The Biological Bases of Empathy

Transcription

1 The Biological Bases of Empathy Jean Decety and Claus Lamm Departments of Psychology and Psychiatry, and Center for Cognitive and Social Neuroscience, The University of Chicago, 5848 S. University Avenue, Chicago, IL 60637, USA Preprint version of book chapter printed in: J. T. Cacioppo & G. G. Berntson (Eds.), Handbook of Neuroscience for the Behavioral Sciences. New York: John Wiley and Sons. Human beings are intrinsically social. Our survival critically depends on social interactions with others, the formation of alliances and accurate social judgments (Cacioppo, 2002). We are motivated to form and maintain positive and significant relationships (Baumeister & Leary, 1995), and most of our actions are directed toward or are responses to others (Batson, 1990). No single factor can account for human social cognitive evolution (e.g., diet and climate), but the single most important factor is the increasing complexity of hominid social groups (Bjorlund & Bering, 2003). It is therefore logical that dedicated mechanisms have evolved to perceive, understand, predict and respond to the internal states (subjective in nature) of other individuals. The construct of empathy accounts for a fundamental aspect of social interaction (see Table 1 for definitions). Philosophers and psychologists have long debated the nature of empathy (e.g., Ickes, 2003; Smith, 1790; Thompson, 2001), and whether the capacity to share and understand other people s emotions sets humans apart from other species (e.g., de Waal, 2005). Here, we consider empathy as a construct accounting for a sense of similarity in feelings experienced by the self and the other, without confusion between the two individuals (Decety & Jackson, 2004; Decety & Lamm, 2006; Eisenberg, Spinrad & Sadovsky, 2006). The experience of empathy can lead to sympathy (concern for another based on the apprehension or comprehension of the other s emotional state or condition), or even personal distress (i.e., an aversive, self-focused emotional 1

2 reaction to the apprehension or comprehension of another s emotional state or condition) when there is confusion between self and other. Knowledge of empathic behavior is essential for an understanding of human social and moral development (Eisenberg et al., 1994). It is generally assumed that people who experience others emotion and feel concerned for them are motivated to help (Hoffman, 2000). Furthermore, various psychopathologies are marked by empathy deficits, and a wide array of psychotherapeutic approaches stress the importance of clinical empathy as a fundamental component of treatment (Farrow & Woodruff, 2007). All of these are good reasons to be interested in investigating the computational and biological mechanisms that underpin the processes involved in interpersonal sensitivity and intersubjectivity. Table 1. Definitional issues. Despite the abundance of definitions of empathy, it is possible and recommended to differentiate emotional contagion, empathy, sympathy and personal distress. - Emotional contagion is the tendency to automatically mimic and synchronize facial expressions, vocalizations, postures, and movements with those of another individual. - Empathy is an emotional response that stems from another s state and that is congruent with the other s emotional state. It involves at least a minimal distinction between self and other. Empathy is not a separate emotion by itself, but a kind of induction process by which emotions, both positive and negative, are shared. - Personal distress is an aversive state (e.g., anxiety, worry) that has not to be congruent with the other s state, and that leads to a self-oriented, egoistic reaction. - Sympathy (or empathic concern) refers to feelings of sorrow, or being sorry for another. It is often the consequence of empathy, although it is possible that sympathy results from cognitive perspective taking. Sympathy is believed to involve an other-oriented, altruistic motivation. Emotion can be considered a process that facilitates appropriate physiological responses to aid the survival of the organism. In recent years, there has been a great upsurge in neuroimaging investigations of empathy. Most of these investigations reflect the approach of social neuroscience, which combines research designs and behavioral measures used in social psychology with neuroscience markers (Cacioppo, Berntson et al., 2000; Cacioppo, 2002). Such an approach plays an important role in disambiguating competing theories in social psychology in general and in empathy-related research in particular (Decety & Hodges, 2006). For instance, one critical question debated among social psychologists is whether 2

3 perspective-taking instructions induce empathic concern and/or personal distress, and to what extent prosocial motivation springs from self-other overlap. In this chapter, we focus on social neuroscience research exploring how people respond behaviorally and neurally to the pain of others. The perception of pain in others constitutes an ecologically valid way to investigate the mechanisms underpinning the experience of empathy for two main reasons: first everybody knows what is pain it is a common and universal experience and knows what are its physical and psychological manifestations; second, we have good knowledge about the neurophysiological pathways that are involved in processing nociceptive information. Findings from recent functional neuroimaging studies of pain empathy demonstrate that the mere perception of another individual in pain results, in the observer, in the activation of the neural network involved in the processing of the first-hand experience of pain. This intimate (yet not complete) overlap between the neural circuits responsible for our ability to perceive the pain of others and those underpinning our own selfexperience of pain supports the shared-representation theory of social cognition (Decety & Sommerville, 2003). This theory posits that perceiving someone else s emotion and having an emotional response or subjective feeling state both fundamentally draw on the same computational processes and rely on somatosensory and motor representations (Sommerville & Decety, 2006). However, we argue that a complete self-other overlap in neural circuits can lead to personal distress and possibly be detrimental to empathic concern and prosocial behavior. Personal distress may even result in a more egoistic motivation to reduce it, by withdrawing from the stressor, for example, thereby decreasing the likelihood of prosocial behavior (Tice et al., 2001). The chapter starts with a discussion of the evolutionary origins of empathy focusing on the role of the autonomic nervous system, followed by a section on the role of hormones. Then, we review the empirical evidence that supports the notion of shared 3

4 neural circuits for the generation of behavior in oneself and its perception from others. A special emphasis is made on recent functional neuroimaging studies showing the involvement of shared neural circuits during the observation of pain in others and during the experience of pain in the self. Next, we discuss how perspective taking and the ability to differentiate the self from the other affect this sharing mechanism. In the last section, we examine how some interpersonal variables modulate empathic concern and personal distress. Finally, this chapter concludes with some cautionary considerations about the social neuroscience approach to intersubjective processes. The evolutionary origins of empathy Natural selection has fine-tuned the mechanisms that serve the specific demands of each species ecology, and social behaviors are best understood in the context of evolution. MacLean (1985) has proposed that empathy emerged in relation with the evolution of mammals (180 million year ago). In the evolutionary transition from reptiles to mammals, three key developments were (1) nursing, in conjunction with maternal care; (2) audiovocal communication for maintaining maternal-offspring contact; and (3) play. The development of this behavioral triad may have depended on the evolution of the thalamocingulate division of the limbic system, a derivative from early mammals. The thalamocingulate division (which has no distinctive counterpart in the reptilian brain) is, in turn, geared in with the prefrontal neocortex that, in human beings, may be inferred to play a key role in familial acculturation. When mammals developed parenting behavior the stage was set for increased exposure and responsiveness to emotional signals of others including signals of pain, separation and distress. Indeed, parenting involves the protection and transfer of energy, information, and social relations to offspring. African hominoids, including chimpanzees (Pan troglodytes), gorillas (Gorilla gorilla), and humans (Homo sapiens), share a number of parenting mechanisms with other placental mammals, including internal gestation, lactation, and attachment mechanisms involving neuropeptides such as oxytocin (Geary & Flinn, 2001). 4

5 The phylogenic origin of behaviors associated with social engagement has been linked to the evolution of the autonomic nervous system and how it relates to emotion. According to Porges (2001, 2007), social approach or withdrawal stem from the implicit computation of feelings of safety, discomfort, or potential danger. Basic to survival is the capacity to react to challenges or stressors and maintain visceral homeostatic states necessary for vital processes such as oxygenation of tissues and supplying nutrients to the body. He proposed that the evolution of the autonomic nervous system (sympathetic and parasympathetic systems) provides a means to understand the adaptative significance of mammalian affective processes including empathy and the establishment of lasting social bonds. Thus empathy can be viewed in terms of adaptive neuroendocrine and autonomic processes, including changes in neuromodulatory systems that regulate bodily states, emotions and reactivity (Carter, Harris & Porges, 2008). These basic evaluative systems are associated with motor responses that aid the adaptive responding of the organism. At this primitive level, appetitive and aversive behavioral responses are modulated by specific neural circuits in the brain that share common neuroarchitectures among mammals (Parr & Waller, 2007). These brain systems are genetically hard-wired to enable animals to respond unconditionally to threatening, or appetitive, stimuli using specific response patterns that are most adaptive to the particular species and environmental condition. The limbic system, which includes the hypothalamus, the parahippocampal cortex, the amygdala, and several interconnected areas (septum, basal ganglia, nucleus accumbens, insula, retrospenial cingulate cortex and prefrontal cortex) is primarily responsible for emotion processing. What unites these regions are their roles in motivation and emotion, mediated by connections with the autonomic system. The limbic system also projects to the cingulate and orbitofrontal cortices, which are involved with the regulation of emotion. 5

6 There is evidence for a lateralization of emotion processing in humans and primates which has been marshaled under two distinct theories. One theory states that the right hemisphere is primarily responsible for emotional processing (Cacioppo & Gardner, 1999), while another one suggests that the right hemisphere regulates negative emotion and the left hemisphere regulates positive emotion (Davidson, 1992). This asymmetry is anatomically based on an asymmetrical representation of homeostatic activity that originates from asymmetries in the peripheral autonomic nervous system, and fits well with the homeostatic model of emotional awareness, which posits that emotions are organized according to the fundamental principle of autonomic opponency for the management of physical and mental energy (Craig, 2005). Supporting evidence for the lateralization of emotion comes from neuroimaging studies and neuropsychological observations with brain damaged patients, but also studies in nonhuman primates. In one study, tympanic membrane temperature (Tty) was used to assess asymmetries in the perception of emotional stimuli in chimpanzees (Parr & Hopkins, 2000). The tympanic membrane is an indirect, but reliable, site from which to measure brain temperature, and is strongly influenced by autonomic and behavioral activity. In that study, chimpanzees were shown positive, neutral, and negative emotional videos depicting scenes of play, scenery, and severe aggression, respectively. During the negative emotion condition, right Tty was significantly higher than the baseline temperature. This effect was relatively stable, long lasting, and consistent across individuals. Temperatures did not change significantly from baseline in the neutral or positive emotion condition, although a significant number of measurements showed increased left Tty during the neutral emotion condition. These data suggest that viewing emotional stimuli results in asymmetrical changes in brain temperature, in particular increased right Tty during the negative emotion condition, evidence of emotional arousal in chimpanzees, and providing support for right hemispheric asymmetry in our closest living ancestor. 6

7 At the behavioral level it is evident from the descriptions of comparative psychologists and ethologists that behaviors homologous to empathy can be observed in other mammalian species. Notably, a variety of reports on ape empathic reactions suggest that, apart from emotional connectedness, apes have an explicit appreciation of the other s situation (de Waal 1996). A good example is consolation, defined as reassurance behavior by an uninvolved bystander towards one of the combatants in a previous aggressive incident (de Waal & van Roosmalen 1979). De Waal (1996) has convincingly argued that empathy is not an all-or-nothing phenomenon, and many forms of empathy exist between the extremes of mere agitation at the distress of another and full understanding of their predicament. Many other comparative psychologists however view empathy as a kind of induction process by which emotions, both positive and negative, are shared, and by which the probabilities of similar behavior are increased in the participants. In the view developed in this paper, this is a necessary but not a sufficient mechanism to account for the full-blown ability of human empathy. However it does provide the basic primitive, yet crucial mechanism on which empathy develops. Indeed, some aspects of empathy are present in other species, such as motor mimicry and emotion contagion (see de Waal & Thompson, 2005). For instance, Parr (2001) conducted an experiment in which peripheral skin temperature (indicating greater negative arousal) was measured in chimpanzees while they were exposed to emotionally negative video scenes. Results demonstrate that skin temperature decreased when subjects viewed videos of conspecifics injected with needles or videos of needles themselves, but not videos of a conspecific chasing the veterinarian. Thus, when chimpanzee are exposed to meaningful emotional stimuli, they are subject to physiological changes similar to those observed during fear in humans, which is similar to the dispositional effects of emotional contagion (Hatfield, 2008). In humans, the construct of empathy accounts for a more complex psychological state than the one associated with the automatic sharing of emotions. Like in other species, emotions and feelings may be shared between individuals, but humans are also able to 7

8 intentionally feel for and act on behalf of other people whose experiences may differ greatly from their own (Batson et al., 1991; Decety & Hodges, 2006). This phenomenon, called empathic concern or sympathy, is often associated with prosocial behaviors such as helping kin, and has been considered as a chief enabling process for altruism. According to Wilson (1988), empathic helping behavior has evolved because of its contribution to genetic fitness (kin selection). In humans and other mammals, an impulse to care for offspring is almost certainly genetically hard-wired. It is far less clear that an impulse to care for siblings, more remote kin, and similar non-kin is genetically hard-wired (Batson, 2006). The emergence of altruism, of empathizing with and caring for those who are not kin is thus not easily explained within the framework of neo- Darwinian theories of natural selection. Social learning explanations of kinship patterns in human helping behavior are thus highly plausible. However, one of the most striking aspects of human empathy is that it can be felt for virtually any target even targets of a different species. In addition, as emphasized by Harris (2000), humans, unlike other primates, can put their emotions into words, allowing them not only to express emotion but also to report on current as well as past emotions. These reports provide an opportunity to share, explain, and regulate emotional experience with others that is not found in other species. Conversation helps to develop empathy, for it is often here that one learns of shared experiences and feelings. Moreover, this self-reflexive capability (which includes emotion regulation) may be an important difference between humans and other animals (Povinelli, 2001). Interestingly two key regions, the anterior insula and anterior cingulate cortex (ACC), involved in affective processing in general and empathy for pain in particular have singularly evolved in apes and humans. Cytoarchitectonic work by Allman and colleagues (2005) indicates that a population of large spindle neurons is uniquely found in the anterior insula and anterior cingulate of humanoid primates. Most notably, they reported a trenchant phylogenetic correlation, in that spindle cells are most numerous 8

9 in aged humans, but progressively less numerous in children, gorillas, bonobos and chimpanzees, and nonexistent in macaque monkeys. Craig (2007) recently suggested that these spindle neurons interconnect the most advanced portions of limbic sensory (anterior insula) and limbic motor (ACC) cortices, both ipsilaterally and contralaterally, which, in sharp contrast to the tightly interconnected and contiguous sensorimotor cortices, are situated physically far apart as a consequence of the pattern of evolutionary development of limbic cortices. Thus, the spindle neurons could enable fast, complex and highly integrated emotional behaviors. In support of this view, convergent functional imaging findings reveal that the anterior insula and the anterior cingulate cortices are conjointly activated during all human emotions. This, according to Craig (2002), indicates that the limbic sensory representation of subjective feelings (in the anterior insula) and the limbic motor representation of volitional agency (in the anterior cingulate) together form the fundamental neuroanatomical basis for all human emotions, consistent with the definition of an emotion in humans as both a feeling and a motivation with concomitant autonomic sequelae (Rolls, 1999). Overall, this evolutionary conceptual view is compatible with the hypothesis that advanced levels of social cognition may have arisen as an emergent property of powerful executive functioning assisted by the representational properties of language (Barrett, Henzi & Dunbar, 2003). However, these higher levels operate on previous levels of organization, and should not be seen as independent of, or conflicting with one another. Evolution has constructed layers of increasing complexity, from nonrepresentational (e.g., emotion contagion) to representational and metarepresentational mechanisms (e.g., sympathy), which need to be taken into account for a full understanding of human empathy. Neuroendocrinological aspects While we prefer to believe that our behavior is largely dependent upon the processes in the central nervous system, other factors such as hormonal status or autonomic 9

10 nervous system activity have to be taken into account if we want to achieve a more accurate understanding of human social interaction. Hormones affect the body and the nervous system in various ways and shape the way in which our bodies and minds are affected by social interactions. For example, an as will be exemplified later on in this chapter, individual differences in stress coping mechanisms may determine how we respond to another s distress. One potential mechanism for these individual differences is the release of stress hormones such as cortisol, which has been shown to affect approach and withdrawal responses to threatening social stimuli (Roelofs et al., 2005). In a similar vein, the neuropeptides oxytocin (OT) and vasopressin (VP) have been repeatedly associated with individual differences in social cognition and behaviors. In non-human mammals, oxytocin is a key mediator of complex emotional and social behaviors, including attachment, social recognition, and aggression. For instance, it has been shown in prairie voles (Microtus ochrogaster) that OT facilitates positive social behaviors and promotes social attachment, such as parental and pair bonding behavior (Carter et al., 1992). Among other mechanisms, OT seems to exert these effects by enabling a more adaptive response of the organism to stressful events. OT achieves this by modulating autonomic arousal, in particular by reducing activity of the hypothalamicpituitary-adrenal (HPA) axis. The HPA axis strongly affects how we react to stressors by the release of stress hormones such as cortisol in humans or corticosterone in rodents. Importantly, the effects of the HPA axis are rather slow and tonic and sometimes persist over extended periods of time acting via changes in gene expression both in the body and in the brain. This is in contrast to the second system involved in stress responses, the sympathetic-adrenomedullary system, which triggers fast mobilization of vital resources by the release of epinephrin and norepinephrin (Gunnar & Quevedo, 2007, for review). Interestingly, OT and VP receptors are found in a number of areas of the nervous system associated with regulation of HPA axis and autonomic nervous system activity. In addition, OT and VP receptors in brain structures associated with social behaviors and 10

11 emotional processing might be the central nervous system substrates for the facilitating effects these hormones have on social interactions. For example, OT receptors are found in the amygdala, the medial prefrontal cortex and the septum. Selective effects of OT on these regions have been demonstrated both in non-human mammals and in humans. The amygdala plays a central role in autonomic function and has been linked to fear and associated automatic responses to environmental threats (e.g., Le Doux, 2000). The anxiolytic and calming role of OT might be achieved by acting upon receptors in this subcortical structure. Indeed, microinjections of OT in the central nucleus of the amygdala inhibit aggressive maternal behavior in female rats (Consiglio et al., 2005), in line with the finding that OT excites inhibitory neurons in the central amygdala (Huber et al., 2005). In humans, neuroimaging demonstrates reduced amygdala responses to social and non-social stimuli after intranasal administration of OT (Kirsch et al., 2005). Individual OT levels also seem to be related to human trust and trustworthiness as shown by a higher level of trust in an economic exchange game requiring participants to accept social risks (Kosfeld et al., 2005) or as expressed by higher OT levels with higher intentional trust in a similar experimental context (Zak et al., 2005). The effects of OT in promoting social attachment led to speculations that it also plays a role for empathic concern, sympathy and prosocial behavior. OT and VP are also potential candidates to explain the etiology of autism spectrum disorders (Carter, 2007, for review). ASD are characterized by deficits in social behavior and communication, with one distinctive feature being deficits in theory of mind (i.e., the ability to reason about intentions and beliefs of others) and empathy. In line with this idea, preliminary evidence from a small sample of ASD participants showed better comprehension of affective speech and the assignment of emotional significance to speech intonation with OT administration (Hollander et al., 2007). In the normal behavioral spectrum, OT enhanced the ability to infer other s mental states by interpreting subtle social cues expressed in the eye region. Notably, this effect was only observed for more difficult emotional-social expressions (Domes, Heinrichs, Glaser et al., 2007). In a companion 11

12 study, the same group demonstrated that activation in the right (but not the left) amygdala is reduced when presenting facial displays of emotions irrespective of their valence (Domes, Heinrichs, Michel et al., 2007). A potential mechanism of OT in enabling empathy and reading the intentions of others might be a general reduction of arousal and anxiety usually triggered by social and non-social stressors. This should allow for a more controlled processing of social cues and a more adaptive response to the emotional state of others. This hypothesis would be in line with evidence from social psychology showing that the need to belong to others - which can be interpreted as a measure of social attachment or the motivation for it - correlates with higher empathic accuracy for both positive and negative affective information (Pickett et al., 2004). Alternatively, and as speculated by Domes and colleagues, OT-induced increases in empathy and mind-reading might reduce social ambiguity and by this means encourage social approach and trusting behavior. While research on OT and human social behavior is still in its infancy, future investigations will have to show which neural and neuronal-hormonal mechanisms are at play when it exerts its effects. The available evidence clearly indicates a promising role of OT for promoting intersubjective understanding and prosocial behavior. Shared neural circuits between self and other It has long been suggested that empathy involves resonating with another person s unconscious affect. For instance, Ax in 1964 proposed that empathy can be thought of as an autonomic nervous system state, which tends to simulate that of another person. In the same vein, Basch (1983) speculated that, because their respective autonomic nervous systems are genetically programmed to respond in a similar fashion, a given affective expression by a member of a particular species can trigger similar responses in other members of that species. This idea fits neatly with the notion of embodiment, which refers both to actual bodily states and to simulations of experience in the brain s modality-specific systems for perception, action, and the introspective systems that 12

13 underlie conscious experiences of emotion, motivation and cognitive operations (Niedenthal et al., 2005). Furthermore, the view that unconscious automatic mimicry of a target generates in the observer the autonomic response associated with that bodily state and facial expression subsequently received empirical support from a variety of behavioral and physiological studies marshaled under the perception-action coupling account of empathy (Preston & de Waal, 2002). The core assumption of the perceptionaction model of empathy is that perceiving a target s state automatically activates the corresponding representations of that state in the observer which in turn activates somatic and autonomic responses. The discovery of sensory-motor neurons (called mirror neurons) in the premotor and posterior parietal cortex discharging both during the production of a given action and during the perception of the same action performed by another individual provides the physiological mechanism for this direct link between perception and action (Rizzolatti & Craighero, 2004). In line with the perception-action matching mechanism, a number of behavioral and electromygraphic studies demonstrated that viewing facial expressions triggers similar expressions on one s own face, even in the absence of conscious recognition of the stimulus (Hatfield, Cacioppo & Rapson, 1994). For example, while watching someone smile, the observer activates the same facial muscles involved in producing a smile at a subthreshold level and this would create the corresponding feeling of happiness in the observer. There is evidence for such a mechanism in the recognition of emotion from facial expression. For instance, viewing facial expressions triggers expressions on one s own face, even in the absence of conscious recognition of the stimulus (Dimberg, Thunberg, & Elmehed, 2000). Making a facial expression generates changes in the autonomic nervous system and is associated with feeling the corresponding emotion. In a series of experiments, Levenson, Ekman and Friesen (1990) instructed participants to produce facial configurations for anger, disgust, fear, happiness, sadness, and surprise while heart rate, skin conductance, finger temperature, and somatic activity were monitored. They found that such a voluntary facial activity produced significant levels of 13

14 subjective experience of the associated emotions as well as specific and reliable autonomic measures. One functional magnetic resonance imaging (fmri) experiment extended these results by showing that when participants are required to observe or to imitate facial expressions of various emotions, increased neurodynamic activity is detected in the brain regions implicated in the facial expressions of these emotions, including the superior temporal sulcus, the anterior insula, and the amygdala, as well as specific areas of the premotor cortex (Carr et al., 2003). Accumulating evidence suggests that a mirroring or resonance mechanism is also at play both when one experiences sensory and affective feelings in the self and perceives them in others. Even at the level of the somatosensory cortex, seeing another s neck or face being touched elicits appropriately organized somatotopic activations in the mind of the observer (Blakemore et al., 2005). Robust support for the involvement of shared neural circuits in the perception of affective states comes from recent neuroimaging and transcranial magnetic stimulation (TMS). For instance, the first-hand experience of disgust and the sight of disgusted facial expressions in others both activate the anterior insula (Wicker et al., 2003). Similarly, the observation of hand and face actions performed with an emotion engages regions that are also involved in the perception and experience of emotion and/or communication (Grosbras & Paus, 2006). Perceiving others in pain Pain is conceived as a subjective experience triggered by the activation of a mental/neural representation of actual or potential tissue damage. This representation involves somatic sensory features, as well as affective-motivational reactions associated with the promotion of protective or recuperative visceromotor and behavioral responses. It is the affective experience of pain that signals an aversive state and motivates behavior to terminate, reduce, or escape exposure to the source of noxious stimulation (Price, 2000). The expression of pain also provides a crucial signal, which can motivate soothing and caring behaviors in others. It is therefore a valuable and 14

15 ecologically valid means to investigate the mechanisms underlying the experience of empathy. A growing body of research demonstrates shared physiological mechanisms for the firsthand experience of pain and the perception of pain in others (Figure 1). One first specific indication for such a shared neural mechanism comes from a single-cell recordings study in neurological patients by Hutchison, Davis and Lozano (1999). These authors recorded with microelectrodes from the dorsal ACC as several types of painful stimuli were delivered to the patients hands, and found stimulus-specific pain responses in Brodmann area 24. Some of these cells displayed mirror-like properties, as they responded to the pinprick whether it was administered to the patient s own hand or to that of the experimenter. Figure 1. Neurophysiological research on pain points out a distinction between the sensory-discriminative aspect of pain processing and the affective-subjective one. These two aspects are underpinned by discrete 15

16 yet interacting neural networks. A growing number of neuroimaging studies recently demonstrated that the observation of pain in others recruits brain areas chiefly involved in the affective and motivational processing of direct pain perception (areas colored in orange). The first functional MRI experiment that investigated neural responses to both the first hand experience of pain and perception of pain in others was conducted by Morrison and collaborators (2004). Study participants were scanned during a condition of feeling a moderately painful pinprick stimulus to the fingertips and another condition in which they watched another person s hand undergo similar stimulation. Both conditions resulted in common hemodynamic activity in a pain-related area in the right dorsal ACC. In contrast, the primary somatosensory cortex showed significant activations in response to noxious tactile, but not visual, stimuli. Another fmri study demonstrated that the dorsal ACC, the anterior insula, cerebellum, and brainstem were activated when healthy participants experienced a painful stimulus, as well as when they observed a signal indicating that another person was receiving a similar stimulus. However, only the actual experience of pain resulted in activation in the somatosensory cortex and a more ventral region of the ACC (Singer et al., 2004). The different response patterns in the two areas are consistent with the ACC s role in coding the motivational-affective dimension of pain, which is associated with the preparation of behavioral responses to aversive events. These findings are supported by an fmri study in which participants were shown still photographs depicting right hands and feet in painful or neutral everyday-life situations, and asked to imagine the level of pain that these situations would produce (Jackson, Meltzoff & Decety, 2005). Significant activation in regions involved in the affective aspects of pain processing, notably the dorsal ACC, the thalamus and the anterior insula was detected, but no activity in the somatosensory cortex. Moreover, the level of activity within the dorsal ACC was strongly correlated with participants mean ratings of pain attributed to the different situations. 16

17 Crying is a universal vocalization in human infants as well as in the infants of other mammals (Newman, 2007). In all studied mammals, young infants emit a speciesspecific cry when in distress, and mothers generally respond with care-taking behavior (e.g., Bell and Ainsworth 1972). One functional MRI study measured brain activity in healthy, breastfeeding first-time mothers with young infants while they listened to infant cries, white noise control sounds, and a rest condition (Lorberbaum et al., 2002). Signal increase was detected in ACC, anterior insula, the medial thalamus, medial prefrontal and right orbitofrontal cortices. Several other structures thought important in rodent maternal behavior also displayed increased activity, including the midbrain, hypothalamus, dorsal and ventral striatum, and the vicinity of the lateral septal region. Facial expressions of pain constitute one important category of facial expression that is readily understood by observers. One study investigated the neural response to pain expressions by performing functional magnetic resonance imaging (fmri) as subjects viewed short video sequences showing faces expressing either moderate pain or, for comparison, no pain (Botvinick et al., 2005). In alternate blocks, the same subjects received both painful and non-painful thermal stimulation. Facial expressions of pain were found to engage cortical areas also engaged by the first-hand experience of pain, including the anterior cingulate cortex and anterior insula. Using fmri, Saarela and colleagues (2006) showed that not only the presence of pain but also the intensity of the observed pain is encoded in the observer s brain as occurs during the observer s own pain experience. When subjects observed pain from the faces of chronic pain patients, activations in bilateral anterior insula, left anterior cingulate cortex, and left inferior parietal lobe in the observer s brain correlated with their estimates of the intensity of observed pain. Furthermore, the strengths of activation in the left anterior insula and left inferior frontal gyrus during observation of intensified pain correlated with subjects self-rated empathy. 17

18 Overall, these fmri studies consistently detected activation of the anterior insula and dorsal ACC (two key regions that belong to the processing of the affective-motivational dimension of pain) during the perception of pain in others, and thus lend support to the idea that common neural substrates are involved in representing one s own and others affective states. However, most of these neuroimaging studies (except Moriguchi et al., 2007) did not report significant signal change in the somatosensory cortex/posterior insula (the region involved in the sensory discriminative dimension of pain). This result seems at odds with the perception-action coupling mechanism (mirror-neuron system) that underlies the automatic resonance between self and others. The somatosensory cortex/posterior dorsal insula contributes to the sensory discriminative dimension of pain as demonstrated by various neuroimaging investigations and lesion studies (e.g., Symonds et al., 2006). However, two recent studies indicate involvement of motor cortex during the perception of pain in others. These studies used transcranial magnetic stimulation (TMS) and found changes in the corticospinal motor representations of hand muscles in individuals observing needles penetrating hands or feet of a human model (Avenanti et al., 2005, 2006). Using electroencephalography (EEG), another study found modulation of somatosensory cortex activity contingent upon observation of others pain (Bufalari et al., 2007). Two possibilities can explain the discrepancy of the EEG and TMS with the fmri studies. One is that the TMS and EEG methods can sense subtle changes in the sensorimotor cortex that are below the significance threshold in fmri techniques. The other possibility is that attending to a specific body part elicits somatosensory activity in the corresponding brain region. This has been demonstrated in a positron emission tomography study in which participants were instructed to focus their attention either on the unpleasantness or on the location of the noxious stimuli delivered on the participants hands (Kulkarni et al., 2005), with the latter condition resulting in increased regional cerebral blood flow in the contralateral primary somatosensory cortex. 18

19 To test if the perception of pain in others involves the somatosensory cortex, Cheng, Yang et al. (2008) measured neuromagnetic oscillatory activity from the primary somatosensory cortex in participants while they observed static pictures depicting body parts in painful and non-painful situations. The left median nerve was stimulated at the wrist, and the post-stimulus rebounds of the ~10-Hz somatosensory cortical oscillations were quantified. Compared to the baseline condition, the level of the ~10-Hz oscillations was suppressed during both of the observational situations, indicating activation of the somatosensory cortex. Importantly, watching painful compared to non-painful situations suppressed somatosensory oscillations to a significantly stronger degree. In addition, these suppressions negatively correlated with the perspective taking subscale of the interpersonal reactivity index. These results, consistent with the mirror-neuron system, demonstrate that the perception of pain in others modulates neural activity in somatosensory cortex and supports the idea that the perception of pain in others elicits subtle somatosensory activity that may be difficult to detect by fmri techniques. Most neuroimaging studies that have explored the overlap in brain response between the observation of behavior performed by others and the generation of the same behavior in self have relied on simple subtraction methods and generally highlight the commonalities between self and other processing, and ignore the differences. This is particularly true for the recent series of fmri studies that have reported shared neural circuits for the first-hand experience of pain and the perception of pain in others (see Jackson, Rainville and Decety, 2006). It is however possible, as argued by Zaki and collaborators, that common activity in ACC and AI may reflect the operation of distinct but overlapping networks of regions that support perception of self or other pain. To address this issue, they scanned participants while they received noxious thermal stimulation (self pain condition) or watched short videos of other people sustaining painful injuries (other pain condition). Analyses identified areas whose activity covaried with ACC and AI activity during self or other pain either across time (intra-individual connectivity) or across participants (inter-individual connectivity). Both connectivity 19

20 analyses identified clusters in the midbrain and periaqueductal gray with greater connectivity to the AI during self pain as opposed to other pain. The opposite pattern was found in the dorsal medial prefrontal cortex, which showed greater connectivity to the ACC and AI during other pain than during self pain using both types of analysis. Intra-individual connectivity analyses also revealed regions in the superior temporal sulcus, posterior cingulate, and precuneus that became more connected to ACC during other pain as compared to self pain. The results of this experiment document distinct neural networks associated with ACC and anterior insula in response to first-hand experience of pain and response to seeing other people in pain. These networks could not have been detected in prior work that examined overlap between self and other pain in terms of average activity, but not connectivity. Morrison and Downing s (2007) analyses of single subject data in generic space similarly suggest that distinct neural networks in anterior and medial cingulate cortex (MCC) are activated during the first- vs. third-hand experience of pain. This is in line with a quantitative meta-analysis of published studies on empathy for pain vs. pain in the self using Activation Likelihood Estimation. This analysis reveals distinct subclusters in both ACC/MCC and the insular cortices (Figures 2 and 3). While activation in MCC seems to be more left-lateralized, caudal and dorsal during empathy for pain, a rostro-caudal activation gradient is evident in the insular cortex. These distinct activation patterns suggest the involvement of only partially overlapping neural subpopulations and indicate the involvement of distinct cognitive and affective processes. Yet, it should also be kept in mind that the effective spatial resolution of fmri, the different experimental paradigms as well as the inherently complex mapping from cognitive to neural/hemodynamic processes make it difficult to achieve a definite conclusion about how much of the activation during empathy for pain can be attributed to shared neural and mental representations. 20

21 Figure 2: Results of a meta-analysis comparing empathy for pain with the 1 st pain experience of pain. Activation differences are projected onto a flattened representation of the left and right hemispheres. Note that empathy predominantly activates the anterior parts of the insula while pain sensations lead to more rostral activation especially in the contralateral (left) hemisphere. 21

22 Figure 3: Results of the same meta-analysis for the MCC/ACC. Note the more left-lateralized and dorsal activations for empathy for pain. Summing up, current neuroscientific evidence suggests that merely perceiving another individual in a painful situation yields responses in the neural network associated with the coding of the motivational-affective and the sensory dimensions of pain in oneself. It is worth noting that vicariously instigated activations in the pain matrix are not necessarily specific to the emotional experience of pain, but to other processes such as somatic monitoring, negative stimulus evaluation, and the selection of appropriate skeletomuscular movements of aversion. Thus, the shared neural representations in the affective-motivational part of the pain matrix might not be specific to the sensory qualities of pain, but instead be associated with more general survival mechanisms such as aversion and withdrawal. 22

23 Perspective-taking There is general consensus among theorists that the ability to adopt and entertain the psychological perspective of others has a number of important consequences in social interaction. Well-developed perspective-taking abilities allow us to overcome our usual egocentrism and tailor our behaviors to others expectations (Davis et al., 1996). Further, successful role-taking has been linked to moral reasoning and altruism (Batson et al., 1991). Adopting another person s perspective involves more than simply focusing our attention on the other. It involves imagining how that person is affected by his or her situation without confusion between the feelings experienced by the self versus feelings experienced by the other person (Decety, 2005). We see others as similar to ourselves on a variety of dimensions and consequently assume that they act as we act, know what we know, and feel what we feel. This default mode is based on a shared representations mechanism between self and other (Decety & Sommerville, 2003) driven by the automatic link between perception and action (Jackson & Decety, 2004). Thus, for successful social interaction, and empathic understanding in particular, an adjustment must operate on the shared representations that are automatically activated through the perception-action coupling mechanism. Whereas the projection of self-traits onto the other does not necessitate any significant storage of knowledge about the other, empathic understanding requires the inclusion of other characteristics within the self. An essential aspect of empathy is to recognize the other person as like the self, while maintaining a clear separation between self and other. Hence, mental flexibility and self-regulation are important components of empathy. One needs to calibrate one s own perspective that has been activated by the interaction with the other, or even by its mere imagination. Such calibration requires the prefrontal cortex executive resources in conjunction with the temporo-parietal junction, as demonstrated by neuroimaging experiments in healthy participants as well as lesion studies in neurological patients. 23

24 Several neuroimaging studies have consistently reported that the medial prefrontal cortex is specifically involved in tasks requiring the processing of information relevant to the self, such as traits and attitudes (e.g., Johnson et al., 2002). An fmri study investigated the neural regions mediating self-referential processing of emotional stimuli and explored how these regions are influenced by the emotional valence of the stimulus (Fossati et al., 2003). Results showed that the self-referential condition induced bilateral activation in the dorsomedial prefrontal cortex, whereas the other referential condition induced activation in lateral prefrontal areas. Activation in the right dorsomedial prefrontal cortex was specific to the self-referential condition regardless of the valence of the words. The authors of that study proposed that one specific role of the right dorsomedial prefrontal cortex is to represent states of an emotional episodic self and then to process emotional stimuli with a personally relevant perspective. This proposition is in line with studies showing activations within both the left and right dorsomedial prefrontal cortex during theory of mind tasks (Brunet-Gouet & Decety, 2006). Because emotions generally signal issues related to the self, subjects may use emotional cues during some theory of mind tasks to differentiate self from other; this self-related emotional processing is indicated by an increase of activity in the right dorsomedial prefrontal cortex. However, the medial prefrontal cortex is not only involved when one reflects on oneself, but also when individuals imagine the subjective perspective of others. Using mental imagery to take the perspective of another is a powerful way to place oneself in the situation or emotional state of that person. Mental imagery not only enables us to see the world of our conspecifics through their eyes or in their shoes, but may also result in similar sensations as the other person s (Decety & Grèzes, 2006). A series of neuroimaging studies with healthy volunteers investigated the neural underpinning of perspective taking in three different modes (i.e., motor, conceptual, and emotional) of self-other representations. In a first study, participants were scanned while they were asked to either imagine themselves performing a variety of everyday actions (e.g., 24

25 winding up a watch), or to imagine another individual performing similar actions (Ruby & Decety, 2001). Both conditions were associated with common activation in the supplementary motor area (SMA), premotor cortex, and the TPJ region. This neural network corresponds to the shared motor representations between the self and the other. Taking the perspective of the other to simulate his or her behavior resulted in selective activation of the frontopolar cortex and right inferior parietal lobule. In a second study, medical students were shown a series of affirmative health-related sentences (e.g., taking antibiotic drugs causes general fatigue) and were asked to judge their truthfulness either according to their own perspective (i.e., as experts in medical knowledge) or according to the perspective of a layperson (Ruby & Decety, 2003). The set of activated regions recruited when the participants put themselves in the shoes of a lay-person included the medial prefrontal cortex, the frontopolar cortex and the right TPJ. In a third study, the participants were presented with short written sentences that depicted real-life situations (e.g., someone opens the toilet door that you have forgotten to lock), which are likely to induce social emotions (e.g., shame, guilt, pride), or other situations that were emotionally neutral (Ruby & Decety, 2004). In one condition, they were asked to imagine how they would feel if they were experiencing these situations. And in another condition, they were asked to imagine how their mothers would feel in those situations. Reaction times were prolonged when participants imagined emotional-laden situations as compared with neutral ones, both from their own perspective and from the perspective of their mothers. Neurodynamic changes were detected in the frontopolar cortex, the ventromedial prefrontal cortex, the medial prefrontal cortex, and the right inferior parietal lobule when the participants adopted the perspective of their mothers, regardless of the affective content of the situations depicted. Cortical regions that are involved in emotional processing, including the amygdala and the temporal poles, were found activated in the conditions that integrated emotional-laden situations. 25

26 A recent functional MRI study used a factorial design to examine the neural correlates of self-reflection and perspective taking (Dargembeau et al., 2007). Participants were asked to judge the extent to which trait adjectives described their own personality (e.g., Are you sociable? ) or the personality of a close friend (e.g., Is Caroline sociable? ) and were also asked to put themselves in the place of their friend (i.e., to take a thirdperson perspective) and estimate how this person would judge the adjectives, with the target of the judgments again being either the self (e.g., According to Caroline, are you sociable? ) or the other person (e.g., According to Caroline, is she sociable? ). The results showed that self-referential processing (i.e., judgments targeting the self vs. the other person) was associated with activation in the ventral and dorsal anterior MPFC, whereas perspective taking (i.e., adopting the other person s perspective, rather than one s own, when making judgments) resulted in activation in the posterior dorsal MPFC; the interaction between the two dimensions yielded activation in the left dorsal MPFC. Findings from this study indicate that self-referential processing and perspective taking recruit distinct regions of the MPFC and suggest that the left dorsal MPFC may be involved in decoupling one s own from other people s perspectives on the self. Social psychology has for a long time been interested in the distinction between imagining the other and imagining oneself, and in particular in the emotional and motivational consequences of these two perspectives. A number of these studies show that focusing on another s feelings may evoke stronger empathic concern, while explicitly putting oneself into the shoes of the target (imagine self) induces both empathic concern and personal distress. In one such study, Batson, Early and Salvarini (1997) investigated the affective consequences of different perspective-taking instructions when participants listened to a story about Katie Banks, a young college student struggling with her life after the death of her parents. This study demonstrated that different instructions had distinct effects on how participants perceived the target s situation. Notably, participants imagining themselves to be in Katie s place showed stronger signs of discomfort and personal distress as participants focusing on the 26

27 target s responses and feelings (imagine other), or as participants instructed to take on an objective, detached point of view. Also, both perspective-taking instructions differed from the detached perspective by resulting in higher empathic concern. This observation may help explain why observing a need situation does not always yield to prosocial behavior: if perceiving another person in an emotionally or physically painful circumstance elicits personal distress, then the observer may tend not to fully attend to the other's experience and as a result lack sympathetic behaviors. Two functional MRI studies recently investigated the neural mechanisms subserving the effects of perspective-taking during the perception of pain in others. One study used pictures of hands and feet in painful scenarios and instructed the participants to imagine and rate the level of pain perceived from two different perspectives (self versus other) (Jackson et al., 2006). Results indicated that both the self and the other perspectives are associated with activation in the neural network involved in the processing of the affective aspect of pain, including the dorsal ACC and the anterior insula. However, the self-perspective yielded higher pain ratings and involved the pain matrix more extensively, including the secondary somatosensory cortex, the mid-insula, and the caudal part of the anterior cingulate cortex. Adopting the perspective of the other was associated with increased activation in the right temporoparietal junction and precuneus. In addition, distinct subregions were activated within the insular cortex for the two perspectives (anterior aspect for others and more posterior for self). These neuroimaging data highlight both the similarities and the distinctiveness of self and other as important aspects of human empathy. The experience of pain in oneself is associated with more caudal activations (within area 24), consistent with spino-thalamic nociceptive projections, whereas the perception of pain in others is represented in more rostral (and dorsal) regions (within area 32). A similar rostro-caudal organization is observed in the insula, which is consistent with its anatomical connectivity and electrophysiological properties (Jackson, Rainville & Decety, 2006). For instance, painful sensations are evoked in the posterior part of the insula 27

28 (and not in the anterior part) by direct electrical stimulation of the insular cortex in neurological patients (Ostrowsky et al., 2002). Altogether, these findings are in agreement with the fact that indirect pain representations (as elicited by the observation of pain in others) are qualitatively different from the actual experiences of pain. In a second neuroimaging study, the distinction between empathic concern and personal distress was investigated more specifically by using a number of behavioral measures and a set of ecological and extensively validated dynamic stimuli (Lamm, Batson & Decety, 2007). Participants watched a series of video-clips featuring patients undergoing painful medical treatment. They were asked to either put themselves explicitly in the shoes of the patient (imagine self), or to focus on their feelings and affective expressions (imagine other). The behavioral data confirmed that explicitly projecting oneself into an aversive situation leads to higher personal distress while focusing on the emotional and behavioral reactions of another s plight is accompanied by higher empathic concern and lower personal distress. The neuroimaging data are consistent with this finding and provide some insights into the neural correlates of these distinct behavioral responses. The self-perspective evoked stronger hemodynamic responses in brain regions involved in coding the motivational-affective dimensions of pain, including bilateral insular cortices, anterior MCC, the amygdala, and various structures involved in action control (Figure 4). The amygdala plays a critical role in fearrelated behaviors, such as the evaluation of actual or potential threats. Imagining oneself to be in a painful and potentially dangerous situation thus might therefore have triggered a stronger fearful and/or aversive response than imagining someone else to be in the same situation. This pattern of results fits well with the pioneering research of Stotland (1969) on the effects of perspective taking on empathy and distress. Participants observed an individual experiencing a painful diathermy using either an imagine self or an imagine 28

29 other instruction. Stotland found higher vasoconstriction for the other-perspective, and more palmar sweat and higher tension and nervousness in the self-perspective. This finding was interpreted as being more in resonance with the feelings of the target when focusing on his affective expressions and motor responses (imagine other), while the first person perspective led to more self-oriented responding that was less closely matched to the actual feelings of the target. Figure 4. Neural and behavioral consequences of two different perspective-taking instructions (adapted from Lamm et al., 2007). The flat-map representation of the left hemisphere shows higher activations during the self-perspective in limbic/paralimbic (medial and anterior cingulate cortex MCC and ACC, insula INS) and cortical brain structures (temporo-parietal junction TPJ, inferior frontal gyrus IFG, postcentral gyrus PCG). The overlay of functional activation on the flattened cortical surface was created using Caret ( Van Essen et al., 2001). Corresponding with Jackson and colleagues (2006), the insular activation found by Lamm & coworkers was also located in a more posterior, mid-dorsal sub-section of this area. The mid-dorsal part of the insula plays a role in coding the sensory-motor aspects of painful stimulation, and it has strong connections with the basal ganglia where activity was also higher during the self-perspective (see also Figure 2). Taken together, it appears that the insular activity during the self-perspective reflects simulation of sensory aspects of the painful experience. Such a simulation might serve to mobilize motor areas for the preparation of defensive or withdrawal behaviors, as well as instigate the interoceptive monitoring associated with autonomic changes evoked by this simulation process (Critchley et al., 2005). Finally, the higher activation in premotor 29

30 structures might connect with a stronger mobilization of motor representations by the more stressful and discomforting first-person perspective. Further support for this interpretation is provided by a positron emission tomography study investigating the relationship between situational empathic accuracy and brain activity which also found higher activation in medial premotor structures, partially extending into MCC, when participants witnessed the distress of others (Shamay-Tsoory et al., 2005). This study also pointed to the importance of prefrontal areas in the understanding of distress. Altogether, the available empirical findings reveal important differences between the neural systems involved in first- and third-person perspective-taking and contradicts the hypothesis/notion that the self and other completely merge in empathy. The specific activation differences in both the affective and sensorimotor aspects of the pain matrix, along with the higher pain and distress ratings, reflect the self-perspective s need for more direct and personal involvement. One key region that might facilitate self vs. other distinctions is the right TPJ. The TPJ is activated in most neuroimaging studies on empathy (Decety & Lamm, 2007), and seems to play a decisive role in self-awareness and the sense of agency. Agency (i.e., the awareness of oneself as an agent who is the initiator of actions, desires, thoughts and feelings) is essential for a successful navigation of shared representations between self and other. Recently, Decety and Lamm (2007) conducted a quantitative meta-analysis of 70 functional neuroimaging studies on agency, empathy, theory of mind, as well as on reorienting of attention. The results demonstrate a substantial overlap in brain activation between low-level processing such as reorienting of attention or the sense of agency and higher-level social-cognitive abilities such as empathy or theory of mind (see Figure 5). These results provide strong empirical support for a domain-general mechanism implemented in the right TPJ, and show that this area is also engaged in lower-level (bottom-up) computational processes associated with the sense of agency and reorienting attention to salient stimuli. 30

31 Figure 5: Activation overlap in right TPJ for empathy for pain and reorienting of attention/change detection (derived from Decety & Lamm, 2007). Thus, self-awareness and a sense of agency both play pivotal roles in empathy and significantly contribute to social interaction. These important aspects are likely to be involved in distinguishing emotional contagion, which relies heavily on the automatic link between perceiving the emotions of another and one s own experience of the same emotion, from empathic responses which call for a more detached and aware relation. The neural responses identified in these studies as non-overlapping between self and other may take advantage of available processing capacities to plan appropriate future actions concerning the other. Furthermore, awareness of one s own feelings and the ability to (consciously and automatically) regulate one s own emotions may allow us to disconnect empathic responses to others from our own personal distress, with only the former leading to prosocial behavior. Modulation of empathic responding 31