Experimental Brain Research

, Volume 204, Issue 3, pp 305–314

The skin as a social organ


    • Institute for Neuroscience and PhysiologyGöteborg University
    • Department of Clinical Neurophysiology, Sahlgrenska University HospitalGöteborg University
  • Line S. Löken
    • Institute for Neuroscience and PhysiologyGöteborg University
    • Department of Clinical Neurophysiology, Sahlgrenska University HospitalGöteborg University
  • Håkan Olausson
    • Institute for Neuroscience and PhysiologyGöteborg University
    • Department of Clinical Neurophysiology, Sahlgrenska University HospitalGöteborg University

DOI: 10.1007/s00221-009-2007-y

Cite this article as:
Morrison, I., Löken, L.S. & Olausson, H. Exp Brain Res (2010) 204: 305. doi:10.1007/s00221-009-2007-y


In general, social neuroscience research tends to focus on visual and auditory channels as routes for social information. However, because the skin is the site of events and processes crucial to the way we think about, feel about, and interact with one another, touch can mediate social perceptions in various ways. This review situates cutaneous perception within a social neuroscience framework by discussing evidence for considering touch (and to some extent pain) as a channel for social information. Social information conveys features of individuals or their interactions that have potential bearing on future interactions, and attendant mental and emotional states. Here, we discuss evidence for an affective dimension of touch and explore its wider implications for the exchange of social information. We consider three important roles for this affective dimension of the cutaneous senses in the transmission and processing of social information: first, through affiliative behavior and communication; second, via affective processing in skin–brain pathways; and third, as a basis for intersubjective representation.


Social neuroscienceCT afferentsPleasant touchEmpathy


“…who shall measure the subtlety of those touches which convey the quality of soul as well as body…?” —George Eliot, Middlemarch, chap 39

The sense of touch helps us to discriminate the location of a stimulus on the skin surface, to explore objects haptically, and to identify and manipulate objects. It also contributes to an integrated sense of our own body (see Serino and Haggard 2009). However, the research emphasis on these wide-ranging functions of touch leaves out a very essential fact: touch can also be pleasant. This aspect of tactile sensation is at the heart of the social domain, allowing positive hedonic experience ranging from the reassurance of a pat on the back to the rills of a sensual caress.

It is possible that functional divisions in the neural organization of touch may resemble that of pain, possessing two major, dissociable dimensions (e.g., Rainville et al. 1997; Hofbauer et al. 2001; Kulkarni et al. 2005; Auvray et al. 2008). The sensory-discriminative dimension supports spatial localization and intensity encoding of a stimulus, and the motivational-affective dimension is involved in coding its valence (e.g., pleasantness/unpleasantness) and motivational relevance. So far, only a handful of studies have explored the possibility of a similar functional dissociation in the domain of non-painful cutaneous sensation (McGlone et al. 2007; Lovero et al. 2009). Though discriminative aspects of touch have been well-studied, the affective aspects have only recently been conceptualized and investigated in the neuroscientific literature (Essick et al. 1999; Francis et al. 1999; Olausson et al. 2002; McGlone et al. 2007; Olausson et al. 2008a, b; Gallace and Spence 2008; Lovero et al. 2009; Löken et al. 2009; Guest et al. 2009).

Hedonically positive touch in human social interactions is ubiquitous despite cultural differences in its regulation, with roles ranging from the casual to the sexual. Sexual and parent–infant interactions are undeniably vital arenas of social touch. For example, the erotic dimension of human touch affects everyday interactions even among people who are not sexually involved, by introducing a culturally influenced “erotic barrier” which precludes certain types of casual touch (Vallbo et al. 2007; Heslin and Alper 1983). Touch also influences developmental pathways: maternal licking of rat pups can influence the behavior of the adult rat (Menard et al. 2004), and monkey infants deprived of tactile contact with a mother or mother surrogate become stressed and even ill-nourished (e.g., Harlow 1958). Here, however, we focus on primarily nonsexual, positively hedonic forms of interaction between adult humans, while acknowledging that these may have sources in and links with sexual and maternal touch behavior.

Touch in affiliative behavior and communication

The most salient nonsexual, positively hedonic forms of social touch can be tentatively divided into categories. “Simple” touch involves brief, intentional contact to a relatively restricted location on the body surface of the receiver during a social interaction; the person who pats the hand of the little old lady on the bus or gently touches the waiter’s elbow while making a request is engaging in “simple” touch. “Protracted” touch involves longer and often mutual skin-to-skin contact between individuals, and usually includes a component of pressure, for example embracing, holding hands, and cuddling. Finally, “dynamic” touch involves continuous movement over the skin from one point to another, and can often be repetitious, as in stroking, rubbing, and caressing. Tickling presents an additional special category, which is associated with playful behavior (especially with children) and is not likely to occur in the context of everyday interactions.

What is the role of “pleasantness”—the positive hedonic facet—in these categories of human social touch? First, pleasant touch may serve as a foundation for affiliative behavior. Second, it may provide a mechanism for the formation and maintenance of social bonds. Third, it is a nonverbal means for the communication of emotions. It is important to note that the above varieties of social touch are not always welcome or even pleasant; touch has an intricate relationship with culture, context, and gender (e.g., Dibiase and Gunnoe 2004), and, depending on these factors, pleasure easily tips into aversion and disgust.

Affiliative behavior is that which reflects or increases the disposition of one or more members in an interaction to seek close contact with another member. This is assumed to be accompanied by positive affective feelings, such as fondness, or with the alleviation of stress or anxiety. The motivational-affective dimension of social touch may provide mechanisms for the promotion of such positive consequences, ultimately pertaining to the desire to seek a touch (“wanting”) or to continue a tactile interaction (“liking”; see Berridge and Kringelbach 2008 for the “liking–wanting” distinction). Evidence that social touch encounters are accompanied by, and engender, positive affective consequences comes mainly from social psychological research investigating the relationship between interpersonal touch interactions and positive evaluations of people and surroundings. These studies usually involve “simple” touch, though without designating it as such (hand-to-hand, hand-to-arm, hand-to-shoulder, etc.).

Effects of touch in social interactions have been found to increase liking of a person or place, and to facilitate a footing of trust or compliance, often manifesting in increased prosocial behavior. For example, a half-second of hand-to-hand touch from a librarian fostered more favorable impressions of the library (Fisher et al. 1976), touching by a salesperson increased positive evaluations of the store (Hornik 1992), and touch can also boost the attractiveness ratings of the toucher (Burgoon et al. 1992). Recipients of such “simple” touches are also more likely to be more compliant or unselfish: returning money left in a public phone (Kleinke 1977), spending more money in a shop (Hornik 1992), tipping more in a restaurant (Crusco and Wetzel 1984), or giving away a cigarette (Joule and Guéguen 2007). The greater degree of compliance, or even generosity, with resources in these studies implies that such rudimentary social touches can potentially provide a platform for trust and cooperation in future exchanges.

“Protracted” touch, “dynamic” touch, and tickling can also have positive concomitants and outcomes in affiliative behavior. Holding a loved one’s hand can reduce the anxiety posed by an impending threat (Coan et al. 2006). Skin-to-skin contact has been shown to have clinical benefits for premature infants (Field 2001), as well as an analgesic effect in human babies undergoing minor tissue-damaging procedures (Gray et al. 2000). Stroking an infant can not only give rise to positive emotions in the baby, but can also modulate negative ones, compared to other forms of touch (Peláez-Nogueras et al. 1997). Tickling is a playful social act and is also seen in other mammals (Panksepp and Burgdorf 2003; Blakemore et al. 1998). It often gives rise to pilomotor responses like goosebumps, the elicitation of which is associated with the insula (Warren 2002). Interestingly, individuals with Asperger’s syndrome (associated both with social impairments and tactile hypersensitivity) rated the intensity and “tickliness” of a piece of foam on the palm as higher when the stimulus was applied by an experimenter than when the subjects touched their own palm with it (Blakemore et al. 2005).

A major context for social touch among primates is grooming. For most primates, grooming one another (allogrooming) is not just about hygiene. Although it does serve this function, the amount of time many primate species devote to this activity is disproportionate to what is required for keeping fur and skin clean. In fact, some species (such as gelada baboons) may spend up to 17% of their time grooming despite requiring perhaps only 1% for strictly hygienic purposes (Lehmann et al. 2007; Dunbar 2008). Instead, allogrooming has taken on a new significance as a form of bonding and reinforcing alliances (Dunbar 1996, 2008).

In contrast to the work on grooming in nonhuman primates, research on human allogrooming is rare. It is possible that grooming is a comparatively more restricted behavior in humans (for example, in intimate or parent–offspring interactions, or in ritual or professional transactions), or that its social role has been largely replaced by language (Dunbar 1996). Nelson and Geher (2007) have defined human grooming broadly as including “any behavior in which an individual removes or mimics removal of something from the skin or body.” In nonhuman primates, such activity involves parting the fur. Typically, allogrooming movements involve a coordinated sweeping through the fur with one hand and plucking at debris with the other (Sparks 1967; Dunbar 2008). Humans, of course, have much less abundant hair than our primate cousins, but our allogrooming practices may have adapted to our nakedness. A homologue to the “sweep” may manifest in the stroking gestures over the skin surface common in the “dynamic” category of interpersonal touch. Such dynamic touch is also pleasant for the toucher, with others’ forearm skin rated more pleasant to touch than one’s own, and forearm skin as more pleasant than glabrous (palm) skin, whether one’s own or another person’s (Guest et al. 2009).

In romantic partnerships, relationship satisfaction, previous experience of familial affection, and trust were positively correlated with self-reports of mutual grooming (Nelson and Geher 2007). The same study showed that individuals who scored higher on anxiety subscales of an attachment questionnaire also reported more frequent grooming behavior, suggesting that an anxious attachment style may be accompanied by behavior likely to lead to more secure bonds. It is also important to consider the stage of an affiliative and/or romantic relationship with respect to the role of intensive touching and grooming (Emmers and Dindia 1995). Indeed, different neural mechanisms may come into play during the initiation of an affiliative relationship and during its maintenance (Depue and Morrone-Strupinsky 2005).

Positive affect and hedonic feelings may be the glue which holds individuals, as necessarily physically separate beings, together in social bonds. Grooming, indeed, may reflect a generalized form of pair-bonding usually seen in reproductive or mother–offspring dyads across numerous taxa (Dunbar 2008). As such it may rely on similar neural and physiological mechanisms as the dyadic cases. For example, in addition to its role in maternal behaviors, the neurotransmitter oxytocin is also an important mediator of grooming (Drago et al. 1986), as demonstrated by the exaggerated grooming behavior that results when it is administered into the cerebrospinal fluid of knockout mice which do not express oxytocin (Amico et al. 2004). Endorphins may also play a central role in grooming. In both talapoin monkeys (Keverne et al. 1989) and rhesus macaques (Martel et al. 1995), opioid receptor blockade results in increased solicitations for grooming. Dopamine may also be important in affiliative behavior and bonding (Depue and Morrone-Strupinsky 2005).

Finally, touch can serve a communicative function. It can be used to convey thoughts and feelings, to regulate them in others, or both (Hertenstein et al. 2006a). Hertenstein et al. (2006a) identifies tactile communication as “systematic changes in another’s perceptions, thoughts, feelings, or behavior as a function of another’s touch in relation to the context in which it occurs.” Tactile communication need not always involve mutual touching, but the giver’s touch may affect participants’ emotions and consequent signals without answering touches. Hertenstein et al. (2006b) demonstrated that individuals from two countries, the US and Spain, were able to discriminate different categories of emotion on the basis of how someone touched them. The most accurately decoded emotions from touch represented both negative and positive categories. For the negative categories these corresponded to emotions that are panculturally recognized when expressed in the face: anger, fear, and disgust. The most accurately decoded positive emotions were ones that implied relationships: love, gratitude, and sympathy. Interestingly, the stroking used to express “sadness” was interpreted by the perceiver as sympathy or love, suggesting that the interpretation of ambiguous interpersonal touch stimuli may be biased towards positive emotions.

Pathways of pleasant touch

If social touch is so central to the way we interact with, perceive, think about, and evaluate others, it is possible that at least some varieties of pleasant touch are subserved by specific neural pathways. In particular, “dynamic” social touch has been the topic of recent investigations both on the level of peripheral neurophysiology and of the brain. This research suggests that such slow, gentle touch may be coded by the nervous system in affective and hedonic, not purely sensory, terms. A recently discovered type of slow-conducting, unmyelinated peripheral nerve fiber supports this hypothesis. The C tactile (CT) afferent has been shown to be sensitive to innocuous tactile stimulation (Vallbo et al. 1999; Wessberg et al. 2003; Nordin 1990). The slow conduction of CT afferents (around 1 m/s) renders them suboptimal for sensory discrimination. However, they show many characteristics consistent with a role in selectively encoding affective touch information.

First, they respond to light mechanical tactile stimulation of hairy, but not glabrous, skin in humans (Johansson and Vallbo 1979; Vallbo et al. 1999). They are abundant in the face and arms (Nordin 1990; Löken et al. 2009) and present in the legs (Löken et al. 2007). The receptive fields of CT fibers in humans consist of 1–9 small responsive spots distributed over an area up to 35 mm2 (Fig. 1a; Wessberg et al. 2003). Using a genetically encoded tracer, Liu et al. (2007) visualized the projections of a population of small-diameter sensory neurons in mice, the likely homologue of CT fibers. The terminal arborizations of these neurons in the epidermis were organized in a pattern of discontinuous patches covering about 50–60% of the area in the hairy skin, whereas they were lacking altogether in the glabrous skin of the pad (Fig. 1b). The pattern of coverage seen in the mouse reflects a large degree of branching in terminal arbors, which is evident in humans as well.
Fig. 1

a Color coded 2-dimensional density plots of receptive fields of 9 human tactile C afferents (reproduced from Wessberg et al. 2003). bLeft distribution of Mas-related G protein-couple receptor B4 peripheral terminals in mouse hairy (outside dotted line) and glabrous (within dotted line) skin, marking the likely mouse homologue of human tactile C afferents. Right higher-magnification view of red boxed area at left (reproduced from Liu et al. 2007). c Correlation between neural discharge rate and perception of pleasantness in response to soft brush stroking in humans (R2 = 0.70). Mean pleasantness ratings are plotted against the corresponding mean firing frequency for each brushing velocity and force (reproduced from Löken et al. 2009)

What is the relationship of CT fiber pathways to pleasantness? Because it is impossible to selectively stimulate CT fibers in healthy individuals, patients with sensory neuropathies provide opportunities to investigate CT fibers’ contribution to touch perception. Psychophysical tests reveal that two unique subjects lacking myelinated tactile (Aβ) afferents, but with spared unmyelinated afferents, detected soft brush stroking and weak monofilament indentation on the forearm skin where CT afferents are abundant (Olausson et al. 2002, 2008c; Cole et al. 2006). Importantly, they failed altogether to detect the same kind of stimuli applied to the glabrous skin of the hand. The sensation associated with a selective CT input (soft brush stroking) was weak, vague, and inconsistent, though nonetheless pleasant, as indicated by forced-choice ratings (Olausson et al. 2002).

A more direct measure of the relationship between CT afferent discharge and pleasantness is provided by microneurography research in which electrode recordings are obtained from the nerves of awake human volunteers as their tactile receptive fields are stimulated by a soft brush. The speed of stroking is crucial to this relationship. Löken et al. (2009) found that CT afferents are sensitive to the velocity of a stimulus moving across the skin, preferring speeds within the range of a gentle caress. We recorded from the afferent nerves of human volunteers during soft brush stroking at different velocities on the forearm. CT units showed a unique preference for speeds of 1–10 cm/s, with decreased discharge for slower or faster speeds. In contrast, this relationship was not seen in Aβ units, which increased firing with increasing stimulus velocity. The highest subjective pleasantness ratings correlated with the intermediate stroking speeds most effective in eliciting CT activation (Fig. 1c). This relationship with affective measures was not seen for Aβ fibers.

Patients with a rare mutation affecting the neural growth factor beta (NGFB) gene show a reduction in C-fiber density (Minde et al. 2004), of which CT fibers are a subtype. Löken et al. (2008) addressed the relationship between C-fiber function and pleasant touch perception in ten carriers of the NGFB mutation. These patients perceived gentle, slow stroking, optimal for eliciting CT afferent responses (1–10 cm/s), as less pleasant than did matched controls, as well as differing in their rating patterns across stimulation velocities. In contrast, patients’ sensitivity to measures of discriminative touch (mediated by intact Aβ afferents) was normal. This indicates that intact CT pathways are likely to be necessary for the normal evaluation of touch pleasantness on hairy skin.

On the cortical level, CT afferent pathways target brain structures associated with affective and homeostatic processing. Functional magnetic resonance imaging (fMRI) studies show increased insula activation for CT-fiber stimulation, and this area may be a target of CT-pathway projections (Olausson et al. 2002, 2008b). CT fibers, like other types of thinly myelinated and unmyelinated fibers, project via lamina I/II of the spinothalamic tract to posterior/basal ventral medial nucleus and posterior insular cortex (Craig 2002, 2008). In contrast to CT afferent pathways, large myelinated exteroceptors conveying discriminative touch project in the dorsal column to the ventral posterior lateral nucleus of the thalamus and from there to first and second somatosensory cortex (SI and SII). Despite having receptors in the skin, CT fibers therefore have more in common anatomically with interoceptive and visceral systems—relevant to a broad set of bodily feelings like pain, itch, and hunger—than to exteroceptive afferent systems processing tactile and nociceptive stimuli impinging from without.

Craig (2008) proposes that a progressively more complex re-representation from posterior to anterior insula comprises the integration of emotionally salient inputs from multiple sensory modalities in the middle insula and emotionally salient inputs from limbic cortical regions (anterior cingulate cortex, orbitofrontal cortex) in the anterior insula. The posterior insula in particular may support an early convergence of sensory and affective processing, as suggested by recent fMRI evidence for somatotopic organization of posterior insula responses to CT stimulation of arm and thigh (Björnsdotter et al. 2009). Such functional organization of insular cortex may reflect an integration of sensory with affective information important for the homeostasis of the body (Craig 2002, 2008). Such integration may also contribute critically to the formation of coherent representations of the body, the sensory environment, motivational conditions, and not least social conditions. The organism’s homeostatic state may in turn set the gain on processing on pathways identified as interoceptive or reward-related (Paulus 2007). Importantly, this view implies that processing in such pathways is modulable and that any given function (e.g., “pleasantness perception”) is dependent on the complex states of other circuits.

CT afferent coding may constitute a first stage for encoding the affective dimension of socially relevant touch, possibly on the basis of velocity information. But considering that Aβ afferents are capable of encoding velocity information orders of magnitude more rapidly than CT fibers can, why posit a special pathway for the encoding of affective velocity-dependent stimulation? A central assumption of the “social touch hypothesis” Olausson et al. (2008a) is that social touch is a distinct domain of touch. A special pathway is required if the affective information of a social touch needs to be distinguished from the “noise” of tactile stimulation not likely to carry such affective meaning. Thus, CT afferents may operate as selectors, activated in parallel with Aβ afferents. Rather than accurately discriminating different stimulation velocities, as Aβ fibers and their central projections can, CT fibers may “pick out” a range of velocities likely to have social-affective relevance, for the purposes of further hedonic processing in affect-related brain areas such as the insula (Fig. 2). Aβ fibers, on the other hand, encode sensory-discriminative aspects of touch with high rapidity and acuity, with pathways projecting ultimately to primary and secondary sensory cortical areas (Fig. 2).
Fig. 2

Schematic model of affective and sensory-discriminative pathways for dynamic touch in hairy skin. CT afferents show an inverted U curve, while Aβ afferent discharge increases linearly with velocity. CT afferents may thus act as selectors of a limited range of velocities likely to carry social or hedonic significance, while Aβ activation is suited to discriminate different stroking speeds. These signals follow dissociable pathways to the cortex but probably interact at several levels. Within cortex, reciprocal connections between posterior insula and secondary somatosensory cortex may allow mutual modulation of affective- and sensory-related processing

Importantly, these two systems are not separate, despite being at least partly dissociable; it is likely that sensory-discriminative and motivational-affective pathways for touch interact. This is especially likely considering the afferent and efferent connections between posterior insular regions and parietal somatosensory areas (Augustine 1996). For example, tactile responses in posterior insula may be influenced by discriminative processing in SI and SII. Likewise, affective coding of tactile stimuli in insular cortex may modulate somatosensory responses in SI (Olausson et al. 2008b) and SII.

Intersubjective representation

The pleasures and pains of cutaneous perception are subjective qualities that differ from objectively measurable perceptual qualities, like color or temperature. For example, though everyone can directly see the coffee mug sitting on the table, or measure the temperature of the coffee, only you can directly feel the pain of the hot ceramic when you grasp it. Another person’s nerve endings cannot send signals to your brain. Yet it is a common experience to “feel” another person’s pain when we see them cut their finger, or “feel” the thrill of a witnessed caress. How is this essentially subjective information conveyed through the channel of visual perception? Research exploring the neural basis of “empathy” points to various roles for tactile-related cortical processing during the observation of others’ touch or pain.

Keysers et al. (2004) used fMRI to investigate brain responses to the direct sensation of touch to the leg, as well as to video clips depicting touch to other people and to inanimate objects. Results showed that responses in SII overlapped between the touch experience conditions and the touch observation conditions. Contact, and not merely the motion of the tactile stimulus, was necessary for such visual responses to occur in SII. Further, where touch was present, visual activity in SII was not limited to human body part stimuli but also extended to inanimate objects. Primary somatosensory cortex (SI) did not show overlapping activity, suggesting that representations of others’ touch are supported by a more integrated level of tactile representation. Ebisch et al. (2008) found overlapping activations in SII for felt and seen touch, with SI responses discriminating whether the touch was intentional or accidental. These results are consistent with the proposition that the same neural pathways are recruited for both felt and seen touch—or even that others’ touch is understood via a “simulation” process or by virtue of “common coding” in similar or overlapping circuits (e.g., Gazzola and Keysers 2009).

In an fMRI case study Blakemore et al. (2005) compared responses to felt and seen touch in a synesthetic individual to those of a group of normal, non-synesthetic controls. The synesthetic individual experiences seen touch in terms of tactile stimulation on her own body, a variety of synesthesia which Blakemore has termed “mirrored touch synaesthesia” (see also Banissy et al. 2009). Seeing touch to others elicited activations in sensory as well as and motor and premotor areas. Such “mirrored touch” synesthetic individuals made more errors than non-synesthetes in a congruency task requiring them to report the location of an actual touch to the face or hand while watching another person being touched, implying that for them, seeing a touch in an incongruent location interferes with feeling touch (Banissy and Ward 2007). They also responded faster than controls in congruent trials and scored higher on an empathy questionnaire.

These studies have examined touch empathy in terms of specificity (body part vs. object) and/or location (e.g., face vs. neck, left vs. right). This is consistent with the activations seen in sensory cortical areas. Yet as reviewed in the sections above, a great degree of the emotional and social significance of a touch is borne by its affective component. What neural pathways are involved during the observation of affective touch? In an fMRI study McCabe et al. (2008) demonstrated that manual application of moisturizing lotion to the hairy (forearm) skin, but not to the glabrous skin of the palm, activated mid-orbitofrontal cortex, associated with affective and reward processing. Viewing of lotion applied to another person’s forearm engaged sensory pathways in SI and adjacent parietal Brodmann’s area 7. The lotion was variously labeled “rich” or “thin”, although the lotion’s consistency did not actually vary with the labels. Underscoring the importance of context for touch pleasantness, word labels modulated pleasantness ratings and processing in somatosensory and reward-related areas, including the insula for directly felt touch and subgenual cingulate for observed touch.

To investigate the role of putative CT afferent pathways in cortical responses to felt and seen pleasant touch, Morrison et al. (2008) used fMRI to compare the effects of stroking at a CT-optimal velocity (3 cm/s) versus a non-CT-optimal velocity (30 cm/s). For felt touch, the posterior insula showed velocity sensitivity similar to that of CT afferents during the stroking of hairy (forearm) skin at 3 cm/s compared to 30 cm/s. Crucially, an adjacent and partially overlapping region showed a similar response pattern during the observation of skin stroking in others. These activations fell in the insula’s posterior lobule, possibly in the caudodorsal granular insula area (Ig), which is associated with somatic and multimodal responses (Augustine 1996). The visual response here was specific to observed social touch, with stroking of a non-body object failing to modulate it. Such velocity-dependent insula responses during direct stimulation may reflect cortical processing of affective touch information from CT fibers. Visual processing may exploit this neural encoding to derive affective information about others’ touch interactions.

A growing body of research investigating brain responses to observed painful experiences of others suggests that brain areas coding the unpleasant aspects of pain might likewise contribute to an observer’s representation of others’ pain. It has been proposed that shared neural processes between feeling and seeing pain may underlie our ability to empathize with others’ distress (e.g., Gallese 2003), perhaps in a fashion resembling that of action-related mirror neurons (Morrison et al. 2004; Morrison 2007; Morrison and Downing 2007). In particular, neuroimaging investigations have shown that pain-related motivational-affective regions, notably the anterior and mid-cingulate cortex (ACC and MCC) and anterior insula, are activated when painful events are observed happening to other people (Jackson et al. 2005) and show overlapping activation when subjects feel pain themselves (Hutchison et al. 1999; Morrison et al. 2004; Singer et al. 2004).

“Empathic” responses in these regions have been shown to be modulated by facial expression (Botvinick et al. 2005; Saarela et al. 2007), attention (Fan and Han 2008), expertise (Cheng et al. 2007), previous interaction with or empathizing with the “victim” (Singer et al. 2006; Loggia et al. 2008), compassion (Farrow et al. 2001), imagination (Ogino et al. 2007), the ability to label emotions (Moriguchi et al. 2007), the administration of oxytocin nasal spray (Singer et al. 2008), pain catastrophizing (Sullivan et al. 2006), realism of the stimuli (Gu and Han 2007a; Fan and Han 2008), self-other distinctions (Lawrence et al. 2006; Jackson et al. 2006a), attentional and social factors (Lamm et al. 2007; Akitsuki and Decety 2009), and perspective-taking (Jackson et al. 2006b; Lamm et al. 2008). Pain-associated words also modulate areas implicated in empathy (Gu and Han 2007b). Observing others’ pain also elicits motor and sensory activity (Avenanti et al. 2005, 2006; Bufalari et al. 2007; Cheng et al. 2007; Lamm et al. 2009), and prompts overt behavioral responses (Morrison et al. 2006, Morrison et al. 2007a, b).

However, the evidence for specificity of empathy-eliciting stimuli, as well as the precise associated cortical loci, is not consistent. The exact location and presence of overlapping cingulate activations between felt and seen pain varies between individuals (Morrison and Downing 2007). It is possible that any objects in the visual field near the observed body part elicit pain-related responses when they are targets of potential pain stimulation (e.g., Armel and Ramachandran 2003) (but when rubber hands in peripersonal space were stabbed with a needle, cingulate and parietal responses showed a specific preference for an anatomically congruent orientation; Lloyd et al. 2006). Nor is it clear that “pain empathy” relies necessarily on first-person pain experience. Patients with congenital insensitivity to pain score normally on empathy questionnaires (Danziger et al. 2006) and show expected activation patterns (Danziger et al. 2009), implying that individuals who have never felt pain may nevertheless find alternate routes to empathy, independent of firsthand experience.

Our ability to “empathize”, in the sense of having relatively reliable, subjective insight into another person’s probable mental state, draws on the perception of features surrounding third-person tactile events. This ability is underpinned by mechanisms that draw attention to events surrounding the skin and facilitate the recognition of others’ likely pain or pleasure, and the consequences of these. Visual cues about such events, such as stroking speed or tissue damage, can inform or constrain an observer’s inferences about the probable subjective state of the observed person, or the relationship between two people in an interaction. This in turn may play an ultimate role in interpreting the emotions and relationships of others, allowing us to infer their emotional and intentional states and to predict their future behavior. The function of the processes leading to empathy may thus be heuristic, subserving learning about touch, pain and the circumstances surrounding them, rather than learning about the other’s mental state per se (although the latter may be a consequence of the former).


The cutaneous senses—especially touch—are crucial yet often-overlooked mediators of social interaction, contributing not only to sensation but to emotion. Social touch may be tentatively divided into the operational categories “simple” (as in a tap), “protracted” (as in a hug), and “dynamic” (as in a caress). The important social roles of the affective aspect of touch can be regarded from the perspective of social neuroscience. Specialized pathways for socially and affectively relevant touch may begin at the level of the skin, with CT afferent fibers, alongside associated processing in affect-related cortical areas such as the insula and orbitofrontal cortex. Major functional roles for social touch include affiliative behavior and communication. Touch- and pain-related representations also provide a basis for intersubjective representations, influencing the understanding of others’ sensory, emotional, and mental states.


I.M. would like to thank Tom Ziemke for his assistance during the writing of this article; thanks also to Christian Keysers and an anonymous reviewer for valuable suggestions.

Copyright information

© Springer-Verlag 2009