Ah

LNa

b

a

ARRAA

KASMHD

1

m&1fMpifgastA

v

0d

Neuropsychologia 49 (2011) 805–812

Contents lists available at ScienceDirect

Neuropsychologia

journa l homepage: www.e lsev ier .com/ locate /neuropsychologia

ttentional modulation of external speech attribution in patients withallucinations and delusions

ana Marija Ilankovica,b,∗, Paul P. Allenb, Rolf Engela, Joseph Kambeitzb, Michael Riedela,orbert Müllera, Kristina Hennig-Fasta

Department of Psychiatry, Ludwig-Maximilians University, Nussbaumstr. 7, 80336 Munich, GermanyInstitute of Psychiatry, Psychosis Clinical Academic Group, P.O. Box 67, Denmark Hill, London SE5 8 AF, UK

r t i c l e i n f o

rticle history:eceived 21 January 2010eceived in revised form 8 January 2011ccepted 10 January 2011vailable online 15 January 2011

eywords:ttentionpeechisattributionallucinations

a b s t r a c t

A range of psychological theories have been proposed to account for the experience of auditory hallucina-tions and delusions in schizophrenic patients. Most influential theories are those implicating the defectiveself-monitoring of inner speech. Some recent studies measured response bias independently of self-monitoring and found the results inconsistent with the defective self-monitoring model, but explainedby an externalizing response bias. We aimed to investigate the role of attentional bias in external mis-attribution of source by modulating participant’s endogenous expectancies. Comparisons were madebetween patients with paranoid schizophrenia (N = 23) and matched healthy controls (N = 23) who par-ticipated in two different versions of an audio–visual task, which differed based upon level of the cuepredictiveness. The acoustic characteristic of voice was altered in half of the trials by shifting the pitch(distortion). Participants passively listened to recordings of single adjectives spoken in their own and

elusions another person’s voice (alien) preceded by their own or another person’s (alien) face and made self/nonself judgments about the source. The patients showed increased error rates comparing to controls, whenlistening to the distorted self spoken words, misidentifying their own speech as produced by others.Importantly, patients made significantly more errors across all the invalid cue conditions. This suggestsnot only the presence of pathological misattribution bias, but also an inadequate balance between top-down and bottom-up attentional processes in the patients, which could be responsible for misattribution

mat

of the ambiguous sensory

. Introduction

Auditory verbal hallucinations (AVH) are one of the most com-on symptoms experienced by patients with schizophrenia (SladeBentall, 1988) and are usually associated with delusions (Liddle,

987). A range of cognitive models have been proposed to accountor AVH and delusions in schizophrenic patients (Allen, Aleman, &

cGuire, 2007; Seal, Aleman, & McGuire, 2004). One such modelroposes that AVH arise because of the defective monitoring of

nner speech (Frith, 1987, 1992). According to this model AVH resultrom defective monitoring of thoughts/inner speech, as they areenerated, leading to misidentification of self-generated thoughts

s external ,alien’ voices. Frith’s model is supported by data fromtudies that have engaged verbal self-monitoring by experimen-ally manipulating auditory verbal feedback while patients withVH spoke aloud (Cahill, Silbersweig, & Frith, 1996; Johns et al.,

∗ Corresponding author at: Department of Psychiatry, Ludwig-Maximilians Uni-ersity, Nussbaumstr. 7, 80336 Munich, Germany.

E-mail address: lana.ilankovic@kcl.ac.uk (L.M. Ilankovic).

028-3932/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.oi:10.1016/j.neuropsychologia.2011.01.016

erial.© 2011 Elsevier Ltd. All rights reserved.

2001). Altering the acoustic characteristics of their speech intro-duced a disparity between what subjects expected to hear andwhat they actually perceived. In both studies patients with hallu-cinations and delusions were more likely than controls to makemisattribution errors and misidentify their own speech as alienwhen it is distorted. However, a later study found that the mis-attribution of self speech to an external source was not specific topatients with AVH and that patients with delusions but no hallu-cinations also demonstrate a significant tendency to misattributetheir own speech (Johns, Gregg, Allen, & McGuire, 2006). The speci-ficity of such a deficit to AVH is therefore equivocal as most studiesreport that patients with delusions also tend to make external mis-attributions when listening to their own distorted speech (Allenet al., 2007). However, the tendency to make misattribution errorsin patients currently experiencing AVH in the context of an affec-tive psychosis was not identical to the one that was seen in patients

experiencing AVH in the context of schizophrenia. Namely, they didnot misattribute their own speech when it was distorted, but theytended to be unsure about the speech source. Hence, the misat-tribution bias was not observed in patients who had a history ofhallucinations but were hallucination-free for a month.

8 psych

cscSBt(t(BSe2oorgsaifirtR2fwbmail

sttRaADardtdrF

pdRctbk1ptb(pedAa

06 L.M. Ilankovic et al. / Neuro

It has also been suggested that positive symptoms, e.g. hallu-inations and delusions, are related to a deficit in reality and/orource monitoring leading to confusion between imagined and per-eived items (Bentall, Baker, & Havers, 1991; Blakemore, Smith,teel, Johnstone, & Frith, 2000; Blakemore, Wolpert, & Frith, 2002;rébion et al., 2000). Much of the experimental evidence is consis-ent with this idea as patients with hallucinations and delusionsH/D patients) in general tend to misattribute items they producedhemselves to an external source on a variety of experimental tasksAllen et al., 2004; Bentall et al., 1991, 1994; Brébion et al., 2000;rébion, Gorman, Amador, Malaspina, & Sharif, 2002; Brébion,mith, Gorman, & Amador, 1997; Johns et al., 2001, 2006). Allent al. (2004) adapted the paradigm used by Johns et al. (2001,006), such that participants made judgments about the sourcef pre-recorded speech rather than speech that was generatednline. As participants were simply required to indicate when theyecognized their own voice, the putative self-monitoring of self-enerated speech was bypassed. Despite this H/D patients weretill more likely to claim their own distorted voice was that ofnother person than patients without hallucinations (and signif-cantly lower levels of delusions) and healthy volunteers. Thisnding suggests that the external misattribution of source mayeflect impairment in not only verbal self-monitoring, but also inhe appraisal of ambiguous sensory material (Allen et al., 2007).ecent event-related brain potentials studies (Ford & Mathalon,005; Maldonado-Heinks et al., 2007) found that hallucinators per-ormed poorly both during self-monitoring their own speech andhen the putative self-monitoring of self-generated speech was

ypassed. Thus they suggested that the self-monitoring of speechay have failed to develop appropriately resulting in uncertainty

bout the source of current perceptions. According to these stud-es, misattributions therefore may result from a “coping” strategyearned over time to navigate through life.

We argue that misattribution of speech in patients with positiveymptoms suggests more than a “coping strategy”. It is our viewhat impairment of attention, that has been cited as a fundamen-al clinical feature of schizophrenia (Kraepelin, 1919; Posner, Early,eiman, Pardo, & Dhawan, 1988) plays an important role in the mis-ttribution of external speech in patients with positive symptoms.bnormalities of visual (Danckert, Saoud, & Maruff, 2003; Maruff,anckert, Pantelis, & Currie, 1998; Sereno & Holzman, 1996) anduditory (Mathalon et al., 2004) attention processing are frequentlyeported in patients with schizophrenia. In particular, attentioneficit had been previously shown to correlate with positive symp-oms of schizophrenia (Rocca et al., 2006). Specifically, cognitiveeficits in both bottom-up and top-down processing have beeneported in patients with schizophrenia (Fuller et al., 2006; Gold,uller, Robinson, Braun, & Luck, 2007).

Interestingly, the neural deficits underlying impaired cognitiveerformance support bottom-up dysregulation more than the top-own cortical dysregulation (Butler et al., 2007; Leavitt, Molholm,itter, Shpaner, & Foxe, 2007). Bottom-up or exogenous attentionalontrol is stimulus driven, i.e. attention is spontaneously orientedowards an oncoming stimulus. Top-down or endogenous control,y contrast, is intentionally and cognitively driven, i.e. directed bynowledge, expectation and current goals (Desimone & Duncan,995; Egeth & Yantis, 1997). Importantly, top-down and bottom-uprocesses represent overlapping organizational principles ratherhan dichotomous constructs, and in most situations, top-down andottom-up processes interact to optimize attentional performanceSarter, Givens, & Bruno, 2001). Deficits in magnocelullar visual

rocessing in schizophrenia suggest a dysfunction even within thearly regions of the visual pathway, which may lead to bottom-upriven dysregulation of higher cortical function (Butler et al., 2007).recent study in auditory processing in schizophrenia reflects

nalogous findings that the earliest afferent input to the primary

ologia 49 (2011) 805–812

auditory cortex, which arrives from subcortical regions, alreadyshow evidence of dysfunction in patients with schizophrenia con-sistent with a bottom-up model of auditory processing deficits(Leavitt et al., 2007). Although there have been few investigationsof top-down and bottom-up attentional processes in patients withpositive symptoms, one study by Aleman, Böcker, Hijman, De Haan,and Kahn (2003) observed a correlation between the severity ofhallucination ratings and top-down influences on auditory percep-tion. This is consistent with the notion that hallucinations mayresult from an increased influence of top-down sensory expecta-tions on conscious perception (Aleman et al., 2003). It has beenalso shown in previous studies that when bottom-up and top-downprocesses conflict, patients with schizophrenia exhibit significantlyworse performance than healthy controls (Maruff et al., 1998).

The aim of the present study is to examine the effect of atten-tional modulation on source decisions for pre-recorded speech inpatients with schizophrenia and in healthy control group. In theprevious versions of the paradigm (Allen et al., 2004; Johns et al.,2001) distortion of participant’s voice was used to manipulate whatthey expected to hear and what they actually perceived. As well asdistortion of participants voice we also manipulated the validity ofcues in a speech appraisal task (Allen et al., 2004) in order to modu-late the participant’s expectancies about the source of speech thatthey heard.

Our hypothesis were:

1) when top-down (cue preceding the voice stimuli) and bottom-up (source of the voice stimuli) mechanisms are placed inconflict (e.g. during invalid trials) patients with hallucinationsand delusion will demonstrate impaired performance relative tocontrols. This is because patients with hallucinations and delu-sions are allowing top-down information to guide them at theexpense of bottom-up information,

2) the patients will demonstrate a significant externalizingresponse bias (misidentifying their own speech as alien) thatwill be particularly evident on invalid cue trials when their ownvoice is preceded by an alien cue. Misattribution errors shouldbe more evident in the predictive cue condition (80% valid cuesand 20% invalid cues), as the high informative power of the cueshould guide source attributions stronger than the unpredictivecue condition (50% valid cues and 50% invalid cues),

3) finally, in the patient group, errors associated with invalid trialsshould correlate with severity of positive symptoms.

2. Method

2.1. Participants

Twenty-three patients who met ICD-10-GM (Band 2; World HealthOrganization, 2008) criteria for paranoid schizophrenia recruited from a poolof volunteers from the psychiatric hospital in Munich (Department of Psychiatryand Psychotherapy, Ludwig-Maximilians-University) and twenty-three healthyGerman-speaking demographically matched (gender, age, education and IQ)volunteers with no history of psychiatric illness recruited through advertisementwere included in this study (see Table 1). All the patients were inpatients and theywere tested 2–6 weeks (M = 19.1 days, SD = 12.3) after the admission to the hospital.They were all symptomatically stable at the time of testing and still reportingpositive symptoms despite medication. They were on regular, stable doses ofantipsychotic medication, either typical (19) or atypical (2) or both (2). In additionto antipsychotic medication, some patients were receiving Benzodiazepine forthe treatment of anxiety (7) and Biperiden to attenuate Parkinson symptoms (4).Patients’ symptoms were assessed on the day of testing using the scale for theassessment of positive symptoms (SAPS; Andreasen, 1984), the scale for assessmentof negative symptoms (SANS; Andreasen, 1984) and the psychotic symptom rating

scale (PSYRATS; Haddock et al., 1999). Patients were selected for inclusion ifthey scored 3 or more on SAPS global hallucinations or delusions subscales (seeTable 1). The exclusion criteria for all the subjects was the following: presenceof neurological diseases, current or recent alcohol abuse or drug addiction. Onlyparticipants with an greater IQ than 85 as determined by the Wortschatztest(Schmidt & Metzler, 1992) were included. After reading a complete description of

L.M. Ilankovic et al. / Neuropsychologia 49 (2011) 805–812 807

Table 1Socio-demographic data of participants.

Patients with hallucinationsand delusions (N = 23)M (SD)

Controls (N = 23)M (SD)

Analysis

Age in years 33.26 (9.30) 33.78 (9.26) t = −.199, p = 0.843Premorbid IQ 104.87 (13.28) 110.65 (12.01) t = 1.48, p = 0.145Years of education 11.91 (2.42) 12.48 (2.46) t = .783, p = 438Gender ratio M/F 11/12 11/12Symptom ratingsAge of onset 29.83 (7.3)Duration of illness 4.70 (5.1)Auditory hallucinations 3.69 (3.8)Non-auditory hallucinations 0.96 (2.3)Delusions 9 (4.2)Other positive symptomsa 6.91 (3.3)Negative symptomsb 8.34 (4.2)Attentional problems 1.57 (1.2)PSYRATS auditory hallucinations 9.26 (10.6)

SANSaffect

ta

2

2

cov

2

ttawatsmhsastbpe

2

fdgr(

2

arscatdi

licic

PSYRATS delusions 13.52 (3.8)

a Mean of global scores for bizarre behaviour and formal thought disorder; SAPS/b Mean of global scores for alogia, anhedonia, inappropriate affect, avolation and

he study all participants gave written informed consent to participate, which waspproved by the local research ethics committee.

.2. Materials

.2.1. Word listsA list of 192 personal adjectives was used. The word list in the present study

onsisted of the same adjectives used by Allen et al. (2004) derived from the studyf Johns et al. (2001). These words were translated into German from the originalersion in English language.

.2.2. Auditory stimuli(1) 96 words that were the subjects own voice (i.e. self speech) and (2) 96 words

hat were a standardized unfamiliar male or female voice that were used to replacehe voice of the subject on half of the trials (alien speech). The speech was recordednd distorted by Audacity 1.2.6. (http://audacity.sourceforge.net). The alien voiceas provided by 2 members of staff at the Psychiatric Hospital in Munich, one male

nd one female native German speaker. The alien pictures were also provided byhe same male and female stuff members, that provided their alien voice to thetudy. We used gender matched images and voices throughout the task. The equip-ent we used consisted of a PC computer (Fujitsu Siemens, P19-2), Sony stereo

eadphones (MDR-XD300) and the microphone (UHER; MDR-XD300). Presentationoftware 10.3. (http://www.neurobs.com) was used to present all auditory stimulind record the responses of the subjects. The volume levels of both the participant’speech and the alien speech were kept as close as possible by normalizing the volumeo ensure that differential volume could not subsequently be used to discriminateetween the two types of speech. The participants sat 60 cm in front of a 19 in. com-uter screen, in a semi-darkened room designated for neuropsychological tests andxperiments.

.2.3. Visual stimuliVisual cues consisted of subjects’ portrait picture (200 × 267 pixel, in

ull color), taken by an Olympus digital camera (D-425: 4.0 megapixel)isplayed vertically in the center of the computer screen on a gray back-round, preceding the upcoming auditory stimuli. Pictures were matched withegard to color intensity, brightness and contrast by using Adobe Photoshophttp://www.adobe.com/de/products/photoshop/)

.3. Design

There were two levels of validity (valid, invalid), two sources of speech (self,lien), two levels of distortion (0, −4 semitones) and two groups (control, patients),esulting in a 2 × 2 × 2 × 2 factorial design. Self speech preceded by a picture of theubject’s own face (self picture cue) and alien speech preceded by an alien pictureue, respectively, constituted valid trials, whereas the self speech preceded by anlien face and alien speech preceded by a subject’s own picture, represented invalidrials. One experimental run consisted of 50% valid cues and 50% invalid cues (unpre-ictive cue condition). The other experimental run consisted of 80% valid and 20%

nvalid cues (predictive cue condition).

We introduced the unpredictive cue condition to be able to assess the task ‘base-

ine’, where participants performance was neither impaired nor facilitated by thenformation provided by the cue, in contrast to predictive cue condition. Each runonsisted of 8 possible combinations of cues (self, alien) and voices (self, alien) mak-ng a total of 192 trials (24 for each experimental condition). In the unpredictive cueondition 12 valid and 12 invalid cue trials were included whereas in the predictive

mean scores.ive flattening; SAPS/SANS mean scores.

cue condition 19 valid and 5 invalid cue trials were included for each experimentalcondition. Valid and invalid trials were pseudorandomized across subjects.

Each trial began with the appearance of the fixation cross for 1000 ms. Then thecue was presented for 200 ms. The word was presented through the headphones300 ms after the onset of the cue, followed by a response screen for 4.5 s. Participantswore headphones and the volume was checked to ensure that it was at the levelsufficient for them to hear the speech without difficulty. The response screen withresponse alternatives appeared on the computer monitor after every trial to avoidthe working memory load in subjects.

If the participants thought the speech they heard was their own they wereinstructed to press the button marked as number one on the keyboard. If theythought the speech belonged to someone else they were asked to press the buttonmarked as number two and if they were unsure of its source they were instructedto press the button number three. After participants made their response by press-ing “1”, “2” or “3” on the keyboard (‘self’, ‘other’, ‘unsure’) the new trial began. Thecomputer recorded the response and the reaction time. In addition to the ‘self’ and‘other’ responses, subjects were also able to register an ‘unsure’ response. This isimportant because when participants were in doubt about the source of speech,they were not obliged to make a forced choice between ‘self’ and ‘other’, makingit more likely that when they did select either of these responses they did so withsome degree of confidence (Fig. 1).

2.4. Procedure

Participants were informed upon recruitment that the experiment would beconducted over three sessions. It was explained by the researcher that the first ses-sion would be used to record their speech and the second and third session wouldbe used to administer the task.

2.4.1. First sessionThe participants were asked to read all 192 words in the microphone even

though half would subsequently be replaced by an alien voice, to ensure that partic-ipants could not make judgments based on source information when subsequentlypresented with the prerecorded words. Hence to ensure that the task relied on per-ceptual discrimination as opposed to source memory. Before the second session 96previously designated words (one half of the list) were replaced with the alien ver-sion of the word. Half of all self and alien words (96 words) were pitch shifted by −4semitones. The degree of pitch was chosen because it made speaker’s voice harder torecognize without making the word incomprehensible. The rest of the words wereunaltered self and alien voice version (48 words each). The same words were usedfor the predictive and unpredictive run. The picture of the subject was always takenat the end of the session. The entire first session took about 30 m.

2.4.2. Second and third sessionThese two sessions were held on separate days in order to avoid practice effects

for the participants. They took place within one week after the first session. Themean interval between the first and the second session was 2.5 days. Participantswere not informed of the level of cue predictability before each assessment, but theywere given a general hint that the cue can be either valid or invalid. The testing time

for the second and third session was about 15 m each.

2.4.3. Statistical analysisThe data were analyzed using Statistical Package for Social Science SPSS 16

(http://www.spss.com/). Separate ANOVAs for repeated measures were conductedwith misattribution errors, unsure responses and reaction times as the depen-

808 L.M. Ilankovic et al. / Neuropsychologia 49 (2011) 805–812

e of tr

dsw‘(bca

3

3

etF

3

picuf

d(wwtdWd

Fig. 1. Exampl

ent variables. Misattribution errors occurred when participants misattributed theource of the speech (i.e. responding with ‘other’ when the source of the speechas self and vice versa), as opposed to errors when participants responded with

unsure’ or made a null response. The within-subjects factors were source of speechself, alien), distortion (0, −4 semitones) and cue predictability (valid, invalid). Theetween subjects factor was group (controls, patients). Correlation analyses wereonducted using Spearman rs statistics. All statistics were two-tailed, and reportedt a significance p < 0.05. All data were normally distributed (Leven’s test).

. Results

.1. Errors

The groups were compared in terms of (a) misattributionrrors and (b) unsure responses. The graphs of mean propor-ions for unsure and misattributed responses are shown inigs. 2(a and b) and 3(a and b).

.1.1. Misattribution errorsThe effect of cue was non-significant (F = 0.139, df = 44,

= 0.711), but there was a significant main effect of source of speechn the unpredictive (F = 10.367, df = 44, p = 0.002) and predictive cueondition (F = 6.598, df = 44, p = 0.014). Across both predictive andnpredictive cueing conditions all participants made more errorsor self words than for alien words.

There was a significant main effect of distortion in both con-itions, unpredictive (F = 48.128, df = 44, p = 0.000) and predictiveF = 50.035, df = 44, p = 0.000). All participants made more errorshen the words were distorted than when they were not. There

as also a significant interaction between source of speech, distor-

ion and group in both cueing conditions, unpredictive (F = 10.203,f = 44, p < 0.003) and predictive (F = 10.429, df = 44, p < 0.002).hen the words were distorted, patients with hallucinations and

elusions made significantly more misattribution errors for self

ial procedure.

words than for alien words comparing to healthy controls (selecting‘other’ when hearing their own voice). See Fig. 2a and b.

In order to bolster the significance of this result, we report theeffect size measure, Cohen’s d, which was large for this comparison(d = 1.07).

In the predictive cue condition, the main effect of cueing wasnot significant (F = 1.48, df = 44, p = 0.229). However, the interac-tion between the cue type and group was significant (F = 4.64,df = 44, p = 0.037). Patients with hallucinations and delusions, butnot healthy controls, made significantly more errors across all theconditions in which the cue was invalid (see Fig. 2b).

Whereas the interaction between the cue type and the groupwas significant the further interaction between cueing, source ofspeech and group was not significant (F = 4.22, df = 44, p = 0.520).

The difference between the patients and the healthy controlswas not significant in invalid cue trials when participant’s ownvoice is preceded by an alien cue (t = 0.346, df = 44, p = 0.168), albeitthe effect size measure was moderate (d = 0.42).

3.1.2. Unsure responsesThe interaction between the cue type and group was not sig-

nificant (F = 0.440, df = 44, p = 0.598). Patients with hallucinationsand delusions did not make significantly more unsure responsescompared to the healthy controls. The main effect of the distortionwas significant both in unpredictive (F = 13.009, df = 43, p = 0.001)and predictive cue condition (F = 0.311, df = 44, p = 0.001) in bothgroups. All participants made more unsure responses when thewords were distorted than when they were not, irrespective of

the source of speech. There was a significant interaction betweencueing (invalid, valid), source of speech and distortion (F = 4.255,df = 43, p = 0.045) in the unpredictive cue condition, showing thatall subjects were particularly prone to give unsure responses forself distorted words preceded by an invalid cue (see Fig. 3a). In

L.M. Ilankovic et al. / Neuropsychologia 49 (2011) 805–812 809

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

valid invalid valid invalid

Mis

a�ri

bu�

on e

rror

rat

e (%

) pa�ents

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

valid invalid valid invalid

self original

self distorted

alien original

alien distorted

controls pa�entscontrolsa

b

Fig. 2. (a) Mean misattribution rate across unpredictive condition and (b) mean misattribution rate across predictive condition.

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Uns

ure

resp

onse

rate

(%) controls pa�ents controls pa�ents

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

self originalself distortedalien originalalien distorted

ab

ition a

at(

3

atdtbtot

0valid invalid valid invalid

Fig. 3. (a) Mean rate of unsure responses across unpredictive cond

ddition, we found a trend for interaction between cueing and dis-ortion (F = 3.953, df = 44, p = 0.053) in the predictive cue conditionsee Fig. 3b)

.2. Correlation analysis

Bivariate correlation analysis was performed to examine thessociation between misattribution errors and symptom ratings inhe patient group. In the unpredictive cue condition, the PSYRATSelusions scores were positively correlated with errors on invalid

rials, in which patients listened to their distorted voice precededy an alien face (r = 0.420, p = 0.046). In the predictive cue condi-ion no correlations between errors on invalid trials and PSYRATSr SAPS scores were found. Also, misattribution errors when lis-ening to the self-distorted voice preceded by the self cue (r = 0.588,

0

500

1000

1500

2000

2500

3000

3500

0

500

1000

1500

2000

2500

3000

3500

valid invalid valid invalid

Reac

�on

�m

e (m

s)

controls pa�entsa b

Fig. 4. (a) Mean reaction time for correct responses across unpredictive condition a

valid invalid valid invalid

nd (b) mean rate of unsure responses across predictive condition.

p = 0.003) and self-undistorted voice preceded by the alien cue werepositively correlated with the SANS global attention deficit score(r = 0.472, p = 0.023).

3.3. Reaction times

To assess if task accuracy across groups was related to responsespeed, mean reaction times were calculated for correct (see Fig. 4aand b) and error trials (see Fig. 5a and b). Reaction times for cor-rect responses on valid and invalid trials did not differ between

the groups. However, for the misattribution errors there was a sig-nificant interaction between cueing, distortion and group in theunpredictive (F = 5.42, df = 41, p = 0.025) and predictive cue condi-tion (F = 4.24, df = 41, p = 0.046). The patients responded faster toundistorted speech falsely when preceded by an invalid cue.

valid invalid valid invalid

self original

self distorted

alien original

alien distorted

controls pa�ents

nd (b) mean reaction time for correct responses across predictive condition.

810 L.M. Ilankovic et al. / Neuropsychologia 49 (2011) 805–812

0

500

1000

1500

2000

2500

3000

3500

0

500

1000

1500

2000

2500

3000

3500

v

self original

self distorted

alien original

alien distorted

controls pa�ents controls pa�ents

Reac

�on

�m

e (m

s)

a b

ition a

3

wtttcp

4

awsamteodt

fap

chbsn(

amMmd

bfitfiamb

valid invalid valid invalid

Fig. 5. (a) Mean reaction time for errors across unpredictive cond

.4. Cross-condition effects

Across the two main conditions (unpredictive and predictive)e found a trend for patients to give less correct answers in invalid

rials in the predictive cue condition compared to the unpredic-ive condition. At the same time, healthy controls were proneo give more correct responses to invalid trials in the predictiveue condition compared to the unpredictive one (F = 0.843, df = 44,= 0.099).

. Discussion

The aim of the present study was to investigate the inter-ction between top-down and bottom-up attention in patientsith schizophrenia with predominantly positive symptoms. More

pecifically, we were interested in testing whether patients’ sourcettributions for their own speech were biased by top-down infor-ation (visual cues of their own or another person’s face) at

he expense of bottom-up stimuli (voice stimuli), in two differ-nt versions of a cueing task. We predicted that the dominancef top-down over bottom-up mechanisms would be expressed byirecting attention to the cue preceding the target, and followinghis cue irrespective of its predictive power.

During the unpredictive cueing task, as expected, there was noacilitation in either controls or patients. The results confirmed ourssumption and there is no evidence that cueing has an effect on theerformance of any of the two groups in the unpredictive condition.

In line with our hypothesis, our results from the predictiveueing task show that patients were far more susceptible thanealthy controls to cue manipulation, resulting in more misattri-ution errors on invalid trials. This is consistent with previoustudies which have demonstrated that patients with schizophre-ia show increased rates of false alarms to invalidly cued targetsJavitt, Rabinowich, Silipo, & Dias, 2007).

Our findings from both predictive and unpredictive conditionre consistent with previous findings using speech appraisal andonitoring task (Allen et al., 2004; Cahill et al., 1996; Ford &athalon, 2005; Johns et al., 2001, 2006), that H/D-patients areore likely to make misattributions about the source of their own

istorted speech than healthy controls.The critical issue, whether misattribution errors are augmented

y invalid top-down information in the invalid trials, was con-rmed. Our second hypothesis that externalizing bias on invalid

rials in the patient group could be source specific was not con-rmed, as the interaction between cueing, source of the speechnd group was not significant. The patients made significantlyore errors across all the conditions in which the cue was invalid,

ut they were not particularly prone to misattribute undistorted

alid invalid valid invalid

nd (b) mean reaction time for errors across predictive condition.

self-generated speech to an external source when the voice waspreceded by an alien (invalid) cue compared to healthy con-trols.

However, the patients experience uncertainty during invalidcueing of the voice stimuli, especially when the majority of all cuesare valid (80–20 condition). As the predictive cue condition hashigher informative power to guide source attributions relative tothe unpredictive cue condition, this strongly suggests that the pat-tern of misattribution responses does differ between patients andcontrols depending on the cue predictability.

Analysis of the participants reaction times showed that patientswere faster when making source judgments after invalid cueing ofundistorted speech relative to healthy controls. This may reflectan overall tendency for patients to make quicker judgments basedon less information. This is in accordance with patients with delu-sions to ‘jump to conclusions’ (Garety, Hemsley, & Wessley, 1991).Additionally, the finding that patients were faster when makingsource judgments after invalid cues relative to healthy controlsis consistent with the study of Nestor et al. (1992). They showedthat patients display an abnormal rapid disengagement of atten-tion or reduced attentional cost for invalid cues. This is presumablybecause patients have difficulty maintaining a mental set, whichwould allow them to benefit during regular, predictable sequences.

Interestingly, the tendency for patients to misattribute self-generated speech was not associated with the severity ofhallucinations. However, the relationship between external mis-attribution of the source and hallucinations isolated from otherpositive symptoms has been challenged through several stud-ies (Allen et al., 2007). It has been suggested that the externalmisattribution of the source is related to the general acute psy-chotic state rather than to a predisposition to hallucinations (Johnset al., 2006). The positive association between misattribution errorsin the invalid cueing conditions with PSYRATS delusions scoresconfirms our assumption that the impaired ability to integrate top-down and bottom-up information is related to an extent to thepositive symptom profile.

As we mentioned before, the tendency to make misattributionerrors in patients currently experiencing auditory hallucinationsin the context of an affective psychosis was not identical to theone that is apparent both in patients experiencing hallucinationsand patients experiencing delusions without hallucinations in thecontext of schizophrenia (Johns et al., 2006). They tend to makesignificantly more ‘unsure’ responses which makes the verbal self-

monitoring deficit in this group of the patients questionable. Hence,the same study reported that defective self-monitoring was notevident in patients who had a history of hallucinations but werecurrently hallucination-free. Therefore, our findings together withprevious reports from verbal self-monitoring studies may suggest

psych

ts

nd(dow2sapsatcbbpbdiuuatmnKpsItiab

bpudfibib

oda

F

Stow

C

aIs

L.M. Ilankovic et al. / Neuro

hat the patients’ response profile is rather related to psychotictate and not to a specific diagnosis.

Taken together our results provide further evidence for exter-al misattribution of source in patients with hallucinations andelusions, in addition to previous studies, on the cross modalaudio–visual) level. An interaction between bottom-up and top-own processes is necessary to achieve rapid and flexible attentionn a cross-modal level and this seems to be impaired in patientsith positive symptoms. As the external misattribution (Allen et al.,

004; Johns et al., 2006) has been shown previously in studies onchizophrenia, as well as the difficulties in integration of voluntarynd reflexive attentional orienting (Maruff et al., 1998), it seemslausible to claim these impairments in patients with positiveymptoms. The current results must be interpreted with cautions study has limitations. We did not test for the specificity to posi-ive symptoms of schizophrenia. Ideally, the relationship betweenognitive deficits, psychotic state and schizophrenia as such, shoulde investigated in the same individuals using a longitudinal design,ut this approach is logistically difficult. In this study, although theatients scored highly on the positive symptoms scales, the possi-ility that their impaired performance might reflected some otherifference to controls cannot be excluded. The influence of med-

cation that patients were taking could be important, but we arenable to confirm this as no studies on verbal self-monitoring havesed the medication free populations. Greater sample sizes wouldlso provide greater power to statistical analyses. As this study ishe first one to integrate attentional modulation and verbal self-

onitoring our results can be interpreted within the framework ofegative symptoms and attentional deficit in schizophrenia (Green,ern, Braff, & Mintz, 2000; Rocca et al., 2006). Our sample com-rised of patients with different lengths of illness and some of themuffered not only from positive, but also from negative symptoms.t may be that some patients experienced greater task difficulty dueo again lack of capacity for attention. The correlation between thenvalid condition and attentional deficit measured by the scale forssessment of negative symptoms (SANS), (Andreasen, 1984) maye suggestive of such a relationship.

In conclusion, the present study is the first to dissociateottom-up and top-down processing using a verbal self-monitoringaradigm and determine a relationship between attentional mod-lation and voice recognition in patients with hallucinations andelusions. Specifically, patients with schizophrenia and this pro-le of symptoms have difficulty switching between top-down andottom-up processing. Further, patients seem not to adequately

nhibit top-down information to guide them at the expense ofottom-up information.

These results indicate that perturbed flexibility and integrationf different attentional modes might be of importance regardingisturbed reality perception which may give rise to hallucinationsnd delusions.

unding

Lana Marija Ilankovic received the Barbara-Wengeler-Stiftungcholarship. However, overall experimental design, data acquisi-ion, statistical analyses, interpretation of the results, preparationf the article and decision to submit the article were implementedithout input from any of the pharmaceutical companies or grants.

onflict of interest

M. Riedel has received research grants/support or has served asconsultant for AstraZeneca, Pfizer, Otsuka Pharma, Janssen-Cilag.

n the context of investigator initiated trials M. Riedel has receivedupport from AstraZeneca and Pfizer. N. Müller received hon-

ologia 49 (2011) 805–812 811

oraria, paid expert testimony, patent applications, registrations,and grants within the last 3 years from Affectis AG, AstraZeneca,Janssen-Cilag, Pfizer, Lundbeck, Otsuka, Lilly. Other authors declareno conflict of interest.

Acknowledgments

We are grateful to the Barbara-Wengeler-Stiftung Scholarshipawarded to Lana Marija Ilankovic. We would like to thank Dr. R.Zimmermann, Dr. Friebel, Dr. M. Holzer, Dr. J. Nürnberger and Dr.S. Dehning for their assistance in the assessment of the patients.The authors are also grateful to D. Rangelov for his helpful com-ments. Finally, authors acknowledge participants volunteering inthis research.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.neuropsychologia.2011.01.016.

References

Aleman, A., Böcker, K., Hijman, R., De Haan, E. H. F., & Kahn, R. (2003). Cognitive basisof hallucinations in schizophrenia: Role of top-down information processing.Schizophrenia Research, 64, 175–189.

Allen, P. P., Johns, L. C., Fu, C. H., Broome, M. R., Vythelingum, G. N., & McGuire, P.K. (2004). Misattribution of external speech in patients with hallucinations anddelusions. Schizophrenia Research, 69(2–3), 277–287.

Allen, P. P., Aleman, A., & McGuire, P. K. (2007). Inner speech models of audi-tory verbal hallucinations: Evidence from behavioral and neuroimaging studies.International Review of Psychiatry, 19(4), 409–417.

Andreasen, N. C. (1984). Scale for the assessment of positive symptoms (SAPS) and scalefor the assessment of negative symptoms (SANS). Iowa: The University of Iowa.

Bentall, R. P., Baker, G. A., & Havers, S. (1991). Reality monitoring and psychotichallucinations. The British Journal of Clinical Psychology, 30, 213–222.

Bentall, R. P., Kinderman, P., & Kaney, S. (1994). The self, attributional process andabnormal beliefs: Towards a model of persecutory delusions. Behaviour Researchand Therapy, 3(32), 331–341.

Blakemore, S. J., Smith, J., Steel, R., Johnstone, E. C., & Frith, C. D. (2000). The percep-tion of self-produced sensory stimuli in patients with auditory hallucinationsand passivity experiences: Evidence for a breakdown in self-monitoring. Psy-chological Medicine, 30, 1131–1139.

Blakemore, S. J., Wolpert, D. M., & Frith, C. D. (2002). Abnormalities in the awarenessof action. Trends in Cognitive Sciences, 6, 237–242.

Brébion, G., Smith, M., Gorman, J., & Amador, X. (1997). Discrimination accuracy anddecision biases in different types of reality monitoring in schizophrenia. TheJournal of Nervous and Mental Disease, 185, 247–253.

Brébion, G., Amador, X., David, A., Malaspina, D., Sharif, Z., & Gorman, J. M. (2000).Positive sympomatology and source monitoring failure in schizophrenia—Ananalysis of symptoms-specific effect. Psychiatry Research, 95, 119–131.

Brébion, G., Gorman, J. M., Amador, X., Malaspina, D., & Sharif, Z. (2002). Sourcemonitoring impairments in schizophrenia: Characterization and associationswith positive and negative symptomatology. Psychiatry Research, 112, 27–39.

Butler, P., Martinez, A., Foxe, J., Kim, D., Zemon, V., Silipo, G., et al. (2007). Subcor-tical visual disfunction in schizophrenia drives secondary cortical impairments.Brain, 130, 417–430.

Cahill, C., Silbersweig, D., & Frith, C. D. (1996). Psychotic experiences induced indeluded patients using distorted auditory feedback. Cognitive Neuropsychiatry,1, 201–211.

Danckert, J., Saoud, M., & Maruff, P. (2003). Attention, motor control and motorimagery in schizophrenia: Implications for the role of the parietal cortex.Schizophrenia Research, 70(2–3), 241–261.

Desimone, R., & Duncan, J. (1995). Neural mechanisms of selective visual attention.Annual Review of Neuroscience, 18, 193–222.

Egeth, H. E., & Yantis, S. (1997). Visual attention: Controls, representation, and timecourse. Annual Review of Psychology, 48, 269–297.

Ford, J., & Mathalon, D. (2005). Corollary discharge dysfunction in schizophrenia:Can it explain auditory hallucinations? International Journal of Psychophysiology,58, 179–189.

Frith, C. D. (1987). The positive and negative symptoms of schizophrenia reflectimpairments in the perception and initiation of action. Psychological Medicine,17, 631–648.

Frith, C. D. (1992). The cognitive neuropsychology of schizophrenia. Hove: LawrenceEarlbaum Associates.

Fuller, R. L., Luck, S. J., Braun, E. L., Robinson, B. M., McMahon, R. P., & Gold, J. M.(2006). Impaired control of visual attention in schizophrenia. Journal of AbnormalPsychology, 115(2), 266–275.

Garety, P., Hemsley, D., & Wessley, S. (1991). Reasoning bias in deluded schizophren-ics and paranoid patients: Biases in performance in probabilistic reasoning. TheJournal of Nervous and Mental Disease, 179, 194–201.

8 psych

G

G

H

J

J

J

K

L

L

M

affective disorder, and normal subjects. Schizophrenia Research, 20(1–2), 33–50.

12 L.M. Ilankovic et al. / Neuro

old, J. M., Fuller, R. L., Robinson, B. M., Braun, E. L., & Luck, S. J. (2007). Impairedtop-down controls of visual search in schizophrenia. Schizophrenia Research, 94,148–155.

reen, M., Kern, R., Braff, D., & Mintz, J. (2000). Neurocognitive deficits and functionaloutcome in schizophrenia: Are we measuring the “right stuff”? SchizophreniaBulletin, 26(1), 119–136.

addock, et al. (1999). Scales to measure dimensions of hallucinations and delu-sions: The psychotic symptom rating scales (PSYRATS). Psychological Medicine,29(4), 879–889.

avitt, D. C., Rabinowich, E., Silipo, G., & Dias, E. (2007). Encoding vs. retention:Differential effects of cue manipulation on working memory performance inschizophrenia. Schizophrenia Research, 91, 159–168.

ohns, L. C., Rossell, S., Frith, C., Ahmad, F., Hemsley, D., Kuipers, E., et al. (2001). Verbalself-monitoring and auditory verbal hallucinations in patients with schizophre-nia. Psychological Medicine, 31, 705–715.

ohns, L. C., Gregg, L., Allen, P., & McGuire, P. (2006). Impaired verbal self-monitoringin psychosis: Effects of state, trait and diagnosis. Psychological Medicine, 36,465–474.

raepelin, E. (1919). Dementia Praecox (translated by R.M. Barclay) (1971). Edinburgh:Livingstone.

eavitt, V. M., Molholm, S., Ritter, W., Shpaner, M., & Foxe, J. (2007). Auditoryprocessing in schizophrenia during the middle latency period (10–50 ms):High-density electrical mapping and source analysis reveal subcorticalantecedents to early cortical deficits. Journal of Psychiatry & Neuroscience, 32,

339–353.

iddle, P. F. (1987). Schizophrenic syndromes, cognitive performance and neurolog-ical dysfunction. Psychological Medicine, 17, 49–57.

aldonado-Heinks, T., Mathalon, D., Houde, J., Gray, M., Faustman, W., & Ford, J.(2007). Relationship of imprecise corollary discharge in schizophrenia to audi-tory hallucinations. Archives of General Psychiatry, 64, 286–296.

ologia 49 (2011) 805–812

Maruff, P., Danckert, J., Pantelis, C., & Currie, J. (1998). Attentional and saccadicabnormalities in schizophrenia. Psychological Medicine, 28, 1091–1100.

Mathalon, D. H., Heinks, T., & Ford, J. (2004). Selective attention in schizophrenia:sparing and loss of executive control. The American Journal of Psychiatry, 161,872–881.

Nestor, P. G., Faux, S. F., McCarley, R. W., Penhune, V., Shenton, M. E., Pollak, S., &Sands, S. F. (1992). Attentional cues in chronic schizophrenia: Abnormal disen-gagement of attention. Journal of Abnormal Psychology, 101, 682–689.

Posner, M. I., Early, T. S., Reiman, E., Pardo, P. J., & Dhawan, M. (1988). Asymmetries inhemispheric control of attention in schizophrenia. Archives of General Psychiatry,45, 814–821.

Rocca, P., Castagna, F., Marchiaro, L., Rasseti, R., Rivoira, E., & Bogetto, F. (2006).Neuropsychological correlates of reality distortion in schizophrenic patients.Psychiatry Research, 145(1), 49–60.

Sarter, M., Givens, B., & Bruno, J. (2001). The cognitive neuroscience of sustainedattention: Where top-down meets bottom-up. Brain Research Reviews, 35,146–160.

Schmidt, K.-H., & Metzler, P. (1992). Wortschatztest (WST). Weinheim: Beltz TestGmbH.

Seal, M. L., Aleman, A., & McGuire, P. K. (2004). Compelling imagery, unanticipatedspeech and deceptive memory: Neurocognitive models of auditory verbal hal-lucinations in schizophrenia. Cognitive Neuropsychiatry, 9, 43–72.

Sereno, A. B., & Holzman, P. S. (1996). Spatial selective attention in schizophrenic,

Slade, P. D., & Bentall, R. P. (1988). Sensory Deception: Towards a Scientific Analysis ofHallucinations. London: Croom Helm.

World Health Organization. (2008). Internationale Klassifikation psychischerStörungen ICD-10. Kapitel V (F). In Diagnostische Kriterien für Forschung undPraxis. Bern: Huber.