Brain substrates underlying auditory speech primingin healthy listeners and listeners with schizophrenia

C Wu123dagger Y Zheng4dagger J Li4dagger H Wu4 S She4 S Liu4 Y Ning4 and L Li145

1School of Psychological and Cognitive Sciences and Beijing Key Laboratory of Behavior and Mental Health Key Laboratory on Machine Perception(Ministry of Education) Peking University Beijing Peoplersquos Republic of China2School of Life Sciences Peking University Beijing Peoplersquos Republic of China3School of Psychology Beijing Normal University Beijing Peoplersquos Republic of China4The Affiliated Brain Hospital of Guangzhou Medical University Guangzhou Peoplersquos Republic of China5Beijing Institute for Brain Disorders Capital Medical University Beijing Peoplersquos Republic of China

Background Under lsquococktail partyrsquo listening conditions healthy listeners and listeners with schizophrenia can use tem-porally pre-presented auditory speech-priming (ASP) stimuli to improve target-speech recognition even though listenerswith schizophrenia are more vulnerable to informational speech masking

Method Using functional magnetic resonance imaging this study searched for both brain substrates underlying theunmasking effect of ASP in 16 healthy controls and 22 patients with schizophrenia and brain substrates underlyingschizophrenia-related speech-recognition deficits under speech-masking conditions

Results In both controls and patients introducing the ASP condition (against the auditory non-speech-priming condi-tion) not only activated the left superior temporal gyrus (STG) and left posterior middle temporal gyrus (pMTG) but alsoenhanced functional connectivity of the left STGpMTG with the left caudate It also enhanced functional connectivity ofthe left STGpMTG with the left pars triangularis of the inferior frontal gyrus (TriIFG) in controls and that with the leftRolandic operculum in patients The strength of functional connectivity between the left STG and left TriIFG was corre-lated with target-speech recognition under the speech-masking condition in both controls and patients but reduced inpatients

Conclusions The left STGpMTG and their ASP-related functional connectivity with both the left caudate and somefrontal regions (the left TriIFG in healthy listeners and the left Rolandic operculum in listeners with schizophrenia)are involved in the unmasking effect of ASP possibly through facilitating the following processes masker-signal inhib-ition target-speech encoding and speech production The schizophrenia-related reduction of functional connectivitybetween the left STG and left TriIFG augments the vulnerability of speech recognition to speech masking

Received 29 October 2015 Revised 6 October 2016 Accepted 11 October 2016 First published online 29 November 2016

Key words Auditory speech priming informational masking pars triangularis of the inferior frontal gyrusschizophrenia speech recognition

Introduction

One of the difficult perceptual tasks with high percep-tual load for human listeners is to recognize speech in anoisy lsquococktail partyrsquo speech-listening environmentwith multiple people talking (Cherry 1953) becausethe attended speech is under not only energetic mask-ing that occurs at the peripheral level but also informa-tional masking that interferes with the processing of

the target talkerrsquos utterances at more central (ie cogni-tive) levels of processing (Schneider et al 2007)

More specifically under lsquococktail partyrsquo environ-ments with multiple people talking (Cherry 1953)two types of masking components contribute to thedifficulty of speech recognition energetic maskingand informational masking (eg Freyman et al 1999Brungart et al 2001 Li et al 2004 Helfer amp Freyman2005 2009 Wu et al 2005 Rakerd et al 2006 Huanget al 2008 2009 Wu et al 2012b) Energetic maskingoccurs when peripheral neural activity elicited by a sig-nal is overwhelmed by that elicited by maskers lead-ing to a degraded or lost neural representation of thesignal Informational masking occurs when signalsand maskers are similar along some informationaldimensions particularly when both target signalsand maskers are speech sounds causing confusion

Address for correspondence L Li School of Psychological andCognitive Sciences Peking University Beijing 100080 PeoplersquosRepublic of China

(Email lianglipkueducn)dagger These authors contributed equally to this work and should be

co-first authors

Psychological Medicine (2017) 47 837ndash852 copy Cambridge University Press 2016doi101017S0033291716002816

ORIGINAL ARTICLE

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

between the target and masker andor uncertaintyregarding the target Relative to energetic maskinginformational masking involves more higher-ordercentral processing (Schneider et al 2007)

Under such adverse listening conditions it is notunusual for a listener to ask the attended speakerrepeat the sentence(s) The beneficial lsquosay-it-againrsquoeffect is caused by using some perceptually andor cog-nitively unmasking primes including prior knowledgeabout part of target-sentence content (Freyman et al2004 Yang et al 2007 Helfer amp Freyman 2009 Wuet al 2012b) familiarity with the target talkerrsquos voice(Yang et al 2007 Huang et al 2010) and visual work-ing memory of the speakerrsquos movements of speecharticulators (Wu et al 2013a b)

In studies of auditory speech priming (ASP) when ameaningless (nonsense) target sentence with a numberof keywords is co-presented with a two-talker speechmasker (which causes informational masking of thetarget sentence at both perceptual and cognitive levelssee Schneider et al 2007) recognition of the last keywordof the target sentence is improved if the early segment ofthe same sentence (including the first or the first two key-words) recited by the target talkerrsquos voice (ie the audi-tory speech prime) is temporally pre-presented in quietbefore the targetmasker co-presentation (Freyman et al2004 Yang et al 2007 Wu et al 2012b) Freyman et al(2004) have suggested that the speech prime helps listen-ers focus attention more quickly on the target speechthereby facilitating recognition of the last keyword inthe target stream against informational masking Up todate the brain substrates specifically underlying theunmasking effect of ASP have not been reported in theliterature It has been reported that subjective clarity ofdegraded (noise-vocoding) speech can be enhanced byprior knowledge of speech content (Davis et al 2005Sohoglu et al 2014) and the underlying mechanismsinclude top-down modulation of speech-signal process-ing at the processing stage of the auditory cortex(Sohoglu et al 2012 Wild et al 2012) It is of interest toknow whether the brain substrates underlying auditoryprime-induced unmasking of speech are similar tothose underlying prior speech-knowledge-inducedimprovement of clarity of degraded speech

Compared with that in healthy listeners the correctpercentage of target-keyword recognition in patientswith schizophrenia is lower not only under noise-masking conditions but also under speech-maskingconditions (Wu et al 2012a 2013b) More in detailour previous studies have revealed that the differencein the threshold (in signalmasker ratio SMR) for rec-ognizing target keywords between people withfirst-episode schizophrenia and their healthy controlsis 10 dB when the masker is steady-state speech-spectrum noise but increases to 17 dB when the

masker is two-talker speech Also the threshold differ-ence between people with chronic schizophrenia andtheir healthy controls is 19 dB when the masker isthe noise but increases to 30 dB when the masker isthe speech Interestingly although people with schizo-phrenia perform worse than healthy listeners in speechrecognition against informational masking (Wu et al2012a 2013b) they retain the ability to use the tempor-ally pre-presented auditory-speech primes to improve(unmask) target-speech recognition under informa-tional masking conditions (Wu et al 2012a) Knowingwhether the brain substrates normally underlying theASP in healthy listeners are also functional in listenerswith schizophrenia is important for understanding thenature of the brain plasticity associated withschizophrenia

It has been reported that people with schizophreniaexperience more difficulties in filtering irrelevantdis-tracting sensory stimuli to prevent informationoverflow that leads to various cognitive dysfunctions(Gottesman amp Gould 2003 Braff amp Light 2005)They also perform worse than healthy listeners inspeech perception particularly under informationalspeech-masking conditions (Wu et al 2012a 2013bZheng et al 2016) Up to date marked progress hasbeen made in understanding the brain regions thatare related to speech (informational) masking of targetspeech in healthy people (Scott et al 2004 2009 Ding ampSimon 2012 Scott amp McGettigan 2013 Evans et al2016) For example the study of Evans et al (2016)showed that both the mid-posterior superior temporalgyrus (STG) and the superior temporal sulcus exhibithigher activity as informational content of maskingsounds is increased suggesting that masking speechis processed within the same pathway for processingtarget speech However only a very small number ofstudies have been reported in the literature on brainsubstrates underlying perceptual cue (spatial ornon-spatial)-induced unmasking of speech recognitionin either healthy listeners or listeners with schizophre-nia (Zheng et al 2016) Using functional magnetic reson-ance imaging (fMRI) in both healthy listeners andlisteners with schizophrenia this study was to searchfor (1) the brain substrates specifically underlying theunmasking effect of ASP and (2) the brain mechanismsunderlying schizophrenia-related speech-recognitiondeficits under speech-masking conditions

Method

Participants

With the recruiting criteria used in our previous stud-ies (Wu et al 2012a 2013b Zheng et al 2016) partici-pants with schizophrenia who were diagnosed with

838 C Wu et al

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

the Structured Clinical Interview for DSM-IV (SCID)(First et al 1996) were recruited in the GuangzhouBrain Hospital Patients with diagnoses of schizo-affective or other psychotic disorders were not includedDemographics-matched healthy listeners (selected forcontrols) were recruited from the community aroundthe hospital with the recruiting criteria used previously(Wu et al 2012a 2013b Zheng et al 2016) They weretelephone interviewed first and then those who passedthe interview were screened with the SCID as used forpatient participants None of the selected healthy con-trols had either a history of Axis I psychiatric disorderas defined by the DSM-IV or a family history of psychi-atric illness based on self-report

Note that some potential patient participants wereexcluded from this study if they had co-morbid diagno-ses substance dependence andor other conditions thataffected experimental tests [including hearing lossabuse andor dependence of alcohol andor drugs atreatment of electroconvulsive therapy (ECT) withinthe past 3 months a treatment of trihexyphenidyl hydro-chloride with a dose of more than 6 mgday andor anage younger than 18 years or older than 59 years] Forthe purpose of improving sleeping some of the patientparticipants received benzodiazepines based on doctorsrsquoadvice

Both patient participants and their guarantees gavetheir written informed consent for participation inthis study Demographics-matched healthy listeners(controls) were recruited from the community aroundthe hospital with the recruiting criteria used previously(Wu et al 2012a) The procedures of this study wereapproved by the Independent Ethics Committee ofthe Guangzhou Brain Hospital

In all 17 healthy participants and 25 patients partici-pated in the study One healthy listener and threepatients were excluded from data analyses due totheir excessive head movement (more than 25 mm intranslation andor 25deg in rotation from the first volumein any axis) during fMRI scanning The remaining 16healthy listeners (seven males and nine females) and22 patients (14 males and 8 females) were included inboth behavioral testing and fMRI data analyses Allparticipants were right-handed with normal pure-tonehearing thresholds at each ear (lt30 dB hearing level) atfrequencies between 125 and 8000 Hz Their first lan-guage was Mandarin Chinese

All patient participants were clinically stable duringtheir participation and they received antipsychotic med-ications during this study with the average chlorpro-mazine equivalent of 521 mgday based on theconversion factors described by Woods (2003) Thelocally validated version of the Positive and NegativeSyndrome Scale Test (Si et al 2004) was conducted onthe day of fMRI scanning for all participants The speech

recognition test was conducted after the fMRI scanningThe characteristics of the patients and healthy partici-pants are shown in online Supplementary Table S1

Stimuli and procedures

There were three types of speech stimuli in total targetspeech masking speech and priming speech Spokenby a young female talker (Talker A) target speechstimuli were Chinese nonsense phrases with threewords and each word contained two syllables Theywere syntactically ordinary but not semantically mean-ingful (see Wu et al 2013b) For example the Englishtranslation of a phrase is lsquoretire his oceanrsquo (keywordsare in italics) Obviously the sentence frame providedno contextual support for recognizing keywords Thespeech masker was a 47-s loop of digitally combinedcontinuous recordings for Chinese nonsense sentences(whose keywords did not appear in target phrases)spoken by two other young female talkers (Talkers Band C Yang et al 2007) There were two types of prim-ing conditions (1) ASP condition ndash the prime (stimu-lus) was identical to the target phrase except that thelast (second) keyword was replaced by a white noisewith both the duration of the longest second keywordamong the target phrases and the sound level 10 dBlower than that of target stimuli (following Freymanet al 2004) (2) auditory non-speech priming (ANSP)condition (ie the speech-priming control condition) ndashthe prime was a white noise with both the durationof the longest target phrase and the sound level 10dB lower than that of target stimuli In addition dur-ing fMRI scanning a non-speech baseline stimulationcondition (the controlling condition for both ASP andANSP conditions) was used with the presentation ofwhite noise whose duration was equal to that of thelongest target phrase and whose sound level was 10dB lower than that of target stimuli

The whole-course scanning consisted of a 10-minauditory-priming functional run and an 8-minstructure-scanning run An event-related fMRI designwas used for the functional run In total there were60 scanning trials for the functional run (20 trials foreach of the three conditions ASP ANSP and baselinestimulation) For an individual participant the 60 trialsacross the three conditions were presented with a ran-dom order The sparse-imaging strategy was used toavoid the effect of machine scanning noise Soundstimuli were presented only during the pause periodbetween successive scanning periods (Hall et al1999) Also to ensure that stimulus-evoked hemo-dynamic responses peaked within the scanning period(Wild et al 2012) the stimulus presentation was sotemporally positioned that the midpoint of stimulus

Informational masking of speech 839

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

presentation occurred about 4200 ms before the onsetof the next scanning

In a scanning trial for examining the priming effect(Fig 1) either the ASP stimulus or the ANSP stimuluswas presented in quiet (without the masker presenta-tion) 600 ms after the offset of the last scanning trialImmediately after the prime presentation the targetand masker were co-presented and terminated simul-taneously To maintain participantsrsquo attention to stimu-lus presentation a two-syllable word either the lastkeyword in the target speech phrase (with the 50possibility) or a different word was presented for500 ms after the co-presentation of the target andmasker speech The participant was instructed to usea button-press with their right index finger to indicatewhether the word on screen was the last keyword inthe target-speech phrase

The sound stimuli used in the fMRI scanning experi-ment were presented through a magnetic resonance-compatible pneumatic headphone system (SAMRTECChina) driven by Presentation software (version 070)without introducing interaural time differences Thetarget-speech level was 60 dB sound pressure level(SPL) (after attenuation by earplugs) and the SMR wasminus4 dB Visual stimuli were presented through a liquidcrystal display screen positioned on the head coil(SAMRTEC China) A brief training was conducted toensure that participants understood the instructionand knew how to conduct their button-press responsesSpeech sentences used in training were different fromthose in experimental scanning

A 30-Tesla Philips Achieva MRI scanner (Veenpluis4-65680 DA Best the Netherlands) was used to acquireblood oxygenation level-dependent (BOLD) gradientecho-planar images (spatial resolution 64 times 64 times 33matrix with the voxel size of 344 times 344 times 46 mm3acquisition time 2000 ms time to repeat 9000 msecho time 30 ms flip angle 90deg field of view 211 times211 mm2) It provided high-resolution T1-weightedstructural images (256 times 256 times 188 matrix with a spatialresolution of 1 times 1 times 1 mm3 repetition time 82 msecho time 38 ms flip angle 7deg)

fMRI data processing and analyses

Pre-processing

All fMRI data were processed and analysed usingStatistical Parametric Mapping software (SPM8 theWellcome Trust Centre for Neuroimaging UK) Thepre-processing of data includes the following fourstages (1) the functional images were corrected forhead movements (2) the anatomical images wereco-registered with the mean realigned images and nor-malized to the standard template [UCLA BrainMapping Center (ICBM) space] using the SPM8

unified segmentation routine (3) all functional imageswere warped using deformation parameters generatedfrom the normalization process including re-samplingto a voxel size of 30 times 30 times 40 mm3 and (4) spatialsmoothing was conducted using a Gaussian kernelwith 8 mm full-width at half maximum (FWHM)Due to the long time to repeat this sparse-imagingparadigm no slice timing was necessary

Random-effect analyses

Random-effect analyses contained two processinglevels At the first level the onsets and durations forthe functional run were modeled using a general linearmodel according to the condition types The threestimulation conditions (ASP ANSP and baseline) wereincluded in the model Six realignment parameters ofhead movement were included to account for residualmovement-related effects (Friston et al 1996) At thesecond level random-effect analyses were conductedbased on the statistical parameter maps from each indi-vidual participant to allow population inference

To assess the overall main effects of participantgroup and prime type and the interaction both thecontrast image of lsquoASP gt baselinersquo and that of lsquoANSPgt baselinersquo from each participant were entered into asecond-level 2 times 2 (group times prime type) full-factoranalysis of variance (ANOVA) F contrasts were used

To localize the brain regions activated by speechstimulation conditions the contrast images of lsquoASP gtbaselinersquo and that of lsquoANSP gt baselinersquo from thefirst-level analyses in each participant were enteredinto second-level one-sample t tests for the patientgroup and the healthy-control group respectively

To reveal the brain regions related to the lsquoASP effect(ASP gt ANSP)rsquo in each participant group contrastimages of lsquoASP gtANSPrsquo from the first-level analysesin each participant were entered into the second-levelgroup one-sample t tests in the healthy control groupand the patient group separately For whole-brain ana-lyses peak signals that were statistically significant atthe p value less than 005 [family-wise error (FWE) cor-rected] were reported

Psychophysiological interaction (PPI) analyses

PPI analyses (Friston et al 1997) were performed toidentify the brain regions showing significantlyincreased functional connectivity with the most criticalbrain structures (seeds) related to the ASP conditioncompared with the ANSP condition The coordinatesof the peak voxel from the contrast of lsquoASP gtANSPrsquoin random-effect analyses were used as the landmarksfor the individual seed voxels A seed region in eachparticipant was defined as a sphere with a 5 mmradius centered at the peak voxel The time series of

840 C Wu et al

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

seed regions were then extracted and the PPI regres-sors which reflected the interaction between psycho-logical variable (ASP v ANSP) and the activationtime course of the seed regions were calculated

The individual contrast images which reflected theeffects of PPI between the seed regions and otherbrain areas were subsequently subjected to thesecond-level one-sample t tests in each of the

participant groups to identify the brain regions show-ing increased co-variation with the activity of theseed regions in analyses of the ASP condition againstthe ANSP condition Then individual participantsrsquo con-trast images were entered into the second-level two-sample t tests for group comparisons In PPI analysespeak signals that were statistically significant at a pvalue less than 005 (FDR corrected) were reported

Fig 1 Illustration of a functional scanning run that comprised 60 trials [20 trials for each of the three conditions auditoryspeech priming (ASP) auditory non-speech priming (ANSP) and non-speech stimulation baseline] presented in randomorder Sparse temporal sampling scanning was used Trial structures of each of the three conditions for the functional run areillustrated separately The temporal midpoint of the sound stimulus was presented 4200 ms prior to the onset of the nextscanning TR Time to repeat

Informational masking of speech 841

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

Functional connectivity analyses (partial correlation)

It has been established that partial correlation can beused as an effective measure of functional connectivitybetween a given pair of brain regions by attenuatingthe contribution of other sources of covariance(Hampson et al 2002 Liu et al 2008 2014) In thisstudy to search for schizophrenia-related changes infunctional connectivity of speech processing-relatedbrain regions both the activation clusters from lsquoASPgt baselinersquo and those from lsquoANSP gt baselinersquo SPMfiles in healthy controls and patient participants wereextracted and used as the seed regions for examiningthe functional connectivity (p lt 005 FWE corrected)(MarsBaR region of interest toolbox for SPM httpmarsbarsourceforgenet) The clusters representingthe same brain region activated by lsquoASP gt baselinersquoand lsquoANSP gt baselinersquo in patients and controls werecombined as the seed for functional connectivity ana-lyses of this brain region The correlation coefficientobtained by partial correlation between a pair of vari-ables after filtering out contributions from all othervariables was included in the dataset (Salvador et al2005 Liu et al 2008 2014) and converted to Z-scoreswith the Fischer transform as the strengths of func-tional connectivity Thus significant differences infunctional connectivity between healthy controls andpatients were identified by significant differences inZ-score (see Liu et al 2014)

Correlation between speech-recognition performance andfunctional connectivity

Spearman correlation analyses were performed usingSPSS 160 software to investigate the correlationbetween the behavioral performance (averaged per-centage correct of target speech recognition across theASP and ANSP listening conditions) and the strengthof functional connectivity (Z-scores of functional con-nectivity) of a pair of brain regions The null hypoth-esis was rejected at the level of 005

Behavioral testing

The behavioral testing was conducted after the fMRIscanning experiment Acoustic signals calibrated bya sound-level meter (AUDit and System 824 LarsonDavis Inc USA) were delivered from a notebook-computer sound card (ATI SB450 AC97) and presentedto participants via earphones (model HDA 600) Thetarget speech level was 60 dB SPL The SPL of thespeech masker was adjusted to produce two SMRsminus4 and minus8 dB There were two within-subject vari-ables (1) priming type (ASP ANSP) and (2) SMR(minus8 minus4 dB) For each participant there were four test-ing conditions and 20 trials (also 20 target-sentence

presentations) for each condition The presentationorder for the four combinations of priming type andSMR were partially counterbalanced across partici-pants using a Latin square order

In a trial the participant who was seated at the cen-ter of a quiet room in the hospital pressed the lsquoEnterrsquokey on a computer keyboard to start the presentationof the auditory priming stimulus Either an ASP stimu-lus or an ANSP stimulus was presented in quiet (with-out the masker presentation) after the key pressImmediately after the prime-presentation phrase thetarget and masker were presented and terminated sim-ultaneously After the maskertarget co-presentationwas finished the participant was instructed to loudlyrepeat the whole target phrases as best as heshecould The experimenters who sat quietly behind theparticipant scored whether each of the two syllablesfor each of the two keywords in the target phraseshad been identified correctly

Target speech recognition performance was valuedby the percentage of correct response of the secondkeywords in target phrases The percentage correctscores were entered into ANOVAs or Spearman correl-ation with the strength of functional connectivity(Z-scores of functional connectivity) of brain regionpairs using SPSS 160 software (USA) The null hypoth-esis was rejected at the level of 005

Results

The speech-priming effect and the participant-groupeffect on speech recognition

Fig 2 shows comparisons in group-mean percentage-correct recognition of the last target keyword (toppanels) and those in group-mean ASP-induced speech-recognition improvement (bottom panels the ASPeffect the difference in percentage correct between theASP condition and the ANSP condition) betweenhealthy listeners and patients when the SMR was eitherminus4 or minus8 dB

To statistically examine the differences in percentageof correct identification of target speech across experi-mental conditions a 2 (group control patient) by 2(priming type ASP ANSP) by 2 (SMR minus4 minus8 dB)three-way ANOVA showed that the main effects ofgroup (F1144 = 13876 p lt 0001) priming type (F1144 =1327 p lt 0001) and SMR (F1144 = 5227 p lt 0001)were all significant However all the two-way interac-tions and the three-way interaction were not signifi-cant Post-hoc tests further confirmed that consistentwith previous studies (Wu et al 2012a 2013b) patientsperformed worse in speech recognition than healthycontrols (Fig 2 top panels) when SMR was either minus4dB (for each of the two priming conditions p lt 0001)

842 C Wu et al

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

or minus8 dB (for each of the two priming conditions p lt0001)

Moreover similar to healthy listeners (Fig 2 bottompanels) patients were able to use the ASP stimulus toimprove their target-speech recognition when the SMRwas either minus4 dB (F172 = 6880 p = 0011) or minus8 dB(F172 = 7192 p = 0009)

Brain regions activated by speech priming

A 2 (group control patient) by 2 (prime type ASPANSP) ANOVA of brain-image data showed that themain effect of priming type (ASP v ANSP) revealedsignificant activation in the bilateral STG bilateral sup-plementary motor area (SMA) bilateral medial super-ior frontal area left posterior middle temporal gyrus(pMTG) right inferior frontal gyrus (IFG) right insu-lar right precentral right putamen and right caudate(Fig 3 top panel and online Supplementary Table S2F contrasts were significant at p lt 005 with FWE cor-rection) The main effect of group (control v patient)elicited significant activation in the left anterior

cingulate cortex left caudate and right putamen at plt 0001 uncorrected but not at p lt 005 correction withFDR (Fig 3 bottom panel) The interaction was notsignificant at either p lt 0001 uncorrected or p lt 005correction with FDR To further explore the brain sub-strates underlying the ASP effects we conducted one-sample t tests and PPI analyses in the healthy controlgroup and the patient group separately

In healthy listeners compared with the ANSP condi-tion introducing the ASP condition significantlyenhanced BOLD signals in the bilateral STG bilateralMTG left pMTG and left putamen (p lt 005 FWE cor-rected) (Fig 4a online Supplementary Table S3)

In patients introducing the ASP condition (againstthe ANSP condition) significantly enhanced BOLD sig-nals only in the left STG and left pMTG (p lt 005 FWEcorrected) (Fig 4a online Supplementary Table S3)Thus both the left STG and left pMTG were activatedby the ASP gt ANSP listening-condition contrast in bothhealthy listeners and patients

We also analysed the data in which the incorrectbutton-press trials were removed in each participant

Fig 2 Top panels Comparisons in group-mean percentage-correct recognition of target keywords between thehealthy-listener group and the patient group when the signalmasker ratio (SMR) was either minus4 or minus8 dB Bottom panelsComparisons in group-mean auditory speech priming (ASP)-induced improvement in recognition of target keywords betweenthe healthy listener group and the patient group when the SMR was either minus4 or minus8 dB ANSP Auditory non-speechpriming plt005

Informational masking of speech 843

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

For healthy controls due to the high correct button-press performance (online Supplementary Fig S1)the brain-image results without incorrect button presstrials were similar to those with both correct trialsand incorrect trials (online Supplementary Fig S2 bot-tom panels and online Supplementary Table S4) Forpatient participants however activation in the bilat-eral STG left MTG and left pMTG became enhancedby removal of incorrect trials (online SupplementaryFig S2 bottom panels and online SupplementaryTable S4)

Functional connectivity of the left STGpMTG inspeech priming

Since both the left STG and left MTG were activated bythe ASP gtANSP condition contrast in both healthy listen-ers and patients PPI analyses were then conducted toidentify the brain regions showing enhanced co-variationspecifically with the left STGMTG induced by the ASP gtANSP contrast In healthy listeners enhanced functionalconnectivity was observed (1) between the left STG andthe following brain regions the left middle frontalgyrus left pars triangularis of the IFG (TriIFG) left insu-lar left caudate left MTG and bilateral putamen and (2)between the left pMTG and the following brain regionsthe left TriIFG left insular left caudate and left supramarginal area (Fig 4b online Supplementary Table S5)

In patients enhanced functional connectivity includedthe ones (1) between the left STG and the following twobrain regions the left Rolandic operculum and the leftcaudate and (2) between the left pMTG and the follow-ing two brain regions the left Rolandic operculum andthe left caudate (Fig 4b online Supplementary Table S5)

Thus the left caudate was the only brain structureshowing significantly enhanced functional connectiv-ity with both the left STG and left pMTG in bothhealthy listeners and patients Moreover particularlyin healthy listeners the left STGpMTG had functionalconnectivity with the left TriIFG and left insular par-ticularly in patients the left STGpMTG had functionalconnectivity with the left Rolandic operculum Nosignificant differences in PPI were found betweenhealthy controls and patients in two-sample t tests

Functional connectivity related to speech recognitionunder speech-masking conditions

Brain regions activated by either the ASP listeningcondition or the ANSP listening condition

As mentioned in the Introduction listeners withschizophrenia perform worse than healthy listenersin speech recognition against informational masking(Wu et al 2012a 2013b) In this study the (non-speech)noise baseline-stimulation condition was used as thereference for examining the brain regions and

Fig 3 Top panels Voxels that exhibited a main effect of priming type The activation map is thresholded at p lt 0001uncorrected and overlaid on the group-average structural image The color scale indicates the p value corrected family-wisefor type I error (FWE) Bottom panels Voxels that exhibited a main effect of group type The activation map is thresholded atp lt 0001 uncorrected and overlaid on the structural image averaged across control group and patient group no voxels survivethe p value false discovery rate corrected for type I error

844 C Wu et al

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

functional connectivity induced by the speech-listeningconditions

In both healthy listeners and patient listeners com-pared with the (non-speech) noise baseline-stimulationcondition the ASP listening condition significantly

enhanced BOLD signals in the bilateral STG left puta-men left TriIFG right precentral and right SMA (p lt005 FWE corrected) (online Supplementary Fig S3and online Supplementary Table S6) In addition spe-cifically in healthy controls the activated brain regions

Fig 4 (a) Activated brain regions associated with the contrast of the auditory speech priming (ASP) listening conditionagainst the auditory non-speech priming (ANSP) listening condition in healthy listeners and patients with schizophrenia Theactivation maps were thresholded at p lt 005 (family-wise error corrected) and overlaid on the group-average structural image(b) Psychophysiological interaction analyses in healthy listeners (middle column) and listeners with schizophrenia (rightcolumn) for revealing the ASP effect (ASP gtANSP)-related functional connectivity of the left superior temporal gyrusmiddletemporal gyrus (STGMTG) Locations of seed regions (left column) are overlaid on the template of SPM8 and the activationmaps are overlaid on a template brain with inflated cortex of SPM8 All peaks are significant at p lt 005 (false discovery ratecorrected) pMTG Posterior middle temporal gyrus IFG inferior frontal gyrus L left

Informational masking of speech 845

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

also included the right putamen and bilateral thal-amus Specifically in patients the activated brainregions also included the left precentral left inferiorparietal lobule (IPL) right insular right lingual andvermis of cerebellum (p lt 005 FWE corrected)

Also as shown by online Supplementary Fig S3 andonline Supplementary Table S6 compared with thenon-speech baseline-stimulation condition in healthycontrols the ANSP condition significantly enhancedBOLD signals in the bilateral STG right putamenand right SMA left precentral and left TriIFG (p lt005 FWE corrected) In patients the ANSP conditionsignificantly enhanced BOLD signals in the bilateralSTG bilateral insular left pars opercularis of the IFG(OperIFG) left IPL left postcentral left cerebellumright mid-cingulate cortex (MCC) right precentraland right lingual cortex (p lt 005 FWE corrected)

Schizophrenia-related changes in functional connectivity forspeech listening

Based on the results of brain regions activated by eitherthe ASP listening condition (ASP gt baseline) or theANSP listening conditions (ANSP gt baseline) func-tional connectivity of these activated brain areas wereexamined using partial correlation analyses Thebrain region clusters activated by either the ASP- orANSP-listening condition (against the non-speechbaseline listening condition) were used as the seedregions (online Supplementary Table S6) (p lt 005FWE corrected) for conducting the comparison in func-tional connectivity between healthy listeners andpatients More specifically in a brain region clustersactivated by the ASP-listening condition and those bythe ANSP-listening condition were combined acrosshealthy listeners and patient listeners to form a seedregion for functional connectivity analyses (forexample all the four activated STG clusters shown inonline Supplementary Table S6 across the two primingconditions and the two participant groups were com-bined to form the STG seed region for the functionalconnectivity analyses) In total 20 clusters of seedregions were obtained and used for examining speechrecognition-related functional connectivity in bothhealthy controls and patients the left STG right STGleft putamen right putamen left TriIFG right TriIFGleft OperIFG left precentral right precentral rightSMA left insular right insular left IPL left postcen-tral right MCC left thalamus right thalamus rightlingual left cerebellum and vermis

Then Z-scores (which represented the strength offunctional connectivity) were obtained based on partialcorrelation analyses between time series of the total of190 seed-region pairs for each participant and com-pared between healthy controls and patients to reveal

the significantly changed functional connectivity(which was related to the schizophrenia-related poorperformance in target-speech recognition) Fig 5ashows Z-scores of all the seed-region pairs whosestrengths of functional connectivity were significantlylower in patients than in healthy listeners

Correlation between strength of functional connectivity andspeech-recognition performance

Spearman correlation analyses were conducted betweenthe percentage correct of target-speech recognition andeach of the Z-scores for the seed-region pairs shownin Fig 5a in both healthy listeners and patients Theresults showed that the percentage correct of target-speech recognition was significantly correlated onlywith the Z-score for functional connectivity betweenthe left STG and the left TriIFG in healthy controlswhen the SMR was minus8 dB (r = 0512 p = 0048) and inpatients when the SMR was either ndash8 dB (r = 0488 p =0021) or minus4 dB (r = 0552 p = 0008) (Fig 5b and c)Also not only was speech-recognition performance inindividual patients generally poorer than that in indi-vidual healthy listeners but also the Z-scores (represent-ing the strength of functional connectivity) in individualpatients were lower than those in individual healthylisteners

Discussion

Brain substrates activated by ASP

This study confirms previous reports that in bothhealthy listeners (Freyman et al 2004 Yang et al2007 Helfer amp Freyman 2009 Huang et al 2010 Wuet al 2012b 2013a) and listeners with schizophrenia(Wu et al 2012b) target-speech recognition againstinformational masking can be improved by temporallypre-presenting an early part of the target speech (iethe speech prime) The behavioral results of thisstudy did not show a significant difference in the prim-ing (unmasking) effect on speech recognition betweenthe two participant groups More importantly thisstudy for the first time reveals the brain substrates spe-cifically underlying the speech prime-induced facilitat-ing effect on speech recognition against speechmasking in both healthy listeners and listeners withschizophrenia

In healthy listeners the BOLD-signal contrastbetween the ASP condition and the ANSP conditionreveals significant enhancement of activation in the fol-lowing brain regions the bilateral STG bilateralpMTG left MTG left superior temporal pole and leftputamen In listeners with schizophrenia the signifi-cantly activated brain regions include the left STGand left pMTG Since both the left STG and left

846 C Wu et al

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

pMTG are activated by the ASP gt ANSP contrast inboth healthy listeners and listeners with schizophreniathese two temporal lobe areas must be the most criticalto the unmasking effect of ASP

It has been suggested that the STG particularly theanterior part of the STG is an early stage in the corticalnetwork for speech perception mediating speech-sound identity information (Hickok amp Poeppel 2004Ahveninen et al 2006 Rauschecker amp Scott 20092015) The involvement of the left STG in speech per-ception at the phonetic lexicalndashsemantic and syntacticlevels has been confirmed by both brain imaging(Friederici et al 2003 Scott amp Wise 2003) and func-tional lesion studies (Boatman 2004) The left pMTGis involved in retrieval of lexicalndashsyntactic information

from the mental lexicon in long-term memory (Snijderset al 2009) and has a functional relationship with theleft IFG in both processing syntactic structures andconstructing grammatical representations of spokenlanguage (Tyler et al 2010 2013 Walz et al 2010Papoutsi et al 2011) Also the left pMTG is involvedin sentence-level prosody processing in Chinese-speech perception (Tong et al 2005) Moreover previ-ous studies have confirmed both the function of theleft rostral STG in working memory of sound identities(Arnott et al 2005) and the function of the left pMTG inthe storage of lexical representation (Lau et al 2008)These memory functions of the left STGpMTG areparticularly critical to the memory-based speech prim-ing effect Thus the speech priming-induced activation

Fig 5 (a) Strengths of functional connectivity (Z-score) of brain seed-region pairs which were significantly different betweenhealthy listeners and patients (b) and (c) Individual participantsrsquo Z-scores for functional connectivity (the abscissa) betweenthe left superior temporal gyrus (L STG) and the left pars triangularis of the inferior frontal gyrus (L TriIFG) weresignificantly correlated with individual participantsrsquo percentage correct recognition of target speech (the ordinate) in thepatient group when the signalmasker ratio (SMR) was minus4 or minus8 dB and in the healthy listener group when the SMR wasminus8 dB R Right Pu putamen PreC precentral

Informational masking of speech 847

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

of the left STGpMTG indicates a speech priming-related facilitation of cortical representation of target-speech signals

It should be noted that in this study only younger-adult femalesrsquo voices were used Since previous studieshave shown that male and female voices are repre-sented as distinct auditory objects in the human non-primary auditory cortex (Weston et al 2015) it isnecessary in the future to examine whether similarresults can be obtained when male-voice stimuli areused

ASP-related functional connectivity of the left STGpMTG

Relative to the ANSP condition the ASP conditionenhances functional connectivity of the left STGpMTG with the left caudate in both healthy listenersand listeners with schizophrenia In addition introdu-cing the ASP condition (against the ANSP condition)enhances functional connectivity of the left STGpMTG with the left TriIFG MFG supramarginalgyrus insular cortex and putamen in healthy listenersand with the left Rolandic operculum in listeners withschizophrenia

The ASP-related functional connectivity of the leftSTGpMTG in healthy listeners suggests some criticalmechanisms normally underlying the unmaskingeffect of ASP (1) since the caudate contributes tospeech inhibition and even more general responseinhibition (Menon et al 2001 Ketteler et al 2008 Liet al 2008 Ali et al 2010) and the left IFG plays arole in signal selections among competing sources(Thompson-Schill et al 2005) the unmasking effect ofASP should be based on both selective attention to tar-get speech and suppression of disruptive speech sig-nals (2) the left IFG also contributes to multiplespeech processes including syntactic unifications onthe basis of interplay with the left pMTG speechmeaning selection sentence re-interpretation andspeech production (Herholz et al 1996 Paulesu et al1997 Papathanassiou et al 2000 Schuhmann et al2009 Snijders et al 2009 Rodd et al 2012) Since theinvolvement of the speech-motor system in speech per-ception is important for facilitating speech perceptionparticularly under lsquococktail partyrsquo listening conditions(Wu et al 2014) functional connectivity of the leftSTGpMTG with the left TriIFG normally contributesto the unmasking effect of ASP through not only facili-tating masking-speech suppression and target-speechrepresentation (eg syntactic unification) but alsoinducing speech production

The results of this study reveal that introducing theASP condition (against the ANSP condition) bothactivates the STGpMTG and enhances functional

connectivity between the STGpMTG and the frontalcortex Thus it is of importance to know whether theenhanced functional connectivity between the STGpMTG and the frontal cortex indicates a top-downmodulation mechanism underlying the ASP-inducedunmasking of speech

It is known that the left Rolandic operculum isinvolved in sentence-level speech prosody processing(Ischebeck et al 2008) speech production (Tonkonogyamp Goodglass 1981) and syntactic encoding duringspeech production (Indefrey et al 2001) It is not clearwhether this schizophrenia-enhanced functional con-nectivity between the left STGpMTG and the leftRolandic operculum reflects schizophrenia-relatedneural plasticity underlying the unmasking effect ofthe speech-priming stimulus The compensatory mech-anism specifically underlying schizophrenia is animportant issue (Tan et al 2007)

Functional connectivity specifically related toschizophrenia-induced deficits in speech recognitionagainst informational masking

The behavioral-testing results of this study supportprevious reports that under lsquococktail partyrsquo listeningconditions (Schneider et al 2007) listeners with schizo-phrenia perform poorly in recognizing target speech(Wu et al 2012a 2013b Zheng et al 2016) This studyalso reveals the brain substrates underlying theschizophrenia-related deficits in speech recognitionunder the priming conditions The strength of functionalconnectivity between a number of speech-processing-related brain regions (including the STG MTG putamenTriIFG IPL precentral cortex and SMA) declines in lis-teners with schizophrenia Particularly in listeners withschizophrenia the strength of functional connectivitybetween the left STGpMTG and the left TriIFG is notonly correlated with speech-recognition performancebut is also lower than that in healthy listenersIwashiro et al (2012) have shown that the gray mattervolume reduces in the TriIFG in both individuals at clin-ical high-risk for psychosis and those with first episodeof schizophrenia The declined functional connectivitybetween the left STGpMTG and left TriIFG as revealedby this study further confirms the schizophrenia-relatedfunctional deficits in the left TriIFG

In addition to the involvement of the STG MTG andTriIFG in speech processing as discussed above the puta-men engages in both syntactic processing (Friederici et al2003) and speech initiation through its functional con-nectivity with both the left IFG and the left lateral tem-poral cortex (Booth et al 2007) Moreover the IPL isinvolved in interpreting sensory information (Zhang ampLi 2014) and works as a sensorimotor interface betweenthe representation of acoustic signals in the auditory

848 C Wu et al

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

cortex and the prediction of sensory consequences ofarticulatory gestures in the premotor cortex (Olson ampBerryhill 2009 Price 2012 Du et al 2014) Also the pre-central gyrus is part of the premotor cortex involved inspeech processing (Callan et al 2010) including phonemesegmentation (Sato et al 2009) categorical speech percep-tion in noise (Du et al 2014) and speech production(Price 2012) Finally the SMA plays a role in planningpreparing controlling and executing complex move-ments (Nachev et al 2008 Price 2012 Laviolette et al2013 Gao et al 2014) The reduced strength of functionalconnectivity between the STG MTG putamen TriIFGIPL precentral cortex and SMA may also account forthe declined speech recognition under speech-maskingconditions in listeners with schizophrenia Indeed previ-ous studies have shown that functional connectivitybetween these brain regions is not only involved in audi-tory andor speech processing (Jeong et al 2009Mhuircheartaigh et al 2010 Price 2010 Turken ampDronkers 2011 Sundermann amp Pfleiderer 2012 Zhangamp Li 2014 Kireev et al 2015 Munoz-Lopez et al 2015)but also impaired in people with schizophrenia (Fordet al 2002 Liu et al 2012 Leroux et al 2014 Pu et al2014 Zhang amp Li 2014) Moreover this study revealsthat the strength of functional connectivity between theleft STG and left TriIFG is significantly correlated withthe speech-recognition performance under the informa-tional masking condition in both healthy listeners and lis-teners with schizophrenia and reduces in listeners withschizophrenia More studies are needed to further exam-ine how schizophrenia-related deficits in functional con-nectivity between the left STG and left TriIFG affectrecognition of speech under the informational maskingcondition

Conclusions

(1) Under lsquococktail partyrsquo listening conditions bothhealthy listeners and listeners with schizophreniaare able to use temporally pre-presented ASP stim-uli to improve their target-speech recognitionagainst informational speech masking The mostcritical brain regions underlying this unmaskingeffect are the left STG and the left pMTG in bothhealthy listeners and listeners with schizophrenia

(2) In both healthy listeners and listeners with schizo-phrenia the left STGpMTG has ASP-related func-tional connectivity with the left caudate suggestingthat the inhibitory function of the left caudate isinvolved in the unmasking effect of ASP

(3) Normally the left STG and left pMTG haveASP-related functional connectivity with the leftTriIFG suggesting an enhanced functional integra-tion among the following three processes speech-

signal selection speech-signal representation andspeech production

(4) In listeners with schizophrenia enhanced ASP-related functional connectivity of the left STGpMTG and the left Rolandic operculummay indicatean ASP-related facilitation in sentence prosody pro-cessing and speech production

(5) The poor performance of speech recognitionagainst informational masking in people withschizophrenia is associated with reduced strengthof functional connectivity between a number ofspeech-processing-related brain regions particu-larly associated with the impaired functional con-nectivity between the left STG and the left TriIFG

Supplementary material

The supplementary material for this article can befound at httpsdoiorg101017S0033291716002816

Acknowledgements

Huahui Li assisted in many aspects of this work Thiswork was supported by the lsquo973rsquo National BasicResearch Program of China (2015|CB3518002011CB707805) the National High Technology Researchand Development Program of China (863 Program2015AA016306) the National Natural ScienceFoundation of China (31170985) the Planned Scienceand Technology Projects of Guangzhou (2014Y2-00105)the Guangzhou Municipal Key Disciplines in Medicinefor Guangzhou Brain Hospital (GBH2014-QN04GBH2014-ZD06) the Chinese National Key ClinicalProgram in Psychiatry (201201004) and the ChinaPostdoctoral Science Foundation General Program(2013M530453)

Declaration of Interest

None

References

Ahveninen J Jaaskelainen IP Raij T Bonmassar G DevoreS Hamalainen M Levanen S Lin FH Sams MShinn-Cunningham BG Witzel T Belliveau JW (2006)Task-modulated lsquowhatrsquo and lsquowherersquo pathways in humanauditory cortex Proceedings of the National Academy ofSciences USA 103 14608ndash14613

Ali N Green DW Kherif F Devlin JT Price CJ (2010) Therole of the left head of caudate in suppressing irrelevantwords Journal of Cognitive Neuroscience 22 2369ndash2386

Arnott SR Grady CL Hevenor SJ Graham S Alain C(2005) The functional organization of auditory workingmemory as revealed by fMRI Journal of CognitiveNeuroscience 17 819ndash831

Informational masking of speech 849

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

Boatman D (2004) Cortical bases of speech perceptionevidence from functional lesion studies Cognition 9247ndash65

Booth JR Wood L Lu D Houk JC Bitan T (2007) The role ofthe basal ganglia and cerebellum in language processingBrain Research 1133 136ndash144

Braff DL Light GA (2005) The use of neurophysiologicalendophenotypes to understand the genetic basis ofschizophrenia Dialogues in Clinical Neuroscience 7 125ndash135

Brungart DS Simpson BD Ericson MA Scott KR (2001)Informational and energetic masking effects in theperception of multiple simultaneous talkers Journal of theAcoustical Society of America 110 2527ndash2538

Callan D Callan A Gamez M Sato MA Kawato M (2010)Premotor cortex mediates perceptual performanceNeuroImage 51 844ndash858

Cherry EC (1953) Some experiments on the recognition ofspeech with one and with two ears Journal of the AcousticalSociety of America 25 975ndash979

Davis MH Johnsrude IS Hervais-Adelman A Taylor KMcGettigan C (2005) Lexical information drives perceptuallearning of distorted speech evidence from thecomprehension of noise-vocoded sentences Journal ofExperimental Psychology General 134 222ndash241

Ding N Simon JZ (2012) Emergence of neural encoding ofauditory objects while listening to competing speakersProceedings of the National Academy of Sciences USA 10911854ndash11859

Du Y Buchsbaum BR Grady CL Alain C (2014) Noisedifferentially impacts phoneme representations in theauditory and speech motor systems Proceedings of theNational Academy of Sciences USA 111 7126ndash7131

Evans S McGettigan C Agnew ZK Rosen S Scott SK(2016) Getting the cocktail party started masking effectsin speech perception Journal of Cognitive Neuroscience 28483ndash500

First MB Gibbon M Spitzer RL Williams JBW (1996)Structured Clinical Interview for DSM-IV Axis I Disorders(SCID) Clinician Version American Psychiatric PressWashington DC

Ford JM Mathalon DH Whitfield S Faustman WO RothWT (2002) Reduced communication between frontal andtemporal lobes during talking in schizophrenia BiologicalPsychiatry 51 485ndash492

Freyman RL Balakrishnan U Helfer KS (2004) Effect ofnumber of masking talkers and auditory priming oninformational masking in speech recognition Journal of theAcoustical Society of America 115 2246ndash2256

Freyman RL Helfer KS McCall DD Clifton RK (1999) Therole of perceived spatial separation in the unmasking ofspeech Journal of the Acoustical Society of America 106 3578ndash3588

Friederici AD Ruschemeyer SA Hahne A Fiebach CJ(2003) The role of left inferior frontal and superior temporalcortex in sentence comprehension localizing syntactic andsemantic processes Cerebral Cortex 13 170ndash177

Friston KJ Buechel C Fink GR Morris J Rolls E Dolan RJ(1997) Psychophysiological and modulatory interactions inneuroimaging NeuroImage 6 218ndash229

Friston KJ Williams S Howard R Frackowiak RS Turner R(1996) Movement-related effects in fMRI time-seriesMagnetic Resonance in Medicine 35 346ndash355

Gao Q Tao Z Zhang M Chen H (2014) Differentialcontribution of bilateral supplementary motor area to theeffective connectivity networks induced by task conditionsusing dynamic causal modeling Brain Connectivity 4 256ndash264

Gottesman II Gould TD (2003) The endophenotype conceptin psychiatry etymology and strategic intentions AmericanJournal of Psychiatry 160 636ndash645

Hall DA Haggard MP Akeroyd MA Palmer ARSummerfield AQ Elliott MR Gurney EM Bowtell RW(1999) lsquoSparsersquo temporal sampling in auditory fMRIHuman Brain Mapping 7 213ndash223

Hampson M Peterson BS Skudlarski P Gatenby JC GoreJC (2002) Detection of functional connectivity usingtemporal correlations in MR images Human Brain Mapping15 247ndash262

Helfer KS Freyman RL (2005) The role of visual speech cuesin reducing energetic and informational masking Journal ofthe Acoustical Society of America 117 842ndash849

Helfer KS Freyman RL (2009) Lexical and indexical cues inmasking by competing speech Journal of the AcousticalSociety of America 125 447ndash456

Herholz K Thiel A Wienhard K Pietrzyk U vonStockhausen HM Karbe H Kessler J Bruckbauer THalber M Heiss WD (1996) Individual functionalanatomy of verb generation NeuroImage 3 185ndash194

Hickok G Poeppel D (2004) Dorsal and ventral streams aframework for understanding aspects of the functionalanatomy of language Cognition 92 67ndash99

Huang Y Huang Q Chen X Qu T Wu X Li L (2008)Perceptual integration between target speech and target-speech reflection reduces masking for target-speechrecognition in younger adults and older adults HearingResearch 244 51ndash65

Huang Y Huang Q Chen X Wu X Li L (2009) Transientauditory storage of acoustic details is associated withrelease of speech from informational masking inreverberant conditions Journal of Experimental PsychologyHuman Perception and Performance 35 1618ndash1628

Huang Y Xu L Wu X Li L (2010) The effect of voice cuing onreleasing speech from informational masking disappears inolder adults Ear and Hearing 31 579ndash583

Indefrey P Brown CM Hellwig F Amunts K Herzog HSeitz RJ Hagoort P (2001) A neural correlate of syntacticencoding during speech production Proceedings of theNational Academy of Sciences USA 98 5933ndash5936

Ischebeck AK Friederici AD Alter K (2008) Processingprosodic boundaries in natural and hummed speech anfMRI study Cerebral Cortex 18 541ndash552

Iwashiro N Suga M Takano Y Inoue H Natsubori TSatomura Y Koike S Yahata N Murakami M Katsura MGonoi W Sasaki H Takao H Abe O Kasai K Yamasue H(2012) Localized gray matter volume reductions in the parstriangularis of the inferior frontal gyrus in individuals atclinical high-risk for psychosis and first episode forschizophrenia Schizophrenia Research 137 124ndash131

850 C Wu et al

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

Jeong B Wible CG Hashimoto R Kubicki M (2009)Functional and anatomical connectivity abnormalities inleft inferior frontal gyrus in schizophrenia Human BrainMapping 30 4138ndash4151

Ketteler D Kastrau F Vohn R Huber W (2008) Thesubcortical role of language processing High levellinguistic features such as ambiguity-resolution and thehuman brain an fMRI study NeuroImage 39 2002ndash2009

Kireev M Slioussar N Korotkov AD Chernigovskaya TVMedvedev SV (2015) Changes in functional connectivitywithin the fronto-temporal brain network induced byregular and irregular Russian verb production Frontiers inHuman Neuroscience 9 36

Lau EF Phillips C Poeppel D (2008) A cortical network forsemantics (de)constructing the N400 Nature ReviewsNeuroscience 9 920ndash933

Laviolette L Nieacuterat M-C Hudson AL Raux M Allard EacuteSimilowski T (2013) The supplementary motor area exertsa tonic excitatory influence on corticospinal projections tophrenic motoneurons in awake humans PLOS ONE 8e62258

Leroux E Delcroix N Dollfus S (2014) Left fronto-temporaldysconnectivity within the language network inschizophrenia an fMRI and DTI study Psychiatry Research223 261ndash267

Li CS Yan P Sinha R Lee TW (2008) Subcortical processesof motor response inhibition during a stop signal taskNeuroImage 41 1352ndash1363

Li L Daneman M Qi JG Schneider BA (2004) Does theinformation content of an irrelevant source differentiallyaffect spoken word recognition in younger and olderadults Journal of Experimental Psychology Human Perceptionand Performance 30 1077ndash1091

Liu H Kaneko Y Ouyang X Li L Hao Y Chen EY Jiang TZhou Y Liu Z (2012) Schizophrenic patients and theirunaffected siblings share increased resting-stateconnectivity in the task-negative network but not itsanticorrelated task-positive network Schizophrenia Bulletin38 285ndash294

Liu Y Liang M Zhou Y He Y Hao Y Song M Yu C Liu HLiu Z Jiang T (2008) Disrupted small-world networks inschizophrenia Brain 131 945ndash961

Liu Y Yu C Zhang X Liu J Duan Y Alexander-Bloch AFLiu B Jiang T Bullmore E (2014) Impaired longdistance functional connectivity and weighted networkarchitecture in Alzheimerrsquos disease Cerebral Cortex 241422ndash1435

Menon V Adleman NE White CD Glover GH Reiss AL(2001) Error-related brain activation during a GoNoGoresponse inhibition task Human Brain Mapping 12 131ndash143

Mhuircheartaigh RN Rosenorn-Lanng D Wise R Jbabdi SRogers R Tracey I (2010) Cortical and subcorticalconnectivity changes during decreasing levels ofconsciousness in humans a functional magnetic resonanceimaging study using propofol Journal of Neuroscience 309095ndash9102

Munoz-Lopez M Insausti R Mohedano-Moriano AMishkin M Saunders RC (2015) Anatomical pathways forauditory memory II information from rostral superior

temporal gyrus to dorsolateral temporal pole and medialtemporal cortex Frontiers in Neuroscience 9 158

Nachev P Kennard C Husain M (2008) Functional role ofthe supplementary and pre-supplementary motor areasNature Reviews Neuroscience 9 856ndash869

Olson IR Berryhill M (2009) Some surprising findings onthe involvement of the parietal lobe in human memoryNeurobiology of Learning and Memory 91 155ndash165

Papathanassiou D Etard O Mellet E Zago L Mazoyer BTzourio-Mazoyer N (2000) A common language networkfor comprehension and production a contribution to thedefinition of language epicenters with PET NeuroImage 11347ndash357

Papoutsi M Stamatakis EA Griffiths J Marslen-WilsonWD Tyler LK (2011) Is left fronto-temporal connectivityessential for syntax Effective connectivity tractographyand performance in left-hemisphere damaged patientsNeuroImage 58 656ndash664

Paulesu E Goldacre B Scifo P Cappa SF Gilardi MCCastiglioni I Perani D Fazio F (1997) Functionalheterogeneity of left inferior frontal cortex as revealed byfMRI Neuroreport 8 2011ndash2017

Price CJ (2010) The anatomy of language a review of 100fMRI studies published in 2009 Annals of the New YorkAcademy of Sciences 1191 62ndash88

Price CJ (2012) A review and synthesis of the first 20 years ofPET and fMRI studies of heard speech spoken languageand reading NeuroImage 62 816ndash847

Pu W Rolls ET Guo S Liu H Yu Y Xue Z Feng J Liu Z(2014) Altered functional connectivity links inneuroleptic-naive and neuroleptic-treated patients withschizophrenia and their relation to symptoms includingvolition NeuroImage Clinical 6 463ndash474

Rakerd B Aaronson NL HartmannWM (2006) Release fromspeech-on-speech masking by adding a delayed masker at adifferent location Journal of the Acoustical Society of America119 1597ndash1605

Rauschecker JP Scott SK (2009) Maps and streams in theauditory cortex nonhuman primates illuminate humanspeech processing Nature Neuroscience 12 718ndash724

Rauschecker JP Scott SK (2015) Pathways and streams in theauditory cortex an update on how work in nonhumanprimates has contributed to our understanding of humanspeech processing In Neurobiology of Language (ed GHickok and SL Small) pp 287ndash298 Elsevier Inc WalthamMA

Rodd JM Johnsrude IS Davis MH (2012) Dissociatingfrontotemporal contributions to semantic ambiguityresolution in spoken sentences Cerebral Cortex 22 1761ndash1773

Salvador R Suckling J Schwarzbauer C Bullmore E (2005)Undirected graphs of frequency-dependent functionalconnectivity in whole brain networks PhilosophicalTransactions of the Royal Society of London Series B BiologicalSciences 360 937ndash946

Sato M Tremblay P Gracco VL (2009) A mediating role ofthe premotor cortex in phoneme segmentation Brain andLanguage 111 1ndash7

Schneider BA Li L Daneman M (2007) How competingspeech interferes with speech comprehension in everyday

Informational masking of speech 851

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at

listening situations Journal of the American Academy ofAudiology 18 559ndash572

Schuhmann T Schiller NO Goebel R Sack AT (2009) Thetemporal characteristics of functional activation in Brocarsquosarea during overt picture naming Cortex 45 1111ndash1116

Scott SK McGettigan C (2013) The neural processing ofmasked speech Hearing Research 303 58ndash66

Scott SK Rosen S Beaman CP Davis JP Wise RJ (2009) Theneural processing of masked speech evidence for differentmechanisms in the left and right temporal lobes Journal ofthe Acoustical Society of America 125 1737ndash1743

Scott SK Rosen S Wickham L Wise RJ (2004) A positronemission tomographystudyof theneural basisof informationaland energetic masking effects in speech perception Journal ofthe Acoustical Society of America 115 813ndash821

Scott SK Wise RJ (2003) PET and fMRI studies of the neuralbasis of speech perception Speech Communication 41 23ndash34

Si T Yang J Shu L Wang X Kong Q Zhou M Li X (2004)The reliability validity of PANSS (Chinese version) and itsimplication Chinese Mental Health Journal 18 45ndash47

Snijders TM Vosse T Kempen G Van Berkum JJPetersson KM Hagoort P (2009) Retrieval and unificationof syntactic structure in sentence comprehension an fMRIstudy using word-category ambiguity Cerebral Cortex 191493ndash1503

Sohoglu E Peelle JE Carlyon RP Davis MH (2012)Predictive top-down integration of prior knowledge duringspeech perception Journal of Neuroscience 32 8443ndash8453

Sohoglu E Peelle JE Carlyon RP Davis MH (2014)Top-down influences of written text on perceived clarity ofdegraded speech Journal of Experimental Psychology HumanPerception and Performance 40 186ndash199

Sundermann B Pfleiderer B (2012) Functional connectivityprofile of the human inferior frontal junction involvementin a cognitive control network BMC Neuroscience 13 119

Tan HY Callicott JH Weinberger DR (2007) Dysfunctionaland compensatory prefrontal cortical systems genes and thepathogenesis of schizophrenia Cerebral Cortex 17 i171ndashi181

Thompson-Schill SL Bedny M Goldberg RF (2005) Thefrontal lobes and the regulation of mental activity CurrentOpinion in Neurobiology 15 219ndash224

Tong Y Gandour J Talavage T Wong D Dzemidzic M XuY Li X Lowe M (2005) Neural circuitry underlyingsentence-level linguistic prosody NeuroImage 28 417ndash428

Tonkonogy J Goodglass H (1981) Language function foot ofthe third frontal gyrus and rolandic operculum Archives ofNeurology 38 486ndash490

Turken AU Dronkers NF (2011) The neural architecture ofthe language comprehension network converging evidencefrom lesion and connectivity analyses Frontiers in SystemsNeuroscience 5 1

Tyler LK Cheung TP Devereux BJ Clarke A (2013)Syntactic computations in the language networkcharacterizing dynamic network properties usingrepresentational similarity analysis Frontiers in Psychology4 271

Tyler LK Wright P Randall B Marslen-Wilson WDStamatakis EA (2010) Reorganization of syntacticprocessing following left-hemisphere brain damage doesright-hemisphere activity preserve function Brain 1333396ndash3408

Walz JM Goldman RI Carapezza M Muraskin J BrownTR Sajda P (2013) Simultaneous EEG-fMRI revealstemporal evolution of coupling between supramodalcortical attention networks and the brainstem Journal ofNeuroscience 33 19212ndash19222

Weston PSJ Hunter MD Sokhi DS Wilkinson IDWoodruff PWR (2015) Discrimination of voice gender inthe human auditory cortex NeuroImage 105 208ndash214

Wild CJ Davis MH Johnsrude IS (2012) Human auditorycortex is sensitive to the perceived clarity of speechNeuroImage 60 1490ndash1502

Woods SW (2003) Chlorpromazine equivalent doses for thenewer atypical antipsychotics Journal of Clinical Psychiatry64 663ndash667

Wu C Cao S Wu X Li L (2013a) Temporally pre-presentedlipreading cues release speech from informationalmasking Journal of the Acoustical Society of America 133El281ndashEl285

Wu C Cao S Zhou F Wang C Wu X Li L (2012a) Maskingof speech in people with first-episode schizophrenia andpeople with chronic schizophrenia Schizophrenia Research134 33ndash41

Wu C Li H Tian Q Wu X Wang C Li L (2013b)Disappearance of the unmasking effect of temporallypre-presented lipreading cues on speech recognition inpeople with chronic schizophrenia Schizophrenia Research150 594ndash595

Wu M Li H Gao Y Lei M Teng X Wu X Li L (2012b)Adding irrelevant information to the content prime reducesthe prime-induced unmasking effect on speech recognitionHearing Research 283 136ndash143

Wu X Wang C Chen J Qu H Li W Wu Y Schneider BA LiL (2005) The effect of perceived spatial separation oninformational masking of Chinese speech Hearing Research199 1ndash10

Wu Z Chen M Wu X Li L (2014) Interaction betweenauditory and motor systems in speech perceptionNeuroscience Bulletin 30 490ndash496

Yang Z Chen J Huang Q Wu X Wu Y Schneider BALiang L (2007) The effect of voice cuing on releasingChinese speech from informational masking SpeechCommunication 49 892ndash904

Zhang S Li C-SR (2014) Functional clustering of the humaninferior parietal lobule by whole-brain connectivitymapping of resting-state functional magnetic resonanceimaging signals Brain Connectivity 4 53ndash69

Zheng Y Wu C Li J Wu H She S Liu S Wu H Mao LNing Y Li L (2016) Brain substrates of perceived spatialseparation between speech sources under simulatedreverberant listening conditions in schizophreniaPsychological Medicine 46 477ndash491

852 C Wu et al

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0033291716002816Downloaded from httpswwwcambridgeorgcore Beijing Normal University on 21 Mar 2017 at 003150 subject to the Cambridge Core terms of use available at