#1. a neural model of voluntary and automatic emotion regulation implications for understanding the...

FEATURE REVIEW

A neural model of voluntary and automatic emotionregulation: implications for understanding thepathophysiology and neurodevelopment ofbipolar disorderML Phillips1,2, CD Ladouceur1 and WC Drevets1,3

1Department of Psychiatry School of Medicine, University of Pittsburgh, Pittsburgh, PA, USA; 2Department of PsychologicalMedicine, Cardiff University School of Medicine, Cardiff, UK and 3Section on Neuroimaging in Mood and Anxiety Disorders,NIH/NIMH, Bethesda, MD, USA

The ability to regulate emotions is an important part of adaptive functioning in society.Advances in cognitive and affective neuroscience and biological psychiatry have facilitatedexamination of neural systems that may be important for emotion regulation. In this criticalreview we first develop a neural model of emotion regulation that includes neural systemsimplicated in different voluntary and automatic emotion regulatory subprocesses. We then usethis model as a theoretical framework to examine functional neural abnormalities in theseneural systems that may predispose to the development of a major psychiatric disordercharacterized by severe emotion dysregulation, bipolar disorder.Molecular Psychiatry (2008) 13, 833857; doi:10.1038/mp.2008.65; published online 24 June 2008

Keywords: emotion regulation; voluntary emotion regulation; automatic emotion regulation;affective neuroscience; neurodevelopment; bipolar disorder

Introduction

Advances in the fields of cognitive and affectiveneuroscience and biological psychiatry have allowedus to identify the neural systems underlying emotionregulation and how abnormalities in these neuralsystems may be associated with the presence ofsymptoms of certain psychiatric disorders, particu-larly symptoms of mood disorders. We previouslydescribed a neural model of emotion perception andregulation based on a critical review of animal,human focal lesion and human neuroimaging stu-dies.1 Here, the main focus was the development of atheoretical framework for the study of functionalabnormalities in neural systems implicated in emo-tional dysregulation in various psychiatric disorders,including bipolar disorder (BD).

Since the publication of our neural model ofemotion perception and regulation there have beenseveral developments in the following research areas.First, there has been an increasing focus in the field ofneuroimaging on the study of neural systems under-lying emotion regulation in healthy, nonpsychiatricpopulations and the application of findings from

these studies to the examination of psychiatricpopulations. Second, the recent research agenda forthe Diagnostic and Statistical Manual of MentalDisorders, Version V (DSM-V) has emphasized theneed to apply these basic and clinical neuroscienceresearch findings to develop a framework for identify-ing biomarkers that reflect pathophysiological pro-cesses to facilitate earlier and more accurate diagnosisof psychiatric disorders.24 This is particularly rele-vant to BD, a disorder that is frequently misdiagnosedor diagnosed too late for early and successfultreatment.5,6 For mood disorders in general, and BDin particular, these biomarkers could include func-tional and/or structural abnormalities in neuralsystems important for emotion regulation. In parallelwith these clinical advances have been an increasingnumber of neuroimaging studies in BD that arebeginning to show specific abnormalities in thestructure and function of neural regions implicatedin emotion regulation in individuals with the dis-order. Finally, there is emerging work in the field ofdevelopmental affective neuroscience examining thedevelopment of neural systems of emotion andemotion regulation in youth with and withoutpsychiatric disorders. Here, there is increasing inter-est in applying clinical neuroscience research find-ings to the study of populations who are genetically atrisk of developing disorders such as BD, where thegoal is to identify neural abnormalities that reflecthigh risk of future development of psychiatric

Received 12 February 2008; revised 17 April 2008; accepted 14May 2008; published online 24 June 2008

Correspondence: Professor ML Phillips, Department of PsychiatrySchool of Medicine, University of Pittsburgh, 121 MeyranAvenue, Pittsburgh, PA 15213, USA.E-mail: [email protected]

Molecular Psychiatry (2008) 13, 833857& 2008 Nature Publishing Group All rights reserved 1359-4184/08 $30.00

www.nature.com/mp

disorder to target individuals for early, preventativeinterventions.

Such advances in the field of affective neuroscienceand biological psychiatry have provided the motiva-tion for the refinement of our original neural model,specifically to expand upon neural systems impli-cated in the different voluntary and automaticcomponent processes and subprocesses inherent toemotion regulation and to develop a new neuralmodel of emotion regulation. We then use this as aframework with which to examine findings fromexisting studies to increase understanding of theneural basis of emotion dysregulation in individualswith BD and in youth at risk for subsequent develop-ment of the disorder. There are two important longer-term implications of this approach. First, this canhelp facilitate identification of biomarkers reflectingpathophysiological mechanisms of BD to aid diag-nosis. Second, this approach can help identifybiomarkers reflecting abnormal neurodevelopmentalprocesses that may characterize individuals most atrisk of future development of BD.

The main goals of this review are therefore asfollows:

(1) Development of a new neural model of emotionperception to focus on examination of the neuralsystems implicated in the different voluntary andautomatic component processes and subprocessesimplicated in emotion regulation

(2) Employment of this new neural model as aframework with which to examine the neuralbasis of emotion dysregulation in BD

(3) Employment of this new neural model as aframework to examine abnormalities in the devel-opment of neural regions implicated in emotionregulation that may be already present in youth atrisk for, and in youth with, BD, and that, in turn,may represent biomarkers associated with thedevelopment of the disorder.

Before introducing the components of our proposedneural model of emotion regulation, we first providean overview of the anatomy of the prefrontal corticalregions upon which we focus in our model, andfindings that have elucidated the functional connec-tivity between these prefrontal cortical regions andsubcortical limbic regions implicated in emotionprocessing.

Anatomy of prefrontal cortical neural regionsimplicated in emotion regulation

The regions of the prefrontal cortex (PFC) that havemost consistently been implicated in cognitivecontrol processes, including decision-making andemotion regulation, include the orbitofrontal cortex(OFC), dorsomedial prefrontal cortex (MdPFC),anterior cingulate gyrus (ACG), dorsolateral prefrontalcortex (DLPFC) and ventrolateral prefrontal cortex(VLPFC; for example see Krawczyk,7 and see follow-ing section). These regions in primates comprise a

mixture of dysgranular, granular and agranular cor-tical areas, including an internal granular layer IV.Ventromedial areas of the PFC (that is, OFC, MdPFCand ACG) develop relatively early, and are involvedespecially in the control of emotional behaviors,whereas lateral prefrontal cortical regions (that is,DLPFC and VLPFC) develop late, and are principallyinvolved in higher executive functions.8 The OFC,MdPFC and the ACG are large, heterogeneous corticalregions that cover the medial wall and ventral surfaceof the frontal cortex. The OFC, including Brodmannareas (BAs) 1114 and medial BA 47, MdPFC (BA 10/32) and parts of the ACG, are regions of the PFC thatare the most densely connected with the amygdalaand other subcortical limbic and paralimbic regions.9

The lateral PFC can be divided into DLPFC, compris-ing BAs 9 and 46,10 and VLPFC, comprising BAs 45and 47.11 BA 44, caudal to BA 46, subserves cognitivefunctions, in particular deductive reasoning andlanguage perception,12 which make it functionallyaligned with the DLPFC. The DLPFC is connected tothe OFC, and to a variety of subcortical neuralregions, including the thalamus, the dorsal striatum(dorsal caudate nucleus), the hippocampus andprimary and secondary cortical association areas,including posterior temporal, parietal and occipitalareas.13 There are also dense connections betweenOFC, MdPFC, and the ACG and DLFPC and VLPFC9

that provide indirect connections between subcorticalregions implicated in processing emotional informa-tion and lateral prefrontal regions implicated incognitive and higher-order executive processing.

The intrinsic cortico-cortical connections of theOFC and the MdPFC can further be described asreflecting two distinct networks.9 The orbital pre-frontal network involves multisensory inputs into theOFC and MdPFC regions and as such providesintegration of sensory information, primarily informa-tion of an affective value.9 The medial prefrontalnetwork involves the integration of visceromotorinformation. Both of these networks have connectionswith subcortical limbic structures. In particular, theorbital network has connections with the ventralmedial part of the basal nucleus of the amygdalawhereas the medial network has connections with theventral lateral part of the nucleus.9 There are alsodistinct connections between these neural networksand the striatum, thalamus, hypothalamus and brain-stem.9 Although distinct, these networks have inter-connections that allow convergence of sensorimotorintegration and visceromotor control in the proces-sing of emotionally salient information, and therebyfacilitate the regulation of emotional behavior.

More recent work most relevant to the formulationof our following neural model of voluntary andautomatic emotion regulation suggests that the pat-tern of projections to and from the OFC and theMdPFC and subcortical regions such as the amygdalaallows for the identification of specific feedback andfocal feedforward pathways. In particular, a recentstudy by Ghashghaei et al.14 examining output and

Neural model of emotion regulationML Phillips et al

834

Molecular Psychiatry

input patterns of connections between prefrontalcortices and the amygdala in nonhuman primatesconfirmed that caudal OFC and ACG areas havestrongest connections with the amygdala. Here, theauthors provided novel evidence for connectionsfrom the amygdala to cortical layers I and II of thesePFC regions, which they interpreted as being im-plicated in the focus of attention to motivationallyrelevant stimuli. Furthermore, they provided evi-dence for focal feedforward projections from theamygdala to the middle layers of caudal OFC andACG, which they interpreted as serving to provideinformation about the emotional salience or signifi-cance of external sensory stimuli, and potentiallycontributing to the role of the OFC in the rapidperception of reward contingencies.15 Input connec-tions from cortical layer III (posterior OFC and areasof the ACG) to the amygdala were also identified andinterpreted as involved in conveying information tothe amygdala about internalized emotions such asjealousy, embarrassment and guilt. Finally, theauthors demonstrated connections between OFC andlateral PFC, which may help mediate voluntaryemotion regulation processes, given that lateral PFCregions have limited output to the amygdala (seesection below).

Prefrontal corticalsubcortical limbic functionalconnectivity

Functional neuroimaging studies employing func-tional connectivity analyses afford the possibility ofdescribing correlations between activities in differentneural regions during task performance. Research inthe field of affective neuroscience using functionalconnectivity analyses is still in its infancy. Never-theless, the few studies that have been conducted todate have yielded findings that can inform under-standing of the role of different prefrontal regions inemotion regulation. One recent study16 employingfunctional connectivity analyses to examine neuralactivity in response to fearful and angry faces inhealthy adults yielded a path model that compriseda system linking the amygdala to subgenual ACG(BA25) to supragenual cingulate cortex (BA32) andback to the amygdala. In addition, the modelindicated a significant information-processing pathfrom OFC to DLPFC (BA46), and also between theamygdala and OFC. These findings support anatomi-cal data demonstrating that the OFC is extensivelyand reciprocally connected with the lateral and dorsalregions of the PFC, and with the amygdala, andsupport the hypothesis that the OFC may mediateconnections between higher-order dorsolateral pre-frontal regions and subcortical limbic regions such asthe amygdala during emotion regulation.

Other studies, using emotion-labeling paradigms,which are associated with voluntary control ofresponses to emotional stimulithinking rather thanfeeling about emotion17have also indicated that theright VLPFC may have an inhibitory role upon the

amygdala, and that this may be mediated by theOFC.1720 These findings are paralleled by reports ofinverse patterns of activity in amygdala and OFC inresponse to ambiguous emotional stimuli, surprisedfacial expressions, depending upon whether thesestimuli were interpreted positively or negatively.21

More negative interpretations of surprised faces wereassociated with greater signal changes in the rightventral amygdala, although more positive interpreta-tions, with greater signal changes in OFC. Anotherstudy provided further evidence for functional inte-gration of OFC, MdPFC and VLPFC in the response toemotionally salient information. Here, functionalconnectivity analyses of neural activity to positivevs negative emotional sentence cues revealed apositive correlation between ventral striatal, OFCand MdPFC activity to positive sentences, and apositive correlation between ventral amygdala,MdPFC and VLPFC activities to negative sentences.Findings in this study also revealed a negativecorrelation between OFC, MdPFC and VLPFC activityto negative sentences, however, suggesting differen-tial roles of OFC and VLPFC in the response topositive vs negative emotional stimuli.22

Functional connectivity findings have also indi-cated a dorsal-rostral division in ACC modulation ofthe thalamussensory cortex pathway in response toemotionally salient stimuli, with the dorsal ACCshowing a positive modulation of this pathway, andthe rostral ACC, an inverse relationship.23 Findingsfrom this study also indicated an inverse interactionbetween the dorsal and rostral ACC with the directthalamusamygdala pathway, providing additionalevidence for roles of dynamic functional relationshipsbetween thalamo-amygdala and cortical regions, andfor a functional differentiation ventral/dorsal prefron-tal neural systems.

These findings therefore implicate different regionsof the PFC, including OFC, MdPFC, VLPFC and ACG,and further suggest distinguishable roles of theseregions in the regulation of activity within subcorticallimbic regions in the response to emotionally salientinformation. We next discuss specific neural modelsof emotion regulation that have been informed bythese data.

Neural models of emotion regulation

Thompson24 proposed that emotion regulation con-sists of extrinsic and intrinsic processes responsiblefor monitoring, evaluating, and modifying emotionalreactions, especially their intensive and temporalfeatures, to accomplish ones goals (page 28). Grossand Thompson25 later added that these processes maybe automatic or controlled, conscious or uncon-scious, and may have their effects at one or morepoints in the emotion generative process (page 8).Although other definitions of emotion regulation havebeen proposed, this definition captures the complexnature of emotion regulation, including the existenceof subprocesses that may be either voluntary or


835


automatic. We highlight two previous models ofemotion processing that focused on neural regionsimplicated in emotion regulation.

Phillips et al. model

In our previous model, we focused on neural regionsimplicated in processes involved in emotion percep-tion, including the regulation of affective state andemotional behavior. Our review of the existinganimal, human lesion and human functional neuroi-maging literature led us to propose a neural modelof emotion perception and regulation.1 We high-lighted the role of a ventral neural system, includingamygdala, insula, ventral striatum (that is, ventralcaudate nucleus, putamen), ventral regions of theACG and ventral regions of the PFC, in particular,ventromedial PFC/OFC, in the identification ofemotionally salient stimuli, and mediation of auto-nomic responses to emotionally salient stimuli asso-ciated with the generation of an emotional state. Incontrast, we highlighted the role of a dorsal neuralsystem, including the hippocampus and dorsalregions of the ACG and PFCbrain regions thatsupport cognitive processes such as selective atten-tion, planning, performance monitoring and volun-tary regulation of emotional states. We proposed thatvoluntary regulation of emotion is supported pre-dominantly by the dorsal system. We also proposed,however, that reciprocal functional relationships mayexist between the ventral and dorsal systems, whichmay be mediated by the ventral medial region of thePFC and which may support both voluntary andautomatic regulation of emotion.

Ochsner and Gross model

More recently, Ochsner and Gross26 proposed a neuralmodel of emotion regulation that focuses on theinteraction between bottom-up emotion appraisal andtop-down cognitive control processes centered onsubcortical affective appraisal systems (amygdala,basal ganglia) and prefrontal cortical and cingulatesystems, respectively.2628 Their model includes twotypes of top-down appraisal systems. The first typeinvolves dorsomedial and dorsolateral prefrontalcortical systems. These systems are implicated intop-down description-based appraisal, allowing thegeneration of mental representations of affectivestates, as well as reappraisal to regulate emotion.The second type involves ventral prefrontal corticalsystems. These systems are implicated in top-downoutcome-based appraisal, important for the learningof associations between outcomes and precedingchoices or events to regulate emotion.

Commonalities and differences between these neuralmodels of emotion regulation

Both of the above models implicate top-down pro-cesses in emotion regulation, and center top-downemotion regulatory processes in dorsal prefrontalcortical regions, and bottom-up emotion generation

in subcortical, limbic neural regions. There are somedifferences, however, between the models.

One issue is the role of the ventral PFC, specificallythe OFC, in emotion regulation. In our previousmodel, we argued for a role of OFC regions in thegeneration of emotional states based upon evidencefrom lesion analysis studies in humans, monkeys androdents and neuroimaging studies in humans duringnormal and pathological emotional states.1,29,30 Incontrast, Ochsner and Gross highlighted the role ofthis region in outcome-based appraisal as a processimportant for emotion regulation. This is paralleledby an increasing number of studies that point to theOFC, in tandem with ventral striatal regions impli-cated in expectancy or anticipation of emotionallysalient future events, in outcome-based learning.31,32

Another major issue is the number of differentsubprocesses implicated in emotion regulation. Inaddition to the separation of emotion regulation intothe two different types of appraisal described above,Ochsner and Gross28 proposed a hypothetical con-tinuum to organize different subprocesses involved inthe cognitive control of emotion. At one end of thecontinuum is the exclusive use of attentional control,including engagement or disengagement of attentionto emotional stimuli, and at the other end is theexclusive use of cognitive change, including theabove top-down appraisal and reappraisal processes.There is also a distinction between suppression andcognitive appraisal and re-appraisal as two differentemotion regulatory strategies. Suppression, definedas a response-focused emotion regulatory strategy,involves the inhibition of ongoing emotion-expressivebehavior. Although suppression alters the behavioralexpression of emotion, it produces mixed physiolo-gical effects (for example, decreased heart rate andincreased sympathetic activity).3336 Reappraisal, de-fined as an antecedent-focused strategy, involves theattempt to alter the emotional meaning of originallysalient stimuli but it does not lead to increasedsympathetic arousal. It has therefore been argued thatreappraisal is a more effective strategy for theregulation of negative affect.26,28,33,34,37 The extent towhich these subprocesses are subserved by differentneural systems remains unclear.

An important, but somewhat neglected issue is theextent to which emotion regulation can be subdividedinto voluntary vs automatic component processes andsubprocesses. It could be argued that these differentsubprocesses operate in parallel and possibly simul-taneously with emotion appraisal and emotion gen-eration processes and, as such, it may be deemedfutile to attempt to study voluntary and automaticsubprocesses as separate entities. If, however, we areto begin to understand the mechanisms of emotionregulation, the underlying neural systems, theirdevelopment and subsequently how they may bealtered in individuals with mood disorders, it willbe necessary to identify the different subprocessesof emotion regulation and their correspondingneural systems.38 Understanding automatic emotion


836


regulatory subprocesses may be particularly relevantto the study of emotion regulation development inchildhood and adolescence and to the study ofindividuals with BD, in whom the inability toregulate emotion and resulting behavioral distur-bances occur frequently without insight or subjectiveawareness.39 Although previous models,24,25 includ-ing our own,1 alluded to the fact that emotionregulation can occur either voluntarily or automati-cally, to our knowledge there has been no integrationof these voluntary and automatic subprocesses intoone neural model of emotion regulation.

The above models of emotion regulation can beintegrated into a theoretical framework to includedifferent types of voluntary and automatic behavioraland cognitive control subprocesses important foremotion regulation. These include voluntary beha-vioral control, attentional control and cognitivechange, and automatic behavioral control, attentionalcontrol and cognitive change subprocesses (Table 1).We, therefore, next briefly review findings fromanimal studies and from human lesion and neuroima-ging studies that can help to increase understandingof the neural systems implicated in these voluntary vsautomatic emotion regulatory subprocesses. Our re-view of the recent literature is not exhaustive butfocuses on examples of studies providing evidence forthe presence of specific subprocesses of voluntaryand automatic emotion regulation.

Voluntary subprocesses

Voluntary behavioral control (suppression)A small number of human neuroimaging studies haveexamined the neural correlates of voluntary behavior-al control in response to emotional stimuli usingdifferent methods to train volunteers to suppressemotional expression. One such method is to trainvolunteers to keep their face still while viewing filmclips such that another person watching their facewould not be able to detect the emotion beingexperienced.33,36,40 Here, suppression of emotion-expressive behavior, associated with reduced negativeemotion experience and behavior, was associatedwith activity within bilateral dorsomedial PFC(MdPFC), right dorsolateral PFC (DLPFC) and leftventrolateral PFC (VLPFC). Greater, rather thanreduced, bilateral subcortical limbic (that is, amygda-la, insula) activity accompanied voluntary suppres-sion, however, suggesting that although suppressionmay constitute an emotion regulation strategy, it maynot be the most beneficial as a therapeutic strategy.28

Another behavioral control method is to trainvolunteers to view emotionally provoking stimuli asdetached observers.41 Here, attempts to inhibitsexual arousal in response to erotic film excerptswere associated with greater right superior PFC/DLPFC and right rostral ACG activity. Using a similarapproach, Levesque et al.42 showed that suppressionof sadness was related to greater activity in rightOFC and right DLPFC in healthy women. Positive

correlations were also demonstrated between self-report ratings of sadness and activity in these regions.

In a recent study, Ohira et al.43 used positronemission tomography (PET) to examine regionalcerebral blood flow changes in PFC that accompaniedautonomic nervous system responses associated withemotion suppression (attempting to remain calm anddiminish any emotional response to emotionallysalient pictures). Suppression of emotional responseswas associated with greater activity within leftMdPFC, VLPFC, bilateral anteromedial and caudalOFC, as well as left rostral-ventral ACG. Suppressionwas also associated with increased amplitude of skinconductance responses, which, in turn, was posi-tively correlated with right OFC activity. In a furtherstudy, suppression of neutral, negative and relation-ship-related thoughts was associated with activitywithin left dorsal ACG and MdPFC.44

Findings from the above studies indicate thatneural regions consistently implicated in voluntarybehavioral control of both positive41 and negative42

emotions include DLPFC and VLPFC, in particular,right DLPFC and left VLPFC. These regions maysupport processes involved in regulating inner statesto achieve desired outcomes.78 Other regions impli-cated in voluntary behavioral control of emotionalbehaviors include left/bilateral MdPFC, left rostralACG and right/bilateral OFC. Given that lateralregions of the PFC share only sparse direct connec-tions with subcortical limbic regions such as theamygdala, it has been proposed that the modulationof subcortical limbic regions by lateral and dorsalregions of the PFC during voluntary behavioralcontrol of emotion may occur indirectly via OFC,79

which has extensive reciprocal anatomical connec-tions with both lateral prefrontal cortical and sub-cortical limbic structures,9,80 and has been implicatedin regulating autonomic nervous system responsesduring emotion suppression.43

Voluntary attentional controlIncluded in voluntary attentional control are (1)selective attention, to direct or redirect attentiontoward goal-related stimuli, and (2) inhibition, todistract from goal-irrelevant stimuli. For example,using an affective Go/NoGo functional magneticresonance imaging (fMRI) paradigm, Goldsteinet al.45 examined the neural correlates of responseinhibition to emotional words in healthy adultvolunteers. The task involved responding to wordstimuli written in normal font (Go trials) and inhibit-ing responses to words written in italics (NoGo trials).The valence of these words was positive, negative orneutral. Findings revealed a frontolimbic networkassociated with a valence by response inhibitioninteraction. Specifically, this network included pre-frontal (right DLPFC, right dorsal ACG and bilateralOFC), left subcortical (amygdala), bilateral paralimbic(hippocampus/parahippocampus) and right parietalcortical regions, particularly during inhibition ofresponses to negative emotional stimuli.


837


Table 1 Voluntary and automatic subprocesses of emotion regulationexamples of research studies

Components ofneural model

Examplesof research studies

Number ofsubjects

Task Key contrast Method Relevant brainregions

Voluntary subprocessesVoluntary behavioralcontrol

Suppressionof emotionexpression

Goldin et al.40 17 (healthysubjects)

Affect ratings after viewingfilm clips and still pictures(negative and neutral);conditions: watch, suppress,reappraise

Suppress > watchnegative

fMRI AmygdalaR VLPFC

Beauregard et al.41 10 (healthysubjects)

Affect ratings after viewingfilm clips (sexually arousingand neutral); conditions:normal reaction, attemptedinhibition

Sexual arousal minusneutral in the attemptedinhibition condition

fMRI R DLPFCR rostral ACG

Levesque et al.42 20 (healthywomen)

Affect ratings after viewingfilm clips (sad and neutral);conditions: normal reaction,suppression

Sad minus neutralin theSuppression condition

fMRI R OFCR DLPFC

Ohira et al.43 10 (healthysubjects)

Affect ratings after viewingpictures (negative, positiveand neutral); conditions:attend, suppress

Suppress minus attend PET L MdPFCVLPFCBi anteromedialand caudal OFCRostralventral ACG

Gillath et al.44 20 (healthywomen)

Thought suppressionparadigm (negative,neutral); conditions:do not think, think, do notthink relationship, thinkrelationship

Do not think > thinkDo not thinkrelationship > thinkrelationshipDo not think negative >think negative

fMRI L dorsal ACGL MdPFC

Voluntary attentionalcontrol

Selective attention Goldstein et al.45 14 (healthysubjects)

Emotional linguisticgo/NoGo task

Contrasts containingNegative NoGocondition

fMRI L amygdalaBi hippocampus andparahippocampusR dorsal ACGR DLPFCBi OFC

Overcominginterference fromemotional distractors

Erk et al.46 12 (healthysubjects)

Sternberg item recognitionparadigm with emotionalpictures as distractors

Valence by workingmemory load

fMRI R amygdalaR DLPFCL ventral striatumL DLPFC

Inhibitionof emotionalmotor response

Perlstein et al.47 10 (healthysubjects)

Emotional working memorytask with emotional picturesas cues and probes

Valence by workingmemory load

fMRI R DLPFCMedial OFC

Neuralm

odelofemotion

regulationM

LPhillips

etal

838

Mo

lecu

lar

Psych

iatry

Table 1 Continued



Number ofsubjects


Luo et al.48 14 (healthysubjects)

Lexical decision-making taskwith supraliminal andsubliminal emotionaldistractors

Valence by visibility fMRI L DLPFCR posterior parietalcortexR dorsal ACG

Voluntary cognitivechange

Reappraisalexpectancyof forthcomingevents

Ochsner et al.49 15 (healthysubjects)

Cognitive reappraisal task(negative and neutral);conditions: attend,reappraise

Reappraise > attend fMRI R amygdalaL OFCL DLPFCL VLPFCL MdPFC

Phan et al.50 14 (healthysubjects)

Cognitive reappraisal task(negative and neutral);conditions: maintain,suppress (reappraise)

Suppress(reappraise) > maintain

fMRI L amygdalaR DLPFCBi dorsal ACGBi OFC

Banks et al.51 14 (healthysubjects)

Cognitive reappraisal task(negative, positive andneutral); conditions:maintain, reappraise

Reappraise > maintain fMRI L amygdalaBi DLPFCR MdPFCR subgenualACGBi OFC

Urry et al.52 19 (healthysubjects)

Cognitive reappraisal task(negative and neutral);conditions: increase,decrease, attend

Decrease minus attend fMRI L amygdalaBi MdPFC

Corlett et al.53 14 (healthysubjects)

Contingency learning task All retrospectiverevaluation events vscontrol items

fMRI R lateralPFC

Kalish et al.84 15 (healthysubjects)

Anxiety Induction Task(for warning of possibleelectric painful stimuli);conditions:(1) anticipation vs noanticipation and (2)regulation vs no regulation

(Regulation > noRegulation)Anxiety >(Regulation > noregulation)No Anxiety

fMRI R DLPFC

Automatic subprocessesAutomatic behavioralcontrol

Extinction Rosenkranz et al.54 Sample of rats Pavlovian conditioningin vivo

NA Intracellularelectrophysio-logical recordings

Lateral nucleus of theamygdala Infralimbiccortex (posteriorsubgenual ACG is theequivalent in humans)

Neuralm

odelofemotion

regulationM

LPhillips

etal

839

Mo

lecu

lar

Psych

iatry

Table 1 Continued



Number ofsubjects


Behavioral regulation Vidal-Gonzalezet al.55

76 rats Fear conditioning andextinction

NA Microstimulation Basal amygdalainfralimbic cortex

Milad and Quirk56 Sample of rats Auditory fear conditioning NA Intracellularelectrophysio-logical recordingsand microstimu-lation

Infralimbic cortex

Morgan et al.57 Male SpragueDawley rats

Classic fear-conditioningparadigm with extinctionand reacquisition

NA Lesion study Infralimbic cortex

McGinty et al.58 Male SpragueDawley rats

Pavlovian odor conditioning NA Intracellularelectrophysio-logical recordingsand micro-stimulation

Basolateral amygdalaInfralimbic cortexNucleus accumbens

Phelps et al.59 18 (healthysubjects)

Simple discrimination,partial reinforcementparadigm

Conditioned stimulus(CS ) minus absenceof conditionedstimulus (CS)

fMRI BisubgenualACGBi MdPFCDorsal ACG

Gottfried & Dolan60 18 (healthysubjects)

Olfactory aversive-conditioningand extinction paradigm

Conditioning minusextinction; extinctionminus conditioning

fMRI Amygdala OFC

Milad et al.61 14 (healthysubjects)

Differential fear-conditioningparadigm

Cortical thickness;conditioned stimulus(CS ) minus absenceof conditionedstimulus (CS)

Structural MRI;fMRI

Dorsal ACG

Automatic attentionalcontrol

Cognitive disengagement Armony and Dolan62 10 (healthysubjects)

Fear-conditioning paradigmand attention-orienting task

Conditioned stimulus(CS ) minus absenceof conditioned stimulus(CS); incongruentminus congruent

fMRI Left OFC

Repressive and avoidantpersonality styles

Pourtois et al.63 15 (healthysubjects)

Modified Dot-Probe task:designed to assess covert spatialselective attention modulated bythreatening stimuli

Fear valid > fear invalid fMRI Lateral OFCIntraparietal sulcus

Whalen et al.64 10 (healthysubjects)

Masked emotional facialexpressions block designparadigm; conditions:fearful, happy and fixation

Fearful face > happyface

fMRI Amygdala

Blair et al.65 22 (healthysubjects)

Affective Stroop withemotional pictures;conditions: negative,positive and neutral

Task by emotion fMRI L rostral ACGBi DLPFC bilateral OFCR MdPFC

Neuralm

odelofemotion

regulationM

LPhillips

etal

840

Mo

lecu

lar

Psych

iatry

Table 1 Continued



Number ofsubjects


Etkin et al.66 19 (healthysubjects)

Emotion conflict resolutionparadigm (emotional facialexpressions with congruentand incongruent emotionalwords)

High conflict minuslow conflict

fMRI; functionalconnectivity

AmygdalaRostral ACG

Vuilleumier et al.67 12 (healthysubjects)

Attentionemotion paradigm(pair of emotional facialexpressions and houses)

(Attended fearfulfaceattended neutralface) > (unattendedfearful faceunattendedneutral face)

fMRI AmygdalaFusiform cortexDorsal ACG

Pessoa et al.68 20 (healthysubjects)

Block design attentionemotion paradigm

Emotion by attentionalload

fMRI AmygdalaPosterior cingulatedMdPFC

Phillips et al.69 8 (healthy men) Emotional facial expressions(fearful, disgust, neutral);conditions: overt, covert

Overt vs covertpresentations of fearOvert vs covertpresentations of disgust

fMRI Amygdalaovert fearfulfacial expressionsInsulaovert disgustfacial expressions

Stein et al.16 83 (healthysubjects)

Block design face-matchingparadigm

Emotional faces minuscontrol task


AmygdalaSubgenual ACGLateral PFCOFCDLPFC

Williams et al.70 15 (healthysubjects)

Emotional facial expressions(fearful, neutral); conditions:conscious, nonconscious

Conscious: fearfulminus neutralNonconscious: fearfulminus neutral


Bi AmygdalaMdPFCDorsal ACGVentral ACG

Automatic cognitivechange

Covert appraisaland reappraisal

Hunkin et al.71 8 (healthy subjects) Associative encodingand single item encodingparadigms

Associative encodingof novel associationsvs associative encodingof familiar associations

fMRI Parahippocampus

Covert response(e.g., error) monitoring

van Veen and Carter72 12 (healthysubjects)

Eriksen flanker paradigm Incorrect responsesvs correct responses

ERP; sourcelocalization

Dorsal ACG

Covert learning thatserves to automaticallyadjust behavior

Luks et al.73

Bechara et al.74 44 controls 7subjects withdamage to OPFC

Iowa gambling task Advantageous decks(CD) minusdisadvantageous decks(AB)

Neuro-psycho-logical data

OFC

Ernst et al.75 20 (healthysubjects)

Risk-taking task:computerized gamblingcard game

Risk-taking task minuscontrol task

PET Bi OFC

Fukui et al.76 19 (healthysubjects)

Iowa gambling task Risky decisions minussafe decisions

fMRI Bi MdPFC

Neuralm

odelofemotion

regulationM

LPhillips

etal

841

Mo

lecu

lar

Psych

iatry

In another study, Erk et al.46 used an affectiveworking memory task. Here, healthy volunteers firstviewed a display of six capital letters followed by adelay period, during which they viewed simulta-neous presentations of a series of emotional pictures,each paired with a single lower case letter. Duringpresentation of the latter, participants indicatedwhether the single letter was part of the initial letterset. Greater right DLPFC activity was associated withreduced right amygdala activity during presentationof negative emotional pictures, whereas greater leftDLPFC activity was associated with reduced leftventral striatal activity during presentation of positiveemotional pictures. The involvement of specificprefrontal regions in downregulating subcorticalactivity to negative vs positive emotion stimuli isconsistent with previous findings using a differentversion of an affective working memory task.47 More-over, recent findings indicate that activity in leftDLPFC, right posterior parietal cortex and right dorsalACG is greater as a function of the emotional salienceand visibility of emotional information during alexical decision-making task.48

Findings from the above studies, therefore, indicatethat bilateral DLPFC, right dorsal ACG and rightparietal cortex are important in supporting voluntaryattentional control subprocesses. Again, it is possiblethat modulation of subcortical limbic regional activityby these lateral and dorsal regions of the PFC duringvoluntary control of attention away from emotionalmaterial may occur indirectly via the OFC. This latterregion has been demonstrated to be part of the neuralnetwork recruited to support this voluntary emotionregulatory subprocess.45

Voluntary cognitive change: reappraisalOne of the first studies to use fMRI to examine theneural correlates of voluntary reappraisal processesinvolved volunteers viewing negative and neutralpictures while being instructed either to experiencethe emotion that each picture elicited, or to reappraisethe emotional salience by reinterpretation of thenegative pictures so that they no longer elicited anegative emotional response.49 Reappraisal was asso-ciated with significantly greater activity in leftDLPFC, left VLPFC and left MdPFC. There was alsoreduced activity in the right amygdala and left OFC.Moreover, the magnitude of activity in left VLPFC wasnegatively correlated with magnitude of activitywithin left amygdala and left OFC.

Another study using a similar paradigm50 askedtrained subjects to either transform a depictedscenario into a positive event or to rationalize thecontent of an emotional picture. Downregulation ofnegative emotions by the use of reappraisal strategiesrecruited several prefrontal regions, including rightDLPFC, bilateral dorsal ACG and bilateral OFC.Activity in these regions was associated with de-creased activity in subcortical limbic regions, includ-ing the left amygdala. In addition, activity in bilateraldorsal ACG, bilateral DLPFC, right anterior insula andT

able

1C

on

tin

ued

Com

pon

en

tsof

neu

ral

mod

el

Exam

ple

sof

rese

arc

hst

ud

ies

Nu

mber

of

subje

cts

Task

Key

con

trast

Meth

od

Rele

van

tbra

inre

gio

ns

Law

ren

ce

et

al.

,u

nd

er

revie

w77

17

(healt

hy

men

)Io

wa

gam

bli

ng

task

(mod

ifie

dfo

rra

pid

even

t-re

late

dfM

RI)

Ris

ky

decis

ion

sm

inu

ssa

fed

ecis

ion

sfM

RI

LM

dP

FC

OF

C

Abbre

via

tion

s:A

CG

,an

teri

or

cin

gu

late

gyru

s;B

i,bil

ate

ral;

DL

PF

C,

dors

ola

tera

lp

refr

on

tal

cort

ex;

fMR

I,fu

ncti

on

al

magn

eti

cre

son

an

ce

imagin

g;

L,

left

;M

dP

FC

,d

ors

om

ed

ial

pre

fron

tal

cort

ex;

NA

,n

ot

ap

pli

cable

;O

FC

,orb

ital

fron

tal

cort

ex;

R,

righ

t;V

LP

FC

,ven

trola

tera

lp

refr

on

tal

cort

ex.

Stu

die

sre

vie

wed

inth

ista

ble

serv

eas

exam

ple

sof

em

pir

ical

evid

en

ce

insu

pp

ort

of

ou

rn

eu

ral

mod

el

of

em

oti

on

regu

lati

on

an

dd

on

ot

con

stit

ute

an

exh

au

stiv

ere

vie

wof

the

lite

ratu

re.


842


bilateral VLPFC was inversely associated with theintensity of negative affect. A recent study alsodemonstrated that activity in left amygdala covariedwith activity in bilateral DLPFC, right MdPFC, theright subgenual region of the ACG and bilateral OFCduring a reappraisal-based strategy to downregulatenegative affect.51 Moreover, the magnitude of amyg-dalaOFC/MdPFC coupling predicted successfulemotion regulation. A further study52 demonstratedan inverse functional coupling between bilateral OFCand left amygdala during the employment of reap-praisal strategies to downregulate negative affect.

Expectancy of forthcoming events and outcomes isan additional type of reappraisal of environmentalstimuli, and may serve as an additional voluntarycognitive control strategy for emotion regulation.Animal studies81 have associated expectancy ofsubsequent outcomes with activity in lateral PFC,whereas human neuroimaging studies have shownthat mismatch between expected and actual outcomes(prediction error) is associated with activity in rightlateral PFC.53,82 It has been also recently demonstratedthat prediction error-related activity in right lateralPFC is reduced in schizophrenia.83

Together, the above findings suggest that reapprai-sal of emotional stimuli to reduce negative affect isassociated with activity in a network of prefrontalcortical regions, including bilateral DLPFC, bilateralMdPFC and bilateral dorsal ACG (see Ochsner andGross26). A specific role of the right DLPFC inreappraisal has also been proposed.84 Activity in thislateral and medial prefrontal cortical network may, inturn, be mediated by bilateral OFC and the rightsubgenual ACG, as activity in these latter regions hasalso been observed during reappraisal of emotionalinformation.51,52

Automatic subprocesses

Automatic behavioral controlKey subprocesses of automatic behavioral controlinclude extinction of previously acquired behaviorand inhibition of the stress response. Rodent studieshave demonstrated that the infralimbic cortex, aregion that shares extensive anatomical connectionswith the amygdala,85,86 has a role in inhibiting activitywithin the amygdala.54,55 The infralimbic cortex hasalso been implicated in the extinction of previouslyconditioned behaviors.56,57,87,88 A recent study inrodents has indicated that projections from thebasolateral amygdala, which encodes affective infor-mation, and the MdPFC converge within the ventralstriatum (nucleus accumbens; McGinty and Grace58),suggesting that the basolateral amygdala may recruitthe MdPFC to drive specific sets of nucleus accum-bens neurons and thereby influence behavioralresponses to conditioned cues. Thus, both infralimbiccortex and MdPFC in rodents appear to be involved inregulation of amygdala-driven behaviors.

Functional neuroimaging studies in humans im-plicate bilateral regions of the subgenual region of the

ACG, together with bilateral MdPFC and dorsal ACGin fear extinction.59 The posterior subgenual ACG hasbeen identified as the human equivalent of the ratinfralimbic cortex.86,89 Other findings highlight therole of bilateral OFC in extinction learning,60,61 and arole for the right OFC in extinction learning more thanaversive conditioning.60

The infralimbic cortex has been also implicated,however, in the encoding of the predictive value ofemotionally salient stimuli in rodents through closeinterconnections with the basolateral amygdala,9,90

whereas the OFC has been associated with themediation of anxious temperament in nonhumanprimates.91 In humans, the left OFC has beenimplicated in aversive conditioning and, in morelateral left OFC, the representation of the value ofstimulus-outcome contingencies.60 The generation ofsad mood92 and depression,93 monitoring of internalstates in individuals with attachment-avoidant per-sonality styles44 have all been associated with greateractivity in left or bilateral subgenual ACG, whereasimprovement in depression after pharmacotherapy94

and deep-brain stimulation93 have been associatedwith decreases in activity within right and leftsubgenual ACG, respectively. One study, for example,demonstrated in individuals with attachment-avoi-dant personality styles relative to individuals withnonavoidant personality styles reduced deactivationof bilateral subgenual ACG during attempted volun-tary suppression of emotionally neutral, negative and/or relationship-related thoughts.44 The authors inter-preted these findings as evidence of a failure of top-down control to decrease monitoring of internal states(mediated by bilateral subgenual ACG) in individualswith avoidant personality styles. These findingstherefore suggest, in humans, roles of bilateral OFCand subgenual ACG in the automatic regulation ofemotional behavior that may be intrinsically linkedwith the roles of these structures in the encoding ofemotional salience.

Automatic attentional controlTo examine neural systems implicated in automaticattentional control of emotion processing, studieshave employed paradigms that implicitly directattention toward or away from emotional material.One such example is the dot probe task.95 Here, afaster response to a probe that appears at the samelocation of a previously presented emotional stimulusrelative to the response to a probe at the oppositelocation is interpreted as an automatic attentionalbias toward the emotional stimulus. Recent findingsimplicate the left OFC in automatic disengagement ofattention away from emotional stimuli (for example,fearful faces) during this type of paradigm.62,63

Another example is the emotional Stroop task.Here, individuals attend to a nonemotional stimulusfeature (for example, color or number of stimuli) atthe expense of the emotional content of the stimulus.Performance of this task has been associated withactivity in left rostral ACG, specifically, greater


843


activity in this region to negative emotional relative toneutral stimuli during early stages of task perfor-mance, although decreased activity in this regionoverall.64 The authors noted here that susceptibilityartifacts may have resulted in signal dropout in moreventral regions of the PFC, such as the OFC. Morerecent findings confirm the role of the left rostralACG, in addition to a wider network of bilateralDLPFC, bilateral OFC and right MdPFC, in perfor-mance of incongruent and congruent trials of anemotional counting Stroop paradigm.65 This studyalso demonstrated a role of the right MdPFC inperformance of negative and positive incongruenttrials. A modified emotional Stroop paradigm has alsobeen developed.66 Here, the task was to identify theexpression and ignore the emotion word labeldisplayed by either congruent or incongruent facialexpressionemotion label stimuli. Incongruent stimu-li were further subdivided into implicitly highconflict resolution trials, where incongruent stimuliwere preceded by incongruent stimuli, and implicitlylow conflict resolution trials, where incongruentstimuli were preceded by congruent stimuli. Facialexpression identification during the high conflictresolution, but not low conflict resolution, trials wasassociated with left rostral ACG, further highlightinga role for this region in the implicit resolution ofemotional conflict.

A further type of paradigm involves redirection ofattention away from facial expressions using coin-cidentally presented spatial cues and/or cognitivetask performance. In one study,67 the task involveddirection of attention toward fearful facial expressionpairs, or away from these pairs and toward nonfacepairs during a same-different task within-pair label-ing task. Here, the caudal region of the ACG wasactivated when fearful faces were task relevant andthat the rostral region of the ACG was activated whenthe fearful faces were task irrelevant. These findingswere interpreted as evidence for a role of the rostralACG in the direction of attention away from emo-tional stimuli by inhibiting processing of emotionalinformation.

Other studies have examined the extent to whichthe magnitude of amygdala activity to facial expres-sions is, in fact, dependent on available attentionalresources by either systematically manipulating theattentional load of a coincidentally presented visuos-patial task68,96 or by presentation of facial expressionscovertly, that is, below the level of subjectiveawareness.69,70,97,98 Findings from the former study68

indicated that increasing the attentional load requiredto perform a coincidentally presented visuospatialtask was associated with reduced amygdala activity.Findings from the latter group of studies provideconflicting data of preserved97,98 but also absent69

amygdala activity to covert presentations of facialexpressions. Functional connectivity analyses havefurthermore suggested that covert processing ofthreat-related facial expressions may be associatedwith a positive feedforward subcortical pathway to

the amygdala, whereas awareness of these facialexpressions was associated with negative connectiv-ity within both cortical and subcortical pathways tothe amygdala.70 Reentrant feedback between theamygdala and cortical and subcortical regions maybe necessary, therefore, for awareness of threat.70

Together, these studies suggest that processing emo-tionally salient stimuli per se may require a certainlevel of automatic or covertly directed attention,associated with amygdala activity, whereas automaticredirection of attention away from emotion stimuli isdependent upon rostral ACG.

Automatic cognitive changeSubprocesses included within automatic cognitivechange comprise appraisal and reappraisal, monitor-ing of behavior and rule learning that all occurwithout subjective awareness. Implicit appraisal ofnovel contexts, covert error monitoring and covertrisk learning paradigms, such as gambling tasks, areall examples of paradigms that are, therefore, appro-priate paradigms for the examination of neuralregions underlying automatic emotion regulation.Regarding the first of these subprocesses, implicitappraisal and reappraisal, Gray and McNaughton99

described a septo-hippocampal system, centeredupon the hippocampus and septum, with the capacityto act as a general-purpose comparator to appraisepotential conflict between different goal-directedbehaviors, to facilitate exploratory rather than defen-sive patterns of behavior and to allow resolution ofgoal conflict. The parahippocampal gyrus, which hasdirect connections with the hippocampus and amyg-dala,100 has also been associated with implicitappraisal and reappraisal processes: for example,with novel face detection in humans (right parahip-pocampus101), and with the encoding of novelassociations occurring without subjective awarenessof the novelty of these associations (left parahippo-campus71). A recent functional connectivity study inhumans has provided, furthermore, evidence ofstrong connections between the amygdala and theparahippocampus (including the hippocampus16).

Covert error monitoring and covert risklearning paradigms, such as gambling tasks areparadigms relevant for examination of neural regionsunderlying the second and third types of subpro-cesses described above that are involved in automaticemotion regulation: behavior monitoring and rulelearning occurring without subjective awareness.Error monitoring has been linked with midline dorsalACG.72,102 More specifically, a role for dorsal ACG hasbeen postulated that involves the monitoring ofconflict as an index of task difficulty, and associatedautonomic activity, during action or strategy selec-tion.102 Informed preparation for conflict processing,furthermore, has been linked with decreased ACGactivity.73 The dorsal ACG has also been associatedwith the mediation of fear expression,61 in furthersupport of the role of this structure in the mediationof autonomic activity.


844


Performance of gambling tasks has been mostconsistently associated with bilateral OFC in humanlesion studies,74,103 and in neuroimaging studies,75

highlighting a role of the OFC in outcome-basedappraisal. Recent neuroimaging findings also impli-cate bilateral (left more than right) MdPFC in theimplicit learning of successful strategies duringgambling task performance,76,77 however, that mayrelate to a potential role for this region in switching ofattention between internally generated thoughts andexternal stimuli104 and the use of internal signalsgenerated by the OFC to select goal-directedresponses.77

Together, these findings implicate bilateral hippo-campus and parahippocampus, midline dorsal ACGand bilateral OFC and MdPFC in automatic cognitivechange subprocesses relevant to automatic emotionregulation. The connections between hippocampusand parahippocampus with OFC and ACG throughthe subiculum and entorhinal cortex9,90 make itplausible that these regions may function as a neuralnetwork during automatic emotion regulation.

Summary and integration: a new neural model ofemotion regulationFindings from human neuroimaging studies indicateroles of several prefrontal cortical regions in sub-processes associated with voluntary emotion regula-tion, including voluntary behavioral control,attention redirection and cognitive change. Dorsalprefrontal cortical regions, including bilateral DLPFC,bilateral MdPFC and bilateral dorsal ACG, have beenmost consistently linked with these subprocesses.The right DLPFC may be implicated in particular withvoluntary cognitive change, that is, reappraisal.Activity in these regions may be mediated by bilateralventromedial prefrontal cortical regions such as theOFC that have direct connections with subcorticalneural regions underlying the identification andinitial processing of emotional material. By contrast,lesion studies in humans and experimental animalsand human neuroimaging studies highlight bilateralsubgenual ACG, bilateral OFC, left rostral ACG andbilateral MdPFC, midline dorsal ACG, and contribut-ing roles of the hippocampus and parahippocampus,in the different subprocesses associated with auto-matic emotion regulation.

In our original neural model of emotion processing,we highlighted that two systems in PFC may beimplicated in emotion regulation: a lateral prefrontalcortical system (including, for example, DLPFC andVLPFC) that is neocortical in origin and operates by afeedback mechanism, and a medial prefrontal corticalsystem (including, for example, OFC subgenual ACG,rostral ACG and MdPFC) that is paleocortical inorigin and operates by a feedforward mechanism.105

Our overview of the more recent literature suggeststhat the former neural system may be involved involuntary subprocesses, whereas the latter neuralsystem may subserve automatic subprocesses. TheMdPFC in particular may use feedforward inputs

from the OFC, as signals of internal states, to selectappropriate behaviors during automatic cognitivechange paradigms. These two neural systems may beactivated concurrently during regulation of emotionalstates and behaviors that are initially generatedby orienting and emotion perceptual processesin amygdala, ventral striatum and thalamus(Figures 1a and b).

Implications for understanding the pathophysiologyand neurodevelopment of bipolar disorderWe previously emphasized the importance of ourearlier neural model of emotion perception andregulation as a framework for understanding patho-physiological mechanisms underlying major psychia-tric disorders, in particular, mood disorders such asBD.29 Using our new neural model of emotionregulation as the framework, we now examine theextent to which recent neuroimaging studies informunderstanding of pathophysiological processes thatlead to the development of BD. We first examinefindings from structural and functional neuroimagingstudies in adult BD. Informed by current under-standing of the normal development of neural regionsimplicated in the different emotion regulatory sub-processes, we then review findings from existingneuroimaging studies of pediatric bipolar and bipolarat-risk populations that point to abnormal develop-ment of these neural regions in BD.

Adult bipolar disorder

Structural neuroimaging studies. There is a growingliterature from volumetric structural neuroimagingstudies indicating greater amygdala volumes in adultBD106,107 relative to age-matched healthy volunteers.More importantly, for the study of neural regionsimplicated in emotion regulation, studies havereported reduced volume and gray matter (GM)density in a variety of different prefrontal corticalregions in BD relative to healthy adults, but noconsistent lateralization of these findings.

Findings include, for example, reduced GM volumeand/or density in left or bilateral ACG and subgenualACG108112 and left, right or bilateral DLPFC, MdPFCand VLPFC,111,113115 although there are some reportsof greater bilateral VLPFC and dorsal ACG GMvolume in adult BD116 or no significant abnormalitiesin PFC volumes.117 Recent findings also indicate GMvolume reductions in left VLPFC in medicated adultBD compared with healthy individuals,118 althoughnot in first-episode BD.119 Genetic risk for BD hasbeen specifically associated with GM deficits in rightACG, including subgenual, rostral and more dorsalcomponents, and ventral striatum,120 and a recentreport provides evidence of an early neurodevelop-mental abnormality in bilateral ACG in adult BD.121

The pattern of volumetric abnormalities in BDresembles that of the changes in dendritic arboriza-tion during stress. Dendritic branching of pyramidalcells increases in some amygdala nuclei, but decreases


845


in the medial PFC and hippocampus, in response torepeated stress in rodents.122124 These findings raisethe possibilities that impaired emotion regulationmay contribute to the volumetric abnormalities found

in these structures in BD, by permitting stressresponses that are exaggerated in magnitude orduration, or by interfering with the modulation orresetting of stress-induced dendritic remodeling.123

Figure 1 Neural model of emotion regulation illustrating neural systems implicated in voluntary and automaticsubprocesses of emotion regulation. (a) Feedforward pathway: medial prefrontal cortical system, including the OFC,subgenual ACG, rostral ACG, hippocampus and parahippocampus and MdPFC. (b) Feedback pathway: lateral prefrontalcortical system, including DLPFC and VLPFC. DLPFC, dorsolateral prefrontal cortex; MdPFC, dorsomedial prefrontal cortex;ACG, anterior cingulate gyrus; VLPFC, ventrolateral prefrontal cortex; OFC, orbital frontal cortex; hipp/parahip,hippocampus-parahippocampus region.


846


Such changes could in turn further exacerbate theemotion dysregulation associated with BD, as inrodents the dendritic atrophy arising in the medialPFC during repeated stress was associated withimpaired modulation (that is, extinction) of behavi-oral responses to fear-conditioned stimuli.125

In parallel, there are reports of reduced whitematter density and volume in bilateral regionscorresponding to subcortical white matter tractsconnecting OFC, thalamus and striatal regions im-plicated in emotion processing in adult BD,112,126 andreduced120 white matter volume in left OFC. Otherfindings suggest more widespread127 or no whitematter abnormalities114,128 in adult BD. These datathus appear compatible with evidence from post-mortem studies indicating that BD subjects showabnormally decreased myelin staining in the deepwhite matter of the DLPFC129 and reduced expressionof oligodendroglia-related genes in multiple PFCregions (reviewed in Sokolov130).

Diffusion tensor imaging studies that allow mea-surement of the structural integrity of specific whitematter tracts have reported reduced integrity of whitematter tracts in bilateral OFC in adult BD.131,132 Recentfindings indicate abnormalities specifically in leftOFC and bilateral dorsal PFC. Here, studies havereported a greater number of reconstructed whitematter fibers between left subgenual ACG andamygdala133 and decreased integrity of bilateral whitematter tracts in dorsal prefrontal cortical regions129,134

in adult BD.Together, these findings point to structural abnorm-

alities in BD in gray and white matter and in whitematter tracts in left OFC, implicated in automaticemotion regulation, and in a variety of differentlateral and dorsal prefrontal cortical regions impli-cated in voluntary emotion regulation.

Functional neuroimaging studies. Functional neuro-imaging studies have consistently demon-strated greater subcortical limbic activity (includingamygdala, ventral striatum and hippocampus) toemotional stimuli in adult BD relative to healthyindividuals during mania,135,136 depression137 andeven when euthymic.138140 Findings in BDremission and depression also include reducedblood flow in bilateral MdPFC during sad moodinduction relative to baseline,141 and reducedactivity in bilateral regions of MdPFC, DLPFC andACG during a modified word-based memory taskdesigned to implicitly evoke affective change ineuthymic adults with BD.142 While these studiessuggest functional impairments in prefrontal corticalneural regions implicated in both voluntary andautomatic emotion regulation, they did not focus onexamination of neural activity during paradigmsmeasuring emotion regulation per se.

Another approach for neuroimaging studies of BD,therefore, has been to examine neural activity duringcognitive control paradigms with nonemotionalstimuli. Some studies have employed voluntary

attentional control paradigms, including executivefunction (for example, working memory; wordgeneration), sustained attention (for example, contin-uous performance task) or response inhibition (forexample, Go/NoGo) tasks. The majority of thesestudies show intact task performance but eitherreduced right DLPFC and right dorsal ACG,143,144 orgreater bilateral medial and lateral145 prefrontalcortical activities in remitted BD vs healthy adults.Findings also show reduced DLPFC, and subgenualACG metabolism in BD depression;146 and reducedactivity in right OFC and right VLPFC,147,148 andalso in right hippocampus and left dorsal ACG,147 inBD mania.

Other studies have employed automatic attentionalcontrol paradigms, such as the nonemotional Stroopcolor-word or counting task. Findings from thesestudies show reduced activity in remitted BD relativeto healthy adults during these paradigms in left-sidedregions, including OFC/VLPFC and MdPFC149,150 andright-sided regions, including MdPFC151 and dorsalACG, but also increased activity in right DLPFC.152

Other studies show reduced left OFC activity,149 or nofunctional abnormalities in PFC,153 in BD depression;and reduced left OFC activity during BD mania149

relative to healthy volunteers. The majority of thesestudies show intact, although some show impaired,154

task performance in BD adults.Together, these studies suggest functional abnorm-

alities in adult BD, including both greater andreduced activity relative to healthy individuals, inlateral and dorsal PFC regions implicated in volun-tary emotion regulation during voluntary attentionalcontrol paradigms, and, largely, reduced activity inOFC and MdPFC implicated in automatic, emotionregulation during automatic attentional controlparadigms. These studies did not, however, specifi-cally examine activity in these neural regions duringperformance of voluntary and automatic emotionregulation paradigms. We next examine findings fromthe few studies that have employed such paradigms tohelp determine the extent to which emotion dysre-gulation in BD may be associated with functionalabnormalities in neural regions underlying voluntary,vs neural regions underlying automatic, emotionregulation.

Voluntary emotion regulation paradigms. One recentstudy employed an affective Go/NoGo paradigm toexamine neural activity during voluntary attentionalcontrol of emotion in BD remission.140 Here, despiteintact task performance, euthymic BD adultsdemonstrated abnormally increased activity inbilateral OFC and left dorsal ACG, in addition toincreased activity in different subcortical limbicregions implicated in emotion processing, wheninhibiting responses to emotional relative to neutraldistractors. Another study employing a similarparadigm showed greater activity to sad relative toneutral distracters in right DLPFC and VLPFC, andgreater activity to happy distracters in bilateral


847


DLPFC, MdPFC and left subgenual ACG in manic BDvs healthy adults.155 One study156 used functionalconnectivity analysis to examine neural activity inmanic BD adults using emotion labeling of facialexpressions, a task previously associated withvoluntary control of responses to emotionalstimulithinking rather than feeling aboutemotionand with greater VLPFC and reducedsubcortical limbic activity than gender labeling offacial expression.17 Findings indicated that manic BDadults had significantly reduced right VLPFCregulation of the left amygdala response during thetask that parallel previous reports of decreasedVLPFC activity in BD adults relative to healthyadults during the above-described voluntaryattentional control paradigms.147,149,150,155

Automatic emotion regulation paradigms. Twostudies employing an emotional Stroop paradigmdemonstrated decreased left OFC activity ineuthymic BD adults,157,158 which was associatedwith slower task performance.158 Another studydemonstrated decreased left MdPFC (BA10), andright VLPFC activity, although no performancedeficit, in manic BD adults during risky decision-makinga gambling paradigm dependent uponautomatic cognitive change as a strategy tomaximize reward and minimize loss.159

Summary. There are inconsistent findings regardingfunction in lateral and dorsal prefrontal corticalregions implicated in voluntary emotion regulationin adult BD. For example, findings from studiesemploying voluntary attentional control paradigmsper se demonstrate patterns of reduced but alsoincreased activity in different bilateral lateral anddorsal prefrontal cortical regions in BD vs healthyadults. Findings from studies employing voluntaryemotion regulation paradigms, however, indicategreater activity in BD than healthy adults bilaterallyin these lateral and dorsal prefrontal cortical regions,together with greater activity in bilateral ventromedialprefrontal cortical regions implicated in automaticemotion regulation, which may mediate the voluntaryemotion regulatory roles of the previous lateral anddorsal prefrontal cortical regions during voluntaryemotion regulation. Structural findings indicatereductions in GM volume and density in lateral anddorsal prefrontal cortical regions implicated involuntary emotion regulation, although there areinconsistent findings of increased GM volume or noabnormalities in these regions, in adult BD. The roleof lateral prefrontal cortical regions in voluntaryemotion regulation in adult BD therefore remainsunclear. It is possible that increased activity in lateralprefrontal cortical regions during voluntaryattentional control and voluntary emotion regulationparadigms may reflect inefficient utilization of theseregions during cognitive control tasks in BD relativeto healthy adults that has been previously proposedin adults with schizophrenia.160

There are more consistent findings regardingfunctional and structural changes in neuralregions implicated in automatic emotion regula-tion in adult BD. Studies employing automaticattentional control paradigms show reducedactivity predominantly in left-sided ventromedialPFC in BD relative to healthy adults. Studiesemploying automatic emotion regulation paradigmsalso show reduced activity predominantly withinleft-sided ventromedial prefrontal cortical regionsimplicated in automatic emotion regulation,both during remission and mania, in BD relativeto healthy adults. These functional neuroimagingfindings are paralleled by structural neuroimagingfindings showing GM structural changes in leftOFC, and abnormal integrity and number ofwhite matter fibers connecting left OFC andsubcortical limbic regions implicated in emotionprocessing, in adult BD. Interestingly, increasedrecruitment of DLPFC in BD vs healthy adultshas been reported during an automatic attentionalcontrol paradigm. This suggests the employmentof more effortful, voluntary, rather than automatic,regulatory control systems during performanceof otherwise automatic control tasks in adultBD. Further studies are required to examinefunctional connectivity between ventromedial andmedial prefrontal cortical systems implicated inautomatic, vs lateral and dorsal prefrontal corticalsystems implicated in voluntary, emotion regulationin adult BD.

The combination of these functional and structuralneural abnormalities, which appear to be decreases inactivity and GM volume or density reductionspredominantly within left-sided ventromedial pre-frontal cortical regions implicated in automatic emo-tion regulation, may underlie the mood instability ofadult BD (Figure 2). The significance of the lateralityof these findings remains unclear, but may suggesta role for the left hemisphere, previously linkedwith perception of positive emotion163 in thepathophysiology of BD. These findings indicate thatfurther insights into the neural basis of mooddysregulation in BD will come especially from studiesthat employ paradigms that measure functionalintegrity of neural regions during performance ofautomatic emotion regulation paradigms. A key focusof these future studies will be to identify persistentand mood-state-dependent functional and, poten-tially, structural abnormalities in these neural regionsin adult BD to dissociate neural abnormalitiesreflecting pathophysiological mechanisms of BD vsneural abnormalities that may be secondary todepression or mania. Moreover, future studies shouldfocus on examination of the relationship betweenfunctional and structural abnormalities in theseneural regions in adult BD, and the extent to whichspecific symptom domains, including comorbidanxiety and substance abuse disorders, are associatedwith distinguishable patterns of abnormality in theseneural regions.


848


Normal development of neural systems of emotionregulationTo better understand how abnormalities in the neuralsystems of emotion regulation may contribute to theneurodevelopment of BD, we first briefly reviewfindings from studies focusing on the normal devel-opment of neural regions implicated in emotionregulation.

During childhood and adolescence, there areimportant maturational changes that occur in brainstructure. Cross-sectional164166 and longitudinal167

studies have shown increases in white mattervolumes, which are thought to reflect increasingmyelination in childhood and adolescence. Maturational

increases in white matter are assumed to be presentglobally, with specific increases shown in frontal,parietal and occipital cortices.166 The overall growthpattern of GM in the telencephalon is more hetero-geneous.168 The general pattern reflects decreases inGM from early childhood to post-adolescence but thiseffect is nonlinear, likely reflecting selective pruningof neuronal connections.164,165,169 Regional analyseshave shown that GM volumes peak at about 12 yearsof age in frontal and parietal cortical regions followedby a decline in post-adolescence thus yielding a netdecrease.167

Furthermore, there is evidence of developmentalchanges in GM volume in subcortical structures

Figure 2 Neural model of emotion regulation illustrating possible functional and structural abnormalities in neural systemsimplicated in voluntary and automatic subprocesses of emotion regulation in adult bipolar disorder. These abnormalities,which appear to be predominantly within the left-sided ventromedial prefrontal cortical regions implicated in automaticemotion regulation, may underlie the mood instability of adult bipolar disorder (BD). For example, functional neuroimagingstudies have demonstrated greater subcortical limbic activity (including amygdala, ventral striatum and hippocampus)to emotional stimuli in adult BD relative to healthy individuals during mania,136,147 depression137 and when euthy-mic.139,140,161,162 Studies employing automatic attentional control paradigms show reduced activity predominantly in left-sided ventromedial PFC in BD relative to healthy adults.149,150 Studies employing automatic emotion regulation paradigmsalso show reduced activity predominantly within left-sided ventromedial prefrontal cortical regions implicated in automaticemotion regulation, both during remission and mania in BD relative to healthy adults.157,159 Structural neuroimaging findingsshow gray matter structural changes in left OFC, and abnormal integrity and number of white matter fibers connecting leftOFC and subcortical limbic regions implicated in emotion processing, in adult BD.112,120,126,131133 There are moreinconsistent findings regarding the roles of lateral and dorsal prefrontal cortical regions implicated in voluntary emotionregulation in adult BD. For example, findings from studies employing voluntary attentional control paradigms per sedemonstrate patterns of reduced,143,144 although also increased,145,152 activity in different bilateral lateral and dorsalprefrontal cortical regions in BD vs healthy adults, whereas findings from studies employing voluntary emotion regulationparadigms indicate greater activity in BD than healthy adults bilaterally in these lateral and dorsal prefrontal cortical regions,together with greater activity in bilateral ventromedial prefrontal cortical regions implicated in automatic emotionregulation.140,155 The latter may mediate the voluntary emotion regulatory roles of the previous lateral and dorsal prefrontalcortical regions during voluntary emotion regulation.14 DLPFC, dorsolateral prefrontal cortex; MdPFC, dorsomedialprefrontal cortex; ACG, anterior cingulate gyrus; VLPFC, ventrolateral prefrontal cortex; OFC, orbital frontal cortex;hipp/parahip, hippocampus-parahippocampus region.


849


implicated in processing emotional information. Forexample, in a study with 4- to 18-year olds, Gieddet al.170 showed that amygdala volume significantlyincreased with age, but only in men, whereashippocampal volume significantly increased withage, but only in women.

Recent research has also revealed important ma-turational changes in brain function throughoutchildhood and adolescence, particularly in regionsinvolved in the voluntary regulation of emotion. Forexample, several studies have reported greater activ-ity with age in brain regions known to supportcognitive control processes.171,172 Some studies, how-ever, demonstrated changes that were specific toadolescence. One study examined neural activityduring voluntary response suppression in 8- to 30-year olds using an anti-saccade task.173 In this study,adolescents had difficulty achieving adult levels ofperformance and compensated by engaging morefrontostriatal circuitry compared with adults. Thesefindings suggest that developmental changes incognitive control processes in adolescence are asso-ciated with a more efficient and functionally inte-grated use of neural circuitry supporting theseprocesses.174 These studies, however, involved none-motional stimuli.

Developmental changes in processing and regulat-ing emotion may be associated with a refinement orincreased efficiency in prefrontalsubcortical connec-tions from childhood through adolescence.175 Forinstance, recent research suggests that explicit pro-cessing of emotional faces continues to develop fromearly childhood through adolescence.176178 Neuralsystems implicated in voluntary emotion regulationmay not be fully mature until adulthood. Forinstance, in a study examining neural activity duringvoluntary reappraisal in 8- and 10-year-old girls,Levesque et al.179 found that in contrast to adultwomen young girls recruited a greater number ofprefrontal regions, possibly reflecting immaturity ofprefrontallimbic connections in childhood. Duringpassive viewing of fearful faces, Monk et al.180

demonstrated that adolescents showed greater activ-ity in right amygdala, bilateral OFC and ACG relativeto adults, suggesting that when attention is uncon-strained, adolescents may be more sensitive toemotionally salient stimuli.

Although research in the normal development ofneural systems of emotion regulation (voluntary andautomatic) remains in its infancy, evidence is emer-ging suggesting that adolescence may be a key periodfor the development of these neural systems.181186 Forexample, there is evidence indicating importantdevelopmental changes during adolescence in neuralsystems involved in reward processing,187 responseinhibition in the context of emotional stimuli,188

response conflict monitoring189191 and risk tak-ing.192,193 The impact of puberty upon the develop-ment of neural systems of emotion regulation remainsto be fully examined.181,182,194 For example, recentstudies indicate puberty-specific changes in brain

structure195 and function,181,196 particularly in neuralregions implicated in emotion processing and regula-tion (for example, amygdala, hippocampus).

Pediatric bipolar disorder and high-risk populationsWe next examine findings from the few structural andfunctional neuroimaging studies that have examinedkey neural regions underlying voluntary and auto-matic emotion regulation in pediatric BD and in-dividuals at high risk for developing BD.

Structural neuroimaging studies. These haverevealed structural alterations in both cortical197 andsubcortical161,198200 regions. Studies in adolescentand first-episode BD198,199 as well as one longi-tudinal study of adolescents161 demonstratesmaller amygdalae and smaller200,201 or normal-sizedhippocampi198,199 in adolescent BD, which differfrom the above findings of enlarged amygdalae inadults.

Although earlier studies failed to detect anydifferences in PFC volumes in pediatric BD, a recentstudy using a more sensitive voxel-based morpho-metric automated technique showed abnormallydecreased GM volume in left DLPFC. With lessstringent statistical thresholds, results also revealeddecreased left amygdala and left nucleus accumbensvolumes in BD children and adolescents.197 Otherrecent findings indicate decreased bilateral OFC inBD adolescents.202 Given the discrepant findingsbetween BD children and adolescents and BD adultsin key areas such as the amygdala, further study mustbe done to determine possible differences in PFCvolumes of those with early vs late onset and/ordiagnosis, to characterize the effects of chronic moodstabilizer treatment on regional brain structure, and totrack change over development within the context ofpossible compensatory mechanisms (including earlytreatment).

There are very few structural studies of individualsat risk for BD. A review of structural neuroimagingstudies in individuals at risk for BD reportedstructural abnormalities in amygdala, striatum,hippocampus and subgenual ACG that were proposedas potential candidate neuroanatomical risk markersfor BD (see Hajek et al.203). A recent study, however,has shown, in children who became diagnosed withBD, increased GM over the left temporal cortex anddecreased GM bilaterally in the ACG (including thesubgenual ACG), which was observed most strikinglyafter the illness onset.204 This neurodevelopmentaltrajectory was observed also in children who had adiagnosis of multidimensional impairment (atypicalpsychosis), who did not convert to a diagnosis of BD.The authors suggested that this pattern of corticaldevelopment may reflect affective dysregulation(lability) in general. More recently, we found abnor-mally increased left hippocampal/parahippocampalGM volume in healthy offspring of parents diagnosedwith BD.205 Moreover, this increase in GM volume inthe BD offspring group was significantly positively


850


correlated with pubertal stage (but not age). Giventhat these offspring of BD parents were healthy at thetime of the study, it is possible that these differencesin GM volume may reflect protective markers for BD.Longitudinal follow-up studies are required to relatethese structural findings in BD at-risk individuals athigh familial risk for BD to future clinical outcome.

Functional neuroimaging studies. BD children andadolescents demonstrate deficits in executivefunction206 and emotion processing,207 and impairedsocial cognition and response flexibility.208 As withadult BD, it has been proposed, therefore, thatpediatric BD may be associated with functionalabnormalities in neural systems supporting emotionregulation processes.209,210 There are several findingsthat support this. First, there is evidence indicatingthat children with BD exhibit deficits in response-reversal learning, suggesting altered functioning ofneural systems implicated in automatic cognitivechange subprocesses.206,211 Euthymic BD adolescentsalso show greater activity within subcortical limbicregions associated with emotion processing ratherthan prefrontal cortical regions such as OFC, rostraland dorsal ACG and MdPFC implicated in automaticattentional control during nonemotional Stroop taskperformance.212 In these BD adolescents, there wasalso no age-related increase in bilateral OFC activitydemonstrated by age-matched healthy adolescentsduring the task.212 Abnormal increases in activity inbilateral DLPFC as well as subcortical limbic regionsduring voluntary attentional control (a workingmemory task) have also been shown in children andadolescents with BD.213

Regarding emotion-processing tasks, when atten-tion is directed toward the emotional aspect of astimulus, BD children and adolescents show abnor-mally increased left amygdala activity to neutralfaces.207 Furthermore, functional connectivity ana-lyses have recently revealed that compared with age-matched controls BD children and adolescents showless functional connectivity between left amygdalaand a network of neural regions implicated inprocessing faces and emotional stimuli, includingright posterior cingulate/precuneus and right fusiformgyrus/parahippocampal gyrus.214 These findings sug-gest that functional abnormalities in neural systemssupporting emotion processing may underlie thepathophysiology of pediatric BD.207,209 Further stu-dies are required, specifically studies employingvoluntary and automatic emotion regulation para-digms, to elucidate the functional abnormalities inneural systems underlying emotion regulation inpediatric and adolescent BD.

There is a paucity of functional neuroimagingstudies in individuals at high risk of BD. One recentneuroimaging study using PET contrasted neuralactivity during sad mood induction in adult BD andtheir healthy siblings.141 Here, healthy siblingsrelative to their affected probands showed greateractivity in left MdPFC, a region activated in healthy

individuals during paradigms involving automaticcognitive change. These differences in neural activitywere interpreted as a potential resiliency marker forBD, given that the adult siblings were free ofpsychopathology. To our knowledge, there are nocurrent published data examining the functioning ofneural systems involved in emotion regulation inoffspring of individuals with BD.

Summary. Findings from existing studies suggeststructural abnormalities in dorsal and ventralprefrontal cortices and functional abnormalitieswithin prefrontal cortical regions implicated involuntary and automatic emotion regulatorysubprocesses in pediatric BD. Clearly, there is aneed for further studies to examine the extent towhich both children and adolescents with, and thoseat risk for, BD have abnormal development of differentprefrontal cortical regions implicated in automaticand voluntary emotion regulation using experimentalparadigms that focus on these different emotionregulatory subprocesses.

Conclusion: understanding the pathophysiologyand neurodevelopment of bipolar disorder

We have described a new neural model of emotionregulation that includes voluntary and automaticregulatory subprocesses, centered in different regionsof PFC, hippocampus and parahippocampus. Wehave then used this model as a theoretical frameworkto examine functional neural abnormalities in theseneural systems that may predispose to the develop-ment of BD, a major psychiatric disorder character-ized by severe emotion dysregulation. The mostconsistent findings in adult BD are those of studiesemploying automatic attentional control and auto-matic emotion regulation paradigms. These findingsindicate abnormally reduced activity in, predomi-nantly, left-sided OFC and MdPFC during automaticattentional control and automatic emotion regulationparadigms in adult BD. Data further suggest structuralabnormalities in dorsal and ventral prefrontal cor-tices, including left-sided abnormalities in the integ-rity of white matter tracts in OFC, in adult BD.Together, these findings point to left-sided abnormal-ities in prefrontal cortical regions implicated inautomatic regulation in adult BD but further study isrequired to elucidate the nature of these abnormalitiesin emotion regulatory n

#1. a neural model of voluntary and automatic emotion regulation implications for understanding the...

Documents