p

s

nAmagsmasbspd(

Exploring the Neural Correlates of Delusionsof ReferenceMahesh Menon, Taylor W. Schmitz, Adam K. Anderson, Ariel Graff, Michele Korostil, David Mamo,Philip Gerretsen, Jean Addington, Gary Remington, and Shitij Kapur

Background: Referential delusions are the most common symptom of schizophrenia and offer an opportunity to examine the neuralcorrelates of delusions because they occur in discrete episodes that can be studied in the scanner. The cortical midline structures (CMS) andsubcortical regions, including the amygdala and striatum, are linked with self-reference in healthy adults. Less is known about the neuralsubstrates of altered self-reference in schizophrenia.

Methods: In this study, patients with schizophrenia experiencing prominent referential delusions (n � 18) and healthy control subjects (n �17) were presented with ambiguous sentences while in the magnetic resonance imaging scanner and asked to rate whether they felt thesentences had been written specifically about them. The sentences were either generic (nonpersonalized) or individually tailored person-alized sentences, designed to induce referential ideation. We hypothesized that both groups would show activity in the CMS, limbic, andstriatal regions and that induced referential ideation would be associated with greater activity in striatal areas in patients with schizophrenia.

Results: A robust main effect of endorsement (endorsed vs. nonendorsed) was observed in the CMS, as well as subcortical regions,including the nucleus accumbens/ventral striatum, amygdala, insula, and midbrain dopamine regions. A group-by-endorsement interac-tion was seen in the medial prefrontal cortex, insula and nucleus accumbens/ventral striatum. Activity in insula and ventral striatum alsocorrelated with the strength of the delusions of reference.

Conclusions: Referential ideation in persons with delusions is associated with heightened CMS, limbic and striatal activity and reduced

differentiation between self- and non-self-relevant information.

ciamtfmnsdcshsd(bbte

eti(r(i(ma(r

Key Words: Delusions of reference, fMRI, neuroimaging, schizo-hrenia, self-reference

J aspers (1) famously asserted, “since time immemorial delu-sions have been taken as the basic characteristic of madness.”Their presence is one of the hallmark criteria in the diagnosis of

chizophrenia (2).One mechanistic theory proposed to explain delusional phe-

omena in schizophrenia is the concept of aberrant salience (3,4).ccording to this theory, the firing of midbrain dopamine neuronsediates the conversion of “cold,” neutral information into person-

lly salient information (4). In this framework, the hyperdopaminer-ic state associated with schizophrenia (5,6) results in instances oftimulus-independent release of dopamine (7), causing the assign-

ent of aberrant salience to neutral objects. Delusions, within suchframework, represent the post hoc explanations of that sense of

alience, and antipsychotics exert their anti-“psychotic” effect bylocking dopamine transmission and attenuating the sensation ofalience (7). The “aberrant salience hypothesis” has received sup-ort from studies examining reward learning (8 –15), aversive con-itioning (16), as well as other probabilistic learning paradigms

17,18).

From the Department of Psychiatry (MM, AG, MK, DM, PG, JA, GR, SK),University of Toronto, Schizophrenia Program (MM, AG, MK, JA, GR, SK)and Multimodal Imaging Group (MM, AG, DM, PG), PET Centre, Centre forAddiction & Mental Health, Department of Psychology (TWS, AKA), Uni-versity of Toronto, Geriatric Mental Health Program (DM, PG), Centre forAddiction & Mental Health, Toronto; Department of Psychiatry (JA), Uni-versity of Calgary, Calgary, Canada; and King’s College London (SK),Institute of Psychiatry, London, United Kingdom.

Address correspondence to Shitij Kapur, M.D., Ph.D., Institute of Psychiatry,Box P001, De Crespigny Park, London, United Kingdom, SE5 8AF; E-mail:Shitij.kapur@kcl.ac.uk.

bReceived Jan 27, 2011; revised May 27, 2011; accepted May 31, 2011.

0006-3223/$36.00doi:10.1016/j.biopsych.2011.05.037

Evidence linking delusions to the aberrant salience hypothesisomes primarily from self-reports of schizophrenia patients regard-

ng delusional beliefs, rather than direct observation of the eventsnd consequent appraisals from which these delusions originate. Aajor obstacle to examining this relationship is that the stimuli that

rigger aberrant salience are inherently unpredictable and there-ore difficult to examine in an experimental setting. One exception

ight be delusions of reference. Referential delusions were origi-ally identified by Kraepelin (19) and are among the most commonymptoms of schizophrenia, present in 67% of persons with theisorder (20). Patients experiencing these symptoms have the per-eption that stimuli, such as something in the newspaper, televi-ion, radio, or overheard statements of strangers walking by, some-ow refers specifically to them. Alternatively, they have thepontaneous feeling that objects are specifically arranged or or-ered so as to convey meaning or messages specifically to them

e.g., the series of license plates of cars parked on the streets) (21). Inoth cases, these instances of heightened self-relevance appear toe discrete moments during which people experience this sensa-

ion of aberrant salience, followed by a post hoc explanation of thatxperience in terms of one’s personal experience.

Although the neural correlates of delusions have not yet beenxamined in detail, an emerging body of literature has examinedhe neural correlates of self-reference in healthy individuals, in tasksn which people make trait attributions about themselves or others22–27), or during subjective preference judgments (28). The brainegions collectively referred to as the cortical midline structuresCMS) have consistently been found to be involved in tasks requir-ng judgments of self-relevance, across various functional domainsverbal, spatial, motor, etc.). The CMS structures include the ventro-

edial (vmPFC) and dorsomedial prefrontal cortex (dmPFC), thenterior cingulate cortex (ACC) and posterior cingulate cortexPCC), and precuneus (29). These structures (medial PFC [mPFC],ostral ACC, and the precuneus) have strong connections with lim-

ic and striatal dopamine regions (30).

BIOL PSYCHIATRY 2011;70:1127–1133© 2011 Society of Biological Psychiatry

rdodthIgRwowsA

T

dgtwcs“omwscnthwItttrsldajisai

Ssscq

srmsy(a

1128 BIOL PSYCHIATRY 2011;70:1127–1133 M. Menon et al.

w

A recent review by Schmitz and Johnson (31) (Figure S1 in Sup-plement 1) proposed the presence of two subsystems involved inappraisal of self-relevant stimuli: 1) a ventral “reflexive” self-referen-tial system implicating a vmPFC–ACC–subcortical ventral pathwayinvolved in the attribution of salience and 2) a dorsal “reflective”self-referential system involving a dmPFC– cortical–subcorticaldorsal pathway that engages introspective processes (e.g., self-reflection, evaluation, recollection), which functions either in con-junction with, or independent of, the former system. Reward-pro-cessing tasks typically engage brain regions associated with theformer system, whereas studies examining self-referential process-ing typically use tasks associated with the latter “reflective” system.These tasks have also been used to examine self-referential infor-mation processing in schizophrenia (32,33) but do not resemble thetypical experience delusions of reference.

In this study, we were interested in capturing the neural activityassociated with the sense of heightened personal self-relevancethat characterizes delusions of reference—in which external stimuliappear to be directly and specifically about the individual. Hence,participants carried out a task in the magnetic resonance imaging(MRI) scanner in which they were presented with ambiguous state-ments written in the third person (e.g., “he likes to drink coffee”) andasked whether they had the feeling that the statement was writtenspecifically about them (rather than simply a judgment of whetherthe statement was true of them). We created a set of “generic”nonpersonalized statements, as well as individualized subject-spe-cific statements. These latter statements were actually “specificallyabout them” and were meant to induce referential ideation.

Because delusions of reference are characterized by the sensa-tion of heightened self-reference to ambiguous (but not actuallyself-referent) stimuli, we hypothesized that 1) patients would en-dorse comparatively more non-self-relevant stimuli—that is, stim-uli outside of their customized set—as being “specifically aboutthem” relative to control subjects; 2) endorsement versus nonen-dorsement of statements as self-relevant would yield significantblood oxygen level– dependent (BOLD) response in the both the“reflective” dorsal and “reflexive” ventral cortico–subcortical ap-praisal pathways; and 3) patients would exhibit greater BOLD acti-vation than control subjects in the dopaminergic components ofthe ventral cortico-subcortical appraisal pathway during stimulusendorsement because of abnormal hyperdopaminergic neuronalresponse in the striatum (5,6).

Methods and Materials

ParticipantsEighteen participants with prominent delusions of reference, as

measured by a “delusions of reference” item score �3 on the Sched-ule for Assessment of Positive Symptoms (SAPS) were recruitedfrom the Schizophrenia Program at the Centre for Addiction &Mental Health. These individuals were compared with 17 controlsubjects with no history of psychiatric illness, recruited via flyersposted within the University of Toronto. Inclusion criteria for pa-tients were as follows: 1) aged 18 to 65 years, 2) English fluency, 3)DSM-IV diagnosis of schizophrenia or schizoaffective disorder, 4)voluntary status, 5) capable of consenting to study participation,and 6) able to undergo functional MRI (fMRI) of approximately 50min duration. Exclusion criteria included 1) serious, unstable medi-cal illness or any concomitant major medical or neurological illness;2) acute suicidal or homicidal ideation; 3) DSM-IV substance depen-dence (except caffeine and nicotine) within 3 months; 4) currentmajor depressive or manic episode; 5) metal implants, cardiac pace-

maker, claustrophobia, or other limitations that would prevent car- s

ww.sobp.org/journal

ying out the MRI component of the study; 6) illegal psychoactiverug use in the past 2 weeks; 7) severe head injury resulting in a lossf consciousness more than 30 min; and 8) SAPS formal thoughtisorder rating �2. Criteria for control participants were similar to

he patient group but included no present or previous psychiatricistory, as measured by the Mini International Neuropsychiatric

nterview (34). Patients and control subjects were matched on age,ender, and estimated premorbid IQ, measured using the Wideange Achievement Test reading subtest (35). Capacity to consentas confirmed for all patient participants with the MacArthur Testf Competence (36), and after complete description of the study,ritten informed consent was obtained from all participants. The

tudy was approved by the Research Ethics Board of the Centre forddiction and Mental Health.

askThe task was designed to evoke sensations similar to ideas or

elusions of reference. During practice trials, participants wereiven the following instructions: “Sometimes people get a feeling

hat something they see or hear (like on TV or when people arealking past them) is actually about them. We want to know what

auses that feeling.” They were then presented with ambiguousentences describing personal characteristics (e.g., “He is lazy” orShe likes to drink coffee” with the pronoun matched to the genderf the subject). Subjects were then told, “As you see these state-ents, we want to know whether you feel that the statement wasritten specifically about you.” Participants used a two-button re-

ponse box in their right hand and pressed “yes” or “no.” Theyarried out three practice trials, and it was clarified that they shouldot press “yes” merely if the sentence was self-descriptive or true of

hem, but rather only if they “had the feeling” that the statementad been written specifically about them, in a manner similar tohat they may have experienced in the situations described earlier.

t was clarified to all participants that this may be uncommon in theask. Following completion of the practice trials, participants wereaken to the MRI scanner where they carried out two runs of theask. In each run, they carried out 60 trials (30 statements, eachepeated twice). Thus, across both runs, participants saw 60 uniquetatements: 20 “neutral” (e.g., “He collects CDs,” “She has a driver’sicense”); 20 “emotionally salient” (e.g., “He was in a horrible acci-ent,” “Everybody hates her”); and 20 “personally salient” individu-lly created subject-specific statements. These personalized sub-

ect-specific statements consisted of information about thendividual’s current life circumstances, hobbies, interests, andymptomatology, all taken from the screening interview conductedseveral weeks before the scanning session, which were meant to

nduce referential ideation.Stimuli were presented using E-Prime software (Psychology

oftware Tools, Pittsburgh, Pennsylvania). Each statement was pre-ented for 5 sec, with a variable interstimulus interval (ISI) of 1.5 to 3ec during which participants saw a fixation cross. Participantsould respond at any point before the presentation of the subse-uent sentence.

After completing the task in the scanner, participants were pre-ented with the stimuli once more and asked to provide specificatings using visual analogue scales for the valence of the state-

ent (rated from 0 � very negative to 10 � very positive), theelf-descriptiveness of the statement (“How true is this statement ofou?” from 0 not true at all to 10 very true), strength of endorsement“How strong was the feeling that the statement was specificallybout you?” from 0 � not very strong to 10 � very strong), and

pecificity of endorsement (“Could this be a statement about peo-

2

2

rm

W

peuroA

wa

R

owtdvaooB

R

pe1p5a

B

pblpsttcpnWrd

T

V

NPGNEVVSSSSRRR

M. Menon et al. BIOL PSYCHIATRY 2011;70:1127–1133 1129

ple in general, or is it specifically about you?” from 0 “a generalstatement about people” to 10 “specifically about me”).

fMRI Data AcquisitionThe MRI scans were acquired by a GE Signa 1.5-T scanner (Gen-

eral Electric, Waukesha, Wisconsin) equipped with a standard headcoil. The subjects used an adjustable mirror located above theireyes to view the back-projected images on a screen placed at thefoot of the scanner bed. In each session, 152 volumes (28 contigu-ous axial 4.4-mm-thick slices) covering the whole brain were ac-quired using a T2*-sensitive spiral sequence (repetition time �

300 msec; echo time � 40 msec; flip angle 85°; matrix 64 � 64; fieldof view 200 � 200 mm). The first three volumes were discarded toallow for T1 equilibrium effects, and the data from the remaining149 volumes were used in the analysis.

For localization purposes, inversion recovery-prepped three-di-mensional fast spoiled gradient recalled T1-weighted anatomic im-ages (120 contiguous axial 1.1 mm thick slices) were acquired (rep-etition time � 12 msec; echo time � 5.4 msec; flip angle 20°; matrix

56 � 256; field of view 200 � 200 mm).

Image PreprocessingThe data were preprocessed and analyzed using SPM5 (The

Wellcome Department of Cognitive Neurology, London, UnitedKingdom; http://www.fil.ion.ucl.ac.uk/spm/software/spm5). Allfunctional images were realigned to the first volume using a six-parameter rigid body transformation, and a mean image was cre-ated. Data from subjects who showed movement of greater thantwo voxels on any axis were discarded. The mean image generatedwas spatially normalized into standard stereotactic space, using theMontreal Neurological Institute echo planar imaging template.Computed transformation parameters were applied to all func-tional images, interpolated to isotropic voxels of 3 mm3, and theesulting images were smoothed using an 8-mm full-width half-

aximum, isotropic Gaussian kernel.

hole-Brain Data AnalysisWe were interested in the difference between the statements

articipants endorsed as being specifically about them versus non-ndorsed statements. We carried out the analysis of the fMRI datasing a 2 group � 2 endorsement (endorsed vs. nonendorsed)

epeated measures analysis of variance. We examined main effectf group, endorsement, and the group � endorsement interaction.

Table 1. Sociodemographic Characteristics of the Final Sample

VariableSchizophrenia

(n � 14/18)Control Subjects

(n � 15/17)

Sex (male/female) 10/4 10/5Age (years) 40.57 (12.8) 35.87 (6.9)School (years) 12.86 (2.48) 16.90 (2.1)a

Premorbid IQ (WRAT) 97.70 (15.3) 107.78 (8.9)SAPS Delusions (Global) 3.8 (1.9) —SAPS Delusions of Reference 4.3 (.9) —SAPS Persecutory Delusions 2.1 (1.6) —SAPS Hallucinations (Global) 2.8 (1.9) —SAPS FTD (Global) .4 (.6) —SANS Affective Flattening 1.2 (.7) —

FTD, Formal Thought Disorder; SANS, Schedule for Assessment of Neg-ative Symptoms; SAPS, Schedule for Assessment of Positive Symptoms;WRAT, Wide Range Achievement Test.

ap � .01.

ll contrasts, unless otherwise specified, were analyzed using a

hole-brain false discovery rate (FDR) corrected p value of � .05nd an extent threshold of 10 voxels.

elationship with SymptomatologyTo examine the relationship between psychotic symptomatol-

gy and the BOLD response during instances of referential ideation,e ran regressions in SPM at the second level, using BOLD response

o endorsed trials as our contrast of interest and the SAPS referentialelusions item as a covariate of interest. These results did not sur-ive whole-brain FDR correction, and therefore a region-of-interestpproach was used, specified using region-of-interest masks basedn the Mawlawi et al. (37) criteria for striatal regions (bilateral masksf caudate, putamen, ventral striatum/nucleus accumbens) androdmann area masks for amygdala, insula, and mPFC.

esults

Data for four participants from the schizophrenia group and twoarticipants from the control group could not be used because ofxcess head motion. The final sample consisted of 14 patients and5 control subjects (see Table 1 for demographic information). Allatients were on atypical antipsychotic medication (olanzapine �; risperidone � 2; quetiapine � 2; clozapine � 5; mean chlorprom-zine equivalent dose � 412.9 mg) (38).

ehavioral ResultsAs can be seen from Table 2 (and Figure S2A in Supplement 1),

atients with delusions of reference endorsed many more items aseing specifically about them. Consistent with the aberrant sa-

ience hypothesis, when we split the endorsed stimuli between theersonalized and nonpersonalized subsets, we found that controlubjects and patients endorsed approximately the same number ofhe personalized statements (Figure S2A in Supplement 1), whereashe patients endorsed significantly more nonpersonal items thanontrol subjects (group � statement type interaction, F25 � 14.64,� .001). This was confirmed by post hoc between groups t test forumber of nonpersonalized items endorsed (t27 � 3.31, p � .01).hen we examined the content of the endorsed nonpersonalized

esponses, we found no within-group differences (i.e., patients en-orse similar number of neutral and emotionally salient “generic”

able 2. Behavioral Results

ariableSchizophrenia

(n � 14)Controls(n � 15)

umber of Items Endorsed 26.64 (7.8) 18.93 (6.9)a

ersonalized Items Endorsed 17.50 (3.8) 16.60 (3.8)eneric Items Endorsed 9.14 (4.9) 3.13 (1.9)a

eutral Generic Items Endorsed 5.08 (2.8) 2.20 (3.6)motional Generic Items Endorsed 4.15 (3.55) .87 (1.0)a

alence Rating (Endorsed Items) 4.76 (1.9) 6.33 (.9)alence Rating (Nonendorsed Items) 3.93 (1.3)c 3.64 (.7)d

elf-descriptiveness (Endorsed Items) 7.50 (1.7) 8.17 (1.4)elf-descriptiveness (Nonendorsed items) 1.62 (.7)d 2.62 (1.1)d

trength of Endorsement 7.50 (1.7) 8.17 (1.4)pecificity of Statement 7.25 (1.6) 7.21 (1.1)eaction Time (Endorsed) 2.20 (.3) 1.96 (.3)eaction Time (Nonendorsed) 2.01 (.3) 1.50 (.1)b

eaction Time Difference (Endorsed –Nonendorsed)

.19 (.2) .46 (.3)a

ap � .01, independent samples t test (patients vs. control subjects).bp � .001, independent samples t test (patients vs. control subjects).c

p � .05, paired samples t test (endorsed vs. nonendorsed).dp � .001, paired samples t test (endorsed vs. nonendorsed).

www.sobp.org/journal

2e

de

mn

Sst

f

vdn

FafF

smssab

pamtan

wnorbppe

R

esbBiptsetttimasoo

D

dAecpscud

oiwmsg

1mt

1130 BIOL PSYCHIATRY 2011;70:1127–1133 M. Menon et al.

w

stimuli as did control subjects). There were between-group differ-ences with patients endorsing more “generic” emotional items(t26 � 3.28, p � .006) and a trend toward more neutral items (t26 �

.39, p � .024) than control subjects, reflecting the fact that patientsndorsed significantly more generic stimuli overall.

When we examined the response latencies for endorsementecisions, we found main effect of group (F25 � 16.27, p � .001),ndorsement (F25 � 61.9, p � .001), as well as a group � endorse-

ment interaction (F25 � 11.08, p � .003), see Figure S2B in Supple-ent 1. Specifically, patients were significantly slower at eliciting

onendorsement responses than control subjects (t27 � 5.75, p �.001), again suggesting greater difficulty in classifying stimuli asbeing non-self-relevant.

On the postscan questionnaire, responses to the question,“How emotional is this statement?” indicated no differencesbetween groups on mean valence ratings (t24 � .78, p � .3).

imilarly, there were no differences between groups for thetrength of endorsement (i.e., “How strong was the feeling thathis statement was written specifically about you?”; (t24 � .35,

p � .7), specificity rating (“Do you think this statement is specif-ically about you?”; t24 � .05, p � .9), or on these measures whenresponses were divided into endorsed and nonendorsed items.For both groups, the endorsed responses were rated as beingmore positive and more self-descriptive than the nonendorseditems (F24 � 368.48, p � .001).

MRI ResultsWhole-Brain Analysis. The group � endorsement analysis of

ariance showed significant main effects of endorsement (en-orsed vs. nonendorsed), a group � endorsement interaction, ando main effect of group.

The main effect of endorsement (endorsed vs. nonendorsed, seeigure 1, Table 3, and Table S1 in Supplement 1) showed significantctivity across most of the brain at our a priori threshold, and there-ore an even more conservative threshold of p � .001 (whole-brain

Figure 1. Main effect of endorsement (endorsed vs. nonendorsed) at athreshold of p � .001 (whole-brain false discovery rate corrected) and a

0-voxel extent threshold) showing significant activity in dorsal and ventraledial prefrontal cortex, anterior cingulate cortex, posterior cingulate cor-

ex, precuneus, caudate, ventral striatum, and midbrain dopamine regions.

DR correction, 10-voxel extent threshold) was used for this analy- p

ww.sobp.org/journal

is. It revealed significant BOLD response in the dorsal and ventraledial prefrontal cortex, ACC, PCC, precuneus, caudate, ventral

triatum/nucleus accumbens (VS/NA), midbrain dopamine regionsubstantia nigra/ventral tegmental area (VTA), and insula. Themygdala was significant at the lower threshold of p � .05 (whole-rain FDR corrected, 10-voxel extent threshold).

The group � endorsement interaction yielded activation inarts of the dMPFC, VS/NA, and the insula (Figure 2, Table 3, andlso Table S2 in Supplement 1). Examining the beta weight esti-ates in all these regions, control subjects showed greater activity

o the endorsed statement relative to the nonendorsed statements,nd the patients showed comparable activity to both endorsed andonendorsed statements.

We also carried out an exploratory analysis to see whether thereas any difference in activity between the personalized and ge-eric items that had been endorsed as self-referential. (This couldnly be done in the patient group.) However, there were no brain

egions where the BOLD activity showed significant differencesetween the two statement types after correcting for multiple com-arisons. This further reinforces the idea that it is the subjectiveerception of self-reference, rather than statement type, that differ-ntiates performance.

elationship with SymptomatologyWe examined the relationship between the BOLD response to

ndorsed items and the intensity of delusions of reference (mea-ured by the SAPS). Although none of the results survived whole-rain correction for multiple comparisons, when we examined theOLD response in our a priori regions of interest, we saw one cluster

n the right VS/NA (Montreal Neurological Institute coordinates ofeak voxel 10, 0, –14, Z � 3.02, 11-voxel cluster), and one cluster in

he right insula (peak 32, 12, 6, Z � 2.94, 49-voxel cluster) thathowed positive correlation with the strength of delusions of refer-nce at a liberal threshold of p � .005 (uncorrected) with an extenthreshold of 10 voxels (Figure S3 and Table S3 in Supplement 1). Toest the specificity of these findings, we ran regressions betweenhe same contrast (endorsed only) and other symptoms, includ-ng the SAPS hallucinations total score and Schedule for Assess-

ent of Negative Symptoms “affective flattening,” and found noctivity in any of our regions of interest. We also examined relation-hip between BOLD response to nonendorsed items and intensityf delusions of reference; we found no significant effects in any ofur regions of interest (Tables S5 and S6 in Supplement 1).

iscussion

Our behavioral results indicate that schizophrenia patients en-orse more ambiguous stimuli as being specifically about them.lthough both patients and control subjects show a similar level ofndorsement to the personalized stimuli (which were actually spe-ifically about them), patients endorsed many more generic, non-ersonalized stimuli as inducing a similar sensation of heightenedelf-reference. The reaction time data indicate that patients andontrol subjects take similar amounts of time to decide which stim-li are self-relevant but that patients are slower at judging stimulieemed non-self-relevant.

In terms of neural activity, our results suggest that experiencesf heightened self-reference of ambiguous stimuli are character-

zed by activity in the CMS, as well as subcortical regions associatedith stimulus salience, such as the insula, VS, and midbrain dopa-inergic regions; these findings are consistent with previous re-

earch on self-reference (29,30,39). Although participants showedreater BOLD response to stimuli evaluated as self-referential (com-

ared with stimuli evaluated as non-self-referential), the magni-

wdt

paopdcedacjawwcsdst

sdenAst

d thr

M. Menon et al. BIOL PSYCHIATRY 2011;70:1127–1133 1131

tude of the difference appears to be smaller in parts of the mPFC,VS/NA, and the insula in patients. The BOLD response in the insulaand VS was also positively correlated with the intensity of referen-tial delusions experienced by the patients.

The behavioral and imaging data thus provide converging evi-dence that in patients, there appears to be reduced differentiationbetween the stimuli considered self-referential and non-self-refer-ential, making it more difficult for them to correctly reject the infor-mation as non-self-relevant. This mirrors findings from reward-learning paradigms that find patients show elevated striatal activityin response to nonreinforced, “neutral” (or nonsalient) stimuli(12,15), consistent with the “aberrant salience” hypothesis. Thesefindings are also consistent with Schmitz and Johnson’s two-pro-cess model of self-referential processing, which suggests that theactivity in VS-limbic-vmPFC pathway is associated with the attribu-tion of salience, and “reflexive” self-reference, which might beheightened in patients. Over time, numerous instances of suchexperiences of heightened self-relevance might increase anxiety ornegative affect and further prime the system in the direction ofincreasing vigilance. The sensations of anxiety and heightened self-reference, in turn, contribute to paranoia. The resulting self-generatedexplanations, based on the individual’s life history and psychodynamicprocesses, are the delusions with which the individual presents.

The finding of systematic hyperactivity in regions involved inself-reference are also consistent with a growing body of researchlooking at activity in the “default mode network” and its link toself-reflective and internally oriented processing (40 – 43). The de-fault mode network has been found to show abnormal connectivity(44) and reduced task-related suppression (45,46) in schizophrenia.We also found activity in regions, such as the superior temporalgyrus (see Tables S2 and S3 in Supplement 1), that have been

Table 3. Main Effect of Condition and Group � Conditi

RegionMNI Co

of Pea

Main Effect of Conditiona

Ventral striatum/nucleus accumbens 10, 1�10, 1

Caudate 12, 1�10, 1

Medial prefrontal cortex �6, 54, 3

Anterior cingulate 6, 1�8, 2

Posterior cingulate/precuneus �6, �4, �

Substantia nigra/VTA �8, �Insula �28, 2Hippocampus �26, �Amygdala �12, �

Group � Condition Interactionb

Ventral striatum/nucleus accumbens �20, 1Putamen �20, 1Medial prefrontal cortex �4, 2

8, 2Posterior cingulate/precuneus 38, �Insula �32, 1

FDR, false discovery rate; MNI, Montreal NeurologicaaAll regions significant at a whole-brain FDR correc

except the amygdala, which was significant at a lower thextent threshold.

bAll regions significant at a whole-brain FDR correcte

implicated in sensations of heightened self-reference associated a

ith mutual eye gaze, in which the patient feels that averted gaze isirect, causing the sensation of “people are looking at me” when

hey might not be (47–50).Neurobiological theories of self, such as the somatic marker hy-

othesis (51), propose that the sense of self lies along a continuum ofwareness, ranging from highly aware explicit (or “reflective”) modesf self-evaluation to more implicit “reflexive” modes of self-referentialrocessing, such as that associated with a “gut feeling” in motivationalisposition. The latter end of this self-awareness continuum has re-eived comparatively less attention in neuroimaging research; how-ver, reflexive self-referential processes guide much of our dailyecision-making behaviors. The paradigm used in this study aimed toddress this issue. Unlike other studies, which looked at the neuralorrelates of “reflective” self-referential processes by asking people toudge whether a presented statement was “true of them,” this studysked subjects to endorse the statement only if they felt it had beenritten specifically about them, irrespective of whether they thought itas self-descriptive, thus engaging both reflective and reflexive pro-

esses. Interestingly, a number of patients following the scanning ses-ion reported that the experience was similar to the sensation evokeduring delusional ideation, necessitating debriefing after the scanningession to reassure participants that the statements were taken fromheir prior screening interview.

All participants were on medication, which might have hadome effect on neural activity. Despite this, all patients showedelusions of reference and had elevated striatal activity, as hypoth-sized a priori. We would therefore predict effects of a larger mag-itude in never-medicated patients with delusions of reference.mbiguous statements may be less naturalistic than videotaped

cenes as used by Park et al. (52); however, our method allowed uso tailor the ambiguous stimuli for each participant. Thus, we were

eraction in A Priori Regions of Interest

atesel F Value Z Value

Cluster Size(Voxels)

29.82 4.71 2927.06 4.52 3437.29 5.17 10228.83 4.64 10838.89 5.26 168737.92 5.2140.03 5.32 75336.17 5.11

6 59.04 6.18 4226 44.12 5.5310 41.82 5.42 402

36.20 5.11 6310 37.33 5.18 50

16 29.21 4.67 21

22.13 4.13 2622.16 4.13 7227.71 4.57 18024.07 4.29

4 22.84 4.19 5818.17 3.77 28

tute; VTA, ventral tegmental area.reshold of p � .001 with a 10-voxel extent threshold,ld of p � .05 whole-brain FDR corrected and a 10-voxel

eshold of p � .05 with a 10 voxel extent threshold.

on Int

ordink Vox

0, 42, 22, 80, 106, 182, 404, 380, 3466, 370, 322, �

0, �428, �6, �

2, �64, �22, 422, 4468, 4

2, 4

l Instited thresho

ble to create experiences akin to ideas of reference as noted by the

www.sobp.org/journal

aaCgGCV

bttrrBiAaANpo

vd

1132 BIOL PSYCHIATRY 2011;70:1127–1133 M. Menon et al.

w

participants themselves. Follow-up studies are required to examinethe specificity of the pattern of deficits seen here with the use of aclinical control group (patients without delusions of reference), aswell as to examine the relationship between these instances ofaberrant salience and other abnormalities in social cognition. Lon-gitudinal studies would be beneficial to determine whether thepatterns of activity change as delusions fluctuate.

Our results demonstrate for the first time that sensations ofheightened self-relevance to ambiguous stimuli (a process thoughtto reflect the underpinnings of delusions of reference) are associ-ated with increased BOLD response in a distributed network, likelyinvolving numerous neurotransmitters, in parts of the mPFC as wellas subcortical limbic and midbrain dopaminergic regions. Althoughit is worth stressing that these changes in BOLD response may alsoreflect downstream consequences of misattribution or other psy-chological mechanisms not reflected in BOLD activity, we speculatethat the differential patterns of activity in the dmPFC, insula andstriatal regions may underlie the phenomenological differencesexperienced between patients and control subjects and suggestthat the decisions of self-reference and specificity of ambiguousstimuli might be the result of differential inputs from the “reflexive”and “reflective” networks in patients versus control subjects.

Following a continuum model approach, these same processesmay, to varying degrees, also underlie the sensation of heightenedself-consciousness experienced as part of “normal” experience inmoments of heightened affect and play a role in other psychiatricconditions, including social anxiety and body dysmorphic disorder.A better understanding of cognitive and neural processes involvedin delusions may also allow us to develop treatment options, suchas transcranial magnetic stimulation or attentional bias modifica-

Figure 2. (A) Group � condition interaction (p � .05, whole-brain false discoentral striatum/nucleus accumbens (VS/NA), insula, and dorsomedial prefromPFC (�4, 22, 42) and the VS/NA (�20, 12, �6). L, left; R, right.

tion, for symptom amelioration in treatment-refractory patients.

ww.sobp.org/journal

We thank Penny Barsoum, Heidi Marcon, Wanna Mar, Sofia Raitsin,nd Fernando Caravaggio for their assistance with data collection anddministrative support. The study was funded by a Canada Researchhair to SK. Shitij Kapur’s contributions were also supported by Strate-ic funding from the Medical Research Council UK (Grant No.0701748 [85253]). A portion of this work was presented at the 1stonference of the Schizophrenia International Research Society inenice, Italy, June 21–25, 2008.

Dr. Graff has received professional services compensation from Ab-ott Laboratories, Janssen-Cilag, Eli Lilly, and Geodon Richter PLC in

he past 3 years. Dr. Mamo has received grants or consultant fees fromhe Bristol-Myers Squibb and Pfizer and has received speaker’s hono-aria from AstraZeneca in the past 3 years. Dr. Remington has receivedesearch support from Novartis Canada, Medicure Inc., and Neurocrineiosciences. He has also received speaker’s honoraria from Novartis

n the past 3 years. Dr. Kapur has received research support fromstraZeneca, Bristol-Myers Squibb, and GlaxoSmithKline. He has alsocted as a consultant or received speaker’s honoraria fromstraZeneca, Bioline, Bristol-Myers Squibb, Eli Lilly, Janssen, Lundbeck,euroSearch, Otsuka, Pfizer, Roche, Servier, and Solvay Wyeth in theast 3 years. All other authors report no biomedical financial interestsr potential conflicts of interest.

Supplementary material cited in this article is available online.

1. Jaspers K (1913): Allgemeine psychopathologie. Berlin: Springer.2. American Psychiatric Association (1994): Diagnostic and Statistical Man-

ual of Mental Disorders (4th ed). Washington, DC: American PsychiatricAssociation.

3. Berridge KC, Robinson TE (1998): What is the role of dopamine in reward:

ate corrected, 10-voxel extent threshold) showing significant activity in thecortex (dmPFC). (B) Beta weight estimates across conditions in voxels in the

very rntal

Hedonic impact, reward learning, or incentive salience? Brain Res Rev28:309 –369.

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

M. Menon et al. BIOL PSYCHIATRY 2011;70:1127–1133 1133

4. Kapur S (2003): Psychosis as a state of aberrant salience: A frameworklinking biology, phenomenology, and pharmacology in schizophrenia.Am J Psychiatry 160:13–23.

5. Abi-Dargham A, Rodenhiser J, Printz D, Zea-Ponce Y, Gil R, Kegeles LS, etal. (2000): Increased baseline occupancy of D2 receptors by dopaminein schizophrenia. Proc Natl Acad Sci U S A 97:8104 – 8109.

6. Laruelle M, Abi-Dargham A (1999): Dopamine as the wind of the psy-chotic fire: New evidence from brain imaging studies. J Psychopharma-col 13:358 –371.

7. Kapur S (2004): How antipsychotics become anti-“psychotic”—fromdopamine to salience to psychosis. Trends Pharmacol Sci 25:402– 406.

8. Juckel G, Schlagenhauf F, Koslowski M, Filonov D, Wustenberg T,Villringer A, et al. (2006): Dysfunction of ventral striatal reward predic-tion in schizophrenic patients treated with typical, not atypical, neuro-leptics. Psychopharmacology 187:222–228.

9. Juckel G, Schlagenhauf F, Koslowski M, Filonow D, Knutson B, Wrase J, etal. (2005): Ventral striatum activation by typical versus atypical neuro-leptics in a fMRI paradigm with monetary incentive stimuli. Pharma-copsychiatry 38:252–252.

10. Juckel G, Schlagenhauf F, Koslowski M, Wustenberg T, Villringer A, Knut-son B, et al. (2006): Dysfunction of ventral striatal reward prediction inschizophrenia. Neuroimage 29:409 – 416.

11. Schlagenhauf F, Juckel G, Koslowski M, Kahnt T, Knutson B, Dembler T, et al.(2008): Reward system activation in schizophrenic patients switched fromtypical neuroleptics to olanzapine. Psychopharmacology 196:673–684.

12. Jensen J, Willeit M, Zipursky RB, Savina I, Smith AJ, Menon M, et al. (2007):The formation of abnormal associations in schizophrenia: Neural andbehavioral evidence. Neuropsychopharmacology 33:473– 479.

13. Roiser JP, Stephan KE, den Ouden HE, Barnes TR, Friston KJ, Joyce EM(2009): Do patients with schizophrenia exhibit aberrant salience? Psy-chol Med 39:199 –209.

14. Roiser JP, Stephan KE, den Ouden HE, Friston KJ, Joyce EM (2010):Adaptive and aberrant reward prediction signals in the human brain.Neuroimage 50:657– 664.

15. Murray G, Corlett P, Clark L, Pessiglione M, Blackwell A, Honey G, et al.(2008): Substantia nigra/ventral tegmental reward prediction error dis-ruption in psychosis. Mol Psychiatry 13:239 –276.

16. Romaniuk L, Honey G, King J, Whalley H, McIntosh A, Levita L, et al.(2010): Midbrain activation during pavlovian conditioning and delu-sional symptoms in schizophrenia. Arch Gen Psychiatry 67:1246 –1254.

17. Corlett P, Murray G, Honey G, Aitken M, Shanks D, Robbins T, et al. (2007):Disrupted prediction-error signal in psychosis: Evidence for an associa-tive account of delusions. Brain 130:2387–2400.

18. Heinz A, Schlagenhauf F (2010): Dopaminergic dysfunction in schizo-phrenia: salience attribution revisited. Schizophr Bull 36:472– 485.

19. Kraepelin E (1906): Uber Sprachstörungen im Traume. Leipzig: Engel-mann.

20. World_Health_Organization (1973): Report of the International PilotStudy of Schizophrenia. Geneva: WHO Publications.

21. Startup M, Startup S (2005): On two kinds of delusion of reference.Psychiatr Res 137:87–92.

22. Johnson SC, Baxter LC, Wilder LS, Pipe JG, Heiserman JE, Prigatano GP(2002): Neural correlates of self-reflection. Brain 125:1808 –1814.

23. Schmitz TW, Kawahara-Baccus TN, Johnson SC (2004): Metacognitiveevaluation, self-relevance, and the right prefrontal cortex. Neuroimage22:941–947.

24. Fossati P, Hevenor SJ, Graham SJ, Grady C, Keightley ML, Craik F, et al.(2003): In search of the emotional self: An fMRI study using positive andnegative emotional words. Am J Psychiatry 160:1938 –1945.

25. Kelley WM, Macrae CN, Wyland CL, Caglar S, Inati S, Heatherton TF(2002): Finding the self? An event-related fMRI study. J Cogn Neurosci14:785–794.

26. Macrae CN, Moran JM, Heatherton TF, Banfield JF, Kelley WM (2004): Medialprefrontal activity predicts memory for self. Cereb Cortex 14:647–654.

27. Mitchell JP, Banaji MR, Macrae CN (2005): The link between social cog-nition and self-referential thought in the medial prefrontal cortex. JCogn Neurosci 17:1306 –1315.

28. Johnson SC, Schmitz TW, Kawahara-Baccus TN, Rowley HA, AlexanderAL, Lee JH, et al. (2005): The cerebral response during subjective choicewith and without self-reference. J Cogn Neurosci 17:1897–1906.

29. Northoff G, Heinzel A, Greck M, Bennpohl F, Dobrowolny H, Panksepp J(2006): Self-referential processing in our brain—A meta-analysis of im-

aging studies on the self. Neuroimage 31:440 – 457.

0. Northoff G, Bermpohl F (2004): Cortical midline structures and the self.Trends Cogn Sci 8:102–107.

1. Schmitz TW, Johnson SC (2007): Relevance to self: A brief review andframework of neural systems underlying appraisal. Neurosci BiobehavRev 31:585–596.

2. Murphy ER, Brent BK, Benton M, Pruitt P, Diwadkar V, Rajarethinam RP, etal. (2010): Differential processing of metacognitive evaluation and theneural circuitry of the self and others in schizophrenia: A pilot study.Schizophr Res 116:252–258.

3. Blackwood NJ, Bentall RP, Ffytche DH, Simmons A, Murray RM, HowardRJ (2004): Persecutory delusions and the determination of self-rele-vance: An fMRI investigation. Psychol Med 34:591–596.

4. Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, etal. (1998): The Mini-International Neuropsychiatric Interview (M.I.N.I.):The development and validation of a structured diagnostic psychiatricinterview for DSM-IV and ICD-10. J Clin Psychiatry 59(suppl 20):22–33;quiz 34 –57.

5. Wilkenson GS, Jastak J (1993): Wide Range Achievement Test—Third Edi-tion (WRAT-3). Wilmington, DE: Wide Range.

6. Appelbaum PS, Grisso T (1995): The Macarthur Treatment CompetenceStudy. I. Mental Illness and Competence to Consent to Treatment. LawHum Behav 19:105–126.

7. Mawlawi O, Martinez D, Slifstein M, Broft A, Chatterjee R, Hwang DR, etal. (2001): Imaging human mesolimbic dopamine transmission withpositron emission tomography: I. Accuracy and precision of D(2) recep-tor parameter measurements in ventral striatum. J Cereb Blood FlowMetab 21:1034 –1057.

8. Woods SW (2003): Chlorpromazine equivalent doses for the newer atyp-ical antipsychotics. J Clin Psychiatry 64:663– 667.

9. van der Meer L, Costafreda S, Aleman A, David AS (2010): Self-reflectionand the brain: A theoretical review and meta-analysis of neuroimagingstudies with implications for schizophrenia. Neurosci Biobehav Rev 34:935–946.

0. Schneider F, Bermpohl F, Heinzel A, Rotte M, Walter M, Tempelmann C,et al. (2008): The resting brain and our self: Self relatedness modulatesresting state neural activity in cortical midline structures. Neuroscience157:120 –131.

1. Gusnard D, Akbudak E, Shulman G, Raichle ME (2001): Role of medialprefrontal cortex in a default mode of brain function. NeuroImage 13:S414 –S414.

2. Gusnard DA, Akbudak E, Shulman GL, Raichle ME (2001): Medial prefron-tal cortex and self-referential mental activity: Relation to a default modeof brain function. Proc Natl Acad Sci U S A 98:4259 – 4264.

3. Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, ShulmanGL (2001): A default mode of brain function. Proc Natl Acad Sci U S A98:676 – 682.

4. Salvador R, Sarró S, Gomar JJ, Ortiz-Gil J, Vila F, Capdevila A, et al. (2010):Overall brain connectivity maps show cortico-subcortical abnormalitiesin schizophrenia. Hum Brain Mapp 31:2003–2014.

5. Seidman LJ, Whitfield-Gabrieli S, Thermenos HW, Rosso I, Makris N,Milanovic S, et al. (2010): Default network and medial prefrontal cortexdysfunction in schizophrenia and in first degree relatives of personswith schizophrenia. Schizophr Res 117:170 –171.

6. Whitfield-Gabrieli S, Thermenos HW, Milanovic S, Tsuang MT, FaraoneSV, McCarley RW, et al. (2009): Hyperactivity and hyperconnectivity ofthe default network in schizophrenia and in first-degree relatives per-sons with schizophrenia. Proc Natl Acad Sci U S A 106:1279 –1284.

7. Hooker C, Park S (2005): You must be looking at me: The nature of gazeperception in schizophrenia patients. Cogn Neuropsychiatry 10:327–345.

8. Rosse R, Kendrick K, Wyatt R, Isaac A, Deutsch S (1994): Gaze discrimina-tion in patients with schizophrenia: Preliminary report. Am J Psychiatry151:919 –921.

9. Menon M, Raitsin S, Graff A, Hooker C, Kapur S (2010): Exploring theneural basis of gaze perception errors in schizophrenia. Schizophr Res117:354.

0. Pelphrey KA, Viola RJ, McCarthy G (2004): When strangers pass. PsycholSci 15:598 – 603.

1. Damasio AR (1994): Descartes’ Error: Emotion, Reason, and the HumanBrain. New York: G. P. Putnam’s Sons.

2. Park IH, Ku J, Lee H, Kim SY, Kim SI, Yoon KJ, et al. (2010): Disrupted

theory of mind network processing in response to idea of referenceevocation in schizophrenia. Acta Psychiatr Scand 123:43–54.

www.sobp.org/journal