Emotion, Cognition, and Mental State Representation in ... · NE33CH09-Salzman ARI 14 May 2010...

NE33CH09-Salzman ARI 14 May 2010 17:24

Emotion, Cognition,and Mental StateRepresentation in Amygdalaand Prefrontal CortexC. Daniel Salzman1,2,3,4,5,6 and Stefano Fusi11Department of Neuroscience, 2Department of Psychiatry, 3W.M. Keck Center on BrainPlasticity and Cognition, 4Kavli Institute for Brain Sciences, 5Mahoney Center for Brainand Behavior, Columbia University, New York, NY 10032; email: [email protected],[email protected], 6New York State Psychiatric Institute, New York, NY 10032

Annu. Rev. Neurosci. 2010. 33:173–202

First published online as a Review in Advance onMarch 23, 2010

The Annual Review of Neuroscience is online atneuro.annualreviews.org

This article’s doi:10.1146/annurev.neuro.051508.135256

Copyright c© 2010 by Annual Reviews.All rights reserved

0147-006X/10/0721-0173$20.00

Key Words

neurophysiology, orbitofrontal cortex, value, reward, aversive,reinforcement learning

Abstract

Neuroscientists have often described cognition and emotion as sepa-rable processes implemented by different regions of the brain, such asthe amygdala for emotion and the prefrontal cortex for cognition. Inthis framework, functional interactions between the amygdala and pre-frontal cortex mediate emotional influences on cognitive processes suchas decision-making, as well as the cognitive regulation of emotion. How-ever, neurons in these structures often have entangled representations,whereby single neurons encode multiple cognitive and emotional vari-ables. Here we review studies using anatomical, lesion, and neurophys-iological approaches to investigate the representation and utilizationof cognitive and emotional parameters. We propose that these mentalstate parameters are inextricably linked and represented in dynamic neu-ral networks composed of interconnected prefrontal and limbic brainstructures. Future theoretical and experimental work is required to un-derstand how these mental state representations form and how shiftsbetween mental states occur, a critical feature of adaptive cognitive andemotional behavior.

173

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


PFC: prefrontalcortex

Contents

INTRODUCTION . . . . . . . . . . . . . . . . . . 174MENTAL STATES:

SYNTHESIZING COGNITIONAND EMOTION . . . . . . . . . . . . . . . . . 175

AN ANATOMICAL SUBSTRATEFOR INTERACTIONSBETWEEN EMOTION ANDCOGNITION . . . . . . . . . . . . . . . . . . . . 177Amygdala . . . . . . . . . . . . . . . . . . . . . . . . . . 177Prefrontal Cortex . . . . . . . . . . . . . . . . . . 177Anatomical Interactions Between

the PFC and Amygdala . . . . . . . . . . 178THE ROLE OF THE AMYGDALA

AND THE PFC INREPRESENTING MENTALSTATES: LESION STUDIES. . . . . 178Amygdala . . . . . . . . . . . . . . . . . . . . . . . . . . 180Prefrontal Cortex . . . . . . . . . . . . . . . . . . 180Prefrontal-Amygdala Interactions . . . 181

NEUROPHYSIOLOGICALCOMPONENTS OFMENTAL STATES . . . . . . . . . . . . . . . 181Neural Representations of

Emotional Valence and Arousalin the Amygdala and the OFC . . . 182

Neural Representations of CognitiveProcesses in the PFC. . . . . . . . . . . . 188

NEURAL NETWORKS ANDMENTAL STATES: ACONCEPTUAL ANDTHEORETICAL FRAMEWORKFOR UNDERSTANDINGINTERACTIONS BETWEENCOGNITION AND EMOTION. . 191

CONCLUSIONS . . . . . . . . . . . . . . . . . . . . 195

INTRODUCTION

The past century has witnessed a debate con-cerning the nature of emotion. When the brainis confronted with a stimulus that evokes emo-tion, does it first respond by activating a rangeof visceral and behavioral responses, which areonly then followed by the conscious experience

of emotion? For example, when we encountera threatening snake, does autonomic reactivity,as well as behaviors such as freezing or flee-ing, emerge prior to the feeling of fear? Thisview, championed by the psychologists WilliamJames and Carle Lange around the turn of thetwentieth century ( James 1884, 1894; Lange1922), has attracted renewed interest because ofthe influential work of Damasio and colleagues(Damasio 1994). Alternatively, do visceral andbehavioral responses occur as a result of centralprocessing in the brain—processing that givesrise to emotional feelings—which then regu-lates or controls a variety of bodily responses[a possibility raised decades ago by WalterCannon (1927) and Philip Bard (1928)]?

Neuroscientists have often sidestepped thisdebate by operationally defining a particularaspect of emotion—e.g., learning about fear—and using a specific behavioral or physiologicalassay—e.g., freezing—to investigate the neu-ral basis of the process (Salzman et al. 2005).This approach is agnostic about which responsecomes first: the visceral and behavioral expres-sion of emotion or the feeling of emotion. Butit has proven powerful in helping to identifyand characterize the neural circuitry respon-sible for specific aspects of emotional expres-sion and regulation. These investigations haveshown that one brain area, the amygdala, playsa vital role in many emotional processes (Baxter& Murray 2002, Lang & Davis 2006, LeDoux2000, Phelps & LeDoux 2005) and that theamygdala and its interconnections with the pre-frontal cortex (PFC) likely underlie many as-pects of the interactions between emotion andcognition (Barbas & Zikopoulos 2007, Murray& Izquierdo 2007, Pessoa 2008, Price 2007).

Today, we still lack a resolution to the origi-nal debate concerning the relationship betweenemotional feelings and the bodily expression ofemotions, in large part because both viewpointsappear to be supported in some circumstances.Emotional feelings do not necessarily involvevisceral and behavioral components andvice versa (Lang 1994). But neurobiologicaladvances—in particular, emerging data on theintimate relationship between the PFC and

174 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


limbic areas such as the amygdala—beginto suggest a solution. As discussed below,the amygdala is essential for many of thevisceral and behavioral expressions of emotion;meanwhile, the PFC—especially its medialand orbital regions—appears to be responsiblefor many of the cognitive aspects of emotionalresponses. However, recent studies suggestthat both the functional and the electrophys-iological characteristics of the amygdala andthe PFC overlap and intimately depend oneach other. Thus, the neural circuits mediatingcognitive, emotional, physiological, and be-havioral responses may not truly be separableand instead are inextricably linked. Moreover,we lack a unifying conceptual frameworkfor understanding how the brain links theseprocesses and how these processes change inunison.

MENTAL STATES:SYNTHESIZING COGNITIONAND EMOTION

Here, we propose a theoretical foundation forunderstanding emotion in the context of itsintimate relation to the cognitive, physiolog-ical, and behavioral responses that constituteemotional expression. We review recent neu-robiological data concerning the amygdala andthe PFC and discuss how these data fit into aproposed framework for understanding inter-actions between emotion and cognition.

The concept of a mental state plays acentral role in our theoretical framework.We define a mental state as a disposition toaction—i.e., every aspect of an organism’s in-ner state that could contribute to its behavioror other responses—which may comprise allthe thoughts, feelings, beliefs, intentions, ac-tive memories, and perceptions, etc., that arepresent at a given moment. Thus mental statescan be described by a large number of variables,and the set of all mental state variables couldprovide a quantitative description of one’s dis-position to behavior. Of note, the identifica-tion of mental state variables is constrained bythe language we use to describe them. Conse-

quently, mental state variables are not neces-sarily unique, and they are not necessarily inde-pendent from each other. Mental state variablesneed not be conscious or unconscious becauseboth types of variables can predispose one toaction. Overall, an organism’s mental state in-corporates internal variables, such as hunger orfear, as well as the representation of a set of en-vironmental stimuli present at a given moment,and the temporal context of stimuli and events.Any given mental state predisposes an organismto respond in certain ways; these actions may becognitive (e.g., making a decision), behavioral(e.g., freezing or fleeing), or physiological (e.g.,increasing heart rate). Mental state variables areuseful theoretical constructs because they pro-vide quantitative metrics for analyzing and un-derstanding behavioral and brain processes.

The concept of a mental state is intimatelyrelated to, but distinct from, what we call a brainstate. Each mental state corresponds to one ormore states of the dynamic variables—firingrates, synaptic weights, etc.—that describe theneural circuits of the brain; the full set of val-ues of these variables constitutes a brain state.How are the variables characterizing a mentalstate represented at the neural circuit level—i.e., the current brain state? This is one way tophrase a fundamental and long-standing ques-tion for neuroscientists. At one end of the spec-trum is the possibility that each neuron encodesonly one variable. For example, a neuron mayrespond only to the pleasantness of a sensorystimulus, and not to its identity, to its mean-ing, or to the context in which the stimulus ap-pears. When neurons encode only one variable,other neurons may easily read out the informa-tion represented, and the representation can, inprinciple, be modified without affecting othermental state variables.

One of the disadvantages of the type of rep-resentation described immediately above is wellillustrated by what is known as the “bindingproblem” (Malsburg 1999). If each neuron rep-resents only one mental state variable, then itis difficult to construct representations of com-plex situations. For example, consider a scenewith two visual stimuli, one associated with

www.annualreviews.org • Amygdala, PFC, and Mental States 175

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


reward and the other with punishment. Thebrain state should contain the information thatpleasantness is associated with the first stimu-lus and not with the other. If neurons repre-sent only one mental state variable at a time,like stimulus identity, or stimulus valence, then“binding” the information about different vari-ables becomes a substantial challenge. In thiscase, there must be an additional mechanismthat links the activation of the neuron repre-senting pleasantness to the activation of theneuron representing the first stimulus. Onesimple and efficient way to solve this problem isto introduce neurons with mixed selectivity toconjunctions of events, such as a neuron that re-sponds only when the first stimulus is pleasant.In this scheme, the representations of pleasant-ness and stimulus identity would be entangledand more difficult to decode, but the numberof situations that could be represented wouldbe significantly larger. As discussed below, dif-ferent brain areas may contain representationswith different degrees of entanglement.

How do emotions fit into the conceptualframework of mental states arising from brainstates? One influential schema for character-izing emotion posits that emotions can varyalong two axes: valence (pleasant versus un-pleasant or positive versus negative) and in-tensity (or arousal) (Lang et al. 1990, Russell1980). These two variables can simply be con-ceived as components of the current mentalstate. Two mental states correspond to differ-ent emotions when at least one of the two men-tal state variables—valence or intensity—is sig-nificantly different. Thus, variables describingemotions have the same ontological status asdo variables that describe cognitive processessuch as memory, attention, decision-making,language, and rule-based problem-solving. Be-low, we describe neurophysiological data doc-umenting that variables such as valence andarousal are strongly encoded in the amygdala–prefrontal circuit, along with variables relatedto other cognitive processes. We suggest thatneural representations in the amygdala maybe more biased toward encoding mental statevariables characterizing emotions (valence and

intensity). PFC neurons may encode a broaderrange of variables in an entangled fashion, re-flecting the complexity of the behavior and cog-nition that are its putative outputs.

The concept of a mental state unites cogni-tion and emotion as part of a common frame-work. How does this framework contribute tothe debate about the relationship between emo-tions and bodily responses? We argue that theissues raised in the debate essentially dissolvewhen one conceptualizes emotions as part ofmental states: Neither emotional feelings norbodily responses necessarily come first or sec-ond. Rather, both of these aspects of emo-tion are outputs of the neural networks thatrepresent mental states. Furthermore, all thethoughts, physiological responses, and behav-iors that constitute emotion are part of an on-going feedback loop that alters the dynamic,ever-fluctuating brain state and generates newmental states from moment to moment.

How do mental states that integrate emo-tion and cognition arise from the activityof neural circuits? Below, we describe a po-tential anatomical substrate—the amygdala–prefrontal circuit—for emotional-cognitive in-teractions in the brain and how neurons inthese areas could dynamically contribute toa subject’s mental state. First, we review thebidirectional connections between the amyg-dala and the PFC that could form the basisof many interactions between cognition andemotion. Second, we review neurobiologicalstudies that used lesions and pharmacologi-cal inactivation to investigate the function ofthe amygdala–PFC circuitry. Third, we reviewneurophysiological data from the amygdala andthe PFC that reveal encoding of variables crit-ical for the representation of mental states andfor the learning algorithms—specifically, rein-forcement learning (RL)—that emphasize theimportance of encoding these parameters foradaptive behavior. For all these topics, we fo-cus on data collected from nonhuman primates.Compared with their rodent counterparts, non-human primates are much more similar to hu-mans in terms of both behavioral repertoire andanatomical development.

176 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


ACC: anteriorcingulate cortex

OFC: orbitofrontalcortex

Finally, we describe a theoretical proposalthat explains how mental states might emergein the brain and how the amygdala and the PFCcould play an integral role in this process. Inparticular, we propose that the interactions be-tween emotion and cognition may be under-stood in the context of mental states that can beswitched or gated by internal or external events.

AN ANATOMICAL SUBSTRATEFOR INTERACTIONS BETWEENEMOTION AND COGNITION

This review focuses on interactions betweenthe amygdala and the PFC because of theirlong-established roles in mediating emotionaland cognitive processes (Holland & Gallagher2004, Lang & Davis 2006, LeDoux 2000,Miller & Cohen 2001, Ochsner & Gross 2005,Wallis 2007). The amygdala is most often dis-cussed in the context of emotional processes;yet it is extensively interconnected with thePFC, especially the posterior orbitofrontal cor-tex (OFC), and the anterior cingulate cortex(ACC). Here we provide a brief overview ofamygdala and PFC anatomy, with an emphasison the potential anatomical basis of interactionsbetween cognitive and emotional processes.

Amygdala

The amygdala is a structurally and functionallyheterogeneous collection of nuclei lying inthe anterior medial portion of each temporallobe. Sensory information enters the amygdalafrom advanced levels of visual, auditory, andsomatosensory cortices, from the olfactory sys-tem, and from polysensory brain areas such asthe perirhinal cortex and the parahippocampalgyrus (Amaral et al. 1992, McDonald 1998,Stefanacci & Amaral 2002). Within the lateralnucleus, the primary target of projections fromunimodal sensory cortices, different sensorymodalities are segregated anatomically. But,owing in part to intrinsic connections, multi-modal encoding subsequently emerges in thelateral, basal, accessory basal, and other nucleiof the amygdala (Pitkanen & Amaral 1998,

Stefanacci & Amaral 2000). Output from theamygdala is directed to a wide range of targetstructures, including the PFC, the striatum,sensory cortices (including primary sensorycortices, connections which are probablyunique to primates), the hippocampus, theperirhinal cortex, the entorhinal cortex, and thebasal forebrain, and to subcortical structuresresponsible for aspects of physiological re-sponses related to emotion, such as autonomicresponses, hormonal responses, and startle(Davis 2000). In general, subcortical projec-tions originate from the central nucleus, andprojections to cortex and the striatum originatefrom the basal, accessory basal, and in somecases the lateral nuclei (Amaral et al. 1992, 2003;Amaral & Dent 1981; Carmichael & Price1995a; Freese & Amaral 2005; Ghashghaeiet al. 2007; Stefanacci et al. 1996; Stefanacci &Amaral 2002; Suzuki & Amaral 1994).

Prefrontal Cortex

The PFC, located in the anterior portion ofthe cerebral cortex and defined by projectionsfrom the mediodorsal nucleus of the thalamus(Fuster 2008), is composed of a group of inter-connected brain areas. The distinctive featureof primate PFC is the emergence of dysgranu-lar and granular cortices, which are completelyabsent in the rodent. In rodents, prefrontal cor-tex is entirely agranular (Murray 2008, Preuss1995, Price 2007, Wise 2008). Therefore, muchof the primate PFC does not have a clear-cuthomolog in rodents. The PFC is often groupedinto different subregions; Petrides & Pandya(1994) have described these as dorsal and lateralareas (Walker areas 9, 46, and 9/46), ventrolat-eral areas (47/12 and 45), medial areas (32 and24), and orbitofrontal areas (10, 11, 13, 14, and47/12). Of note, there are extensive intercon-nections between different PFC areas, allowinginformation to be shared within local networks(Barbas & Pandya 1989, Carmichael & Price1996, Cavada et al. 2000), and informationalso converges from sensory cortices in mul-tiple modalities (Barbas et al. 2002). In general,dorsolateral areas receive input from earlier


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


sensory areas (Barbas et al. 2002). Orbitofrontalareas receive inputs from advanced stages ofsensory processing from every modality, in-cluding gustatory and olfactory (Carmichael &Price 1995b, Cavada et al. 2000, Romanski et al.1999). Thus, extrinsic and intrinsic connectionsmake the PFC a site of multimodal convergenceof information about the external environment.

In addition, the PFC receives inputs thatcould inform it about internal mental state vari-ables, such as motivation and emotions. Or-bital and medial PFC are closely connected withlimbic structures such as the amygdala (see be-low) and also have direct and indirect connec-tions with the hippocampus and rhinal cortices(Barbas & Blatt 1995, Carmichael & Price1995a, Cavada et al. 2000, Kondo et al. 2005,Morecraft et al. 1992). Medial and part of or-bital PFC has connections to the hypothalamusand other subcortical targets that could medi-ate autonomic responses (Ongur et al. 1998).Neuromodulatory input to the PFC fromdopaminergic, serotonergic, noradrenergic,and cholinergic systems could also convey in-formation about internal state (Robbins &Arnsten 2009). Finally, outputs from the PFC,especially from dorsolateral PFC, are directedto motor systems, consistent with the notionthat the PFC may form, represent, and/ortransmit motor plans (Bates & Goldman-Rakic1993, Lu et al. 1994). Altogether, the PFC re-ceives inputs that provide information aboutmany external and internal variables, includ-ing those related to emotions and to cogni-tive plans, providing a potential anatomical sub-strate for the representation of mental states.

Anatomical Interactions Betweenthe PFC and Amygdala

Although there are diffuse bidirectional projec-tions between amygdala and much of the PFC[see, e.g., figure 4 of Ghashghaei et al. (2007)],the densest interconnections are between theamygdala and orbital areas (e.g., caudal area13) and medial areas (e.g., areas 24 and 25).The extensive anatomical connections amongthe amygdala, the PFC, and related structuresare summarized in Figure 1. Amygdala input

to the PFC often terminates in both superficialand deep layers. OFC output to the amygdalaoriginates in deep layers, and in some cases alsoin superficial layers, suggesting both feedfor-ward and feedback modes of information trans-mission (Ghashghaei et al. 2007).

Previous work has established that the OFCoutput to the amygdala is complex and segre-gated, targeting multiple systems in the amyg-dala (Ghashghaei & Barbas 2002). Some OFCoutput is directed to the intercalated masses, aribbon of inhibitory neurons in the amygdalathat inhibits activity in the central nucleus(Ghashghaei et al. 2007, Pare et al. 2003). In ad-dition, the OFC projects directly to the centralnucleus, providing a means by which the OFCcan activate this output structure in addition toinhibiting it (Ghashghaei & Barbas 2002, Ste-fanacci & Amaral 2000, 2002). Finally, the OFCprojects to the basal, accessory basal, and lateralnuclei, where it may influence computationsoccurring within the amygdala (Ghashghaei &Barbas 2002, Stefanacci & Amaral 2000, 2002).Overall, the bidirectional communication be-tween the amygdala and the OFC, as well as theconnections with the rest of the PFC, providesa potential basis for the integration of cogni-tive, emotional, and physiological processesinto a unified representation of mental states.

THE ROLE OF THE AMYGDALAAND THE PFC INREPRESENTING MENTALSTATES: LESION STUDIES

Recent studies using lesions or pharmaco-logical inactivation combined with behavioralstudies in monkeys have begun to reveal thespecific roles of the primate amygdala andvarious regions of the PFC in cognitive andemotional processes. We focus here on studiesthat have helped demonstrate the roles of thesebrain structures in processes such as valuation,rule-based actions, emotional processes, at-tention, goal-directed behavior, and workingmemory—processes that are likely to set someof the variables that constitute a subject’smental state.

178 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Amygdala

Central nucleus

Lateralnucleus

Prefrontalcortex

Autonomicnervous system

HPA axis (hormonal responses)Behavioral expressions of emotion

(e.g., freezing, startle reaction)

Hippocampus, rhinal cortices

Basal andaccessory

basal nuclei

Orbitofrontalcortex

Anteriorcingulate

cortex

Intercalatedmasses

Sensorycortices

Sensorycortices

Heavy

a

LightModerate AmyAmyAmy

M25

D24

MPAII

M14

M10

M9

32

c

1311

10

O12 OProOProOPro

O14 O25

AmyAmyAmy

OPAIIOPAIIOPAIIOLF

e f

HPA axis (hormonal responsesBehavioral expressions of emotion

(e.g., freezing, startle reaction)

pocampus,hinall co trtiices

V8 131110

O12

O14 O25

D9

10

D8D46

V46L12

AmyAmyAmy

AmyAmyAmy

OPAIIOPAIIOPAIIOLF

Heavy

d

LightModerate AmyAmyAmy

M25

D24

MPAII

M14

M10

M9

32

5 mm

Dorsal

Rostral

b

5 mm

Dorsal

Rostral

V8

D9

10

D8D46V46L12

AmyAmyAmy

OProOProOPro

Inhibitory connections

Excitatory connections

Figure 1Overview of anatomical connections of the amygdala and the prefrontal cortex (PFC). Schematic showing some (but not all) the mainprojections of the amygdala and the PFC. The interconnections of the amygdala and the PFC (and especially the OFC) areemphasized. (a–c) Summary of projections from the amygdala to the PFC (density of projections is color coded). (d– f ) Summary ofprojections from the PFC to the amygdala (projection density is color coded). The complex circuitry between the amygdala and theOFC is also highlighted (red arrows connect the structures). Medial amygdala nuclei not shown. Many additional connections of bothamygdala and PFC are not shown. Panels a–f were adapted with permission from figures 5 and 6 of Ghashghaei et al. (2007).


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Amygdala

Historically, lesions of primate amygdala pro-duced a wide range of behavioral and emotionaleffects (Aggleton & Passingham 1981, Jones &Mishkin 1972, Kluver & Bucy 1939, Spiegler &Mishkin 1981, Weiskrantz 1956); but in recentyears, scientists have increasingly recognizedthe importance of using anatomically preciselesions that spare fibers of passage. Many olderstudies had employed aspiration or radiofre-quency lesions, which destroy both gray andwhite matter. By contrast, recent studies usingexcitotoxic chemical injections, which specifi-cally kill cell bodies, have revised our under-standing of cognitive and emotional functionsthat require the amygdala (Baxter & Murray2000, Izquierdo & Murray 2007). Some conclu-sions, however, have been confirmed over manystudies using both old and new techniques.In particular, scientists have most prominentlyused two types of behavioral tasks to estab-lish the amygdala’s role in forming or updat-ing associations between sensory stimuli andreinforcement. First, consistent with findingsfrom rodents, the primate amygdala is requiredfor fear learning induced by Pavlovian condi-tioning (Antoniadis et al. 2009). Second, theamygdala is required for updating the value of arewarding reinforcer during a devaluation pro-cedure (Machado & Bachevalier 2007, Malkovaet al. 1997, Murray & Izquierdo 2007). In thistype of task, experimenters satiate an animal ona particular type of reward and test whethersatiation changes subsequent choice behaviorsuch that the animal chooses the satiated foodtype less often; amygdala lesions eliminate theeffect of satiation. Pharmacological inactiva-tion of the amygdala has confirmed the amyg-dala’s role in updating a representation of areinforcer’s value; however, once this updatingprocess finishes, the amygdala does not appearto be required (Wellman et al. 2005). In addi-tion, the amygdala is important for other as-pects of appetitive conditioned reinforcement(Parkinson et al. 2001) and for behavioral andphysiological responding to emotional stim-uli such as snakes and intruders in a manner

consistent with its playing a role in processingboth emotional valence and intensity (Izquierdoet al. 2005, Kalin et al. 2004, Machado et al.2009). Finally, experiments using ibotenic acidinstead of aspiration lesions in the amygdalahave led to revisions in our understanding ofthe amygdala for reversal-learning task perfor-mance (during which stimulus-reinforcementcontingencies are reversed). Recent evidenceindicates that the amygdala is not requiredfor reversal learning on tasks involving onlyrewards, unlike previous accounts (Izquierdo& Murray 2007). Overall, these data link theamygdala to functions that rely on neural pro-cessing related to both emotional valence andintensity.

Prefrontal Cortex

A long history of studies have used lesions toestablish the importance of the PFC in goal-directed behavior, rule-guided behavior, andexecutive functioning more generally (Fuster2008, Miller & Cohen 2001, Wallis 2007).These complex cognitive processes form an in-tegral part of our mental state. In addition,lesions of orbitofrontal cortex (OFC) causemany emotional and cognitive deficits remi-niscent of amygdala lesions, including deficitsin reinforcer devaluation and in behavioraland hormonal responses to emotional stim-uli (Izquierdo et al. 2004, 2005; Kalin et al.2007; Machado et al. 2009; Murray & Izquierdo2007). Recently, investigators have employeddetailed trial-by-trial data analysis to enhancethe understanding of the effects of lesions; thiswork led investigators to propose that ACCand OFC are more involved in the valuationof actions and stimuli, respectively (Kennerleyet al. 2006, Rudebeck et al. 2008, Rushworth &Behrens 2008).

In addition, a recent study separately exam-ined lesions of the dorsolateral PFC, the ven-trolateral PFC, the principal sulcus (PS), theACC, and the OFC on a task analogous to theWisconsin Card Sorting Test (Buckley et al.2009) used to assay PFC function in humans(Stuss et al. 2000). In the authors’ version of

180 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


the task, monkeys must discover by trial anderror the current rule that is in effect; sub-jects needed to employ working memory for therule, as well as to utilize information about re-cent reward history to guide behavior. Lesionsin different PFC regions caused distinct deficitprofiles: Deficits in working memory, reward-based updating of value representations, and ac-tive utilization of recent choice-outcome valueswere ascribed primarily to PS, OFC, and ACClesions, respectively (Buckley et al. 2009). Ofnote, this study used aspiration lesions of thetargeted brain regions, which almost certainlydamaged fibers of passage located nearby.

A classic finding following aspiration lesionsof the OFC is a deficit in learning about rever-sals of stimulus-reward contingencies ( Jones &Mishkin 1972). However, a recent study usedibotenic acid to place a discrete lesion in OFCareas 11 and 13 (Kazama & Bachevalier 2009)and failed to find a deficit in reversal learning.We therefore may need to revise our under-standing of how OFC contributes to reversallearning (similar to revisions made with refer-ence to amygdala function, see above); however,the lack of an effect in the recent study mayhave been due to the anatomically restricted na-ture of the lesion. This issue will require furtherinvestigation.

Prefrontal-Amygdala Interactions

The amygdala is reciprocally connected withthe PFC, primarily OFC and ACC, but alsodiffusely to other parts of the PFC (Figure 1).Studies have begun to examine possible func-tional interactions between the amygdala andthe OFC in mediating different aspects ofreinforcement-based and emotional behavior.In one powerful set of experiments, Baxter andcolleagues (2000) performed a crossed surgicaldisconnection of the amygdala and the OFCby lesioning amygdala on one side of the brainand the OFC in the other hemisphere [con-nections between the amygdala and the OFCare ipsilateral (Ghashghaei & Barbas 2002)]. Asnoted above, bilateral lesions of monkey amyg-dala or the OFC impair reinforcer devaluation;

consistent with this finding, the authors foundthat surgical disconnection also impaired rein-forcer devaluation, indicating that the amyg-dala and the OFC must interact to update thevalue of a reinforcer. Notably, in humans, neu-roimaging studies on rare patients with focalamygdala lesions have revealed that the BOLDsignal related to reward expectation in the ven-tromedial PFC is dependent on a functioningamygdala (Hampton et al. 2007). Investigatorshave also described functional interactions be-tween the amygdala and the OFC in rodents(Saddoris et al. 2005, Schoenbaum et al. 2003);however, as noted above, rodent OFC may notnecessarily correspond to any part of the pri-mate granular/dysgranular PFC (Murray 2008,Preuss 1995, Wise 2008).

The lesion studies described above supportthe notion that the PFC and the amygdala,often in concert with each other, participate inexecutive functions such as attention, rule rep-resentation, working memory, planning, andvaluation of stimuli and actions. In addition,these structures mediate aspects of emotionalprocessing, including processing related toemotional valence and intensity. Togetherthese variables form an integral part of whatwe have termed a mental state. However, onemust exercise some caution when interpretingthe results of lesion studies: Owing to potentialredundancy in neural coding among braincircuits, a negative result does not necessarilyimply that the lesioned area is not normally in-volved in the function in question. As discussedin the next section, neurons in many parts ofthe PFC have complex, entangled physiologicalproperties. Given redundancy in encoding,it is therefore not surprising that lesions inthese parts of the PFC often do not impairfunctioning related to the full range of responseproperties.

NEUROPHYSIOLOGICALCOMPONENTS OFMENTAL STATES

We have defined mental states as action dis-positions, where actions are broadly defined to


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


CS: conditionedstimulus

US: unconditionedstimulus

include cognitive, physiological, or behavioralresponses. Here, we focus on neural signals inthe PFC and the amygdala that may encodekey cognitive and emotional features of a men-tal state: the valuation of stimuli, the valenceand intensity of emotional reactions to stimuli,our knowledge of the context of sensory stimuliand the requisite rules in that context, and ourplans for interacting with stimuli in the envi-ronment. We review recent neurophysiologicalrecordings from behaving nonhuman primatesthat demonstrate coding of all these variables,and they often feature entangled encoding ofmultiple variables.

Neural Representations of EmotionalValence and Arousal in the Amygdalaand the OFC

In recent years, a number of physiological ex-periments have been directed at understandingthe coding properties of neurons in the amyg-dala and the OFC. The amygdala has long beeninvestigated with respect to aversive processingand its prominent role in fear conditioning,primarily in rodents (Davis 2000, LeDoux2000, Maren 2005). However, a number ofscientists have recognized that the amygdalaalso plays a role in appetitive processing(Baxter & Murray 2002). Early neurophysio-logical experiments in monkeys established theamygdala as a potential locus for encoding theaffective properties of stimuli (Fuster & Uyeda1971; Nishijo et al. 1988a, b; Sanghera et al.1979; Sugase-Miyamoto & Richmond 2005).

To determine whether neurons in theprimate amygdala preferentially encodedrewarding or aversive associations, Paton andcolleagues (2006) recorded single neuronactivity while monkeys learned that visualstimuli—novel abstract fractal images—predicted liquid rewards or aversive air puffsdirected at the face, respectively. The ex-periments employed a Pavlovian procedurecalled trace conditioning, in which there is abrief temporal gap (the trace interval) betweenthe presentation of a conditioned stimulus(CS) and an unconditioned stimulus (US)

(Figure 2a). Monkeys exhibited two behav-iors that demonstrated their learning of thestimulus-outcome contingencies: anticipatorylicking (an approach behavior) and anticipatoryblinking (a defensive behavior). After mon-keys learned the initial CS-US associations,reinforcement contingencies were reversed.Neurophysiological recordings revealed thatthe amygdala contained some neurons thatrespond more strongly when a CS is pairedwith a reward (positive value-coding neurons),and other neurons respond more strongly whenthe same CS is paired with an aversive stimulus(negative value-coding neurons). Althoughindividual neurons exhibited this differentialresponse during different time intervals (e.g.,during the CS interval or parts of the traceinterval), across the population of neurons, thevalue-related signal was temporally extendedacross the entire trial (Figure 2b,c). Positiveand negative value-coding neurons appeared tobe intermingled in the amygdala; both types ofneurons dispersed within (and perhaps beyond)the basolateral complex (Belova et al. 2008,Paton et al. 2006).

Theoretical accounts of reinforcementlearning often posit a neural representation ofthe value of the current situation as a whole(state value). Data from Belova et al. (2008) sug-gest that the amygdala could encode the valueof the state instantiated by the CS presentation.Neural responses to the fixation point, whichappeared at the beginning of trials, were con-sistent with a role of the amygdala in encodingstate value. One can argue that the fixation pointis a mildly positive stimulus because monkeyschoose to look at it to initiate trials; and indeed,positive value-coding neurons tend to increasetheir firing in response to fixation point presen-tation, and negative value-coding neurons tendto decrease firing (Figure 3). Neural signalingafter reward or air-puff presentation also indi-cates that amygdala neurons track state value,as differential levels of activity, on a popula-tion level, extending well beyond the termina-tion of USs (Figure 2b,c). All these signals re-lated to reinforcement contingencies could beused to coordinate physiological and behavioral

182 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Visual stimulus (CS)Initial Reversal

Outcome (US)

Fixation

1000 ms

300 or 350 ms

Trace

1,500 ms

a

–1.0

–0.5

0

0.5

1.0

1.5

2.0

Negative amygdala cellsc

–0.5

0

0.5

1.0

1.5

2.0

2.5

3.0

Nor

mal

ized

neu

ral r

espo

nse

(z-s

core

uni

ts)

–1.0

–0.5

0

0.5

1.0

1.5Negative OFC cellse

0 1 2 3

–1

0

1

2

3

4

5

Positive amygdala cells

Reward trials

b

Air-puff trials

Time from CS onset (s)

Positive OFC cellsd

0 1 2 3

Figure 2Neural representation of positive and negative valence in the amygdala and the OFC. (a) Trace-conditioning task involving bothappetitive and aversive conditioning. Monkeys first centered gaze at a fixation point. Each experiment used novel abstract images asconditioned stimuli (CS). After fixating for 1 s, monkeys viewed a CS briefly, and following a 1.5-ms trace interval, unconditionedstimulus (US) delivery occurred. One CS predicted liquid reward, and a second CS predicted an aversive air puff directed at the face.After monkeys learned these initial associations, as indicated by anticipatory licking and blinking, the reinforcement contingencies werereversed. A third CS appeared on one-third of the trials, and it predicted either nothing or a much smaller reward throughout theexperiment (not depicted in the figure). (b–e) Normalized and averaged population peri-stimulus time histograms (PSTHs) for positiveand negative encoding amygdala (b,c) and OFC (d,e) neurons.

responses specific to appetitive and aversive sys-tems; therefore, they form a potential neuralsubstrate for positive and negative emotionalvariables.

As discussed earlier, however, valence is onlyone dimension of emotion; a second dimen-sion is emotional intensity, or arousal. Re-cent data also link the amygdala to this second


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


c

0 0.5 1.0

–0.5

0

0.5

1.0

1.5

2.0 Positive value-coding cells

Time from fixation point onset (s)

Nor

mal

ized

resp

onse

(z-s

core

uni

ts)

Negative value-coding cellsNon-value-coding cells

a

0 1 2

50

100

150

Firi

ng ra

te (s

p/s)

Example cell 1

5

10

bb0 1 2

Firi

ng ra

te (s

p/s)

Example cell 2

Large reward trialsSmall reward trialsAir-puff trials



0

50

100

Positivevalue-coding

cells

Non-value-coding

cells

Negativevalue-coding

cells

Perc

enta

ge o

f cel

ls

d Excitatory response to FPInhibitory response to FPNo response to FP

9

8

30

29

19

27

6

14

3

Figure 3Amygdala neurons track state value during the fixation interval. (a,b) PSTHs aligned on fixation point onsetfrom two example amygdala neurons (a, positive encoding; b, negative encoding) revealing responses to thefixation point consistent with their encoding state value. (c) Averaged and normalized responses to thefixation point for positive, negative, and nonvalue-coding amygdala neurons. (d ) Histograms showing thenumber of cells that increased, decreased, or did not change their firing rates as a function of which valencethe neuron encoded. Note that the fixation point may be understood as a mildly positive stimulus, so positiveneurons tend to increase their response to it and negative neurons decrease their response. Adapted withpermission from Belova et al. (2008, figure 2).

dimension. Belova and colleagues (2007) mea-sured responses to rewards and aversive airpuffs when they were either expected or un-expected. Surprising reinforcement is gener-ally experienced as more arousing than whenthe same reinforcements occur predictably;consistent with this notion, expectation oftenmodulated responses to reinforcement in theamygdala—in general, neural responses wereenhanced when reinforcement was surprising.For some neurons, this modulation occurredonly for rewards or for air puffs, but not for both(Figure 4a–d ). These neurons therefore couldparticipate in valence-specific emotional andcognitive processes. However, many neurons

modulated their responses to both rewards andair puffs (Figure 4e, f ). These neurons couldunderlie processes such as arousal or enhancedattention, which occur in response to intenseemotional stimuli of both valences. Consis-tent with this role, neural correlates of skinconductance responses, which are mediated bythe sympathetic nervous system, have been re-ported in the amygdala (Laine et al. 2009).Moreover, this type of valence-insensitivemodulation of reinforcement responses byexpectation could be appropriate for driv-ing reinforcement learning through attention-based learning algorithms (Pearce & Hall1980).

184 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


4

2

0

4

2

0

0 0.5 0 0.5 0 0.51.0 1.0 1.0

Expected

Example cell 1:Response to reward

modulated by expectation

Example cell 2:Response to air-puff


Example cell 3:Response to both reinforcements


Unexpected

Nor

mal

ized

resp

onse

(z-s

core

uni

ts)

a c e

b d f

Time from US onset (s)

Reward

Air-puff

Figure 4Valence-specific and valence-nonspecific encoding in the amgydala. (a–f ). Normalized and averaged neural responses to reinforcementwhen it was expected (magenta) and unexpected (cyan) for reward (a,c,e) and air puff (b,d, f ). Expectation modulated reinforcementresponses for only one valence of reinforcement in some cells (a–d) but modulated reinforcement responses for both valences in manycells (e,f ). These responses are consistent with a role of the amygdala in valence-specific processes, as well as valence-nonspecificprocesses, such as attention, arousal, and motivation. Adapted with permission from Belova et al. (2007, figure 3).

Of course, the amygdala does not operatein isolation; in particular, its close anatomicalconnectivity and functional overlap with theOFC raises the question of how OFC process-ing compares with and interacts with amygdalaprocessing. Using a paradigm similar to thatdescribed above, Morrison and Salzman dis-covered that the OFC contains neurons thatprefer rewarding or aversive associations, as inthe amygdala, and that, across the population,the signals extend from shortly after CS on-set until well after US offset (Figure 2d,e; datalargely collected from area 13) (Morrison &Salzman 2009, Salzman et al. 2007). OFC re-sponses to the fixation point are also modu-lated according to whether a cell has a positiveand negative preference, in a manner similar tothe amygdala (S. Morrison & C.D. Salzman,unpublished data). Together, these data

suggest that the OFC could also participate ina representation of state value.

Both positive and negative valences are rep-resented in the amygdala and the OFC, buthow might OFC and amygdala interact witheach other? Unpublished data indicate that theappetitive system—composed of cells that pre-fer positive associations—updates more quicklyin the OFC, adapting to changes in reinforce-ment contingencies faster than the appetitivesystem in the amygdala (S. Morrison & C.D.Salzman, personal communication, 2009).However, the opposite is true for the aversivesystem: Negative-preferring amygdala neuronsadapt to changes in reinforcement contingen-cies more rapidly than do their counterparts inthe OFC. Thus, the computational steps thatupdate representations in appetitive and aver-sive systems are not the same in the amygdala


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


and the OFC, even though the neurons appearto be anatomically interspersed in both struc-tures. In contrast, after reinforcement contin-gencies are well learned in this task, the OFCsignals upcoming reinforcement more rapidlythan does the amygdala in both appetitive andaversive cells. This finding is consistent with arole for the OFC in rapidly signaling stimulusvalues and/or expected outcomes once learningis complete—a signal that could be used to exertprefrontal control over limbic structures such asthe amygdala or to direct behavioral respondingmore generally.

The studies described above used Pavlo-vian conditioning—a procedure in which noaction is required of the subject to re-ceive reinforcement—to characterize neuralresponse properties in relation to appetitiveand aversive processing. However, many otherstudies have used decision-making tasks toquantify the extent to which neural responseproperties are related to reward values (Dorris& Glimcher 2004; Kennerley et al. 2008; Kimet al. 2008; Lau & Glimcher 2008; McCoy& Platt 2005; Padoa-Schioppa & Assad 2006,2008; Platt & Glimcher 1999; Roesch & Olson2004; Samejima et al. 2005; Sugrue et al. 2004;Wallis 2007; Wallis & Miller 2003); moreover,similar tasks are often used to examine humanvaluation processes using fMRI (Breiter et al.2001, Gottfried et al. 2003, Kable & Glimcher2007, Knutson et al. 2001, Knutson & Cooper2005, McClure et al. 2004, Montague et al.2006, O’Doherty et al. 2001, Rangel et al. 2008,Seymour et al. 2004). The strength of decision-making tasks is that the investigator can di-rectly compare the subjects’ preferences, on afine scale, with neuron signaling. For exam-ple, Padoa-Schioppa & Assad (2006) trainedmonkeys to indicate which of two possiblejuice rewards they wanted; they offered thejuices in different amounts by presenting visualtokens that indicated both juice type and juiceamount (Figure 5a). Using this task, they dis-covered that multiple signals were present indifferent populations of neurons in the OFC.Some OFC neurons encoded what the authorstermed “chosen value”: Firing was correlated

with the value of the chosen reward; some neu-rons preferred higher and lower values, respec-tively (Figure 5b,c). These cell populations arereminiscent of the positive and negative value-coding neurons uncovered using the Pavlovianprocedure described above. However, becausenegative valences were not explored in these ex-periments, these neurons may represent moti-vation, arousal, or attention, which are corre-lated with reward value (Maunsell 2004, Roesch& Olson 2004). Other OFC neurons encodedthe value of one of the rewards offered (offervalue cells; Figure 5d ) and others still simplyencoded the type of juice offered (taste neurons;Figure 5e), consistent with previous identifi-cation of taste-selective neurons in the OFC(Pritchard et al. 2007, Wilson & Rolls 2005).Further data suggested that the OFC responseswere menu-invariant—i.e., if a cell prefers Ato B, and B to C, it will also prefer A to C(Padoa-Schioppa & Assad 2008). This charac-teristic is called transitivity; it implies the abilityto use the representation of value as a context-independent economic currency that could sup-port decision-making. However, this findingmay depend on the exact design of the taskbecause other studies, focusing on partiallyoverlapping regions of the OFC, have reportedneural responses that reflect relative rewardpreferences, i.e., responses that vary with con-text and do not meet the standard of transitivity(Tremblay & Schultz 1999).

This rich variety of response properties inthe OFC and the amygdala still represents onlya subset of the types of encoding that havebeen observed in these brain areas. For exam-ple, amygdala neurons recorded during traceconditioning often exhibited image selectivity(Paton et al. 2006), and similar signals havebeen observed in the OFC (S. Morrison & C.D.Salzman, personal communication). Moreover,investigators have also described amygdalaneural responses to faces, vocal calls, and com-binations of faces and vocal calls (Gothardet al. 2007, Kuraoka & Nakamura 2007,Leonard et al. 1985). Meanwhile, the OFCneurons also encode gustatory working mem-ory and modulate their responses depending on

186 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


aFixate, 1.5 s

Offer on1–2 s delay Go, RT 0.75 s to juice

b

0

10

20

30

40

50

c

0

10

20

30

40

d e

0

10

0:1 1:3 1:2 1:1 2:1 3:1 4:1 6:1 2:0

Offers (#B:#A)

Firi

ng ra

te (s

p/s)

0

10

0:1 1:3 1:2 1:1 2:1 3:1 4:1 6:1 2:0

Figure 5OFC neural responses during economic decision-making. (a) Behavioral task. Monkeys centered gaze at afixation point and then viewed two visual tokens that indicate the type and quantity of juice reward beingoffered for potential saccades to each location (tokens, yellow and blue squares). After fixation point extinction,the monkey is free to choose which reward it wants by making a saccade to one of the targets. The amountsof juices offered of each type are titrated against each other to develop a full psychometric characterization ofthe monkey’s preferences as a function of the two juice types offered. (b–e) Activity of four neurons revealingdifferent types of response profiles. X-axis shows the quantity of each offer type. Chosen value neuronsincreased (b) or decreased (c) their firing when the value of their chosen option increased. Offer valueneurons (d ) increased their firing when the value of one of the juices offered increased. Juice neurons(e) increased their firing for trials with a particular juice type offered, independent of the amount of juiceoffered. Adapted from Padoa-Schioppa & Assad (2006, figures 1 and 3) with permission.


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


reward magnitude, reward probability, andthe time and effort required to obtain a re-ward (Kennerley et al. 2008). Overall, in ad-dition to encoding variables related to valenceand arousal/intensity—two variables central tothe representation of emotion—amygdala andOFC neurons encode a variety of other vari-ables in an entangled fashion (Paton et al. 2006,Rigotti et al. 2010a).

Neural Representations of CognitiveProcesses in the PFC

We have reviewed briefly the encoding of va-lence and arousal in the amygdala and the OFC,and now we turn our attention to the encodingof other mental state variables in the PFC. Ourgoal is not to discuss systematically every aspectof PFC neurophysiology, but instead to high-light response properties that may play an espe-cially vital role in setting the variables that con-stitute a mental state: encoding of rules, whichare essential for appropriately contextualizingenvironmental stimuli and other variables; flex-ible encoding of stimulus-stimulus associationsacross time and sensory modality; and encodingof complex motor plans.

Encoding of rules in the PFC. Understand-ing rules for behavior forms the basis for muchof our social interaction; therefore, rules mustroutinely be represented in our brains. A crit-ical feature of our cognitive ability is the abil-ity to apply abstract, as opposed to concrete,rules, i.e., rules that can be generalized andflexibly applied to new situations. In a strik-ing demonstration of this type of rule encodingin the PFC, Wallis and Miller recorded fromthree parts of the PFC (dorsolateral, ventrolat-eral, and the OFC) while monkeys performed atask requiring them to switch flexibly betweentwo abstract rules (Figure 6) (Wallis et al.2001). In this task, monkeys viewed two sequen-tially presented visual cues that could be eithermatching or nonmatching. In different blocksof trials, monkeys had to apply either a matchrule or a nonmatch rule—indicated by the pre-sentation of another cue at the start of the

trial—to guide their responding. The visualstimuli utilized in the blocks were identical;thus, the only difference between the blocks wasthe rule in effect, and this information must bea part of the monkey’s mental state. Many neu-rons in all three parts of the PFC exhibited se-lective activity depending on the rule in effect;some neurons preferred match and others non-match (Figure 6). Of note, rule-selective ac-tivity was only one type of selectivity that waspresent: Neurons often responded selectively tothe stimuli themselves, as well as to interactionsbetween the stimuli and the rules. Therefore, itappears that these neurons represent abstractrules along with other variables in an entangledmanner.

In the work by Wallis and Miller, the rulein effect was cued on every trial, and the mon-keys switched from one rule to the other on atrial-by-trial basis. In contrast, Mansouri andcolleagues (2006) used a task in which the ruleswitched in an uncued manner on a block-by-block basis, and monkeys had to discover therule in effect in a given block (an analog of theWisconsin Card Sorting Task). In one block oftrials, monkeys had to apply a color-match ruleto match two stimuli, and in the other block,monkeys had to apply a shape-match rule. Theauthors discovered that neural activity in thedorsolateral PFC encoded the rule in effect; dif-ferent neurons encoded color and shape rules(Figure 7a,b). Rule encoding occurred duringthe trial itself but also during the fixation in-terval, and even during the intertrial interval(ITI) (Figure 7c). This observation implies thata neural signature of the rule in effect was main-tained throughout a block of trials—even whenthe monkey was not performing a trial—as if themonkey had to keep the rule in mind. We sug-gest that this representation of rules thereforerepresents a distinctive component of a mentalstate.

Temporal integration of sensory stimuliand actions. One’s current situation is de-fined not only in terms of the stimuli currentlypresent, but also by the temporal context inwhich those stimuli appear, as well as by the

188 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


a Test 1

Sample

Matchrule

Response

Response

Test 2Delay 2

+ cue

(e.g., juice drop)

Delay 1

Nonmatchrule

500 ms

500 ms500 ms

1500 ms Response

Response

+ cue

(e.g., green

background)

800 ms

c

ed

b c Sample Delay

0 500 1000 1500 2000

e

Time from sample onset (ms)

Sample Delay

0 500 1000 1500 20000

10

20

30

40

50

d

Firi

ng ra

te (h

z)

Sample Delay

Cue epoch

Nonmatchrule

Matchrule

10

20

30

40

50

b Sample DelayJuice

Low tone

No juice

High tone

Figure 6Single neurons encode rules in PFC. (a) Behavioral task. Monkeys grasped a lever to initiate a trial. They then had to center gaze at afixation point while viewing a sample object, wait during a brief delay, and then view a test object. Two types of trials are depicted (doublehorizontal arrows). On match rule trials, monkeys had to release the lever if the test object matched the sample object. On nonmatch ruletrials, monkeys had to release the lever if the test object did not match the sample. Otherwise, they had to hold the lever until a thirdobject appeared that always required lever release. The rules in effect varied trial-by-trial by virtue of a different sensory cue (e.g., tonesor juice) presented during viewing of the sample object. (b,c) PFC neurons encoding match (b) or nonmatch (c) rules. Activity was higherin relation to the rule in effect regardless of the stimuli shown. Adapted with permission from Wallis et al. (2001, figures 1 and 2).

associations those stimuli have with other stim-uli. Fuster (2008) proposed that a cardinal func-tion of the PFC is to provide a representa-tion that reflects the temporal integration ofrelevant sensory information. Indeed, Fuster

and colleagues (2000) have demonstrated thistype of encoding in areas 6, 8, and 9/46 ofthe dorsolateral PFC. In this study, monkeysperformed a task in which they had to asso-ciate an auditory tone (high or low) with a


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


–1 0 3s

Start cue Start cue

Fixation

Sample ITI

Decision

Shape blocks

p < 0.00112

2–1.3 1.3s0

Spik

es/s

Cells

a

p < 0.0114

0

b

71

0

ITI Fix Decision

Start-cue Sample Beforereward

Afterreward

c

Color blocks

Figure 7PFC neurons encode rules in effect across time within a trial. Monkeys performed a task in which they had tomatch either the shape or color of two simultaneously presented objects with a sample object viewed earlierin the trial. Monkeys learned by trial and error whether a shape or color rule was in effect within a block oftrials, and block switches were uncued to the monkey. (a,b) Two PFC cells that fired differentially dependingon the rule in effect; activity differences emerged during the fixation (a) and intertrial intervals (ITI) (b).Activity is aligned on a start cue, which occurs before fixation on every trial. During the sample interval, onestimulus is presented over the fovea. During the decision interval, two stimuli are presented to the left andright; one matched the sample stimulus in color, and the other matched in shape. The correct choice can bechosen only if one has learned the rule in effect for the current block. (c) Distribution of activity differencesbetween shape and color rules for each cell studied in each time interval of a trial. Each line corresponds to asingle cell, and the solid parts of a line indicate when the cell fired differentially between color and shapeblocks. Encoding of rules occurred in all time epochs, indicating that PFC neurons encode the rule in effectacross time within a trial. Adapted with permission from Mansouri et al. (2006, figures 2 and 3).

subsequently presented colored target (red orgreen). The authors discovered cells that re-sponded selectively to associated tones and col-ors, e.g., cells that fired strongly only for thehigh tone and its associated target. Meanwhile,failure to represent the correct association ac-

curately was correlated with behavioral errors;thus, PFC neurons’ ability to form and repre-sent cross-temporal and cross-modality repre-sentations was linked to subsequent actions.

The integration of sensory stimuli in theenvironment, as described above, is key for

190 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


setting mental state variables; moreover, if werecall that a mental state can be defined as adisposition to action, any representation ofplanned actions must clearly be an importantelement of our mental state. Neural signalsrelated to planned actions have been reportedin numerous parts of the PFC in several tasks(Fuster 2008, Miller & Cohen 2001). In recentyears, scientists have used more complex motortasks to explore encoding of sequential move-ment plans. In the dorsolateral PFC, Tanjiand colleagues have described activity relatedto cursor movements that will result from aseries of planned arm movements (Mushiakeet al. 2006). This activity therefore reflectsfuture events that occur as a result of plannedmovements. Other studies of PFC neuronshave discovered neural ensembles that predicta sequence of planned movements (Averbecket al. 2006); when the required sequence ofmovements changes from block to block, theneural ensemble coding changes, too. In a man-ner reminiscent of rule encoding, this codingof planned movements was also present duringthe ITI, as if these cells were keeping note of theplanned movement sequence throughout theblock of trials (Averbeck & Lee 2007). Thus,the PFC not only tracks stimuli across time,but also represents the temporal integrationof planned actions and the events that hingeon them. Dorsolateral PFC may well interactwith the OFC and the ACC, and, via theseareas, the amygdala, to make decisions basedon the values of both environmental stimuliand internal variables and then to execute thesedecisions via planned action sequences.

NEURAL NETWORKS ANDMENTAL STATES: ACONCEPTUAL ANDTHEORETICAL FRAMEWORKFOR UNDERSTANDINGINTERACTIONS BETWEENCOGNITION AND EMOTION

We have so far reviewed how neurons in theamygdala and the PFC may encode neural sig-nals representing variables—some more closely

tied to emotional processes, and others to cog-nitive processes—that are components of men-tal states and how these representations are of-ten entangled (i.e., more than one variable isencoded by a single neuron). But how do theseneurons interact within a network to repre-sent mental states in their entirety? Moreover,how can cognitive processes regulate emotionalprocesses?

A central element of emotional regulationinvolves developing the ability to alter one’semotional response to a stimulus. In general,one can consider at least two basic ways inwhich this can occur. First, learning mecha-nisms may operate to change the representationof the emotional meaning of a stimulus. In-deed, one could simply forget or overwrite apreviously stored association. Moreover given astimulus previously associated with a particularreinforcement, such that the stimulus elicitsan emotional response, re-experiencing thestimulus in the absence of the associated re-inforcement can induce extinction. Extinctionis thought to be a learning process wherebypreviously acquired responses are inhibited. Inthe case of fear extinction, scientists currentlybelieve that original CS-US associations con-tinue to be stored in the brain (so that they arenot forgotten or overwritten), and inhibitorymechanisms develop that suppress the fearresponse (Quirk & Mueller 2008). Second,mechanisms must exist that can change orswitch one’s emotional responses depending onone’s knowledge of his/her context or situation.A simple example of this phenomenology oc-curs when playing the game of blackjack. Here,the same card, such as a jack of clubs, can berewarding, if it makes a total of 21 in your hand,or upsetting, if it makes a player go bust. Emo-tional responses to the jack of clubs can therebyvary on a moment-to-moment basis dependingon the player’s knowledge of the situation (e.g.,his/her understanding of the rules of the gameand of the cards already dealt). Emotionalvariables here depend critically on the cognitivevariables representing one’s understanding ofthe game and one’s current hand of cards. Al-though mechanisms for this type of emotional


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


regulation remain poorly understood, itpresumably involves PFC-amygdala neuralcircuitry.

What type of theoretical framework coulddescribe these different types of emotional reg-ulation? Are there qualitative differences be-tween the neural mechanisms that underliethem? Here we briefly describe one possibleapproach for explaining this phenomenology.Our proposal is built on the assumption thateach mental state corresponds to a large num-ber of states of dynamic variables that describeneurons, synapses, and other constituents ofneural circuits. These components must inter-act such that neural circuit dynamics can ac-tively maintain a representation of the currentdisposition of behavior, i.e., the current men-tal state. Complex interactions between thesecomponents must therefore correspond to theinteractions between mental state variables suchas emotional and cognitive parameters. Indeed,when brain states change, these changes typ-ically and inherently involve correlated modi-fications of multiple mental state variables. Inthis section, we discuss how a class of neuralmechanisms could underlie the representationof mental states and the potential interactionbetween cognition and emotion. We constructa conceptual framework whereby cognition-emotion interactions can occur via two sortsof mechanisms: associative learning and switch-ing between mental states representing differ-ent contexts or situations.

A natural candidate mechanism for rep-resenting mental states is the reverberatingactivity that has been observed at the singleneuron level in the form of selective persistentfiring rates, such as that which has beendescribed in the PFC (e.g., Figure 7) and otherstructures (Miyashita & Chang 1988, Yakovlevet al. 1998). Each mental state could be rep-resented by a self-sustained, stable pattern ofreverberating activity. Small perturbations ofthese activity patterns are damped by the inter-actions between neurons so that the state of thenetwork is attracted toward the closest patternof persistent activity representing a particularmental state. For this reason, these patterns

are called attractors of the neural dynamics.Attractor networks have been proposed asmodels for associative and working memory(Amit 1989, Hopfield 1982), for decision-making (Wang 2002), and for rule-basedbehavior (O’Reilly & Munakata 2000, Rolls& Deco 2002). Here, we suggest a scenarioin which attractors represent stable mentalstates and every external or internal eventencountered by an organism may steer theactivity from one attractor to a different one.This type of mechanism could provide stableyet modifiable representations for the mentalstates, just like the on and off states of a switch.Thus mental states could be maintained overrelatively long timescales but could also rapidlychange in response to brief events.

Attractor networks can be utilized to modelassociative learning. Consider again the ex-periment performed by Paton and colleagues(2006), described in the section on neural rep-resentation of emotional variables. In a simplemodel, one can assume that learning involvesmodifying connections from neurons repre-senting the CS (for simplicity, called externalneurons) to some of the neurons representingthe mental state, in particular those that rep-resent the value of the CS in relation to re-inforcement (called internal valence neurons).When the CSs are novel, a monkey does notknow what to expect (reward or air puff ). Themonkey may know that it will be one of thetwo outcomes. Therefore, the CS in that par-ticular context could induce a transition intoone of the preexistent attractors representingthe possible states. Some of these states corre-spond to the expectation of positive or nega-tive reinforcement, and other states could cor-respond to neutral valence states. The externalinput starts a biased competition between allthese different states. If the reinforcement re-ceived differs from the expected one, then thesynapses connecting external and internal neu-rons will be modified such that the competitionbetween mental states will generate a bias to-ward the correct association (see e.g., Fusi et al.2007). This learning process is typical of situa-tions in which there are one-to-one associations

192 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


and, for example, the same CS always has thesame value. The monkey can simply learn thestimulus-value associations by trial and error;with appropriate synaptic learning rules, the ex-ternal connections are modified as needed.

In the situation described above, one CSalways predicts reward, and the other pun-ishment. But such conditions do not alwaysexist. For example, Paton and colleagues re-versed reinforcement contingencies after learn-ing had occurred. In principle, learning thesereversed contingencies could involve modify-ing the external connections to the neural cir-cuit, thereby having new associations overwriteor override the previous associations. However,reversal tasks may not simply erase or unlearnassociations; instead, reversal tasks may rely onprocesses similar to those invoked during ex-tinction (Bouton 2002, Myers & Davis 2007).Increasing evidence implicates the amygdala-PFC circuit as playing a fundamental role inextinction (Gottfried & Dolan 2004, Izquierdo& Murray 2005, Likhtik et al. 2005, Milad &Quirk 2002, Olsson & Phelps 2004, Pare et al.2004, Quirk et al. 2000).

For the second type of emotional regula-tion, during which emotional responses to stim-uli depend on knowledge of one’s situation orcontext, we need a qualitatively different learn-ing mechanism. Consider a hypothetical vari-ant of the experiment by Paton et al. (2006), inwhich the associations are reversed and changedmultiple times. For example, stimulus A mayinitially be associated with a small reward andB with a small punishment. Then, in a secondblock of trials, A becomes associated with a largereward, and B with a large punishment. As-sume that as the experiment proceeds, subjectsgo back and forth between these two types ofblocks of trials so that the two contexts are alter-nated many times. In this case, if we can store arepresentation of both the two alternating con-texts, we can adopt a significantly more efficientcomputational strategy. Instead of learning andforgetting associations, we can simply switchfrom one context to the other. For example,on the first trial of a block, if a large punish-ment follows B, the monkey can predict that

seeing A on subsequent trials will result in itsreceiving a large reward. Overall, in the firstcontext, A and B can lead to only small rewardsand punishments, respectively. In the secondcontext, A and B always predict large rewardsand punishments. To implement this switchingtype of computational strategy, internal synap-tic connections within the neural network mustbe modified to create the neural representationsof the mental states corresponding to the twocontexts (Rigotti et al. 2010a).

To illustrate how a model employing anattractor neural network can describe the caseof mental state switching described above, weemploy an energy landscape metaphor [seee.g., Amit (1989) and Figure 8]. Each networkstate can be described as a vector containingthe activation states of all neurons, and it canbe characterized by its energy value. If we knowthe energy for each state, then we can predictthe network behavior because the network statewill evolve toward the state correspondingto the closest minimum of the energy. InFigure 8, we represent each network state as apoint on a plane and the corresponding energyas a surface that resembles a hilly landscape.To describe the hypothetical experiment underdiscussion, which involves switching betweencontexts, we assume that two variables, contextand valence (with two contexts and five differentvalences represented), characterize each mentalstate and that only one brain state correspondsto each mental state. As a consequence, for eachpoint on the context-valence plane there is onlyone energy value (Figure 8, red surface). Thenetwork naturally relaxes toward the bottomof the valleys (minima of the energy), whichrepresent different mental states. At the neurallevel, each of these points corresponds to aparticular pattern of persistent activity. The sixvalleys in Figure 8 correspond to six potentialmental states created after modifying internalconnections to represent the two different con-texts and the related CS-US associations. As aresult of the interactions due to recurrent con-nections, the cognitive variable correspondingto context constrains the set of accessibleemotional states. In both contexts, interactions


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Trace

Expectinglarge reward

Expecting smallpunishment

following CS B

Trace Fixation

Wait(second context)

CS A

External inputencoding CS A

+

Valence–

Context

1

2

Surprisesignal

Large punishment (US)

Figure 8The dynamics of context-dependent values. We consider a hypothetical variation of the experiment by Paton et al. (2006) in whichthere are two contexts. In the first context, CS A and B predict small rewards and punishments, respectively. In the second context, CSA and B are associated with large rewards and punishments. Six mental states now correspond to six valleys in the energy landscape. Inpanel 1, we consider the first trial of context 2, immediately after switching from context 1. Stimulus B has just been presented (notshown), and it is believed to predict small punishment. However a large punishment is delivered (not shown), and a surprise signal tiltsthe energy function (panel 2), inducing a transition to the neutral mental state of context 2 at the beginning of the next trial (panel 3; weassume for this example that the fixation interval has a neutral value). Now the system has already registered that it is in context 2.Consequently, the appearance of CS A tilts the energy landscape, and the mental state settles at a large positive value (panel 4). After CSdisappearance, the network relaxes into the high positive value mental state of context 2. Thus the network does not need to relearnthat CS A predicts a large reward because the network has already formed a representation for all the mental states contained withinthis simple experiment. Just knowing that the context has changed is sufficient for subjects to make an accurate prediction aboutimpending reinforcement. For a detailed attractor model implementing this form of context dependency see Rigotti et al. (2010b).

with external neurons representing CS A orB can tilt temporarily the energy surface andbring the neural network into a different valley,corresponding to a different mental state. Thefinal destination depends on the initial mentalstate representing the context. The valencesof the states differ in the two contexts becausevalences associated with large rewards andpunishments exist only in the second context.

This example illustrates the cognitive reg-ulation of emotion because changes in a cog-nitive variable (context) cause a change inthe possible associated emotional parameters(valence). Analogous mechanisms could under-lie how other cognitive variables can influenceemotional responses. For example, different so-cial situations can demand different emotionalresponses to similar sensory stimuli, and knowl-edge of the social situation (essentially a contextvariable) can thereby constrain the emotionalresponses possible.

We based the forgoing discussion on the as-sumption that the mental states are represented

by attractors of the neural dynamics. Alterna-tive and complementary solutions are basedon neural representations of mental states thatchange in time (Buonomano & Maass 2009,Jaeger & Hass 2004). For these neural systems,every trajectory or set of trajectories in thespace of all possible brain states represents aparticular mental state. These dynamic systemscan generate complex temporal sequences thatare important for motor planning (Susillo &Abbott 2009). However, they cannot instanta-neously generalize to situations in which eventsare timed differently and they can be difficult todecode. Generally speaking, all known modelsof mental states provide a useful conceptualframework for understanding the principlesof the dynamics of neural circuits, but theyfall short of capturing the richness and com-plexity of real biological neural networks. Forexample, brain states are not encoded solelyin the neuronal spiking activity. Investigatorsonly now have begun to study interactionsamong dynamic variables operating on diverse

194 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


timescales, all contributing to a particularbrain state [see e.g., Mongillo et al. (2008) fora working memory model based on short-termsynaptic facilitation].

The theoretical framework we proposeprovides a means for representing mentalstates in the distributed activity of networks ofneurons encoding entangled representations.Because we have defined mental states as actiondispositions, it is natural to wonder how mentalstates are linked to the selection and executionof actions. In recent years, reinforcementlearning (RL) algorithms have provided anelegant framework for understanding bothhow subjects choose their actions to maximizereward and minimize punishment and howthe brain may represent modeled parametersduring this process (Daw et al. 2006, Dayan& Abbott 2001, Sutton & Barto 1998). Inparticular, scientists have attempted to linktwo types of RL algorithms onto specificneural structures: a model-based algorithmand a model-free algorithm (Daw et al. 2005).Model-based algorithms are suited for goal-directed actions and likely involve the PFC,whereas model-free algorithms could mediatethe generation of habitual behavior (Graybiel2008) and may involve the striatum. Of note,both types of RL algorithms actually require analready formed representation of states (i.e., ofthe relevant variables of one’s current situation)to enable one to assign values to them to guideaction selection. If one drives in an unfamiliarneighborhood packed with restaurants andneeds to choose a restaurant, one must firstbuild a mental map of the environment as it isexperienced before assigning values to possibledestinations. This process of creating mentalstate representations is not provided for by RLalgorithms, which involve only the assignmentand updating of values to already createdstates.

Creating a representation of mental statesinvolves forging links between the many men-tal state variables that neurons represent.Recent work on the neural basis of object per-ception represents an initial step toward un-derstanding how variables may be combined

to support the formation of a mental staterepresentation. The perception of objects re-quires one to develop a representation that isinvariant for many viewing conditions, such asthe precise retinal position of the object, thesize, scale, or pose of the object, or the amountof clutter the object appears within the visualfield. Di Carlo and colleagues have now pro-vided evidence that unsupervised learning aris-ing from the temporal contiguity of stimuli ex-perienced during natural viewing leads to theformation of an invariant representation of avisual object (Li & DiCarlo 2008). Object per-ception corresponds to only one component ofa mental state, but the scientific approach pio-neered by Di Carlo’s group may provide a pathfor understanding how other mental state vari-ables also become linked to create a unified rep-resentation of a particular state. Indeed, sometheoreticians have proposed that simple mech-anisms such as temporal contiguity might un-derlie new mental state formation (O’Reilly &Munakata 2000, Rigotti et al. 2010a).

CONCLUSIONS

The conceptual framework we have put forthposits that mental states are composed of manyvariables that together correspond to an actiondisposition. These variables include parame-ters, such as valence and arousal, which areoften ascribed to emotional processes, as wellas parameters ascribed to cognitive processes,such as perceptions, memories, and plans. Ofcourse, mental state parameters also includevariables encoding our visceral state. Much inthe way that Wittgenstein argued that philo-sophical controversies dissolve once one care-fully disentangles the different ways in whichlanguage is being used (Wittgenstein 1958), weargue that the debate between scientists aboutthe origin of emotional feelings—whether vis-ceral processes precede or follow emotionalfeeling—dissolves. Instead, all these parametersmay be linked and together form the represen-tation of our mental state.

This conceptual framework has broadimplications for understanding interactions


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


between cognition and emotion in the brain.On the one hand, emotional processes can in-fluence cognitive processes; on the other hand,cognitive processes can regulate or modify ouremotions. Both of these interactions can be im-plemented by changing mental state variables(either emotional or cognitive ones); emotionsand thoughts shift together, correspondingto the new mental state. Of course, differentmechanisms may exist for implementing theseinteractions between cognition and emotion,such as mechanisms involving learning and

extinction, as well as mechanisms that supportthe creation of new mental state represen-tations, such as when one learns a new rule.The process of understanding the complexencoding properties of the amygdala, the PFC,and related brain structures, as well as under-standing their functional interactions, is in itsinfancy. Somehow the intricate connectivityof these brain structures gives rise to mentalstates and accounts for interactions betweencognition and emotion that are fundamental toour well-being and our existence.

DISCLOSURE STATEMENT

The authors are not aware of any affiliations, memberships, funding, or financial holdings thatmight be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS

We thank B. Lau and S. Morrison for helpful discussion, as well as comments on the manuscript,and H. Cline for invaluable support. C.D.S. gratefully acknowledges funding support from NIH(R01 MH082017, R01 DA020656, and RC1 MH088458) and the James S. McDonnell Founda-tion. S.F. receives support from DARPA SyNAPSE, the Gatsby Foundation, and the Sloan-SwartzFoundation.

LITERATURE CITED

Aggleton J, ed. 2000. The Amygdala—A Functional Analysis. Oxford: Oxford Univ. PressAggleton JP, Passingham RE. 1981. Syndrome produced by lesions of the amygdala in monkeys (Macaca

mulatta). J. Comp. Physiol. Psychol. 95:961–77Amaral D, Price J, Pitkanen A, Carmichael S. 1992. Anatomical organization of the primate amygdaloid com-

plex. In The Amygdala: Neurobiological Aspects of Emotion, Memory, and Mental Dysfunction, ed. J Aggleton,pp. 1–66. New York: Wiley-Liss

Amaral DG, Behniea H, Kelly JL. 2003. Topographic organization of projections from the amygdala to thevisual cortex in the macaque monkey. Neuroscience 118:1099–120

Amaral DG, Dent JA. 1981. Development of the mossy fibers of the dentate gyrus: I. A light and electronmicroscopic study of the mossy fibers and their expansions. J. Comp. Neurol. 195:51–86

Amit D. 1989. Modeling Brain Function—The World of Attractor Neural Networks. New York: Cambridge Univ.Press

Antoniadis EA, Winslow JT, Davis M, Amaral DG. 2009. The nonhuman primate amygdala is necessary forthe acquisition but not the retention of fear-potentiated startle. Biol. Psychiatry 65:241–48

Averbeck BB, Lee D. 2007. Prefrontal neural correlates of memory for sequences. J. Neurosci. 27:2204–11Averbeck BB, Sohn JW, Lee D. 2006. Activity in prefrontal cortex during dynamic selection of action se-

quences. Nat. Neurosci. 9:276–82Barbas H, Blatt GJ. 1995. Topographically specific hippocampal projections target functionally distinct pre-

frontal areas in the rhesus monkey. Hippocampus 5:511–33Barbas H, Ghashghaei H, Rempel-Clower N, Xiao D. 2002. Anatomic basis of functional specialization in

prefrontal cortices in primates. In Handbook of Neuropsychology, ed. J Grafman, pp. 1–27. Amersterdam:Elsevier Science B.V.

196 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Barbas H, Pandya DN. 1989. Architecture and intrinsic connections of the prefrontal cortex in the rhesusmonkey. J. Comp. Neurol. 286:353–75

Barbas H, Zikopoulos B. 2007. The prefrontal cortex and flexible behavior. Neuroscientist 13:532–45Bard P. 1928. A diencephalic mechanism for the expression of rage with special reference to the sympathetic

nervous system. Am. J. Physiol. 84:490–515Bates JF, Goldman-Rakic PS. 1993. Prefrontal connections of medial motor areas in the rhesus monkey.

J. Comp. Neurol. 336:211–28Baxter M, Murray E. 2000. Reinterpreting the behavioural effects of amygdala lesions in nonhuman primates.

See Aggleton 2000, pp. 545–68Baxter M, Murray EA. 2002. The amygdala and reward. Nat. Rev. Neurosci. 3:563–73Baxter M, Parker A, Lindner CC, Izquierdo AD, Murray EA. 2000. Control of response selection by reinforcer

value requires interaction of amygdala and orbital prefrontal cortex. J. Neurosci. 20:4311–19Belova MA, Paton JJ, Morrison SE, Salzman CD. 2007. Expectation modulates neural responses to pleasant

and aversive stimuli in primate amygdala. Neuron 55:970–84Belova MA, Paton JJ, Salzman CD. 2008. Moment-to-moment tracking of state value in the amygdala.

J. Neurosci. 28:10023–30Bouton ME. 2002. Context, ambiguity, and unlearning: sources of relapse after behavioral extinction.

Biol. Psychiatry 52:976–86Breiter H, Aharon I, Kahneman D, Dale A, Shizgal P. 2001. Functional imaging of neural responses to

expectancy and experience of monetary gains and losses. Neuron 30:619–39Buckley MJ, Mansouri FA, Hoda H, Mahboubi M, Browning PG, et al. 2009. Dissociable components of

rule-guided behavior depend on distinct medial and prefrontal regions. Science 325:52–58Buonomano DV, Maass W. 2009. State-dependent computations: spatiotemporal processing in cortical net-

works. Nat. Rev. Neurosci. 10:113–25Cannon W. 1927. The James-Lange theory of emotions: a critical examination and an alternative theory.

Am. J. Psychol. 39:106–24Carmichael ST, Price JL. 1995a. Limbic connections of the orbital and medial prefrontal cortex in macaque

monkeys. J. Comp. Neurol. 363:615–41Carmichael ST, Price JL. 1995b. Sensory and premotor connections of the orbital and medial prefrontal

cortex of macaque monkeys. J. Comp. Neurol. 363:642–64Carmichael ST, Price JL. 1996. Connectional networks within the orbital and medial prefrontal cortex of

macaque monkeys. J. Comp. Neurol. 371:179–207Cavada C, Company T, Tejedor J, Cruz-Rizzolo RJ, Reinoso-Suarez F. 2000. The anatomical connections

of the macaque monkey orbitofrontal cortex. A review. Cerebral. Cortex 10:220–42Damasio A. 1994. Descarte’s Error: Emotion, Reason, and the Human Brain. New York: Harcourt BraceDavis M. 2000. The role of the amygdala in conditioned and unconditioned fear and anxiety. See Aggleton

2000, pp. 213–87Daw ND, Niv Y, Dayan P. 2005. Uncertainty-based competition between prefrontal and dorsolateral striatal

systems for behavioral control. Nat. Neurosci. 8:1704–11Daw ND, O’Doherty JP, Dayan P, Seymour B, Dolan RJ. 2006. Cortical substrates for exploratory decisions

in humans. Nature 441:876–79Dayan P, Abbott LF. 2001. Theoretical Neuroscience. Cambridge, MA: MIT PressDorris MC, Glimcher PW. 2004. Activity in posterior parietal cortex is correlated with the relative subjective

desirability of action. Neuron 44:365–78Freese JL, Amaral DG. 2005. The organization of projections from the amygdala to visual cortical areas TE

and V1 in the macque monkey. J. Comp. Neurol. 486:295–317Fusi S, Asaad WF, Miller EK, Wang XJ. 2007. A neural circuit model of flexible sensorimotor mapping:

learning and forgetting on multiple timescales. Neuron 54:319–33Fuster J. 2008. The Prefrontal Cortex. London: ElsevierFuster JM, Bodner M, Kroger JK. 2000. Cross-modal and cross-temporal association in neurons of frontal

cortex. Nature 405:347–51Fuster JM, Uyeda AA. 1971. Reactivity of limbic neurons of the monkey to appetitive and aversive signals.

Electroencephalogr. Clin. Neurophysiol. 30:281–93


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Ghashghaei H, Barbas H. 2002. Pathways for emotion: interactions of prefrontal and anterior temporalpathways in the amygdala of the rhesus monkey. Neuroscience 115:1261–79

Ghashghaei HT, Hilgetag CC, Barbas H. 2007. Sequence of information processing for emotions based onthe anatomic dialogue between prefrontal cortex and amygdala. Neuroimage 34:905–23

Gothard KM, Battaglia FP, Erickson CA, Spitler KM, Amaral DG. 2007. Neural responses to facial expressionand face identity in the monkey amygdala. J. Neurophysiol. 97:1671–83

Gottfried J, O’Doherty J, Dolan RJ. 2003. Encoding predictive reward value in human amygdala and or-bitofrontal cortex. Science 301:1104–7

Gottfried JA, Dolan RJ. 2004. Human orbitofrontal cortex mediates extinction learning while accessing con-ditioned representations of value. Nat. Neurosci. 7:1144–52

Graybiel AM. 2008. Habits, rituals, and the evaluative brain. Annu. Rev. Neurosci. 31:359–87Hampton AN, Adolphs R, Tyszka MJ, O’Doherty JP. 2007. Contributions of the amygdala to reward ex-

pectancy and choice signals in human prefrontal cortex. Neuron 55:545–55Holland PC, Gallagher M. 2004. Amygdala-frontal interactions and reward expectancy. Curr. Opin. Neurobiol.

14:148–55Hopfield JJ. 1982. Neural networks and physical systems with emergent collective computational abilities.

Proc. Natl. Acad. Sci. USA 79:2554–58Izquierdo A, Murray EA. 2005. Opposing effects of amygdala and orbital prefrontal cortex lesions on the

extinction of instrumental responding in macaque monkeys. Eur. J. Neurosci. 22:2341–46Izquierdo A, Murray EA. 2007. Selective bilateral amygdala lesions in rhesus monkeys fail to disrupt object

reversal learning. J. Neurosci. 27:1054–62Izquierdo A, Suda RK, Murray EA. 2004. Bilateral orbital prefrontal cortex lesions in rhesus monkeys disrupt

choices guided by both reward value and reward contingency. J. Neurosci. 24:7540–48Izquierdo A, Suda RK, Murray EA. 2005. Comparison of the effects of bilateral orbital prefrontal cortex lesions

and amygdala lesions on emotional responses in rhesus monkeys. J. Neurosci. 25:8534–42Jaeger H, Haas H. 2004. Harnessing nonlinearity: predicting chaotic systems and saving energy in wireless

communication. Science 304:78–80James W. 1884. What is an emotion? Mind 9:188–205James W. 1894. The physical basis of emotion. Psychol. Rev. 1:516–29Jones B, Mishkin M. 1972. Limbic lesions and the problem of stimulus-reinforcement associations. Exp. Neurol.

36:362–77Kable JW, Glimcher PW. 2007. The neural correlates of subjective value during intertemporal choice.

Nat. Neurosci. 10:1625–33Kalin NH, Shelton SE, Davidson RJ. 2004. The role of the central nucleus of the amygdala in mediating fear

and anxiety in the primate. J. Neurosci. 24:5506–15Kalin NH, Shelton SE, Davidson RJ. 2007. Role of the primate orbitofrontal cortex in mediating anxious

temperament. Biol. Psychiatry 62:1134–39Kazama A, Bachevalier J. 2009. Selective aspiration or neurotoxic lesions of orbital frontal areas 11 and 13

spared monkeys’ performance on the object discrimination reversal task. J. Neurosci. 29:2794–804Kennerley SW, Dahmubed AF, Lara AH, Wallis JD. 2008. Neurons in the frontal lobe encode the value of

multiple decision variables. J. Cogn. Neurosci. 21:1162–78Kennerley SW, Walton ME, Behrens TE, Buckley MJ, Rushworth MF. 2006. Optimal decision making and

the anterior cingulate cortex. Nat. Neurosci. 9:940–47Kim S, Hwang J, Lee D. 2008. Prefrontal coding of temporally discounted values during intertemporal

choice. Neuron 59:161–72. Erratum Neuron 59(3):522Kluver H, Bucy P. 1939. Preliminary analysis of functions of the temporal lobes in monkeys. Arch. Neurol.

Psychiatry 42:979–1000Knutson B, Adams CM, Fong GW, Hommer D. 2001. Anticipation of increasing monetary reward selectively

recruits nucleus accumbens. J. Neurosci. 21:RC159Knutson B, Cooper JC. 2005. Functional magnetic resonance imaging of reward prediction. Curr. Opin.

Neurol. 18:411–17Kondo H, Saleem KS, Price JL. 2005. Differential connections of the perirhinal and parahippocampal cortex

with the orbital and medial prefrontal networks in macaque monkeys. J. Comp. Neurol. 493:479–509

198 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Kuraoka K, Nakamura K. 2007. Responses of single neurons in monkey amygdala to facial and vocal emotions.J. Neurophysiol. 97:1379–87

Laine CM, Spitler KM, Mosher CP, Gothard KM. 2009. Behavioral triggers of skin conductance responsesand their neural correlates in the primate amygdala. J. Neurophysiol. 101:1749–54

Lang PJ. 1994. The varieties of emotional experience: a meditation on James-Lange theory. Psychol. Rev.101:211–21

Lang PJ, Bradley MM, Cuthbert BN. 1990. Emotion, attention, and the startle reflex. Psychol. Rev. 97:377–95Lang PJ, Davis M. 2006. Emotion, motivation, and the brain: reflex foundations in animal and human research.

Prog. Brain Res. 156:3–29Lange C. 1922. The Emotions. Baltimore, MD: Williams & WilkinsLau B, Glimcher PW. 2008. Value representations in the primate striatum during matching behavior. Neuron

58:451–63LeDoux JE. 2000. Emotion circuits in the brain. Annu. Rev. Neurosci. 23:155–84Leonard CM, Rolls ET, Wilson FA, Baylis GC. 1985. Neurons in the amygdala of the monkey with responses

selective for faces. Behav. Brain Res. 15:159–76Li N, DiCarlo JJ. 2008. Unsupervised natural experience rapidly alters invariant object representation in visual

cortex. Science 321:1502–7Likhtik E, Pelletier JG, Paz R, Pare D. 2005. Prefrontal control of the amygdala. J. Neurosci. 25:7429–37Lu MT, Preston JB, Strick PL. 1994. Interconnections between the prefrontal cortex and the premotor areas

in the frontal lobe. J. Comp. Neurol. 341:375–92Machado CJ, Bachevalier J. 2007. The effects of selective amygdala, orbital frontal cortex or hippocampal

formation lesions on reward assessment in nonhuman primates. Eur. J. Neurosci. 25:2885–904Machado CJ, Kazama AM, Bachevalier J. 2009. Impact of amygdala, orbital frontal, or hippocampal lesions

on threat avoidance and emotional reactivity in nonhuman primates. Emotion 9:147–63Malkova L, Gaffan D, Murray EA. 1997. Excitotoxic lesions of the amygdala fail to produce impairment in

visual learning for auditory secondary reinforcement but interfere with reinforcer devaluation effects inrhesus monkeys. J. Neurosci. 17:6011–20

Mansouri FA, Matsumoto K, Tanaka K. 2006. Prefrontal cell activities related to monkeys’ success and failurein adapting to rule changes in a Wisconsin Card Sorting Test analog. J. Neurosci. 26:2745–56

Maren S. 2005. Synaptic mechanisms of associative memory in the amygdala. Neuron 47:783–86Maunsell JH. 2004. Neuronal representations of cognitive state: reward or attention? Trends Cogn. Sci. 8:261–

65McClure SM, Laibson DI, Loewenstein G, Cohen JD. 2004. Separate neural systems value immediate and

delayed monetary rewards. Science 306:503–7McCoy AN, Platt ML. 2005. Risk-sensitive neurons in macaque posterior cingulate cortex. Nat. Neurosci.

8:1220–27McDonald AJ. 1998. Cortical pathways to the mammalian amygdala. Prog. Neurobiol. 55:257–332Milad MR, Quirk GJ. 2002. Neurons in medial prefrontal cortex signal memory for fear extinction. Nature

420:70–74Miller EK, Cohen JD. 2001. An integrative theory of prefrontal cortex function. Annu. Rev. Neurosci. 24:167–

202Miyashita Y, Chang H. 1988. Neuronal correlate of pictorial short-term memory in the primate temporal

cortex. Nature 331:68–70Mongillo G, Barak O, Tsodyks M. 2008. Synaptic theory of working memory. Science 319:1543–46Montague PR, King-Casas B, Cohen JD. 2006. Imaging valuation models in human choice. Annu. Rev. Neurosci.

29:417–48Morecraft RJ, Geula C, Mesulam MM. 1992. Cytoarchitecture and neural afferents of orbitofrontal cortex in

the brain of the monkey. J. Comp. Neurol. 323:341–58Morrison S, Salzman C. 2009. The convergence of information about rewarding and aversive stimuli in single

neurons. J. Neurosci. 29:11471–83Murray EA. 2008. Neuropsychology of primate reward processes. In New Encyclopedia of Neuroscience, Vol. 6,

ed. LR Squire, pp. 993–99. Oxford: Academic Press


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Murray EA, Izquierdo A. 2007. Orbitofrontal cortex and amygdala contributions to affect and action inprimates. Ann. N.Y. Acad. Sci. 1121:273–96

Mushiake H, Saito N, Sakamoto K, Itoyama Y, Tanji J. 2006. Activity in the lateral prefrontal cortex reflectsmultiple steps of future events in action plans. Neuron 50:631–41

Myers KM, Davis M. 2007. Mechanisms of fear extinction. Mol. Psychiatry 12:120–50Nishijo H, Ono T, Nishino H. 1988a. Single neuron responses in amygdala of alert monkey during complex

sensory stimulation with affective significance. J. Neurosci. 8:3570–83Nishijo H, Ono T, Nishino H. 1988b. Topographic distribution of modality-specific amygdalar neurons in

alert monkey. J. Neurosci. 8:3556–69Ochsner KN, Gross JJ. 2005. The cognitive control of emotion. Trends Cogn. Sci. 9:242–49O’Doherty J, Kringelbach ML, Rolls ET, Hornak J, Andrews C. 2001. Abstract reward and punishment

representations in the human orbitofrontal cortex. Nat. Neurosci. 4:95–102Olsson A, Phelps EA. 2004. Learned fear of “unseen” faces after Pavlovian, observational, and instructed fear.

Psychol. Sci. 15:822–28Ongur D, An X, Price JL. 1998. Prefrontal cortical projections to the hypothalamus in macaque monkeys.

J. Comp. Neurol. 401:480–505O’Reilly R, Munakata Y. 2000. Computational Explorations in Cognitive Neuroscience. Cambridge, MA: MIT

PressPadoa-Schioppa C, Assad JA. 2006. Neurons in the orbitofrontal cortex encode economic value. Nature

441:223–26Padoa-Schioppa C, Assad JA. 2008. The representation of economic value in the orbitofrontal cortex is

invariant for changes of menu. Nat. Neurosci. 11:95–102Pare D, Quirk GJ, Ledoux JE. 2004. New vistas on amygdala networks in conditioned fear. J. Neurophysiol.

92:1–9Pare D, Royer S, Smith Y, Lang EJ. 2003. Contextual inhibitory gating of impulse traffic in the intra-

amygdaloid network. Ann. N. Y. Acad. Sci. 985:78–91Parkinson J, Crofts HS, McGuigan M, Tomic DL, Everitt BJ, Roberts AC. 2001. The role of the primate

amygdala in conditioned reinforcement. J. Neurosci. 21:7770–80Paton J, Belova M, Morrison S, Salzman C. 2006. The primate amygdala represents the positive and negative

value of visual stimuli during learning. Nature 439:865–70Pearce J, Hall G. 1980. A model for Pavlovian conditioning: variations in the effectiveness of conditioned but

not unconditioned stimuli. Psychol. Rev. 87:532–52Pessoa L. 2008. On the relationship between emotion and cognition. Nat. Rev. Neurosci. 9:148–58Petrides M, Pandya D. 1994. Comparative architectonic analysis of the human and macaque frontal cortex.

In Handbook of Neuropsychology, ed. F Boller, J Grafman, pp. 17–57. New York: ElsevierPhelps EA, LeDoux JE. 2005. Contributions of the amygdala to emotion processing: from animal models to

human behavior. Neuron 48:175–87Pitkanen A, Amaral DG. 1998. Organization of the intrinsic connections of the monkey amygdaloid complex:

projections originating in the lateral nucleus. J. Comp. Neurol. 398:431–58Platt ML, Glimcher PW. 1999. Neural correlates of decision variables in parietal cortex. Nature 400:233–38Preuss TM. 1995. Do rats have prefrontal cortex? The Rose-Woolsey-Akert program reconsidered. J. Cogn.

Neurosci. 7:1–24Price JL. 2007. Definition of the orbital cortex in relation to specific connections with limbic and visceral

structures and other cortical regions. Ann. N. Y. Acad. Sci. 1121:54–71Pritchard TC, Schwartz GJ, Scott TR. 2007. Taste in the medial orbitofrontal cortex of the macaque.

Ann. N. Y. Acad. Sci. 1121:121–35Quirk GJ, Mueller D. 2008. Neural mechanisms of extinction learning and retrieval. Neuropsychopharmacology

33:56–72Quirk GJ, Russo GK, Barron JL, Lebron K. 2000. The role of ventromedial prefrontal cortex in the recovery

of extinguished fear. J. Neurosci. 20:6225–31Rangel A, Camerer C, Montague PR. 2008. A framework for studying the neurobiology of value-based decision

making. Nat. Rev. Neurosci. 9:545–56

200 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Rigotti M, Ben Dayan Rubin D, Morrison SE, Salzman CD, Fusi S. 2010a. Attractor concretion as a mechanismfor the formation of context representations. NeuroImage doi: 10.1016/j.neuroimage.2010.01.047. In press

Rigotti M, Ben-Dayan Rubin D, Wang X-J, Fusi S. 2010b. The importance of the diversity in neural responsesin context-dependent tasks. Submitted

Robbins T, Arnsten A. 2009. The neuropsychopharmacology of fronto-executive function: monoaminergicmodulation. Annu. Rev. Neurosci. 32:267–87

Roesch MR, Olson CR. 2004. Neuronal activity related to reward value and motivation in primate frontalcortex. Science 304:307–10

Rolls E, Deco G. 2002. Computational Neuroscience of Vision. Oxford: Oxford Univ. PressRomanski LM, Bates JF, Goldman-Rakic PS. 1999. Auditory belt and parabelt projections to the prefrontal

cortex in the rhesus monkey. J. Comp. Neurol. 403:141–57Rudebeck PH, Bannerman DM, Rushworth MF. 2008. The contribution of distinct subregions of the ven-

tromedial frontal cortex to emotion, social behavior, and decision making. Cogn. Affect. Behav. Neurosci.8:485–97

Rushworth MF, Behrens TE. 2008. Choice, uncertainty and value in prefrontal and cingulate cortex.Nat. Neurosci. 11:389–97

Russell JA. 1980. A circumplex model of affect. J. Pers. Soc. Psychol. 39:1161–78Saddoris MP, Gallagher M, Schoenbaum G. 2005. Rapid associative encoding in basolateral amygdala depends

on connections with orbitofrontal cortex. Neuron 46:321–31Salzman CD, Belova MA, Paton JJ. 2005. Beetles, boxes and brain cells: neural mechanisms underlying

valuation and learning. Curr. Opin. Neurobiol. 15:721–29Salzman CD, Paton JJ, Belova MA, Morrison SE. 2007. Flexible neural representations of value in the primate

brain. Ann. N. Y. Acad. Sci. 1121:336–54Samejima K, Ueda Y, Doya K, Kimura M. 2005. Representation of action-specific reward values in the striatum.

Science 310:1337–40Sanghera MK, Rolls ET, Roper-Hall A. 1979. Visual responses of neurons in the dorsolateral amygdala of the

alert monkey. Exp. Neurol. 63:610–26Schoenbaum G, Setlow B, Saddoris MP, Gallagher M. 2003. Encoding predicted outcome and acquired

value in orbitofrontal cortex during cue sampling depends upon input from basolateral amygdala. Neuron39:855–67

Seymour B, O’Doherty JP, Dayan P, Koltzenburg M, Jones AK, et al. 2004. Temporal difference modelsdescribe higher-order learning in humans. Nature 429:664–67

Spiegler BJ, Mishkin M. 1981. Evidence for the sequential participation of inferior temporal cortex andamygdala in the acquisition of stimulus-reward associations. Behav. Brain Res. 3:303–17

Stefanacci L, Amaral DG. 2000. Topographic organization of cortical inputs to the lateral nucleus of themacaque monkey amygdala: a retrograde tracing study. J. Comp. Neurol. 421:52–79

Stefanacci L, Amaral DG. 2002. Some observations on cortical inputs to the macaque monkey amygdala: ananterograde tracing study. J. Comp. Neurol. 451:301–23

Stefanacci L, Suzuki WA, Amaral DG. 1996. Organization of connections between the amygdaloid complexand the perirhinal and parahippocampal cortices in macaque monkeys. J. Comp. Neurol. 375:552–82

Stuss DT, Levine B, Alexander MP, Hong J, Palumbo C, et al. 2000. Wisconsin Card Sorting Test performancein patients with focal frontal and posterior brain damage: effects of lesion location and test structure onseparable cognitive processes. Neuropsychologia 38:388–402

Sugase-Miyamoto Y, Richmond BJ. 2005. Neuronal signals in the monkey basolateral amygdala during rewardschedules. J. Neurosci. 25:11071–83

Sugrue LP, Corrado GS, Newsome WT. 2004. Matching behavior and the representation of value in theparietal cortex. Science 304:1782–87

Susillo D, Abbott L. 2009. Generating coherent patterns of activity from chaotic neural networks. Neuron.63:544–57

Sutton R, Barto A. 1998. Reinforcement Learning. Cambridge, MA: MIT PressSuzuki WA, Amaral DG. 1994. Perirhinal and parahippocampal cortices of the macaque monkey: cortical

afferents. J. Comp. Neurol. 350:497–533


Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.


Tremblay L, Schultz W. 1999. Relative reward preference in primate orbitofrontal cortex. Nature 398:704–8von der Malsburg C. 1999. The what and why of binding: the modeler’s perspective. Neuron 24:95–104Wallis JD. 2007. Orbitofrontal cortex and its contribution to decision-making. Annu. Rev. Neurosci. 30:31–56Wallis JD, Anderson KC, Miller EK. 2001. Single neurons in prefrontal cortex encode abstract rules. Nature

411:953–56Wallis JD, Miller EK. 2003. Neuronal activity in primate dorsolateral and orbital prefrontal cortex during

performance of a reward preference task. Eur. J. Neurosci. 18:2069–81Wang XJ. 2002. Probabilistic decision making by slow reverberation in cortical circuits. Neuron 36:955–68Weiskrantz L. 1956. Behavioral changes associated with ablation of the amygdaloid complex in monkeys.

J. Comp. Neurol. 49:381–91Wellman LL, Gale K, Malkova L. 2005. GABAA-mediated inhibition of basolateral amygdala blocks reward

devaluation in macaques. J. Neurosci. 25:4577–86Wilson FA, Rolls ET. 2005. The primate amygdala and reinforcement: a dissociation between rule-based and

associatively-mediated memory revealed in neuronal activity. Neuroscience 133:1061–72Wise SP. 2008. Forward frontal fields: phylogeny and fundamental function. Trends Neurosci. 31:599–608Wittgenstein W. 1958. Philosophical Investigations. Oxford: BlackwellYakovlev V, Fusi S, Berman E, Zohary E. 1998. Inter-trial neuronal activity in inferior temporal cortex: a

putative vehicle to generate long-term visual associations. Nat. Neurosci. 1:310–17

202 Salzman · Fusi

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.

NE33-FM ARI 14 May 2010 19:47

Annual Review ofNeuroscience

Volume 33, 2010Contents

Attention, Intention, and Priority in the Parietal LobeJames W. Bisley and Michael E. Goldberg � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 1

The Subplate and Early Cortical CircuitsPatrick O. Kanold and Heiko J. Luhmann � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �23

Fly Motion VisionAlexander Borst, Juergen Haag, and Dierk F. Reiff � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �49

Molecular Pathways of Frontotemporal Lobar DegenerationKristel Sleegers, Marc Cruts, and Christine Van Broeckhoven � � � � � � � � � � � � � � � � � � � � � � � � � � � � �71

Error Correction, Sensory Prediction, and Adaptationin Motor ControlReza Shadmehr, Maurice A. Smith, and John W. Krakauer � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �89

How Does Neuroscience Affect Our Conception of Volition?Adina L. Roskies � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 109

Watching Synaptogenesis in the Adult BrainWolfgang Kelsch, Shuyin Sim, and Carlos Lois � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 131

Neurological ChannelopathiesDimitri M. Kullmann � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 151

Emotion, Cognition, and Mental State Representation in Amygdalaand Prefrontal CortexC. Daniel Salzman and Stefano Fusi � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 173

Category Learning in the BrainCarol A. Seger and Earl K. Miller � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 203

Molecular and Cellular Mechanisms of Learning Disabilities:A Focus on NF1C. Shilyansky, Y.S. Lee, and A.J. Silva � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 221

Wallerian Degeneration, WldS, and NmnatMichael P. Coleman and Marc R. Freeman � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 245

v

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.

NE33-FM ARI 14 May 2010 19:47

Neural Mechanisms for Interacting with a World Fullof Action ChoicesPaul Cisek and John F. Kalaska � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 269

The Role of the Human Prefrontal Cortex in Social Cognitionand Moral JudgmentChad E. Forbes and Jordan Grafman � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 299

Sodium Channels in Normal and Pathological PainSulayman D. Dib-Hajj, Theodore R. Cummins, Joel A. Black,and Stephen G. Waxman � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 325

Mechanisms of Synapse and Dendrite Maintenance and TheirDisruption in Psychiatric and Neurodegenerative DisordersYu-Chih Lin and Anthony J. Koleske � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 349

Connecting Vascular and Nervous System Development: Angiogenesisand the Blood-Brain BarrierStephen J. Tam and Ryan J. Watts � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 379

Motor Neuron Diversity in Development and DiseaseKevin C. Kanning, Artem Kaplan, and Christopher E. Henderson � � � � � � � � � � � � � � � � � � � � � � 409

The Genomic, Biochemical, and Cellular Responses of the Retina inInherited Photoreceptor Degenerations and Prospects for theTreatment of These DisordersAlexa N. Bramall, Alan F. Wright, Samuel G. Jacobson, and Roderick R. McInnes � � � 441

Genetics and Cell Biology of Building Specific Synaptic ConnectivityKang Shen and Peter Scheiffele � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 473

Indexes

Cumulative Index of Contributing Authors, Volumes 24–33 � � � � � � � � � � � � � � � � � � � � � � � � � � � 509

Cumulative Index of Chapter Titles, Volumes 24–33 � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 513

Errata

An online log of corrections to Annual Review of Neuroscience articles may be found athttp://neuro.annualreviews.org/

vi Contents

Ann

u. R

ev. N

euro

sci.

2010

.33:

173-

202.

Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Col

umbi

a U

nive

rsity

on

07/0

7/10

. For

per

sona

l use

onl

y.

Emotion, Cognition, and Mental State Representation in ... · NE33CH09-Salzman ARI 14 May 2010...

Documents

Transcript of Emotion, Cognition, and Mental State Representation in ... · NE33CH09-Salzman ARI 14 May 2010...