the theory of constructed emotion: an active inference ... · pdf filethe theory of...

The theory of constructed emotion an active inference

account of interoception and categorizationLisa Feldman Barrett123

1Department of Psychology Northeastern University Boston MA USA 2Athinoula A Martinos Center forBiomedical Imaging and 3Psychiatric Neuroimaging Division Department of Psychiatry MassachusettsGeneral Hospital and Harvard Medical School Charlestown MA USA

Correspondence should be addressed to Lisa Feldman Barrett Department of Psychology Northeastern University Boston MA USA E-mail lbarrettneuedu

Abstract

The science of emotion has been using folk psychology categories derived from philosophy to search for the brain basis ofemotion The last two decades of neuroscience research have brought us to the brink of a paradigm shift in understandingthe workings of the brain however setting the stage to revolutionize our understanding of what emotions are and howthey work In this article we begin with the structure and function of the brain and from there deduce what the biologicalbasis of emotions might be The answer is a brain-based computational account called the theory of constructed emotion

Key words emotion predictive coding construction interoception categorization concepts affect

Ancient philosophers and physicians believed a human mind tobe a collection of mental faculties They divided the mind notwith an understanding of biology or the brain but to capturethe essence of human nature according to their concerns abouttruth beauty and ethics The faculties in question havemorphed over the millennia but generally speaking theyencompass mental categories for thinking (cognitions) feeling(emotions) and volition (actions and in more modern versionsperceptions) These mental categories symbolize a cherishednarrative about human nature in Western civilization thatemotions (our inner beast) and cognitions (evolutionrsquos crowningachievement) battle or cooperate to control behavior1 The clas-sical view of emotion (Figure 1) was forged in these ancientideas Affective neuroscience takes its inspiration from thisfaculty-based approach Scientists begin with emotion conceptsthat are most recognizably English (Pavlenko 2014 Wierzbicka2014) such as anger sadness fear and disgust and search fortheir elusive biological essences (ie their neural signatures or

fingerprints) usually in subcortical regions This inductive ap-proach assumes that the emotion categories we experience andperceive as distinct must also be distinct in nature

If the history of science has taught us anything however itis that human experiences rarely reveal the way that the nat-ural world works lsquoPhysical concepts are free creations of thehuman mind and are not however it may seem uniquelydetermined by the external worldrsquo (Einstein et al 1938 p 33)The last two decades of neuroscience research have brought usto the brink of a paradigm shift in understanding the workingsof the brain setting the stage to revolutionize our study of emo-tions (or any mental category) So in this article we turn thetypical inductive approach on its head We begin not with men-tal categories but with the structure and function of the brainand from there deduce what the biological basis of emotionsmight be The answer I suggest will look something like thetheory of constructed emotion (Barrett 2017) formerly the con-ceptual act theory of emotion (Barrett 2006b 2011a 2012 20132014)

To begin this discussion I first outline enough backgroundon brain structure and function to start asking informed ques-tions about the biological basis of emotion I then introduce thetheory of constructed emotion contrast it briefly with the clas-sical view when instructive to do so and consider selected

Received 13 August 2016 Revised 13 August 2016 Accepted 11 October 2016

VC The Author (2016) Published by Oxford University Press For Permissions please email journalspermissionsoupcom

1 Throughout the millennia with few exceptions cognitions werethought to reside in the brain emotions in the body and then lateremotions were relocated to the parts of the brain that control the bodyFor example Aristotle placed both thinking and feeling in organs ofthe body Descartes kept emotions in the body and placed cognition inthe pineal gland of the brain)

1

Social Cognitive and Affective Neuroscience 2017 1ndash23

doi 101093scannsw154Advance Access Publication Date 19 October 2016Original article

findings on the brain basis of emotion through the theoryrsquoslens In the process I offer a series of novel hypotheses aboutwhat emotions are and how they work

The biological backgroundWhat is a brain

A brain is a network of billions of communicating neurons bathedin chemicals called neurotransmitters which permit neurons to

pass information to one another (Doya 2008 Bargmann 2012) Thefiring of a single neuron (or a small population of neurons) repre-sents the presence or absence of some feature at a moment intime (Deneve 2008 Deneve and Jardri 2016) However a given neu-ron (or group of neurons) represents different features from mo-ment to moment (eg Stokes et al 2013 Spillmann et al 2015)because many neurons synapse onto one (many-to-one connectiv-ity) and a neuronrsquos receptive field depends on the information itreceives (ie depends on its neural context in the momentMcIntosh 2004) Conversely one neuron also synapses on many

Fig 1 The classical view of emotion The classical view of emotion includes basic emotion theories (eg for a review see Tracy and Randles 2011) causal appraisal theo-

ries (eg Scherer 2009 Roseman 2011) and theories of emotion that rely on black-box functionalism (Davis 1992 Anderson and Adolphs 2014) Each emotion faculty is

assumed to have its own innate lsquoessencersquo that distinguishes it from all other emotions This might be a Lockean essence (an underlying causal mechanism that all in-

stances of an emotion category share making them that kind of emotion and not some other kind of emotion depicted by the circles in the figure) Lockean essences

might be a biological such as a set of dedicated neurons or psychological such as a set of evaluative mechanisms called lsquoappraisalsrsquo An emotion category is usually

assumed to have a Platonic essence [a physical fingerprint that instances of that emotion share but that other emotions do not such a set of facial movements (an lsquoex-

pressionrsquo) a pattern of autonomic nervous system activity andor a pattern of appraisals] Of course no one is expecting complete invariance but it is assumed that in-

stances of a category are similar enough to be easily diagnosed as the same emotion using objective (perceiver-independent) measures alone (A) is adapted from Davis

(1992) (B) is adapted Anderson and Adolphs (2014) (C) is adapted from Barrett (2006a) which reviews the growing evidence that contracts the classical view of emotion

2 | Social Cognitive and Affective Neuroscience 2017 Vol 0 No 0

other neurons [one-to-many connectivity (Sporns 2011 Sterlingand Laughlin 2015)] to help implement instances of different psy-chological categories As a consequence neurons are multipurpose[for evidence and discussion see (Barrett and Satpute 2013Anderson 2014 Anderson and Finlay 2014)] even in subcortical re-gions like the amygdala (Cerf personal communication 30 July2015)2

When the brain is viewed as a massive network rather thana single organ or a collection of lsquomental modulesrsquo it becomesapparent that this one anatomic structure of neurons can createan astounding number of spatiotemporal patterns making thebrain a network of high complexity (Sporns 2011 Bullmore andSporns 2012 Rigotti et al 2013) Natural selection prefers highcomplexity systems as they can reconfigure themselves into amultitude of different states (Whitacre 2010 Whitacre andBender 2010 Sterling and Laughlin 2015)

The brain achieves complexity through lsquodegeneracyrsquo(Edelman and Gally 2001) the capacity for dissimilar represen-tations (eg different sets of neurons) to give rise to instances ofthe same category (eg anger) in different contexts (ie many-to-one mappings of structure to function) Degeneracy is ubiqui-tous in biology from the workings inside a single cell to distrib-uted brain networks (eg see Tononi et al 1999 Edelman andGally 2001 Marder and Taylor 2011) Natural selection favorssystems with degeneracy because they are high in complexityand robust to damage (Whitacre and Bender 2010) Degeneracyexplains why Roger the patient who lost his limbic circuitry toherpes simplex type I encephalitis still experiences emotions(Feinstein et al 2010) and why monozygotic twins with fully cal-cified basolateral sectors of the amygdala [due to Urbach-Wiethe disease (UWD)] have markedly different emotional cap-acity despite genetic and environmental similarity (Becker etal 2012 Mihov et al 2013) Degeneracy also explains how acharacteristic can be highly heritable even without a single setof necessary and sufficient genes (eg Turkheimer et al 2014)

In emotion research degeneracy means that instances of anemotion (eg fear) are created by multiple spatiotemporal pat-terns in varying populations of neurons Therefore it is unlikelythat all instances of an emotion category share a set of core fea-tures (ie a single facial expression autonomic pattern or set ofneurons see Clark-Polner et al 2016) This observation is anexample of population thinking pioneered in Darwinrsquos On theOrigin of Species (Mayr 2004)3 By observing the natural world

Darwin realized that biological categories such as a species areconceptual categories (highly variable instances groupedtogether by lsquoa goalrsquo rather than by similar features or a singleshared underlying cause (Mayr 2004) My hypothesis followingDarwinrsquos insight is that fear (or any other emotion) is alsquocategoryrsquo that is populated with highly variable instances(Clark-Polner et al 2016 Clark-Polner Johnson amp Barrett 2016eg Wilson-Mendenhall et al 2011 2015) The summaryrepresentation of any emotion category is an abstraction thatneed not exist in nature (as is true for any biological categoryfor a discussion of population thinking see Mayr 2004 asapplied to emotion concepts and categories see Barrett 2017and as applied to concepts and categories more generally seeBarsalou 1983 Voorspoels et al 2011) The fact that humanbrains effortlessly and automatically construct suchrepresentations (eg Murphy 2002 Posner and Keele 1968)helps to explain why scientists continue to believe in theclassical view and even propose it as an innovation (egAnderson and Adolphs 2014) even as evidence continues tocall it into doubt (eg Barrett 2006a 2016b 2012 Barrett et al2007a see Table 1 for specific neuroscience examples with aparticular focus on fear as the category that has garnered themost support for the classical view)

What is a brain for

A brain did not evolve for rationality happiness or accurate per-ception All brains accomplish the same core task (Sterling andLaughlin 2015) to efficiently ensure resources for physiologicalsystems within an animalrsquos body (ie its internal milieu) so thatan animal can grow survive and reproduce This balancing actis called lsquoallostasisrsquo (Sterling 2012) Growth survival and repro-duction (and therefore gene transmission) require a continualintake of metabolic and other biological resources Metabolicand other expenditures are required to plan and execute thephysical movements necessary to acquire those resources inthe first place (and to protect against threats and dangers)Allostasis is not a condition of the body but a process for howthe brain regulates the body according to costs and benefits lsquoef-ficiencyrsquo requires the ability to anticipate the bodyrsquos needs andsatisfy them before they arise (Sterling 2012 Sterling andLaughlin 2015)4 An animal thrives when it has sufficient re-sources to explore the world and to consolidate the details of

2 Cells in the medial temporal lobe (including the amygdala) appear toact as a memory cache for important things (eg photos of friendsfamily famous people the patients themselves landscapes direc-tions some cells donrsquot respond to anything for a few days and thenbegin to respond when the experimenters walk into the room) atsome other point the cells might adopt and code for something en-tirely different that becomes important (Cerf personal communica-tion 30 July 2015) Even primary sensory neurons are not coding forsingle sensory features but for associations between one feature (likethe presence or absence of a line) with other sensory features eg V1neurons have receptive fields that include auditory and sensorimotorchanges (eg Liang et al 2013)

3 Before On the Origin of Species a lsquospeciesrsquo was defined as biological type(ie with a set of unchanging physical characteristics or features thatare passed down through the generations) This typological character-ization fundamentally underestimates within-category variation (in itsphenotypic features and in itrsquos gene pool) and over-estimates betweencategory variation (and borderline cases are often encountered Mayr2004 Gelman and Rhodes 2012) One of Darwinrsquos greatest conceptualinnovations in Origin was to revolutionize the concept of a species as abiopopulation of highly variable individuals (instead of a group ofhighly similar creatures who share a set of co-occurring biological

features) (Mayr 2004) Since then the concept of a lsquospeciesrsquo has beencharacterized on the basis of what category members do (ie function-ally) not on the basis of a shared gene pool or a set of physical fea-tures A species is a reproductive community (sometimes members ofdifferent species are reproductively incompatible sometimes theydonrsquot such as when they are geographically isolated) Fundamentallythis translates into the insight that a biological category (a lsquospeciesrsquo) isa conceptual category rather than a typological one a species is apopulation of physically unique individuals who similarities aredefined functionally not physically

4 Sometimes allostasis involves dynamically regulating resource alloca-tion (ie diverting glucose electrolytes water etc from one system toanother) to meet the bodyrsquos spending needs eg in advance of stand-ing up the heart beats stronger and faster blood vessels constrict andblood pressure raises to ensure that the brain continues to receive theblood (and oxygen) Sometimes allostasis involves signaling the needfor resources before the body runs out ( eg drinking before dehydra-tion occurs) or preparing for the intake of resources in advance of theiringestion eg saliva example in humans and some other mammalssaliva is made of alpha-amylase which is an enzyme that breaks downglucose When the body is in need of glucose saliva is pre-emptivelysecreted (even before anything is ingested) Even just imaging foodcauses glucose secretion

L F Barrett | 3

Table 1 Examples of neuroscience evidence that disconfirm the classical view of emotion

Observation Method Example Citations

Different emotion categories cannot be specifically and consistently localizedto distinct populations of neurons within a single region of the humanbrain

Human neuroimagingtask-related data

(Vytal and Hamann 2010Lindquist et al 2012)

Different emotion categories cannot be consistently localized to specific in-trinsic networks in the human brain

Human neuroimaging in-trinsic connectivity data

(Barrett and Satpute 2013Touroutoglou et al 2015)

The instances of an emotion category need not share a population of neuronsthat are necessary or sufficient to implement them

Human neuroimagingmulti-voxel patternanalysis

(Clark-Polner Johnson ampBarrett 2016)

Individual neurons when stimulated in studies of experience expressionand perception do not have emotion-specific receptive fields

Intracranial stimulation inhumans

(Guillory and Bujarski 2014)

Lesions to the amygdala produce variable functional consequencesMonozygotic twins whose basolateral nuclei of both amygdalae are calci-fied due to UWD do not show equivalent deficits in experiencing and per-ceiving fear patient BG has deficits similar to patient SM (who hascomplete loss of both amygdalae due to UWD) whereas her sister AM isable to experience and perceive fear when BG cannot Other people withbasolateral lesions from UWD show different problems in fear perception(they are vigilant to rather than neglectful of posed fear faces) Patient SMcan experience intense fear in the real world under certain circumstancesand her impairments in fear perception appear to be limited to experi-ments where she is asked to view stereotyped fear poses and explicitlycategorize them as fearful There is ample evidence that she is able to per-ceive fear in various circumstances in real life (see Box 1)

Behavioral observations inhumans with amygdalalesions

(Bechara et al 1995 1999Adolphs et al 1999Adolphs and Tranel1999 2003 Atkinsonet al 2007 de Gelderet al 2014 Hamptonet al 2007 Tsuchiyaet al 2009 Hurlemannet al 2007 De Martinoet al 2010 Boes et al2011 Feinstein et al2011 2013 2016 Beckeret al 2012 Terburg et al2012 Feinstein 2013Mihov et al 2013)

Various circuits acting in parallel or collaboratively support learning when toperform behaviors that are typically referred to as lsquofear behaviorsrsquo Somehave argued that these circuits represent distinct pathways from theamygdala to the periaqueductal gray through the hypothalamus to controldifferent situation-specific fear behaviors but others find that there aremany (circuits) to one (behavior) mappings Others find one (circuit) tomany (behavior) mappings Still others find that the amygdala or specificparts (eg the basolateral nuclei) are not necessary for the expression ofpreviously learned aversive responses (ie the way that learning is ex-pressed depends on the context and the available options for behavior)Also cortical regions (eg dmPFC and vmPFC) appear necessary for aver-sive learning when the context is more ecologically valid and less artifi-cially simple One thing is certain scientists routinely engage in mentalinference and refer to circuits as controlling different types of fear when infact they are studying context-dependent behaviors that may not bear aone-to-one correspondence to fear

Optogenetic research andsome lesion research inrodents

(Furlong et al 2010 Grossand Canteras 2012 Herryand Johansen 2014Sharpe and Killcross2015 Tovote et al 2015McGaugh 2016)

The expression of aversive learning depends on the state of the animal Forexample when the conditioned stimulus (a tone) is presented alone afterhaving been paired with the unconditioned stimulus (an electric shock)the animal typically freezes its heart rate increases and its skin conduct-ance goes up which is usually taken as evidence that the animal haslearned fear Yet when an animal is restrained in position as it hears thetone its heart rate decreases

Classical conditioning inrats

(Iwata and LeDoux 1988)

Adult monkeys with amygdala lesions are more likely to explore novel ob-jects right away (which is usually interpreted as a lsquolack of fearrsquo) but an al-ternative explanation is that the amygdala helps to regulate exploratorybehavior in novel situations which in turn will increase sensory process-ing when there is substantial prediction error lsquoFearrsquo is not necessary Anamygdala might be required for aversive learning but not the behavioralresponse learned (paralleling observations in rodent experiments)

Lesions in non-humanprimates

(Mason et al 2006Antoniadis et al 2009)


experience within the brainrsquos synaptic connections makingthose experiences available to guide later decisions about futureexpenditures and deposits Too much of a resource (eg obesityin mammals) or not enough (eg fatigue Dantzer et al 2014) issuboptimal Prolonged imbalances can lead to illness (egHunter and McEwen 2013 McEwen et al 2015) that remodelsthe brain (Crossley et al 2014 Goodkind et al 2015) and thesympathetic nervous system with corresponding behaviorchanges (Sloan et al 2007 Capitanio and Cole 2015 for a re-view see Sloan et al 2008)

Whatever else your brain is doingmdashthinking feeling per-ceiving emotingmdashit is also regulating your autonomic nervoussystem your immune system and your endocrine system as re-sources are spent in seeking and securing more resources Allanimal brains operate in the same manner (ie even insectbrains coordinate visceral immune and motor changes Sterlingand Laughlin 2015 p 91) This regulation helps explain why inmammals the regions that are responsible for implementingallostasis (the amygdala ventral striatum insula orbitofrontalcortex anterior cingulate cortex medial prefrontal cortex(mPFC) collectively called lsquovisceromotor regionsrsquo) are usuallyassumed to contain the circuits for emotion In fact many ofthese visceromotor regions are some of the most highly

connected regions in the brain and they exchange informationwith midbrain brainstem and spinal cord nuclei that coordin-ate autonomic immune and endocrine systems with one an-other as well as with the systems that control skeletomotormovements and that process sensory inputs Therefore theseregions are clearly multipurpose when it comes to constructingthe mental events that we group into mental categories(see Figure 2)

How does a brain perform allostasis

For a brain to effectively regulate its body in the world it runsan internal model of that body in the world5 In psychology werefer to this modeling as lsquoembodied simulationrsquo (Barsalou 2008Barsalou et al 2003) (eg see Figure 3) An internal model ismetabolic investment implemented by intrinsic activity (eg

Fig 2 Hubs in the human brain (A) Hubs of the rich club adapted from van den Heuvel and Sporns (2013) These regions are strongly interconnected with one another and it is

proposed that they integrate information across the brain to create large-scale patterns of information flow (ie synchronized activity van den Heuvel and Sporns 2013) They are

sometimes referred to as convergence or confluence zones (eg Damasio 1989 Meyer and Damasio 2009) (B) Results of a forward inference analysis revealing lsquohot spotsrsquo in the

brain that show a better than chance increase in BOLD signal across 5633 studies from the Neurosynth database Activations are thresholded at FWE P lt 005 Limbic regions (ie

agranulardysgranular with descending projections to visceromotor control nuclei) include the cingulate cortex [midcingulate cortex (MCC) pregenual anterior cingulate cortex

(pgACC)] ventromedial prefrontal cortex (vmPFC) supplementary motor and premotor areas (SMA and PMC) medial temporal lobe the anterior insula (aINS) and ventrolateral pre-

frontal cortex (vlPFC) (eg Carrive and Morgan 2012 Bar et al 2016) for a discussion and additional references see (Kleckner et al in press) AG angular gyrus MC motor cortex

5 There is a well-known principle of cybernetics anything that regulates(ie acts on) a system must contain an lsquointernal modelrsquo of that system(Conant and Ross Ashby 1970) From a brainrsquos perspective the lsquosystemrsquoin question includes itrsquos body and itrsquos ecological niche a body must bewatered fed and cared for so that a creature can grow thrive and ul-timately reproduce and care for itrsquos young so as to pass its genes tothe subsequent generation

L F Barrett | 5

Berkes et al 2011) that in humans occupies 20 of our total en-ergy consumed (Raichle 2010)6 Given these considerationsmodeling the world lsquoaccuratelyrsquo in some detached disembodiedmanner would be metabolically reckless Instead the brainmodels the world from the perspective of its bodyrsquos physio-logical needs7 As a consequence a brainrsquos internal model in-cludes not only the relevant statistical regularities in theextrapersonal world but also the statistical regularities of theinternal milieu Collectively the representation and utilizationof these internal sensations is called lsquointeroceptionrsquo (Craig2015) Recent research suggests that interoception is at the coreof the brainrsquos internal model and arises from the process of allo-stasis (Barrett and Simmons 2015 Chanes and Barrett 2016Barrett 2017 for related discussions see Seth et al 2012 Seth2013 Pezzulo et al 2015 Seth amp Friston 2016) Interoceptivesensations are usually experienced as lower dimensional feel-ings of affect (Barrett and Bliss-Moreau 2009 Barrett 2017) Assuch the properties of affectmdashvalence and arousal (Barrett andRussell 1999 Kuppens et al 2013)mdashare basic features of con-sciousness (Damasio 1999 Dreyfus and Thompson 2007Edelman and Tononi 2000 James 18902007 Searle 1992 2004Wundt 1897) that importantly are not unique to instances ofemotion

All animals run an internal model of their world for the pur-pose of allostasis (ie the notion of an internal model is species-general) Even single-celled organisms that lack a brain learnremember make predictions and forage in service to allostasis(Freddolino and Tavazoie 2012 Sterling and Laughlin 2015)The content of any internal model is species-specific howeverincluding only the parts of the animalrsquos physical surroundings

that its brain has judged relevant for growth survival and repro-duction (ie a brain creates its affective niche in the presentbased on what has been relevant for allostasis in the past)Everything else is an extravagance that puts energy regulationat risk

As an animalrsquos integrated physiological state changes con-stantly throughout the day its immediate past determines theaspects of the sensory world that concern the animal in the pre-sent which in turn influences what its niche will contain in theimmediate future This observation prompts an important in-sight neurons do not lie dormant until stimulated by the out-side world denoted as stimulusresponse8 Ample evidenceshows that ongoing brain activity influences how the brainprocesses incoming sensory information (eg Sayers et al1974)9 and that neurons fire intrinsically within large networkswithout any need for external stimuli (Swanson 2012) The im-plications of these insights are profound namely it is very un-likely that perception cognition and emotion are localized indedicated brain systems with perception triggering emotionsthat battle with cognition to control behavior (Barrett 2009)This means classical accounts of emotion which rely on thisSR narrative are highly doubtful

An internal model is predictive not reactive

An increasingly popular hypothesis is that the brainrsquos simula-tions function as Bayesian filters for incoming sensory inputdriving action and constructing perception and other psycho-logical phenomena including emotion Simulations are thoughtto function as prediction signals (also known as lsquotop-downrsquo orlsquofeedbackrsquo signals and more recently as lsquoforwardrsquo models) thatcontinuously anticipate events in the sensory environment10

Fig 3 Neural activity during simulation N frac14 16 (data from Wilson-Mendenhall et al 2013) Participants listened with eyes closed to multimodal descriptions rich in

sensory details and imagined each real-world scenario as if it was actually happening to them (ie the experiences were high in subjective realism) Contrast presented

is scenario immersion gt resting baseline maps are FDR corrected P lt 005 Left image x frac14 1 right image xfrac1442 Heightened neural activity in primary visual cortex

(not labeled) somatosensory cortex (SSC) and MC during scenario immersion replicated prior simulation research (McNorgan 2012) and established the validity of the

paradigm Notice that simulation was associated with an increase in BOLD response within primary interoceptive cortex (ie the pINS) in the sensory integration net-

work of lateral orbitofrontal cortex (lOFC) (Ongur et al 2003) and in the thalamus increased BOLD responses were also seen as expected in limbic and paralimbic re-

gions such as the vmPFC the aINS the temporal pole (TP) SMA and vlPFC as well as in the hypothalamus and the subcortical nuclei that control the internal milieu

PAG periacquiductal gray PBN parabrachial nucleus

6 Long-range neural connections like those that form the human brainrsquosbroadly distributed intrinsic networks are particularly expensive(Bullmore and Sporns 2012 Sterling and Laughlin 2015) with most ofthe energy costs going to signaling between neurons particularly inpost-synaptic processes (Attwell and Laughlin 2001 Attwell andIadecola 2002 Alle et al 2009 Harris et al 2012)

7 A trivial example of course is that infrared light is not normally some-thing a human can see and so your brain (and mine) does not normallyrepresent it In this regard we humans have been able to expand ourecological niche (and therefore our internal models) with technology

8 This mistaken belief is an artifact of studying neurons in isolation (egHodgkin and Huxley 1952) which creates a misleading picture of howthe nervous system functions (Marder 2011) For a similar view see(Dewey 1896)

9 Also see Makeig et al 2002 2004 Mazaheri and Jensen 2010Laxminarayan et al 2011 Qian and Di 2011 Scheeringa et al 2011

10 The term lsquofeedbackrsquo derives from a time when the brain was thoughtto be largely stimulus driven (Sartorius et al 1993) Nonetheless the


This hypothesis is variously called predictive coding active in-ference or belief propagation (eg Rao and Ballard 1999Friston 2010 Seth et al 2012 Clark 2013ab Hohwy 2013 Seth2013 Barrett and Simmons 2015 Chanes and Barrett 2016Deneve and Jardri 2016)11 Without an internal model the braincannot transform flashes of light into sights chemicals intosmells and variable air pressure into music Yoursquod be experien-tially blind (Barrett 2017) Thus simulations are a vital ingredi-ent to guide action and construct perceptions in the present12

They are embodied whole brain representations that anticipate(i) upcoming sensory events both inside the body and out aswell as (ii) the best action to deal with the impending sensoryevents Their consequence for allostasis is made available inconsciousness as affect (Barrett 2017)

I hypothesize that using past experience as a guide thebrain prepares multiple competing simulations that answer thequestion lsquowhat is this new sensory input most similar torsquo (seeBar 2009ab) Similarity is computed with reference to the cur-rent sensory array and the associated energy costs and poten-tial rewards for the body That is simulation is a partiallycompleted pattern that can classify (categorize) sensory signalsto guide action in the service of allostasis Each simulation hasan associated action plan Using Bayesian logic (Deneve 2008Bastos et al 2012) a brain uses pattern completion to decideamong simulations and implement one of them (Gallivan et al2016) based on predicted maintenance of physiological effi-ciency across multiple body systems (eg need for glucose oxy-gen salt etc)

From this perspective unanticipated information from theworld (prediction error) functions as feedback for embodied simu-lations (also known as lsquobottom-uprsquo or confusingly lsquofeedforwardrsquosignals) Error signals track the difference between the predictedsensations and those that are incoming from the sensory world(including the bodyrsquos internal milieu) Once these errors are mini-mized simulations also serve as inferences about the causes ofsensory events and plans for how to move the body (or not) to dealwith them (Lochmann and Deneve 2011 Hohwy 2013) By modu-lating ongoing motor and visceromotor actions to deal with up-coming sensory events a brain infers their likely causes

In predictive coding as we will see sensory predictions arisefrom motor predictions simulations arise as a function of vis-ceromotor predictions (to control your autonomic nervous sys-tem your neuroendocrine system and your immune system)

and voluntary motor predictions which together anticipate andprepare for the actions that will be required in a moment fromnow These observations reinforce the idea that the stimu-lusresponse model of the mind is incorrect13 For a givenevent perception follows (and is dependent on) action not theother way around Therefore all classical theories of emotionare called into question even those that explain emotion as it-erative stimulusresponse sequences

The computational architecture of the brain is aconceptual system plus pattern generators

The mechanistic details of predictive coding provide yet an-other deep insight a brain implements its internal model withlsquoconceptsrsquo that lsquocategorizersquo sensations to give them meaning(Barrett 2017)14 Predictions are concepts (see Figure 4)Completed predictions are categorizations that maintainphysiological regulation guide action and construct perceptionThe meaning of a sensory event includes visceromotor andmotor action plans to deal with that event As detailed in Figure5 meaning does not trigger action but results from it Thismakes classical appraisal theories highly doubtful because theyassume that a response derives from a stimulus that is eval-uated for its meaning (eg Lazarus 1991 Scherer 2009Roseman 2011 for a discussion see (Barrett et al 2007b Grossand Barrett 2011) Appraisals as descriptions of the world (egClore and Ortony 2008) however are produced by categoriza-tion with concepts (eg Si et al 2010)

Traditionally a lsquocategoryrsquo is a population of events or objectsthat are treated as similar because they all serve a particulargoal in some context a lsquoconceptrsquo is the population of represen-tations that correspond to those events or objects15 I

history of science is laced with the idea that the mind drives percep-tion [eg in the 11th century by Ibn al-Haytham who helped to inventthe scientific method in the 18th century by Kant (1781) and in the19th century by Helmholtz] In more modern times see Craikrsquos con-cept of internal models (1943) Tolmanrsquos cognitive maps (1948)Johnson-Lairdrsquos internal models and for more recent referencesNeisser (1967) and Gregory (1980) The novelty in recent formulationscan be found in (i) the hypothesis that predictions are lsquoembodiedrsquosimulations of sensory-motor experiences (ii) they are ultimately inthe service of allostasis and therefore interoception is at their coreand of course (iii) the breadth of behavioral functional and anatomicevidence supporting the hypothesis that the brainrsquos internal modelimplements active inference as prediction signals including (iv) thespecific computational hypotheses implementing a predictive codingaccount

11 Notably Buzsaki (2006) wrote that lsquoBrains are foretelling devicesrsquoThere is accumulating evidence that prediction and prediction errorsignals oscillate at different frequencies within the brain (eg Arnaland Giraud 2012 Bressler and Richter 2015 Brodski et al 2015)

12 Simulations also constitute representations of the past (ie memories) andthe future (ie prospections) and implement imagination mind wander-ing and daydreams (Schacter et al 2007 Buckner et al 2008)

13 For an excellent treatment of the conceptual baggage in words likelsquoorganismrsquo lsquostimulusrsquo and lsquoresponsersquo see Danziger (1997) in particu-lar see the thoughtful historical discussion of how motives and emo-tions became conceptualized as sources of lsquointernalrsquo stimulation andhow physiological changes became lsquoresponsesrsquo to that stimulation

14 For those who are unfamiliar with predictive coding consider theproblem of allostasis from the perspective of your own brain For yourentire life your brain is entombed in a dark silent box (ie a skull) Ithas to figure out the causes of sensory events outside your skull toguide action in the service of allostasis but all it has access to theirconsequences in the form of sights sounds smells touches tastesand interoceptive sensations (ie sensations from your heart pump-ing your lungs expanding from inflammation from metabolic proc-esses and so on) So your brain is faced with a problem of reverseinference any given sensationmdasha flash of light or a sound or an acheor crampmdashcan have many different causes In addition the sensoryinformation is dynamically changing noisy and ambiguous Yourbrain solves this puzzle by using the only other source of informationavailable to itmdashpast experiencesmdashto create simulations that predictincoming sensory events before their consequences arrive to thebrain In this way your brain efficiently uses the statistical regular-ities from its past to anticipate future events that must be dealt with

15 Traditionally categories are supposed to exist in the world whereasconcepts are supposed to exist in the brain This distinction makessense for natural kind categories (where the boundaries exist inde-pendent of perceivers) or when the instances of a category sharephysical similarities ndash some set of statistical regularities in their sen-sory aspects or perceptual features (eg most human faces have acertain set of visual features ndash two eyes two ears a nose some hairand a mouth ndash in approximately the same orientation) In manycases however the boundary between a category and its concept isblurred For example consider a category whose instances share asimilar function but do not share any physical features (eg currencythroughout the course of human history) These conceptual

L F Barrett | 7

Fig 4 The brain is a concept generator (A) Brodmann areas are shaded to depict their degree of laminar organization including the insula (bottom right) The brainrsquos

computational architecture is depicted (adapted from Barbas 2015) where prediction signals flow from the deep layers of less granular regions (cell bodies depicted

with triangles) to the upper layers of more granular regions this can also be thought of concept construction [as described in Barrett (2017)] I hypothesize that agranu-

lar (ie limbic) cortices generatively combine past experiences to initiate the construction of embodied concepts multimodal summaries cascade to sensory and motor

systems to create the simulations that will become motor plans and perceptions Prediction error processing in turn is akin to concept learning The upper layers of

cortex compress prediction errors and reduce error dimensionality eventually creating multimodal summaries by virtue of a cytoarchitectural gradient prediction

error flows from the upper layers of primary sensory and motor regions (highly granular cortex) populated with many small pyramidal cells with few connections to-

wards less granular heteromodal regions (including limbic cortices) with fewer but larger pyramidal cells having many connections (Finlay and Uchiyama 2015) (B)

Evidence of conceptual processing in the default mode network Multimodal summaries for emotion concepts [adapted from Skerry and Saxe (2015) Figure 1B] sum-

mary representations of sensory-motor properties (color shape visual motion sound and physical manipulation [Fernandino et al (2016) Figure 5] and semantic pro-

cessing [adapted from Binder and Desai (2011) Figure 2] (C) Regions that consistently increase activity during emotional experience (green) emotion regulation (blue)

and their overlap (red) [as appears in Clark-Polner et al (2016) adapted from Buhle et al (2014) and Satpute et al (2015)] Overlaps are observed in the aIns vlPFC the

MCC SMA and posterior superior temporal sulcus Studies of emotional experience show consistent increase in activity that is consistent with manipulating predic-

tions (ie the default mode and salience networks) whereas reappraisal instructions appear to manipulate the modification of those predictions (ie the frontoparietal

and salience networks) (D) Intensity maps for five emotion categories examined by Wager et al (2015) Maps represent the expected activations or population centers

given a specific emotion category Maps also reflect expected co-activation patterns Notice that population centers for all emotion categories can be found within the

default mode and salience networks These are probabilistic summaries not brain states for emotion Adapted from Wager et al (2015)


hypothesize that in assembling populations of predictions eachone having some probability of being the best fit to the currentcircumstances (ie Bayesian priors) the brain is constructingconcepts (Barrett 2017) or what Barsalou refers to as lsquoad hocrsquoconcepts (Barsalou 1983 2003 Barsalou et al 2003) In the lan-guage of the brain a concept is a group of distributed lsquopatternsrsquoof activity across some population of neurons Incoming sen-sory evidence as prediction error helps to select from or modifythis distribution of predictions because certain simulations willbetter fit the sensory array (ie they will have stronger priors)with the end result that incoming sensory events are catego-rized as similar to some set of past experiences This in effectis the original formulation of the conceptual act theory of emo-tion (see Barrett 2006b 2017) the brain uses emotion conceptsto categorize sensations to construct an instance of emotionThat is the brain constructs meaning by correctly anticipating(predicting and adjusting to) incoming sensations Sensationsare categorized so that they are (i) actionable in a situated wayand therefore (ii) meaningful based on past experience Whenpast experiences of emotion (eg happiness) are used to cat-egorize the predicted sensory array and guide action then oneexperiences or perceives that emotion (happiness)

In other words an instance of emotion is constructed thesame way that all other perceptions are constructed using thesame well-validated neuroanatomical principles for informa-tion flow within the brain Barbas and colleaguesrsquo structuralmodel of corticocortical connections (Barbas and Rempel-Clower 1997 Barbas 2015) provides specific hypotheses abouthow concepts categorize incoming sensory inputs to guide ac-tion and create perception and in doing so fills the computa-tional and neural gaps in my initial theoretical formulation ofthe theory providing novel hypotheses about how a brain con-structs emotional events [outlined in Barrett and Simmons(2015) and Chanes and Barrett (2016) Barrett (2017)] The firstkey observation is that prediction signals are carried via lsquofeed-backrsquo connections that originate in cortical regions with theleast well-developed laminar structure referred to as lsquoagranu-larrsquo Agranular regions are cytoarchitecturally arranged to sendbut not receive prediction signals within the cerebral cortexAnother name for agranular cortices is lsquolimbicrsquo Limbic corticessuch as the anterior cingulate cortex and the ventral portion ofthe anterior insula (aINS) allostatically control physiology by re-laying descending prediction signals to the internal milieu via asystem of subcortical regions (Bar et al 2016 Kleckner et al inpress) including the central nucleus of the amygdala(Ghashghaei et al 2007) the ventral and dorsal striatum andthe central pattern generators (Swanson 2012) across hypothal-amus the parabrachial nucleus periaqueductal grey and thesolitary nucleus (see Figure 5A and B) Cortical regions with adysgranular structure which are referred to as limbic (Barbas2015) or paralimbic (Mesulam 1998) also issue descending pre-diction signals to the bodyrsquos internal milieu [eg midcingulatecortex mPFC (MCC) ventrolateral prefrontal cortex (vlPFC) pre-motor cortex (PMC) etc see Figure 5A and B] My hypothesis isthat these lsquovisceromotorrsquo regions of the brain that are respon-sible for implementing allostasis and that are usually assignedan emotional function are lsquodrivingrsquo the perception signals iethe lsquoconceptsrsquo that constitute the brainrsquos internal model in

conjunction with the hippocampus (eg see Davachi andDuBrow 2015 Hasson et al 2015)

A concept is not only the descending prediction signals thatcontrol the viscera It also includes the efferent copies of thosesignals that cascade to primary motor cortex (MC) as skeletomo-tor prediction signals as well as to all primary sensory corticesas sensory prediction signals (see Figure 5C and D respectivelyfor a discussion see Bastos et al 2012 Adams et al 2013Barbas 2015 Barrett and Simmons 2015 Chanes and Barrett2016) Following the evidence for how the cytoarchitectural gra-dients in the cortical sheet predict information flow across cor-tical regions (Barbas et al 1997 p 6608) prediction signals flowfrom deep layers of limbic cortices and terminate in the upperlayers of cortical regions with more developed (ie more granu-lar) structure such as gustatory and olfactory cortex primaryMC primary interoceptive cortex and the primary visual audi-tory and somatosensory regions16

Because MC has a laminar organization that is less well de-veloped than primary visual auditory somatosensory and in-teroceptive sensory regions (Barbas and Garcıa-Cabezas 2015) Ihypothesize that MC sends efferent copies to those sensory re-gions as sensory predictions (see Figure 5D red lines) A similararrangement between motor and somatosensory cortex hasbeen proposed by Friston and colleagues (Adams et al 2013)Furthermore because of their differential laminar developmentI hypothesize that primary interoceptive cortex in mid-to-posterior dorsal insula forwards sensory predictions to visualauditory and somatosensory cortices (propagating across eithera single or multiple synapses Figure 5D gold lines) The skeleto-motor prediction signals prepare the body for movement theinteroceptive prediction signals initiate a change in affect (iethe expected sensory consequences of allostatic changes withinthe bodyrsquos internal milieu) and the extrapersonal sensory pre-diction signals prepare upcoming perceptions This hypothesisis consistent not only with over three decades of tract tracingstudies in non-human animals but also with engineering de-sign principles (ie compute locally and relay only the informa-tion that is needed to assemble a larger pattern Sterling andLaughlin 2015) Predictions literally change the firing of primarysensory and motor neurons even though the incoming sensoryinput has not yet arrived (and may never arrive eg Kok et al2016) Accordingly all action and perception are created withconcepts All concepts contribute to allostasis and representchanges in affect not just those that construct the events thatfeel affectively intense or are created with emotion concepts

To consider how this works try this thought experiment inthe past you have experienced diverse instances of happinessmay be lying outdoors on a sunny day finishing a strenuousworkout hugging a close friend eating a piece of delectablechocolate or winning a competition Each instance is differentfrom every other and when the brain creates a concept of hap-piness to categorize and make sense of the upcoming sensory

categories have also been called abstract or nominal categoriesBiological categories are conceptual categories as we learned in On

the Origin of Species The categories of social reality such as flowersand weeds or emotion categories are conceptual because functionsare imposed on physically disparate instances by virtual of collectiveagreement (Barrett 2012)

16 Primary visual auditory and somatosensory cortices have the mostdeveloped laminar structure within the cerebral cortex and thereforereceive prediction signals but are unable to send them Primary in-teroceptive cortex is relatively less developed than these regions andtherefore sends multimodal sensory predictions to these exterocep-tive regions Primary motor cortex (with a small granular layer(Barbas and Garcıa-Cabezas 2015) has an even less well-developedlaminar structure and therefore sends predictions to somatosensorycortex (Adams et al 2013 Shipp et al 2013) as well as all the primarysensory regions mentioned so far Agranular limbic cortices send butdo not receive prediction signals because they have the least well-developed laminar structure of the entire cortical mantle

L F Barrett | 9

A

B

C

D

Fig 5 A depiction of predictive coding in the human brain (A) Key limbic and paralimbic cortices (in blue) provide cortical control the bodyrsquos internal milieu Primary

MC is depicted in red and primary sensory regions are in yellow For simplicity only primary visual interoceptive and somatosensory cortices are shown subcortical

regions are not shown (B) Limbic cortices initiate visceromotor predictions to the hypothalamus and brainstem nuclei (eg PAG PBN nucleus of the solitary tract) to

regulate the autonomic neuroendocrine and immune systems (solid lines) The incoming sensory inputs from the internal milieu of the body are carried along the

vagus nerve and small diameter C and Ad fibers to limbic regions (dotted lines) Comparisons between prediction signals and ascending sensory input results in predic-

tion error that is available to update the brainrsquos internal model In this way prediction errors are learning signals and therefore adjust subsequent predictions (C)

Efferent copies of visceromotor predictions are sent to MC as motor predictions (solid lines) and prediction errors are sent from MC to limbic cortices (dotted lines) (D)

Sensory cortices receive sensory predictions from several sources They receive efferent copies of visceromotor predictions (black lines) and efferent copies of motor

predictions (red lines) Sensory cortices with less well developed lamination (eg primary interoceptive cortex) also send sensory predictions to cortices with more

well-developed granular architecture (eg in this figure somatosensory and primary visual cortices gold lines) For simplicityrsquos sake prediction errors are not depicted

in panel D sgACC subgenual anterior cingulate cortex vmPFC ventromedial prefrontal cortex pgACC pregenual anterior cingulate cortex dmPFC dorsomedial pre-

frontal cortex MCC midcingulate cortex vaIns ventral anterior insula daIns dorsal anterior insula and includes ventrolateral prefrontal cortex SMA supplementary

motor area PMC premotor cortex mpIns midposterior insula (primary interoceptive cortex) SSC somatosensory cortex V1 primary visual cortex and MC motor

cortex (for relevant neuroanatomy references see Kleckner et al in press)


events it constructs a population of simulations (as potentialactions and perceptions) according to the rules of BayesrsquoTheorem whose priors reflect their similarity to the currentsituation (before the evidence is taken into account) The simi-larity need not be perceptualmdashit can be goal-based So the brainconstructs an on-line concept of happiness not in absoluteterms but with reference to a particular goal in the situation (tobe with friends to enjoy a meal to accomplish a task) all in theservice of allostasis This implies that lsquohappinessrsquo has a specificmeaning but its specific meaning changes from one instance tothe next

As prediction signals cascade across the synapses of a brainincoming sensory signals arriving to the brain (ie from the ex-ternal environment and the internal periphery) simultaneouslyallow for computations of prediction error that are encoded toupdate the internal model (correcting visceromotor and motoraction plans as well as sensory representations see Figure 5dotted lines) Viscerosensory prediction errors arise fromphysiologic changes within the internal milieu and ascend viavagal and small diameter afferents in the dorsal horn of the spi-nal cord through the nucleus of the solitary tract the parabra-chial nucleus the periaqueductal gray and finally to the ventralposterior thalamus before arriving in granular layer IV of theprimary interoceptive insular cortex (Damasio and Carvalho2013 Craig 2015) Notice that in the context of this frameworkperception (ie the lsquomeaningrsquo of sensory inputs) is constructedwith reference to allostasis and sensory prediction errors aretreated at a very basic level as information that guides a lsquopre-dictedrsquo visceromotor and motor action plan

Prediction errors also arise within the amygdala the basalganglia and the cerebellum and are forwarded to the cortex tocorrect its internal model (see (Beckmann et al 2009 Buckneret al 2011 Haber and Behrens 2014 Kleckner et al in press)I hypothesize that information from the amygdala to the cortexis not lsquoemotionalrsquo per se but signals uncertainty (Whalen 1998)about the predicted sensory input (via the basolateral complex)and helps to adjust allostasis (via the central nucleus) as a re-sult17 The arousal signals that are associated with increases inamygdala activity (eg Wilson-Mendenhall et al 2013) can beconsidered a learning signal (Li and McNally 2014) Similarlyprediction errors from the ventral striatum to the cortex(referred to as lsquoreward prediction errors Schultz 2016) conveyinformation about sensory inputs that impact allostasis morethan expected (ie that this information should be encoded andconsolidated in the cortex and acted upon in the moment)Dopamine is associated with engaging in vigorous action andlearning that is necessary to achieve the rewards that maintainefficient allostasis (or restore it in the event of disruption) ra-ther than playing a necessary or sufficient role in rewardsthemselves (Salamone and Correa 2012 Guitart-Masip et al2014) Other neuromodulators such as opioids seem to be moreintrinsic to reward in that regard (eg Fields and Margolis 2015)

The cerebellum models prediction errors from the peripheryand relays them to cortex to modify motor predictions [ie itpredicts the sensory consequences of a motor command muchfaster than actual sensory prediction errors can manage(Shadmehr et al 2010) and helps the cortex reduce the sensoryconsequences caused by onersquos own movements] The samemay be true for visceromotor predictions given the connectivitybetween the cerebellum and the cingulate cortex hypothal-amus and amygdala (Schmahmann and Pandya 1997 Stricket al 2009 Schmahmann 2010 Buckner et al 2011)18 Thiswould give the cerebellum a major role in allostasis conceptgeneration and the construction of emotion (eg see meta-analytic evidence in Kober et al 2008 Lindquist et al 2012Wager et al 2015)

A brain implements an internal model of the world withconcepts because it is metabolically efficient to do so Even be-fore birth a brain begins to build its internal model by process-ing prediction error from the body and the world (see Barrett2017 for discussion) Prediction errors (ie unanticipated sen-sory inputs) cascade in a feedforward cortical sweep originatingin the upper layers of cortices with more developed laminarorganization and terminating in the deep layers of cortices withless well-developed lamination As information flows from sen-sory regions (whose upper layers contain many smaller pyram-idal neurons with fewer connections) to limbic and otherheteromodal regions in frontal cortex (whose upper layers con-tain fewer but larger pyramidal neurons with many more con-nections see Figure 4A) it is compressed and reduced indimensionality (Finlay and Uchiyama 2015) This dimension re-duction allows the brain to represent a lot of information with asmaller population of neurons reducing redundancy andincreasing efficiency because smaller populations of neuronsare summarizing statistical regularities in the spiking patternsin larger populations with in the sensory and motor regionsAdditional efficiency is achieved because conceptually similarrepresentations reuse neural populations during simulation(eg Rigotti et al 2013) As a result different predictions are sep-arable but are not spatially separate (ie multimodal summa-ries are organized in a continuous neural territory that reflectstheir similarity to one another) Therefore the hypothesis isthat all new learning (eg the processing of prediction error) isconcept learning because the brain is condensing redundantfiring patterns into more efficient (and cost-effective) multi-modal summaries This information is available for later use bylimbic cortices as they generatively initiate prediction signalsconstructed as low-dimensional multimodal summaries (ielsquoabstractionsrsquo) these summaries consolidated from priorencoding of prediction errors become more detailed and par-ticular as they propagate out to more architecturally granularsensory and motor regions to complete embodied conceptgeneration

Taking a network perspective

In a keynote address in 2006 I first proposed that several of thebrainrsquos intrinsic networks (what would come to be called the de-fault mode salience and frontoparietal control networks) aredomain-general or multi-use networks that are involved in con-structing emotional episodes (eg see Figure 6 in Kober et al2008) Building on the findings so far as well as the anatomicaldistribution of limbic cortices within the brain (see Figure 5) I

17 Even more interestingly there is some evidence to suggest that thecortical regions projecting to the brainstem nuclei which originatethese neuromodulators (such as the locus coeruleus for norepineph-rine) are largely entrained by limbic cortical regions via descendingvisceromotor predictions that project directly from the anterior cin-gulate cortex and dorsomedial prefrontal cortex as well as indirectlyvia projections from the central nucleus of the amygdala and thehypothalamus (see Counts and Mufson 2012) The locus coeruleusalso receives ascending interoceptive and nociceptive predictionerrors (see Counts and Mufson 2012) This is yet another way thatallostasis is altered by modulating the gain or excitability of neuronsthat represent sensory and motor prediction errors

18 In a fly brain the mushroom bodies may play an analogous role inpredictive coding (Sterling and Laughlin 2015)

L F Barrett | 11

have refined these hypotheses (see Figure 6) I hypothesize asothers do (Mesulam 2002 Hassabis and Maguire 2009 Buckner2012) that the default mode network is necessary for the brainrsquosinternal model Regardless of the other mental categoriesmapped to default mode network activity the simulations initi-ated within this network cascade to create concepts that even-tually categorize sensory inputs and guide movements in theservice of allostasis This hypothesis is partially consistent withthe hypothesis that the default mode network represents se-mantic concepts (Binder et al 2009 Binder and Desai 2011) (seeFigure 4B) I hypothesize that the default mode network hostslsquopartrsquo of their patterns but simulations are more than justmultimodal sensorimotor summaries they are fully embodiedbrain states They emerge as default mode summaries cascadeout to primary sensory and motor regions to become detailedand particularized [ie to modulate the spiking patterns of sen-sory and motor neurons (Barrett 2017) for supporting evidenceon embodied representations of concepts see (Pulvermuller2013 Fernandino et al 2015 2016 Barsalou 2016)19

I further hypothesize that the salience network tunes the in-ternal model by predicting which prediction errors to pay atten-tion to [ie those errors that are likely to be allostaticallyrelevant and therefore worth the cost of encoding and consoli-dation called precision signals (Feldman and Friston 2010Clark 2013ab Moran et al 2013 Shipp et al 2013)]20

Specifically I hypothesize that precision signals optimize thesampling of the sensory periphery for allostasis and they aresent to every sensory system in the brain (for anatomical andfunctional justifications see (Chanes and Barrett 2016)] Theydirectly alter the gain on neurons that compute prediction errorfrom incoming sensory input (ie they apply attention) to signalthe degree of confidence in the predictions (ie the priors) con-fidence in the reliability or quality of incoming sensory signalsandor predicted relevance for allostasis Unexpected sensoryinputs that are anticipated to have resource implications (ieare likely to impact survival offering reward or threat or are ofuncertain value) will be treated as lsquosignalrsquo and learned (ieencoded) to better predict energy needs in the future with allother prediction error treated as lsquonoisersquo and safely ignored (Liand McNally 2014 for discussion see Barrett 2017) Limbic re-gions within the salience network may also indirectly signal theprecision of incoming sensory inputs via their modulation ofthe reticular nucleus that encircles that thalamus and controlsthe sensory input that reaches the cortex via thalamocorticalpathways [for relevant anatomy see (Zikopoulos and Barbas2006 2012 John et al 2016)]21 My hypothesis then is that cor-tical limbic regions within the salience network are at the coreof the brainrsquos ability to adjust its internal model to the condi-tions of the sensory periphery again in the service of allostasis(eg see Figure 6) This is consistent with the salience networkrsquos

role in attention regulation eg (Power et al 2011 Touroutoglouet al 2012 Ullsperger et al 2014 Uddin 2015)

In addition I hypothesize that neurons with the frontoparie-tal control network sculpt and maintain simulations for longerthan the several hundred milliseconds it takes to process immi-nent prediction errors) and they may also help to suppress orinhibit simulations whose priors are very low (because thosepriors are influenced not only by the current sensory array butalso by what the brain predicts for the future) It pays to be flex-ible to be able to construct and use patterns that extend overlonger periods of time (different animals have different time-scales that are relevant to their behavioral repertoire and ecolo-gical niche) Itrsquos also valuable to learn on a single trial withoutbeing guided by recurring statistical regularities in the worldparticularly if you reside in a quickly changing environment Asa prediction generator the brain is constructing simulations (asconcepts) across many different timescales (ie integrating in-formation across the few moments that constitute an event butalso across longer time frames at various scales for similarideas see Wilson et al 2010 Hasson et al 2015) Therefore abrain may be pattern matching to categorize not only on shortprocessing timescales of milliseconds but also on much longertimescales (seconds to minutes to hours or even longer) Thelesson here for the science of emotion is that the brain doesnot process individual stimulimdashit processes events across tem-poral windows Emotion perception is event perception not ob-ject perception22

The theory of constructed emotion

Now we can see how a multi-level constructionist view like thetheory of constructed emotion offers an approach to under-standing the brain basis of emotion that is consistent withemerging computational and evolutionary biological views ofthe nervous system23 A brain can be thought of as running aninternal model that controls central pattern generators in theservice of allostasis (for more on pattern generators seeBurrows 1996 Sterling and Laughlin 2015 Swanson 2000) Aninternal model runs on past experiences implemented as con-cepts A concept is a collection of embodied whole brain repre-sentations that predict what is about to happen in the sensoryenvironment what the best action is to deal with impendingevents and their consequences for allostasis (the latter is madeavailable to consciousness as affect) Unpredicted information(ie prediction error) is encoded and consolidated whenever it ispredicted to result in a physiological change in state of perceiver(ie whenever it impacts allostasis) Once prediction error isminimized a prediction becomes a perception or an experienceIn doing so the prediction explains the cause of sensory eventsand directs action ie it categorizes the sensory event In thisway the brain uses past experience to construct a

19 Notice that this hypothesis is species-general rats have a defaultmode network and are not able to engage in mental time travel as faras we can tell (Hsu et al 2016) but this in no way disconfirms the hy-pothesis that the network is running an internal model of the ani-malrsquos world in the service of allostasis

20 This allows for the encoding of statistical patterns of uncertain valuethat can later be reconstructed when they are of use (Dunsmoor et al2015)

21 Salience regions may also play a role in maintaining the internalmodel that is unconstrained by the sensory world (such as when con-structing memories imaginings dreams reveries mind-wanderingand so on) Salience regions also help accomplish multimodal inte-gration [compare eg the topography of the salience network and themultimodal integration network found in Sepulcre et al (2012)]

22 The frontopartial control network (which contains key limbic richclub hubs in the midcingulate cortex and anterior insula) may alsohave a role to play in managing sensory prediction errors by applyingattention to select those body movements that will generate the ex-pected sensory inputs presumably with help from cerebellar and stri-atal prediction errors

23 The theory of constructed emotion integrates ideas and empiricalfindings from neuroconstruction (Mareschal et al 2007 Westermannet al 2007 Karmiloff-Smith 2009) rational constructivism (Xu andKushnir 2013) psychological construction (Russell 2003 Barrett2006b 2012 2013 Barrett et al 2015) and social construction (Boigerand Mesquita 2012) as well as descriptive appraisal theories (egClore and Ortony 2008)


categorization [a situated conceptualization (Barsalou 1999Barsalou et al 2003 Barrett 2006b Barrett et al 2015)] that bestfits the situation to guide action The brain continually con-structs concepts and creates categories to identify what the sen-sory inputs are infers a causal explanation for what causedthem and drives action plans for what to do about them Whenthe internal model creates an emotion concept the eventualcategorization results in an instance of emotion

This hypothesis is consistent with the conceptual innov-ations in Darwinrsquos On the Origin of Species [(Darwin 18591964)even as it is inconsistent with lsquoThe Expression of the Emotions in

Man and Animalsrsquo (Darwin 18722005) for a discussion seeBarrett 2017] Some of the psychological constructs used in thetheory of constructed emotion are species-general (eg allosta-sis interoception affect and concept) while others require thecapacity for certain types of concepts (Barrett 2012) and aremore species-specific (eg emotion concepts) It is necessary tounderstand which constructs are species-general vs species-specific to solve the puzzle of the biological basis of emotionMistaking one for the other is a category error that interfereswith scientific progress

Constructionism as a scientific paradigm makes differentassumptions than the classical paradigm (Barrett 2015) asksdifferent questions and requires different methods and analyticprocedures than those of the classical view (whose methods areill-suited to testing it) As a consequence constructionism isoften profoundly misunderstood [for two recent examples see(Anderson and Adolphs 2014 Kragel and LaBar 2016)] Withthese observations in mind here is a partial list of claims I amnot making to avoid further confusion

i I am not saying that emotions are illusions Irsquom sayingthat emotion categories donrsquot have distinct dedicatedneural essences Emotion categories are as real as anyother conceptual categories that require a human per-ceiver for existence such as lsquomoneyrsquo (ie the various ob-jects that have served as currency throughout humanhistory share no physical similarities Barrett 2012 2017)

ii I am not saying that all neurons do everything (aka equi-potentiality) I am suggesting that a given neuron doesmore than one thing (has more than one receptive field)and that there are no emotion-specific neurons

iii I am not claiming that networks are Lego blocks with a staticconfiguration and an essential function I am suggesting thatwhen it comes to understanding the physical basis of psycho-logical categories it is necessary to focus on ensembles ofneurons rather than individual neurons A neuron does notfunction on its own and many neurons are part of more thanone network Moreover networks function via degeneracymeaning that a given network has a repertoire of functionalconfigurations (ie functional motifs) that is constrained byits anatomical structure (ie its structural motif)

iv I am not claiming that subcortical regions are irrelevant toemotion I hypothesize that an instance of emotion is a brainstate that makes the sensory array meaningful and in sodoing engages the pattern generators for whatever actionsare functional in the context given a personrsquos current state

v I am not saying that the default mode and salience net-works implement allostasis and therefore should not bemapped to other psychological categories I am claimingthat these (and other) domain-general networks can bemapped to many psychological categories at the same time

Fig 6 A large-scale system for allostasis and interoception in the human brain (A) The system implementing allostasis and interoception is composed of two large-scale

intrinsic networks (shown in red and blue) that are interconnected by several hubs (shown in purple for coordinates see Kleckner et al in press) Hubs belonging to the

lsquorich clubrsquo are labeled These maps were constructed with resting state BOLD data from 280 participants binarized at p lt 105 and then replicated on a second sample of

270 participants vaIns ventral anterior insula MCC midcingulate cortex PHG parahippocampal gyrus PostCG postcentral gyrus PAG periaqueductal gray PBN para-

brachial nucleus NTS the nucleus of the solitary tract vStriat ventral striatum Hypothal hypothalamus (B) Reliable subcortical connections thresholded P lt 005 un-

corrected replicated in 270 participants

L F Barrett | 13

vi I am not saying that concepts are stored in the defaultmode network Irsquom saying that the default mode networkrepresents efficient multimodal summaries from which acascade of predictions issues through the entire corticalsheet terminating in primary sensory and motor regionsThe whole cascade is an instance of a concept

vii I am not saying that emotions are deliberate nor denyingthat automaticity exists I am saying that in humans ac-tual executive control (eg via the frontoparietal controlnetwork in primates) and the experience of feeling in con-trol are not synonymous (Barrett et al 2004) All animalbrains create concepts to categorize sensory inputs andguide action in an obligatory and automatic way outsideof awareness Automaticity and control are different brainmodes (each of which can be achieved with a variety ofnetwork configurations) not two battling brain systems

viii I am not saying that non-human animals are emotionlessIrsquom saying that emotion is perceiver-dependent so ques-tions about the nature of emotion must include a

perceiver lsquoIs the fly fearfulrsquo is not a scientific questionbut lsquoDoes a human perceive fear in the flyrsquo and lsquoDoes thefly feel fearrsquo can be answered scientifically (and the an-swers are lsquoyesrsquo and lsquonorsquo) Notice that I am not claiming thata fly feels nothing it may feel affect (Barrett 2017)

Selected implications of the theory

Scientific revolutions are difficult At the beginning new para-digms raise more questions than they answer They may ex-plain existing anomalies or redefine lingering questions out ofexistence but they also introduce a new set of questions thatcan be answered only with new experimental and computa-tional techniques This is a feature not a bug because it fostersscientific discovery (Firestein 2012) A new paradigm barelygets started before it is criticized for not providing all the an-swers But progress in science is often not answering old ques-tions but asking better ones The value of a new approach isnever based on answering the questions of the old approach

Table 2 Selected neuroscience evidence supporting the theory of constructed emotion


Degeneracy mapping many neurons regionsnetworks or patterns to one emotion category

Human neuroimaging task-relateddata

(Vytal and Hamann 2010 Lindquist et al2012 Wilson-Mendenhall et al 2011 2015Oosterwijk et al 2015)


Human neuroimaging multi-voxelpattern analysis

(Clark-Polner Johnson and barrett 2016)compare the different patterns for a givenemotion category across (Kragel and LaBar2015 Wager et al 2015 Saarimaki et al2016)


Intracranial stimulation in humans (Guillory and Bujarski 2014)


Behavioral observations in humanswith amygdala lesions

(Becker et al 2012 Mihov et al 2013)


Optogenetic research showing manyto one mappings for behaviors inrodents

(Herry and Johansen 2014)

Neural reuse Mapping one neural assembly tomany emotion categories


(Vytal and Hamann 2010 Wilson-Mendenhall et al 2011 Lindquist et al2012)


Human neuroimaging intrinsic con-nectivity data

(Wilson-Mendenhall et al 2011 Barrett andSatpute 2013 Touroutoglou et al 2015)


Optogenetic research and some lesionresearch in rodents

(Tovote et al 2015)

Predictive coding explains aversive (lsquofearrsquo)learning


(Furlong et al 2010 McNally et al 2011 Liand McNally 2014)

Emotion concepts are embodied Human neuroimaging task-relateddata

(Oosterwijk et al 2012 2015)

Multimodal summaries of emotion concepts arerepresented in the default mode network


(Peelen et al 2010 Skerry and Saxe 2015)

Default mode and salience network interconnec-tivity is associated with the intensity of emo-tional experiences (as distinct from arousal)


(Raz et al 2016)

Embodied simulations are associated withincreased activity in primary interoceptivecortex


(Wilson-Mendenhall et al under review)


Such is the case with the theory of constructed emotionEvidence from various domains of research is consistent withthe proposed hypotheses (for select neuroscience examples seeTable 2) even as it casts aside some of the old unansweredquestions of the classical view

Ironically perhaps the strongest evidence to date for thetheory comes from studies that use pattern classification to dis-tinguish categories of emotion Several recent articles takingthis approach have reported success in differentiating one emo-tion category from anothermdasha finding that is routinely con-strued as providing the long awaited support for the classicalview (Kassam et al 2013 Kragel and LaBar 2015 Saarimaki etal 2016) However patterns that distinguish among the catego-ries in one study do not replicate in the other studies The sameis true for studies that successfully created different patterns ofautonomic physiology despite using the same stimuli and ex-perimental method and sampling from the same population(eg Stephens et al 2010 Kragel and LaBar 2013)

Generally speaking pattern classification results in the sci-ence of emotion are routinely misinterpreted A pattern thatdiagnoses sadness is not the brain state for sadness but merelya statistical summary of a highly variable set of instances Toassume otherwise is an essentialist error that mistakes a statis-tical summary for the norm24 Indeed the voxels that make upa pattern for a category need not observed in every (or even any)single instance of that category A classic study by Posner andKeele (1968) demonstrated a similar general phenomenon

almost half a century ago and we have confirmed this with asimple mathematical simulation (see Clark-Polner Johnson andBarrett 2016)

The theory of constructed emotion is consistent with theolder literature on decorticate animals that appears to supportthe classical view For example consider the experiment byWoodworth and Sherrington (1904) who surgically removed thecerebral cortex thalamus and hypothalamus of cats andobserved what they referred to as lsquopseud-affectiversquo (for pseudo-affective) reflexesmdashreflexive motor actions were left intact butthe lsquoaffectiversquo (ie allostatic) driver of these responses was goneAs a consequence these animals appeared to behave emotion-ally but the actions were no longer in service of survival Thesefindings can be interpreted as demonstrating the existence ofpattern generators when the machinery of the conceptual sys-tem has been removed A similar observation can be made forCannon and Brittonrsquos (1925) lsquosham ragersquo where a decorticatedcat upon waking from anesthesia spontaneously spit clawedand arched its back Importantly Cannon referred to these ac-tions as lsquofuryrsquo and in doing so inferred the presence of a mentalstate from a set of actions

Scientists carefully map the circuitry for behaviors (or meremovements in some cases) in non-human animals but somemistakenly believe that they are mapping the circuitry for emo-tions An example is observing freezing behavior in a rat (eg inresponse to electric shock) and calling it lsquofearrsquo Rats donrsquot freezeonly in situations that we perceive as threatening (ie one be-havior maps to many categories) and rats exhibit a variety ofbehaviors in threatening situations (ie many behaviors map toone category) This error assuming that an action is equivalentto an emotion which I call the mental inference fallacy (Barrett

Box 1 The curious case of SM

Much of our understanding of the neural basis of fear comes from studying Patient SM She has difficulty experiencing fearin many normative circumstances (eg horror movies haunted houses) but she experiences intense fear during experimentswhere she is asked to breathe air with higher concentrations of CO2 She is able to mount a normal skin conductance re-sponse to an unexpectedly loud sound but her brain seems not to use arousal as a learning cue in mild situations (eg stand-ard lsquofear learningrsquo paradigms) and she therefore has difficulty learning from prior errors (eg she does not mount an anticipa-tory skin conductance response to aversive stimuli during the Iowa Gambling Task and she does not show loss aversionwhen gambling) Interestingly however there is other evidence that SM is capable of what is typically called lsquofear learningrsquoSM is averse to breaking the law for fear of getting in trouble She also spontaneously reports feeling worried She is ablelsquolearn fearrsquo in the real world (eg she is averse to seeking medical treatment or visiting the dentist because of pain she expe-rienced on a previous occasions)

SMrsquos impairments in fear perception appear to be clearest in experiments where she is asked to view stereotyped fearposes and explicitly categorize them as fearful (although she shows more widespread deficits in explicitly perceiving arousalin posed faces and she appears to have no difficulty rapidly processing fear faces outside of consciousness) She can perceivefear in bodies and voices (as is evidenced by her efforts to help her friend or call the police for others in danger) SM caneven perceive fear in posed faces when her attention is directed to the eyes of the stimulus face consistent with evidencethat some amygdala neurons are particularly sensitivity to the sclera of eyes (the faces that depict stereotyped fear posescontain widen eyes) By contrast other patients with Urbach-Wiethe Disease spend a longer time looking at the widenedeyes of stereotyped fear poses and have no difficulty correctly categorizing those faces as fearful

It should be noted (but rarely is) that SMrsquos brain shows abnormalities that extend beyond the amygdala including the an-terior entorhinal cortex and ventromedial prefrontal cortex both of which show dense reciprocal connections to the amyg-dala and very likely play a role in SMrsquos specific behavioral profile It is also important to note that SM has had difficultiessustaining long-term relationships including friendships and is distressed by this There seems to be one clear exceptionSM has been able to maintain a relationship with the scientists she has worked with for almost two decades SM callsthem for support (eg when she is worried or afraid such as when did not want to return for painful medical treatment)They help her with the details of daily life (financial and otherwise) It would be interesting to examine whether SM is awareof their hypothesis that the amygdala contains the circuitry for fear

For references see Table 1 and also Feinstein et al (2016)

24 The average size of a US family in 2015 was 314 persons but thatdoes not mean that every family (or even any family) contained 314people

L F Barrett | 15

2017) has wreaked havoc with the scientific accumulation ofknowledge about emotion (also see Barrett 2012 LeDoux 2012)Motor movements do not provide a direct indication of an in-ternal state be it in a rodent a monkey or a human [eg see dec-ades of research on mental inference processes (Gilbert 1998)and opacity of mind (Robbins and Rumsey 2008)]

When viewed in this light it is an error to claim that studiesof the human brain using functional magnetic resonance imag-ing (fMRI) yield different results than studies of non-human ani-mal brains with lesions or optogenetics (eg Adolphs 2013) Inreality all are making physical measurements and mappingemotion concepts to them and no set of findings is describedappropriately by classical emotion concepts For example in anattempt to deal with all the variation within the category lsquofearrsquoscientists create finer-grained typologies in an attempt to bringnature under control and make it easier to identify their es-sences (eg Gross and Canteras 2012) But this does not avoidthe problem of the mental state fallacy Furthermore it makesno sense to elevate categories for anger sadness fear disgustand happiness to a common ethological framework for compar-ing humans with other animals when there is ample evidencefrom linguistics anthropology and psychology that these cate-gories do not offer a robust universal framework for comparinghumans of different cultures (Russell 1991 Gendron et al2014ab 2015 Wierzbicka 2014 Crivelli et al 2016)

Conclusions and future directions

Scientists must abandon essentialism and study emotions in alltheir variety We must not merely focus on the few stereotypesthat have been stipulated based on a very selective reading ofDarwin We must assume variability to be the norm ratherthan a nuisance to be explained after the fact It will never bepossible to measure an emotion by merely measuring facialmuscle movements changes in autonomic nervous system sig-nals or neural firing within the periaqueductal gray or theamygdala To understand the nature of emotion we must alsomodel the brain systems that are necessary for making mean-ing of physical changes in the body and in the world

This article is a mere sketch of a much larger scientific land-scape The theory of constructed emotion proposes that emo-tions should be modeled holistically as whole brain-bodyphenomena in context My key hypothesis is that the dynamicsof the default mode salience and frontoparietal control net-works form the computational core of a brainrsquos dynamic in-ternal working model of the body in the world entrainingsensory and motor systems to create multi-sensory representa-tions of the world at various time scales from the perspective ofsomeone who has a body all in the service of allostasis (for evi-dence consistent with this view see van den Heuvel et al 2012van den Heuvel and Sporns 2011 2013) In other words allosta-sis (predictively regulating the internal milieu) and interocep-tion (representing the internal milieu) are at the anatomical andfunctional core of the nervous system These insights offer arange of new hypothesesmdasheg that reappraisal and other regu-lation processes (Etkin et al 2015 Gross 2015) are accomplishedwith predictions that categorize sensory inputs and control ac-tion with concepts (see Figure 4C)

The theory of constructed emotion also views the distinctionbetween the central and peripheral nervous systems as histor-ical rather than as scientifically accurate For example ascend-ing interoceptive signals bring sensory prediction errors fromthe internal milieu to the brain via lamina I and vagal afferentpathways and they are anatomically positioned to be

modulated by descending visceromotor predictions that controlthe internal milieu (eg Fields 2004) This suggests the hypoth-esis that concepts (ie prediction signals) act like a volume dialto influence the processing of prediction errors before they evenreach the brain This provides new hypotheses about the chron-ification of pain (see Barrett 2017) that considers pain and emo-tion as two sides of the same coin rather than separatephenomena that influence one another

Emotions are constructions of the world not reactions to itThis insight is a game changer for the science of emotion It dis-solves many of the debates that remained mired in philosoph-ical confusion and allows us to better understand the value ofnon-human animal models without resorting to the perils ofessentialism and anthropomorphism It provides a commonframework for understanding mental physical and neurodege-nerative disorders (eg Barrett and Simmons 2015 BarrettQuigley amp Hamilton 2016 Barrett 2017) and collapses the artifi-cial boundaries between cognitive affective and social neuro-sciences (see Barrett amp Satpute 2013) Ultimately the theory ofconstructed emotion equips scientists with new conceptualtools to solve the age-old mysteries of how a human nervoussystem creates a human mind

Funding

This article was prepared with support from the NationalInstitute on Aging (R01 AG030311) the National CancerInstitute (U01 CA193632) the National Science Foundation(CMMI 1638234) and the US Army Research Institute for theBehavioral and Social Sciences (W911NF-15-1-0647 andW911NF- 16-1-0191) The views opinions andor findingscontained in this paper are those of the authors and shallnot be construed as an official Department of the Army pos-ition policy or decision unless so designated by otherdocuments

Conflict of interest None declared

Glossary

Agranular Cerebral cortex with the least developed laminar or-ganization involving no definable layer IV and no clear distinc-tion between the neurons in layers II and III

Allostasis Regulating the internal milieu by anticipatingphysiological needs and preparing to meet them before theyarise

Concept Traditionally a category is a group of instances thatare similar for some function or purpose a concept is the men-tal representations of those category members In the theory ofconstructed emotion a concept is a collection of embodiedwhole brain representations that predicts what is about to hap-pen in the sensory environment what the best action is to dealwith these impending events and their consequences forallostasis

Degeneracy Degeneracy refers to the capacity for biologicallydissimilar systems or processes to give rise to identical func-tions Degeneracy is different from redundancy (which is ineffi-cient and to be avoided)

Dysgranular Cerebral cortex with a moderately developedlaminar organization involving a rudimentary layer IV and bet-ter developed layers II and III

Hub A group of the brainrsquos most inter-connected neuronsThe hubs with the most dense connections are referred to as


lsquorich clubrsquo hubs and include visceromotor regions as well asother heteromodal regions They are thought to function as ahigh-capacity backbone for synchronizing neural activity inte-grating information (and segregating noise) across the entirebrain

Internal Milieu An integrated sensory representation of thephysiological state of the body

Laminar Organization The architectural organization of neu-rons in a cortical column

Naıve Realism The belief that onersquos senses provide an accur-ate and objective representation of the world

Pattern Generators Groups of neurons (ie nuclei) that imple-ment the sequences of actions for coordinated behaviors likefeeding running and mating An action is a single movementbut a behavior is an event Pattern generators are in the hypo-thalamus and down in the brainstem near their effectormuscles and organs (Sterling and Laughlin 2015 Swanson2005)

Visceromotor Internal movements involving autonomic neu-roendocrine and immune systems

Acknowledgements

We thank to Ajay Satpute Amitai Shenhav SuzanneOosterwijk and Eliza Bliss-Moreau who read andor madeinvaluable comments on an earlier draft of this article

ReferencesAdams RA Shipp S Friston KJ (2013) Predictions not com-

mands active inference in the motor system Brain Structureand Function 218(3) 611ndash43

Adolphs R (2013) The biology of fear Current Biology 23(2)R79ndash93

Adolphs R Russell JA Tranel D (1999) A role for the humanamygdala in recognizing emotional arousal from unpleasantstimuli Psychological Science 10(2)167ndash71

Adolphs R Tranel D (1999) Intact recognition of emotionalprosody following amygdala damage Neuropsychologia37(11)1285ndash92

Adolphs R Tranel D (2003) Amygdala damage impairs emo-tion recognition from scenes only when they contain facial ex-pressions Neuropsychologia 41(10)1281ndash9

Alle H Roth A Geiger JR (2009) Energy-efficient action po-tentials in hippocampal mossy fibers Science325(5946)1405ndash8 doi101126science1174331

Anderson DJ Adolphs R (2014) A framework for studyingemotion across species Cell 157 187ndash200

Anderson ML (2014) After Phrenology Neural Reuse and theInteractive Brain Cambridge MIT Press

Anderson ML Finlay BL (2014) Allocating structure to func-tion the strong links between neuroplasticity and natural se-lection Frontiers in Human Neuroscience 7 918

Antoniadis EA Winslow JT Davis M Amaral DG (2009)The nonhuman primate amygdala is necessary for the acquisi-tion but not the retention of fear-potentiated startle BiologicalPsychiatry 65(3) 241ndash8

Arnal LH Giraud AL (2012) Cortical oscillations and sensorypredictions Trends in Cognitive Sciences 16(7) 390ndash8

Atkinson AP Heberlein AS Adolphs R (2007) Spared ability torecognise fear from static and moving whole-body cues followingbilateral amygdala damage Neuropsychologia 45(12) 2772ndash82

Attwell D Iadecola C (2002) The neural basis of functionalbrain imaging signals Trends in Neuroscience 25(12) 621ndash5

Attwell D Laughlin SB (2001) An energy budget for signalingin the grey matter of the brain Journal of Cerebral Blood Flow andMetabolism 21(10) 1133ndash45

Bar KJ de la Cruz F Schumann A et al (2016) Functionalconnectivity and network analysis of midbrain and brainstemnuclei NeuroImage 134 53ndash63

Bar M (2009a) A cognitive neuroscience hypothesis of moodand depression Trends in Cognitive Sciences 13(11) 456ndash63

Bar M (2009b) The proactive brain memory for predictionsPhilosophical Transactions of the Royal Society B Biological Sciences364(1521) 1235ndash43

Barbas H (2015) General cortical and special prefrontal connec-tions principles from structure to function Annual Review ofNeuroscience 38 269ndash89

Barbas H Garcıa-Cabezas M (2015) Motor cortex layer 4 less ismore Trends in Neurosciences 38(5)259ndash61

Barbas H Rempel-Clower N (1997) Cortical structure predicts thepattern of corticocortical connections Cerebral Cortex 7(7)635ndash46

Bargmann CI (2012) Beyond the connectome how neuromo-dulators shape neural circuits Bioessays 34(6)458ndash65doi101002bies201100185

Barrett LF (2006a) Emotions as natural kinds Perspectives onPsychological Science 1 28ndash58

Barrett LF (2006b) Solving the emotion paradox categorizationand the experience of emotion Personality and Social PsychologyReview 10(1) 20ndash46

Barrett LF (2009) The future of psychology connecting mind tobrain Perspectives on Psychological Science 4(4) 326ndash39

Barrett LF (2011a) Bridging token identity theory and superve-nience theory through psychological constructionPsychological Inquiry 22(2) 115ndash27

Barrett LF (2011b) Was Darwin wrong about emotional expres-sions Current Directions in Psychological Science 20 400ndash6

Barrett LF (2012) Emotions are real Emotion 12(3) 413ndash29Barrett LF (2013) Psychological construction A Darwinian ap-

proach to the science of emotion Emotion Review 5 379ndash89Barrett LF (2014) The conceptual act theory a precis Emotion

Review 6 292ndash7Barrett LF (2015) Ten common misconceptions about the

psychological construction of emotion In Barrett LFRussell JA editors The psychological construction of emotion p45-79 New York Guilford

Barrett LF (2017) How Emotions Are Made The Secret Life the BrainNew York NY Houghton-Mifflin-Harcourt

Barrett LF Bliss-Moreau E (2009) Affect as a psychologicalprimitive Advances in Experimental Social Psychology 41167ndash218

Barrett LF Lindquist KA Bliss-Moreau E et al (2007a) Ofmice and men natural kinds of emotions in the mammalianbrain A response to Panksepp and Izard Perspectives onPsychological Science 2(3) 297ndash311

Barrett LF Mesquita B Ochsner KN Gross JJ (2007b) Theexperience of emotion Annual Review of Psychology 58373ndash403

Barrett LF Russell JA (1999) Structure of current affectControversies and emerging consensus Current Directions inPsychological Science 8(1) 10ndash4

Barrett LF Satpute AB (2013) Large-scale brain networks inaffective and social neuroscience towards an integrative func-tional architecture of the brain Current Opinion in Neurobiology23(3) 361ndash72

Barrett LF Simmons WK (2015) Interoceptive predictions inthe brain Nature Reviews Neuroscience 16 419ndash29

L F Barrett | 17

Barrett LF Tugade MM Engle RW (2004) Individual differ-ences in working memory capacity and dual-process theoriesof the mind Psychological Bulletin 130(4)553ndash73

Barrett LF Wilson-Mendenhall CD Barsalou LW (2015) Theconceptual act theory a road map In Barrett LF Russell JAeditors The Psychological Construction of Emotion New YorkGuilford

Barsalou LW (1983) Ad hoc categories Memory and Cognition11(3) 211ndash27

Barsalou LW (1999) Perceptual symbol systems Behavioral andBrain Science 22(4) 577ndash609 discussion 610-560

Barsalou LW (2003) Situated simulation in the human concep-tual system Language and Cognitive Processes 18 513ndash62

Barsalou LW (2008) Grounded cognition Annual Review ofPsychology 59 617ndash45

Barsalou LW (2016) On staying grounded and avoidingQuixotic dead ends Psychonomic Bulletin and Review 23(4)1122ndash42

Barsalou LW Kyle Simmons W Barbey AK Wilson CD(2003) Grounding conceptual knowledge in modality-specificsystems Trends in Cognitive Sciences 7(2) 84ndash91

Bastos AM Usrey WM Adams RA Mangun GR Fries PFriston KJ (2012) Canonical microcircuits for predictive cod-ing Neuron 76(4) 695ndash711

Bechara A Damasio H Damasio AR Lee GP (1999)Different contributions of the human amygdala and ventro-medial prefrontal cortex to decision-making Journal ofNeuroscience 19(13) 5473ndash81

Bechara A Tranel D Damasio H Adolphs R Rockland CDamasio AR (1995) Double dissociation of conditioning anddeclarative knowledge relative to the amygdala and hippo-campus in humans Science 269(5227) 1115ndash8

Becker B Mihov Y Scheele D et al (2012) Fear processing andsocial networking in the absence of a functional amygdalaBiological Psychiatry 72(1) 70ndash7

Beckmann M Johansen-Berg H Rushworth MF (2009)Connectivity-based parcellation of human cingulate cortexand its relation to functional specialization Journal ofNeuroscience 29(4) 1175ndash90

Berkes P Orban G Lengyel M Fiser J (2011) Spontaneouscortical activity reveals hallmarks of an optimal internalmodel of the environment Science 331(6013) 83ndash7

Binder JR Desai RH (2011) The neurobiology of semanticmemory Trends in Cognitive Sciences 15(11) 527ndash36

Binder JR Desai RH Graves WW Conant LL (2009) Whereis the semantic system A critical review and meta-analysis of120 functional neuroimaging studies Cerebral Cortex 19(12)2767ndash96

Boes AD Mehta S Rudrauf D et al (2011) Changes in corticalmorphology resulting from long-term amygdala damageSocial Cognitive and Affective Neuroscience 7(5) 588ndash95

Boiger M Mesquita B (2012) The construction of emotion ininteractions relationships and cultures Emotion Review 4

Bressler SL Richter CG (2015) Interareal oscillatory syn-chronization in top-down neocortical processing CurrentOpinion in Neurobiology 31 62ndash6

Brodski A Paach GF Helbling S Wibral M (2015) The facesof predictive coding The Journal of Neuroscience 35 8997ndash9006

Buckner RL (2012) The serendipitous discovery of the brainrsquosdefault network NeuroImage 62(2) 1137ndash45

Buckner RL Andrews-Hanna JR Schacter DL (2008) Thebrainrsquos default network anatomy function and relevance todisease Annals of the New York Academy of Sciences 1124 1ndash38

Buckner RL Krienen FM Castellanos A Diaz JC Yeo BT(2011) The organization of the human cerebellum estimatedby intrinsic functional connectivity Journal of Neurophysiology106(5) 2322ndash45 doi101152jn003392011

Buhle JT Silvers JA Wager TD et al (2014) Cognitive re-appraisal of emotion a meta-analysis of human neuroimagingstudies Cerebral Cortex

Bullmore E Sporns O (2012) The economy of brain network or-ganization Nature Reviews Neuroscience 13(5) 336ndash49

Burrows M (1996) The Neurobiology of an Insect Brain OxfordNew York Oxford University Press

Buzsaki G (2006) Rhythms of the Brain Oxford New York OxfordUniversity Press

Cannon WB Britton SW (1925) Studies on the conditions ofactivity in endocrine glands American Journal of PhysiologyndashLegacy Content 72(2) 283ndash94

Capitanio JP Cole SW (2015) Social instability and immunityin rhesus monkeys the role of the sympathetic nervous sys-tem Philosophical Transactions of the Royal Society B BiologicalScicnes 370(1669) 20140104

Carrive P Morgan MM (2012) Periaquiductal gray In Mai JKPaxinos G editors The Human Nervous System 3rd edn pp367ndash400 Boston Elsevier

Chanes L Barrett LF (2016) Redefining the Role of LimbicAreas in Cortical Processing Trends in Cognitive Sciences 20(2)96ndash106

Clark A (2013a) Whatever next Predictive brains situatedagents and the future of cognitive science Behavioral and BrainSciences 36 281ndash53

Clark A (2013b) The many faces of precision (Replies to com-mentaries on ldquoWhatever next Neural prediction situatedagents and the future of cognitive sciencerdquo) Frontiers inPsychology 4 270

Clark-Polner E Johnson T Barrett LF (2016) Multivoxel pat-tern analysis does not provide evidence to support the exist-ence of basic emotions Cerebral Cortex doi 101093cercorbhw028

Clark-Polner E Wager TD Satpute AB Barrett LF (2016)Neural fingerprinting Meta-analysis variation and the searchfor brain-based essences in the science of emotion In BarrettLF Lewis M Haviland-Jones JM editors The handbook ofemotion 4th edn p 146ndash165 New York Guilford

Clore GL Ortony A (2008) Appraisal theories how cognitionshapes affect into emotion In Lewis M Haviland-Jones JMBarrett LF editors Handbook of Emotions 3rd edn pp 628ndash42New York Guilford Press

Conant RC Ross Ashby W (1970) Every good regulator of asystem must be a model of that systemdagger International Journal ofSystems Science 1(2) 89ndash97

Counts SE Mufson EJ (2012) The human nervous system InMai JK Paxinos G editors The Human Nervous System pp425ndash38 Boston MA Elsevier

Craig AD (2015) How Do You Feel an Interoceptive Moment withYour Neurobiological Self Princeton Princeton UniversityPress

Crivelli C Jarillo S Russell JA Fernandez-Dols JM (2016)Reading emotions from faces in two indigenous societiesJournal of Experimental Psychology-General 145(7) 830ndash43

Crossley NA Mechelli A Scott J et al (2014) The hubs of thehuman connectome are generally implicated in the anatomyof brain disorders Brain 137(8) 2382ndash95

Damasio A Carvalho GB (2013) The nature of feelings evolu-tionary and neurobiological origins Nature ReviewsNeuroscience 14(2) 143ndash52


Damasio AR (1989) Time-locked multiregional retroactivationa systems-level proposal for the neural substrates of recall andrecognition Cognition 33(1ndash2) 25ndash62

Damasio AR (1999) The Feeling of What Happens Body andEmotion in the Making of Consciousness Houghton HoughtonMifflin Harcourt

Dantzer R Heijnen CJ Kavelaars A Laye S Capuron L(2014) The neuroimmune basis of fatigue Trends inNeurosciences 37(1) 39ndash46

Danziger K (1997) Naming the Mind How Psychology Found ItsLanguage London England Sage

Darwin C (18591964) On the Origin of Species A Facsimile of theFirst Edition Cambridge MA Harvard University Press

Darwin C (18722005) The Expression of the Emotions in Man andAnimals Stilwell KS Digireads com

Davachi L DuBrow S (2015) How the hippocampus preservesorder the role of prediction and context Trends in CognitiveSciences 19(2) 92ndash9

Davis M (1992) The role of the amygdala in fear and anxietyAnnual Review of Neuroscience 15 353ndash75

de Gelder B Terburg D Morgan B Hortensius R Stein D Jvan Honk J (2014) The role of human basolateral amygdala inambiuous social threat perception Cortex 52 28ndash34

De Martino B Camerer CF Adolphs R (2010) Amygdala dam-age eliminates monetary loss aversion Proceedings of theNational Academy of Sciences of the United States of America107(8) 3788ndash92

Deneve S (2008) Bayesian spiking neurons I inference NeuralComputation 20 91ndash117

Deneve S Jardri R (2016) Circular inference mistaken be-lief misplaced trust Current Opinion in Behavioral Sciences 1140ndash8

Dewey J (1896) The reflex arc concept in psychology ThePsychological Review 3 357ndash70

Doya K (2008) Modulators of decision making NatureNeuroscience 11(4) 410ndash6

Dreyfus G Thompson E (2007) Asian perspectives Indian the-ories of mind In Zelazo PD Moscovitch M Thompson Eeditors The Cambridge Handbook of Consciousness pp 89ndash114Cambridge Cambridge University Press

Dunsmoor JE Murty VP Davachi L Phelps EA (2015)Emotional learning selectively and retroactively strengthensmemories for related events Nature 520(7547) 345ndash8

Edelman GM Tononi G (2000) A Universe of Consciousness HowMatter Becomes Imagination New York Basic books

Edelman GM Gally JA (2001) Degeneracy and complexity inbiological systems Proceedings of the National Academy ofSciences of the United States of America 98 13763ndash8

Einstein A Infeld L (1938) Evolution of physics CambridgeUnited Kingdom Cambridge University Press

Etkin A Buchel C Gross JJ (2015) The neural bases of emo-tion regulation Nature Reviews Neuroscience 16(11) 693ndashthorn

Feinstein JS (2013) Lesion studies of human emotion and feel-ing Current Opinion in Neurobiology 23(3) 304ndash9

Feinstein JS Adolphs R Damasio A Tranel D (2011) Thehuman amygdala and the induction and experience of fearCurrent Biology 21(1) 34ndash8

Feinstein JS Adolphs R Tranel D (2016) A tale of survivalfrom the world of Patient SM In Amaral DG Adolphs R edi-tors Living without an Amygdala pp 1ndash38 New York Guilford

Feinstein JS Buzza C Hurlemann R et al (2013) Fear andpanic in humans with bilateral amygdala damage NatureNeuroscience 16(3) 270ndash2

Feinstein JS Rudrauf D Khalsa SS et al (2010) Bilateral lim-bic system destruction in man Journal of Clinical andExperimental Neuropsychology 32(1) 88ndash106

Feldman H Friston KJ (2010) Attention uncertainty and free-energy Frontiers in Human Neuroscience 4 215

Fernandino L Binder JR Desai RH et al (2016) Concept rep-resentation reflects multimodal abstraction A framework forembodied semantics Cerebral Cortex 26 2018ndash34

Fernandino L Humphries CJ Seidenberg MS Gross WLConant LL Binder JR (2015) Predicting brain activation pat-terns associated with individual lexical concepts based on fivesensory-motor attributes Neuropsychologia 76 17ndash26

Fields H (2004) State-dependent opioid control of pain NatureReviews Neuroscience 5(7) 565ndash75

Fields HL Margolis EB (2015) Understanding opioid rewardTrends in Neurosciences 38(4) 217ndash25

Finlay BL Uchiyama R (2015) Developmental mechanisms chan-neling cortical evolution Trends in Neurosciences 38(2) 69ndash76

Firestein S (2012) Ignorance How It Drives Science Oxford OxfordUniversity Press

Freddolino PL Tavazoie S (2012) Beyond homeostasis apredictive-dynamic framework for understanding cellular be-havior Annual Review of Cell and Developmental Biology 28363ndash84

Friston K (2010) The free-energy principle A unified brain the-ory Nature Reviews Neuroscience 11 127ndash38

Furlong TM Cole S Hamlin AS McNally GP (2010) The roleof prefrontal cortex in predictive fear learning BehavioralNeuroscience 124(5) 574

Gallivan JP Logan L Wolpert DM Flanagan JR (2016) Parallelspecification of competing sensorimotor control policies for al-ternative action options Nature Neuroscience 19(2) 320ndash6

Gelman SA Rhodes M (2012) Two-thousand years of stasisHow psychological essentialism impedes evolutionary under-standing In Rosengren KS Brem S Evans EM Sinatra Geditors Evolution Challenges Integrating Research and Practice inTeaching and Learning About Evolution pp 3ndash21Oxford OxfordUniversity Press

Gendron M Roberson D van der Vyver JM Barrett LF(2014a) Cultural relativity in perceiving emotion from vocal-izations Psychological Science 25(4) 911ndash20

Gendron M Roberson D van der Vyver JM Barrett LF(2014b) Perceptions of emotion from facial expressions are notculturally universal evidence from a remote culture Emotion14(2) 251ndash62

Gendron M Roberson D Barrett LF (2015) Cultural variatonin emotion perception is real a response to Sauter et alPsychological Science 26 357ndash9

Ghashghaei H Hilgetag C Barbas H (2007) Sequence of infor-mation processing for emotions based on the anatomic dia-logue between prefrontal cortex and amygdala NeuroImage34(3) 905ndash23

Gilbert DT (1998) Ordinary personology In Fiske STGardner L editors The Handbook of Social Psychology Vol 1 and2 pp 89ndash150 New York NY McGraw-Hill

Goodkind M Eickhoff SB Oathes DJ et al (2015)Identification of a common neurobiological substrate for men-tal illness JAMA Psychiatry 72(4) 305ndash15

Gross JJ Barrett LF (2011) Emotion generation and emotionregulation one or two depends on your point of view EmotionReview 3(1) 8ndash16

Gross CT Canteras NS (2012) The many paths to fear NatureReviews Neuroscience 13(9) 651ndash8

L F Barrett | 19

Gross JJ (2015) Emotion regulation current status and futureprospects Psychological Inquiry 26(1) 1ndash26

Guillory SA Bujarski KA (2014) Exploring emotions using in-vasive methods review of 60 years of human intracranial elec-trophysiology Social Cognitive and Affective Neuroscience 9(12)1880ndash9

Guitart-Masip M Duzel E Dolan R Dayan P (2014) Actionvserus valence in decision making Trends in Cognitive Sciences18 194ndash202

Haber SN Behrens TE (2014) The neural network underlyingincentive-based learning implications for interpreting circuitdisruptions in psychiatric disorders Neuron 83(5) 1019ndash39

Hampton AN Adolphs R Tyszka JM OrsquoDoherty JP (2007)Contributions of the amygdala to reward expectancy and choicesignals in human prefrontal cortex Neuron 55(4) 545ndash55

Harris JJ Jolivet R Attwell D (2012) Synaptic energy use andsupply Neuron 75(5) 762ndash77

Hassabis D Maguire EA (2009) The construction system ofthe brain Philosophical Transactions of the Royal Society BBiological Sciences 364(1521) 1263ndash71

Hasson U Chen J Honey CJ (2015) Hierarchical processmemory memory as an integral component of informationprocessing Trends in Cognitive Sciences 19(6) 304ndash13

Herry C Johansen JP (2014) Encoding of fear learning andmemory in distributed neuronal circuits Nature Neuroscience17(12) 1644ndash54

Hodgkin AL Huxley AF (1952) A quantitative description ofmembrane current and its application to conduction and exci-tation in nerve Journal of Physiology 117(4) 500ndash44

Hohwy J (2013) The Predictive Mind Oxford OUP OxfordHunter RG McEwen BS (2013) Stress and anxiety across the

lifespan structural plasticity and epigenetic regulationEpigenomics 5(2)177ndash94

Hurlemann R Wagner M Hawellek B et al (2007) Amygdala con-trol of emotion-induced forgetting and remembering evidencefrom Urbach-Wiethe disease Neuropsychologia 45(5)877ndash84

Iwata J LeDoux JE (1988) Dissociation of associative and non-associative concomitants of classical fear conditioning in thefreely behaving rat Behavioral Neuroscience 102(1)66ndash76

James W (18902007) The Principles of Psychology Vol 1 NewYork Dover

John YJ Zikopoulos B Bullock D Barbas H (2016) The emo-tional gatekeeper A computational model of attention selec-tion and suppression through the pathway from the amygdalato the inhibitory thalamic reticular nucleus PLOSComputational Biology 12(2)e1004722

Karmiloff-Smith A (2009) Nativism versus neuroconstructi-vism rethinking the study of developmental disordersDevelopmental Psychology 45(1)56ndash63

Kassam KS Markey AR Cherkassky VL Loewenstein GJust MA (2013) Identifying emotions on the basis of neuralactivation PLoS One 8(6) e66032

Kleckner IR Zhang J Touroutoglou A et al (in press)Evidence for a large-scale brain system supporting allostasisand interoception in humans Nature Human Behavior doihttpsdoiorg101101098970

Kober H Barrett LF Joseph J Bliss-Moreau E Lindquist KWager TD (2008) Functional grouping and cortical-subcorticalinteractions in emotion a meta-analysis of neuroimaging stud-ies NeuroImage 42(2) 998ndash1031

Kok P Bains LJ van Mourik T Norris DG de Lange FP(2016) Selective activation of the deep layers of the human pri-mary visual cortex by top-down feedback Current Biology26(3) 371ndash6

Kragel PA LaBar KS (2013) Multivariate pattern classificationreveals autonomic and experiential representations of discreteemotions Emotion 13(4) 681ndash90

Kragel PA LaBar KS (2015) Multivariate neural biomarkers ofemotional states are categorically distinct Social Cognitive andAffective Neuroscience 10(11) 1437ndash48

Kragel PA LaBar KS (2016) Decoding the nature of emotion inthe brain Trends in Cognitive Sciences 20(6)444ndash55

Kuppens P Tuerlinckx F Russell JA Barrett LF (2013) Therelation between valence and arousal in subjective experiencePsychological Bulletin 139(4) 917ndash40

Laxminarayan S Tadmor G Diamond SG Miller EFranceschini MA Brooks DH (2011) Modeling habituationin rat EEG-evoked responses via a neural mass model withfeedback Biological Cybernetics 105(5ndash6) 371ndash97

Lazarus RS (1991) Emotion and Adaptation New York NYOxford University Press

LeDoux JE (2012) Rethinking the emotional brain Neuron73(4)653ndash76

Li SSY McNally GP (2014) The conditions that promote fearlearning prediction error and Pavlovian fear conditioningNeurobiology of Learning and Memory 108 14ndash21

Liang M Mouraux A Hu L Iannetti G (2013) Primary sensorycortices contain distinguishable spatial patterns of activity foreach sense Nature Communications 4

Lindquist KA Wager TD Kober H Bliss-Moreau E BarrettLF (2012) The brain basis of emotion a meta-analytic reviewBehavioral and Brain Sciences 35(3)121ndash43

Lochmann T Deneve S (2011) Neural processing as causal in-ference Current Opinion in Neurobiology 21(5)774ndash81

Makeig S Debener S Onton J Delorme A (2004) Miningevent-related brain dynamics Trends in Cognitive Sciences8(5)204ndash10

Makeig S Westerfield M Jung TP et al (2002) Dynamic brainsources of visual evoked responses Science 295(5555) 690ndash4

Marder E Taylor AL (2011) Multiple models to capture thevariability in biological neurons and networks NatureNeuroscience 14 133ndash8

Mareschal D Johnson MH Sirois S Spratling M ThomasMS Westermann G (2007) Neuroconstructivism-I How theBrain Constructs Cognition Oxford Oxford University Press

Mason WA Capitanio JP Machado CJ Mendoza SPAmaral DG (2006) Amygdalectomy and responsiveness tonovelty in rhesus monkeys (Macaca mulatta) generality andindividual consistency of effects Emotion 6(1) 73

Mayr E (2004) What Makes Biology Unique Considerations on theAutonomy of a Scientific Discipline Cambridge CambridgeUniversity Press

Mazaheri A Jensen O (2010) Rhythmic pulsing linking on-going brain activity with evoked responses Frontiers in HumanNeuroscience 4 177

McEwen BS Bowles NP Gray JD et al (2015) Mechanisms ofstress in the brain Nature Neuroscience 18(10) 1353ndash63

McGaugh JL (2016) Consolidating memories Annual Review ofPsychology 66 1ndash24

McIntosh AR (2004) Contexts and catalysts a resolution of thelocalization and integration of function in the brainNeuroinformatics 2(2) 175ndash82

McNally GP Johansen JP Blair HT (2011) Placing predictioninto the fear circuit Trends in Neurosciences 34(6) 283ndash92

McNorgan C (2012) A meta-analytic review of multisensory im-agery identifies the neural correlates of modality-specific andmodality-general imagery Frontiers in Human Neuroscience 6Article 285


Mesulam MM (1998) From sensation to cognition Brain 121(Pt6) 1013ndash52

Mesulam MM (2002) The human frontal lobes Transcendingthe default mode through contingent encoding In Stuss DTKnight RT editors Principles of Frontal Lobe Function pp 8ndash30New York NY Oxford University Press

Meyer K Damasio A (2009) Convergence and divergence in aneural architecture for recognition and memory Trends inCognitive Sciences 32 376ndash82

Mihov Y Kendrick KM Becker B et al (2013) Mirroring fearin the absence of a functional amygdala Biological Psychiatry73(7)e9ndash11

Moran RJ Campo P Symmonds M Stephan KE Dolan RJFriston KJ (2013) Free energy precision and learning the roleof cholinergic neuromodulation The Journal of Neuroscience33(19)8227ndash36

Murphy G (2002) The Big Book of Concepts Cambridge MA MITpress

Ongur D Ferry AT Price JL (2003) Architectonic subdivisionof the human orbital and medial prefrontal cortex Journal ofComparative Neurology 460(3) 425ndash49

Oosterwijk S Lindquist KA Adebayo M Barrett LF (2015)The neural representation of typical and atypical experiencesof negative images comparing fear disgust and morbid fas-cination Social Cognitive and Affective Neuroscience 11 11ndash22

Oosterwijk S Lindquist KA Anderson E Dautoff RMoriguchi Y Barrett LF (2012) States of mind Emotionsbody feelings and thoughts share distributed neural net-works NeuroImage 62(3) 2110ndash28

Oosterwijk S Mackey S Wilson-Mendenhall C WinkielmanP Paulus MP (2015) Concepts in context processing mentalstate concepts with internal or external focus involves differ-ent neural systems Social Neuroscience 10(3) 294ndash307

Pavlenko A (2014) The Bilingual Mind And What It Tells Us aboutLanguage and Thought Cambridge Cambridge University Press

Peelen MV Atkinson AP Vuilleumier P (2010) Supramodalrepresentations of perceived emotions in the human brainJournal of Neuroscience 30(30) 10127ndash34

Pezzulo G Rigoli F Friston K (2015) Active Inference homeo-static regulation and adaptive behavioural control Progress inNeurobiology 134 17ndash35

Posner MI Keele SW (1968) On the genesis of abstract ideasJournal of Experimental Psychology 77(3p1) 353ndash63

Power JD Cohen AL Nelson SM et al (2011) Functional net-work organization of the human brain Neuron 72(4)665ndash78

Pulvermuller F (2013) How neurons make meaning brainmechanisms for embodied and abstract-symbolic semanticsTrends in Cognitive Sciences 17(9)458ndash70

Qian C Di X (2011) Phase or amplitude The relationship be-tween ongoing and evoked neural activity Journal ofNeuroscience 31(29)10425ndash6

Raichle ME (2010) Two views of brain function Trends inCognitive Sciences 14(4)180ndash90

Rao RPN Ballard DH (1999) Predictive coding in the visualcortex a functional interpretation of some extra-classical re-ceptive-field effects Nature Neuroscience 2 79ndash87

Raz G Touroutoglou A Gilam G et al (2016) Functional con-nectivity dynamics during film viewing reveal common net-works for different emotional experiences Cognitive Affectiveand Behavioral Neuroscience 16 709ndash23

Rigotti M Barak O Warden MR et al (2013) The importanceof mixed selectivity in complex cognitive tasks Nature497(7451)585ndash90

Robbins J Rumsey A (2008) Introduction Cultural and linguis-tic anthropology and the opacity of other mindsAnthropological Quarterly 81(2)407ndash20

Roseman IJ (2011) Emotional behaviors emotivational goalsemotion strategies Multiple levels of organization integratevariable and consistent responses Emotion Review 3 1ndash10

Russell JA (1991) Culture and the Categorization of EmotionsPsychological Bulletin 110(3)426ndash50

Saarimaki H Gotsopoulos A Jaaskelainen IP et al (2016)Discrete Neural Signatures of Basic Emotions Cerebral Cortex26(6)2563ndash73

Salamone JD Correa M (2012) The mysterious motivationalfunctions of mesolimbic dopamine Neuron 76 470ndash85

Sartorius N Kaelber CT Cooper JE et al (1993) Progress to-ward achieving a common language in psychiatry resultsfrom the field trial of the clinical guidelines accompanying theWHO classification of mental and behavioral disorders in ICD-10 Archives of General Psychiatry 50(2)115ndash24

Satpute AB Wilson-Mendenhall CD Kleckner IR BarrettLF (2015) Emotional experience In Toga AW editor BrainMapping An Encyclopedic Reference pp 65ndash72 Boston MAElsevier

Sayers BM Beagley HA Henshall WR (1974) Themechanism of auditory evoked EEG responses Nature247 481ndash3

Schacter DL Addis DR Buckner RL (2007) Rememberingthe past to imagine the future the prospective brain NatureReviews Neuroscience 8(9) 657ndash61

Scheeringa R Mazaheri A Bojak I Norris DG KleinschmidtA (2011) Modulation of visually evoked cortical FMRI re-sponses by phase of ongoing occipital alpha oscillationsJournal of Neuroscience 31(10) 3813ndash20

Scherer KR (2009) Emotions are emergent processes they re-quire a dynamic computational architecture PhilosophicalTransactions of the Royal Society B Biological Sciences 364 (1535)3459ndash74

Schmahmann JD (2010) The role of the cerebellum incognition and emotion personal reflections since 1982 onthe dysmetria of thought hypothesis and its historicalevolution from theory to therapy Neuropsychology Review20(3)236ndash60

Schmahmann JD Pandya DN (1997) The cerebrocere-bellar system International Review of Neurobiology 4131ndash60

Schultz W (2016) Dopamine reward prediction-error signallinga two-component response Nature Reviews Neuroscience 17(3)183ndash95

Searle JR (1992) The Rediscovery of the Mind Cambridge MITpress

Searle JR (2004) Mind A Brief Introduction Oxford OxfordUniversity Press

Sepulcre J Sabuncu MR Yeo TB Liu H Johnson KA (2012)Stepwise connectivity of the modal cortex reveals the multi-modal organization of the human brain Journal of Neuroscience32(31)10649ndash61

Seth AK (2013) Interoceptive inference emotion and theembodied self Trends in Cognitiive Sciences 17(11) 565ndash73

Seth AK Suzuki K Critchley HD (2012) An interoceptive pre-dictive coding model of conscious presence Frontiers inPsychology 2 395

Seth A K Friston K J (2016) Active interoceptive inferenceand the emotional brain Philosophical Transactions of the RoyalSociety B doi 101098rstb20160007

L F Barrett | 21

Sharpe MJ Killcross S (2015) The prelimbic cortex directs at-tention toward predictive cues during fear learning Learningand Memory 22(6) 289ndash93

Shipp S Adams RA Friston KJ (2013) Reflections on agranu-lar architecture predictive coding in the motor cortex Trendsin Neurosciences 36(12) 706ndash16

Si M Marsella S Pynadath D (2010) Modeling appraisal inTheory of Mind Reasoning Journal of Autonomous Agents andMulti-Agent Systems 20 14ndash31

Skerry AE Saxe R (2015) Neural representations of emotionare organized around abstract event features Current Biology25(15) 1945ndash54

Sloan EK Capitanio JP Cole SW (2008) Stress-induced re-modeling of lymphoid innervation Brain Behavior andImmunity 22(1) 15ndash21

Sloan EK Capitanio JP Tarara RP Mendoza SP MasonWA Cole SW (2007) Social stress enhances sympathetic in-nervation of primate lymph nodes mechanisms and implica-tions for viral pathogenesis The Journal of Neuroscience 27(33)8857ndash65

Spillmann L Dresp-Langley B Tseng CH (2015) Beyond theclassical receptive field The effect of contextual stimuliJournal Vision 15(9) 7

Sporns O (2011) Networks of the Brain Cambridge MIT pressStephens CL Christie IC Friedman BH (2010) Autonomic

specificity of basic emotions evidence from pattern classifica-tion and cluster analysis Biological Psychology 84(3) 463ndash73

Sterling P (2012) Allostasis a model of predictive regulationPhysiology and Behavior 106(1) 5ndash15

Sterling P Laughlin S (2015) Principles of Neural DesignCambridge MIT Press

Stokes MG Kusunoki M Sigala N Nili H Gaffan DDuncan J (2013) Dynamic coding for cognitive control in pre-frontal cortex Neuron 78(2) 364ndash75

Strick PL Dum RP Fiez JA (2009) Cerebellum and NonmotorFunction Annual Review of Neuroscience 32 413ndash34

Swanson LW (2000) Cerebral hemisphere regulation of moti-vated behavior Brain Research 886(1ndash2) 113ndash64

Swanson LW (2012) Brain Architecture Understanding the BasicPlan Oxford Oxford University Press

Terburg D Morgan B Montoya E et al (2012) Hypervigilancefor fear after basolateral amygdala damage in humansTranslational Psychiatry 2(5) e115

Tononi G Sporns O Edelman GM (1999) Measures of degen-eracy and redundancy in biological networks Proceedings of theNational Academy of Sciences of the United States of America 96(6)3257ndash62

Touroutoglou A Hollenbeck M Dickerson BC Barrett LF(2012) Dissociable large-scale networks anchored in the rightanterior insula subserve affective experience and attentionNeuroImage

Touroutoglou A Lindquist KA Dickerson BC Barrett LF(2015) Intrinsic connectivity in the human brain does not re-veal networks for lsquobasicrsquo emotions Social Cognitive and AffectiveNeuroscience 10(9) 1257ndash65

Tovote P Fadok JP Luthi A (2015) Neuronal circuits for fearand anxiety Nature Reviews Neuroscience 16(6) 317ndash31

Tracy JL Randles D (2011) Four models of basic emotions areview of Ekman and Cordaro Izard Levenson and Pankseppand Watt Emotion Review 3(4) 397ndash405

Tsuchiya N Moradi F Felsen C Yamazaki M Adolphs R(2009) Intact rapid detection of fearful faces in the absence ofthe amygdala Nature Neuroscience 12(10) 1224ndash5

Turkheimer E Pettersson E Horn EE (2014) A phenotypicnull hypothesis for the genetics of personality Annual Reviewof Psychology 65 515ndash40

Uddin LQ (2015) Salience procesing and insular cortical func-tion and dysfunction Neuron 16 55ndash61

Ullsperger M Danielmeier C Jocham G (2014)Neurophysiology of performance monitoring and adaptive be-havior Physiological Reviews 94 35ndash79

van den Heuvel MP Kahn RS Goni J Sporns O (2012) High-cost high-capacity backbone for global brain communicationProceedings of the National Academy of Sciences of the United Statesof America 109(28) 11372ndash7

van den Heuvel MP Sporns O (2011) Rich-club organization ofthe human connectome The Journal of Neuroscience 31(44)15775ndash86

van den Heuvel MP Sporns O (2013) An anatomical substratefor integration among functional networks in human cortexThe Journal of Neuroscience 33(36) 14489ndash500

van den Heuvel MP Sporns O (2013) Network hubs in thehuman brain Trends in Cognitive Sciences 17 683ndash96

Voorspoels W Vanpaemel W Storms G (2011) A formalideal-based account of typicality Psychonomic Bulletin andReview 18(5) 1006ndash14

Vytal K Hamann S (2010) Neuroimaging support for dis-crete neural correlates of basic emotions a voxel-basedmeta-analysis Journal of Cognitive Neuroscience 22(12)2864ndash85

Wager TD Kang J Johnson TD Nichols TE Satpute ABBarrett LF (2015) A Bayesian model of category-specific emo-tional brain responses PLOS Computational Biology 11(4)e1004066

Westermann G Mareschal D Johnson MH Sirois SSpratling MW Thomas MSC (2007) NeuroconstructivismDevelopmental Science 10(1) 75ndash83

Whalen PJ (1998) Fear vigilance and ambiguity Initial neuroi-maging studies of the human amygdala Current Directions inPsychological Science 7(6) 177ndash88

Whitacre J Bender A (2010) Degeneracy a design principle forachieving robustness and evolvability Journal of TheoreticalBiology 263(1) 143ndash53

Whitacre JM (2010) Degeneracy a link between evolvabilityrobustness and complexity in biological systems TheoreticalBiology and Medical Modelling 7(6) 1ndash17

Wierzbicka A (2014) Imprisoned in English The Hazards ofEnglish as a Default Language Oxford Oxford UniversityPress

Wilson CR Gaffan D Browning PG Baxter MG (2010)Functional localization within the prefrontal cortex missingthe forest for the trees Trends in Neurosciences 33(12)533ndash40

Wilson-Mendenhall C Barrett LF Barsalou LW (2013)Neural evidence that human emotions share core affectiveproperties Psychological Science 24 947ndash56

Wilson-Mendenhall CD Barrett LF Barsalou LW (2015)Variety in emotional life within-category typicality of emo-tional experiences is associated with neural activity in large-scale brain networks Social Cognitive and Affective Neuroscience10(1)62ndash71 doi101093scannsu037

Wilson-Mendenhall CD Barrett LF Simmons WK BarsalouLW (2011) Grounding emotion in situated conceptualizationNeuropsychologia 49 1105ndash27

Woodworth RS Sherrington CS (1904) A pseudaffective re-flex and its spinal path Journal of Physiology 31(3ndash4) 234ndash43


Wundt W (1897) Outlines of Psychology In Judd CH (Trans)Leipzig Wilhelm Engelmann

Xu F Kushnir T (2013) Infants are rational constructivistlearners Current Directions in Psychological Science 22(1) 28ndash32

Zikopoulos B Barbas H (2006) Prefrontal projections tothe thalamic reticular nucleus form a unique circuit for

attentional mechanisms The Journal of Neuroscience 26(28)7348ndash61

Zikopoulos B Barbas H (2012) Pathways for emotionsand attention converge on the thalamic reticular nu-cleus in primates Journal of Neuroscience 32(15)5338ndash50

L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24

findings on the brain basis of emotion through the theoryrsquoslens In the process I offer a series of novel hypotheses aboutwhat emotions are and how they work

The biological backgroundWhat is a brain

A brain is a network of billions of communicating neurons bathedin chemicals called neurotransmitters which permit neurons to

pass information to one another (Doya 2008 Bargmann 2012) Thefiring of a single neuron (or a small population of neurons) repre-sents the presence or absence of some feature at a moment intime (Deneve 2008 Deneve and Jardri 2016) However a given neu-ron (or group of neurons) represents different features from mo-ment to moment (eg Stokes et al 2013 Spillmann et al 2015)because many neurons synapse onto one (many-to-one connectiv-ity) and a neuronrsquos receptive field depends on the information itreceives (ie depends on its neural context in the momentMcIntosh 2004) Conversely one neuron also synapses on many

Fig 1 The classical view of emotion The classical view of emotion includes basic emotion theories (eg for a review see Tracy and Randles 2011) causal appraisal theo-

ries (eg Scherer 2009 Roseman 2011) and theories of emotion that rely on black-box functionalism (Davis 1992 Anderson and Adolphs 2014) Each emotion faculty is

assumed to have its own innate lsquoessencersquo that distinguishes it from all other emotions This might be a Lockean essence (an underlying causal mechanism that all in-

stances of an emotion category share making them that kind of emotion and not some other kind of emotion depicted by the circles in the figure) Lockean essences

might be a biological such as a set of dedicated neurons or psychological such as a set of evaluative mechanisms called lsquoappraisalsrsquo An emotion category is usually

assumed to have a Platonic essence [a physical fingerprint that instances of that emotion share but that other emotions do not such a set of facial movements (an lsquoex-

pressionrsquo) a pattern of autonomic nervous system activity andor a pattern of appraisals] Of course no one is expecting complete invariance but it is assumed that in-

stances of a category are similar enough to be easily diagnosed as the same emotion using objective (perceiver-independent) measures alone (A) is adapted from Davis

(1992) (B) is adapted Anderson and Adolphs (2014) (C) is adapted from Barrett (2006a) which reviews the growing evidence that contracts the classical view of emotion







What is a brain for






L F Barrett | 3









































L F Barrett | 5




































L F Barrett | 7




























L F Barrett | 9

A

B

C

D

























L F Barrett | 11





























L F Barrett | 13































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24






What is a brain for






L F Barrett | 3









































L F Barrett | 5




































L F Barrett | 7




























L F Barrett | 9

A

B

C

D

























L F Barrett | 11





























L F Barrett | 13































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24









































L F Barrett | 5




































L F Barrett | 7




























L F Barrett | 9

A

B

C

D

























L F Barrett | 11





























L F Barrett | 13































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24




































L F Barrett | 7




























L F Barrett | 9

A

B

C

D

























L F Barrett | 11





























L F Barrett | 13































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24




























L F Barrett | 9

A

B

C

D

























L F Barrett | 11





























L F Barrett | 13































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24

A

B

C

D

























L F Barrett | 11





























L F Barrett | 13































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24










L F Barrett | 11





























L F Barrett | 13































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24





























L F Barrett | 13































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24































(Tovote et al 2015)











(Raz et al 2016)

















L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24













L F Barrett | 15









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24









Funding



Glossary














Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24







Acknowledgements








































L F Barrett | 17

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24

























































































L F Barrett | 19























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24























































































L F Barrett | 21

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24

















































L F Barrett | 23

nsw154-fn1

nsw154-fn2

nsw154-fn3

nsw154-fn4

nsw154-fn5

nsw154-fn6

nsw154-fn7

nsw154-fn8

nsw154-fn9

nsw154-fn10

nsw154-fn11

nsw154-fn12

nsw154-fn13

nsw154-fn14

nsw154-fn15

nsw154-fn16

nsw154-fn17

nsw154-fn18

nsw154-fn19

nsw154-fn20

nsw154-fn21

nsw154-fn22

nsw154-fn23

nsw154-fn24

the theory of constructed emotion: an active inference ... · pdf filethe theory of...

Documents