Escolar Documentos
Profissional Documentos
Cultura Documentos
The phenomenon of empathy entails the ability to share the affective experiences of
others. In recent years social neuroscience made considerable progress in revealing the
mechanisms that enable a person to feel what another is feeling. The present review pro-
vides an in-depth and critical discussion of these findings. Consistent evidence shows
that sharing the emotions of others is associated with activation in neural structures
that are also active during the first-hand experience of that emotion. Part of the neural
activation shared between self- and other-related experiences seems to be rather auto-
matically activated. However, recent studies also show that empathy is a highly flexible
phenomenon, and that vicarious responses are malleable with respect to a number
of factors—such as contextual appraisal, the interpersonal relationship between em-
pathizer and other, or the perspective adopted during observation of the other. Future
investigations are needed to provide more detailed insights into these factors and their
neural underpinnings. Questions such as whether individual differences in empathy
can be explained by stable personality traits, whether we can train ourselves to be more
empathic, and how empathy relates to prosocial behavior are of utmost relevance for
both science and society.
Key words: empathy; social neuroscience; pain; fMRI; anterior insula (AI); anterior
cingulate cortex (ACC); prosocial behavior; empathic concern, altruism; emotion
contagion
81
82 Annals of the New York Academy of Sciences
“empathy,” is only one part of a large spec- nitive mechanisms enable us to “put ourselves
trum of a person’s possible vicarious responses into someone else’s shoes.” Surprisingly, it took
toward others. In the second part, current ev- quite some time for the field of neuroscience
idence from social neuroscience studies of em- and in particular of functional neuroimaging
pathy will be selectively reviewed. This re- to dare to make contributions to this challeng-
view will focus on a discussion of the shared ing pursuit (Carr et al. 2003; Keysers et al.
representations account of empathy, which is 2004; Morrison et al. 2004; Singer et al. 2004;
currently the dominant neuroscientifically mo- Wicker et al. 2003). This might be attributed
tivated approach to understanding the mech- to the complexities inherent in this multidi-
anisms underlying empathy. In addition, we mensional psychological phenomenon as well
will reflect on the role of bottom-up and top- as to the methodological challenges of bringing
down influences in generating and modulat- such an idiosyncratic and context-dependent
ing empathic responses and discuss the dif- phenomenon into a scientific environment
ferent motivational consequences carried by that requires well-controlled and reproducible
the vicarious responses of empathy versus per- experiments.
sonal distress. In the final part, we propose At a basic phenomenological level, empa-
research questions and domains that should thy denotes an affective response to the directly
be given special attention by future empathy perceived, imagined, or inferred feeling state
researchers. of another being (for an excellent overview
of the various “things called empathy,” see
Batson 2009). In our own understanding, em-
Definition of Terms: Empathy pathy occurs when an observer perceives or
and Its Sisters imagines someone else’s (i.e., the target’s) affect
and this triggers a response such that the ob-
Despite the word’s linguistic roots in ancient server partially feels what the target is feeling.
Greek—from empatheia (passion), which is com- De Vignemont and Singer (2006, p. 435), for
posed of “en” (in) and “pathos” (feeling)—the example, define empathy as follows: We “em-
scientific scrutiny of empathy has a relatively pathize” with others when we have (1) an af-
short history that can be dated back to its use fective state (2) which is isomorphic to another
in philosophical aesthetics. From there the En- person’s affective state, (3) which was elicited
glish term originated as a direct translation of by observing or imagining another person’s af-
the German Einfühlung (“feeling into” some- fective state, and (4) when we know that the
thing), a term that was originally proposed as other person’s affective state is the source of
a tool for analyzing works of art and nature, our own affective state. However, there are al-
but later developed into a more general mech- most as many definitions of empathy as there
anism for recognizing each other as “minded are researchers in the field. It is therefore es-
creatures” (Stüber 2008). After this initial philo- sential to briefly review and define the rele-
sophically motivated period of scrutiny, most vant key concepts and components generally
of the research on empathy was performed associated with the broad concept of empathy,
by developmental and social psychologists (e.g., such as mimicry, emotional contagion, sym-
Batson 1991, 2009; Hoffman 2000; Eisenberg pathy, and compassion. Although these con-
2000; Eisenberg & Strayer 1987). Naturally, so- cepts each refer to a different phenomenon,
cial psychologists showed particular interest in they usually occur in concert. In most cases,
this inherently social phenomenon (e.g., Batson mimicry or emotional contagion precedes em-
1991; Davis 1994). A great deal of social psy- pathy, which precedes sympathy and compas-
chological research has been devoted to the sion, which in turn may precede prosocial
question which perceptual, affective, and cog- behavior.
Singer & Lamm: Social Neuroscience of Empathy 83
The Big and Little Sisters of Empathy communicative rather than for epistemological
reasons (Chartrand & Bargh 1999; van Baaren
First, we need to distinguish between mimicry et al. 2004).
and emotional contagion, which can both con- Emotional contagion is another process that
tribute substantially to an empathic response. is related to but distinct from empathy. It
Mimicry is defined as the tendency to automat- denotes the tendency to “catch” other peo-
ically synchronize affective expressions, vocal- ple’s emotions and has alternately been labeled
izations, postures, and movements with those “primitive empathy” (Hatfield et al. 2009). For
of another person (Hatfield et al. 1994). Our example, babies start crying when they hear
understanding of its role as a low-level mech- other babies crying—long before they develop
anism contributing to empathy derives from a a sense of a self separate from others. Re-
multitude of studies using facial electromyog- cently, initial evidence for involuntary pupil-
raphy. These studies demonstrate that when an lary contagion was found in an fMRI study
observer perceives another person’s affective fa- (Harrison et al. 2006). Participants were pre-
cial expressions, such as a smile or a frown, sented with photos of sad faces with various
corresponding affective expressions result in pupil sizes. Their own pupil size was signifi-
the observer (see Dimberg & Oehman 1996, cantly smaller when they viewed sad faces with
for a review). Relying on the facial feedback small as compared to larger pupils, and the
hypothesis, according to which one appraises Edinger−Westphal nucleus in the brainstem,
one’s own emotions by perceiving their bodily which controls pupil size, was specifically en-
concomitants, Sonnby-Borgstrom (2002) pro- gaged by this contagious effect. Activation in
posed that mimicry enables one to automati- this subcortical structure provides evidence that
cally share and understand another’s emotions. pupillary contagion occurs outside of aware-
Her proposal also receives support from studies ness and may represent a precursor of empathy.
showing a (notably, weak) correlation between This study also demonstrates the strong over-
the strength of the mimicry response and trait lap between mimicry and emotional contagion
measures of empathy. (e.g., Hatfield et al. 1993). Note, however, that
However, facial responses are not only trig- there are cases in which mimicry occurs with-
gered when we observe others, but also when out an emotional component and other cases
we are exposed to negatively or positively va- in which emotions are automatically elicited by
lenced visual stimuli without social relevance observing others’ emotional states without the
(Cacioppo et al. 2000, for a review). In addition, involvement of motor mimicry.
several investigations demonstrate the influence On a conceptual level, neither emotional
of top-down processes on mimicry, such as contagion nor mimicry cannot account for the
those associated with the relationship between full-blown experience of empathy. In our un-
empathizer and target (Lakin & Chartrand derstanding empathy crucially depends upon
2003), the affective state of the observer (Moody self-awareness and self/other distinction; in
et al. 2007; Niedenthal et al. 2001), or the per- other words, on our ability to distinguish be-
spective from which pain in others is witnessed tween whether the source of our affective ex-
(Lamm et al. 2008). These observations cast perience lies within ourselves or was triggered
doubt on the assumption that mimicry repre- by the other (de Vignemont & Singer 2006;
sents some sort of automatic or “hard-wired” Decety & Jackson 2004; Decety & Lamm 2006).
motor resonance with another person’s affec- Without this ability, witnessing someone else’s
tive display. Furthermore, mimicry seems to emotions could, for example, result, purely, in
serve a social function in increasing rapport personal distress and a self-centered response in
and fondness between self and other, raising the observer. We therefore regard mimicry and
the question whether this function evolved for emotional contagion as important, yet distinct
84 Annals of the New York Academy of Sciences
and neither necessary nor sufficient processes empathy in order to sense how to increase his
for the experience of empathy. victim’s suffering; in competitive environments
On the level of vicariously felt responses, we (from sports to business operations to warfare),
have to distinguish between empathy, sympa- successful tactics take into account the negative
thy, empathic concern, and compassion. In all affective effects that an action will have on the
four cases, affective changes are induced in the opponent; and experiencing too much empa-
observer in response to the perceived or imag- thy can lead to an aversive distress response and
ined affective state of another person. How- selfish instead of other-oriented behavior (see
ever, while empathy involves feelings that are below). In general, however, empathy is con-
isomorphic to those of the other person, sym- ceived to be a first necessary step in a chain that
pathy, empathic concern, and compassion do begins with affect sharing, followed by under-
not necessarily involve shared feelings. For ex- standing the other person’s feelings, which then
ample, empathizing with a person feeling sad motivates other-related concern and finally en-
will result in a feeling of sadness in the self, gagement in helping behavior. Empathy and
whereas sympathizing with, being empathically prosocial behavior are thus closely linked on a
concerned, or feeling compassion for a sad per- conceptual level. Notably, while consistent ev-
son will result in either pity or compassionate idence for the link between “feeling for” (em-
love for the person, but not sadness. Also, when pathic concern) and prosocial behavior exists
an observer notices that someone is jealous of (e.g., Batson 1991; Eisenberg 2000; Eisenberg
him, he will most likely not start feeling jeal- et al. 1989), a clearcut empirical demonstra-
ous himself—though he might show sympa- tion of a link between empathy and prosocial
thy or compassion for the jealous person (for behavior is still missing.
similar arguments, see de Vignemont & Singer
2006). Note also that the terms empathic concern,
sympathy, and compassion have sometimes been Empathy Research in the Context
treated as synonymous (cf. Batson 2009) and of Social Neuroscience: The Shared
that the most widely used psychometric mea- Network Hypothesis
sures of empathic concern (see below) involve
self-report of compassionate, sympathetic, or Various scholars have proposed that we come
tender feelings. Therefore, the crucial distinc- to understand the actions, sensations, and emo-
tion between the term empathy and those like tions of others by the activation of neural
sympathy, empathic concern, and compassion is that representations corresponding to those states.
empathy denotes that the observer’s emotions Inspired by earlier perception−action models
reflect affective sharing (“feeling with” the other in the domain of action understanding (Prinz
person) while compassion, sympathy, empathic 2005), Preston and de Waal (2002) proposed
concern denote that the observer’s emotions a neuroscientific model of empathy, one which
are inherently other oriented (“feeling for” the suggests that observing or imagining another
other person). person in a particular emotional state auto-
Further, folk psychological accounts usually matically activates a representation of that state
relate the occurrence of empathy to proso- in the observer, along with its associated auto-
cial and altruistic, other-oriented motivations nomic and somatic responses. Other authors
(i.e., a motivation with the goal to increase the have also suggested that shared neural repre-
other person’s well-being or welfare). Empathy, sentations play a general role in understand-
though, does not necessarily carry such moti- ing other people’s mental states. They claim
vations, and real-life examples of how empa- that shared representations provide us with a
thy can “go awry” (from a prosocial point of simulation of their corresponding sensorimo-
view) abound. For example, a torturer may use tor, affective, or mental states (Gallese 2003a;
Singer & Lamm: Social Neuroscience of Empathy 85
Goldman 2006). Such accounts hold that the 2000), which consists of the brain areas in-
capacity to project ourselves imaginatively into volved in the processing of pain, were activated
another person’s perspective by simulating their when participants experienced pain themselves
mental activity using our own mental appara- as well as when they saw a signal indicating
tus lies at the root of our mature mind-reading that their loved one would experience pain.
abilities, and the reasoning of these accounts These areas—in particular, bilateral anterior
has been extended to the domains of actions insula (AI), the dorsal anterior cingulate cortex
and feelings: To understand what another per- (ACC), brain stem, and the cerebellum—are
son is doing we simulate his movements using involved in the processing of the affective com-
our own motor program; to understand what ponent of pain; in other words, they encode
another person is feeling, we simulate his feel- how unpleasant or aversive the subjectively felt
ings using our own affective programs (see also pain is. Thus, both the firsthand experience of
Keysers & Gazzola 2006). Indeed, this so-called pain and the knowledge that a beloved partner
shared representations account of social inter- is experiencing pain activates the same affective
action and intersubjectivity has become the brain circuits—suggesting that our own neural
dominant explanation of the hemodynamic ac- response reflects our partner’s negative affect.
tivation patterns observed in recent fMRI stud- This initial finding of shared neural activa-
ies of empathy. tion between self and other has been replicated
The majority of social neuroscience studies and extended using a variety of paradigms and
on empathy used the observation of pain in oth- methods. While initially evidence was restricted
ers as a model paradigm to evoke empathic re- to the affective component of pain—as indi-
sponses (de Vignemont & Singer 2006; Decety cated by activation restricted to areas involved
& Lamm 2006; Singer & Leiberg 2009; see, in coding the affective−motivational aspects of
e.g., Jabbi et al. 2007 or Wicker et al. 2003, for the feeling of pain—there is now evidence that
examples using other emotional states). One areas associated with somatosensory processing
common finding of these investigations is that can also be activated when we witness another
vicariously experiencing pain activates part of person’s pain, in particular, when our attention
the neural network that is also activated when is directed to the somatosensory aspects of the
we are in pain ourselves. For example, Singer pain experience (e.g., Bufalari et al. 2007;
and colleagues (2004) recruited couples and Cheng et al. 2008; Lamm & Decety 2008;
measured hemodynamic responses triggered by Lamm et al. 2007b). Bufalari and colleagues
painful stimulation of the female partner via an (2007), for example, demonstrated that the am-
electrode attached to her right hand. In another plitude of an event-related potential compo-
condition the same painful stimulation was ap- nent known to be generated in primary somato-
plied to the male partner who was seated next sensory cortex (P45) is modulated by seeing a
to the MRI scanner and whose hand could be needle piercing another person’s hand. In a si-
seen via a mirror system by the female part- milar vein, a recent fMRI study showed that
ner lying in the scanner. Differently colored (contralateral) right primary somatosensory co-
flashes of light on a screen pointed to either rtex was activated when participants wit-
the male or the female partner’s hand, indi- nessed another person’s left hand being
cating which of them would receive painful pierced (Lamm & Decety 2008; Lamm et al.
stimulation. This procedure enabled the mea- 2009). This activation overlapped with indivi-
surement of pain-related brain activation when dually determined somatosensory represen-
pain was applied either to the scanned subject tations of touch of the hand that had
(firsthand experience of pain) or to her part- been determined in a separate localizer run.
ner (empathy for pain). The results suggest that The latter finding is an important contri-
parts of the so-called pain matrix (Derbyshire bution to the fMRI literature, which has
86 Annals of the New York Academy of Sciences
only when participants were experiencing pain designs (Grill-Spector & Malach 2001; Henson
themselves (Lamm et al. 2007a; Singer 2006; & Rugg 2003). Such an approach has already
Singer et al. 2004), whereas activity in AI was been used in the domain of action observation
observed when participants were experiencing (Dinstein et al. 2007) and mentalizing (Jenkins
pain themselves and when vicariously feeling et al. 2008), and similar studies are now under-
someone else’s pain. In a similar vein, instruc- way in the domain of empathy research. These
tions to imagine pain from a first-person per- studies will show whether activation overlaps in
spective or to specifically assess the somatosen- the observer result from the activation of the
sory consequences of vicariously perceived pain same neural resources.
revealed more posterior activation clusters in Finally, a word of caution is required as to
the insular cortex (Jackson et al. 2006a; Lamm whether the observed involvement of AI or
et al. 2007a,b). ACC is specifically related to empathy or to
The important role of anterior insular cor- emotional contagion (or both). One indication
tex for empathy has also been demonstrated that shared activations during empathy for pain
in other domains. AI is known to be crucially are not simply related to emotional contagion
involved in processing sensations and emo- is provided by studies of Singer et al. (2004,
tions such as taste or disgust. Similar neu- 2006). Here, arbitrary visual cues rather than
ral responses were elicited when participants explicit affective displays or other potentially
viewed movies of disgusted faces and when they contagious stimuli such as depictions of painful
smelled disgusting odors themselves (Wicker events were used to assess empathic responses.
et al. 2003) as well as when they viewed videos In addition, a recent fMRI study by Lamm
showing people sampling pleasant or unpleas- et al. (2009) detected substantial responses in
ant tastes and when they sampled the different AI and ACC when participants were presented
tastes themselves (Jabbi et al. 2007). Interest- with visual stimuli depicting situations that were
ingly, the latter study also revealed that brain clearly nonpainful for them (touch of another
responses in anterior insular cortex were posi- person’s hand by a Q-tip), but known to be
tively correlated with individual differences in painful for the target. Together, these studies
empathy measured by empathy questionnaires. render it less plausible that their neural re-
The observation of similar neural activations sponses were directly triggered by the percep-
during the firsthand vs. the vicarious experi- tion of overt (and emotionally contagious) cor-
ence of various sensations and emotions (e.g., relates of the other person’s emotional state.
disgust, taste, pain) raises the question whether Nevertheless, additional studies are required to
these activations can indeed be interpreted as investigate how the brain actually differentiates
shared representations. Shared activations are between self- and other-related feelings and to
certainly a good indicator of shared represen- dissociate the respective contributions of the
tations. However, apart from the fact that we two related phenomena.
do not know how observations on one level In sum, while substantial empirical evidence
(the psychological/representational level) are suggests that shared neural activations are at
mapped to those on another (the neural one), the root of sharing feelings, sensations, and ac-
none of the currently available human neu- tions of others, additional research is required
roimaging methods directly measures activity to clarify whether these activations are actu-
of single neurons or neural networks (how- ally shared on the neural level, to what extent
ever, see Hutchison et al. 1999). Therefore, two we share both the somatosensory antecedents
fMRI activation maps with overlapping clus- or only the affective consequences of another’s
ters might still result from differing neural ac- affective state, what constitutes the functional
tivity. One way to resolve this problem is the significance of these shared activations, and
use of repetition-priming or fMRI-adaptation how we can distinguish between emotional
88 Annals of the New York Academy of Sciences
contagion vs. empathy on the neural level. tributed to attention, inhibitory, and other ex-
More sophisticated experimental designs, an- ecutive control processes also documents that
alytical approaches, as well as closer col- empathy is not a purely sensory-driven process
laborations between philosophers of mind, in which affective states are induced in the ob-
neurophysiologists, introspection-based con- server solely by means of bottom-up processes.
templatives, cognitive scientists, and social neu- On the contrary, it has long been argued that
roscientists are required to yield more detailed contextual appraisal, cognitive processes, and
answers to these questions. top-down control are important constituents of
human empathy. For example, in the eighteenth
Empathy—Bottom-Up vs. Top-Down century, the philosopher and economist Adam
Processes Smith proposed that imagination enables us to
project ourselves into the place of other persons,
An important aspect of most neuroscien- experiencing sensations that are generally sim-
tifically motivated models of empathy is that ilar to, although typically weaker than, those of
the activation of shared representations in the the other person (Smith 1759/1976). Most con-
observer is initiated mostly automatically and temporary neuroscientific models of empathy
without conscious awareness. For example, in endorse this view and stress the importance of
most studies of empathy in the domains of top-down control and contextual appraisal for
pain, touch, and disgust, participants are not either the generation of an empathic response
informed that the goal of the study is to investi- or for modulating an existing one induced by
gate empathy-related neural responses. Rather, the above-mentioned bottom-up processes (de
they are instructed to passively watch a scene or Vignemont & Singer 2006; Decety & Lamm
movie depicting a person expressing an emo- 2006; Hein & Singer 2008; Singer 2006).
tion or being touched (Blakemore et al. 2005; Decety and Lamm (2006), for example, pro-
Keysers et al. 2004; Singer et al. 2004, 2006; posed a model in which bottom-up (i.e., direct
Wicker et al. 2003). Nevertheless, this situa- matching between perception and action) and
tion alone is sufficient to engage brain networks top-down (i.e., regulation, contextual appraisal,
representing the firsthand experience of affect and control) information processes are fun-
or touch. Some authors have therefore sug- damentally intertwined in the generation and
gested that we automatically share other peo- modulation of empathy. In this model, bottom-
ple’s feelings, a hypothesis in line with earlier up processes account for direct emotion sharing
perception−action models of motor behavior which is automatically activated (unless inhib-
and imitation and with their extension to the ited) by perceptual input. On the other end, ex-
domain of empathy (Gallese 2003b; Preston & ecutive functions implemented in the prefrontal
de Waal 2002). Preston and de Waal’s model and cingulate cortex serve to regulate both cog-
(2002), for example, stresses the importance of nition and emotion through selective attention
automatic and perceptually driven processes and self-regulation. This meta-cognitive level is
such as emotional contagion and mimicry. It continuously updated by bottom-up informa-
should be noted though that the term automatic tion, and in return controls the lower level by
in this case refers to a process that does not providing top-down feedback. Thus, top-down
require conscious and effortful processing but regulation, through executive functions, mod-
can be inhibited or controlled (cf. Bargh 1994). ulates lower levels and adds flexibility, mak-
In addition, attention to the target’s affective ing the individual less dependent on exter-
state is required for triggering the postulated nal cues. The meta-cognitive feedback loop
cascade of events starting with emotional con- also plays a crucial role in taking into account
tagion and, ultimately, resulting in a full-blown one’s own mental competence in order to react
empathic experience. The crucial role that is at- (or not) to the affective states of others. This
Singer & Lamm: Social Neuroscience of Empathy 89
model should be supplemented by top-down study the same group demonstrated that it is
processes that are not classically associated with mainly a late event-related potential compo-
executive function and its associated neural nent that differs between the attention and no-
structures, in particular those in the medial attention conditions, while an early component
and dorsolateral prefrontal cortex. Mental im- is not affected by the focus of attentional set
agery, for example, has been shown to result in (Fan & Han 2008). The early response might
shared representations in both the motor and correspond to the postulated automatic activa-
the sensory-affective domain without consider- tion of shared representations, representations
able prefrontal activation (e.g., Decety & Grezes which might then be interpreted by later neural
2006; Ogino et al. 2007). In addition, con- processes either inhibiting or augmenting the
textual appraisal and target evaluation in behavioral relevance of the initial affective res-
empathy-for-pain paradigms predominantly onance. Surprisingly, however, only the early
activated areas such as the orbitofrontal cortex response correlated with subjective reports of
or the ventral striatum (e.g., Lamm et al. 2007b; the amount of perceived pain in the other per-
Singer et al. 2006), in other words, structures son as well as the unpleasantness experienced
involved in affective evaluation, reward, and by the observer.
punishment. Recent results also support the assumption
Several recent findings provide strong sup- that the contextual appraisal of a situation
port for top-down influences on empathy. rather than its sensory input alone determines
These findings document the flexibility of the the empathizer’s neural and behavioral re-
human mind in responding to others and sponse. In one fMRI study, participants viewed
show that empathy is not an all-or-none phe- pictures of needle injections and of tissue biop-
nomenon. Here, we want to focus on the role sies performed on human hands—with the lat-
of attention, contextual appraisal, perspective- ter consisting of the insertion of a biopsy needle
taking, and the relationship between em- into an anesthetized hand (Lamm et al. 2007b).
pathizer and target as salient examples of how Hence, the actual painful consequences of two
top-down control affects empathic responses basically identical visual stimuli had to be taken
(see Hein & Singer 2008 for a more exhaus- into account in order to show appropriate vicar-
tive account). ious responses. The neural structures support-
Two recent fMRI studies and an event- ing this appraisal process included the medial
related potential study demonstrate that the dorsal and orbitofrontal cortex (OFC) and the
way we attend to the emotions of others sub- right temporoparietal junction (rTPJ). Involve-
stantially modulates our empathic responses to ment of the OFC was associated with reeval-
them. In one study participants were exposed uating the valence of the seemingly aversive,
to pictures of painful situations—such as cut- yet actually neutral biopsy stimulus. Activation
ting one’s finger or getting one’s hand caught of the rTPJ and dorsal medial prefrontal cor-
in the door. When evaluating the painful conse- tex, however, was linked to self/other distinc-
quences of those situations, participants showed tion and self-awareness. These capacities en-
activation in large parts of the pain matrix— abled the observers to distinguish between a
a now typical response given the evidence probably automatically triggered, self-centered
reviewed above. However, when participants response to the aversive stimulus and the knowl-
were instructed to count the number of hands edge that such a response was actually not
shown in a picture—an experimental manipu- appropriate given the contextual information
lation that was devised to distract their attention about the biopsy’s affective consequences. No-
from the inflicted pain—no activation in insular tably, a recent meta-analysis of neuroimaging
or cingulate cortices was observed (Gu & Han studies of attention, theory of mind, the expe-
2007). In a follow-up electroencephalographic rience of agency, and empathy demonstrated
90 Annals of the New York Academy of Sciences
that all these phenomena resulted in largely rior insula, the amygdala, as well as various
overlapping activations in the rTPJ (Decety & structures involved in action control. Imagining
Lamm 2007). This finding also raises awareness oneself in a potentially harmful situation might
that, in many cases, so-called high-level phe- therefore trigger a stronger aversive response
nomena can be explained more parsimoniously than imagining someone else in the same sit-
by underlying low-level processes—such as the uation. In addition, it might engage systems
orienting of attention to external cues, or, as that help the observer to evaluate the affective
recently proposed, the switching between ex- meaning of similar situations, rendering these
ternally and internally oriented modes of infor- representations a basis for future projections
mation processing (Corbetta et al. 2008). into another person’s situation. Avenanti et al.
Perspective taking is another prominent (2006) used similar perspective-taking manip-
example of how top-down processes shape ulations to assess their effects on motor inhi-
empathic responses and social understanding bition. In line with fMRI findings, TMS re-
(Avenanti et al. 2006; Jackson et al. 2006a; sults suggest that perspective-taking effects do
Lamm et al. 2007a, 2008; Ruby & Decety 2004; not operate on the level of primary sensorimo-
Stotland 1969). There is long-standing interest tor representations. Together with the absence
in how placing of oneself into another’s shoes of activation differences in visual areas in the
affects one’s vicarious response to the other, and fMRI study by Lamm et al. (2007a), this sug-
whether this differentially affects altruistic and gests that information about the other person
prosocial behavior (Batson et al. 1997; Batson “enters” the neural system in a similar way for
et al. 1987; Underwood & Moore 1982). So- both perspectives but is then processed and ap-
cial psychology studies suggest that adopting praised differently (but see also Avenanti et al.
a so-called imagine-other perspective (focusing 2008).
on the feelings and thoughts of the other) pro- As anyone knows from personal experience,
motes empathic concern and an other-oriented our attitudes toward others vary greatly and
(altruistic) motivation while explicitly imagin- therefore can affect empathic responses to
ing oneself (“imagine-self”) to be in the target’s them. Lanzetta’s group provided the first in-
distressful situation results in heightened per- vestigations of the psychophysiological corre-
sonal distress and an egoistic motivation to re- lates of how our relationship to someone else
duce that distress (via withdrawal or an aversive affects our response to their emotions (Englis
response). et al. 1982; Lanzetta & Englis 1989). Facial
Recent fMRI and TMS studies investi- EMG and other measures of autonomic ner-
gated the neural correlates of this phenomenon vous system activity demonstrated that being
(Avenanti et al. 2006; Jackson et al. 2005; in a competitive gaming relationship results in
Lamm et al. 2007a). In the study by Lamm counterempathetic responses—in other words,
et al. (2007a), participants watched videos joyful affect when the competitor loses and neg-
of patients undergoing painful auditory treat- ative affect when he or she wins. In a similar
ment either by imagining that they were in vein, a recent fMRI study (Singer et al. 2006)
the patient’s place (imagine-self) or by fo- assessed the modulation of empathic brain re-
cusing on the patient’s feelings and affective sponses to another person’s pain as a function
expressions (imagine-other). The results indi- of the perceived fairness of the other person and
cated higher personal distress and less em- the gender of the observer. As in previous empa-
pathic concern during the imagine-self per- thy studies, empathy-related activation in ACC
spective, which was associated with stronger and AI was observed for both genders when a
hemodynamic responses in brain regions cod- fair, liked player was in pain. However, men,
ing the motivational−affective dimensions of but not women, showed an absence of such
pain, such as the bilateral medial and ante- empathy-related activity when seeing an
Singer & Lamm: Social Neuroscience of Empathy 91
unfair and disliked player in pain. Instead, men other’s state in the absence of any bottom-up
showed increased activation in areas associ- stimulation.
ated with reward (nucleus accumbens), with As for the former, based on available evi-
activation in these areas correlating positively dence, we speculate that cortical areas asso-
with their desire for revenge (as assessed by ciated with executive function, contextual ap-
self-report after MRI scanning). These results praisal, and attention play a major role. This
suggest that, at least in men, a desire for re- would include cortical structures such as the
venge won over empathic motivation when dorsolateral prefrontal cortex, medial and an-
they were confronted with someone experi- terior cingulate cortex, right ventral premo-
encing pain who they believed deserved to be tor cortex, inferior and superior parietal cor-
punished. tex, and orbitofrontal cortex (e.g., Aron et al.
These results also point to gender as another 2004; Corbetta et al. 2008; Duncan & Owen
important variable in the investigation of em- 2000; Ridderinkhof et al. 2004; Rolls 2004).
pathetic responses. In fact, there has been a These areas interact with structures encoding
longstanding debate about whether women ac- the bottom-up driven affective responses, such
tually possess more empathy, as expressed by as the anterior insula, amygdala, and possibly
higher scores in various self-report measures, also parts of the ventral striatum. The latter
or whether the different questionnaire scores way in which top-down processes affect empa-
can be explained by social desirability and de- thy through mental imagery and top-down gen-
mand effects (Davis 1994; Eisenberg & Lennon eration of feelings becomes particularly impor-
1983). While social neuroscience should be able tant when minimal sensory information about
to make a valid contribution to this question the other is available, requiring the use of con-
soon (see Han et al. 2008 for preliminary evi- text information, affective memory, and self-to-
dence), the same amount of caution as in behav- other projection to infer the affective condition
ioral studies has to be applied with respect to of the other person. Depending on the situa-
experimental demand and social desirability ef- tion and the information available to the em-
fects. It will be a major challenge to disentangle pathizer, these processes might be implemented
neural responses that can be attributed to inter- in distributed networks that are recruited dur-
nal (“true”) or external (“socially desirable,” de- ing the imagery of action, sensation, and affect.
mand characteristics) empathetic motivations. In line with this reasoning, a recent fMRI study
In addition, gender-specific physiological dif- showed that the first-person imagery of pain
ferences such as head size, skull thickness, but activates structures also involved in empathy
also individual differences in hormonal state at for pain as well as in the direct experience of
the time of investigation have to be considered pain (Ogino et al. 2007). Similar evidence is
as potential confounds. reported by Jabbi et al. (2008), whose analyses
To summarize, there is considerable evi- demonstrate that it is not only the activation of a
dence that empathy is substantially modulated brain region, but also its interaction with other
by top-down processes such as attention or the brain areas which should be taken into account
contextual appraisal of a situation. Notably, when one interprets functional neuroimaging
there seem to be at least two different ways data (see also Jabbi et al. 2008; Zaki et al. 2007).
in which top-down processes can affect the em- The existence of different pathways for top-
pathic response. One way is to either inhibit down control is an important factor that should
or amplify representations that have been acti- be taken into account when assessing individual
vated via sensory channels and mechanisms as- differences in empathy as well as their link to
sociated with perception-action coupling. The prosocial behavior. For example, the responses
second way is to generate empathic responses of more “sensory-driven” persons might differ
by means of imagination or anticipation of the from those with better imaginative capacities.
92 Annals of the New York Academy of Sciences
Chartrand, T. L., & Bargh, J. A. (1999). The chameleon the human frontal lobe recruited by diverse cognitive
effect: The perception-behavior link and social in- demands. Trends Neurosci., 23, 475–483.
teraction. J. Pers. Soc. Psychol., 76, 893–910. Eisenberg, N. (2000). Emotion, regulation, and moral de-
Cheng, Y., Lin, C.-P., Yang, C.-Y., et al. (2008). The per- velopment. Annu. Rev. Psychol., 51, 665–697.
ception of pain in others modulates somatosensory Eisenberg, N., Fabes, R. A., Miller, P. A., et al. (1989). Re-
oscillations. NeuroImage, 40, 1833–1840. lation of sympathy and personal distress to prosocial
Corbetta, M., Patel, G., & Shulman, G. L. (2008). The behavior: A multimethod study. J. Pers. Soc. Psychol.,
reorienting system of the human brain: From envi- 57, 55–66.
ronment to theory of mind. Neuron, 58, 306–324. Eisenberg, N., & Lennon, R. (1983). Sex differences in
Craig, A. D. (2002). How do you feel? Interoception: The empathy and related capacities. Psycholog. Bull., 94,
sense of the physiological condition of the body. Nat. 100–131.
Rev. Neurosci., 3, 655–666. Eisenberg, N., & Strayer, J. A. (1987). Empathy and Its
Craig, A. D. (2003). Interoception: The sense of the phys- Development. Cambridge, UK: Cambridge University
iological condition of the body. Curr.Opin. Neurobiol., Press.
13, 500–505. Englis, B. G., Vaughan, K. B., & Lanzetta, J. T. (1982).
Critchley, H. D. (2005). Neural mechanisms of auto- Conditioning of counter-empathetic responses. J.
nomic, affective, and cognitive integration. J. Comp. Exp. Soc. Psychol., 18, 375–391.
Neurol., 493, 154–166. Fan, Y., & Han, S. (2008). Temporal dynamic of neu-
Critchley, H. D., Wiens, S., Rotshtein, P., et al. (2004). ral mechanisms involved in empathy for pain: An
Neural systems supporting interoceptive awareness. event-related brain potential study. Neuropsychologia,
Nature Neurosci., 7, 189–195. 46, 160–173.
Damasio, A. R. (1994). Descartes’ error and the future of Gallese, V. (2003a). The manifold nature of interpersonal
human life. Sci. Am., 271, 144. relations: The quest for a common mechanism. Phi-
Davis, M. H. (1994). Empathy: A Social Psychological Approach. los. Trans. R Soc. Lond. B Biol. Sci., 358, 517–528.
Boulder, CO: Westview. Gallese, V. (2003b). The roots of empathy: The shared
de Vignemont, F., & Singer, T. (2006). The empathic manifold hypothesis and the neural basis of inter-
brain: How, when and why? Trends Cogn. Sci., 10, subjectivity. Psychopathol., 36, 171–180.
435–441. Goldman, A. (2006). The Simulating Mind. New York: Ox-
Decety, J., & Grezes, J. (2006). The power of simulation: ford University Press.
Imagining one’s own and other’s behavior. Brain Res., Grill-Spector, K., & Malach, R. (2001). fMR-adaptation:
1079, 4–14. A tool for studying the functional properties of hu-
Decety, J., & Jackson, P. L. (2004). The functional archi- man cortical neurons. Acta Psychol. (Amst.), 107, 293–
tecture of human empathy. Behav. Cogn. Neurosci. Rev., 321.
3, 71–100. Gu, X., & Han, S. (2007). Attention and reality con-
Decety, J., & Lamm, C. (2006). Human empathy through straints on the neural processes of empathy for pain.
the lens of social neuroscience. Scientific World Journal, NeuroImage, 36, 256–267.
6, 1146–1163. Han, S., Fan, Y., & Mao, L. (2008). Gender difference in
Decety, J., & Lamm, C. (2007). The role of the right tem- empathy for pain: An electrophysiological investiga-
poroparietal junction in social interaction: How low- tion. Brain Res., 1196, 85–93.
level computational processes contribute to meta- Harrison, N. A., Singer, T., Rotshtein, P., et al. (2006).
cognition. Neuroscientist, 13, 580–593. Pupillary contagion: Central mechanisms engaged
Decety, J., & Lamm, C. (2009). The biological basis of in sadness processing. Soc. Cogn. Affect Neurosci., 1, 5–
empathy. In J. T. Cacioppo & G. G. Berntson (Eds.), 17.
Handbook of Neuroscience for the Behavioral Sciences. New Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1993). Em-
York: John Wiley and Sons. otional contagion. Curr. Dir. Psychol. Sci., 2, 96–99.
Derbyshire, S. W. G. (2000). Exploring the pain “neuro- Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1994). Emo-
matrix.” Curr. Rev. Pain, 4, 467–477. tional Contagion. Cambridge, UK: Cambridge Univer-
Dimberg, U., & Oehman, A. (1996). Behold the wrath: sity Press.
Psychophysiological responses to facial stimuli. Mo- Hatfield, E., Rapson, R. L., & Le, Y. L. (2009). Emotional
tivation and Emotion, 20, 149–182. contagion and empathy. In J. Decety & W. Ickes
Dinstein, I., Hasson, U., Rubin, N., et al. (2007). (Eds.), The Social neuroscience of Empathy. Cambridge,
Brain areas selective for both observed and ex- MA: MIT.
ecuted movements. J. Neurophysiol., 98, 1415– Hein, G., & Singer, T. (2008). I feel how you feel but
1427. not always: The empathic brain and its modulation.
Duncan, J., & Owen, A. M. (2000). Common regions of Curr.Opin. Neurobiol., 18, 153–8.
Singer & Lamm: Social Neuroscience of Empathy 95
Henson, R. N., & Rugg, M. D. (2003). Neural response sponses during empathy for pain. Brain Res., 1227,
suppression, haemodynamic repetition effects, and 153–161.
behavioural priming. Neuropsychologia, 41, 263–270. Lanzetta, J. T., & Englis, B. S. (1989). Expectations of
Hoffman, M. L. (2000). Empathy and Moral Development. cooperation and competition and their effects on
Cambridge, UK: Cambridge University Press. observers’ vicarious emotional responses. J. Pers. Soc.
Hutchison, W. D., Davis, K. D., Lozano, A. M., et al. Psychol., 56, 543–554.
(1999). Pain-related neurons in the human cingulate Moody, E. J., McIntosh, D. N., Mann, L. J., et al. (2007).
cortex. Nature Neurosci., 2, 403–405. More than mere mimicry? The influence of emotion
Jabbi, M., Bastiaansen, J., & Keysers, C. (2008). A com- on rapid facial reactions to faces. Emotion, 7, 447–
mon anterior insula representation of disgust ob- 457.
servation, experience and imagination shows diver- Morrison, I., & Downing, P. E. (2007). Organization of felt
gent functional connectivity pathways. PLoS ONE, 3, and seen pain responses in anterior cingulate cortex.
e2939. NeuroImage, 37, 642–651.
Jabbi, M., Swart, M., & Keysers, C. (2007). Empathy Morrison, I., Lloyd, D., di Pellegrino, G., et al. (2004). Vi-
for positive and negative emotions in the gustatory carious responses to pain in anterior cingulate cortex:
cortex. NeuroImage, 34, 1744–1753. Is empathy a multisensory issue? Cogn. Affect Behav.
Jackson, P., Brunet, E., Meltzoff, A., et al. (2006a). Em- Neurosci., 4, 270–278.
pathy examined through the neural mechanisms in- Niedenthal, P. M., Brauer, M., Halberstadt, J. B., et al.
volved in imagining how I feel versus how you feel (2001). When did her smile drop? Facial mimicry and
pain. Neuropsychologia, 44, 752–761. the influences of emotional state on the detection of
Jackson, P., Meltzoff, A., & Decety, J. (2005). How do change in emotional expression. Cognition and Emotion,
we perceive the pain of others? A window into the 15, 853–864.
neural processes involved in empathy. NeuroImage, 24, Ogino, Y., Nemoto, H., Inui, K., et al. (2007). Inner ex-
771–779. perience of pain: Imagination of pain while viewing
Jackson, P., Rainville, P., & Decety, J. (2006b). To what images showing painful events forms subjective pain
extent do we share the pain of others? Insight from representation in human brain. Cerebral Cortex, 17(5),
the neural bases of pain empathy. Pain, 125, 5–9. 1139–1146.
Jenkins, A. C., Macrae, C. N., & Mitchell, J. P. (2008). Ploghaus, A., Tracey, I., Gati, J. S., et al. (1999). Dissoci-
Repetition suppression of ventromedial prefrontal ating pain from its anticipation in the human brain.
activity during judgments of self and others. Proc. Science, 284, 1979–1981.
Natl. Acad. Sci. USA, 105, 4507–4512. Preston, S. D., & de Waal, F. B. M. (2002). Empathy: Its
Keysers, C., & Gazzola, V. (2006). Towards a unifying ultimate and proximate bases. Behav. Brain Sci., 25,
neural theory of social cognition. Prog. Brain Res., 156, 1–72.
379–401. Prinz, W. (2005). Experimental approaches to action. In
Keysers, C., Wicker, B., Gazzola, V., et al. (2004). A touch- J. Roessler & N. Eilan (Eds.), Agency and Self-Awareness
ing sight: SII/PV activation during the observation (pp. 165–187). New York: Oxford University Press.
and experience of touch. Neuron, 42, 335–346. Ridderinkhof, K. R., Ullsperger, M., Crone, E. A., et al.
Lakin, J. L., & Chartrand, T. L. (2003). Using noncon- (2004). The role of the medial frontal cortex in cog-
scious behavioral mimicry to create affiliation and nitive control. Science, 306, 443–447.
rapport. Psychol. Sci., 14, 334–339. Rolls, E. T. (2004). The functions of the orbitofrontal
Lamm, C., Batson, C. D., & Decety, J. (2007a). The neural cortex. Brain and Cognition, 55, 11–29.
substrate of human empathy: Effects of perspective- Ruby, P., & Decety, J. (2004). How would you feel versus
taking and cognitive appraisal. J. Cogn. Neurosci., 19, how do you think she would feel? A neuroimaging
42–58. study of perspective-taking with social emotions. J.
Lamm, C., Meltzoff, A. N., & Decety, J. (2009). How Cogn. Neurosci., 16, 988–999.
do we empathize with someone who is not like us? Singer, T. (2006). The neuronal basis and ontogeny of em-
A functional magnetic resonance imaging study. J. pathy and mind reading: Review of literature and im-
Cogn. Neurosci., in press. plications for future research. Neurosci. Biobehav. Rev.,
Lamm, C., Nusbaum, H. C., Meltzoff, A. N., et al. 30, 855–863.
(2007b). What are you feeling? Using functional Singer, T., & Leiberg, S. (2009). Sharing the emotions
magnetic resonance imaging to assess the modula- of others: The neural bases of empathy. In M. S.
tion of sensory and affective responses during empa- Gazzaniga (Ed.), The Cognitive Neurosciences IV . Cam-
thy for pain. PLoS ONE, 12, e1292. bridge, MA: MIT Press.
Lamm, C., Porges, E., Cacioppo, J. T., et al. (2008). Per- Singer, T., Seymour, B., O’Doherty, J., et al. (2004). Em-
spective taking is associated with specific facial re- pathy for pain involves the affective but not the
96 Annals of the New York Academy of Sciences
sensory components of pain. Science, 303, 1157– Underwood, B., & Moore, B. (1982). Perspective-taking
1161. and altruism. Psychol. Bull., 91, 143–173.
Singer, T., Seymour, B., O’Doherty, J. P., et al. (2006). van Baaren, R. B., Horgan, T. G., Chartrand, T. L., et al.
Empathic neural responses are modulated by the (2004). The forest, the trees, and the chameleon:
perceived fairness of others. Nature, 439, 466–469. Context dependence and mimicry. J. Pers. Soc. Psy-
Smith, A. (1976). The Theory of Moral Sentiments. Oxford, chol., 86(3), 453–459.
UK: Oxford University Press. (Original work pub- van der Gaag, C., Minderaa, R. B., & Keysers, C. (2007).
lished 1759). Facial expressions: What the mirror neuron system
Sonnby-Borgstrom, M. (2002). Automatic mimicry reac- can and cannot tell us. Soc. Neurosci., 2, 179–222.
tions as related to differences in emotional empathy. Wicker, B., Keysers, C., Plailly, J., et al. (2003). Both of us
Scand. J. Psychol., 43, 433–443. disgusted in my insula: The common neural basis of
Stotland, E. (1969). Exploratory investigations of em- seeing and feeling disgust. Neuron, 40, 655–664.
pathy. In L. Berkowitz (Ed.), Advances in Experi- Zahn-Waxler, C., Radke-Yarrow, M., Wagner, E., et al.
mental Social Psychology (pp. 271–314). New York: (1992). Development of concern for others. Develop.
Academic. Psych., 28, 126–136.
Stüber, K. (2008). Empathy. In E. N. Zalta (Ed.), Zaki, J., Ochsner, K. N., Hanelin, J., et al. (2007). Differ-
The Stanford Encyclopedia of Philosophy (http://plato. ent circuits for different pain: Patterns of functional
stanford.edu/entries/empathy/). Stanford: The connectivity reveal distinct networks for processing
Metaphysics Research Lab, Stanford University. pain in self and others. Soc. Neurosci., 2, 276–291.