U, Baker SC, Dolan RJ,Frackowiak RS, Frith CD. Other minds in the brain: Afunctional imaging study of ‘‘theory of mind’’ in storycomprehension. Cognition 1995;57:109–128.

28. Goel V, Grafman J, Sadato N, Hallett M. Modelling otherminds. Neuroreport 1995;6:1741–1746.

29. Gallagher HL, Happe F, Brunswick N, Fletcher PC, Frith U,Frith CD. Reading the mind in cartoons and stories:An fMRI study of ‘theory of mind’ in verbal and nonverbaltasks. Neuropsychologia 2000;38:11–21.

30. Baron-Cohen S, Goodhart F. The ‘‘seeing leads to knowing’’deficit in autism: The Pratt and Bryant probe. British Journalof Developmental Psychology (A) 1994;12:397–402.

31. Gallagher HL, Frith CD. Functional imaging of ‘theory ofmind’. Trends in Cognitive Sciences 2003;7:77–83.

32. Saxe R, Kanwisher N. People thinking about thinking people:The role of the temporo-parietal junction in ‘theory of mind’.Neuroimage 2003;19:1835–1842.

33. Stuss DT, Gallup Jr GG, Alexander MP. The frontal lobesare necessary for ‘theory of mind’. Brain 2001;124:279–286.

34. Tager-Flusberg H. A re-examination of the theory of mindhypothesis of autism. In: Burack J, Charman T, Yirmiya N,Zelazo P, editors. The development of autism: Perspectivesfrom theory and research. Mahwah, NJ: Lawrence ErlbaumAssociates; 2001.

35. Sabbagh MA. Understanding orbitofrontal contributions totheory-of-mind reasoning: Implications for autism. Brain andCognition 2004;55:209–219.

36. Brothers L, Ring B. A neuroethological framework for therepresentation of minds. Journal of Cognitive Neuroscience1992;4:107–118.

37. Leslie AM. ‘Theory of mind’ as a mechanism of selectiveattention. In: Gazzaniga M, editor. The new cognitiveneuroscience. Cambridge, MA: MIT Press; 2000. pp. 1–10.

38. Surian L, Leslie AM. Competence and performance in falsebelief understanding: A comparison of autistic and normal3-year-old children. British Journal of DevelopmentalPsychology 1999;17:141–155.

39. Hynes CA, Baird AA, Grafton ST. Differential role of theorbital frontal lobe in emotional versus cognitive perspective-taking. Neuropsychologia 2006;44:374–483.

40. McDonald S, Flanagan S. Social perception deficits aftertraumatic brain injury; the interaction between emotionalrecognition, mentalizing ability and social communication.Neuropsychology 2004;18:572–579.

41. Shamay-Tsoory S, Aharon-Peretz J. Dissociable prefrontalnetworks for cognitive and affective theory of mind: A lesionstudy. Neuropsychologia 2007;45:3054–3067.

42. Shamay-Tsoory S, Aharon-Peretz J, Perry D. Two systemsfor empathy: A double dissociation between emotional andcognitive empathy in inferior frontal gyrus versus ventrome-dial prefrontal cortex. Brain 2009;132:617–627.

43. Appollonio I, Leone M, Isella V, Piamarta F, Consoli T,Villa ML, Forapani E, Russo A, Nichelli P. The FrontalAssessment Battery (FAB): Normative values in an Italianpopulation sample. Neurological Science 2005;26:108–116.

44. Hagen C, Malkmus D, Durham P. Levels of cognitivefunctioning. Rehabilitation of the head injured adult:Comprehensive cognitive management. Downey, CA:Professional Staff Association of the Rancho Los AmigosHospital, Inc.; 1979.

45. Damasio H, Damasio A. Lesion analysis in neuropsychology.New York: Oxford University Press; 1989.

986 A. Geraci et al.

46. Serafin M, Surian L. Il Test degli Occhi: Uno strumento pervalutare la ‘teoria della mente’. Giornale Italiano di Psicologia2004;31:213–236.

47. Baron-Cohen S, O’Riordan M, Stone V, Jones R, Plaisted K.Recognition of faux pas by normally developing children andchildren with Asperger syndrome or high-functioning autism.Journal of Autism and Developmental Disorders 1999;29:407–418.

48. Heaton KR, Chelune GJ, Talley Kay GG, Curtiss G.Wisconsin card sorting test manual revised and expanded.P.E.A., Odessa, Florence; 1993.

49. Laiacona M, Inzaghi MG, De Tanti A, Capitani E.Wisconsin card sorting test: A new global score,with Italian norms, and its relationship with the Weiglsorting test. Journal of Neurological Science 2001;21:279–291.

50. Shamay-Tsoory S, Tomer R, Berger B, Goldsher D,Aharon-Peretz J. Impaired ‘‘Affective Theory of Mind’’ isassociated with right Ventromedial Prefrontal Damage.Cognitive and Behavioral Neurology 2005;18:55–67.

51. Channon S, Crawford S. The effects of anterior lesions onperformance on a story comprehension test: Left anterior

impairment on a theory of mind-type task. Neuropsychologia2000;38:1007–1017.

52. Baron-Cohen S, Leslie AM, Frith U. Does the autistic childhave a ‘theory of mind’? Cognition 1985;21:37–46.

53. Mazza M, De Risio A, Surian L, Roncone R, Casacchia M.Selective impairments of theory of mind in people withschizophrenia. Schizophrenia Research 2001;47:299–308.

54. Frith CD, Frith U. Interacting minds—a biological basis.Science 1999;286:1692–1695.

55. Frith U, Frith CD. The biological basis of social interaction.Current Directions in Psychological Science 2001;10:151–155.

56. Jenkins A, Macrae CN, Mitchell JP. Repetition suppressionof ventromedial prefrontal activity during judgments of selfand others. Proceedings of the National Academy of Sciences(USA) 2008;105:4507–4512.

57. Grezes J, Frith C, Passingham RE. Brain mechanisms forinferring deceit in the actions of others. Journal ofNeuroscience 2004;24:5500–5505.

58. Slessor G, Phillips LH, Bull R. Exploring the specificity ofage-related differences in theory of mind tasks. Psychologyand Aging 2006;22:639–643.

ToM following TBI 987