REVIEWS AhoughtheesonmvegentetdenCetgtnOi stress rly in the c text of s with a A he the superio amygdala damag sine th same stimul.Th of the ant s d the leftp m poin ature sup et of v-of-mi a ents of ting anothe al cort cortex in functic iaggtte uch n hat t n the ac een血thed ce o n sup in both isdoigsoi of othe ons orbit acial ex d th activation du and the can be interpret taro for right - of s n.or the rella of the So.there is goo devidence that we can figureou ine the n th re likely to act in part byn in d dire s might th hat there in the fit bet t Th uld be nt peop lat dge ith th that rontalortexaidorbitg pres sions rely on the inte tices have been linked to the regulation doest relyn the generation ofa more 172 MARCH 2003 VOLUME 4 172 | MARCH 2003 | VOLUME 4 www.nature.com/reviews/neuro REVIEWS relationships, social cooperativity, moral behaviour, and social aggression93–96. In this case, its role has been stressed particularly in the context of social develop￾ment and its pathologies (see the section on neuro￾psychiatry below). It could be that the integration of information about other people and oneself 97, and the social relationship between the two, are the hallmarks of medial prefrontal processing. Many of the same stimuli that engage the superior temporal gyrus, and lead viewers to attribute actions, intentions and goals, also activate regions of the neo￾cortex that are involved in representing actions70. These regions include premotor- and somatosensory-related cortices — the efferent and afferent sides of actions, respectively. A series of recent studies have investigated the role of the right somatosensory-related cortices and the left premotor cortex in making emotional and personality attributions from point-light displays and movements of geometric shapes. Damage in both regions impairs the ability to make such attributions98. Simulation. There is a rapidly growing literature sup￾porting the idea that we understand other people’s behaviour, in part, by simulation99. Observing another person’s actions results in desynchronization of motor cortex activity measured with MEG100. Imitating another subject’s actions through observation activates the pre￾motor cortex in functional imaging studies101; such acti￾vation is somatotopic with respect to the body part that is observed to perform the action, even in the absence of any overt action on the part of the observing subject102. In fact, in both humans103 and monkeys104, so-called ‘mirror neurons’ have been discovered. These neurons respond both when the subject is doing something spe￾cific, and when he or she observes another person doing the same thing. Damage restricted to somatosensory cor￾tex impairs the ability to recognize complex blends of emotions in facial expressions105 (FIG. 6), and there is an association between the impaired somatic sensation of one’s own body and the impaired ability to judge other people’s emotions105. Functional imaging studies also support a role for right somatosensory-related cortices in representing the actions that we observe other people performing, as being distinct from those that we perform ourselves106. So, there is good evidence that we can figure out how others are feeling, what they intend and how they are likely to act, in part by putting ourselves in their shoes, so to speak. This process could be entirely auto￾matic and covert, but it seems likely that there are con￾siderable differences in how skilled different people are at employing it. These differences would be expected to correlate with differences in empathy, emotional awareness or their dysfunction (as seen in sociopathy and ALEXITHYMIA, for example). There are also some unanswered questions about the extent of simulation that is necessary to construct social knowledge. For example, does the recognition of emotions from facial expressions rely on the internal generation of a motor or somatosensory representation of the face alone? Or does it rely on the generation of a more comprehensive Although there is convergent evidence that theory￾of-mind abilities emerge in a coordinated fashion during development, so far there is only preliminary evidence to indicate that they are a neuroanatomical package. The evidence for a role for the amygdala in the￾ory-of-mind abilities comes from a small number of patients with amygdala lesions74,75. A single functional imaging study has argued for amygdala activation in a theory-of-mind task requiring judgements about facial expressions76, and another study has found impairments after amygdala damage using the same stimuli77. The evidence is stronger for the medial prefrontal cortex, as several functional imaging studies have found that it is activated when subjects perform theory-of-mind tasks71,78,79 (FIG. 5c). In addition, a few studies have found that patients with damage to the frontal lobes are impaired on theory-of-mind tasks69,80. Furthermore, there is some evidence that the role of the medial pre￾frontal cortex in theory-of-mind tasks can be dissociated from its broader role in behavioural control and execu￾tive function that is also engaged by most of the tasks that are commonly used81,82. Rather than attempting to assign the whole set of theory-of-mind abilities to a particular neural struc￾ture or system, it might be more promising to explore the dependency of specific components of this ability on specific neural structures. In one study, it was found that damage to orbitofrontal cortex impaired the ability to detect a faux pas 83, perhaps indicating that this brain region contributes to our understand￾ing of other people in part by engaging the emotions and feelings that accompany social interaction. In sup￾port of this idea, it was found that appreciation of humour84, social-norm transgression resulting in embarrassment85, viewing of erotic stimuli86 and elici￾tation of other moral emotions87, all activate the medial prefrontal cortex. The role for the medial orbital and anterior cingulate regions in monitoring and regulating social emotions is consistent with their activation during interactions between attention, awareness and emotion88–90. The data can be interpreted along two different directions: the specialization of prefrontal cortices for aspects of social cognition, or the reliance of social cog￾nition on more general resources that are provided by this region of the brain. In line with the second interpre￾tation, sectors of prefrontal cortex seem to be crucial for integrating the allocation of cognitive resources on the basis of automatic emotional evaluation and volitional, effortful direction90,91. These mechanisms might there￾fore reflect aspects of a more general function in regulat￾ing the fit between goals and behaviour. There could then be specific examples of such domain-general pro￾cessing on which social behaviour draws: contextual inhibition by prefrontal cortex of emotional responses triggered by the amygdala92, or response inhibition and reversal in the face of a changing social context82. In line with the alternative interpretation that sectors of the prefrontal cortex are specialized for processing social information, medial and orbital prefrontal cor￾tices have been linked to the regulation of interpersonal ALEXITHYMIA Cognitive disturbance that is characterized by the difficulty in describing one’s own emotions. WASON SELECTION TASK The most popular experimental design for probing deductive reasoning. It consists of a conditional statement, the truth of which the subject must decide. Typically, conditionals about social rules, threats and promises all show a facilitation in the proportion of logically correct choices, and it has been argued that humans evolved a specialized skill to detect deception in the context of social contracts (for example, cheating). SOMATIC MARKERS Emotional states that are triggered during the consideration of potential future outcomes of choices