Review

Cognitive-emotional interactions

The role of social cognition in emotion

Andreas Olsson and Kevin N. Ochsner
Department of Psychology, Columbia University, 1190 Amsterdam Avenue, New York, NY 10027, USA

Although recent research has shown that social Understanding emotions in self and others
cognition and emotion engage overlapping regions of The ability to understand both another person’s and one’s
the brain, few accounts of this overlap have been offered. own emotional states is essential for virtually all aspects of
What systems might be commonly or distinctively social behavior and crucially depends upon MSA. Indeed,
involved in each? The close functional relationship be- emotion understanding – by definition – requires a causal
tween social cognition and emotion might be under- attribution about the intentions behind an action. Evi-
stood in terms of a central role for mental state dence suggests that MSA contributes to emotion under-
attribution in the understanding, learning and regulation standing through the operation of both rapid stimulus-
of emotion. In each of these cases, mental state attribu- driven processes [4,5] and more deliberative, reflective and
tions might be supported by either stimulus-driven or conceptually driven processes [6–8].
more reflective processes. Evidence for the neural bases of stimulus-driven MSA
came initially from imaging studies showing that some
Exploring the role of mental state attribution in motor regions respond during both the execution and
emotion observation of specific movements [4]. The idea was that
Whether viewed from a phyologenetic or an ontogenetic if motor regions code the intentions behind one’s own
perspective, it is clear that the abilities to understand, action, then if activated when observing another person
learn from and behave appropriately towards one engaging in the same action, they could support a direct
another were as essential for our homonid ancestors experiential understanding of that person’s intention [4,9].
as they are for a developing child [1]. In the past decade, This ‘shared representation’ logic guided subsequent
insight into the neural mechanisms supporting these studies of the direct experience and observation of pain
abilities has been provided by two burgeoning fields of or emotion that also showed activation of overlapping
research: social cognitive neuroscience and affective neural systems, including most prominently the two cor-
neuroscience. Although these fields developed largely tical regions that receive ascending viscerosensory inputs:
independently [2], for multiple reasons the boundaries the anterior insula (AI) and the midportion of the anterior
between the phenomena they study are becoming cingulate cortex (mACC) [4,7,10–15]. The AI is believed to
increasingly blurred (Box 1). Perhaps chief among them support affective experience in part through interoceptive
is the finding that ostensibly different types of social awareness of these body state inputs [16,17], whereas the
cognitive or emotional abilities recruit similar suites of ACC is thought to code affective attributes of pain, such as
cortical and subcortical neural systems. Although this the perceived unpleasantness (as opposed to sensory-dis-
fact has been noted [3], to date few accounts of the criminative properties, such as location and intensity) [18–
apparent interdependence of social cognition and 20] and motivate appropriate behavior through projections
emotion have been advanced. to motor and autonomic centers [16,21]. The engagement of
Here, we review recent work suggesting that this the AI, ACC and other regions is thought to facilitate the
relationship can be understood – at least in part – in terms automatic sharing of – and hence direct experiential un-
of a fundamental role of one type of social cognitive derstanding of the intentions behind – affective states,
capacity in processing emotion: the ability to explain beha- which in turn might provide a substrate for empathic
vior in terms of intentional mental states, which we will connection [7,13].
refer to as mental state attribution (MSA; Box 2). The role Emotion understanding is not always so simple, how-
of MSA can be considered in three domains: (i) understand- ever, because nonverbal cues to emotion are often ambig-
ing emotion, (ii) learning emotionally significant infor- uous. In such cases, additional information is needed to
mation and (iii) regulation of emotional responses. For constrain attributions about a person’s intentions and
each domain, we discuss the roles in emotion processing hence their emotional state. One source of information is
of systems supporting MSAs that vary in their degree of our prior experience with that person and knowledge about
representational complexity and reflective control. Based antecedent events. For example, wide open eyes could
on this review, we propose a neural framework for un- mean that someone is either afraid or surprised – and
derstanding the role of social cognition in emotion that can elicit amygdala activity accordingly – depending on what
guide future research. one knows has just happened to them [22]. Similarly,
activation of shared representations – and presumably
Corresponding authors: Olsson, A. (aolsson@paradox.psych.columbia.edu);
empathic connection – might be blocked if one perceives
Ochsner, K.N. (kochsner@paradox.psych.columbia.edu). another to be a past or potential competitor [23,24].
1364-6613/$ – see front matter ß 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.tics.2007.11.010 65
Review Trends in Cognitive Sciences Vol.12 No.2

Box 1. Investigating social cognition and emotion: separate Box 2. MSA

approaches to inseparable phenomena? As used here, MSA is an umbrella term encompassing the various
The fact that social cognition and emotion depend upon highly ways in which we use mental state concepts to describe, understand
overlapping neural systems might seem at odds with the fact that and predict behavior, and corresponds to the ’intentional stance’, as
behavioral work on these topics has historically proceeded in largely described by philosophers [76]. MSA has figured importantly but
independent disciplines. Close inspection of their respective theo- differently in four types of behavioral work that have provided
retical and methodological approaches, however, suggests that springboards for current imaging research.
social cognition and emotion researchers might often be studying
intertwined aspects of the same phenomenon. Social cognition
To understand these relationships, consider how each field would Although social cognition research has not explicitly focused on
approach study of a man and woman on their first date. A social unpacking the mechanisms of MSA, it has appreciated that MSA is
cognition researcher might ask how they make attributions about central to the goal of understanding the causes of behavior. Social
the situational or dispositional causes of each other’s behavior, and cognition researchers have asked, for example, how much behavior
how such judgments are influenced by their motivations, attitudes, is due to internal (e.g. beliefs, intentions, dispositions and attitudes)
moods, group memberships and available cognitive resources as opposed to situational forces, in addition to how we weigh the
[53,72,73]. By contrast, an emotion researcher might be interested contributions of each when judging the causes of either our own
in describing the cognitive antecedents and/or expressions of each behavior or the behavior of others [53,77].
person’s affective experience, emotion-related behavior and auto-
nomic arousal, with an emphasis on understanding their cross- Emotion
species or cross-cultural consistencies [74,75]. In the study of emotion, MSA figures prominently, if not focally, in
Two types of similarities between social cognition and emotion appraisal theories, which specify how cognitive interpretations can
can be highlighted here. The first is that the attributions of interest shape emotions. MSAs about intentionality of another’s action, for
to social cognition researchers might determine how one responds example, will determine if one feels anger, and are central to
emotionally. In the present example, whether you think someone is secondary appraisals of one’s ability to cope with current stressors.
anxious by nature, or because anyone would be on a first date, MSAs are also used to introspect upon and describe one’s current or
could determine how you respond emotionally to them [53]. The past emotions [75].
second is that the stimuli used to study social cognition and
emotions are often related, if not identical, especially in neu- Development
roscience research on these topics. Stimuli in both emotion and Developmental psychologists coined the term ‘theory of mind’
social cognition experiments include real, simulated, described or (TOM) to explain a child’s emerging use of MSAs to explain
photographed social interactions, in addition to images of faces and behavior, and have debated whether TOM involves learning to
facial expressions, all of which might evince both social cognitive simulate another’s mind or the acquisition of abstract theoretical
attributions and emotional responses to varying degrees. knowledge about mental states [1,9,56]. Efforts to identify the age of
It might not be surprising, therefore, that social cognition and TOM onset has led to the development of false belief tasks thought
emotion depend upon similar neural systems, given that social to be diagnostic of the ability to use MSAs, as opposed to other
information carries great emotional and motivational significance. schemes, for explaining behavior [78].
Of course, we can respond emotionally to non-social stimuli (e.g.
odors, nature, etc.), and some emotions, such as pity and shame, Social cognitive and affective neuroscience
might require social cognitive attributions; space constraints Functional imaging and lesion work has informed the study of MSA
prevent further discussion of this topic here, but it is discussed by by identifying a network of associated brain systems and the
Harris and Fiske [64] and by Scherer et al. [75]. It is not clear, somewhat inconsistent deficits in MSA that follow from lesions to
however, when and how social interaction can be devoid of emotion some of it components. In general, these data have illustrated how
or affect. One goal for future neuroscience research should be to neuroscience is useful for understanding how complex abilities,
incorporate behavioral methods that could differentiate the roles of such as MSA, might fractionate into component computational
social cognition and emotion in social interaction. systems. A primary goal of current work is to characterize these
computations and describe how they give rise to different types of
MSAs [6,26,27,68].

Engagement of control processes that support more

cognitively complex MSAs might enable understanding activity [29]. In this study, MPFC was active for the
of the emotional state of a target when stimulus-triggered perception of pain in self and others versus perceiving
processing of situational information is not sufficient. damage to artificial limbs, which makes its unclear
These controlled MSAs enable us actively to take other whether MPFC involvement is related to mental attribu-
peoples’ perspectives and make judgments about their tions specific to empathy, non-empathy-related attribu-
emotions or diagnostic elements of stable emotional dis- tions or both. The lack of clear findings of MPFC
positions, thereby changing empathic responding [25] and activation could indicate a functional distinction between
activation in the AI and mACC [8]. By and large, they the affective experience of directly sharing another’s feel-
depend on a network of regions, including the right ings and the meta-cognitive task of thinking about and
temporal parietal junction (TPJ) and dorsal-rostral regions rating them [6,30], which is consistent with the obser-
of the medial prefrontal cortex (MPFC), including Brod- vation that MPFC activity during others’ pain (although
mann area (BA) 10 [6,26,27]. Interestingly, a recent meta- not significantly greater than baseline) might be positively
analysis singles out BA 10 – which is especially developed related to self-rated empathic concern [13]. Alternatively,
in humans – as being particularly sensitive to tasks invol- it is possible that experiential and reflective processing do
ving both emotions and MSA [28]. If the mACC and AI work together but only in certain circumstances. This
support direct experiential awareness of intentional states, possibility is supported by a recent study examining func-
it is possible that the MPFC network supports meta-cog- tional connectivity of mACC and AI regions commonly
nitive reflective awareness of them. active during the direct experience and observation of pain
Strikingly, only one study of stimulus-triggered empa- [15]. Here, mACC and AI activity was correlated with
thy showing mACC and AI activity has also shown MPFC activity in pain-related brainstem nuclei during self-pain

66
 Review Trends in Cognitive Sciences Vol.12 No.2

and with the MPFC when observing others in pain when, and how, we use similar processes for judging
(Figure 1a, b). The fact that these regions were not ident- ourselves and others, and the extent to which these pro-
ified in the vast majority of standard contrasts suggests cesses are dedicated to the processing of social information
that MPFC involvement might have been missed in prior as opposed to more general meta-cognitive functions used
work examining activity averaged across subjects and time when categorizing and re-representing affective and other
points. The coactivation of these networks suggests that types of input in a symbolic format, is currently a matter of
emotion understanding might in some cases involve the debate [6,32].
use of reflective cognition drawing on the MPFC to inter- In sum, the research surveyed in this section suggests
pret activity in networks supporting experiential proces- that understanding the emotions of others through MSA is
sing of emotions. supported by a distributed functional network including
Interestingly, some of the same regions involved in cingulate and insula regions that are important for
reflecting upon others’ emotional states are involved in stimulus-driven processing of social cues and their under-
reflecting upon our own emotions [6,26], consistent with lying intentions, in addition to a network of regions cen-
theories suggesting that in some cases we treat ourselves tered on the MPFC used to reflect upon and reason about
as an ‘other’ when making self-judgments [31]. The reverse them.
might also be true: we use information about our own
states and traits when we reflect upon the states and traits Emotional learning
of others who seem to be like ourselves. Thus, judgments The role of MSA in emotion is not limited to understanding
about known or similar, as compared with less familiar or emotions in the present moment but additionally helps us
dissimilar, others draw on medial frontal regions similar to to learn about emotion-eliciting events and also to
those used for self-referential processing [6,27]. Exactly form lasting impressions of others’ emotionally relevant

Figure 1. (a, b) Sagittal views of areas with increased connectivity with seed regions commonly active during self and other pain in the (a) AI and (b) ACC [15]. Activations
during the direct experience of pain are marked in red, and during the observation of others’ pain are marked in blue. Importantly, when observing others’ pain, activation in
the MPFC (circled) co-varied with activation in seed regions in both the AI {6,40,24}* and ACC {12,48,12}*. (c,d) Functional activation during an observational fear learning
task [35]; (c) shows a coronal view of activation in the right AI {–28, 15, –4}y when observing the pain response of a learning model to a shock. The adjacent graph shows that
the magnitude of this activation predicts the strength of the conditioned response (indexed by the skin conductance response) at a later time to a cue associated with the
pain of the learning model (P < 0.05). (d) A sagittal view of activation in the (1) MPFC {1,46,24}y and (2) ACC {3,27,32}y during the observation of the pain response of a
learning model to a shock. As in (c), adjacent graphs display the positive relationship between the magnitude of activation during observation and the subsequent
conditioned response (P = 0.08 and P < 0.05 for 1 and 2, respectively).z*x, y, z coordinates for local maxima in Montreal Neurological Institute (MNI) space; yx, y, z
coordinates for local maxima in Talairach space; zafter one individual who was an outlier was removed.

67
Review Trends in Cognitive Sciences Vol.12 No.2

dispositions. For example, watching another’s fear [50] or knowing that they are in pain [24] – perhaps
expression to an unfamiliar dog could provide valuable because of an expectation that the partner’s behavior will
information about potential danger, that individual’s be brought back in line with expected norms. It should be
anxious disposition or both. These abilities to learn from noted, however, that growing evidence suggests that the
and about others crucially depend on understanding amygdala and striatum are not simply ‘fear’ and ‘reward’
others’ emotions using both stimulus-driven and reflective systems, and instead are sensitive to the arousal-related
MSAs. and motivational properties of valenced stimuli in general
[51,52].
Learning from others The affective value of a social target is determined, of
Neuroscience models of emotional learning are based course, both by their dispositions and by situational vari-
primarily on studies of fear conditioning showing that ables [53]. An identical punch might be taken as aggressive
the amygdala is crucial for learning through direct experi- or playful, for example, depending on one’s assessment of
ence which stimuli predict aversive outcomes [33,34]. the puncher’s intent. Contextual information that precedes
Emotional learning can also take place through indirect or follows initial stimulus-driven assessments might flex-
means, however, such as verbal communication and obser- ibly constrain and update them using bidirectional inter-
vation. Although these social–emotional routes to learning connections between the amygdala, striatum and
are crucial to everyday life for most humans [35], their orbitofrontal cortex (OFC) [54–56]. For example, OFC
neural bases have only recently begun to be explored damage might cause an individual to be unable to appreci-
[36,37]. ate faux pas, or to consider that the amount they tease
A common example is observational fear learning in someone, or are praised by them, is contextually inap-
which one vicariously learns to fear a stimulus that elicits propriate [54,56,57]. In such situations, OFC damage
expressions of fear in a conspecific. Cross-species research impairs the ability quickly and implicitly to use MSAs
demonstrates that learning fear through observing others (e.g. my girlfriend is embarrassed. . .) to regulate
shares behavioral [38,39] and neural features with con- moment-to-moment behavior (. . .so I should stop teasing
ditioned fear, including involvement of the amygdala her). By contrast, rostral and dorsal MPFC might support
[37,40]. Although this suggests that low-level stimulus- explicit MSAs used when reflecting upon the intentions
driven processes have an important role in observational behind and consequences of actions, as evidenced by its
fear learning across species, recent work in humans activation during strategic games [50,58] and when for-
suggests that, in addition to explicit knowledge about giving others’ transgressions [59].
the stimulus contingencies, fear learning through obser- Taken together, current work suggests that both rapid
vation might also involve higher-level reflective MSAs. and reflective MSAs support emotional learning from and
Olsson et al. [37] found that overlapping regions of the about others. As the complexity of required MSAs
amygdala, AI and ACC were active during both observa- increases, processing might move from stimulus-driven
tional learning and subsequent expression of fear subcortical to orbitofrontal regions and on to rostral and
responses, whereas the dorsal MPFC was active only dorsal MPFC regions if processing becomes explicitly eva-
during observation of another’s distress. Importantly, luative or reflective.
the magnitude of the conditioned response was predicted
by activity in the AI, ACC and dorsal MPFC (Figure 1c,d). Regulation of emotional responses
This suggests that shared representations supporting The MSA processes used for emotion understanding and
experiential understanding of emotion – in addition to learning also enable us adaptively to regulate our own
regions supporting reflective MSAs – jointly support emotional responses. To date, studies have investigated
social–emotional learning (Figure 1c,d). primarily the use of higher-level MSAs to regulate emotion
in two ways.
Learning about others The first ‘situation focused’ or ‘other focused’ strategy
The two types of MSAs that support learning from others involves reinterpreting the situational meaning of others’
also provide crucial diagnostic information about others’ intentions or feelings, as when, for example, thinking
stable social–emotional dispositions. Consider a social positively or negatively about the dispositions (she is
interaction that unfolds over time. During the initial hearty or weak) and future emotions (she will feel fine
moments of contact, stimulus-driven systems might assess or get worse) experienced by someone who is sick and in
the affective value of social targets with varying degrees of pain [60–62]. Interestingly, recent work suggests that
complexity. For example, the simple perception or evalu- simply making an attribution about the feelings of another
ation of either attractive or potentially threatening and person can have the unintended consequence of disrupting
untrustworthy faces activates either striatal regions impli- amygdala-mediated negative evaluations of them [63,64].
cated in reward and reinforcement learning [41] or the One reason for this might be that MSAs can direct atten-
amygdala and AI regions implicated in aversive learning, tion to the nonthreatening intentions (e.g. thinking about
as described earlier [42–45]. With repeated interaction, their food preferences) of a social target, thereby disambig-
imaging studies suggest that the striatum might encode uating them as potential sources of threat.
which actions produce desired social outcomes, consistent The second ‘self-focused’ strategy involves mentally
with its general role in reinforcement learning [41,46]. manipulating one’s personal connection to an emotionally
Striatum activity accompanies both cooperation and trust charged situation, as when, for example, one experiences
[46–49] – but also punishing a previously unfair partner aversive events from the simulated perspective of an

68
 Review Trends in Cognitive Sciences Vol.12 No.2

Figure 2. A functional–anatomical organization of regions supporting the role of MSA in emotion. (a) Midline regions (e.g. MPFC) interconnected with emotion centers
support representations of internal states and might be coding emotional qualities of MSAs. By contrast, lateral regions [e.g. the lateral prefrontal cortex (LPFC) and TPJ]
interconnected with visuospatial centers support externally generated representations and might be coding cognitive aspects of MSA. (b) Closely aligned to the midline
section along with the insula (1), posterior regions [e.g. the posterior cingulate cortex (PCC), posterior insula (PI) and the mACC] support simple ‘first-order’ sensory aspects
of MSA, whereas representational complexity increases as the information is re-represented in more anterior regions (e.g. MPFC); (2) ventral regions [e.g. the amygdala (A),
OFC, striatum (Stri) and ventral MPFC (vMPFC)] are predominantly engaged in stimulus-driven processes, whereas dorsal regions [e.g. dorsal MPFC (dMPFC) and LPFC]
support performance monitoring and reflective processes of MSA. Abbreviation: cogn, cognitive; R, right.

objective and detached third person observer [65–67]. key role integrating information about the internal state
Although each strategy can modulate stimulus-driven of the body with higher-level mental state knowledge
amygdala activity, they seem to depend differentially upon needed to categorize affective states. Finally, as processing
complementary PFC regions: the MPFC – which receives moves from ventral to dorsal regions, it becomes less
direct inputs about the internal state of the body – might be stimulus driven and increasingly controlled and reflective,
relatively more important for self-focused regulation, enabling one to judge explicitly and be aware of one’s own
whereas other-focused regulation might depend more on or others’ emotional states. Given the early stages of
ventrolateral PFC regions used to select from memory research in this area, of course, the functional distinctions
information that helps to interpret others’ feelings [6]. drawn between medial-lateral, posterior-anterior and ven-
In sum, it seems that MSAs can regulate emotion either tral-dorsal regions are by necessity highly simplified and
through reinterpretations of others’ feelings or by simu- heuristic. It is expected that continued research will be
lating what it would be like to be experiencing the event crucial for their further specification.
from a third person perspective. These strategies might
differentially depend on medial and lateral regions of the Future directions
PFC, respectively. This survey of research on the neural bases of MSA in
emotion understanding, learning and regulation provides
A neural framework for the role of social cognition a framework for understanding current work but also
in emotion highlights how much there remains to be clarified in future
One way to understand the relationship between social research. At least four types of question will be important
cognition and emotion is to delineate the way in which the to address.
processes and mental representations underlying them are The first concerns the fact that research has focused on
distributed along general functional axes in the brain how the average person perceives emotions in static social
[30,68]. Based on the preceding review, we propose that stimuli, but less is known about how dispositional or
the role of social cognition – and MSA in particular – in situational motivations (e.g. to seek affiliation or fear
emotion can be understood in terms of three related but rejection, or be empathic versus competitive) might influ-
distinct dimensions of functional–anatomic organization. ence the neural systems supporting emotion understand-
The proposed framework should be viewed as complemen- ing, learning and regulation during real-time social
tary to previously formulated functional–anatomical interaction. Methods and approaches are now being devel-
models, which have attempted to capture a broader array oped to address these issues [48,69].
of social cognitive phenomena and have often done so along Second, an interactive approach will also be important
two dimensions [30,68]. According to the scheme proposed for the analysis of data, to clarify how different components
here (Figure 2, Box 3), lateral regions preferentially pro- of the functional networks supporting MSA work together
cess information about external inputs, whereas medial in social cognitive and emotional attributions. Indeed, the
regions have an essential role in MSA by representing fact that both simple and complex MSAs can be impaired
information about internal states [68]. Along with following damage to just one of the regions described ear-
the AI, posterior medial regions such as the mACC lier [54,70,71], and that each region might have general
represent body state information that is represented with functional roles beyond MSA [26,32], makes it clear that
increasing complexity as processing moves anteriorly understanding time- and context-varying interactions be-
towards the frontal pole and BA 10 [30]. BA 10 has a tween regions will be essential [15].

69
Review Trends in Cognitive Sciences Vol.12 No.2

schizophrenia to anxiety disorders and depression, are

Box 3. A three-dimensional neural framework for
characterized by impaired social cognition in addition to
understanding the role of MSA in social cognition and
emotion understanding, learning and regulation. Examin-
emotion
ing possible deficits in core neural systems supporting
The role of social cognition – and MSA in particular – in emotion can MSA might be crucial to understanding better the etiology
be broadly understood in terms of three related but distinct and possible therapies for these disorders.
dimensions of functional–anatomical organization that generally
follow patterns of inter-regional connectivity and neurodevelop-
mental trends in cortical evolution. Acknowledgements
The first concerns the medial–lateral axis (Figure 2a), with midline This work was supported by grants MH076137 from NIH and DA022541
regions interconnected with visceral centers and stimulus-driven from NIDA, and a Margaret and Herman Sokol Postdoctoral Fellowship
emotion centers such as the amygdala and striatum [79]. Whereas (AO).
the midline regions are important for representing information
about internal states, lateral regions interconnected with visuospa- References
tial centers are important for externally generated representations. 1 Tomasello, M. (1999) The Cultural Origins of Human Cognition,
In this view, medial and lateral regions might be more engaged by, Harvard University Press
respectively, the emotional and cognitive qualities of MSAs [51], 2 Ochsner, K.N. Social cognitive neuroscience: historical development,
such that attributions about feelings recruit the MPFC, and complex core principles, and future promise. In Social Psychology: A Handbook
elaborations of MSA dependent on abstract rules and retrieving of Basic Principles (Kruglanksi, A. and Higgins, E.T., eds) (in press)
information from memory recruit the LPFC [6,68,80]. Interestingly, 3 Adolphs, R. (2001) The neurobiology of social cognition. Curr. Opin.
similar functional–anatomical distinctions have been drawn pre- Neurobiol. 11, 231–239
viously in relation to both social cognitive processes [68] and other 4 Gallese, V. et al. (2004) A unifying view of the basis of social cognition.
nonsocial domains, such as response selection and motor behavior Trends Cogn. Sci. 8, 396–403
[81] and hedonic experiences [55]. 5 Preston, S.D. and de Waal, F.B. (2002) Empathy: its ultimate and
The second involves the midline posterior–anterior axis and proximate bases. Behav. Brain Sci. 25, 1–20
concerns the degree of complexity of MSA-related information 6 Ochsner, K.N. et al. (2004) Reflecting upon feelings: an fMRI study of
(Figure 2b), which in the frontal lobe follows a radial pattern of neural systems supporting the attribution of emotion to self and other.
cortical evolution, whereby a three-layered cingulate cortex devel- J. Cogn. Neurosci. 16, 1746–1772
oped into a six-layered rostral MPFC [82]. Posterior regions provide 7 Decety, J. and Jackson, P.L. (2004) The functional architecture of
the ‘first-order’ perceptual substrate of the socio-emotional proces- human empathy. Behav. Cogn. Neurosci. Rev. 3, 71–100
sing stream by initially representing nonverbal [superior temporal 8 Lamm, C. et al. (2007) The neural substrate of human empathy: effects
sulcus (STS)] and body state (posterior insula and middle ACC) of perspective-taking and cognitive appraisal. J. Cogn. Neurosci. 19,
cues. As this information is passed anteriorly along the insula and 42–58
MPFC, it is re-represented with increasingly complexity [30,31,80] in 9 Blakemore, S.J. and Decety, J. (2001) From the perception of action to
a ‘second-order’ format that supports awareness of, and explicit the understanding of intention. Nat. Rev. Neurosci. 2, 561–567
judgments about, the meaning of affective mental states. A similar 10 Carr, L. et al. (2003) Neural mechanisms of empathy in humans: a relay
distinction has been outlined by Amodio and Frith [30]. from neural systems for imitation to limbic areas. Proc. Natl. Acad. Sci.
The third dimension tracks the frontal midline from its ventral U. S. A. 100, 5497–5502
regions predominantly engaged in stimulus-driven processes to its 11 Iacoboni, M. and Dapretto, M. (2006) The mirror neuron system and
dorsal regions supporting performance monitoring and reflective the consequences of its dysfunction. Nat. Rev. Neurosci. 7, 942–951
processes (Figure 2b). At present, this division is perhaps the most 12 Morrison, I. et al. (2004) Vicarious responses to pain in anterior
speculative, and is intended to distinguish the amygdala, striatum cingulate cortex: is empathy a multisensory issue? Cogn. Affect.
and OFC, which support stimulus-driven MSAs, from dorsomedial Behav. Neurosci. 4, 270–278
and lateral PFC regions, which support explicit and reflective MSAs 13 Singer, T. et al. (2004) Empathy for pain involves the affective but not
[6,83]. sensory components of pain. Science 303, 1157–1162
Using this framework, it can be seen that different neural systems 14 Jackson, P.L. et al. (2005) How do we perceive the pain of others? A
might support MSAs, depending upon the task and processing window into the neural processes involved in empathy. Neuroimage 24,
demands: subcortical (amygdala and striatum) and posterior (STS, 771–779
mACC and insula) systems will support emotion understanding, 15 Zaki, J. et al. (2007) Different circuits for different pain: patterns of
learning and regulation unless it becomes necessary to engage functional connectivity reveal distinct networks for processing pain in
anterior medial and lateral PFC regions to categorize and reason self and others. Soc. Neurosci. 2, 276–291
about affective states [83]. 16 Critchley, H.D. et al. (2005) Anterior cingulate activity during error
and autonomic response. Neuroimage 27, 885–895
17 Craig, A.D. (2002) How do you feel? Interoception: the sense of the
Third, careful comparisons of human and non-human physiological condition of the body. Nat. Rev. Neurosci. 3, 655–666
18 Eisenberger, N.I. et al. (2003) Does rejection hurt? An FMRI study of
animal models of social–emotional behavior will be useful social exclusion. Science 302, 290–292
for understanding what processes uniquely characterize 19 Hutchison, W.D. et al. (1999) Pain-related neurons in the human
each species. In some areas of research – such as social– cingulate cortex. Nat. Neurosci. 2, 403–405
emotional learning – it remains to be seen whether strong 20 Wager, T.D. et al. (2004) Placebo-induced changes in FMRI in the
homologies between species belie the use of common MSA anticipation and experience of pain. Science 303, 1162–1167
21 Dum, R.P. and Strick, P.L. (2002) Motor areas in the frontal lobe of the
processes or a reliance on variants of the same basic primate. Physiol. Behav. 77, 677–682
Observational Fear Learning paradigms. In other areas 22 Kim, H. et al. (2004) Contextual modulation of amygdala responsivity
of work – such as the use of MSA to regulate emotion – the to surprised faces. J. Cogn. Neurosci. 16, 1730–1745
mapping might be complicated because humans are 23 Lanzetta, J.T. and Englis, B.G. (1989) Expectations of cooperation and
capable of complex attributions about mental states that competition and their effects on observers’ vicarious emotional
responses. J. Pers. Soc. Psychol. 56, 543–554
might be difficult to capture in animal models. 24 Singer, T. et al. (2006) Empathic neural responses are modulated by the
Fourth, basic work on the role of social cognition in perceived fairness of others. Nature 439, 466–469
emotion is particularly well suited for translational 25 Batson, C.D. et al. (2002) Moral hypocrisy: addressing some
efforts. Myriad psychological disorders, ranging from alternatives. J. Pers. Soc. Psychol. 83, 330–339

70
Review Trends in Cognitive Sciences Vol.12 No.2

26 Saxe, R. et al. (2006) Overlapping and non-overlapping brain regions 56 Stone, V.E. et al. (1998) Frontal lobe contributions to theory of mind.
for theory of mind and self reflection in individual subjects. Soc. Cogn. J. Cogn. Neurosci. 10, 640–656
Affect. Neurosci. 1, 229–234 57 Shamay-Tsoory, S.G. et al. (2003) Characterization of empathy deficits
27 Mitchell, J.P. et al. (2006) Dissociable medial prefrontal contributions following prefrontal brain damage: the role of the right ventromedial
to judgments of similar and dissimilar others. Neuron 50, 655–663 prefrontal cortex. J. Cogn. Neurosci. 15, 324–337
28 Gilbert, S.J. et al. (2006) Functional specialization within rostral 58 Gallagher, H.L. et al. (2002) Imaging the intentional stance in a
prefrontal cortex (area 10): a meta-analysis. J. Cogn. Neurosci. 18, competitive game. Neuroimage 16, 814–821
932–948 59 Farrow, T.F. et al. (2001) Investigating the functional anatomy of
29 Jackson, P.L. et al. (2006) Empathy examined through the neural empathy and forgiveness. Neuroreport 12, 2433–2438
mechanisms involved in imagining how I feel versus how you feel 60 Kim, S.H. and Hamann, S. (2007) Neural correlates of positive and
pain. Neuropsychologia 44, 752–761 negative emotion regulation. J. Cogn. Neurosci. 19, 776–798
30 Amodio, D.M. and Frith, C.D. (2006) Meeting of minds: the medial 61 Urry, H.L. et al. (2006) Amygdala and ventromedial prefrontal cortex
frontal cortex and social cognition. Nat. Rev. Neurosci. 7, 268–277 are inversely coupled during regulation of negative affect and
31 Ochsner, K.N. et al. (2005) The neural correlates of direct and reflected predict the diurnal pattern of cortisol secretion among older adults.
self-knowledge. Neuroimage 28, 797–814 J. Neurosci. 26, 4415–4425
32 Mitchell, J.P. (2007) Activity in right temporo-parietal junction is not 62 Phan, K.L. et al. (2005) Neural substrates for voluntary suppression of
selective for theory-of-mind. Cereb. Cortex, DOI: 10.1093/cercor/ negative affect: a functional magnetic resonance imaging study. Biol.
bhm051 (http://cercor.oxfordjournals.org/) Psychiatry 57, 210–219
33 Davis, M. and Whalen, P.J. (2001) The amygdala: vigilance and 63 Lieberman, M.D. et al. (2007) Putting feelings into words: affect
emotion. Mol. Psychiatry 6, 13–34 labeling disrupts amygdala activity in response to affective stimuli.
34 LeDoux, J.E. (2000) Emotion circuits in the brain. Annu. Rev. Neurosci. Psychol. Sci. 18, 421–428
23, 155–184 64 Harris, L.T. and Fiske, S.T. (2007) Social groups that elicit disgust are
35 Bandura, A. (1977) Social Learning Theory, General Learning Press differentially processed in mPFC. Soc. Cogn. Affect. Neurosci. 2, 45–51
36 Olsson, A. and Phelps, E.A. (2007) Social learning of fear. Nat. 65 Ochsner, K.N. et al. (2004) For better or for worse: neural systems
Neurosci. 10, 1095–1102 supporting the cognitive down- and up-regulation of negative emotion.
37 Olsson, A. et al. (2007) Learning fears by observing others: the neural Neuroimage 23, 483–499
systems of social fear transmission. Soc. Cogn. Affect. Neurosci. 2, 3–11 66 Kalisch, R. et al. (2005) Anxiety reduction through detachment:
38 Mineka, S. and Cook, M. (1993) Mechanisms involved in the subjective, physiological, and neural effects. J. Cogn. Neurosci. 17,
observational conditioning of fear. J. Exp. Psychol. Gen. 122, 23–38 874–883
39 Olsson, A. and Phelps, E.A. (2004) Learned fear of ‘unseen’ faces after 67 Beauregard, M. (2007) Mind does really matter: evidence from
Pavlovian, observational, and instructed fear. Psychol. Sci. 15, 822–828 neuroimaging studies of emotional self-regulation, psychotherapy,
40 Hooker, C.I. et al. (2006) Amygdala response to facial expressions and placebo effect. Prog. Neurobiol. 81, 218–236
reflects emotional learning. J. Neurosci. 26, 8915–8922 68 Lieberman, M.D. (2007) Social cognitive neuroscience: a review of core
41 O’Doherty, J. et al. (2004) Dissociable roles of ventral and dorsal processes. Annu. Rev. Psychol. 58, 259–289
striatum in instrumental conditioning. Science 304, 452–454 69 Kross, E. et al. (2007) Neural dynamics of rejection sensitivity. J. Cogn.
42 Adolphs, R. et al. (1998) The human amygdala in social judgment. Neurosci. 19, 945–956
Nature 393, 470–474 70 Heberlein, A.S. and Adolphs, R. (2004) Impaired spontaneous
43 Cunningham, W.A. et al. (2003) Neural components of social anthropomorphizing despite intact perception and social knowledge.
evaluation. J. Pers. Soc. Psychol. 85, 639–649 Proc. Natl. Acad. Sci. U. S. A. 101, 7487–7491
44 Todorov, A. et al. (2006) Toward socially inspired social neuroscience. 71 Adolphs, R. et al. (2002) Impaired recognition of social emotions
Brain Res. 1079, 76–85 following amygdala damage. J. Cogn. Neurosci. 14, 1264–1274
45 Winston, J.S. et al. (2002) Automatic and intentional brain responses 72 Fiske, S.T. (1998) Stereotyping, Prejudice, and Discrimination,
during evaluation of trustworthiness of faces. Nat. Neurosci. 5, 277–283 McGraw-Hill
46 Delgado, M.R. (2007) Reward-related responses in the human 73 Forgas, J.P. et al., eds (2003) Social Judgments: Implicit and Explicit
striatum. Ann. N.Y. Acad. Sci. 1104, 70–88 Processes, Cambridge University Press
47 Delgado, M.R. et al. (2005) Perceptions of moral character modulate the 74 Ekman, P. (1999) Basic emotions. In The Handbook of Cognition and
neural systems of reward during the trust game. Nat. Neurosci. 8, Emotion (Dalgleish, T. and Power, T., eds), pp. 45–60, John Wiley &
1611–1618 Sons, Ltd
48 King-Casas, B. et al. (2005) Getting to know you: reputation and trust 75 Scherer, K.R. et al., eds (2001) Appraisal Processes in Emotion: Theory,
in a two-person economic exchange. Science 308, 78–83 Methods, and Research, Oxford University Press
49 Sanfey, A.G. et al. (2006) Neuroeconomics: cross-currents in research 76 Dennett, D.C. (1987) The Intentional Stance. MIT Press
on decision-making. Trends Cogn. Sci. 10, 108–116 77 Baumeister, R.F. (1998) The Self, McGraw-Hill
50 de Quervain, D.J. et al. (2004) The neural basis of altruistic 78 Saxe, R. et al. (2004) Understanding other minds: linking
punishment. Science 305, 1254–1258 developmental psychology and functional neuroimaging. Annu. Rev.
51 Phan, K.L. et al. (2002) Functional neuroanatomy of emotion: a meta- Psychol. 55, 87–124
analysis of emotion activation studies in PET and fMRI. Neuroimage 79 Ongur, D. et al. (2003) Architectonic subdivision of the human orbital
16, 331–348 and medial prefrontal cortex. J. Comp. Neurol. 460, 425–449
52 Seymour, B. et al. (2007) The neurobiology of punishment. Nature Rev. 80 Burgess, P.W. et al. (2007) The gateway hypothesis of rostral prefrontal
Neurosci. 8, 300–311 cortex (area 10) function. Trends Cogn. Sci. 11, 290–298
53 Gilbert, D.T. (1998) Ordinary personology. In The Handbook of Social 81 Thaler, D. (1995) The functions of the medial premotor cortex. I. Simple
Psychology (4th edn) (Gilbert, D.T. et al., eds), pp. 89–150, McGraw Hill learned movements. Exp. Brain Res. 102, 445–460
54 Beer, J.S. et al. (2003) The regulatory function of self-conscious 82 Paus, T. (2001) Primate anterior cingulate cortex: where motor control,
emotion: insights from patients with orbitofrontal damage. J. Pers. drive and cognition interface. Nat. Rev. Neurosci. 2, 417–424
Soc. Psychol. 85, 594–604 83 Lieberman, M.D. et al. (2002) Reflection and reflexion: a social cognitive
55 Kringelbach, M.L. (2005) The human orbitofrontal cortex: linking neuroscience approach to attributional inference. Adv. Exp. Soc.
reward to hedonic experience. Nat. Rev. Neurosci. 6, 691–702 Psychol. 34, 199–249

71