Mirror Neuron and Theory of Mind Mechanisms Involved in Face-to-Face Interactions: A Functional Magnetic Resonance Imaging Approach to Empathy Martin Schulte-Ru¨ther1,2, Hans J. Markowitsch3, Gereon R. Fink1,4, and Martina Piefke1,2 Abstract & Empathy allows emotional psychological inference about MPFC, the posterior cingulate cortex (PCC)/precuneus, and the other person’s mental states and feelings in social contexts. We temporo-parietal junction bilaterally. Empathy-related process- aimed at specifying the common and differential neural mech- ing of emotional facial expressions recruited brain areas anisms of ‘‘self’’- and ‘‘other’’-related attribution of emotional involved in mirror neuron and theory-of-mind (ToM) mecha- states using event-related functional magnetic resonance nisms. The differential engagement of the MPFC, the PCC/ imaging. Subjects viewed faces expressing emotions with direct precuneus, and temporo-parietal regions in the self-task indi- or averted gaze and either focused on their own emotional cates that these structures act as key players in the evaluation response to each face (self-task) or evaluated the emotional of one’s own emotional state during empathic face-to-face in- state expressed by the face (other-task). The common network teraction. Activation of mirror neurons in a task relying on activated by both tasks included the left lateral orbito-frontal empathic abilities without explicit task-related motor compo- and medial prefrontal cortices (MPFC), bilateral inferior frontal nents supports the view that mirror neurons are not only in- cortices, superior temporal sulci and temporal poles, as well volved in motor cognition but also in emotional interpersonal as the right cerebellum. In a subset of these regions, neural cognition. An interplay between ToM and mirror neuron mech- activity was significantly correlated with empathic abilities. The anisms may hold for the maintenance of a self–other distinction self- (relative to the other-) task differentially activated the during empathic interpersonal face-to-face interactions. & INTRODUCTION other distinction during interpersonal interaction. Em- A key aspect of social interaction is the inference of pathic feelings allow for socially appropriate emotional other persons’ emotional states by evaluating their facial responses, be it a shared or reactive emotional state. expressions (Ekman, 1982). These can help us to gain Therefore, empathy is different from pure emotional access to someone’s feelings and may act as triggers of contagion as the resulting emotional state can be dif- empathy. Empathy is based upon processes of psycho- ferent from the one being observed (see, for example, logical inferences about other persons’ mental and Davis, 1996; Stotland, 1969). emotional states occurring within a specific social con- It has been suggested that shared emotional feelings text. This context may provide the frame of reference for might be traced back to mirror neuron mechanisms the integration of emotion and cognition to yield in- (Gallese, 2003), which had initially been proposed in sights into the intentions and feelings of others (Decety the field of motor representations. Mirror neurons have & Jackson, 2004). Three core aspects are likely to play a been discovered in the monkey brain in frontal area role in the emergence and maintenance of empathy: (i) F5 and parietal area PF (see, e.g., Rizzolatti, Fogassi, an intuitive feeling of having something in common with & Gallese, 2001, for a review). There is increasing evi- the other person which relies on socially shared emo- dence from neuroimaging studies that core components tional experiences; (ii) cognitive mechanisms of per- of a human mirror neuron system (hMNS) comprise spective-taking; and (iii) the ability to maintain a self– parts of the inferior frontal cortex and the posterior parietal cortex (e.g., Koski, Iacoboni, Dubeau, Woods, & Mazziotta, 2003; Iacoboni et al., 1999). The mirror neu- 1Research Center Ju¨lich, Germany, 2University Hospital Aachen, ron account has recently been adapted to processes op- Rheinisch-Westfa¨lische Technische Hochschule (RWTH) Aachen, erating during interpersonal cognition (Gallese, Keysers, Germany, 3University of Bielefeld, Germany, 4University Hospital & Rizzolatti, 2004; however, see Jacob & Jeannerod, Cologne, Germany 2005 for a critique). It has been suggested that the D 2007 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 19:8, pp. 1354–1372 brain networks controlling motor behavior and social however, did not take into account that the empathic interaction share (at least in part) computational parallels switching between the self- and the other-perspective goes (Wolpert, Doya, & Kawato, 2003), and might be imple- beyond the ‘‘classic’’ first-person/third-person concept mented in similar or overlapping brain regions. Recent adopted by studies on perspective-taking (e.g., Vogeley studies indeed indicate that not only motor plans but also & Fink, 2003; Vogeley et al., 2001). Rather, empathy relies more abstract intentions (Iacoboni et al., 2005) or emo- on a dyadic social situation where the inferred intentions tional states (Carr, Iacoboni, Dubeau, Mazziotta, & Lenzi, and feelings of the other directly affect the self as the tar- 2003) recruit the same areas in the inferior frontal cortex get of the other’s emotion. This account is also supported which are supposed to contain mirror neurons. by studies of cognitive aspects of interpersonal interac- Carr et al. (2003) demonstrated that components of tions, suggesting that ‘‘on-line’’ social reasoning is most the hMNS (including the inferior frontal cortex) are in- efficiently triggered in a context where the observed volved in the imitation and observation of facial expres- behavior of others is highly relevant to one’s own inten- sions of emotions. Reports on hMNS involvement in the tions and actions (Kampe, Frith, & Frith, 2003; Gallagher, mere perception of facial expressions of emotions, how- Jack, Roepstorff, & Frith, 2002). To our knowledge, no ever, are yet rare. This might be due to the fact that most functional neuroimaging data are currently available on previous studies focused on the implicit processing of the neural circuits supporting empathic self–other inter- emotional facial expressions using distractor tasks such actions unfolding in the dynamics of emotional interper- as gender detection or conditioning paradigms (Winston, sonal contexts. O’Doherty, & Dolan, 2003; Critchley et al., 2000), and/or We used functional magnetic resonance imaging explicit tasks such as emotion categorization or judgment (fMRI) to investigate empathy-related neural processes, of emotion intensity (e.g., Winston et al., 2003; Critchley applying a paradigm which required both the decod- et al., 2000; George et al., 1993). All these tasks can be ing of other persons’ emotional states from facial cues accomplished by the use of perceptual strategies, for (‘‘other’’-condition) and the evaluation of one’s own example, pattern matching. The above studies may thus emotional response to these faces (‘‘self’’-condition). have failed to detect hMNS activity because the percep- Subjects evaluated either the emotional state of the tual processing of emotional facial expressions was not observed person or their own emotional reaction when embedded in an interpersonal context. viewing that person. Responses were given by choosing For pain and disgust, functional neuroimaging studies appropriate emotional descriptions from a list of adjec- showed evidence that the processing of one’s own emo- tives depicting emotional states. The experimental setup tional experience and the observation of another person thus differed from previous approaches in that it did having a similar emotional experience may engage over- not focus on a perceptual decision on the emotion ex- lapping brain networks (Singer et al., 2004; Wicker et al., pressed by a face (e.g., a decision between basic emotion 2003). Interestingly, in the study by Singer et al. (2004), categories), but rather assessed the interactive switching the relative engagement of these networks was corre- between the self- and the other-perspective during the lated with individual differences in empathic abilities. attribution of emotion. Using this paradigm, we were The cognitive aspects of empathic interpersonal per- able to experimentally construct an interpersonal con- spective taking are closely related to theory of mind text in which empathic social cognition could emerge. In (ToM). The ToM concept depicts meta-cognitive abilities addition, we manipulated the head direction of the such as the inference of beliefs, intentions, and desires stimulus faces such that they were either directed to of other people (Premack & Woodruff, 1978). Impor- the observer or averted by 458. An age/gender decision tantly, ToM also requires the distinction between one’s task on neutral faces was used as a high-level baseline. own thoughts and intentions and those of others, which Standard empathy scales were administered to assess is a key component of interpersonal interactions. individual differences in empathic abilities. Individual The neural network supporting cognitive aspects of scores on these scales were used to determine whether ToM include temporo-parietal areas, the medial prefron- measures of empathic capabilities may predict the am- tal cortex (MPFC), and the temporal poles (e.g., Vogeley plitude of empathy-related brain activity, especially in