Sei sulla pagina 1di 12

REVIEWS

Mammalian empathy: behavioural


manifestations and neural basis
Frans B.M.deWaal1* and StephanieD.Preston2*
Abstract | Recent research on empathy in humans and other mammals seeks to dissociate
emotional and cognitive empathy. These forms, however, remain interconnected in evolution,
across species and at the level of neural mechanisms. New data have facilitated the development
of empathy models such as the perceptionaction model (PAM) and mirror-neuron theories.
According to the PAM, the emotional states of others are understood through personal,
embodied representations that allow empathy and accuracy to increase based on the observers
past experiences. In this Review, we discuss the latest evidence from studies carried out across
a wide range of species, including studies on yawn contagion, consolation, aid-giving and
contagious physiological affect, and we summarize neuroscientific data on representations
related to anothers state.

Empathy In lay parlance as well as the scientific literature, the term This umbrella term includes obvious instances of
Any process that emerges empathy refers to both the sharing of emotions between affective resonance, such as when an observer feels con-
from the fact that observers individuals and the adopting of anothers point of view. tagiously distressed by the targets distress or mimics their
understand others states by This capacity allows individuals to quickly relate to emotions. However, such outward signs are not required,
activating personal, neural and
mental representations of that
anothers state, which is essential in species that provide as shared neural representations do not need to produce
state, including the capacity to extensive parental care and work cooperatively towards downstream bodily responses7. This phenomenon exists
be affected by and share the common goals1. A mentalistic definition of empathy, for all states, but we focus on distress because of the pre-
emotional state of another; closer to theory-ofmind, is also common, and describes ponderance of relevant data and the clear application to
assess the reasons for the
how empathy allows one to access anothers mind by helping behaviour. Empathic accuracy exists on a con-
others state; and identify
with the other, adopting his simulating how one would feel in his or her shoes2. For tinuum that is dependent upon the observers attention,
or her perspective. example, Baron-Cohen3 describes empathy as a leap of motivation to understand and relevant personal past expe-
imagination into someone elses headspace. Definitions riences. Empathy can but does not always lead to helping.
that favour cognition over emotion, however, exclude Conversely, helping can emerge from many routes aside
young children and most non-human species, which, from empathy and shared affect. As higher-level processes
nonetheless, are affected by the emotional states of such as simulation, theory-ofmind and cognitive empathy
others, and provide striking care and aid as aresult. require observers to activate their own representations of
The notion that empathy is an affective fusion the targets state from the top down, they are included
1
Psychology Department and
Living Links, Yerkes National with another individuals state, rather than a cognitive under the umbrella term empathy, even if they require
Primate Research Center, deduction, had already been proposed in early 20th additional cognitive and executive processes.
EmoryUniversity, century philosophy 4. Lipps5, who gave us the concept of Arranging related yet distinguishable phenomena
36EagleRow, Atlanta, Einfhlung (German for feeling into), emphasized the into a single concept reflects the evolutionary view
Georgia30322, USA.
2
Department of Psychology,
almost involuntary nature of this response, which he according to which complex organs and capacities arise
University of Michigan, regarded as possible without extensive learning, asso- incrementally. They evolve in layers, with each new layer
530Church Street, ciation or reasoning. It took a century, however, for the being built on top of and dependent on older ones with-
AnnArbor, Michigan48109, quick spontaneous nature of human affective empathy out ever replacing them, as in the socalled Russian-doll
USA.
to be confirmed by empirical research. In this Review, model (FIG.1). There exists a tendency to treat each aspect
*Both authors contributed
equally to this work. we propose a definition that balances the emotional separately and dwell on the distinctions, but in doing
Correspondence to
and cognitive sides, and use empathy as an umbrella so we lose sight of the functionally integrated whole.
F.B.M.d.W. term for all processes that emerge from the fact that Accordingly, empathy is defined here as emotional and
dewaal@emory.edu observers understand others states by activating their mental sensitivity to anothers state, from being affected
doi:10.1038/nrn.2017.72 own personal, neural and mental representations of by and sharing in this state to assessing the reasons for it
Published online 29 Jun 2017 that state1,6. and adopting the others point ofview.

NATURE REVIEWS | NEUROSCIENCE ADVANCE ONLINE PUBLICATION | 1



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

other requires and hand it to them15. Similarly, capuchin


monkeys vary their food-sharing with what they know
Perspective- about their partners previous food consumption16. There
taking and is no apriori reason to exclude perspective-taking in
targeted helping smaller-brained species, and studies on rodent aid-giving
are indeed suggestive of related capacities (see below).
The shared vocabulary adopted here for expressions of
empathy in both humans and other animals reflects the
Darwinian assumption that if related species show sim-
Theory-ofmind
The ability to attribute mental ilar responses under similar circumstances, the underly-
states to others, such as ing proximate processes are probably homologous (that is,
Empathic
knowledge, intentions
concern and derived from a common ancestor) rather than analogous
and beliefs. consolation (that is, independently evolved).
Affective empathy
The first review of this framework was presented in
Also known as emotional 2002 (REF.1) and was based on evidence available at the
empathy. Empathy that is time that came from across different species and levels
directly affected by the of analysis. According to the perceptionaction model
emotional state of another by
(PAM), individuals understand and have a sense of oth-
matching or feeling with it, as
a result of perceiving this state. Perceptionaction Motor mimicry ers emotions because the nervous system evolved to map
mechanism and emotional others states onto their own individual representations
Cognitive empathy contagion for experiencing those states1. As such, when an observer
Empathy derived from attends to anothers state, he or she spontaneously
a top-down process in which
the observer imagines how the
accesses information about the other (the target), their
target feels, even if the target is feelings, the situation and other related concepts through
not present or their feelings a distributed associative process. The fact that observers
cannot be directly observed. understand others using their own individual representa-
Figure 1 | The Russian-doll Nature
model ofReviews | Neuroscience
the evolution of tions, which are developed through experience, explains
Empathic perspective-taking empathy. Various components of the empathic response,
The capacity to take anothers why empathy and accuracy increase with familiarity and
which have been added layer upon layer during evolution, similarity with the other and their situation. Extensive
affective perspective: for
remain functionally integrated. At its core is the
example, understanding their
perceptionaction mechanism, which induces a similar
empirical support for the PAM has accumulated since it
specific situation and needs, was first presented, and thus an update is warranted. In
separate from ones own, emotional state in the observer as in the target. Its most
which still requires access to basic expressions are motor mimicry and emotional fact, so much research has been performed since 2002
personal representations of contagion. The dolls outer layers, such as empathic that we cannot comprehensively review it here. We focus
the others state. concern and perspective-taking, build upon this core on summarizing the most robust and central findings
socio-affective basis while increasingly requiring emotion and debates, and often cite review articles to avoid selec-
False-belief task regulation, selfother distinction and cognition. Even tive citation. We place empathy within the Tinbergian
A crucial theory-ofmind task though the dolls outer layers depend on prefrontal
that determines whether an framework according to which every behavioural phe-
functioning, they remain fundamentally linked to the core
observer knows what another nomenon can be investigated from different yet comple-
perceptionaction mechanism. Adapted with permission
knows, even if this knowledge
from: de Waal, F. B. M. in Feelings & Emotions: The mentary angles, such as why these mechanisms were
is incongruent with the selected during evolution, and how behavioural and
observers own.
Amsterdam Symposium (eds Manstead T., Frijda, N. &
Fischer A.) 379399 (Cambridge Univ. Press, 2003). neural mechanisms regulate their expression17,18.
Targeted helping The existence of empathy outside of our own species
Assistance and care based on is hardly a novel idea. Even while avoiding the term itself,
a cognitive appreciation of the Thus defined, empathy is common in our species, early studies on altruistic behaviour often employed par-
others specific need or
starts early in life8, and shows neural and physiological adigms suggesting that the emotional contagion of pain or
circumstances.
correlates as well as inheritability 9. Whereas emotional distress is a motivating factor. Altruism is defined as an
Tinbergian framework responses to the emotions of others are commonplace act that benefits another individual at a cost to the self.
A series of why questions that in many species10,11, the best evidence for empathic Having trained rats to obtain food by pressing a lever,
we may ask about any perspective-taking, which we have distinguished as cogni- Church19 found that if the same action delivers an elec-
observed behaviour, as
proposed by Niko Tinbergen.
tive empathy 12, comes from intensely social animals that tric shock to a visible second rat, the first rat will tem-
The questions concern have a high degree of encephalization, such as humans, porarily cease lever-pressing, thus foregoing its rewards.
different levels of causation: apes, elephants and dolphins. Some species show signs of Entitled Emotional reactions of rats to the pain of oth-
for example, why did a theory-ofmind in a false-belief task13, while others per- ers, Churchs 1959 paper inspired a flurry of research,
behaviour evolve (what are its
form targeted helping that demonstrates an appreciation which initially assumed a conditioning explanation.
benefits), why does the
behaviour occur (what caused of anothers specific predicament. For example, dolphins Later, it turned out that such behaviour did not require
it) and what is the behaviours may lift an incapacitated companion to the surface so prior experience with the protocol. For example, Rice and
phylogenetic origin? that they can breathe and chimpanzees may bring down Gainer 20 found that rats spontaneously help distressed
fruits from a tree to an elderly female who has lost her companions that have been hoisted off of the cage floor
Emotional contagion
Emotional state matching
climbing abilities14. Chimpanzees seem to understand the by lowering them back down. Monkeys tested in the
between a target and an needs and desires of others because after having judged same era showed even more dramatic responses, such
observer. anothers situation they will find the specific tool that the as macaques refusing to pull a chain that delivers food

2 | ADVANCE ONLINE PUBLICATION www.nature.com/nrn



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Consolation behaviour to themselves if doing so also shocks a visible conspe- the aversion was stronger to the former. None of these
Reassurance behaviour cific21. Subjects sometimes went more than a week with early results supported explanations based on learning
directed at a distressed party, only minimal food to avoid shocking another monkey, and conditioning, even though these processes do have a
such as a victim of aggression and this effect was augmented by prior shock experience role in empathy and altruism. The assumption of a basic
(also see the definition of
empathic concern).
but did not require it 22. Such sacrifices relate to the tight empathy mechanism is in fact more parsimonious than is
social linkage among macaques and the fact that these any learning-based explanation, which would need to be
Empathic concern responses were more pronounced between familiar than rather demanding in its range of proposed contingencies.
Also known as sympathetic unfamiliar individuals22. With regards to our closest relatives, the apes, observa-
concern. Concern about
Early research stressed the untrained, spontaneous tions of empathic responses go back to early last century.
anothers state, and attempts
to ameliorate this state (also
nature of helping and its social bias in order to counter Yerkes24 reported sympathetic behaviour by a bon-
see the definition of explanations that were based on conditioning and associ- obo towards a sickly companion, and Ladygina-Kohts25
consolation behaviour). ative learning. If the avoidance of aversive stimuli (such as described similar tendencies for her juvenile chimpanzee.
distress vocalizations) or the seeking of extrinsic rewards By the 1970s, these responses were defined as consolation
was the sole reason for altruistic behaviour, there would behaviour and were systematically documented in natural-
be no reason for a familiarity bias and effects should be istic settings26. The way in which apes comfort others in
similar between strangers, yet with few exceptions famil- distress (for example, embracing, touching and kissing)
iarity promotes altruism. In addition, when Miller et al.23 (FIG.2) is morphologically similar to how young children
compared macaque responses to images of fearful con- express socalled empathic concern for family members
specifics versus negatively conditioned non-social stimuli, who have been instructed to feign pain or discomfort8,27.

Behaviour Denition Mechanisms Non-human species


a Mirroring Rapid face matching Motor mimicry Chimpanzees
and movement mapping Orangutans
Gelada baboons
Macaques

b Yawn contagion Yawning in response to Motor mimicry Chimpanzees


anothers yawns Macaques
Canines
Budgerigars

c State matching Sharing the emotional state Emotional contagion Mice


of another Voles
Chickens

d Consolation Comforting a distressed party Empathic concern that is based Chimpanzees


on emotional contagion and Bonobos
requires self-regulation Macaques
Canines
Elephants
Voles
Mice
Rooks
e Learned helping Liberating another or relieving Aiding that is motivated by Macaques
its pain emotional contagion, and requires Rats
self-regulation and trial-and-
error learning

f Targeted helping Help adapted to the specic Aiding that is motivated by Apes
need or situation of another emotional contagion, and requires Capuchin monkeys
self-regulation and Dolphins
perspective-taking Elephants

Figure 2 | Behavioural manifestations of animal empathy. The table mates matching each others physiologicalNature
stress Reviews
level (partc);a juvenile
| Neuroscience
summarizes behavioural patterns that are considered to be expressions bonobo wrapping her arms around another who has just lost a fight to
of empathy in non-human vertebrates, ranging from motor mirroring provide consolation (partd);a rat that has learned to liberate another rat
and yawn contagion to targeted helping. Most of these behaviours have that was trapped in a container (part e); and an adult chimpanzee
been characterized by experimental research, whereas others have been showing targeted helping by assisting a juveniles descent from a tree
documented observationally (see the references cited in the main text). (partf). The images in parts af are courtesy of PierFrancescoFerrari,
The mechanisms and the non-human species in which these behaviours University of Parma, Italy; TeresaRomero, University of Lincoln, UK;
have been observed are shown. From top to bottom, the images show the ZackJohnson, Georgia Institute of Technology, USA; ZannaClay,
following:two gelada baboon juveniles with playfaces (that is, they are DurhamUniversity, UK; and InbalBen-AmiBartal, University of California,
showing rapid facial mimicry) (parta);a yawning wolf (partb);prairie vole Berkeley, USA.

NATURE REVIEWS | NEUROSCIENCE ADVANCE ONLINE PUBLICATION | 3



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Even infants younger than 1year old exhibit significant Emotional contagion. Following the tradition of Church
attention and concern for a peer in distress28,29, which con- and Miller from the mid20th century, researchers have
trasts with long-held stereotypes that human infants do recently confirmed the existence of contagious affect in
little more than cry contagiously when otherscry. rodents. For example, the pain of one rodent may increase
the pain response of an observing conspecific, and this
Behavioural manifestations effect is highly influenced by familiarity with the suffering
Behavioural studies of empathy in mammals and other target47,48. Moreover, this effect is blocked by the presence
animals follow one of three approaches. The first focuses of strangers, who induce a glucocorticoid stress response
on motor mimicry, which is one of many behaviours that suppresses the natural motivating contagion of dis-
that reveal an underlying perceptionaction mechanism tress in rodents and humans49. Contagious emotional
(as described in the PAM). The second approach meas- responses are at least partially conveyed by the transfer of
ures resonating physiological and behavioural responses negative physiological states between the target and the
to the emotions of others, such as pain contagion. The observer. For example, both prairie voles50 and humans51
third is to focus on altruistic behaviour, such as con- match the hormonal stress response of a stressed target. In
solation behaviour and the helping of others in need, humans, this effect correlates with self-reported empathic
which are often motivated by shared affect between concern and perspective-taking in dailylife51.
an observer and a target along with an evaluation of Contagious anxiety in observers probably follows an
the targets situation. All three areas have seen recent invertedU-shaped function, with moderate activation
progress, as summarized in FIG.2. facilitating a response but high activation inhibiting it 52.
Emotion self-regulation may be key to keeping the activa-
Motor mirroring. Motor mimicry is a natural consequence tion at a moderate level. Thus, a bonobos ability to over-
of a neural system that maps another individuals bodily come its own distress correlates with the frequency with
and facial expressions onto the observers own representa- which it offers consolation to others who are in distress53.
tions for action. This perceptionaction mechanism As emotion self-regulation is compromised in bono-
permits the adaptive synchronization of states between bos that are orphaned early in life53 as also reported
targets and observers, such as in the rapid facial mimicry for human orphans54 fully developed empathy may
described in humans30 and other primates31,32. Supporting require an affectionate upbringing.
a possible causal link with empathy, humans who report
being more emotionally empathic in daily life are the most Altruistic helping. It has been claimed that humans are
accurate mimics of others facial expressions33. A typical the only truly altruistic species, and that other animals
example of motor mimicry is yawn contagion, which can are self-interested and only care about return benefits
be studied with experimental protocols in which subjects (for examples, see REFS55,56). This view erroneously
watch videos of yawning conspecifics or through natural- conflates evolution and motivation. Although the evo-
istic observations of the spreading of yawns across a group. lution of altruism is indeed explained by kin selection
Yawn contagion has been found in species as diverse as and reciprocal altruism both of which emphasize the
humans, non-human primates, canines and birds3437. As eventual return benefits from helping this explana-
with empathy more broadly, yawn contagion is biased tion does not consider the role of motivation. Animals
towards socially close others, which cannot be explained do not know about genetic benefits such as inclusive
by attention as controlled studies report either similar or fitness. There is also not much evidence indicating that
greater attention to unfamiliar yawn stimuli34,38. they anticipate future return favours from their acts,
Even though monkeys are sometimes said to lack as in reciprocal altruism. One cannot be motivated by
true imitation39, they do mimic human facial and hand what one does not know. There is no conflict, therefore,
gestures40, and they socially prefer partners who mimic between the assumption that altruistic behaviour evolved
them41. In the wild, monkeys copy one anothers feeding for self-serving reasons and the suggestion that it reflects
techniques, resulting in cultural behavioural variation42. a spontaneous motivational impulse that originates at
In the laboratory, monkeys replicate the motor actions of least in humans and other mammals from empathy
a group member in, for example, the way in which they with the recipients need, pain or distress6. Tinbergen18
open a box 43. Anthropoid apes show additional capaci- encouraged researchers to separate why a behaviour
ties in that they imitate with an understanding of cause evolved (the ultimate reason for its existence) from how
Perceptionaction
mechanism
and effect44. However, even in apes, motor mimicry and it is instantiated in the individual (its proximate causa-
Spontaneous activation of an shared representations probably provide the foundation tion). Theories regarding these issues are different yet
individuals own personal for such behaviour: for example, when an observing ape compatible as they occupy separate levels of analysis.
representations for a target, spontaneously synchronizes his movements with those of The view that humans are uniquely altruistic is also
their state and their situation
a conspecific model demonstrating tool use45. Studies inconsistent with the increasing amount of evidence
when perceiving the targets
state. of ape imitation indicate that it is not merely about indicating that even rodents are affected by the pain
emulating a beneficial outcome because ghost demon- of familiar conspecifics and offer help. For example,
Emotion self-regulation strations of the correct outcome without an actual con- monogamous voles show a contagious stress response
Control over ones own specific demonstrating the required actions are ignored. to familiar partners, and this is mediated by stress
emotions to promote adaptive
responding, including response
For imitation to occur, apes need to see a conspecific per- and anxiety induced in the observer; they also groom
delay, recovery from upsets form the action sequence46, which suggests the crucial stressed partners, and this seems to have a calming effect
and selective attention. role of bodily identification and motor mimicry. as it reduces anxiety-related behaviour 50. Rats learn by

4 | ADVANCE ONLINE PUBLICATION www.nature.com/nrn



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

trial-and-error how to free a trapped conspecific by since it has the property of transforming another per-
opening a door 57,58, and this is again mediated by shared sons misfortune into ones own feeling of distress. This
anxiety but impeded by excessive stress in the observer 57. is the beauty of the empathy mechanism and makes its
The transfer of negative affect from a target to an operation similar to most activities that are required
observer is also thought to motivate another important for mammalian survival (for example, eating, breeding
prosocial behaviour in primates: consolation. Defined as and nursing), which are all inherently rewarding. If the
comforting body contact aimed at distressed others, con- same intrinsic rewards apply to helping behaviour, this
solation was first described in the great apes26,53,59,60, but has is no reason to denigrate the response or to refuse to call
since also been found in some macaque species61, canines62, it altruistic. Evolutionary biology already assumes that
elephants63, rodents50 and corvids64. In primates, responses altruistic tendencies yield long-term net benefits.
such as placing an arm around another or grooming a Altruistic behaviour is far from limited to mammals,
victim of aggression have been shown to reduce behav- yet it remains unclear whether the same mechanism
ioural indicators of anxiety, such as self-scratching and underlies helping in non-mammals. This is quite pos-
self-grooming 60,65, while also reducing heartrate66. sible in other vertebrates (see below), but it is less likely
In line with the PAM1, these studies support the crucial that homologous, neural perceptionaction mechanisms
role of caught distress and arousal in prosocial behaviour, underpin the well-documented altruism of invertebrates
which nonetheless must be kept within limits to facilitate such as social insects. Through convergent evolution,
a prosocial response. The same mechanism probably genetically distant species often arrive at functionally
underlies rescue attempts by rodents, dogs, elephants, pri- similar, analogous behaviours while using different
mates and humans, given that all of these species possess mechanisms. However, even if the sophisticated rescue
mammalian brains that have highly conserved character- of trapped conspecifics by ants is not based on emo-
istics. Although many acts that are defined by biologists as tional or cognitive empathy 75,76, it has been suggested
altruistic (for example, alarm calls and defence of the hive) that contagious ant alarm and digging is induced in
do not require emotional empathy, there is a category of nearby conspecifics by the release of a volatile chemical
directed altruism in mammals that does seem to depend on from the mandible of a trapped ant 77. If true, the mech-
an affective sensitivity to anothers state and situation6,14. anism in ants may not be as different from the PAM as
There is little support for the main alternative hypoth- one might think, as their helping response is still trig-
esis, which is that helping behaviour is driven by tangible gered by the transfer of a physiological state from target
rewards. For example, rats prefer to release a trapped cage to observer, and thus represents a pheromonal analogy
mate over accessing attractive food for themselves57,58,67. to the perceptionaction mechanism of mammals.
They share any available rewards after the rescue and per-
form helping responses even if a subsequent reunion with The evolution of empathy. Offspring care most likely
the target (which could be rewarding) is blocked57. Both strengthened the evolution of empathy because of the
human children and chimpanzees spontaneously assist selection pressure to develop a rapid, motivating emo-
a human experimenter, but children given an extrinsic tional connection between helpless neonates and their
reward for helping actually help less than do those who caregivers8,12,78. Signalling their state through smiling and
are merely praised or not rewarded at all68,69. Thus, aid can
be given in direct response to anothers need or distress in
a manner that is independent of extrinsic rewards.
Taken together, altruistic and consoling responses Target 1 Observer
Emotion
seem to arise from the transfer of emotion from the tar- transfer
get to the observer, which in turn motivates the observer Shared
Distress
aect
to approach and console the target, thus reducing the 2
Consolation Self-regulation
negative state of both parties (FIG.3). The central role of and/or aid
affect transfer in helping behaviour was demonstrated
in rats in the above helping paradigm, in an experiment 3
that showed that the administration of the anxiolytic Emotion
transfer
midazolam reduced rats helping responses without 4
Reduced
impairing the instrumental act required to obtain food67. Relief Self-reward
distress
Affect transfer provides a builtin reward to helping
because it gives the observer a stake in anothers well- Figure 3 | From affect transfer to altruism. As shown in
Nature Reviews | Neuroscience
Directed altruism being, producing both shared pain and vicarious relief multiple animal studies, empathy may promote aid-giving
Helping or comforting or joy when the pain is ameliorated; this is accompanied behaviour between conspecifics6,57,58. First, the targets
behaviour directed at an
by the warm glow that humans feel after doing a good distress induces stress or distress in the observer through
individual in need, pain or emotional transfer (step1). The observer needs to
distress. deed70 or the vicarious rewards that monkeys seem to
downregulate its own distress in order to effectively attend
gain from prosocial behaviour 6,71,72.
to the target, such as through helping or consolation
Convergent evolution Blurring the line between the self and the other at (step2). The resulting reduction of the targets distress as
A process in which unrelated the proximate level undermines the usefulness of a
species independently evolve a result of being helped is then transferred back to the
similar traits or capacities in
dichotomy between selfish and altruistic motives6,73. As observer, ameliorating the observers caught distress
response to similar expressed by Hoffman74, empathy may be uniquely well (step3); this reduction constitutes an intrinsic reward for
environmental pressures. suited for bridging the gap between egoism and altruism, performed altruism (step4).

NATURE REVIEWS | NEUROSCIENCE ADVANCE ONLINE PUBLICATION | 5



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Ideomotor action crying, human infants prompt caregivers into action. demands on it (for an example, see REF.84). The fact that
An action that is covertly Equivalent signals operate in other animals in which mammals retain distress vocalizations into adulthood
mimicked when anothers reproduction requires feeding, cleaning and warm- hints at the continued survival value of care-inducing
action or movement of an ing of the young. Parents that paid close attention to signals, such as those that prompt primates to carefully
object is being observed, such
as moving your arm when you
the emotional signals from their offspring probably lick and clean the wounds of unrelated conspecifics85.
watch someone bowl or tilting out-reproduced the indifferent ones, and this selective Despite the fact that the spread of emotion from
your head in synchrony with advantage applies at least to mammals and birds. The one individual to another occurs spontaneously, empa-
a pendulum swing. signals of the young are not just responded to but induce thy is still subject to appraisals, filters and inhibitions86
an agitated state in observers, suggesting that parents that prevent it from being expressed when it would be
Motor imitation
Reenactment by the observer
become distressed at the perception of offspring distress79. maladaptive. For example, chimpanzees are capable
of a targets motor movements Transferred negative affect provides a natural motivation of brutally killing each other 87,88, as are humans. Social
or facial expressions. to act1,12. In addition, neuropeptide systems that subserve species must be able to suppress their empathic response
offspring care are homologous across mammals and facil- to anothers pain when, for example, it is better for the
Affordances
Motor or action properties of
itate altruism in humans (for a review, see REF.78). Given observer to flee, keep a safe distance or confront a poten-
objects that are activated by this origin, the capacity to be motivated to help distressed tial adversary. Conflicting individual goals will prevent
their percepts, which are and vulnerable targets should occur in a wide variety of the activation of shared representations from proceeding
intrinsic to the mental vertebrates that show social attachment and parental care, to compassion, concern or altruistic action, but they are
representation of the object.
including rodents80, birds81 and possibly reptiles82. Given still needed to at least initially decode the targets state.
Simon effect the highly conserved nature of neuropeptide systems, Conversely, in humans and other species, the empathic
A behavioural effect in which an even older origin of empathy could be envisioned, response is increased by similarity, familiarity and social
motor actions are facilitated particularly in species that show social attachments83. closeness (for examples, see REFS1,47,89), and this is
when they are consistent with Once the neural underpinnings of an empathic off- consistent with where evolutionary theory expects it to
the spatial location of the
stimulus, and slowed when
spring care system existed, they could serve outside of the occur: that is, in close interdependent social relationships
they are inconsistent or rearing context and have a role in the wider social fabric. that involve either genetic relatedness or reciprocation.
opposing the location. They could be activated by the perception of a distressed Thus, the empathy mechanism is biased in precisely the
adult conspecific, such as in rodent consolation behav- way that is predicted by the current evolutionary models
iour 50. For every species, the reach of empathy varies with of cooperation6.
its social organization, creating differences across solitary,
pair-bonded, cooperatively hunting or group-living ani- Neural basis
mals, not because the empathy mechanism works differ- The PAM1 offers a proximate explanation for how we
ently, but because species and their ecology place different can feel into others states. The PAM derives from
a perceptionaction theory of motor behaviour that
reveals overlapping representations for performing and
observing actions, as exemplified in phenomena such as
Box 1 | Mirror neurons as part of the bigger picture
ideomotor action, motor imitation, affordances and the Simon
The extension of mirror neurons to social-cognitive phenomena135,136 has received both effect90. Applied to emotional phenomena, attending to
attention and criticism. Mirror cells cannot fully explain empathy, but their relevance anothers distress is assumed to activate observers own
should not be dismissed. Mirror neurons perse only represent a small subset of monkey distributed representations for the target and their state
F5 inferior premotor cells91 and, thus, must be part of a larger coding scheme. Yet, even
and situation including related feelings, memories and
these single cells represent motor plans and goals, which are bidirectionally connected
associations in a connectionistmanner.
to the resulting sensations90, allowing primates to infer goals from simple observed acts.
Extended to emotional phenomena, observing an affective posture or expression can A few key early studies offered the first notable neuro
feed back from peripheral motor representations to activate associated emotional scientific evidence for the PAM. First, mirror neurons were
states. After the discovery of mirror neurons in macaques91, their existence was recorded in macaque monkeys that responded similarly
confirmed in human patients137. Given that mirror cells were found in humans outside of to performed and observed actions, and these recordings
traditional mirror regions, mirroring may be a generalized neural coding strategy. In provided cellular-level evidence for perceptionaction
addition, some human cells are activated during execution but are inhibited during coding in the brain91 (BOX1). Extending this to humans,
observation, which may assist in selfother distinction138. a subsequent imaging study found premotor activity in
Brain areas that constitute the wider mirror system namely, the inferior frontal the brains of participants who simply observed pictures of
gyrus (IFG; also known as Brodmann area 44) and rostral posterior parietal cortex (rPPC)
tools (for example, hammers) or silently named the use
clearly participate in the imitation of motor acts, expressions and empathy. These
of the tool in their mind92. Finally, emotional evidence for
human mirror regions are homologous with the macaque mirror regions F5, and PF and
PFG (which are subdivisions of the rostral inferior parietal lobule). These human regions using personal representations to perceive anothers affect
are also interconnected with the insula and show activation in functional MRI (fMRI) came from a brain-lesion study in which patients were
studies of empathy and imitation136,139. The role of the IFG is supported by a meta-analysis most impaired at decoding anothers facial expressions of
that found consistent activation of the IFG and adjacent anterior insula during empathic emotion when they had somatosensory damage93. Since
pain118. The ventromedial prefrontal cortex can integrate such affective signals to guide these results were summarized in the original theoretical
emotional decisions134. Both mirror theories and the perceptionaction model (PAM) review paper 1, many studies have further tested this view
presume that motor acts and affective states can transfer from a target to an observer in through a variety of methods, including positron emission
a bottomup, goal-relevant manner through shared representations for perception and tomography, functional neuroimaging (functional MRI
action. However, the PAM explains a wider range of phenomena and empathic
(fMRI)), psychophysiology, electromyography, transcra-
biases through its focus on distributed, person-specific representations that are
nial magnetic stimulation and electroencephalography
formed through experience7.
(for reviews, see REFS9496).

6 | ADVANCE ONLINE PUBLICATION www.nature.com/nrn



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Bottomup Research has confirmed the existence of overlapping expression and empathy (BOX1). There are more studies in
Describes a neural and mental neural signatures for experiencing and observing affective this rich field than can be described here, but we summa-
process that is stimulus-driven states, including happiness, anger, fear, disgust, sadness9799 rize the main findings and controversies below, focusing
on the basis of directly and most often pain (for a review, see REF.100). The on the implications for thePAM.
observed information without
requiring explicit cognitive
data are largely consistent, logical and supportive of the
processes or capacities. PAM. However, there is controversy in the field that orig- Levels of empathy in the brain. Even though the PAM
Empathy arises from inates from divergent views of what overlap between rep- emphasizes the spontaneous activation of the observers
bottomup processes that are resentations means (for example, it might be mistaken for representations for the targets state7, it is not limited
shared across species such as
signatures that are identical or without distinct proper- to reflexive processes such as mimicry and emotional
motor mimicry, emotional
contagion and state matching.
ties). In addition, controversy persists regarding the role contagion, but also addresses complex forms of empa-
of mirror neurons or mirror systems in empathy. Mirror thy, including the effects of attention, regulation, expe-
theories are similar to the PAM in that both rely on shared rience and culture. Perceptionaction processes are
representations for perception and action. However, the indeed more noticeable during contagion and mimicry.
PAM focuses more on distributed representations that However, the PAM is a proximate-level theory that pre-
include relevant feelings, memories and associations sumes a shared, distributed, neural representation for
that are related to the target, and to the targets state and the self and other that is engaged whenever the state
situation. Even if mirror theories are sometimes maligned of another individual is decoded, even if the shared
because of popularized media accounts about them, they property is not always obvious to or felt by the observer.
should not be dismissed, but rather understood in con- If observers attend to anothers state, this produces
junction with information about neural coding strate- empathy and accuracy in a highly graded fashion that
gies and the function of the inferior frontal gyrus (IFG) increases with personal and contextual factors, and
in the perception and production of language, gesture, can be inhibited in many different ways (for example,
avoiding attention, focusing on the response and motor
inhibition at multiple levels of the system). As representa-
Target Observer
tions are person-specific and developed through experi-
ence, they are fundamentally sensitive to the observers
Cognitive unique history, which explains why empathy and accu-
racy increase with the relevance of the targets state to
the observers own past experiences, and to their simi-
Aective larity and familiarity; these effects occur across species
(as discussed above). As a result of this graded quality,
In cognitive empathy, top-down one cannot make binary statements about experiences
imagination activates the
aective representations that that are the same or different, or in which empathy or
are also involved in bottom-up accuracy either does or does not exist7,101.
aective empathy, and this Observers must process the targets state using what-
provides the imagination
with information regarding ever relevant representations they possess at least
the emotional states of others initially as part of the process of basic information
decoding. Degrees of understanding that result from
this process range from a rudimentary sense that the
other feels bad to a highly tailored, accurate under-
Distributed associated representations that standing of that specific state and its entailments98,102.
are involved in aective and cognitive empathy: There is also information about the target in the observ-
Physiological states Facial expressions ers activated representation, which makes it possible
Direct perception leads to Semantic concepts Body postures
bottom-up, aective empathy Associated memories
for the observer to understand dissimilar others on the
basis of specific knowledge about them. Finally, shared
representations do not always lead to felt resonance,
Figure 4 | Both cognitive and affective empathy access Nature Reviewsperson-specific
distributed, | Neuroscience
compassion or help, particularly when the observers
affective representations. Bottomup, affective empathy (red box) occurs when an
goals conflict with those of the target; this explains the
observer directly perceives the emotional state (such as a sad facial expression) of the
target. This naturally activates distributed, personal representations of the targets state
importance of interdependence for prosocial outcomes.
in the observer (purple box). These representations have developed over time with Whereas neuroscience has shown that there are dis-
experience in the observers life and include associated memories, semantic concepts, tinct brain correlates for emotional and cognitive empathy
and bodily states and expressions. When empathy proceeds in a top-down, cognitive (for examples, see REFS103106), and clinicians observe
manner (blue box), the neural regions that support working memory, executive distinct emotional or cognitive empathy impairments in
function, emotion regulation and visuospatial processes instead access the affective clinical disorders107,108, these distinctions do not under-
empathy representations from the top-down (indicated by the arrows). Thus, although mine the shared-representations model. It is sometimes
the stimulation emerges from inside the mind rather than from the outside world, the useful to bifurcate empathy into affective and cognitive
affective regions and the associated representations are shared between cognitive
and affective processes. Thus, subtractions of affective from cognitive forms of
components, but this does not deny shared processes.
empathy reveal greater brain activity in the blue regions than in the red regions. Shared processing is required for both bottomup and top-
However, the cognitive process must still access the affective regions and their shared down forms of empathy because it is the distributed affective
associated representations (purple) to provide the imagination or simulation with representation that imbues the percept with content and
content and meaning. meaning, and this is needed for all forms of empathy (FIG.4).

NATURE REVIEWS | NEUROSCIENCE ADVANCE ONLINE PUBLICATION | 7



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Top-down This initial decoding phase of processing is mostly brief during theory-ofmind tasks is more medial when the
Describes a neural and mental and not subjectively experienced, except in rare cases task is more affective (for example, involving the orbito
process that requires of sustained resonance. Moreover, their shared proper- frontal cortex and/or VMPFC) and more lateral when
a conscious, cognitive ties cannot be demonstrated using procedures designed it is more cognitive (involving the DLPFC) (for exam-
evaluation to take into account
information that is not directly
to emphasize differences, such as the fMRI subtraction ples, see REFS103106). Mentalizing perse that is,
observable, such as taking method or even multi-voxel pattern analysis. By analogy, thinking about what the other thinks or believes, or tak-
anothers perspective or our concept of a chair includes the affordance of sitting in ing their perspective is particularly associated with
reasoning about their state on a chair, which can be associated with motor activity when activation of the STG and TPJ106,109,110. The amygdala is
the basis of conceptual
viewing chairs. However, it does not follow from this that not always implicated in empathy because it habituates
knowledge. These processes
participate in more advanced
our neural correlates for viewing chairs and tree stumps quickly to repeated stimuli, but it is engaged by tasks that
forms of empathy that are are indistinguishable, or that either will always be accom- involve salient states such as fear and distress, or learning
more sophisticated in humans panied by subjectively felt pressure on our backsides. associations between these states and outcomes111113.
and with age, but they are not When human subjects are presented with differ- However, just because neural regions are engaged to
required to simply understand
how someone feels.
ent tasks, concepts, instructions or stimuli, we obvi- different extents, or are engaged more reliably in one task
ously expect differences in brain activity. For example, than in another, does not mean that they are completely
emotional empathy tasks engage brain regions that separate and do not share properties. Researchers need
are associated with affective and motor-motivational to focus on the inherent relationship between the task,
processes to a greater extent (for example, the insula, instructions and analysis strategy and their resulting
anterior cingulate cortex (ACC), thalamus, amygdala, neural correlates. Even if we argue against oversim-
fusiform gyrus, somatosensory and motor cortices, and plified claims about neural distinctiveness or overlap,
ventromedial prefrontal cortex (VMPFC)). Cognitive it is true that we need to study the precise way in which rep-
empathy and perspective-taking tasks engage exec- resentations overlap versus the precise way in which they
utive, working memory and visuospatial processes are distinct, particularly at the scale of neurons and their
to a greater extent (for example, the dorsolateral PFC distributed interconnections.
(DLPFC), dorsomedial PFC (DMPFC), superior tempo-
ral gyrus (STG), temporoparietal junction (TPJ), supe- Empathic pain. Owing to its ease of study, physical pain
rior parietal lobule (SPL) and inferior parietal lobule is the most common context for testing shared rep-
(IPL)) (FIG.5). In addition, the frontal activity observed resentations between the self and the other. Typically,

aMCC, SMA DLPFC M1 S1


ACC and CMA Thalamus
S2
IPL

TPJ

VMPFC PI
AI and
IFG
Amygdala TP
Thalamus FG
STG Hypothalamus FG

Regions emphasized by Regions emphasized by Regions shared between cognitive


aective empathy tasks cognitive empathy tasks and aective tasks

Nature Reviews | Neuroscience


Figure 5 | Neural regions that participate in human empathy. The figure shows a medial sagittal view (far left), a left
lateral side view (centre) and a coronal view (far right) of the human brain, and indicates the relative locations and roles of
brain areas that are involved in human empathy. Affective representations are required to imagine how another person
feels. Depending on the task, however, the observed neural activity will emphasize bottomup, affective brain areas or
top-down, cognitive ones. Regions that are more associated with affective empathy tasks such as direct perception of
the emotion or pain of another person are shown in red: the anterior cingulate cortex (ACC) including perigenual and
subgenual regions of the ACC, amygdala, thalamus, hypothalamus, primary motor cortex (M1), premotor cortex, the
primary and secondary somatosensory cortices (S1 and S2), and the temporal pole (TP). Brain regions that are more
associated with higher-level, top-down forms of empathy such as imagining how you would feel in the place of another
or taking their perspective are shown in blue: the dorsolateral prefrontal cortex (DLPFC), inferior parietal lobule (IPL),
temporoparietal junction (TPJ), superior temporal gyrus (STG) and fusiform gyrus (FG). Two regions that are commonly
activated in both affective and cognitive tasks are shown in green: the anterior insula (AI) and the anterior middle
cingulate cortex (aMCC), which extends dorsally into the supplementary motor area (SMA) and cingulate motor area
(CMA). Neural locations are approximated so that the regions can be viewed in just three images (for example, the FG is
normally too medial to be seen in the lateral image, the amygdala is shown in the medial view as emerging from behind the
visible midline slice in right temporal cortex, and differences in laterality by region and task are not represented). IFG,
inferior frontal gyrus; PI, posterior insula; VMPFC, ventromedial PFC.

8 | ADVANCE ONLINE PUBLICATION www.nature.com/nrn



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

participants view images of injury to another (for exam- effects are influenced by personal experience, patients
ple, a hand closed in a door or cut with a knife) or sym- with congenital insensitivity to pain (CIP) who do
bols that are associated with the application of a noxious not experience peripheral pain are also impaired
stimulus to another (for example, an electrical shock or at detecting others pain and can feel indifferent to it,
a hand submerged in ice-cold water). In other phases of although their performance can be improved by trait
the experiment, the same participants receive a noxious concern for others122 (BOX2).
stimulus to their own body. Neural activity is operation- The precise meaning of the empathic pain network
alized as empathic pain when the same brain region is is debated (see REFS118,123126). Generally, researchers
activated during pain to the self and pain to the other. assume that the anterior insula and aMCC do not repre-
Data from these studies strongly support the PAM in sent the sensation of physical pain, such as the feeling of
that pain representations are shared but not identical, a prick on your arm or a burn on your fingertip. Instead,
and are widely distributed and highly specific to the these regions are thought to encode something about
state, task and the observers prior experiences. Almost the salient, aversive or arousing properties that pain
all studies find that both experiencing pain (self pain) shares with other experiences. Even the earliest reports
and perceiving anothers pain (other pain or empathic of empathic pain suggested that it represents affective
pain) activate the anterior insula and cingulate cortex rather than sensory components114. Similarly, a recent
(from the posterior ACC (pACC) to the anterior middle meta-analysis found that the bilateral anterior insula and
cingulate cortex (aMCC))114117. A meta-analysis has con- the dorsal ACC (dACC)aMCCsupplementary motor
firmed early empirical reports, and found that viewing area (SMA) were reliably activated during empathy for
either pain or symbols of pain to the other consistently pain and for basic emotions125. Moreover, self and other
activate the bilateral anterior insula (merging into the experiences activate the left anterior insula and mACC
IFG), and the pACC or aMCC118. for three distinct negative experiences (pain, disgust
There is also pharmacological evidence that ones and economic unfairness)126. In addition, patients with
own pain influences how one processes the pain of oth- CIP still show anterior insula and aMCC activity when
ers. Acetaminophen decreases self pain and empathic observing others pain, which suggests that the activation
pain for physical and social forms of distress 119. does not derive from peripheral sensations127 (BOX2).
Analgesia reduces self pain and empathic pain, and this Of course, introspection alone is sufficient to tell us
is accompanied by reduced pain ratings and alterations that empathic pain does not usually make people feel
in the pain related P2 type event-related potential120. the precise peripheral sensation that they witness being
Even placebo analgesia reduces empathic pain and experienced by another. People do not look at their own
associated activity in the anterior insula and aMCC, hand when they see someone touch a hot stove or say
and this effect can be blocked by the opioid antago- Ouch!. They are more likely to feel an aversive sicken-
nist naltrexone121. Moreover, demonstrating that pain ing feeling in their stomach and wrinkle their nose in
disgust while turning away and exclaiming Ooh! or
Ugh!. Their experiences differ from those of the other.
Box 2 | The mysterious case of empathic pain in patients who do not feel any
As such, they must also differ at the level of the brain
while sharing affective properties aswell.
Functional MRI (fMRI) has shown that the anterior insula and anterior middle cingulate Although the representations in the anterior insula
cortex (aMCC) are activated in patients with congenital insensitivity to pain (CIP) are unlikely to be specific or exclusive to felt pain (at
during empathic pain122,127. Does this mean that empathy does not require a similar least in this anterior region), they probably represent
experience or access to personal representations of pain? The data tell a different story.
feeling states that are more like the internal gut feelings
Patients with CIP are actually impaired at perceiving others pain and empathizing
with them, and this impairment scales with their pain insensitivity. In one study,
that are experienced in a wide variety of intense states,
patients with CIP frequently suspected other people of exaggerating their pain and including disgust, pain, dread, anxiety, stomach aches
often considered others including their friends or spouse as sissies (REF.122). and even desire. The anterior insula is known to repre-
Some patients showed no reaction to others pain at all, in experiments or real life122. sent interoceptive states across species128, and patients
Patients with CIP also rate others pain as less aversive, they have a less intense with CIP who have spared anterior insula activation in
empathic pain response in the left anterior insula and inferior frontal gyrus, and the response to empathic pain do experience internal sen-
aMCC, and they lack the aMCC response to others facial pain. In short, patients with sations such as headaches and stomach aches127. The
CIP are impaired at empathizing with others pain. aMCC is assumed to respond to any unexpected out-
Patients with CIP also do not have a deficit for all types of pain, only peripheral pain, come that requires a fast or reissued response, including
and some somatic pain (for example, headaches and migraines) remains intact122.
pain, non-pain negative states, cognitive-motor errors
These intact representations allow them to relate to the noxious, aversive, motivating
properties of pain that are represented in the anterior insula and aMCC. Patients with
and even unexpected rewards (see REFS71,129131).
CIP also learn how to interpret pain cues, allowing them to pass easy behavioural tests This motor-motivational view of the aMCC is consist-
such as rating facial pain on a three-point scale122. ent with other proposals71,125,131 and with the caregiving
Therefore, data from patients with CIP support the perceptionaction model by model of altruism78. These interpretations are consist-
showing that it is harder to feel concerned about anothers state in the absence of ent with the core principle of the PAM that the brain
a highly relevant, elaborate representation of the same state from personal integrates sensation and action more than linguistic and
experience. It also supports the view that the brain uses whatever representations disciplinary categories tendto.
are available to understand others, in a way that scales with the relevance of the Even if the anterior insula and aMCC are not spe-
observers experiences7. Experience with ones own pain thus influences ones ability cific to felt pain, one should not conversely infer that
to empathize with anothers.
empathy never includes bodily sensations or peripheral

NATURE REVIEWS | NEUROSCIENCE ADVANCE ONLINE PUBLICATION | 9



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

stimulation. Ideomotor actions and spontaneous mim- Research supports an evolutionary, perceptionaction
icry demonstrate the effector-specific, peripheral effects view in which observers spontaneously imitate, mimic
of the underlying neural perceptionaction process90. In and feel into the states of those they attend to through
the emotional domain, the somatosensory cortices are a neural process by which the targets states are mapped
activated during a variety of empathy tasks includ- onto the observers distributed learned and personal
ing those involving pain104, facial emotion recognition93 representations for the target, state and situation. This
and emotional empathy 105 thus indicating a role for shared component of self and other representations
sensation in observing emotions, particularly when the explains the strong biases for similarity, familiarity and
task emphasizes sensation. The anterior insula which past experience that are found across species.
is consistently activated during empathic pain (see All species may not possess complex capacities to
above) is also active during observed and imitated put themselves into anothers shoes in the same goal-
facial emotion132, felt and observed disgust 97,133, and the directed manner as is seen in humans, but most car-
feeling of butterflies in the stomach while anticipating egiving mammals that live in interdependent social
risk134. Thus, there are times when sensations have a groups are affected by the pain and distress of familiar
more prominent role in empathic emotion perception, others in ways that may propel them to act on their
particularly when the sensation is made salient by the behalf. For some species, there is experimental evi-
stimuli and task instructions. For example, rating pain dence of well-developed empathic perspective-taking
intensity on the targets limb activates more sensory and targeted helping. Shared representations of affec-
areas than does rating pain aversiveness or viewing tive states are activated from the top down in more
facial pain expressions without knowing the cause124. cognitive forms of empathy, which recruit additional
Tasks that focus attention on the location of the injury executive and visuospatial processes. However, the lit-
activate more sensory activity (for example, in S1, S2 erature overestimates distinctions between emotional
and the SMA), whereas tasks that ask subjects to place and cognitive empathy, following traditional practices
themselves into the shoes of the other recruit more to dichotomize in science and philosophy. Despite each
visuospatial, executive and working-memory processes having unique features, affective and cognitive empa-
(mentalizing; for example, the STG, TPJ, IPL, DMPFC thy both require access to the shared representations of
and VMPFC)100,118,124. By carefully linking neural regions emotion that provide simulations with content and an
to the focus of the task and the experience of the subject, embodied meaning.
the many seemingly puzzling results from different stud- Future research must examine empathy and its mech-
ies begin to makesense. anisms in a wider variety of species, even beyond the
mammals, to better understand the underlying mech-
Conclusion anisms and how empathy shifts with social ecology. In
Our scientific understanding of empathy has greatly addition, neuroscience must move beyond oversimplified
improved within the past two decades owing to a huge claims of same or different processes, and apply a more
amount of research into how this capacity evolved, exists phenomenological approach that contextualizes results
across species, and is instantiated in the body and brain. within the tasks, instructions and analyses employed.

1. Preston,S.D. & de Waal,F.B.M. Empathy: its ultimate 12. de Waal,F.B.M. Good Natured (Harvard Univ. Press, 23. Miller,R.E., Murphy,J.V. & Mirsky,I.A. Relevance of
and proximate bases. Behav. Brain Sci. 25, 171 1996). facial expression and posture as cues in
(2002). 13. Krupenye,C., Kano,F., Hirata,S., Call,J. & Tomasello,M. communication of affect between monkeys. AMA Arch.
This paper offers the first summary of empathy as Great apes anticipate that other individuals will act Gen. Psychiatry 1, 480488 (1959).
a perceptionaction process and reviews evidence according to false beliefs. Science 354, 110114 (2016). 24. Yerkes,R.M. Almost Human (Century, 1925).
from multiple levels of analysis. 14. de Waal,F.B.M. The Age of Empathy (Harmony 25. Ladygina-Kohts,N.N. Infant Chimpanzee and Human
2. Goldman,A.I. Simulating Minds: the Philosophy, Books, 2009). Child (ed. de Waal,F.B.M.) (Oxford Univ. Press,
Psychology, and Neuroscience of Mindreading (Oxford 15. Yamamoto,S., Humle,T. & Tanaka,M. Chimpanzees 2001)
Univ. Press, 2006). flexible targeted helping based on an understanding of 26. de Waal,F.B.M. & van Roosmalen,A. Reconciliation
3. Baron-Cohen,S. in The Lost Self: Pathologies of the conspecifics goals. Proc. Natl Acad. Sci. USA 109, and consolation among chimpanzees. Behav. Ecol.
Brain and Identity (eds Feinberg,T.E. & Keenan,J.P.) 35883592 (2012). Sociobiol. 5, 5566 (1979).
166180 (Oxford Univ. Press, 2005). 16. Hattori,Y., Leimgruber,K., Fujita,K. & de Waal,F.B. This is the first non-human animal study to define
4. Zahavi,D. Simulation, projection and empathy. M. Food-related tolerance in capuchin monkeys and analyse consolation behaviour.
Conscious. Cogn. 17, 514522 (2008). (Cebus apella) varies with knowledge of the partners 27. Zahn-Waxler,C., Radke-Yarrow,M., Wagner,E. &
5. Lipps,T. Einfhlung, innere nachahmung und previous food-consumption. Behaviour 149, 171 Chapman,M. Development of concern for others. Dev.
organenempfindungen [German]. Arch. Gesamte 185 (2012). Psychol. 28, 126136 (1992).
Psychol. 1, 465519 (1903). 17. Bateson,P. & Laland,K.N. Tinbergens four questions: This is a pioneering study of the early development
6. de Waal,F.B.M. Putting the altruism back into an appreciation and an update. Trends Ecol. Evol. 28, of human consolation behaviour.
altruism: the evolution of empathy. Annu. Rev. Psychol. 712718 (2013). 28. Liddle,M.J.E., Bradley,B.S. & Mcgrath,A. Baby
59, 279300 (2008). 18. Tinbergen,N. On aims and methods of ethology empathy: infant distress and peer prosocial responses.
7. Preston,S.D. & Hofelich,A.J. The many faces of [German]. Z.Tierpsychol. 20, 410433 (1963). Infant Ment. Health J. 36, 446458 (2015).
empathy: parsing empathic phenomena through 19. Church,R.M. Emotional reactions of rats to the pain of 29. Roth-Hanania,R., Davidov,M. & Zahn-Waxler,C.
a proximate, dynamic-systems view of representing the others. J.Comp. Physiol. Psychol. 52, 132134 (1959). Empathy development from 8to 16months: early
other in the self. Emot. Rev. 4, 2433 (2012). This is the first rodent study to analyse emotional signs of concern for others. Infant Behav. Dev. 34,
8. Zahn-Waxler,C. & Radke-Yarrow,M. The origins of responses to the distress of others. 447458 (2011).
empathic concern. Motiv. Emot. 14, 107130 (1990). 20. Rice,G.E. & Gainer,P. Altruism in the albino rat. 30. Dimberg,U., Thunberg,M. & Elmehed,K.
9. Walter,H. Social cognitive neuroscience of empathy: J.Comp. Physiol. Psychol. 55, 123125 (1962). Unconscious facial reactions to emotional facial
concepts, circuits, and genes. Emot. Rev. 4, 917 (2012). 21. Masserman,J.H., Wechkin,S. & Terris,W. Altruistic expressions. Psychol. Sci. 11, 8689 (2000).
10. Darwin,C. The Descent of Man, and Selection in behavior in rhesus monkeys. Am. J.Psychiatry 121, This study is part of a series of studies on facial
Relation to Sex (Princeton Univ. Press, 1982). 584585 (1964). mimicry that first demonstrated the spontaneous
11. Plutchik,R. in Empathy and its Development (eds 22. Wechkin,S., Masserman,J.H. & Terris,W. Shock to nature of human empathic reactions.
Eisenberg,N. & Strayer,J.) 346 (Cambridge Univ. a conspecific as an aversive stimulus. Psychon. Sci. 1, 31. Mancini,G., Ferrari,P.F. & Palagi,E. Rapid facial
Press, 1987). 4748 (1964). mimicry in geladas. Sci. Rep. 3, 1527 (2013).

10 | ADVANCE ONLINE PUBLICATION www.nature.com/nrn



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

32. Ross,M.D., Menzler,S. & Zimmermann,E. Rapid 58. Sato,N., Tan,L., Tate,K. & Okada,M. Rats 85. Dittus,W.P. & Ratnayeke,S.M. Individual and social
facial mimicry in orangutan play. Biol. Lett. 4, 2730 demonstrate helping behavior toward a soaked behavioral responses to injury in wild toque macaques
(2008). conspecific. Anim. Cogn. 18, 10391047 (2015). (Macaca sinica). Int. J.Primatol. 10, 215234 (1989).
33. Dimberg,U., Andrasson,P. & Thunberg,M. 59. Romero,T., Castellanos,M.A. & de Waal,F.B. M. 86. De Vignemont,F. & Singer,T. The empathic brain:
Emotional empathy and facial reactions to facial Consolation as possible expression of sympathetic how, when and why? Trends Cogn. Sci. 10, 435441
expressions. J.Psychophysiol. 25, 2631 (2011). concern among chimpanzees. Proc. Natl Acad. Sci. (2006).
34. Campbell,M.W. & de Waal,F.B.M. Ingroup-outgroup USA 107, 1211012115 (2010). 87. de Waal,F.B.M. Chimpanzee Politics (Johns Hopkins
bias in contagious yawning by chimpanzees supports 60. Clay,Z. & de Waal,F.B. M. Bonobos respond to Univ. Press, 1998).
link to empathy. PLoS ONE 6, e18283 (2011). distress in others: consolation across the age 88. Wilson,M.L. etal. Lethal aggression in Pan is better
35. Norscia,I. & Palagi,E. Yawn contagion and empathy in spectrum. PLoS ONE 8, e55206 (2013). explained by adaptive strategies than human impacts.
Homo sapiens. PLoS ONE 6, e28472 (2011). 61. Palagi,E., DallOlio,S., Demuru,E. & Stanyon,R. Nature 513, 414417 (2014).
36. Romero,T., Konno,A. & Hasegawa,T. Familiarity bias Exploring the evolutionary foundations of empathy: 89. Lanzetta,J.T. & Englis,B.G. Expectations of
and physiological responses in contagious yawning by consolation in monkeys. Evol. Hum. Behav. 35, cooperation and competition and their effects on
dogs support link to empathy. PLoS ONE 8, e71365 341349 (2014). observers vicarious emotional responses. J.Pers. Soc.
(2013). 62. Cools,A.K., Van Hout,A.J.M. & Nelissen,M.H. Psychol. 56, 543554 (1989).
37. Gallup,A.C., Swartwood,L., Militello,J. & Sackett,S. Canine reconciliation and third-party-initiated 90. Prinz,W. Perception and action planning. Eur. J.Cogn.
Experimental evidence of contagious yawning in postconflict affiliation: do peacemaking social Psychol. 9, 129154 (1997).
budgerigars (Melopsittacus undulatus). Anim. Cogn. mechanisms in dogs rival those of higher primates? 91. di Pellegrino,G., Fadiga,L., Fogassi,L., Gallese,V. &
18, 10511058 (2015). Ethology 114, 5363 (2008). Rizzolatti,G. Understanding motor events:
38. Silva,K., Bessa,J. & de Sousa,L. Auditory contagious 63. Plotnik,J.M. & de Waal,F.B. M. Asian elephants a neurophysiological study. Exp. Brain Res. 91,
yawning in domestic dogs (Canis familiaris): first (Elephas maximus) reassure others in distress. PeerJ 176180 (1992).
evidence for social modulation. Anim. Cogn. 15, 2, e278 (2014). This study reports the discovery of mirror neurons,
721724 (2012). 64. Seed,A.M., Clayton,N.S. & Emery,N.J. Postconflict which are a subclass of cells that activate similarly
39. Visalberghi,E. & Fragaszy,D.M. in Language and third-party affiliation in rooks, Corvus frugilegus. Curr. when a macaque or the experimenter performs
Intelligence in Monkeys and Apes: Comparative Biol. 17, 152158 (2007). a goal-directed motor action.
Developmental Perspectives (eds Parker,S.T. & 65. Fraser,O.N., Stahl,D. & Aureli,F. Stress reduction 92. Grafton,S.T., Fadiga,L., Arbib,M.A. & Rizzolatti,G.
Gibson,K.R.) 247273 (Cambridge Univ. Press, through consolation in chimpanzees. Proc. Natl Acad. Premotor cortex activation during observation and
1990). Sci. USA 105, 85578562 (2008). naming of familiar tools. Neuroimage 6, 231236
40. Ferrari,P.F. etal. Neonatal imitation in rhesus 66. Aureli,F., Preston,S.D. & de Waal,F.B.M. Heart rate (1997).
macaques. PLoS Biol. 4, e302 (2006). responses to social interactions in free-moving rhesus 93. Adolphs,R., Damasio,H., Tranel,D., Cooper,G. &
41. Paukner,A., Suomi,S.J., Visalberghi,E. & macaques (Macaca mulatta): a pilot study. J.Comp. Damasio,A.R. A role for somatosensory cortices in
Ferrari,P.F. Capuchin monkeys display affiliation Psychol. 113, 5965 (1999). the visual recognition of emotion as revealed by
toward humans who imitate them. Science 325, 67. Ben-Ami Bartal,I. etal. Anxiolytic treatment impairs three-dimensional lesion mapping. J.Neurosci. 20,
880883 (2009). helping behavior in rats. Front. Psychol. 7, 850 26832690 (2000).
42. Perry,S. etal. Social conventions in wild white-faced (2016). This study offers early support for the PAM by
capuchin monkeys: evidence for traditions in 68. Warneken,F., Hare,B., Melis,A.P., Hanus,D. & showing that damage to the somatosensory cortex
a neotropical primate. Curr. Anthropol. 44, 241268 Tomasello,M. Spontaneous altruism by chimpanzees (where the body is represented) impairs an
(2003). and young children. PLoS Biol. 5, e184 (2007). individuals ability to recognize the facial emotions
43. Dindo,M., Whiten,A. & de Waal,F.B.M. Ingroup 69. Warneken,F. & Tomasello,M. Extrinsic rewards of others.
conformity sustains different foraging traditions in undermine altruistic tendencies in 20month-olds. 94. Jackson,P.L., Brunet,E., Meltzoff,A.N. & Decety,J.
capuchin monkeys (Cebus apella). PLoS ONE 4, Motiv. Sci. 1, 4348 (2014). Empathy examined through the neural mechanisms
e7858 (2009). 70. Andreoni,J. Giving with impure altruism: applications involved in imagining how I feel versus how you feel
44. Horner,V. & Whiten,A. Causal knowledge and to charity and Ricardian equivalence. J.Polit. Econ. pain. Neuropsychologia 44, 752761 (2006).
imitation/emulation switching in chimpanzees (Pan 97, 14471458 (1989). 95. Singer,T. & Lamm,C. The social neuroscience of
troglodytes) and children (Homo sapiens). Anim. 71. Wager,T.D. etal. Pain in the ACC? Proc. Natl Acad. empathy. Ann. NY Acad. Sci. 1156, 8196 (2009).
Cogn. 8, 164181 (2005). Sci. USA 113, E2474E2475 (2016). 96. Zaki,J. & Ochsner,K.N. The neuroscience of
45. Fuhrmann,D., Ravignani,A., Marshall-Pescini,S. & 72. Ballesta,S. & Duhamel,J.R. Rudimentary empathy in empathy: progress, pitfalls and promise. Nat.
Whiten,A. Synchrony and motor mimicking in macaques social decision-making. Proc. Natl Acad. Neurosci. 15, 675680 (2012).
chimpanzee observational learning. Sci. Rep. 4, 5283 Sci. USA 112, 1551615521 (2015). 97. Wicker,B. etal. Both of us disgusted in my insula: the
(2014). 73. Christov-Moore,L. & Iacoboni,M. Selfother common neural basis of seeing and feeling disgust.
46. Hopper,L.M., Lambeth,S.P., Schapiro,S.J. & resonance, its control and prosocial inclinations: Neuron 40, 655664 (2003).
Whiten,A. Observational learning in chimpanzees and brainbehavior relationships. Hum. Brain Mapp. 37, 98. Preston,S.D. etal. The neural substrates of cognitive
children studied through ghost conditions. Proc. 15441558 (2016). empathy. Soc. Neurosci. 2, 254275 (2007).
R.Soc. B Biol. Sci. 275, 835840 (2008). 74. Hoffman,M.L. Is altruism part of human nature? 99. Morelli,S.A. & Lieberman,M.D. The role of
47. Langford,D.J. etal. Social modulation of pain as J.Pers. Soc. Psychol. 40, 121137 (1981). automaticity and attention in neural processes
evidence for empathy in mice. Science 312, 75. Nowbahari,E., Scohier,A., Durand,J.L. & Hollis,K.L. underlying empathy for happiness, sadness, and
19671970 (2006). Ants, Cataglyphis cursor, use precisely directed rescue anxiety. Front. Hum. Neurosci. 7, 160 (2013).
This paper provides compelling evidence of behavior to free entrapped relatives. PLoS ONE 4, 100. Zaki,J., Wager,T.D., Singer,T., Keysers,C. &
emotional contagion in mice and set the stage for e6573 (2009). Gazzola,V. The anatomy of suffering: understanding
subsequent rodent research. 76. Vasconcelos,M., Hollis,K., Nowbahari,E. & the relationship between nociceptive and empathic
48. Langford,D.J. etal. Social approach to pain in Kacelnik,A. Pro-sociality without empathy. Biol. Lett. pain. Trends Cogn. Sci. 20, 249259 (2016).
laboratory mice. Soc. Neurosci. 5, 163170 (2010). 8, 910912 (2012). 101. Mohr,A.H., Kross,E. & Preston,S.D. Devil in the
49. Martin,L.J. etal. Reducing social stress elicits 77. Wilson,E.O. A chemical releaser of alarm and digging details: effects of depression on the prosocial
emotional contagion of pain in mouse and human behavior in the ant Pogonomyrmex badius (Latreille). response depend on timing and similarity.
strangers. Curr. Biol. 25, 326332 (2015). Psyche 65, 4151 (1958). Adapt. Hum. Behav. Physiol. 2, 281297 (2016).
50. Burkett,J. etal. Oxytocin-dependent consolation 78. Preston,S.D. The origins of altruism in offspring care. 102. Hodges,S.D., Kiel,K.J., Kramer,A.D., Veach,D. &
behavior in rodents. Science 351, 375378 (2016). Psychol. Bull. 139, 13051341 (2013). Villanueva,B.R. Giving birth to empathy: the effects of
51. Buchanan,T.W., Bagley,S.L., Stansfield,R.B. & This is a comprehensive review of the similar experience on empathic accuracy, empathic
Preston,S.D. The empathic, physiological resonance offspring-care perspective on how altruism and concern, and perceived empathy. Pers. Soc. Psychol.
of stress. Soc. Neurosci. 7, 191201 (2012). heroic rescue evolved and are encoded in the brain. Bull. 36, 398409 (2010).
This paper provides the first demonstration that 79. MacLean,P.D. Brain evolution relating to family, play, 103. Vllm,B.A. etal. Neuronal correlates of theory of
stress at the physiological level spreads from and the separation call. Arch. Gen. Psychiatry 42, mind and empathy: a functional magnetic resonance
one stressed human participant to empathic 405417 (1985). imaging study in a nonverbal task. Neuroimage 29,
observers. 80. Sivaselvachandran,S., Acland,E.L., Abdallah,S. & 9098 (2006).
52. Buchanan,T.W. & Preston,S.D. Stress leads to Martin,L.J. Behavioral and mechanistic insight into 104. Lamm,C., Nusbaum,H.C., Meltzoff,A.N. &
prosocial action in immediate need situations. rodent empathy. Neurosci. Biobehav. Rev. http:// Decety,J. What are you feeling? Using functional
Front. Behav. Neurosci. http://dx.doi.org/10.3389/ dx.doi.org/10.1016/j.neubiorev.2016.06.007 magnetic resonance imaging to assess the modulation
fnbeh.2014.00005 (2014). (2016). of sensory and affective responses during empathy for
53. Clay,Z. & de Waal,F.B. M. Development of socio- 81. Edgar,J., Lowe,J., Paul,E. & Nicol,C. Avian maternal pain. PLoS ONE 2, e1292 (2007).
emotional competence in bonobos. Proc. Natl Acad. response to chick distress. Proc. R.Soc. B Biol. Sci. 105. Nummenmaa,L., Hirvonen,J., Parkkola,R. &
Sci. USA 110, 1812118126 (2013). 278, 31293134 (2011). Hietanen,J.K. Is emotional contagion special? An
54. Fries,A.B.W. & Pollak,S.D. Emotion understanding 82. Doody,J.S., Burghardt,G.M. & Dinets,V. Breaking fMRI study on neural systems for affective and
in postinstitutionalized Eastern European children. the socialnon-social dichotomy: a role for reptiles in cognitive empathy. Neuroimage 43, 571580 (2008).
Dev. Psychopathol. 16, 355369 (2004). vertebrate social behavior research? Ethology 119, 106. Schnell,K., Bluschke,S., Konradt,B. & Walter,H.
55. Kagan,J. Human morality is distinctive. J.Conscious. 95103 (2013). Functional relations of empathy and mentalizing: an
Stud. 7, 4648 (2000). 83. Donaldson,Z.R. & Young,L.J. Oxytocin, fMRI study on the neural basis of cognitive empathy.
56. Fehr,E. & Fischbacher,U. The nature of human vasopressin, and the neurogenetics of sociality. Neuroimage 54, 17431754 (2011).
altruism. Nature 425, 785791 (2003). Science 322, 900904 (2008). 107. Blair,R.J.R. Responding to the emotions of others:
57. Bartal,I.B.A., Decety,J. & Mason,P. Empathy and 84. Chen,Q., Panksepp,J.B. & Lahvis,G.P. Empathy is dissociating forms of empathy through the study of
pro-social behavior in rats. Science 334, 14271430 moderated by genetic background in mice. PLoS ONE typical and psychiatric populations. Conscious. Cogn.
(2011). 4, e4387 (2009). 14, 698718 (2005).

NATURE REVIEWS | NEUROSCIENCE ADVANCE ONLINE PUBLICATION | 11



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

108. Derntl,B., Seidel,E.M., Schneider,F. & Habel,U. 121. Rutgen,M., Seidel,E.M., Riecansky,I. & Lamm,C. 132. Carr,L., Iacoboni,M., Dubeau,M.C., Mazziotta,J.C.
How specific are emotional deficits? A comparison of Reduction of empathy for pain by placebo analgesia & Lenzi,G.L. Neural mechanisms of empathy in
empathic abilities in schizophrenia, bipolar and suggests functional equivalence of empathy and first- humans: a relay from neural systems for imitation to
depressed patients. Schizophr. Res. 142, 5864 hand emotion experience. J.Neurosci. 35, limbic areas. Proc. Natl Acad. Sci. USA 100,
(2012). 89388947 (2015). 54975502 (2003).
109. Samson,D., Apperly,I.A., Chiavarino,C. & 122. Danziger,N., Prkachin,K.M. & Willer,J.C. Is pain the 133. Calder,A.J., Keane,J., Manes,F., Antoun,N. &
Humphreys,G.W. Left temporoparietal junction is price of empathy? The perception of others pain in Young,A.W. Impaired recognition and experience of
necessary for representing someone elses belief. patients with congenital insensitivity to pain. Brain disgust following brain injury. Nat. Neurosci. 3,
Nat. Neurosci. 7, 499500 (2004). 129, 24942507 (2006). 10771078 (2000).
110. Saxe,R. & Wexler,A. Making sense of another mind: 123. Lieberman,M.D. & Eisenberger,N.I. The dorsal 134. Bechara,A., Damasio,H. & Damasio,A.R. Emotion,
the role of the right temporo-parietal junction. anterior cingulate cortex is selective for pain: results decision making and the orbitofrontal cortex. Cereb.
Neuropsychologia 43, 13911399 (2005). from large-scale reverse inference. Proc. Natl Acad. Cortex 10, 295307 (2000).
111. Adolphs,R., Baron-Cohen,S. & Tranel,D. Impaired Sci. USA 112, 1525015255 (2015). This paper shows how the orbitofrontal cortex
recognition of social emotions following amygdala 124. Krishnan,A. etal. Somatic and vicarious pain are informs decisions from ones own simulated
damage. J.Cogn. Neurosci. 14, 12641274 represented by dissociable multivariate brain affective state, allowing humans to imagine how
(2002). patterns. eLife 5, e15166 (2016). another person feels and how they would feel as
112. Blair,R.J.R. Fine cuts of empathy and the 125. Fan,Y., Duncan,N.W., de Greck,M. & Northoff,G. a result of possible decision outcomes.
amygdala: dissociable deficits in psychopathy and Is there a core neural network in empathy? An fMRI 135. Gallese,V., Keysers,C. & Rizzolatti,G. A unifying view
autism. Q.J.Exp. Psychol. (Hove) 61, 157170 based quantitative meta-analysis. Neurosci. Biobehav. of the basis of social cognition. Trends Cogn. Sci. 8,
(2008). Rev. 35, 903911 (2011). 396403 (2004).
113. Marsh,A.A. Neural, cognitive, and evolutionary This paper offers a meta-analysis of the role of the 136. Iacoboni,M. Imitation, empathy, and mirror neurons.
foundations of human altruism. Wiley Interdiscip. Rev. dACCaMCCSMA and the bilateral anterior insula Annu. Rev. Psychol. 60, 653670 (2008).
Cogn. Sci. 7, 5971 (2016). across empathy tasks, demonstrating 137. Mukamel,R., Ekstrom,A.D., Kaplan,J., Iacoboni,M.
114. Singer,T. etal. Empathy for pain involves the affective a domain-general role with similar recruitment for & Fried,I. Single-neuron responses in humans during
but not sensory components of pain. Science 303, self and other experiences. execution and observation of actions. Curr. Biol. 20,
11571162 (2004). 126. Corradi-DellAcqua,C., Tusche,A., Vuilleumier,P. & 750756 (2010).
115. Morrison,I., Lloyd,D., di Pellegrino,G. & Roberts,N. Singer,T. Cross-modal representations of first-hand 138. Keysers,C. & Gazzola,V. Social neuroscience: mirror
Vicarious responses to pain in anterior cingulate and vicarious pain, disgust and fairness in insular and neurons recorded in humans. Curr. Biol. 20,
cortex: is empathy a multisensory issue? Cogn. Affect. cingulate cortex. Nat. Commun. 7, 10904 (2016). R353R354 (2010).
Behav. Neurosci. 4, 270278 (2004). In this study, the anterior insula and aMCC are 139. Shamay-Tsoory,S.G., Aharon-Peretz,J. & Perry,D.
116. Jackson,P.L., Meltzoff,A.N. & Decety,J. How do we engaged across pain, disgust and unfairness, showing Two systems for empathy: a double dissociation
perceive the pain of others? A window into the neural that these regions are not pain specific but participate between emotional and cognitive empathy in inferior
processes involved in empathy. Neuroimage 24, in various negative self and other experiences. frontal gyrus versus ventromedial prefrontal lesions.
771779 (2005). 127. Danziger,N., Faillenot,I. & Peyron,R. Can we share Brain 132, 617627 (2009).
117. Morrison,I. & Downing,P.E. Organization of felt and a pain we never felt? Neural correlates of empathy in
seen pain responses in anterior cingulate cortex. patients with congenital insensitivity to pain. Neuron Acknowledgements
Neuroimage 37, 642651 (2007). 61, 203212 (2008). The authors thank their sources of funding during the prepa-
118. Lamm,C., Decety,J. & Singer,T. Meta-analytic 128. Craig,A.D. How do you feel? Interoception: the sense ration of this manuscript, including support for the Living
evidence for common and distinct neural networks of the physiological condition of the body. Links Center from Emory College, Georgia, USA (to
associated with directly experienced pain and Nat. Rev. Neurosci. 3, 655666 (2002). F.B.M.d.W.) and a grant from Rackham Graduate School at
empathy for pain. Neuroimage 54, 24922502 129. de la Vega,A., Chang,L.J., Banich,M.T., Wager,T.D. the University of Michigan, USA (to S.D.P.). The authors also
(2011). & Yarkoni,T. Large-scale meta-analysis of human thank Y. Lei for assistance with manuscript preparation, and
119. Mischkowski,D., Crocker,J. & Way,B.M. From medial frontal cortex reveals tripartite functional L. Bickel for creating the original brain figures.
painkiller to empathy killer: acetaminophen organization. J.Neurosci. 36, 65536562 (2016).
(paracetamol) reduces empathy for pain. Soc. Cogn. 130. Shackman,A.J. etal. The integration of negative Competing interests statement
Affect. Neurosci. 11, 13451353 (2016). affect, pain and cognitive control in the cingulate The authors declare no competing interests.
120. Rtgen,M. etal. Placebo analgesia and its cortex. Nat. Rev. Neurosci. 12, 154167 (2011).
opioidergic regulation suggest that empathy for pain 131. Morrison,I., Peelen,M.V. & Downing,P.E. The sight of Publishers note
is grounded in self pain. Proc. Natl Acad. Sci. USA others pain modulates motor processing in human Springer Nature remains neutral with regard to jurisdictional
112, E5638E5646 (2015). cingulate cortex. Cereb. Cortex 17, 22142222 (2007). claims in published maps and institutional affiliations.

12 | ADVANCE ONLINE PUBLICATION www.nature.com/nrn



2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.