
2023 Impact Factor
The mirror neuron system (MNS) is a brain network activated when we move our body parts and when we observe the actions of other agent. Since the mirror neuron’s discovery in research on monkeys, several studies have examined its network and properties in both animals and humans. This review discusses MNS studies of animals and human MNS studies related to high-order social cognitions such as emotion and empathy, as well as relations between MNS dysfunction and mental disorders. Finally, these evidences are understood from an evolutionary perspective.
The mirror neuron system (MNS) is a network of neuron groups that discharge when an individual performs an action and/or observes an action of another agent. The MNS is divided into two principal hubs; the premotor area in the frontal lobe and the inferior parietal lobule (IPL).1,2) Additionally, the superior temporal sulcus (STS) is considered to be a key area of the MNS.3) Mirror neurons were first discovered in area F5 of the ventral premotor cortex (PMv) in macaques.4–6) They fire when monkeys observe other individuals grasp toward objects and when the monkeys execute the grasping motion themselves. Some neurons in the F5 posterior (F5p) area that respond to the presentation of objects as the monkey grasps an object are called “canonical neurons.”7,8) The F5p is also known as a hand-related area that encodes goal-directed actions. Both motor and mirror neurons are located in the F5 convexity (F5c) area and they fire during both observation and execution of specific goal-directed actions involving the hand and mouth.9)
In macaques, the frontal mirror area of the brain is divided into the lateral surface area 4 (primary motor cortex; M1), caudal part of area 6 (also a part of M1), medial surface area 6 (supplementary motor area; SMA) and premotor area. The premotor area is divided into caudal part (PMc)—with a direct connection to subcortical areas—and rostral part (PMr) that indirectly affects motor generation. The F5 area is a part of the parietal-frontal and prefrontal-frontal networks.10) Kraskov et al.11) have revealed that pyramidal tract neurons (PTN) in F5 suppress firing during observation but activate firing during movement. They interpreted the results as an inhibition effect suppressing motor generation during observation. Contrastingly, Vigneswaran et al.12) found that some PTN fired more during action observation when they recorded a signal from PTN in the area F1. Early mirror neuron studies suggested that mirror neurons are only engaged in action execution and in understanding the intention and transformation of visual perception into action execution. They suggest that the PMv has a connection with the caudal part of the inferior frontal gyrus (IFG). The IFG is also known as Broca’s area, a special motor area for human language generation.13,14)
Mirror aspects were also discovered in the IPL of macaque brains.15) Mirror neurons in the parietal area PF/PFG of the IPL also discharge during both the observation and execution of complex actions such as grasping an object to place it at a target locus or to eat it. Moreover, activity is modulated by the final goal of the action.16) The IPL receives visual information from eyes and somatosensory information from mouth, hands, and arms.17) This suggests that the visuo-motor organization of the IPL may build the neural basis of the ability to understand the intention of others’ action.18)
Research shows that the STS also has a mirror property. The STS is a brain area which responds to biological motions.19–21) It is not generally considered a part of the MNS because it does not have motor properties.2) The STS provides an audiovisual-to-motor link, that is, integration between seeing, hearing, and doing.22) Via the IPL, the area F5 is connected to higher-order visual areas of the STS.2) The first stream originates from a sector in the upper bank of the STS (STPm), reaches the parietal area PFG, and terminates in the area F5c. The second stream arises in the lower bank of the STS, reaches the anterior intraparietal area (AIP), and then the area F5a.23,24) Thus, recent literature suggests that the STS be considered a part of temporo-parieto-premotor pathways and the MNS.25,26)
Experiments comparing execution, imitation, observation, and imagination conditions using various neuroimaging techniques has revealed a network of the MNS.26–29) We present key regions and subregions of the human MNS in Figure 1. The key areas consist of the SMA, IPL, and STS. Subregions in the frontal lobe are the dorsal and PMv, IFG (including Broca’s area), and M1, which are related to motor function. The subregions in the parietal lobe include the primary somatosensory cortex (S1). The temporal and occipital subregions include the posterior middle temporal gyrus (pMTG) and middle temporal (MT/V5) area. Interestingly, the fusiform face area (FFA) is related with specific conditions such as facial expressions.30,31)
In this review, we first introduce various properties of mirror neurons based on animal electrophysiology. We then discuss the relationship between human MNS and higher-order cognitive abilities such as empathy and social cognition and conclude with an exploration of clinical connotations related to MNS dysfunction and mental disorders.
Multiple properties of mirror neurons have been revealed in studies on primates’ premotor mirror neuron. Rizzolatti et al.32) reported that some F5 neurons fire during grasping, but not holding, while others fire during holding, but not grasping. Umiltà et al.33) showed that more than half of the F5 mirror neurons discharged even in the invisible motor-action condition. Some neurons respond to observations of goal-directed actions (experimenter’s hand or mouth actions related to food items) but not to observation of the object alone or mimicking of action without the object.4,5) Some mirror neurons fire when macaques break a peanut and when they observe someone else breaking a peanut, as well as when the macaques hear the sound of a peanut breaking.10,34) These were named audio-visual mirror neurons.23) These data show that intentions of the actions of others can be recognized by the mirror mechanism. Thus, it strongly suggests that macaques can “read minds.”35) Later, Ishida et al.36) found visuo-tactile bimodal mirror neurons in the ventral intraparietal area of macaque brains, implying that there are not only multimodal properties in the MNS, but also bodily-action representations of self and others.
Mirror neurons may be located on the summit of the where/how visual pathway. The responses of some premotor mirror neurons were significantly modulated when the observed actions were executed in macaques’ peripersonal space, while the remaining neurons discharged more vigorously when the observed actions were executed in macaques’ extrapersonal space.37) These mirror neurons encode the spatial position of an observed action. More interestingly, Caggiano et al.38) reported that some of premotor mirror neurons seem to visually encode actions in a view-dependent manner (frontal vs. side view). In their study, the visual responses of mirror neurons were studied by presenting the same actions as seen from different points of view, including also the actor’s point of view. The discharges of mirror neurons were modulated not only by the action being observed, but also by the perspective from which it was observed.
Studies have recently examined the social implications of the MNS such as emotion and facial gestures in macaques. Ferrari et al.39) found evidence supporting that facial-imitation capabilities during the first week of life predicts the development of voluntary motor skills during the first year of life in macaques. This suggests that imitation abilities reflect not only general motor functions during development but also mirror neuron activity.
Songbirds have an auditory-vocal MNS involved in singing and learning of new songs. Their song system is commonly divided into two functionally distinct pathways; the posterior descending pathway (PDP), necessary for both the acquisition of new songs and production of learned song, and the anterior forebrain pathway (AFP), necessary only for acquisition.40,41) HVC (high vocal center with a former terminology), is a telencephalic nuclei, where auditory and vocal motor information emerge.42) Although it has been revealed that the HVC is not the highest order song related nuclei, it is the origin in which projection to the song-motor pathway (SMP) and AFP begins.43) In mammalian terms, HVC of songbirds is similar to the Broca’s area, PDP can be regarded as a homolog of a motor pathway which descends from the cerebral cortex to the brainstem, whereas the AFP can be considered as cortical-basal ganglia loop and thalamus involved in working memory.40,41,44–48) The HVC contains two types of neurons; HVCRA neurons which innervates the song motor nucleus RA (robust nucleus of the arcopallium) for singing pre-learned songs and HVCX neurons which provides an input to a striatopallidal structure area X (avian medial striatum) in the AFP for acquisition of a new song or arrangement of a pre-learned song.42,43,45)
Similar to Broca’s aphasia, adult songbirds with HVC lesions lose their singing ability, although they continue to produce innate vocalizations such as alarm calls and simpler vocalizations that resemble the babbling vocalizations produced by juvenile songbirds at the earliest stages of song learning.44,49) Interestingly, songbirds with HVC lesions also show deficits in the ability to recognize the songs of other birds of their own species,50) or to learn new contingencies to these songs.51) These results suggest that the HVC serves as auditory premotor and mirror neurons.41)
Researchers argue that dolphins may also have the MNS. Bottlenose dolphins can use parts of their bodies and comprehend gestural symbols, suggesting that they understand a mirror image and have self-recognition.52–54) Their mirroring system supports social intelligence, such as imitating other dolphins and humans.55) As Jeannerod notes that self-recognition is a prerequisite for establishing social communication with others,56) it is sufficient to say that dolphins may have the MNS.
Von Economo neurons (VENs) have been proposed as neurohistological evidence of higher cognitive abilities such as social and emotional cognition, awareness, and intuition and have been described in humans, great apes, elephants, cetaceans.52,57–60) The VENs are spindle-shaped, large bipolar projection neurons located in layer 3 and 5 of the anterior cingulate cortex (ACC), anterior insular (AI), and frontopolar (FP) cortex. Considering dolphins’ social intelligence and the existence of VENs, the supposition that dolphins may have the MNS is justified.
Electroencephalography (EEG) methods have mainly focused on mu rhythms. Mu rhythms refer to sensorimotor activity in alpha bands (8–13 Hz in adults, 6–9 Hz in children) recorded over central scalp locations C3, Cz, and C4; it is also referred to as “rolandic alpha.”22,61) Mu power decreases during both execution and observation of movements, and even when imagining movements.22,62) It is therefore called mu suppression or mu power desynchronization. Some studies have reported infant’s mu rhythm in central regions of the brain.61,63) A recent EEG study indicated greater mu suppression (putatively reflecting mirror neuron activation) in infants and adults when observing transitive movements compared to intransitive movements.64)
Hari et al.’s source localization study65) of mu rhythm activity based on magnetoencephalography (MEG) identified sensorimotor regions. Concurrent functional magnetic resonance imaging (fMRI) studies suggest that mu suppression may represent activity in brain areas of the MNS.66–69) On the other hand, some researchers suggest that mu suppression is not a perfect index of MNS activity. Aleksandrov and Tugin70) compared mu suppression from various types of motor activity and found that mu suppression was not significantly lower than it was in conditions where participants viewed human movement. They argued that mu suppression instead depends on the attention demands of the task, which decreases over time. Perry and Bentin71) found a similar pattern of changes in EEG power at the occipital and central electrodes. They argued that the significant differences could be attributed to differences in attentional demands between conditions, rather than differences in mirror neuron activity.
Studies using EEG and EMG have also included betaand theta-frequency band activity. Beta band is usually defined as 13 to 35 Hz, with a typical peak frequency of ~20 Hz. Researchers have also suggested that changes in beta activity index mirror neuron activity.72–74) Beta suppression is more related to motor processing and the primary motor cortex, while mu rhythm is linked predominantly to the somatosensory system.75,76) Theta rhythm (4–8 Hz) is also related to movement and activity in non-mu rhythm frequency bands by voluntary movement from the somatosensory area have been considered as an alternative index of the MNS function.77,78)
A number of transcranial magnetic stimulation (TMS) studies provide another putative measure of the human MNS. For example, Heiser et al.79) stimulated the Broca’s area with repetitive TMS (rTMS) while participants perform a key-press experiment. In their study, participants imitated a key-pressing movement done by another individual in experimental condition. In the control condition, participants pressed the keys in response to a red dot indicating the key to be pressed. rTMS lowered the participants’ performance during imitation but not during the control task. That is, a transient lesion by rTMS over left or right IFG induced a selective impairment of action imitation.
fMRI has been a popular research method in MNS studies. Iacoboni et al.80) conducted the first fMRI study with imitation. Participants were asked to observe, execute, or imitate the hands action on the screen. The IFG pars opercularis (IFGpo), posterior parietal cortex, and parietal operculum showed higher activation during imitation than during execution, and the former two areas were activated during observation but less than they were during execution. Buccino et al.’s fMRI study81) found bilateral premotor and parietal activations during object-related movements, and only premotor activation during non-object-related movements.
Lui et al.82) reported activations in key regions of the MNS when human participants observed mimed and meaningless movements. In addition, fMRI studies have consistently shown activation of MNS areas when participants perform, observe, and imagine a movement; when participants look at point-light biological animations or non-biological motions likely to evoke associations with biological action; and when participants hear object-directed action sounds.2,83–86) Caspers et al.’s meta-analysis28) on human MNS studies with fMRI and positron emission tomography confirmed basic key regions such as SMA, IPL, and STS activation in observation, imitation, and execution conditions and presented multiple associative areas. Molenberghs et al.87) also conducted a similar meta-analysis study on human MNS and consolidated MNS regions.
Some researchers argue that the MNS is engaged during the experience of emotion. For example, disgust induced by unpleasant odor increase activity level of the anterior insular, and the insula is also activated by the observation of disgust in others. These data strongly suggest that the insula contains neural populations that are activated both, when participants experience disgust and when they see it in others.88,89) Disgust is a very basic emotion whose expression has an important survival value for the conspecifics. Thus, the insula is the primary cortical area not only for chemical exteroception, such as gustation and olfaction, but also for the interoceptive state of the body. A mirror mechanism is also present in the insula and rostral cingulate cortex (rACC).90,91) This emotional mirroring system gives the observer a direct feeling of what others feel. The functional connection between the motor/premotor regions and insula limbic system is considered as a core neural system for emotion processing.88,92)
The posterior sector of the insula is characterized by connections among auditory, somatosensory, and premotor areas. From these data, it is clear that the insula is not exclusively a sensory area. In particular, emotional reactions to pain investigated using an fMRI paradigm showed that the same sites of the anterior insula and cingulate cortex were active in both conditions: when participants received an electric stimulation to induce pain and when participants observed their partner receiving a painful stimulation.90,93) The mirroring mechanism in empathic pain is similar to mechanism in empathic disgust.
Molnar-Szakacs94) suggested that perception of actions extends empathy and even morality through MNS. Moll et al.95) revealed that STS was more activated when participants saw pictures conveying moral violations than pictures that were emotionally unpleasant but without moral connotations. The results suggest that a network consisted of core regions of MNS, including the orbitofrontal cortex (OFC), medial frontal gyrus (MFG) and STS, is engaged in social cognition such as moral processing.
The idea that motor system may support social cognition and behavior has been adopted in MNS studies.96) Theory-of-mind (ToM) refers to the metacognitive ability to infer another person’s mental state, including their beliefs and desires, from their experiences and behavior.97) Empathy-related processing of emotional facial expressions recruits brain areas involved in mirror neuron and ToM mechanisms. Thus, once actions of another individual are represented and understood in terms of one’s own actions, it is possible to predict the mental state of the observed individual, leading to a ToM.22,98,99) The neural network supporting cognitive aspects of ToM includes temporo-parietal areas, the medial prefrontal cortex (mPFC), and temporal poles.35,100,101) The brain regions supporting ToM and structures previously associated with the human MNS are activated during the attribution of emotion to oneself and the other person. The activation of mirror neurons in a task relying on empathic abilities without explicit task-related motor components supports the view that mirror neurons are not only involved in motor cognition, but also in emotional interpersonal cognition.102) Gallese103,104) proposed the “shared manifold hypothesis” that people recognize other human beings as similar to us, suggesting that people can “read minds.” This hypothesis is associated with the argument that people perceive emotion in others by activating the same emotion in themselves.91,92,105–109)
Oberman et al.98) measured EEG when participants were watching a social interaction video clip. The MNS is less responsive to out-groups and most responsive to ingroups, and therefore holds implications for empathy and prejudice.110,111) A recent study investigated whether the MNS is more closely related to a type-1 (automatic, reflexive mental process) or type-2 process (deliberate decision-making and inferring of internal states). Christov-Moore and Iacoboni112) compared brain activations when participants imitated facial expression after pain observation and engaging in a dictator game. Activation levels in the pain observation situation predicted an altruistic decision in dictator game and a positive correlation was found in the superior parietal lobe, the pain-processing area. A network comprising cortical midline structures (CMS), including the mPFC, ACC, and precuneus, has been associated with self-processing and social cognition.113,114) As the MNS and CMS both seem to be involved in self/other representations, the direct connections between the precuneus—a major node of the CMS and the IPL—and the posterior component of the MNS can be an evidence of the interaction between two systems.27,115) Lou et al.116) suggest that this is one possible pathway in which such interactions might occur. The CMS shares a part of neural structures with the default mode network (DMN) that is generally active under the resting state. Thus, future studies on the association between these two networks can help understand the link between human cerebral processes and social abilities.
A reduction in mirror neuron activity may be involved in the pathophysiology of psychiatric conditions that are characterized by social cognitive deficits. Prime examples are autism, schizophrenia, and psychopathy.98,117–119)
It is well established that children with autism spectrum disorders (ASD) have deficits in imitation.120,121) Because the MNS is implicated in imitation learning, recent studies have sought to examine the development of the MNS in children with ASD. In a MEG study, individuals with Asperger’s syndrome showed delayed activation of the MNS in the inferior frontal lobe during an imitation task.122) Oberman et al.123) compared mu suppression in high-functioning individuals with ASD and age- and gender-matched control subjects while watching videos of a moving hand, a bouncing ball, and virtual noise, and while moving their own hand. Control subjects showed significant mu suppression in both, self and observed hand movements, whereas the ASD group showed significant mu suppression only in self-performed hand movements, but not in observed hand movements.
In a follow-up study, Oberman et al.124) on mu suppression in high-functioning children with ASD and ageand gender-matched control subjects, mu suppression was measured while the ASD children watched videos of a stranger opening and closing their right hand, watched their guardian or sibling performing the same hand motion, watched their own hand performing the same action, and watched a video of two bouncing balls moving vertically toward and away from each other. The results revealed that children with ASD and typically developing children showed greater mu suppression for actions performed by familiar individuals (such as their guardians/parents) compared to those of strangers. These data showed a lack of mu suppression in the autistic children during observation, implying a possible dysfunction in the MNS. This was later developed into the “broken mirror” hypothesis.
Dapretto et al.’s fMRI study117) specifically investigated the neural correlates of the capacity of imitating facial expressions of basic emotions in high-functioning ASD individuals. Their results showed that during observation and imitation, autistic individuals did not show activation of the MNS in the IFGpo. Importantly, activity in this area was inversely related with symptom severity in the social domain. The authors concluded that “a dysfunctional MNS may underlie the social deficits observed in autism.”125)
Limited studies have examined the association between other mental disorders and problems in the MNS. They suggest that schizophrenia and psychopathy may be related to a dysfunction in the MNS. According to McCormick et al.,126) participants with active schizophrenia-spectrum and psychotic disorders showed significantly greater mu suppression in the sensorimotor cortex of the left hemisphere than they did in their residual phase and when compared to healthy individuals. Those in the residual phase and healthy individuals showed similar levels of mu suppression. That is, greater left-sided mu suppression was positively correlated with psychotic symptoms, suggesting that abnormal mirror neuron activity may exist among patients with schizophrenia during the active phase of the illness.
Negative symptoms of schizophrenia include anhedonia, affective flattening and impaired socio-emotional processing.127–131) Empirical research has argued an association between negative symptoms and anomalous MNS activity.80,132–134) Using fMRI, Derntl et al.135) identified emphatic deficit in schizophrenia patients. They reported decreased brain activities in schizophrenia patients compared to healthy participants during three tasks related to empathy. Specific brain regions included those related to the MNS, such as IFG and MTG, and those associated to social cognition and emotion, such as the precuneus, cingulate, and amygdala. Using diffusion tensor imaging (DTI), Clark et al.136) have revealed abnormality in the temporal part of the superior longitudinal fasciculus (tSLF) in patients with schizophrenia. Considering that the SLF connects the frontal, parietal, temporal, and occipital lobes and it terminates the SMA, PMv and IFG in the frontal lobe, their results suggest that damages in neural tracts connected to frontal subregions of MNS might be associated with schizophrenia. Voineskos et al.137) have discovered abnormality in neural tracts of patients with deficit schizophrenia compared to patients with non-deficit schizophrenia and healthy participants. Their study suggests a significant relationships between negative symptoms of schizophrenia and abnormality of MNS, namely the arcuate fasciculus (AF) which connects two key regions of MNS, the inferior frontal and temporo-parietal areas.
Mehta et al.138) have systematically reviewed neuroimaging studies on mirror neuron activity in schizophrenia or patients with psychosis. Most of the studies included in their review reported decreased activity of MNS in the patients; only a few reported increased activity of MNS. In addition, most studies described premotor and motor cortices as the specific brain regions for mirror activity. The IFG, IPL, and posterior superior temporal sulcus (pSTS) were also identified as parts of MNS that exhibited abnormal activity in patients.
Psychopathy appears to be the inverse of autism. In psychopathy, emotional empathy was impaired with ToM and motor empathy intact; however, in autism, emotional empathy was fully functional and ToM and motor empathy showed deficits.119) The study of psychopathy suggested that distancing should not be pushed too far. In both cases, the diseases were understood through a refiguring of the two differences in the MNS.139)
Since its incidental discovery in macaques, the definition of mirror neurons has been extended to include a huge cerebral network. The MNS is located on the top of visual dorsal stream. This system contains supplementary-motor and temporo-parietal association areas that process visual, auditory, and somatosensory information. Table 1 includes the names and main functions of each region that constitute the MNS. Table 2 lists researches that examined specific functions of the brain in relations to specific brain regions and subregions of the MNS. Table 3 presents researches on specific psychological/psychiatric disorders related with certain areas of the brain and the subregions of MNS. The MNS is considered to be related not only with motor imitation and reading intention of other agents, but also with concepts in social interaction, such as emotion, empathy, and ToM. Hence, it is speculated that the cause of psychiatric disorders such as ASD may be strongly associated with dysfunctions of the MNS.
Indeed, MNS is important in language, learning and development itself. For example, a linkage between action and perception is thought to be important in language acquisition and utilization.140) Subtle evidences include how the IFG shows strong activation when participants imitate the actions of an experimenter.141) More direct evidences include how language acquisition by imitation of speech sounds have been supported by the acoustic mirror neurons in Broca’s area.2,34) In their study, Tettamanti et al.142) measured motor evoked potentials (MEP) after a single pulse TMS and simultaneously presented action-related sentences and found that the premotor mirror system modulates verbal language processing.
Regardless of the empirical support, there are skeptical opinions and controversies surrounding the existence and the role of the human MNS. Hickok143) made a critical opinion that strongly opposed motor simulation in the Broca’s area. Moreover, he argued that sensory systems are essential to recognize the actions of others. Heyes144) also has remarked that human MNS does not work as a dominant network specialized in understanding the actions of others and has hypothesized that the mirror neurons are a byproduct of associative learning. Critiques against the human MNS, have commonly mentioned that the relation between primates’ mirror neurons and the human MNS is either non-parallel or undetermined.143–145) For such reason, Hickok143) has argued that propositions suggesting a significant relation between the human MNS and higher social cognitive functions are mostly based on assumptions with little empirical evidence. However, these critiques do not deny the existence of mirror neurons in primates, and the existence of human mirror neurons based on a single neuron recording study by Mukamel et al.146) These opposing arguments on the human MNS are worth considering when interpreting its relations to human behaviors.
The MNS is hypothesized to be an evolutionary precursor and launching point for much of higher cognition and complex social behavior unique and fundamental to humans.78) Plotnik et al.60) suggested that elephants’ and dolphins’ mirror self-recognition could be linked to a convergent evolution related to complex sociality and cooperation. Reiss and Marino52) also suggested different neurological substrates of mirror self-recognition for dolphins and primates, as their brain physiology have diverged about 65 million to 70 million years ago.
Research on MNS across human, primates and other non-primate animals emphasize the idea of an ancient phylogenetic origin of the mirror mechanism and propose the function of MNS in the course of evolution.147) A more recent epigenetic perspective can provide new insights to understand the development of mirror neurons and the differences of mirror neurons between monkeys and humans.148) The epigenetic hypothesis proposes that the development of the MNS, which functions to encode goals and understands actions, is not simply a result of associative learning or natural selection, but is influenced by complicated epigenetic regulations that underlie neuronal plasticity, satisfying neuronal, cognitive and environmental “niches.”149) For example, research by Matsunaga et al.150) suggests that the demethylation of the Gadd45 gene family may be one of the major epigenetic regulation mechanisms responsible for the growth and differentiation of neurons in the parietal cortex of mice and marmosets.
Although there are a many studies on the MNS, several questions remain unanswered. First, the relationship between the MNS and DMN needs to be clarified, as they appear to share some brain areas related to cognition regarding self. Further, some research suggests that ASD is a dysfunction of the DMN.151) Finally, the relationship between psychiatric disorders and MNS dysfunction needs in-depth understanding, requiring more robust evidence beyond functional connectivity and neural correlates.
This work was supported by the Brain Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF-2015M3C7A1028252), and a grant from the Korea Science and Engineering Foundation (KOSEF), funded by the Korean government (NRF-2015 R1A2A2A01003564). The authors declare that there is no conflict of interest.
![]() |
![]() |