PersPectivesBox 1 | <strong>The</strong> co-activation of the AIC <strong>and</strong> the ACCIt is underst<strong>and</strong>ably mystifying that a region ofthe ventrolateral prefrontal cortex (the <strong>anterior</strong><strong>insula</strong>r cortex (AIC)) <strong>and</strong> a region of the medialprefrontal cortex (the <strong>anterior</strong> cingulate cortex(ACC)) are co-active in so many behaviours,because such widely separated regions in thecortex generally have distinct roles. <strong>The</strong><strong>insula</strong>r cortex was long regarded simply as avisceral sensory region, based on findings byPenfield, Mesulam <strong>and</strong> Saper 84–86 , whereas themedial prefrontal cortex has been associatedwith conflict responses, impulsive behaviour<strong>and</strong> autonomic activity. In an earlier article Isuggested that the <strong>insula</strong> <strong>and</strong> the ACC beregarded as limbic sensory <strong>and</strong> motor corticesthat respectively engender the <strong>feel</strong>ing <strong>and</strong> themotivation (agency) that constitute anyemotion 1 . This suggestion was based on thedual lamina I spinothalamocortical projectionto both the <strong>insula</strong> <strong>and</strong> the ACC, theco-activation of these areas in virtually allstudies of emotion, their respective descendingprojections to sensory (parabrachial nucleus)<strong>and</strong> motor (periaqueductal grey) brainstemregions, the overall anatomical organization ofthe frontal cortex into sensory <strong>and</strong> motornetworks, <strong>and</strong> the evolutionarily ancient limbicrole of the cingulate cortex in integratedbehavioural control. I regarded the AIC as theprobable site for <strong>awareness</strong> on the basis of itsafferent representation of the ‘<strong>feel</strong>ings’ fromthe body, <strong>and</strong> the ACC as the probable sitefor the initiation of behaviours. A recentreview 87 offered support for this view <strong>and</strong> anexplanation for the anatomical separation ofthe <strong>insula</strong> <strong>and</strong> the ACC. <strong>The</strong> ACC evolved firstas a motor-control region aligned with thesensory integration, in the hippocampus <strong>and</strong>the amygdala, of olfactory-guided groupbehaviour in mammals. <strong>The</strong> <strong>insula</strong> evolved laterfor cortical processing of homeostatic sensoryactivity in the individual animal. <strong>The</strong> tworegions naturally became linked for integrativeautonomic control, <strong>and</strong> in mammalian evolutionthe <strong>insula</strong> grew as limbic behavioural activitybecame aligned more with autonomic activitythan with olfactory activity. This anatomicalperspective 87 is illustrated in the ventral view ofthe brain in the figure, which reveals thecommon relationship of these structures to theolfactory epithelium. Figure is reproduced, withpermission, from ReF. 87 © (1999) Elsevier.aCingulate sulcusRhinal sulcusbInsulaOlfactory cortexEntorhinal areaHippocampusCollateral sulcusLaterobasal cortical amygdalaOlfactorybulbOlfactorypeduncleOlfactory tractAnterior olfactory nucleusDiagonal b<strong>and</strong>Basolateral amygdalaHippocampusRisk, uncertainty <strong>and</strong> anticipation. Of thearticles listed in Supplementary informationS1 (table), I note particularly one report that‘emotional value’ in the immediate present,as specified in a secondorder temporaldifferencemodel of learning, correlatedselectively with activity in the right AIC 24 ;another study (in which this learning modelwas adapted for neuroeconomic studies)reported that both the riskprediction signals<strong>and</strong> the riskprediction error signals specifiedin the model were present only in thebilateral AIC 25 . No activation was observed inthe ACC in this study. Activation in the AICis also correlated with <strong>feel</strong>ings of anticipatedvalue during purchase <strong>and</strong> sales decisions 26 .Visual <strong>and</strong> auditory <strong>awareness</strong> of themoment. Several reports that associated activationin the AIC with <strong>awareness</strong> of sensoryNature Reviews | Neurosciencebistable percepts must be explicitly notedhere 27–29 , but I highlight three additionalstudies. <strong>The</strong> first is a PET study which foundthat the right AIC <strong>and</strong> the ACC are sensitiveto crossmodal sensory time synchronization<strong>and</strong> display a graded response to amismatch in timing between auditory <strong>and</strong>visual stimuli that should normally be synchronous(for example, a speaking mouth) 30 .<strong>The</strong> second is an fMRI study that examined62 | JANuARy 2009 | VOLuME 10 www.nature.com/reviews/neuro© 2009 Macmillan Publishers Limited. All rights reserved
PersPectivesBox 2 | VEns <strong>and</strong> <strong>awareness</strong>An extraordinary morphological characteristic of the <strong>anterior</strong> <strong>insula</strong>r cortex (AIC) <strong>and</strong> the <strong>anterior</strong>cingulate cortex (ACC) in hominoid primates is the unique concentration of clusters of largespindle-shaped neurons among the pyramidal neurons in layer 5, called von Economo neurons(VENs) after an early neuroanatomist 88,89 . <strong>The</strong>ir connections are not k<strong>now</strong>n, but I have proposed 78that VENs are the substrate for fast interconnections between the physically separated advancedlimbic sensory (the AIC) <strong>and</strong> motor (the ACC) cortices. Analogous to the tight interconnectionsbetween the contiguous somatosensory <strong>and</strong> motor cortices (so-called U-fibres) needed for manualdexterity (for example, for playing a musical instrument), the VENs might enable fast, highlyintegrated representations of emotional moments <strong>and</strong> behaviours. <strong>The</strong>y may underlie the jointactivity in the AIC <strong>and</strong> the ACC reported in most studies. <strong>The</strong> loss of emotional <strong>awareness</strong> <strong>and</strong>self-conscious behaviours in patients with frontotemporal dementia that correlates with thedegeneration of VENs directly supports this notion 75,76,90,91 .Hof, Allman <strong>and</strong> colleagues reported that many VENs are present in aged <strong>human</strong>s, butprogressively fewer are found in infants, gorillas, bonobos <strong>and</strong> chimpanzees, <strong>and</strong> they are notpresent at all in macaque monkeys 88,89 . This clear phylogenetic progression parallels the results ofthe mirror test for self-<strong>awareness</strong> 2,55 . VENs have recently been reported in elephants 92 <strong>and</strong>whales 93 . <strong>The</strong> possibility that this implies sentience in these animals received support recentlywhen elephants were reported to pass the mirror test 94 . Interestingly, there are reports thatelephants sing or make music communally (as captured in a film on the National Geographicwebsite), which would be consistent with the present model for <strong>awareness</strong>, in which music (viewedas a rhythmic temporal progression of emotionally laden moments) is an emergent property of<strong>awareness</strong>. By contrast, although most people <strong>feel</strong> that particular <strong>do</strong>gs <strong>and</strong> cats have some senseof <strong>awareness</strong>, the cortex of these animals receives an integrated brainstem homeostatic pathway<strong>and</strong> only a primordial homologue of the ascending lamina I pathway that underlies the emergenceof interoceptive re-representations in the AIC of <strong>human</strong>oid primates 1 . Whether that is a sufficientbasis for sentience is unresolved: cats <strong>and</strong> <strong>do</strong>gs <strong>do</strong> not pass the mirror test, <strong>and</strong> a lesion of thepresumptive homologous pathway in cats <strong>do</strong>es not produce the same devastating effects oninteroception that it produces in <strong>human</strong>s 1 . Rats <strong>and</strong> lizards apparently <strong>do</strong> not have a homologousanatomical substrate at all 1 .“inspection time” using a briefly displayedasymmetric visual test stimulus (a onesidedfork) followed immediately by an ambiguousstimulus (socalled backward masking) 31 .<strong>The</strong> authors reported that subjects’ performancein detecting the asymmetry decreasedprogressively from 100% to chance levels forpresentation times shorter than 150 ms,yet activation in the AIC <strong>and</strong> the ACCselectively <strong>and</strong> progressively increased withshorter presentation times. <strong>The</strong> authorsinferred from these data an “effortrelatedprocess” that guides goaldirected attention,which they related to psychometric intelligence;however, this observation can also beinterpreted as evidence for a role for the AIC<strong>and</strong> ACC in heightened <strong>awareness</strong> of theimmediate moment. <strong>The</strong> third study 32 usedthe attentionalblink paradigm, in which asecond target cannot be perceived if it occurstoo soon after a primary target in a rapidseries of visual stimuli. <strong>The</strong> authors reportedjoint activation of the AIC/IFG <strong>and</strong> theACC (plus activation of the DLPFC <strong>and</strong>parietal regions) when the second target wascorrectly detected at the shortest intervals(100–200 ms) but not when it was not perceived.<strong>The</strong>y suggested that the AIC/IFG <strong>and</strong>the ACC might contain a “conceptual shorttermmemory buffer prone to decay <strong>and</strong>replacement by other stimuli” that receivesprocessed input from a parietal network<strong>and</strong> requires a finite period of time totransfer information into working memory(in the DLPFC).Time perception. Joint activation of theAIC <strong>and</strong> the ACC or nearby areas has beenreported in many studies of ‘mental timekeeping’<strong>and</strong> interval estimation across therange of seconds to subseconds, but littleexplanation has been given for this activity(for example, see ReF. 33). In a recent report,task difficulty was manipulated in orderto isolate time estimation from other taskrelatedcognitive dem<strong>and</strong>s 34 . <strong>The</strong> authorsfound three small regions that seem to becrucial for time perception: one in the <strong>do</strong>rsalputamen (bilaterally), another in the leftinferior parietal cortex <strong>and</strong> another at thejunction of the AIC <strong>and</strong> the IFG bilaterally(with no ACC activation). <strong>The</strong>y suggestedthat the AIC/IFG focus must be “of centralimportance” in time perception.Attention. Activation of the AIC (<strong>and</strong> theACC) is reported by most studies of goaldirectedattention, but these studies oftenlack comments regarding the role of thisactivation. One influential model for theneural correlates of the executive control ofattention <strong>do</strong>es not include the <strong>insula</strong> (forexample, see ReF. 35), <strong>and</strong> some have arguedthat attention <strong>and</strong> consciousness must bedifferent processes 36 . Nevertheless, tworecent studies of attention described activationin the AIC/IFG 37,38 . <strong>The</strong> authors of onestudy reported decreased activation in theright IFG, ACC <strong>and</strong> middle frontal gyrus(DLPFC) just before lapses of attention duringa monotonous selectiveattention task(in which lapses of attention were markedby increased reaction times), <strong>and</strong> theyobserved increased activation after suchlapses in approximately the same regions(bilateral AIC, ACC, DLPFC, occipital <strong>and</strong>parietal visual regions), possibly correspondingto renewed attention 37 . <strong>The</strong>y suggestedthat the AIC/IFG is involved in the stimulustriggeredreorienting of attention <strong>and</strong> thatthe ACC is involved in the detection <strong>and</strong>/orresolution of processing conflicts; however,if the AIC/IFG <strong>and</strong> the ACC are regardedas complementary limbic sensory <strong>and</strong>motor regions, respectively (BOX 1), then thisresponse profile would also be consistentwith the interpretation that target <strong>awareness</strong>is engendered in the AIC/IFG <strong>and</strong> control ofdirected effort is engendered in the ACC. Inanother study, a wellpractised behaviouraltask was used to encourage mindw<strong>and</strong>ering(‘stimulusindependent thought’), <strong>and</strong> theauthors found activation associated withselfreported mindw<strong>and</strong>ering periodsin several regions, particularly the <strong>insula</strong>(middle <strong>and</strong> posterior) <strong>and</strong> the ACC 38 .<strong>The</strong>y interpreted the <strong>insula</strong>r activity withreference to interoception <strong>and</strong> emotional<strong>awareness</strong>.Perceptual decision making. I highlightthree seminal reports in this category. First,Ploran et al. 39 tracked brain activation insubjects watching a screen on which animage was slowly being revealed. <strong>The</strong>yfound a gradual increase in activation inbrain regions that are involved in objectidentification, but a sudden burst of activityin the AIC <strong>and</strong> the ACC at the momentof recognition (that is, coinciding with the<strong>awareness</strong> of the percept itself). Second,Thielscher <strong>and</strong> Pessoa 40 examined a twochoiceperceptual task using a graded seriesof morphed emotional faces (an experimentallygenerated bistate percept). <strong>The</strong>y foundan invertedushaped correlation betweenperceptual choice <strong>and</strong> both reaction time(which represented the decisionmakingprocess) <strong>and</strong> bilateral AIC/IFG (R>L) <strong>and</strong>ACC activation. <strong>The</strong>y concluded that theAIC “may have been important in theactual generation of the perceptual choice”.Finally, Kikyo et al. 41 examined theNATuRE REVIEwS | NeuroscieNce VOLuME 10 | JANuARy 2009 | 63© 2009 Macmillan Publishers Limited. All rights reserved