EEG correlates of self-referential processing

14 Pages·2013·0.68 MB·English

by  Gennady G. Knyazev

REVIEWARTICLE
HUMAN NEUROSCIENCE published:06June2013
doi:10.3389/fnhum.2013.00264
EEG correlates of self-referential processing
GennadyG.Knyazev*
InstituteofPhysiology,SiberianBranchofRussianAcademyofMedicalSciences,Novosibirsk,Russia
Editedby: Self-referentialprocessinghasbeenprincipallyinvestigatedusingfunctionalmagneticres-
GeorgNorthoff,UniversityofOttawa, onance imaging (fMRI). However, understanding of the brain functioning is not possible
Canada
withoutcarefulcomparisonoftheevidencecomingfromdifferentmethodologicaldomains.
Reviewedby:
This paper aims to review electroencephalographic (EEG) studies of self-referential pro-
AlexanderFingelkurts,BM-Science
BrainandMindTechnologies cessing and to evaluate how they correspond, complement, or contradict the existing
ResearchCentre,Finland fMRI evidence.There are potentially two approaches to the study of EEG correlates of
AndrewFingelkurts,BM-Science self-referentialprocessing.Firstly,becausesimultaneousregistrationofEEGandfMRIhas
BrainandMindTechnologies
becomepossible,thedegreeofoverlapbetweenthesetwosignalsinbrainregionsrelated
ResearchCentre,Finland
RexCannon,UniversityofTennessee, toself-referentialprocessingcouldbedetermined.Secondandmoredirectapproachwould
USA bethestudyofEEGcorrelatesofself-referentialprocessingperse.Inthisreview,Idiscuss
*Correspondence: studies,whichemployedboththeseapproachesandshowthatinlinewithfMRIevidence,
GennadyG.Knyazev,Instituteof EEG correlates of self-referential processing are most frequently found in brain regions
Physiology,SiberianBranchof
overlapping with the default network, particularly in the medial prefrontal cortex. In the
RussianAcademyofMedical
Sciences,TimakovaStreet4, timedomain,thediscriminationofself-andothers-relatedinformationismostlyassociated
Novosibirsk630117,Russia withtheP300ERPcomponent,butsometimesisobservedevenearlier.Inthefrequency
e-mail:[email protected] domain,differentfrequencyoscillationshavebeenshowntocontributetoself-referential
processing,withspontaneousself-referentialmentationbeingmostlyassociatedwiththe
alphafrequencyband.
Keywords:self-referentialprocessing,defaultmodenetwork,EEG,ERP,oscillations
Self-referentialprocessinghasbeenprincipallyinvestigatedusing Reliability and validity of a particular method is also a very
functional magnetic resonance imaging (fMRI) and positron important issue. Reliability is the cornerstone of any scientific
emissiontomography(PET),whichcurrentlydominatethefield enterprise.Ifameasurementisunreliable,itcannotbevalid.How-
of human neuroscience. Electroencephalographic (EEG) studies ever,if amethodisreliableitcanalsobeinvalid(Carminesand
are less numerous and, to the best of my knowledge, have not Zeller,1979). In this review,it is not possible to cover the issue
been systematically reviewed. Understanding of the brain func- of reliability and validity of EEG and fMRI methods in detail
tioningisnotpossiblewithoutcarefulcomparisonoftheevidence (forrecentreviews,seee.g.,BennettandMiller,2010;Thatcher,
comingfromdifferentmethodologicaldomains.Ideally,different 2010). High levels of reliability (i.e.,>0.95) of several quantita-
methods are expected to complement each other. For example, tiveEEGmeasureshavebeenshowninmanystudies(e.g.,Lund
excellent spatial resolution of fMRI could be complemented by etal.,1995;McEvoyetal.,2000;Corsi-Cabreraetal.,2007;Gud-
excellenttemporalresolutionofEEG.Inreality,however,different mundsson et al., 2007; Näpflin et al., 2008; Towers and Allen,
methodsmaygivecontradictingresults.Insuchacase,acareful 2009;Schmidtetal.,2012).Somewhatsmallerreliabilitiesareusu-
analysisofpossiblecausesofthediscrepancyisnecessary.Inthis ally found for event-related potential (ERP) components. Thus,
paper,IaimedtoreviewEEGstudiesofself-referentialprocessing test-retestcorrelationcoefficientsforoddballtaskP300amplitude
andtoevaluatehowtheycorrespond,complement,orcontradict range from 0.50 to 0.80 and for peak latency from 0.40 to 0.70
to the existing fMRI evidence. It is important to keep in mind (Polich,1986; Fabiani et al.,1987; Segalowitz and Barnes,1993;
that fMRI and EEG represent different aspects of brain activity Walhovd and Fjell, 2002). Hall et al. (2006) found higher test-
and there may be a degree of incongruence between hemody- retestreliabilityfortheP300amplitude(0.86)andlatency(0.88).
namicandelectrophysiologicalsignals.Therelationshipbetween Less evidence exists regarding reliability of fMRI measures. Vul
EEGsignalandconcurrentchangesinneuronalspikingandlocal et al. (2009), summarizing several studies, conclude that fMRI
field potentials are relatively well understood (e.g., Buzsaki and measurescomputedatthevoxellevelwillnotoftenhavereliabil-
Draguhn,2004;Basar,2008).Ontheotherhand,itisnotyetclear itiesgreaterthanabout0.7.Liebermanetal.(2009)arguedthat
how the changes in the blood oxygen level dependent (BOLD) fMRI reliability was likely around 0.90. Friedman et al. (2008)
signal relate to concurrent changes in neuronal events (Huettel showthatformedianpercentsignalchangemeasure,themedian
etal.,2004;Debeneretal.,2006).Thequesttoelucidatehowthe test-retest reliability was 0.76. Aron et al. (2006) found 1-year
selfisprocessedinthebrainrequiresasolidunderstandingofthe test-retestfMRIreliabilityinaclassification-learningtaskexceed-
linkbetweenneuroimagingfindingsandtheirelectrophysiological ing 0.8. Similar test-retest reliability of fMRI activation during
underpinnings. prosaccades and antisaccades at the group level was shown by
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|1
Knyazev EEGcorrelatesofself-referentialprocessing
Raemaekers et al. (2007). However, these authors showed that Event-relatedpotentialisapowerfulandverypopulartoolfor
reliable results could be obtained in some but not all subjects, thestudyof corticaldynamicsthatarephase-lockedto(mostly)
mostlyduetoindividualdifferencesintheglobaltemporalsignal external stimuli and events. By calculating the mean of EEG
to noise ratio (SNR). Comprehensive discussion of the reliabil- epochs, the activity phase-locked to the stimulus is preserved,
ity of fMRI and effects of SNR could be found in Bennett and whereasnon-phase-lockedactivitycancelsitselfout.Itshouldbe
Miller(2010).Thus,itcouldbesummarizedthattest-retestreli- borne in mind that ERP is not the only kind of electrical corti-
ability, at least for some EEG measures, tends to be excellent calresponses.Aportionof theseresponsesistime-lockedtothe
and is at the border between good and excellent for most fMRI stimulus,butisnottemporallysynchronizedwithit,meaningthat
studies. this activity will cancel itself out during averaging. This kind of
responses is usually labeled induced responses, as distinct from
SELF-REFERENTIALPROCESSINGANDTHEDEFAULTMODE evokedresponsesthatarephase-lockedtothestimulus.Therehas
NETWORK beenalongdebateabouthowERPsarerelatedtoongoingoscil-
Theconceptofthedefaultmodenetwork(DMN)wasfirstintro- lationsandinducedresponses(e.g.,KolevandYordanova,1997;
duced by Raichle et al. (2001) basing on the evidence showing Makeig et al., 2002; Jansen et al., 2003; Klimesch et al., 2004).
a consistent pattern of deactivation across a network of brain Mostresearchersagreethatevokedandinducedresponsesrepre-
regions that occurs during the initiation of task-related activ- sentdifferentaspectsofbrainfunction.Muchevidenceshowsthat
ity (Raichle et al., 2001; Raichle and Snyder, 2007). The DMN evokedresponses(e.g.,differentERPcomponents)areinvolvedin
includes the precuneus/posterior cingulate cortex (p/PCC), the stimulusperceptionandprocessing,thatis,bottom-upprocesses.
medialprefrontalcortex(MPFC),andmedial,lateral,andinferior Induced responses,on the other hand,do not probably directly
parietalcortex.Thisnetworkisactiveintherestingbrainwitha participateinstimulusperceptionandprocessing.However,they
highdegreeoffunctionalconnectivity(FC)betweenregions.The areinvolvedinconcomitanttop-downprocesses,suchasalloca-
more demanding the task the stronger the deactivation appears tion of attention,memory retrieval,decision-making,and emo-
to be (McKiernan et al., 2006; Singh and Fawcett, 2008). A tion. Linking evoked responses with bottom-up and induced
notable exception to this general pattern of deactivation during responses with top-down processes is consistent with the theo-
goal-directed activity occurs in relation to tasks requiring self- reticalframeworksuggestedbyDavidetal.(2006)whoassociate
referentialthoughtandsocialcognition(Mitchell,2006;Gobbini evoked and induced responses with“drivers”and“modulators,”
etal.,2007),whichsuggeststhattheDMNlikelymediatesactive respectively.Themechanismsofactionofdriversrefertoclassical
cognitiveprocessesratherthanbeingstrictlya“default”network, neuronaltransmission,eitherbiochemicalorelectrical.Modula-
which only shows inactivation. Recent studies show that these tory effects can engage a complex cascade of highly non-linear
processes include first-person perspective (Greicius et al., 2003; cellularmechanisms(Davidetal.,2006).
Vogeley et al., 2004), task-independent thoughts (Binder et al., OscillationsarethemostsalientfeatureofEEG.Theycouldbe
1999; McKiernan et al., 2003), episodic memory (Greicius and studiedbothinrestandduringprocessingofexternalstimulior
Menon, 2004), social cognition and sense of agency processes tasks.Ongoingandevent-relatedoscillationsareusuallycatego-
(DecetyandSommerville,2003;GallagherandFrith,2003),dis- rizedintofivefrequencybands:delta(0.5–3.5Hz),theta(4–7Hz),
tinctionbetweenself-andnon-self-relatedstimuli(seeNorthoff alpha(8–12Hz),beta(13–30Hz),andgamma(>30Hz),although
etal.,2006;Buckneretal.,2008;forareview),andsocialinterac- thereisgenerallyalackofconsistencybetweenstudieswithmain-
tiontasks(Rillingetal.,2004,2008).Allthisevidenceimpliesthat taining a standard range of EEG bands. The five major bands
the DMN appears to be the seat of self-referential processing in are frequently subdivided into narrower sub-bands and there is
thebrain. no general agreement as to the boundaries of these sub-bands.
This is a potential source of discrepancies in results of different
APPROACHESTOTHESTUDYOFEEGCORRELATESOF studies.Itwasalsosuggestedthattherearesubstantialindividual
SELF-REFERENTIALPROCESSES differences in EEG frequency band boundaries and they should
Electroencephalogram and fMRI represent different aspects of beindividuallyadjustedusingalphapeakfrequencyastheanchor
brainactivity.Moreover,differentEEGmeasuresmayalsorelate (Klimesch,1999).Thesedebateshavepartlylosttheiractualitydue
todifferentaspectsofneuronalactivityandshowlittleornocor- totheadventofmodernmethodsoftime-frequencyrepresenta-
relation with each other. Therefore, a brief description of most tion,suchaswavelettransform,andadoptionofmass-univariate
popularmeasuresthatareusedinEEGdomainseemsnecessary statisticalapproaches(e.g.,DelormeandMakeig,2004).
forclearerunderstandingof laterdiscussedstudies.Firstly,EEG It is increasingly becoming clear that oscillations may have a
measurescouldbeobtainedinarestingconditionorduringper- specialandveryimportantroleintheintegrationofbrainfunc-
formanceofdifferenttasksorpresentationofdifferentstimuli.In tions(Nunez,2000;Varelaetal.,2001;CanteroandAtienza,2005;
theformercasetheyrepresent“spontaneous”orongoingelectrical Palvaetal.,2005;Knyazev,2007;Basar,2008;FingelkurtsandFin-
activityandcouldbeusedtoinvestigateEEGcorrelatesofspon- gelkurts,2010).TwodifferentaspectsofEEGoscillationscouldbe
taneous self-referential processes, such as mind wondering and potentiallymeasured:thepowerofaparticularoscillationatdif-
task-unrelated-thoughts.Inthelattercase,differentmeasuresof ferentcorticallocationsanditssynchrony(i.e.,phaseconsistency)
event-relatedchangesinelectricalactivity,suchasERPandevent- over these locations. The former is usually measured by means
related oscillations,are used to study the processing of external ofdifferenttime-frequencytransforms,suchasFourierorwavelet
self-relatedinformation. transform,thelatterbymeansofcoherenceorsimilarmeasures.To
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|2
Knyazev EEGcorrelatesofself-referentialprocessing
evaluateevent-relatedchangesinoscillatoryactivityEEGisusually simultaneousregistrationofEEGandfMRIhasbecomepossible,
recordedbefore(thebaseline)andduring(thetestperiod)presen- thedegreeof overlapbetweenthesetwosignalsinbrainregions
tationofstimuliorperformanceofatask;EEGchangesinthetest related to self-referential processing (e.g., the DMN) could be
periodrelativetobaselinearetreatedas“event-related”activityand determined.Secondandmoredirectapproachwouldbethestudy
arebelievedtoreflectbrainactivationinvolvedintheprocessing of EEG correlates of self-referential processing per se. Below, I
ofthetaskinhand.Event-relatedoscillationsaresubdividedinto willdiscussstudies,whichemployedboththeseapproachesand
evoked(phase-lockedtothestimulus)andinduced(non-phase- willtrytoshowwhethertheresultscorrespond,complement,or
lockedtothestimulus)parts,thelatterusuallybeingmuchlarger contradicttheexistingfMRIframework.
in amplitude than the former. According to the currently most
populartheory,theformeroscillationsarethebuildingblocksof EEGCORRELATESOFTHEDEFAULTMODENETWORK
theERP(e.g.,Makeigetal.,2002;Klimeschetal.,2004).Beyond BecauseDMNmostlyoperatesinarestingstate,manysimultane-
ERPsandoscillations,theglobal“microstates”(i.e.,quasistableand ous EEG-fMRI studies attempted to reveal correlations between
unique topographic distributions of the whole-cortex electrical spontaneousfluctuationsofBOLDandcorticalelectricalactivity
fieldpotential,Lehmann,1990)andlocal“microstates”(i.e.,qua- inthisstate.Sinceoscillationsconstitutethemostsalientfeatureof
sistablestateswithinindividualcortexlocations,Fingelkurtsand thespontaneousEEG,manyofthesestudiescorrelatedBOLDwith
Fingelkurts,2010)couldbeinvestigatedbothinrestandduring different EEG frequency bands.Alpha oscillations have received
performanceoftasks. most attention because they characterize quiet wakefulness and,
Spatial localization of observed effects is an important and likeDMN,areinverselyrelatedtobottom-upsensoryprocessing
rather complicated issue in EEG research. Scalp EEG samples (Goldmanetal.,2002;Laufsetal.,2003a,b;Moosmannetal.,2003;
a volume-conducted,spatially degraded version of the electrical Goncalvesetal.,2006;deMuncketal.,2007,2008;Tyvaertetal.,
activity, where the potential at any location can be considered 2008; Jann et al.,2009,2010; Sadaghiani et al.,2010). The gen-
a mixture of multiple sources (Makeig et al., 2004). To over- eral pattern that has been revealed in these studies is consistent
comethislimitation,differentblindsourceseparationandsource with the picture in which thalamus shows positively correlated
reconstructiontechniqueshavebeendevised.Blindsourcesepara- activity,whilefronto-parietalandoccipitalregionsexhibitnega-
tiontechniques,likeindependentcomponentanalysis(ICA),are tivelycorrelatedactivity.Togetherwithstudiesreportingreduced
increasinglybecomingpopularbothinEEGandinfMRIresearch, attentiontotheexternalenvironment,thesecorrelationssuggest
butthereareseveralprincipaldifferencesinhowthesetechniques areductionofactivityinbrainregionsassociatedwithexternally
areappliedinthetwodomains.InEEGresearch,temporalICA directedattentionandapotentialincreaseinactivityintheDMN
(TICA) is usually used, whereas in fMRI research, spatial ICA (Larson-Prioretal.,2011).However,therearesignificantdiffer-
(SICA) is almost exclusively applied. There are several reasons ences in reported positive alpha-band correlations to elements
for this,of which the most important is that the spatial dimen- of the DMN (e.g., Laufs et al., 2003b; Ben-Simon et al., 2008;
sion is much larger than the temporal dimension in fMRI data, Jannetal.,2009).Laufs(2008)notedthatthefailureacrossstud-
whereas for EEG data, the temporal dimension is much larger iestoidentifyanaveragecorticalBOLDsignalpattern,whichis
thanthenumberofsources(Calhounetal.,2001).Thismethod- positivelycorrelatedwithalphapower,maybeexplainedbynon-
ological difference may impede the direct comparison of EEG uniformbrainactivityatthepopulationlevelduringperiodsof
andfMRIICA results.To overcomethisobstacle,Knyazev etal. prominentalphaoscillationswhichfMRIgroupanalysismustfail
(2011)developedamethod,whichallowsapplicationofSICAto todetect.Laterstudies,whichusedmoresophisticatedapproaches
EEG data. A series of simulations showed that both SICA and todataanalysis,tendtoshowpositivecorrelationsofalphaoscilla-
TICAperformedadequatelywithspatiallyandtemporallyinde- tionswiththeDMNmorefrequently.Thus,Mantinietal.(2007)
pendentsources,butSICAoutperformedTICAwhensourceswere incorporatedintotheiranalysisEEGbandsbetween1and50Hz
temporallycorrelated(Knyazev,2013b). averagedacrosstheentirescalpandcorrelatedwiththesebands
Thesourcereconstructiontechniquescouldberoughlydivided thefMRItimecoursesofresting-statenetworks(RSNs)identified
into two categories: 3D imaging (or distributed) reconstruction bytheuseofICA.TheDMNandthedorsalattentionalnetwork
methods and equivalent current dipole approaches. The former had strong relationship with alpha and beta rhythms, albeit in
considerallpossiblesourcelocationssimultaneously,allowingfor oppositedirections,withtheDMNshowingpositivewhereasthe
large and widely spread clusters of activity. The latter rely on a attentionalnetworkshowingnegativecorrelationwiththeseoscil-
hypothesisthatonlyafewsourcesareactivesimultaneouslyand lations.Jannetal.(2010)reportonthetopographicassociationof
those sources are focal. It should be emphasized that all EEG EEGspectralfluctuationsandRSNsdynamicsusingEEGcovari-
source reconstruction methods are probabilistic modeling tech- ancemapping.T-mappingofthecovariancemapsindicatedthat
niques,whichatbestpointtothemostprobablelocationanddo thestrongesteffectswereagaininthealphaandbetabands.DMN
notgivethe“true”localizationofsources.Besides,theytypically activitywasfoundtobeassociatedwithincreasedalphaandbeta1
have low spatial resolution. However,it should be kept in mind bandactivity.Brookesetal.(2011b)analyzedmagnetoencephalo-
that fMRI data also represent results of statistical procedures to graphic(MEG)datausingacombinationofbeamformerspatial
compare signals between groups or within subjects and do not filteringandICA.ThismethodresultedinRSNswithsignificant
showthedirectstructurallocalizationofobservedeffects. similarityintheirspatialstructurecomparedwithRSNsderived
There potentially are two different approaches to the study independentlyusingfMRI.Inthisstudy,theDMNwasidentified
of EEG correlates of self-referential processing. Firstly, because using MEG data filtered into the alpha band. Wu et al. (2010)
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|3
Knyazev EEGcorrelatesofself-referentialprocessing
using parallel ICA decompositions of the fMRI data in the spa- (Basar et al., 1997; Nunez et al., 2001). Thus, Ben-Simon et al.
tial domain and of the EEG data in the spectral domain found (2008)demonstratedtwospatiallysegregatedyetsimultaneously
widespread alpha hemodynamic responses and high functional activenetworksassociatedwithalpharhythmmodulations,which
connectivity(FC)duringeyes-closedrestwithpredominantneg- theycalltheinducedandthespontaneous.Thesenetworksmight
ative peaks in occipital, temporal, and frontal regions, biphasic berelatedtotwoendogenousprocessesofthe“restingbrain,”one,
responsesintheDMN,andapositivepeakinthethalamus.Eyes- whichistunedoutwardandisperiodic,theother,whichisfocused
openrestingabolishedmanyofthehemodynamicresponsesand inwardandispersistent(Ben-Simonetal.,2008).Thelatternet-
markedlydecreasedFC.Ontheotherhand,Moetal.(2013)found workshowedaconsiderableoverlapwiththeDMN.Twoseparable
thatvisualalphapowerwaspositivelycorrelatedwithDMNonly alpha-bandnetworkswererevealedalsoinastudybyChenetal.
whentheeyeswereopen.Thisfindinghasbeeninterpretedasindi- (2013)whoemployedafour-stepanalyticapproachtotheEEG:
catingthatundertheeyes-opencondition,strongDMNactivity (1)groupICAtoextractindependentcomponents;(2)standard-
isassociatedwithreducedvisualcorticalexcitability,whichserves izedlow-resolutiontomographyanalysis(sLORETA)forcortical
toblockexternalvisualinputfrominterferingwithintrospective sourcelocalizationofICnetworknodes;(3)graphtheoryforFC
mentalprocessingmediatedbyDMN,whileweakDMNactivity estimationof epoch-wiseICbandpower;(4)circumscribingIC
isassociatedwithincreasedvisualcorticalexcitability,whichhelps similaritymeasuresviahierarchicalclusteranalysisandmultidi-
tofacilitatestimulusprocessing.Hlinkaetal.(2010)showedthat mensionalscaling.Duringeyes-opencomparedwitheyes-closed
DMN’sFCcorrelatespositivelywithrelativealphaandbetapower. condition,graphanalysesrevealedtwosalientfunctionalnetworks
Ros et al. (2013) used neurofeedback to reduce alpha rhythm. with fronto-parietal connectivity: a medial network with nodes
Comparedtosham-feedback,neurofeedbackinducedanincrease in the MPFC/precuneus, which overlaps with the DMN, and a
of connectivity within regions of the salience network involved more lateralized network comprising the middle frontal gyrus
inintrinsicalertnessandadecreaseofconnectivityintheDMN. and inferior parietal lobule. Interestingly, there is a hypothesis
ThechangeinDMNconnectivitywaspositivelycorrelatedwith that an internal train of thought unrelated to external reality is
changesin“ontask”mindwanderingaswellasresting-statealpha produced through cooperation between autobiographical infor-
rhythm.Moreover,bothmindwanderingandalphachangecorre- mation provided by the DMN and the fronto-parietal control
latedpositivelywithconnectivityinclustersoftheprecuneusboth networkwhichhelpssustainandbufferinternaltrainsofthought
intheneurofeedbackandintheshamgroup.Besides,forthesham againstdisruptionbytheexternalworld(Smallwoodetal.,2012).
group only, a more extensive positive correlation with resting- Thishypothesisexplainswhyactivationofthefronto-parietalnet-
statealphachangewasobservedinaregionoftheMPFC.Hence, workandtheDMNisoftenobservedtogetherduringperiodsof
both neurofeedback and sham groups remained consistent with internallyguidedthought.Ifthishypothesisistrue,theexistenceof
thereportsof apositiveassociationbetweenalphasynchroniza- twoseparablealpha-bandnetworksassociatedwiththeDMNand
tionandDMNconnectivity(Mantinietal.,2007;Jannetal.,2009; thefronto-parietalnetwork,respectively,wouldmakefunctional
Hlinkaetal.,2010).Meyeretal.(2013)investigatedtherelation- sense. In any case,the involvement of alpha oscillations in both
shipofICA-derivedRSNsandtheircorrelatedelectrophysiological thetop-downattentionalcontrolandtheintegrationof internal
signals in eyes-open resting state. In 4 of the 12 subjects,nega- mental processes are supported by numerous observations (see
tivealphacorrelationwithvisualRSNswasfound,however,due e.g.,Klimeschetal.,2007;Knyazev,2007forreviews).
to large inter-subject variability,no significant correlations were OtherEEGfrequencybands(mostnotablythetaandgamma)
foundonthegrouplevel. alsoshowedcorrelationswithDMNBOLDsignal.Medialfrontal
SomeinvestigatorscorrelatedfMRIBOLDsignalwithmeasures theta power changes were negatively correlated with the BOLD
of EEGsynchronizationinthealphafrequencyband.Jannetal. responseinmedialfrontal,inferiorfrontal,p/PCC,inferiorpari-
(2009)showthattheBOLDcorrelatesofglobalEEGsynchroniza- etal, middle temporal cortices, and the cerebellum (Scheeringa
tioninthealphafrequencyarelocatedinbrainareasinvolvedin etal.,2008).Meltzeretal.(2007)alsofoundthatfronto-medial
theDMN.Sadaghianietal.(2010,2012)adaptedthephase-locking thetamoststronglynegativelycorrelateswiththeMPFC,although
valuetoassessfluctuationsinsynchronythatoccurovertimein negativecorrelationswerealsofoundwithotherDMNareassuch
ongoingEEGalphaactivity.Fluctuationsinglobalsynchronyin asPCC.InthestudybyMizuharaetal.(2004),thefrontalmidline
the upper alpha band correlated positively with activity in sev- theta showed negative correlation with BOLD signal over ante-
eralprefrontalandparietalregions,asmeasuredbyfMRI.fMRI rior medial regions. The inverse relationship between theta and
intrinsicconnectivityanalysisconfirmedthattheseregionscorre- BOLDintheDMNwasalsoobservedinthestudybyWhiteetal.
spondtothewell-knownfronto-parietalnetworkwhichhasbeen (2012). There is some evidence that delta,like theta,also shows
consistentlyshowntoberecruitedbytasksthatinvolvetop-down negative correlation with the DMN. Thus, Hlinka et al. (2010)
attentional control processes. This apparent disagreement with showed that DMN’s FC correlates negatively with relative delta
theJannetal.’s(2009)studyisexplainedbythefactthatdifferent power.InastudybyDimitriadisetal.(2010),deltaactivityshowed
measures of phase synchrony and a fixed vs. individually deter- awidespreadincreaseinareasoverlappingwiththeDMNduring
minedhighalpharangeareemployedinthetwostudiesimplying the performance of arithmetic tasks,which are known to cause
thatresultsmightcorrespondtofunctionallydifferentoscillations DMN’s deactivation. Since delta and theta are indicated as the
(Sadaghiani et al., 2012). This latter notion is in line with the primaryEEGfrequenciesinlimbicstructures(i.e.,thetainhip-
frameworkstatingthatthescalp-recordedalphaistheend-product pocampus and delta in orbito-frontal cortex, see e.g., Knyazev,
of manyalpharhythmsthatarespatiallyaveragedoverthescalp 2007,2012 for review) the negative correlations with the DMN
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|4
Knyazev EEGcorrelatesofself-referentialprocessing
maybeinfluencedbyprojectionsfromthesestructurestomidline spontaneousBOLDactivityintheDMN(Britzetal.,2010;Musso
frontalregions(e.g.,Brazier,1967,1968,1969). etal.,2010;Yuanetal.,2012).
Contrarytothetaanddelta,gamma(30–50Hz)powershows Insum,thestudyofspontaneousEEGcorrelatesoftheDMN
positivecorrelationswithDMNBOLDsignalatrest(Mantinietal., appear to suggest that low-frequency EEG oscillations of delta
2007)anddecreasesduringthetransitionfromrestingstatetoan and theta bands predominantly at frontal cortical sites correlate
attention task which is interpreted as a correlate of DMN deac- negativelywiththeDMN,whereashigherfrequencyoscillations
tivation (Lachaux et al., 2008; Hayden et al., 2009; Jerbi et al., (mostnotablyalphaatparietalandoccipitalregions)showpositive
2010;Berkovich-Ohana et al.,2012). Moreover,slow changes in correlationswiththisnetwork.Itshouldbenotedthatalthough
the power of gamma oscillations make a significant contribu- alpha, beta, and gamma oscillations show positive correlations
tiontothespontaneouslocalfluctuationsofresting-stateBOLD with the DMN,specificities of these relationships are not equal
signals (Nir et al.,2007,2008; He et al.,2008; Scholvinck et al., forthethreebands.Itappearsthatalpha(andpossiblyslowbeta)
2010)supportingthenotionthatgammaprocessingreflectslocal correlates positively with DMN and negatively with attentional
neural computations (Canolty and Knight, 2010; Siegel et al., networkswhereasgammashowspositivecorrelationswithmost
2012). Most interesting data on gamma-correlates of the DMN cognitive processes including attention (e.g.,Muller et al.,2000;
havebeenobtainedinstudieswithdepthrecordingsinhumans Fanetal.,2010;Hippetal.,2011;Ossandónetal.,2012).Verylow
(e.g.,Jerbietal.,2010).However,Wangetal.(2012)haveshown EEGfrequenciescouldalsobeconsideredaspromisingcandidates,
thatlow-frequencyoscillations(<20Hz),andnotgammaactiv- althoughthefunctionalsignificanceoftheseoscillationshasyetto
ity,predominantlycontributedtointer-arealBOLDcorrelations. bedetermined.
Thelow-frequencyoscillationsalsoinfluencelocalprocessingby
modulatinggammaactivitywithinindividualareas(Wangetal., EEGSTUDIESOFSELF-REFERENTIALPROCESSING
2012). AllEEGstudiesofself-referentialprocessingcouldbesubdivided
Basing on PET and fMRI findings of DMN localization and intoseveralcategories basingonthenature of EEGphenomena
properties,someinvestigatorsattemptedtoderiveEEGcorrelates under study and the kind of self-referential processing. Firstly,
of theDMNwithoutsimultaneousEEG-fMRIrecordings.Chen some studiesattempted tocorrelate spontaneousEEG measures
etal.(2008)comparedthespatialdistributionandspectralpower in a resting state with measures of spontaneous self-referential
of seven bands of resting-state EEG activity in eyes-closed and thoughts (e.g., retrospective self-reports). Secondly, EEG corre-
eyes-openconditionandtermedthedefinedsetof regionaland latesof theprocessingof self-relatedvs.notself-relatedexternal
frequency specific activity the EEG-DMN. Fingelkurts and Fin- stimulihavebeeninvestigated.Thelatterinturncouldbecatego-
gelkurts(2011)usedmeasuresof“operationalsynchrony”ofalpha rizedintostudiesusingERPsoroscillationsastheoutcomeEEG
oscillations and found a constellation of operationally synchro- measure. I will describe these three groups of studies separately
nizedcorticalareasincludingtwosymmetricaloccipito-parieto- andwilltrytosummarizehowtheyagreeordisagreewitheach
temporal and one frontal spatio-temporal patterns (indexed as otherandtheexistingfMRIframework.
DMN)thatwaspersistentacrossallstudiedexperimentalcondi-
tions.Interestingly,itwasfurthershown,thatsuchDMNopera- SPONTANEOUSEEGSTUDIES
tionalsynchronywassmallestorevenabsentinpatientsinvege- Therearefewresting-stateEEGstudies,whichattemptedtocorre-
tativestate,intermediateinpatientsinminimallyconsciousstate, latespontaneousEEGmeasureswithmeasuresofself-referential
and highest in healthy fully self-conscious subjects (Fingelkurts thoughts. Cannon and Baldwin (2012) sought to determine
et al., 2012). Because fMRI research has shown that functional whetherthecurrentsourcedensitylevelsintheDMNasmeasured
synchrony across elements of the DMN coheres through brain bysLORETAwouldcorrespondtootherneuroimagingtechniques
oscillations at very low frequencies (i.e.,0.1Hz,Fransson,2005; andtounderstandthesubjectivementalactivityassociatedwith
Foxetal.,2006),somestudiesinvestigatedverylowEEGfrequen- theDMNduringbaselinerecordingsandthreeexperimentalcon-
cies (VLF,Vanhatalo et al.,2004; Helps et al.,2008,2009,2010; ditions. Participants completed subjective reports regarding the
Broydetal.,2011).IthasbeenshownthatVLFhasatemporally mentalactivitiesemployedduringbaselinerecordings.Inallfre-
stable and distinctive spatial distribution across the scalp with quencybandsfromdeltatobeta,theDMNappearedtobepref-
maximal power distributed across frontal midline and posterior erentiallyinvolvedinself-relevant,self-specific,orself-perceptive
regions(Helpsetal.,2008,2010).Thisscalpnetworkshowsdeac- processes.Knyazevetal.(2011)usedacombinationof ICAand
tivationof EEGpowerfollowingthetransitionfromresttotask sLORETAsourceimagingtorevealRSNsintraditionalEEGfre-
(Helps et al.,2009,2010) and these deactivations are correlated quencybands.Ashortself-reportscalewasusedtomeasureindi-
withattentionperformance(Helpsetal.,2010;Broydetal.,2011). vidualdifferencesintheintensityofself-referentialthoughts.Only
UsingsLORETA,thesourcesofthisdeactivationwerelocalizedto alpha-bandspatialpatternssimultaneouslyshowedaconsiderable
medial prefrontal regions,p/PCC,and temporal regions (Broyd overlapwiththeDMNandapositivecorrelationwiththemeasure
etal.,2011).TheseresultssuggestsimilaritiesbetweentheDMN of self-referential thoughts. This group of researchers has repli-
asidentifiedbyfMRIandtheVLFEEGnetwork. catedtheirfindingsinlargeanddiversegroupsofsubjectscoming
Some authors propose that the neural activity at a specific fromtwodifferentculturesandfoundculture-relateddifferences
frequency band is unlikely to constitute the electrophysiologi- inEEGcorrelatesofself-referentialthoughts(Knyazevetal.,2012).
cal correlate of an RSN. Instead, microstates of the EEG signal Specifically,theself-referentialthought-relatedincreaseof alpha
havebeenproposedaspotentialelectrophysiologicalcorrelatesof activity prevailed in the posterior DMN hub in Russian, but in
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|5
Knyazev EEGcorrelatesofself-referentialprocessing
theanteriorDMNhubinTaiwaneseparticipants.Theseculture- describedthemselvesintermsofanabstract,independentsense-
relateddifferencescouldbeexplainedbydifferentself-construal of-self underlying thought (predominantly self-referral mode)
styles that prevail in different cultures (Markus and Kitayama, exhibitedhigheralphaandlowergammapower.
1991), but they could be also explained by systematic culture- Default mode network is one among several networks with
related differences in personality (see e.g.,Gartstein et al.,2005; differentfunctionalproperties,includingthosefororientingatten-
Knyazevetal.,2008bfortheevidenceonpersistentdifferencesin tion (Corbetta et al.,2008) and memory encoding and retrieval
temperament and personality across the lifespan between Russ- (MaguireandFrith,2004;Habeckaetal.,2005;Burianovaetal.,
ianandothercultures).Thislatterexplanationseemsparticularly 2010).Whereastask-specificnetworksareactivatedwhenatten-
feasible in view of the evidence that similar differences in EEG tion is directed toward relevant stimuli, the DMN increases in
correlates of self-referential thoughts have been found between activity during rest (Buckner et al., 2008). It is still unknown,
extraverts and introverts (Knyazev, 2013a) and there is ample however,howthebrainswitchesfunctionallybetweendefaultand
evidencethatEasternpopulationsingeneralandTaiwanesepop- task-specificnetworks.Oneinterestinghypothesisisthattransient
ulation in particular are lower in Extraversion than most more functionalorganizationofneuralassembliesisbroughtaboutby
westernpopulationsincludingRussia(seee.g.,AllikandMcCrae, synchronizationofneuraloscillations(vonSteinetal.,2000;Varela
2004). This evidence gives interesting hint about the relation- etal.,2001;Ward,2003).Itshouldbeborneinmindhoweverthat
shipbetweenEEGcorrelatesof self-referentialthoughtsandthe sometimessynchronizationofanoscillationwithinanetworkmay
dopaminergic basis of extraversion (Depue and Collins, 1999). actuallyreflecttheinhibitionof thisnetwork(seee.g.,Klimesch
Indeed, it has been shown that the association between extra- etal.,2007).SeveralEEGstudiescomparedsynchronyandspectral
version and posterior vs. frontal EEG activity is mediated by powermeasureswithinthetask-specificnetworks(attentionand
dopamine (Wacker et al.,2006;Wacker and Gatt,2010; Koehler memory)andtheDMNduringattention/workingmemorytasks
etal.,2011)andthereisampleevidencethattheposteriorandthe vs. mind wandering. More power and phase synchronization in
anteriorDMNhubsaredifferentiallysusceptibletodopaminergic theta,alpha,andgammafrequencybandshasbeenfoundduring
influences(seeKnyazev,2013aforareviewofthisevidence). mindwanderingbetweenbrainregionsassociatedwiththeDMN,
AnumberofstudiesinvestigatedEEGcorrelatesofself-related whereasduringperiodswhensubjectswerefocusedonperforming
mentalprocessesduringmeditation.Lehmannetal.(2001)using avisualtask,therewassignificantlymorephasesynchronywithin
LORETAimagesoftheEEGgammafrequencybandinvestigated atask-specificbrainnetwork(Kirschneretal.,2012).Increasesin
locations of intra-cerebral source gravity centers and showed thetaoscillationsinthemedialfrontalcortex,whichareaccom-
that self-induced meditational dissolution and reconstitution of paniedbydecreasesinbetaandgammaoscillationsatthesame
the experience of the self involves the right fronto-temporal spatialcoordinatesandotherbrainareas,includingnodesofthe
area.Travis(2001)comparedEEGpatternsduringtranscending DMN,havebeenshownduringworkingmemorytasks(Brookes
(described as“silence and full awareness of pure consciousness, etal.,2011a).Theincreaseinfrontalthetapowerduringworking
where the experiencer is left all by himself” Mahesh, 1963, p. memorytaskshasbeenshowntocorrelatewithBOLDdecrease
288,cited from Travis,2001) to other experiences during Tran- inregionsthattogetherformtheDMN(Scheeringaetal.,2009).
scendental Meditation practice. To correlate specific meditation The same study showed a right posterior alpha power increase,
experiences with physiological measures,the experimenter rang whichwasfunctionallyrelatedtoBOLDdecreasesintheprimary
a bell three times during the session. Subjects categorized their visual cortex and in the posterior part of the right middle tem-
experiencesaroundeachbellring.Transcending,incomparison poral gyrus. No correlations were observed between oscillatory
to“other”experiences,wasmarkedbyhigherEEGalphaampli- EEGphenomenaandBOLDinthetraditionalworkingmemory
tudeatparietalsitesandhigheralphacoherencebetweenFzand areas. These findings prompt an assumption that the observed
Pz.Travisetal.(2010)showedthat,comparedtoeyes-closedrest, increasesinoscillatorypowerduringworkingmemorytasksactu-
TranscendentalMeditationledtohigheralpha1frontalpowerand allyreflectinhibitionofneuronalactivitythatmayinterferewith
lowerbeta1andgammafrontalandparietalpower,higherfrontal working memory maintenance,with theta power increase being
andparietalalpha1interhemisphericcoherenceandhigherfrontal relatedtotheinhibitionoftheDMNwhilealphapowerincrease
andfronto-centralbeta2intra-hemisphericcoherence.eLORETA beingrelatedtotheinhibitionofsensoryperception(Scheeringa
analysis identified sources of alpha1 activity in midline cortical etal.,2009).Childrendemonstrateastrongernegativecorrelation
regionsthatoverlappedwiththeDMN.TravisandShear(2010) between global theta power and the BOLD signal in the DMN
summarized that different meditation techniques are associated during a working memory task relative to adults implying that
withdifferentEEGbands.Focusedattentiontechniquesarechar- children suppress this network even more than adults,probably
acterizedbybeta/gammaactivity;openmonitoringtechniquesare from an increased level of task-preparedness to compensate for
characterizedbythetaactivity;andself-transcendingischaracter- notfullymaturecognitivefunctions(Michelsetal.,2012).Incon-
izedbyalphaactivity.Lastly,Travisetal.(2004)showthatoscilla- trast to power, correlations between instantaneous theta global
toryactivity(spontaneousandtask-related)correlateswithtrait- fieldsynchronizationandtheBOLDsignalwereexclusivelypos-
like psychological characteristics along an object-referral/self- itive in both adults and children, but only significant in adults
referralcontinuumofself-awareness.Specifically,individualswho inthefrontal-parietalandposteriorcingulatecortices.Moreover,
described themselves in terms of concrete cognitive and behav- theta synchronization,in contrast to EEG power,was positively
ioral processes (predominantly object-referral mode) exhibited correlatedwithresponseaccuracyinbothagegroups.Thus,these
lower alpha and higher gamma power,whereas individuals who studies show that increase of theta power correlates with DMN
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|6
Knyazev EEGcorrelatesofself-referentialprocessing
suppression;increaseofthetasynchronycorrelateswithworking junctionat330–400msafterstimulusonset(Blankeetal.,2005);
memoryperformance;increaseofalphapower,ontheotherhand, duration of this activation, but not its strength, were found to
correlateswithasuppressionofsensorynetworks. correlatepositivelywithperceptualaberrationscores(Arzyetal.,
Summingup,theaboveoutlinedEEGstudiesappeartocon- 2007).AhigherP300wavetothesubject’sownfacethanfamiliaror
vergeinshowingthatinrestingcondition,self-relatedthoughtsare unfamiliarfaceswasobservedinseveralstudies(Ninomiyaetal.,
accompaniedbyanincreaseofspectralpowerincorticalregions 1998;Scottetal.,2005;Suietal.,2006).Cahareletal.(2002)didnot
overlapping with the DMN and these changes are most consis- observethiseffect,probablybecauseoftheveryhighoccurrence
tently found in the alpha band of frequencies. During working ofthesubject’sownface,illustratingthemajorhabituationeffect
memorytasks,however,thedeactivationoftheDMNisreflected ofsuchparadigms.Keyesetal.(2010)observeddifferencesinthe
in an increase of medial frontal theta power with concomitant ERPwaveformsmuchearlier,withincreasedN170andvertexpos-
decreaseofbetaandgammaoscillationsandanincreaseofalpha itive potential amplitude over posterior and fronto-central sites,
powerinsensorycorticesreflectinginhibitionofneuronalactivity respectively,forselfrelativetobothfriendandstrangerfaces.Cul-
thatmayinterferewithworkingmemorymaintenance. turaldifferenceinneuralmechanismsofself-recognitionhasbeen
investigatedbothwithregardtothelong-termculturalexperiences
EEGCORRELATESOFTHEPROCESSINGOFSELF-RELATED (Suietal.,2009)andaftermodulationoftemporaryaccesstoother
INFORMATION culturalframeworksusingaself-construalprimingparadigm(Sui
Becauseself-relatedinformationcouldbepresentedviadifferent etal.,2013).ForBritishparticipants,theown-faceinducedfaster
sensoryandfunctionaldomains(e.g.,auditory,visual,sensorimo- responsesandalargernegativeactivityat280–340ms(N2)relative
tor,verbal,spatial,emotional,andsoon),therecouldbedomain- tothefamiliarface,whereasChineseparticipantsshowedreduced
specificandself-specificeffects.Ameta-analysisbyNorthoffetal. N2 amplitude to the own-face compared with the familiar face
(2006)ofPETandfMRIstudiesofself-referentialprocessingiden- (Suietal.,2009).Furthermore,forBritishparticipants,priming
tifiedactivationincorticalmidlinestructuresoccurringacrossall aninterdependentself-construalreducedthedefaultanteriorN2
functional domains (e.g., verbal, spatial, emotional, and facial). totheirownfaces.ForChineseparticipants,however,primingan
Cluster and factor analyses indicated functional specialization independentself-construalsuppressedthedefaultanteriorN2to
into ventral, dorsal, and posterior cortical midline areas. The theirfriend’sfaces(Suietal.,2013).
latterencompassesthep/PCCandisconsideredinvolvedinself- Similarlytotheprocessingofownface,participant’sownname
integration(i.e.,linkageofself-referentialstimulitothepersonal elicitsahigherP300amplitude(e.g.,Fischleretal.,1987;Berlad
context,NorthoffandBermpohl,2004).Itisinteresting,therefore, andPratt,1995;MullerandKutas,1996;Holeckovaetal.,2006).
to look how EEG studies corroborate or contradict this frame- Bypresentationtheparticipant’sfirstnameagainstanumberof
work.IwillfirstpresentERPandthenoscillationstudiesof the other first names in strict equiprobable fashion, it was possible
processingofself-relatedstimuli. to record an electrophysiological response to the subject’s own
Own body, own name, and the image of own face are the name,whichisindependentofitsprobabilityofoccurrence(Per-
kind of stimuli that are frequently used in the studies of self- rinetal.,1999,2005).ThecharacteristicsofthisERPareconsistent
processing. It has been suggested that social cognition is one of with those of the classical P300, but the latency (500ms) was
the functions of the DMN (e.g.,Mitchell,2006) and it certainly muchlongerthanthatusuallyobtainedinresponsetopuretones
constitutes a part of the self (e.g., Markus and Kitayama, 1991; (300ms),thisbeingprobablytheconsequenceofthedifferencein
HanandNorthoff,2009).Therefore,theprocessingofsocialstim- thelengthofthestimulus(Perrinetal.,1999).DifferentialERPs
uliandeffectsofsocialandculturalcontextsarealsorelevantto to the own name were shown in altered states of consciousness,
thestudyofself-referentialprocessing.Becauserealsocialbehav- suchassleep(Perrinetal.,1999,2005;Prattetal.,1999)andin
ior(i.e.,interactionswithotherpeople)isnotalwayspossibleto patientsinavegetativestate(Perrinetal.,2006),suggestingthat
organizeinalaboratoryinacontrolledmanner,whichisneeded theidentificationof self-relevantstimuliremainsinthesestates.
forEEGregistrationandsubsequentmeaningfulanalysis,virtual Using an EEG-PET paradigm, Perrin et al. (2005) have shown
(i.e.,modeledbymeansof acomputergame)socialinteractions thattheamplitudeoftheP300component,elicitedwhenhearing
arefrequentlyused. one’sownname,correlateswithregionalcerebralbloodchangesin
rightsuperiortemporalsulcus,precuneus,andMPFC.Addition-
ERPSTUDIES ally,the latter was more correlated to the P300 obtained for the
ManyERPstudiesofself-referentialprocessingshowthatthedis- subject’snamecomparedtothatobtainedforotherfirstnames.
criminationof self fromothersisfrequentlyassociatedwiththe TheseresultsareingoodagreementwithfMRIstudiesshowing
well-knownP300ERPcomponent,anevokedresponsetostimuli differences in activation in MPFC and right paracingulate cor-
thatareunexpected,salient,ormotivationallyrelevant(Polichand tex (Kampe et al., 2003; Staffen et al., 2006) when comparing
Kok,1995).Sourcelocalizationofthisresponsefrequentlyshows activationtopresentationof thesubject’sownnametotheacti-
activations in DMN structures associated with self-processing. vation to presentation of other names. These results are also in
Thus,theownhandelicitedagreaterpositivecomponent(P350– goodagreementwiththeproposedcriticalroleofmidlinestruc-
500) than did other hand and the generator of this component turesinself-referentialprocessing(NorthoffandBermpohl,2004;
was localized in the anterior cingulate cortex (ACC, Su et al., Lou et al., 2005). Similar effects were observed when the self-
2010). Mental imagery tasks with respect to the own body have relevanceeffectinobjectrecognitionwasstudied(Miyakoshietal.,
been shown to elicit selective activation of the temporo-parietal 2007).
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|7
Knyazev EEGcorrelatesofself-referentialprocessing
Effects of the self-relevant possessive pronouns compared to evaluation induced stronger desynchronization and decreased
non-self-relevant possessive pronouns were studied in several phasesynchronyinalphaandgammabands,whereaspreparatory
studies. These studies have shown that self-relevant possessive self-relatedattentionalorientationwasmarkedbysynchronization
pronoun elicited significantly larger P300 amplitude than non- in these same bands (Mu and Han,2013). However,in another
self-relevant possessive pronouns (Zhou et al., 2010; Shi et al., study, the same authors show that relative to other referential
2011)withsourcesofthisactivitybeingidentifiedinMPFC,ante- traits,self-referentialtraitsinducedevent-relatedsynchronization
riorcingulate,andpostcentralcortex(Shietal.,2011).Wallaetal. oftheta-bandactivityoverthefrontalareaat700–800msandof
(2007, 2008) showed that in the time range between 250 and alpha-bandactivityoverthecentralareaat400–600ms(Muand
400mstheinformationrelatedto“my”andto“his”couldbedis- Han,2010).
tinguishedoveroccipitalelectrodesandinthetemporalregion.In SeveralstudiesinvestigatedEEGcorrelatesofsocialcognition
astudybyEsslenetal.(2008),self-vs.other-referencewasinvesti- andbehavior.Billekeetal.(2013)usedEEGtostudytheneuro-
gatedusingtraitadjectivesinreferencetotheselforaclosefriend. biologyofperceptionofsocialriskinsubjectsplayingtheroleof
TheMPFCwasfoundtobemoreactiveduringtheself-reference proposersinaniteratedultimatumgame.Theplayersweresepa-
condition. In an interesting study by Herbert et al. (2011), the ratedtohigh-riskandlow-riskoffers.Priortofeedback,high-risk
effectof emotionalvalenceonERPselicitedbyself-relevantand offersgeneratedadropinalphaactivityinanextendednetwork.
non-self-relevant pronoun-noun expressions was investigated. Moreover,trial-by-trial variation in alpha activity in the medial
From350msonward,processingofself-relatedunpleasantwords prefrontal, posterior temporal, and inferior parietal cortex was
elicited larger frontal negativity, whereas processing of pleasant specifically modulated by risk and, together with theta activity
wordselicitedlargerpositiveamplitudesoverparietalelectrodes in the prefrontal and PCC,predicted the proposer’s subsequent
from 450ms after stimulus onset. This evidence is in line with behavior.Rejectionsoflow-riskofferselicitedahigherprefrontal
above discussed evidence linking anterior DMN hub with pro- thetaactivitythanrejectionsofhigh-riskoffers.Usingacombina-
cessing of negative and posterior DMN hub with processing of tionofICAandsLORETAimagingKnyazevetal.(2011)showed
positiveself-relatedinformation(Knyazev,2013a).However,Wat- that cortical patterns of alpha desynchronization in response to
sonetal.(2007)observedlargerN400amplitudesforwordswith facialstimuliweredifferentdependingonwhetherthesestimuli
the self-positivity bias at fronto-central electrode sites. Further werepresentedinacontextof socialinteractionsorajudgment
researchisneededtodisentangletheeffectsofself-referenceand of facialaffecttask.Intheformercase,alphadesynchronization
emotionalvalenceoncorticalelectricalresponses. wasfoundintheposteriorDMNhub,whereasinthelattercaseit
Insum,thediscussedERPstudiesgenerallyconcurwithfMRI appearedattheterminalfieldoftheventralvisualstream.Knyazev
studies in suggesting that medial cortices (most notably MPFC etal.(2013)usedacomputergametomodelsocialinteractions
andACC)arethecrucialstructuresforprocessingofself-relevant withvirtual“persons,”whichincludedthreemajorkindsofsocial
information.Additionally,theyshowthatthetimeframeof this behavior: aggressive, friendly, and avoidant. Most salient differ-
processing most frequently coincides with the well-known P300 ences were found between avoidance and approach behaviors,
ERPcomponent. whereasthetwokindsofapproachbehavior(i.e.,aggressionand
friendship)didnotdifferfromeachother.Comparativetoavoid-
OSCILLATIONSTUDIES ance, approach behaviors were associated with higher induced
Contrary to ERP,which reflects only the evoked (i.e., stimulus- responsesinmostfrequencybands,whichweremostlyobservedin
phase-locked) response, oscillations could be spontaneous, corticalareasoverlappingwiththeDMN.Thedifferencebetween
induced,orevoked.Spontaneousoscillationsascorrelatesofself- approach-andavoidance-relatedoscillatorydynamicswasmore
referentialprocesseshavebeenalreadydiscussedearlier.Thischap- salient in subjects predisposed to approach behaviors (i.e., in
terwillreviewstudiesdealingwithinducedandevokedresponses aggressive or sociable individuals) and was less pronounced in
toself-relatedstimuli(seeearlierinthisreviewadiscussiononpos- subjectspredisposedtoavoidancebehavior(i.e.,inhightraitanx-
siblefunctionalmeaningofthesetwokindsofresponses).Manyof iety scorers). These findings are in line with previous findings
thesestudiesshowthatalphasuppressionappearstobethemost showing the effect of these personality traits on the perception
salientfeatureofinducedresponsestosuchkindofstimuli.Thus, ofsocialemotionalstimuli(Knyazevetal.,2008a)andoscillatory
bymeansofvirtualrealitytechnology,ithasbeenshownthathand responsestoapproach-andavoidance-relatedcues(Knyazevand
ownershipandtheexperienceofself-locationarereflectedinalpha Slobodskoj-Plusnin,2007).
(ormu)bandpower(8–13Hz)modulationsinbilateralsensori- The role of gamma activity in the p/PCC in autobiograph-
motorcorticesandposteriorparietalareas(Lenggenhageretal., ical memory retrieval in humans was investigated by means of
2011;Evans and Blanke,2013). Electrical neuroimaging showed intracranialrecordings(Dastjerdietal.,2011;Fosteretal.,2012).
that alpha power in the MPFC was correlated with the degree Late-onset (>400ms) increases in broad high gamma power
ofexperimentallymanipulatedself-location(Lenggenhageretal., (70–180Hz)withinp/PCCsub-regionsduringepisodicautobio-
2011).Alphaactivityinhighlysimilarfronto-parietalregionswas graphicalmemoryretrievalwasobserved,whileitwassignificantly
alsomodulatedduringamotorimagerytask(EvansandBlanke, reducedorabsentwhensubjectsretrievedself-referentialsemantic
2013).Hearingsubject’sowncomparedtoothernameswasasso- memoriesorrespondedtoself-judgmentstatements,respectively.
ciated with increased alpha-band desynchronization at frontal Asignificantdeactivationofhighgammapowerwasalsoobserved
sites in time window of 500–1000ms (Höller et al.,2011). Self- duringtasks,whichrequireexternallydirectedattention,suchas
relatedevaluationonpersonalitytraitscomparedtofriend-related arithmeticcalculation(Fosteretal.,2012).
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|8
Knyazev EEGcorrelatesofself-referentialprocessing
All these studies show that induced oscillatory responses to DMN and self-referential processes.At this stage of our knowl-
self-relatedstimuliaremostlyfoundincorticalareasbelongingto edge,itseemsprudentialnottolinktheseprocessestoaparticular
theDMNandaremostsalientinthealphabandof frequencies, oscillation. Depending on situational context and the kind of
althoughresponsesinotherfrequencybands(mostnotablytheta self-referential processes, different oscillations may be involved.
andgamma)arealsofrequentlyobserved. However,somepatternoftheirinvolvementisalreadydiscernible.
Few studies investigated evoked oscillatory responses to self- Itappearsthatdeltaandthetaoscillations(mostprominentlyat
referentialstimuli.Miyakoshietal.(2010)usingtheimageofpar- frontalmidlineregions)correlatenegativelywithDMN.Increase
ticipant’sownfaceobservedphaseresetting(i.e.,evokedresponse, of theta power during working memory tasks is related to inhi-
asmeasuredbyITCvalues)inthethetabandwithinthemedial bition of DMN regions (Scheeringa et al., 2009; Brookes et al.,
frontalareaduring270–390mspost-stimulus.Royeetal.(2010) 2011a; Michels et al., 2012). Alpha (and possibly beta) oscilla-
during passive listening observed enhanced evoked oscillatory tions appear to be positively related to DMN and spontaneous
activity in the 35–75Hz band for subject’s own telephone ring- self-referential processes and negatively to attentional networks.
tone, starting as early as 40ms after sound onset, and found a Alphaalsoshowsmostprominentpowerdecreaseduringprocess-
co-activation of left auditory areas and left frontal gyri. Active ing of external self-related information. The notion of different
detection of sounds additionally activated the superior parietal “alphas”involvedindifferentaspectsofattentionregulationand
lobe supporting the existence of a fronto-parietal network of top-downprocessesseemsveryattractive(Ben-Simonetal.,2008;
selectiveattention.Lastly,Knyazevetal.(2011)observedevoked Sadaghiani et al., 2012; Chen et al., 2013). Gamma oscillations
alpha-bandresponsestofacialstimuliinasocialinteractiontask correlatepositivelywithDMNandaredefinitelyinvolvedinself-
inthePCC. referential processing,but specificity of their involvement raises
doubtsbecausemanystudiesshowtheirinvolvementinvirtually
GENERALSUMMARYANDUNRESOLVEDQUESTIONS anycognitiveprocess.Finally,oscillationsofverylowfrequencies
Itcouldbesummarizedthatingeneral,thereisagoodcorrespon- correlate with DMN, but their involvement in self-related cog-
dence between imaging and EEG studies in localizing the self- nitive processes, which typically occur at much faster temporal
referentialprocessinginthebrain.AcrossdifferentEEGmeasures scales,seems doubtful. I would like to stress that all this relates
andexperimentalparadigms,moststudiesfindEEGcorrelatesof tospontaneousandinducedoscillations.Therearetoofewstud-
these processes within the DMN; most frequently in the MPFC iesmeasuringevokedoscillatoryresponsestoself-relatedstimuli,
and other midline structures. This is remarkable, because mid- whichmakeitimpossibletoderiveevenpreliminaryconclusions.
linestructuresarenotdirectlyaccessiblefromthescalpandtheir Giventheabovediscussedassociationbetweenself-referentialpro-
activitycouldbeonlymodeledbymeansofsourceimagingtech- cessing and the P300 and existing evidence on crucial role of
niques,whichhavelowspatialresolutionandwell-knownother deltaandthetaoscillationsinshapingthisERPcomponent(see
limitations. New information, which comes from EEG research e.g., Knyazev, 2007, 2012 for a review), one would expect that
and may not be obtained in fMRI studies concerns the tempo- evoked responses in these frequency bands, particularly in the
ral dynamics of self-referential processing and involvement of MPFC,shouldbeassociatedwithself-referentialprocessing(see
oscillations. Although some studies find self-processing-related e.g.,Miyakoshietal.,2010).AnotherverypromisingfieldofEEG
differencesintheERPwaveforms(Keyesetal.,2010)orevoked research,whichsofarhasattractedonlylimitedattentioninthe
gamma response (Roye et al., 2010) very early (170 and 40ms, studyofself-referentialprocessing,isthestudyofphaserelation-
respectively), most other studies show these differences at later shipsbetweendifferentcorticalregionsinafrequencybandand
stages,which are most frequently associated with the P300 ERP betweendifferentfrequencies(PalvaandPalva,2012;Schutterand
component. Given the well-known functional correlates of this Knyazev,2012).
component(i.e.,saliencedetection),thisevidencehighlightsthe
salience of self-related information and the tendency to pick it ACKNOWLEDGMENTS
outfromthestreamofexternalstimuli.Mostimportantandstill ThisworkwassupportedbygrantsoftheRussianFoundationfor
most disputable question is the relation of EEG oscillations to BasicResearch(RFBR)No.11-06-00041-aand13-04-00182-a.
REFERENCES and not strength of activa- integrativetheory.Int.J.Psychophys- doi:10.1371/journal.pone.000
Allik, J., and McCrae, R. R. (2004). tion in temporo-parietal cortex iol. 26, 5–29. doi:10.1016/S0167- 3984
Toward a geography of per- positively correlates with schizo- 8760(97)00753-8 Berkovich-Ohana, A., Glicksohn,
sonality traits: patterns of typy. Neuroimage 35, 326–333. Bennett, C. M., and Miller, M. B. J., and Goldstein, A. (2012).
profiles across 36 cultures. J. doi:10.1016/j.neuroimage.2006. (2010).Howreliablearetheresults Mindfulness-induced changes in
Cross Cult. Psychol. 35, 13–28. 11.027 fromfunctionalmagneticresonance gammabandactivity–implications
doi:10.1177/0022022103260382 Basar, E. (2008). Oscillations in imaging? Ann. N. Y. Acad. Sci. for the default mode network,
Aron, A. R., Gluck, M. A., and Pol- “brain-body-mind” – a holistic 1191, 133–155. doi:10.1111/j.1749- self-reference and attention.
drack,R.A.(2006).Long-termtest- view including the autonomous 6632.2010.05446.x Clin. Neurophysiol. 123, 700–710.
retest reliability of functional MRI system. Brain Res. 1235, 2–11. Ben-Simon, E., Podlipsky, I., Arieli, doi:10.1016/j.clinph.2011.07.048
in a classification learning task. doi:10.1016/j.brainres.2008. A., Zhdanov, A., and Hendler, Berlad,I.,andPratt,H.(1995).P300in
Neuroimage 29,1000–1006.doi:10. 06.102 T. (2008). Never resting brain: responsetothesubject’sownname.
1016/j.neuroimage.2005.08.010 Basar,E.,Schurmann,M.,Basar-Eroglu, simultaneous representation of Electroencephalogr.Clin.Neurophys-
Arzy,S.,Mohr,C.,Michel,C.M.,and C., and Karakas, S. (1997). Alpha two alpha related processes in iol.96,472–474.doi:10.1016/0168-
Blanke, O. (2007). Duration oscillationsinbrainfunctioning:an humans. PLoS ONE 3:e3984. 5597(95)00116-A
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|9
Knyazev EEGcorrelatesofself-referentialprocessing
Billeke, P., Zamorano, F., Cosmelli, status. PLoS ONE 6:e17325. from environment to theory Dimitriadis, S. I., Laskaris, N. A.,
D., and Aboitiz, F. (2013). Oscil- doi:10.1371/journal.pone.0017325 of mind. Neuron 58, 306–324. Tsirka,V.,Vourkas,M.,and Mich-
latory brain activity correlates Buckner,R.L.,Andrews-Hanna,J.R., doi:10.1016/j.neuron.2008.04.017 eloyannis, S. (2010). What does
with risk perception and predicts and Schacter, D. L. (2008). The Corsi-Cabrera, M., Galindo-Vilchis, delta band tell us about cogni-
social decisions. Cereb. Cortex brain’s default network: anatomy, L., del-Río-Portilla, Y., Arce, tiveprocesses:amentalcalculation
doi:10.1093/cercor/bhs269 function and relevance to disease. C., and Ramos-Loyo, J. (2007). study. Neurosci. Lett. 483, 11–15.
Binder, J. R., Frost, J. A., Hammeke, Ann. N. Y. Acad. Sci. 1124, 1–38. Within-subjectreliabilityandinter- doi:10.1016/j.neulet.2010.07.034
T. A., Bellgowan, P. S. F., Rao, S. doi:10.1196/annals.1440.011 sessionstabilityofEEGpowerand Esslen, M., Metzler, S., Pascual-
M., and Cox, R. W. (1999). Con- Burianova, H., McIntosh, A. R., and coherent activity in women eval- Marqui, R., and Jancke, L.
ceptualprocessingduringthecon- Grady, C. L. (2010). A com- uated monthly over nine months. (2008). Pre-reflective and
sciousrestingstate:afunctionalMRI mon functional brain network Clin. Neurophysiol. 118, 9–21. reflective self-reference: a spa-
study.J.Cogn.Neurosci.11,80–93. for auto-biographical, episodic, doi:10.1016/j.clinph.2006.08.013 tiotemporal EEG analysis.
doi:10.1162/089892999563265 and semantic memory retrieval. Dastjerdi,M.,Foster,B.L.,Nasrullah,S., Neuroimage 42, 437–449. doi:10.
Blanke, O., Mohr, C., Michel, C. Neuroimage 49, 865–874. doi:10. Rauschecker,A.M.,Dougherty,R.F., 1016/j.neuroimage.2008.01.060
M., Pascual-Leone, A., Landis, 1016/j.neuroimage.2009.08.066 Townsend,J.D.,etal.(2011).Differ- Evans,N.,andBlanke,O.(2013).Shared
T., and Thut, G. (2005). Linking Buzsaki, G., and Draguhn,A. (2004). entialelectrophysiologicalresponse electrophysiology mechanisms of
out-of-body experience and self Neuronal oscillations in cortical during rest, self-referential, and bodyownershipandmotorimagery.
processing to mental own-body networks. Science 304, 1926–1929. non-self-referentialtasksinhuman Neuroimage 64, 216–228. doi:10.
imagery at the temporoparietal doi:10.1126/science.1099745 posteromedial cortex. Proc. Natl. 1016/j.neuroimage.2012.09.027
junction. J. Neurosci. 25, 550–557. Caharel, S., Poiroux, S., Bernard, Acad. Sci. U.S.A. 108, 3023–3028. Fabiani,M.,Gratton,G.,Karis,D.,and
doi:10.1523/JNEUROSCI.2612- C., Thibaut, F., Lalonde, R., and doi:10.1073/pnas.1017098108 Donchin,E.(1987).“Thedefinition,
04.2005 Rebai, M. (2002). ERPs associ- David,O.,Kilner,J.M.,andFriston,K.J. identificationandreliabilityofmea-
Brazier, M. A. (1967). The EEG in ated with familiarity and degree (2006).Mechanismsofevokedand surementoftheP300componentof
stress. Physiological and psycho- of familiarity during face recogni- induced responses in MEG/EEG. theevent-relatedbrainpotential,”in
logical aspects. Introduction. The tion.Int.J.Neurosci.112,1499–1512. Neuroimage 31,1580–1591.doi:10. Advancesinpsychophysiology,Vol.2,
search for the mechanisms of the doi:10.1080/00207450290158368 1016/j.neuroimage.2006.02.034 edsP.Ackles,J.Jennings,andM.G.
brain’s reactions to stress. Elec- Calhoun,V.D.,Adali,T.,Pearlson,G.D., deMunck,J.C.,Gonçalves,C.I.,Faes,T. H.Coles(Greenwich,CT:JAIPress),
troencephalogr. Clin. Neurophysiol. andPekar,J.J.(2001).Spatialand J.C.,Kuijer,J.P.A.,Pouwels,P.J.W., 1–78.
25(Suppl.),209. temporal independent component Heethaar,R.M.,etal.(2008).Astudy Fan,J.,Byrne,J.,Worden,M.S.,Guise,
Brazier,M.A.(1968).StudiesoftheEEG analysisoffunctionalMRIdatacon- ofthebrain’srestingstatebasedon K. G., McCandliss, B. D., Fossella,
activityoflimbicstructuresinman. tainingapairoftask-relatedwave- alpha band power, heart rate and J., et al. (2010). The relation of
Electroencephalogr.Clin.Neurophys- forms.Hum.BrainMapp.13,43–53. fMRI. Neuroimage 42, 112–121. brainoscillationstoattentionalnet-
iol.25,309–318.doi:10.1016/0013- doi:10.1002/hbm.1024 doi:10.1016/j.neuroimage.2008. works. J. Neurosci. 27, 6197–6206.
4694(68)90171-5 Cannon, R. L., and Baldwin, D. 04.244 doi:10.1523/JNEUROSCI.1833-
Brazier,M.A.(1969).AnalysisofEEG R. (2012). EEG current source deMunck,J.C.,Gonçalves,C.I.,Huij- 07.2007
activity recorded from electrodes density and the phenomenol- boom,L.,Kuijer,J.P.A.,Pouwels,P. Fingelkurts, A. A., and Fingelkurts,
implanted in deep structures of ogy of the default network. J.W.,Heethaar,R.M.,etal.(2007). A. A. (2010). Short-term EEG
thehumanbrain.Electroencephalogr. Clin. EEG Neurosci. 43, 257–267. Thehemodynamicresponseofthe spectral pattern as a single
Clin.Neurophysiol.26,535–536. doi:10.1177/1550059412449780 alpharhythm:anEEG/fMRIstudy. event in EEG phenomenology.
Britz,J.,VanDeVille,D.,andMichel, Canolty, R. T., and Knight, R. T. Neuroimage 35,1142–1151.doi:10. Open Neuroimag. J. 4, 130–156.
C. M. (2010). BOLD correlates (2010). The functional role of 1016/j.neuroimage.2007.01.022 doi:10.2174/1874440001004010130
of EEG topography reveal rapid cross-frequency coupling. Trends Debener,S.,Ullsperger,M.,Siegel,M., Fingelkurts, A. A., and Fingelkurts,
resting-state network dynam- Cogn.Sci.(Regul.Ed.)14,506–515. Fiehler, K., von Cramon, D. Y., A. A. (2011). Persistent oper-
ics. Neuroimage 52, 1162–1170. doi:10.1016/j.tics.2010.09.001 and Engel, A. K. (2006). Single- ational synchrony within brain
doi:10.1016/j.neuroimage.2010. Cantero,J.L.,andAtienza,M.(2005). trialEEG/fMRIrevealsthedynam- default-mode network and self-
02.052 The role of neural synchroniza- ics of cognitive function. Trends processing operations in healthy
Brookes, M. J., Wood, J. R., Steven- tion in the emergence of cogni- Cogn.Sci.(Regul.Ed.)10,558–563. subjects. Brain Cogn. 75, 79–90.
son, C. M., Zumer, J. M., White, tion across the wake-sleep cycle. doi:10.1016/j.tics.2006.09.010 doi:10.1016/j.bandc.2010.11.015
T. P., Liddle, P. F., et al. (2011a). Rev. Neurosci. 16, 69–83. doi:10. Decety, J., and Sommerville, J. A. Fingelkurts, A. A., Fingelkurts, A.
Changes in brain network activ- 1515/REVNEURO.2005.16.1.69 (2003). Shared representations A., Bagnato, S., Boccagni, C.,
ity during working memory tasks: Carmines,E.G.,andZeller,R.A.(1979). between self and others: a social and Galardi, G. (2012). DMN
a magnetoencephalography study. Reliability and Validity Assessment. cognitiveneuroscienceview.Trends operational synchrony relates
Neuroimage 55,1804–1815.doi:10. NewburyPark:SagePublications. Cogn. Sci. (Regul. Ed.) 7,527–533. to self-consciousness: evidence
1016/j.neuroimage.2010.10.074 Chen, A. C. N., Feng, W., Zhao, H., doi:10.1016/j.tics.2003.10.004 from patients in vegetative and
Brookes,M.J.,Woolrich,M.,Luckhoo, Yin,Y.,and Wang,P. (2008). EEG Delorme, A., and Makeig, S. (2004). minimally conscious states.
H.,Price,D.,Hale,J. R.,Stephen- defaultmodenetworkinthehuman EEGLAB: an open source tool- Open Neuroimag. J. 6, 55–68.
son,M. C.,et al. (2011b). Investi- brain: spectral regional field pow- box for analysis of single-trial doi:10.2174/1874440001206010055
gatingtheelectrophysiologicalbasis ers.Neuroimage41,561–574.doi:10. EEG dynamics including inde- Fischler, I., Jin, Y. S., Boaz, T. L.,
ofrestingstatenetworksusingmag- 1016/j.neuroimage.2007.12.064 pendent component analysis. Perry, N. W. Jr., and Childers,
netoencephalography. Proc. Natl. Chen, J. L., Ros, T., and Gruzelier, J. Neurosci. Methods 134, 9–21. D. G. (1987). Brain potentials
Acad.Sci.U.S.A.108,16783–16788. J. H. (2013). Dynamic changes doi:10.1016/j.jneumeth.2003.10.009 related to seeing one’s own
doi:10.1073/pnas.1112685108 of ICA-derived EEG functional Depue, R. A., and Collins, P. name. Brain Lang. 30, 245–262.
Broyd, S. J., Helps, S. K., and connectivity in the resting state. F. (1999). Neurobiology of doi:10.1016/0093-934X(87)90101-5
Sonuga-Barke, E. J. S. (2011). Hum. Brain Mapp. 34, 852–868. the structure of personality: Foster,B.L.,Dastjerdi,M.,andParvizi,
Attention-induced deactivations doi:10.1002/hbm.21475 dopamine, facilitation of incen- J. (2012). Neural populations
in very low frequency EEG oscil- Corbetta,M.,Patel,G.,and Shulman, tive motivation, and extraversion. in human posteromedial cor-
lations: differential localisation G. L. (2008). The reorienting Behav. Brain Sci. 22, 491–569. tex display opposing responses
according to ADHD symptom system of the human brain: doi:10.1017/S0140525X99002046 during memory and numerical
FrontiersinHumanNeuroscience www.frontiersin.org June2013|Volume7|Article264|10