Honore ton père et ta mère, afin que tes jours se prolongent dans le pays que l’Éternel, ton Dieu, te donne. Exode 20: 12
Instruis l’enfant selon la voie qu’il doit suivre; Et quand il sera vieux, il ne s’en détournera pas. Proverbes 22: 6
Si donc, méchants comme vous l’êtes, vous savez donner de bonnes choses à vos enfants, à combien plus forte raison votre Père qui est dans les cieux donnera-t-il de bonnes choses à ceux qui les lui demandent. Jésus (Matthieu 7: 11)
Comme le Père m’a aimé, je vous ai aussi aimés. (…) Aimez-vous les uns les autres, comme je vous ai aimés. Jésus (Jean 15: 9-12)
En vérité, en vérité, je vous le dis, le Fils ne peut rien faire de lui-même, il ne fait que ce qu’il voit faire au Père; et tout ce que le Père fait, le Fils aussi le fait pareillement. Car le Père aime le Fils, et lui montre tout ce qu’il fait; (…) Je ne puis rien faire de moi-même (…) je ne cherche pas ma volonté, mais la volonté de celui qui m’a envoyé. Jésus (Jean 5: 19-30)
Nul ne peut servir deux maîtres. Car, ou il haïra l’un, et aimera l’autre; ou il s’attachera à l’un, et méprisera l’autre. Vous ne pouvez servir Dieu et Mamon. Jésus (Matthieu 6: 24)
Car mon joug est doux, et mon fardeau léger. Jesus (Matthieu 11: 38)
Faites les gestes de la foi et vous croirez. Pascal
Car il ne faut pas se méconnaître, nous sommes automate autant qu’esprit. Et de là vient que l’instrument par lequel la persuasion se fait n’est pas la seule démonstration. Combien y a‑t‑il peu de choses démontrées ? Les preuves ne convainquent que l’esprit, la coutume fait nos preuves les plus fortes et les plus crues. Elle incline l’automate qui entraîne l’esprit sans qu’il y pense. Qui a démontré qu’il sera demain jour et que nous mourrons, et qu’y a‑t‑il de plus cru ? C’est donc la coutume qui nous en persuade. C’est elle qui fait tant de chrétiens, c’est elle qui fait les Turcs, les païens, les métiers, les soldats, etc. Il y a la foi reçue dans le baptême de plus aux chrétiens qu’aux païens. Enfin il faut avoir recours à elle quand une fois nous avons l’esprit a vu où est la vérité afin de nous en abreuver et nous teindre de cette créance qui nous échappe à toute heure, car d’en avoir toujours les preuves présentes c’est trop d’affaire. Il faut acquérir une créance plus facile qui est celle de l’habitude qui sans violence, sans art, sans argument nous fait croire toutes les choses et incline toutes nos puissances à cette croyance, en sorte que notre âme y tombe naturellement. Quand on ne croit que par la force de la conviction et que l’automate est incliné à croire le contraire ce n’est pas assez. Il faut donc faire croire nos deux pièces, l’esprit par démonstration les raisons qu’il suffit d’avoir vues une fois en sa vie et l’automate, par la coutume, et en ne lui permettant pas de s’incliner au contraire. Inclina cor meum Deus. Pascal
The moment I wake up before I put on my make up I say a little prayer for you … I run for the bus dear, while riding I think of us dear I say a little prayer for you … At work I just take time and all through my coffee break time I say a little prayer for you … Aretha Franklin
It may be the devil or it may be the Lord but you’re gonna have to serve somebody. Bob Dylan
But it’s so hard loving you … The Beatles
Jack (…) set aside an hour and a half each day for this. He’d spend the first 40 minutes or so relaxing and clearing his mind. Then he visualized a fox (he liked foxes). After four weeks, he started to feel the fox’s presence, and to have feelings he thought were the fox’s.(…) For a while he was intensely involved with her, and said it felt more wonderful than falling in love with a girl. Then he stopped spending all that time meditating — and the fox went away. It turned out she was fragile. (…) The mere fact that people like Jack find it intuitively possible to have invisible companions who talk back to them supports the claim that the idea of an invisible agent is basic to our psyche. But Jack’s story also makes it clear that experiencing an invisible companion as truly present — especially as an adult — takes work: constant concentration, a state that resembles prayer. It may seem paradoxical, but this very difficulty may be why evangelical churches emphasize a personal, intimate God. While the idea of God may be intuitively plausible — just as there are no atheists in foxholes, there are atheists who have prayed for parking spots — belief can be brittle. Indeed, churches that rely on a relatively impersonal God (like mainstream Protestant denominations) have seen their congregations dwindle over the last 50 years. To experience God as walking by your side, in conversation with you, is hard. Evangelical pastors often preach as if they are teaching people how to keep God constantly in mind, because it is so easy not to pray, to let God’s presence slip away. But when it works, people experience God as alive. Secular liberals sometimes take evolutionary psychology to mean that believing in God is the lazy option. But many churchgoers will tell you that keeping God real is what’s hard. T. M. Luhrmann
The essence of this mechanism — called the mirror mechanism — is the following: each time an individual observes another individual performing an action, a set of neurons that encode that action is activated in the observer’s cortical motor system. (…) The mirrormechanism was originally discovered in the ventral premotor cortex of the macaque monkey … Single-neuron recordings showed that this area contains neurons — mirror neurons — that discharge both when a monkey executes a specific motor act and when it observes another individual performing the same motor act. Mirror neurons do not fire in response to a simple presentation of objects, including food. Most of them do not respond or respond only weakly to the observation of the experimenter performing a motor act (for example, grasping) without a target object. (…) There is convincing evidence that an action observation–action execution mirror circuit also exists in humans. This evidence comes from brain imaging, transcranial magnetic stimulation (TMS), electroencephalography (eeG) and magnetoencephalography (MeG) studies. (…) The crucial issue concerning the parieto-frontal mirror neurons is their role in cognition. If this mirror mechanism is fundamental to understanding actions and intentions, the classical view — that the motor system has a role only in movement generation — has to be rejected and replaced by the view that the motor system is also one of the major players in cognitive functions. (…) Further evidence of goal encoding by the parieto-frontal mirror circuit was obtained in an fMRI experiment in which two aplasic individuals, born without arms and hands, and control volunteers were asked to watch video clips showing hand actions. All participants also performed actions with their feet, mouth and, in the case of controls, hands. The results showed that the parieto-frontal mirror circuit of aplasic individuals that was active during movements of the feet and mouth was also recruited by the observation of hand motor acts that they have never executed but the motor goals of which they could achieve using their feet or mouth. The issue of whether the human parieto-frontal mirror network encodes motor goals was also addressed by fMRI and TMS studies investigating the activation of motor areas in subjects listening to action-related sounds. Hearing and categorizing animal vocalizations preferentially activated the middle portion of the superior temporal gyri bilaterally (a region that is not related to motor act coding), whereas hearing and categorizing sounds of tools that were manipulated by hands activated the parieto-frontal mirror circuit. Similarly, it was shown that listening to the sound of hand and mouth motor acts activated the parieto-frontal mirror network. This activation was somatotopically organized in the left premotor cortex and was congruent with the motor somatotopy of hand and mouth actions. (…) In support of this view, two studies showed that the meaning of the motor acts of other individuals could be understood in the absence of visual information describing them. In one study, monkeys heard the sounds of a motor act (such as ripping a piece of paper) without seeing it; in the other study, the monkeys knew that behind a screen was an object and saw the experimenter’s hand disappear behind the screen, but they could not see any hand–object interaction. The results showed that in both experiments F5 mirror neurons in the monkeys fired in the absence of visual information describing the motor act of the experimenter. The neuronal activation therefore underpinned the comprehension of the goal of the motor act of the other individual, regardless of the sensory information that described that motor action. (…) There is no doubt that, in some cases, understanding the motor behaviour of others might require a mechanism different from mirroring. A typical example is the capacity of humans to recognize the actions of animals that do not belong to the human motor repertoire and cannot be captured by a motor generalization. The evidence for a non-mirror mechanism in action recognition was provided by an fMRI study in which volunteers were presented with video clips showing motor acts that did or did not belong to the human motor repertoire. Although all volunteers recognized the observed motor acts regardless of whether or not they belonged to their own motor repertoire, no activation of parieto-frontal mirror areas was found in response to acts that did not belong to their motor repertoire (for example, a dog barking). The areas that became active in such cases were occipital visual and STS areas. By contrast, the sight of motor acts that were within the motor repertoire of the observer (for example, a dog biting) recruited the parieto-frontal mirror network. (…) Finally, there is evidence that the mirror mechanism, possibly located in this case in the fronto-mesial areas, also has a role in setting up an anticipatory representation of the motor behaviour of another individual. It has been shown that the ‘Bereitschaftspotential’, an electrophysiological marker of the readiness to act, occurs not only when an individual actively performs a motor act, but also when the nature and the onset time of an upcoming action performed by another individual is predictable on the basis of a visual cue. (…) Such motor-based understanding seems to be a primary way in which individuals relate to one another, as shown by its presence not only in humans and monkeys, but also in evolutionarily distant species, such as swamp sparrows and zebra finches. (…) Saxophone playing has been used as an example to show that the mirror view of action understanding is “untenable”: no motor competence is required to understand that someone is playing a saxophone. This is true, but such competence leads to a different understanding of saxophone playing. The non-motor-based understanding implies a mere semantic knowledge of what a saxophone is for, whereas the motor experience allows an individual to understand what saxophone playing really means — that is, it provides a musical knowledge ‘from the inside’ (…) Furthermore, this mechanism indicates the existence of a profound natural link between individuals that is crucial for establishing inter-individual interactions. Finally, preliminary evidence suggests that the impairment of this natural link may be one of the causes of the striking inability of people with autism to relate to other individuals. Giacomo Rizzolatti and Corrado Sinigaglia
A l’heure où nos savants font la fine bouche (mais c’est aussi leur boulot et comme ça que la science avance) devant l’une des découvertes peut-être les plus révolutionnaires du siècle …
A savoir celle des neurones miroirs …
Sans lesquels, des primates aux humains mais aussi aux oiseaux, tant l’apprentissage que l’emphatie ne seraient possibles …
Comment ne pas voir avec ce récent article de l’anthropologue de Stanford T.M. Luhrman et le cas particulier de la religion …
L’importance, comme pour l’amour (voir Aretha Franklin) et comme le Christ lui-même l’a montré, de l’imitation active …
Pour initier une relation avec Dieu …
Mais, aussi et surtout comme par exemple la brillante mais brève période born again d’un chanteur comme Bob Dylan l’a si spectaculairement montré …
Pour l’entretenir et la maintenir …
T. M. Luhrmann
The New York Times
October 14, 2013
BIG SUR, Calif. — “AMERICANS are obsessed with the supernatural,” Jeffrey J. Kripal, a scholar of religion, told me here at Esalen, an institute dedicated to the idea that “we are all capable of the extraordinary.”
Surveys support this. In 2011, an Associated Press poll found that 8 in 10 Americans believed in angels — even 4 in 10 people who never went to church. In 2009 the Pew Research Center reported that 1 in 5 Americans experienced ghosts and 1 in 7 had consulted a psychic. In 2005, Gallup found that 3 out of 4 Americans believed in something paranormal, and that 4 in 10 said that houses could be haunted.
One interpretation of these data is that belief in the supernatural is hard-wired. Scholars like the anthropologist Pascal Boyer, author of “Religion Explained: The Evolutionary Origin of Religious Thought,” and the psychologist Justin L. Barrett, author of “Why Would Anyone Believe in God?” argue that the fear that one would be eaten by a lion, or killed by a man who wanted your stuff, shaped the way our minds evolved. Our hunter-gatherer ancestors were more likely to survive if they interpreted ambiguous noise as the sound of a predator. Most of the time it was the wind, of course, but if there really was danger, the people who worried about it were more likely to live.
That inclination to search for an agent has evolved into an intuition that an invisible agent, or god, may be there. (You can argue this theory from different theological positions. Mr. Boyer is an atheist, and treats religion as a mistake. Mr. Barrett is an evangelical Christian, who thinks that God’s hand steered evolution.)
However, intuitive plausibility is one thing, and measured, sober faith is another. These are the two kinds of thinking that the Nobel laureate Daniel Kahneman, author of “Thinking, Fast and Slow,” calls “system one” (quick intuitions) and “system two” ( deliberative judgment). When we’re scared in the dark, we populate the world with ghosts. When we consider in full daylight whether the ghosts were real — ah, that is another matter.
Consider how some people attempt to make what can only be imagined feel real. They do this by trying to create thought-forms, or imagined creatures, called tulpas. Their human creators are trying to imagine so vividly that the tulpas start to seem as if they can speak and act on their own. The term entered Western literature in 1929, through the explorer Alexandra David-Néel’s “Magic and Mystery in Tibet.” She wrote that Tibetan monks created tulpas as a spiritual discipline during intense meditation. The Internet has been a boon for tulpa practice, with dozens of sites with instructions on creating one.
Jack, a young man I interviewed, decided to make a tulpa when he was in college. He set aside an hour and a half each day for this. He’d spend the first 40 minutes or so relaxing and clearing his mind. Then he visualized a fox (he liked foxes). After four weeks, he started to feel the fox’s presence, and to have feelings he thought were the fox’s.
Finally, after a chemistry exam, he felt that she spoke to him. “I heard, clear as day, ‘Well, how did you do?’ ” he recalled. For a while he was intensely involved with her, and said it felt more wonderful than falling in love with a girl.
Then he stopped spending all that time meditating — and the fox went away. It turned out she was fragile. He says she comes back, sometimes unexpectedly, when he practices. She calms him down.
The mere fact that people like Jack find it intuitively possible to have invisible companions who talk back to them supports the claim that the idea of an invisible agent is basic to our psyche. But Jack’s story also makes it clear that experiencing an invisible companion as truly present — especially as an adult — takes work: constant concentration, a state that resembles prayer.
It may seem paradoxical, but this very difficulty may be why evangelical churches emphasize a personal, intimate God. While the idea of God may be intuitively plausible — just as there are no atheists in foxholes, there are atheists who have prayed for parking spots — belief can be brittle. Indeed, churches that rely on a relatively impersonal God (like mainstream Protestant denominations) have seen their congregations dwindle over the last 50 years.
To experience God as walking by your side, in conversation with you, is hard. Evangelical pastors often preach as if they are teaching people how to keep God constantly in mind, because it is so easy not to pray, to let God’s presence slip away. But when it works, people experience God as alive.
Secular liberals sometimes take evolutionary psychology to mean that believing in God is the lazy option. But many churchgoers will tell you that keeping God real is what’s hard.
T. M. Luhrmann, an anthropologist at Stanford, is a contributing opinion writer.
November 6, 2012
In the early 1990s, a team of neuroscientists at the University of Parma made a surprising discovery: Certain groups of neurons in the brains of macaque monkeys fired not only when a monkey performed an action – grabbing an apple out of a box, for instance – but also when the monkey watched someone else performing that action; and even when the monkey heard someone performing the action in another room.
In short, even though these “mirror neurons” were part of the brain’s motor system, they seemed to be correlated not with specific movements, but with specific goals.
Over the next few decades, this “action understanding” theory of mirror neurons blossomed into a wide range of promising speculations. Since most of us think of goals as more abstract than movements, mirror neurons confront us with the distinct possibility that those everyday categories may be missing crucial pieces of the puzzle – thus, some scientists propose that mirror neurons might be involved in feelings of empathy, while others think these cells may play central roles in human abilities like speech.
Some doctors even say they’ve discovered new treatments for mental disorders by reexamining diseases through the mirror neuron lens. For instance, UCLA’s Marco Iacoboni and others have put forth what Iacoboni called the “broken mirror hypothesis” of autism – the idea that malfunctioning mirror neurons are likely responsible for the lack of empathy and theory of mind found in severely autistic people.
Ever since these theories’ earliest days, though, sharp criticism has descended on the claims they make. If it turns out that mirror neurons play only auxiliary roles – and not central ones – in action understanding, as many opponents of these claims contend, we may be looking in entirely the wrong place for causes of autism and speech disorders. We could be ignoring potential cures by focusing on a hypothesis that’s grown too popular for its own good.
And through it all, the mirror neuron field continues to attract new inquisitive minds. September 2012 marked the first-ever Mirror Neurons: New Frontiers Summit in Erice, Sicily, where researchers championing all sides of the debate gathered to share their findings and hash out their differences.
In the wake of the Summit, I caught up with some of the world’s top mirror neuron experts, and asked them to bring me up to date on their latest findings, debates, and discussions. Their insights paint a more subtle, nuanced picture of mirror neurons’ role than anyone originally suspected.
Can mirror neurons understand?
There’s something strange about the range of actions mirror neurons respond to. They don’t respond to pantomimes, or to meaningless gestures, or to random animal sounds. They seem specially tuned to respond to actions with clear goals – whether those actions are perceived through sight, sound, or any other sensory pathway.
This realization led the discoverers of mirror neurons to put forth what they call the “action understanding” hypothesis – that mirror neurons are the neural basis for our ability to understand others’ actions. On this hypothesis rests a kingdom: If it’s true, Iacoboni may be right that we can treat autism and speech disorders by repairing the human mirror neuron system. But this kingdom’s borders have fallen under relentless attack since its very earliest days.
One of the first scientists to question the “action understanding” hypothesis was UC Irvine’s Greg Hickok. Though Hickok doesn’t dispute the existence of mirror neurons, he’s highly skeptical about their supposed central role in empathy, speech, autism and understanding – and he’s spent the past 10 years publishing research regarding those doubts.
The question of whether mirror neurons allow us to understand movement gestures, Hickok explains, is only one of the “action understanding” school’s unsupported claims – researchers who argue for a mirror neuron-centric model of speech comprehension also bear the burden of proving their claim that the motor system is involved in representing the meaning of action-related language.
What the “action understanding” school originally claimed, Hickok says, was that mirror neurons provide the neural mechanism for attaching meanings to motor actions – but in recent years, many of those same researchers have been leaning away from that claim, and toward the contention that mirror neurons themselves actually encode the meanings of actions. And both of these claims, according to Hickok, remain unsupported by hard evidence.
“Iacoboni and the other ‘action understanding’ supporters are conflating two logically independent questions,” Hickok explains. “Their original claim was that mirror neurons provide the mechanism for attaching meaning to actions like hand and speech gestures. But the second question – which they conflate with the first – is whether the meanings of actions are coded in motor systems.” In other words, before we can say for sure whether mirror neurons are necessary for understanding others’ actions, we first need to establish whether these neurons associate actions with their meanings, code the meanings themselves, or neither.
“It could be that mirror neurons facilitate your understanding a reaching movement,” Hickok adds, “but don’t themselves represent the semantics of the concept ‘reach’ generally.” In short, even if mirror neurons do enable your brain to access the concept ‘reach,’ that doesn’t mean they themselves are the neurons that encode that concept.
Over the years, Hickok has led several dozen studies that find dissociations between motor control and conceptual understanding. If he’s right, and mirror neurons help code movements but not semantic concepts of them, researchers may be looking for the causes of autism and speech disorders in areas that merely reflect, rather than produce, the symptoms – like picking trash out of a creek while ignoring the garbage dump upstream.
Take patients with Broca’s aphasia, for instance. These patients, who’ve suffered severe damage to the motor areas of their brain’s left hemisphere, have major trouble joining words into coherent phrases. Ask a person with Broca’s aphasia about the last time he visited the hospital, and he’ll say something like, “hospital… and ah… Wednesday… Wednesday, nine o’clock… and oh… Thursday… ten o’clock, ah doctors.” Even so, a patient with Broca’s aphasia can still understand sentences he hears others say. “If the neural system supporting speech production were critical to speech recognition,” Hickok says, “Broca’s aphasia should not exist.”
To use a more familiar example, babies – and, arguably, even dogs – clearly understand the meanings of many words without having the motor ability to say them. By the same token, we can understand the meaning of a verb like “echolocate” without having any understanding of how to perform it.
Thus, Hickok says, “hearing the word ‘kiss’ activates motor lip systems not because you need lips to understand the action,” but because your previous experiences with the word “kiss” are associated with movements involved in kissing. Mirror neurons, then, don’t encode the meaning of the word “kiss” itself; they simply happen to fall downstream of that understanding in your brain’s river of associations.
What all this implies, Hickok says, is that “action understanding is clearly not a function of the motor system.” If we want to find the neural correlates of understanding itself, Hickok suggests, we should concentrate our search upstream from the motor cortex, in brain regions like the superior temporal sulcus (STS), which plays a central role in our ability to associate objects with goals – to decide, in other words, what an action or object is “for.”
Not everyone’s thrilled by this line of argument, though. “When one looks at the data,” Iacoboni says, “true examples of dissociation between action understanding and action production are very rare.” Action understanding doesn’t always require motor-cortex activity, he agrees; but in many instances, mirror neurons do indeed appear to be crucial for it.
For example, patients with damaged motor cortices seem to have trouble placing photos of people’s actions in chronological order – though they have no trouble ordering photos of, say, a falling ball. Cases like these, Iacoboni says, argue strongly for mirror neurons’ importance in understanding the intentions of other people’s actions. This means, he says, that the concepts of “action” and “understanding” need to be integrated into a single model of mirror neuron function – not picked further apart.
But action execution and action understanding fall apart naturally, Hickok contends. “This is evident in the fact that the inability to produce speech following brain damage or in developmental speech disorders, for example, does not cause speech recognition deficits. It is also plainly evident in the fact that we can understand actions that we can’t perform, such as fly, slither, or coil.”
As you may have noticed by now, a specter that’s even harder to pin down lurks throughout this whole debate: We have no empirical rubric for action understanding; no experiment that can tell us for sure whether it’s happening – because there’s no real agreement about what exactly “understanding” is. It’s a weirdly recursive question: Understanding implies meaning; and so far, neither Hickok nor his opponents have been able to pin down what “meaning” means in neurological terms. “The fact is, we don’t know exactly how semantic understanding is achieved neurally,” Hickok says. “I certainly don’t know.”
Does association mean understanding?
It doesn’t always take a brand-new discovery to shake up an old debate – sometimes what’s needed is a new way of seeing the data. In the mirror neuron debate, that fresh approach comes courtesy of Cecilia Heyes, a professor of psychology at Oxford’s All Souls College. At the 2012 New Frontiers Summit, Heyes presented her case for an altogether different approach to studying mirror neuron function. The really important question, she says, isn’t whether mirror neurons encode understanding, but whether they qualify as a special class of neuron at all.
Mirror neurons, in Heyes’ view, aren’t evolved specifically “for” understanding, imitation, or any other purpose – rather, they’re simply ordinary motor-cortex neurons that happen to take on special roles as we learn to associate motor actions with sounds, feelings, goals and so on. “Special-purpose mechanisms can be forged by evolution or by learning,” Heyes says – and if we can figure out what makes certain neurons, but not others, take on mirror properties in the first place, we’ll be in a much better position to examine what they’re up to.
As for the question of whether mirror neurons “do” meaning, association, or both, Heyes thinks it may boil down to how we choose to define “meaning” and “understanding.” “I don’t think it’s right to contrast meaning and association,” she says. “In principle, mirror neurons could be a product of associative learning and help us to understand the meaning of actions.” But before we can find that out with a lab experiment, she adds, supporters and defenders of the “action understanding” hypothesis will need to explain what exactly it is that they’re claiming or denying, so we know what we’re looking for.
Hickok, for his part, says Heyes’ hypothesis actually supports his argument that mirror neurons don’t constitute the basis of action understanding – after all, he explains, if mirror neurons associate incoming stimuli with motor responses, why does the concept of “understanding” need to enter the picture at all? “The mirror neuron system links sensory stimuli to the motor system for the control of action,” he says. “It’s a system that acts reflexively and adaptively.” So as far as describing mirror neurons’ function in terms of sensory-motor association, Hickok says, Heyes is right on the money.
While Iacoboni also agrees that Heyes’ hypothesis is reasonable, he cautions that mirror neurons are still a special kind of associative cell: One that’s specialized for action-oriented associations. “Why should mirror neurons respond to specific actions,” Iacobini asks, “if they’re just learning visuomotor associations?” Why, in other words, do they respond not to just any action-related stimulus, but only to actions that have goals?
And it’s on this question of goal-orientedness – and what it implies about the human mind – that the views of Hickok, Heyes, and the Parma school all diverge once again.
Does empathy depend on mirror neurons?
No matter whose side of the debate you’re on, Vittorio Gallese cuts an imposing figure. One of the original discoverers of macaque mirror neurons – and a father of the “action understanding” theory – Gallese has spent the past three decades vigorously defending the centrality of mirror neurons in our ability to know what others’ actions are “for.”
“The data strongly suggest that mirror neurons map between an observer’s goals and the acting animal’s motor goals,” Gallese says. These neurons fire in relation to the goal of grasping, he explains, whether it’s performed by a hand, a pincer, or another tool; whether it’s performed by oneself or another individual; whether the other’s movement is seen or merely heard. The only common factor in all these events, Gallese says, is the goal they aim to achieve.
Gallese actually agrees with Hickok that understanding can take place without mirror neuron activation. However, he notes, “only through the activation of mirror neurons can we grasp the meaning of others’ behavior from within.” In other words, mirror neurons enable us to understand other people’s actions in terms of our own movements and goals – to empathize with them.
Hickok will have none of it. Gallese, he says, is trying to quietly slip out of his original hypothesis that mirror neurons associate meanings with actions, and into a more evasive “claim that they allow ‘understanding from the inside,’ whatever that means.”
Gallese has an answer at the ready: If not in mirror neurons, then where else should we look for action understanding? Surely not in the STS, as Hickok advocates. “Evidence demonstrates that only the motor system – not the STS – can generalize a motor goal independently from the effector accomplishing it,” Gallese says: When it comes to directly mapping others’ motor goals against our own, mirror neurons are still the only serious contenders in town. That kind of perceptual mapping, says Gallese, is what he means by “understanding from the inside.” More work is necessary, he acknowledges, to establish the exact nature of this kind of understanding – but nevertheless, its dependence on mirror neurons is clear.
Iacoboni is somewhat less sanguine. “Admittedly, it is very difficult to obtain empirical evidence that unequivocally proves this hypothesis,” he says – though he’s quick to add that “both imaging and neurological evidence are compellingly consistent with it.” The evidence is also consistent, he adds, with the idea that mirror neuron function is significantly altered in people on the autism spectrum of disorders (ASD) – implying a correlation between autism and “broken” mirror neurons.
That may be so, Heyes interjects – but ASD is too complex a range of disorders to lay at the feet of a single malfunctioning neuron system. “Iacoboni doesn’t ask,” she says, “whether atypical mirror mechanism activity generates – rather than merely accompanies – autism spectrum disorders.” If, as Hickok contends, mirror neurons lie far downstream in the process of action understanding, this abnormal mirror-neuron activation may simply be another symptom of autism, rather than its cause.
Gallese agrees – partially. “It is very unlikely that autism can be simply equated to a mere malfunctioning of the mirror neuron mechanism,” he says – but nevertheless, “many of the social cognitive impairments manifested by ASD individuals might be rooted in their incapacity to organize and directly grasp the intrinsic goal-related organization of motor behavior.” Mirror neurons map others’ motor goals to our own; autistic individuals have trouble grasping others’ goals; therefore, Gallese argues, some kind of correlation clearly exists.
But there’s an even more serious problem with this line of reasoning, says Morton Ann Gernsbacher, a prominent autism researcher at the University of Wisconsin-Madison. “It has been repeatedly demonstrated,” Gernsbacher says, “that autistic persons of all ages have no difficulty understanding the intention of other people’s actions.” Not only that – decades of research have also shown that autistic people can perform imitation tasks as well as or better than non-autistic participants, and that they can be highly responsive to imitation by others.
And so, once again, we come back to the question of what kind of understanding it is that we’re talking about here: Can people with autism really be said to “understand” an action they can’t readily imitate it? Gernsbacher says that, obviously, the answer’s yes. Gallese would argue that this isn’t “understanding from the inside,” but a more abstract kind.
Iacoboni, as usual, takes a more integrative view: “Current theories of empathy suggest a multilayer functional structure, with a core layer of automatic responses to reproduce the affective states of others. Mirror neurons are likely cellular candidates for the core layer of empathy.” And it’s that core layer of empathy, Iacobini says, that likely lies at the root of true action understanding.
In the final analysis, the one conclusion that’s emerged loud and clear from all these debates is that mirror neurons aren’t the end-all of understanding, empathy, autism, or any other brain function. The closer we examine the parts these neurons play, the more we find ourselves peering between the cracks of these mental processes – watching them unravel into threads that run throughout the brain. It may very well turn out that “meaning” and “understanding” aren’t single processes at all, but tangled webs of processes involving motor emulation, abstract cognition, and other emotional and instinctual components whose roles we’re only beginning to guess.
After decades of research, these strange cells continue to astound and confound us – not only with their unique abilities, but with the hidden complexity to which they may provide a key. But, as so often happens in neuroscience, we may end up having to pick the lock before we understand exactly how the key fits into it.
About the Author: Ben Thomas is an author, journalist, inventor and independent researcher who studies consciousness and the brain. A lifelong lover of all things mysterious and unexplained, he weaves tales from the frontiers of science into videos, podcasts and unique multimedia events. Lots more of his work is available at http://the-connectome.com. Follow on Twitter @theconnectome.
Giacomo Rizzolatti*and Corrado Sinigaglia
The parieto-frontal cortical circuit that is active during action observation is the circuit with mirror properties that has been most extensively studied. Yet, there remains controversy on its role in social cognition and its contribution to understanding the actions and intentions of other individuals. Recent studies in monkeys and humans have shed light on what the parieto-frontal cortical circuit encodes and its possible functional relevance for cognition. We conclude that, although there are several mechanisms through which one can understand the behaviour of other individuals, the parieto-frontal mechanism is the only one that allows an individual to understand the action of others ‘from the inside’ and gives the observer a first-person grasp of the motor goals and intentions of other individuals.
One of the most intriguing and exciting developments in neuroscience in recent years has been the discovery of a mechanism that unifies action perception and action execution 1–3 . The essence of this mechanism — called the mirror mechanism — is the following: each time an individual observes another individual performing an action, a set of neurons that encode that action is activated in the observer’s cortical motor system. The mirror mechanism is present in many cortical areas and brain centres of birds, monkeys and humans. The basic functions of these areas and centres vary con – siderably, from song production to the organization of goal-directed motor acts , to emotional processes. Thus, like other basic mechanisms (for example, excitatory postsynaptic potentials), the functional role of the mir – ror mechanism depends on its anatomical location, with its function ranging from recognition of the song of conspecifics in birds 4,5 to empathy in humans 6 . The aim of this article is not to review the vast literature on the mirror mechanism, but to focus on one spe – cific circuit endowed with mirror properties: the parieto- frontal action observation–action execution circuit. The reason for this choice is twofold. First, the proposed interpretation of the function of the parieto-frontal circuit as a mechanism that enables individuals to under – stand the actions and intentions of others ( mirror-based action understanding ) represented a paradigm shift in the classical view that these cognitive functions depend on higher-level mental processes. Second, mostly as a reaction to this new perspective, there have been attempts to interpret the functions of the action observation–action execution circuit in a way that minimizes or even denies its role in cognition. For these reasons, a review of the data on the mirror mechanism in the action observation–action execution network seems timely and necessary. In this Review, we examine first what the parieto-frontal action observation–action execution circuit encodes in monkeys and humans and then discuss its possible func – tional relevance for cognition. After examining different views on these issues, we conclude that the parieto-fron – tal mechanism allows an individual to understand the actions of another individual ‘from the inside’ and gives the observing individual a first-person grasp of the motor goals and intentions of another individual. The parieto-frontal mirror network The monkey parieto-frontal network. The mirror mechanism was originally discovered in the ventral premotor cortex of the macaque monkey (area F5) 1–3 . Single-neuron recordings showed that this area contains neurons — mirror neurons — that discharge both when a monkey executes a specific motor act and when it observes another individual performing the same motor act. Mirror neurons do not fire in response to a simple presentation of objects, including food. Most of them do not respond or respond only weakly to the observation of the experimenter performing a motor act (for example, grasping) without a target object 7 . Area F5 has recently been divided into three sectors: F5c, F5p and F5a 8–9 (FIG. 1) . Mirror neurons were originally recorded in the cortical convexity that corre – sponds to F5c 1–3 . However, functional MRI (fMRI) data showed that the other two areas also respond to observing a grasping action 8 . Mirror neurons are also present in the rostral part of the inferior parietal lobule (I pl ), particularly in area p FG 10 – 12 and the anterior intraparietal area (AI p ) 9,13 (FIG. 1) . Both these areas are heavily connected with F5: p FG mostly with F5c, and the AI p with F5a 14 . Both area p FG and the AI p receive higher-order visual infor – mation from the cortex located inside the superior temporal sulcus (STS) 13 – 14 . STS areas, like mirror areas, encode bio – logical motion, but they lack motor properties. They are therefore not part of the mirror system in a strict sense. The AI p also receives connections from the middle temporal gyrus 15 . This input could provide the mirror areas with information concerning object identity. Finally, area F5 is connected with area F6 — the pre- supplementary motor area (pre-SMA) — and with the prefrontal cortex (area 46) 16 . The prefrontal cortex is also richly connected with the AI p 16 . The frontal inputs con – trol the selection of self-generated and stimulus-driven actions according to the intentions of the agent 17 . It was recently shown that, in addition to areas p FG and AI p , two other areas of the parietal lobe contain mirror neurons: the lateral intraparietal area and the ventral intraparietal area. The mirror properties of neurons in these areas are not the focus of this Review but are briefly discussed in BOX 1 . The human parieto-frontal network. There is convinc – ing evidence that an action observation–action execu – tion mirror circuit also exists in humans. This evidence comes from brain imaging, transcranial magnetic stimulation (TMS), electroencephalography ( ee G) and magnetoencephalography (M e G) studies. Brain imaging studies have shown that, as in the mon – key, this action observation–action execution mirror cir – cuit is formed by two main regions: the inferior section of the precentral gyrus plus the posterior part of the inferior frontal gyrus; and the inferior parietal lobule, includ – ing the cortex located inside the intraparietal sulcus 18 . Additional cortical areas (such as the dorsal premotor cor – tex and the superior parietal lobule) have also been occa – sionally found to be active during action observation and execution 19–21 . Although it is possible that their activation is due to a mirror mechanism, it is equally possible that it reflects motor preparation. In support of this interpreta – tion are single-neuron data from monkeys showing that these areas are involved in covert motor preparation 22–23 . As for the superior parietal lobule, although its activation is typically absent in studies in which the experimenters use distal motor acts as visual stimuli, it is prominent when volunteers observe proximal arm movements that are directed to a particular location in space 24 . Single-subject fMRI analyses have recently provided evidence that other cortical areas (for example, the pri – mary and secondary somatosensory cortices and the middle temporal cortex) also become active during action observation and action execution 21 . It has been suggested 21 that these activations outside of the ‘classi – cal’ mirror areas are caused by additional mechanisms (for example, internal models) that are triggered by the mirror mechanism. These activations would enrich the information about the actions of other individuals that the mirror mechanism provides. A tale of two populations. Some authors have recently argued that the activation of the same areas during action observation and action execution is not suffi – cient to prove the existence of the mirror mechanism in humans 25 . Instead, they have suggested that, in humans, motor areas have distinct, segregated populations of vis – ual and motor neurons, the visual neurons discharging during action observation and the motor neurons during action execution. They proposed to use the ‘repetition– suppression’ technique — that is, a technique based on the progressive reduction of a physiological response to repeated stimuli to prove this point 25 . If mirror neurons exist in humans, they should ‘adapt’ when the observa – tion of a motor act is followed by the execution of that motor act, and vice versa . The ‘adaptation’ effects are, in general, difficult to interpret 26 . Adaptation occurs at the synaptic level and should therefore be present only when information repeatedly reaches a neuron through the same or largely common pathways 27 . This input commonality is typically absent when mirror neurons are activated during action observation and execution. During action observation, the input to the parieto-frontal circuit arrives from higher- order visual areas (for example, the STS) 16 whereas, during voluntary movement, it mostly comes from the frontal lobes 17 . The results of adaptation experiments therefore depend on the design of the experimental paradigm and on the stimuli used. These considerations could explain why the results of repetition–suppression experiments have been contradictory. Although some authors found evidence of the mirror mechanism in the parietal 28 or the frontal nodes 29 , others obtained negative results 30–31 . Regardless of the empirical data that may help to define some properties of the parieto-frontal mirror mechanism, the logic of the two-population story is flawed. Assuming that neurons in motor areas respond – ing to action observation are merely visual neurons implies that motor areas contain a large number of ‘dis – placed’ visual neurons and that these neurons do not communicate with their ‘neighbour’ motor neurons. Both these assumptions are hard to reconcile with what is known about the organization of the cerebral cortex. Most importantly, TMS studies have shown a clear con – gruence between the observed motor act and the acti – vated motor representation 32–36 . Thus, if higher-order sensory information describing a motor act reaches motor neurons that encode that same motor act, these motor neurons are mirror neurons by definition. Humans do not differ from monkeys in this respect. What do parieto-frontal mirror neurons encode? Evidence for goal coding in monkeys. The crucial issue concerning the parieto-frontal mirror neurons is their role in cognition. If this mirror mechanism is fundamental to understanding actions and intentions, the classical view — that the motor system has a role only in movement generation — has to be rejected and replaced by the view that the motor system is also one of the major players in cognitive functions. To address this fundamental issue, a preliminary problem must first be solved: what do the parieto-frontal mirror neurons encode when they discharge in response to the observation of the actions of others? A way to solve this problem is to examine what mir – ror neurons encode when they discharge during motor behaviour. w hat is recorded in single-neuron studies during both action execution and observation are action potentials — that is, neuronal output. Thus, having deter – mined what neurons encode during the execution of an agent’s own motor act, one also knows what they encode when they are triggered by the agent’s observation of a motor behaviour of others. e arly experiments on area F5 found that most of the motor neurons in this area encode motor acts (that is, goal-related movements, such as grasping) rather than movements (that is, body-part displacements without a specific goal, such as finger flexion) 3 7 –38 . A recent study provided compelling evidence that this is the case 39 . This study describes single-neuron recordings from monkeys that were trained to grasp objects using two types of pliers: normal pliers, which require typical grasping movements of the hand, and ‘reverse’ pliers, which require hand move – ments executed in the reverse order (that is, first closing and then opening the fingers). The results showed that F5 neurons discharged during the same phase of grasp – ing in both conditions, regardless of whether this involved opening or closing of the hand (FIG. 2) . The functional properties of I pl motor neurons are similar to those of F5 neurons: the goal of the executed motor acts is the parameter that is encoded by I pl neurons that fire during the execution of motor acts 11,40 – 42 . The mirror neurons in F5 and I pl do not differ in their motor properties from parieto-frontal motor neu – rons that do not have visual properties 1–3 . Thus, when they fire in response to motor act observation, they send information about the goal of the observed motor acts. This information can be encoded with different degrees of generality: some mirror neurons (strictly congruent mir – ror neurons) fire when the observed and executed motor acts are the same (for example, grasping with precision grip), whereas other mirror neurons (broadly congruent mirror neurons) fire when the observed motor act has the same goal as the executed motor act (for example, grasp – ing), but can be achieved in a different way (for example, with both precision and whole-hand grips) 43–44 . Recently, a single-neuron study investigated the effect of the spatial relationships between an agent and an observer, comparing F5 mirror neuron responses to motor acts performed near the monkey (in the peripersonal space) or outside its reach (in the extra – personal space) 45 (FIG. 3) . The results showed that many F5 mirror neurons were differentially modulated by the location of the observed motor act. Some neurons were selective for actions executed in the monkey’s peripersonal space, whereas others were selective for stimuli in the extrapersonal space. These findings indicate that mirror neurons may encode the goal of the motor acts of another individual in an observer-centred spatial framework, thus providing the observer with crucial information for organizing their own future behaviour in cooperation or competition with the observed individuals. Goal and single-movement coding in humans. In accordance with early findings 46–49 , a series of new fMRI studies provided strong evidence that the human parieto- frontal mirror circuit encodes the goal of observed motor acts. Volunteers were instructed to observe video clips in which either a human or a robot arm grasped objects 50 . Despite differences in shape and kinematics between the human and robot arms, the parieto-frontal mirror circuit was activated in both conditions. Another group extended these results by investigating cortical activation in response to the observation of motor acts performed by a human hand, a robot hand or a tool 51 . Here, bilat – eral activation of a mirror network formed by intra – parietal and ventral premotor cortex occured, regardless of the effector. In addition, the observation of tool actions produced a specific activation of a rostral sector of the left anterior supramarginal gyrus, suggesting that this sector specifically evolved for tool use. Further evidence of goal encoding by the parieto- frontal mirror circuit was obtained in an fMRI experi – ment in which two aplasic individuals, born without arms and hands, and control volunteers were asked to watch video clips showing hand actions 52 . All partici – pants also performed actions with their feet, mouth and, in the case of controls, hands. The results showed that the parieto-frontal mirror circuit of aplasic individuals that was active during movements of the feet and mouth was also recruited by the observation of hand motor acts that they have never executed but the motor goals of which they could achieve using their feet or mouth. The issue of whether the human parieto-frontal mir – ror network encodes motor goals was also addressed by fMRI and TMS studies investigating the activation of motor areas in subjects listening to action-related sounds. Hearing and categorizing animal vocalizations preferentially activated the middle portion of the supe – rior temporal gyri bilaterally (a region that is not related to motor act coding), whereas hearing and categoriz – ing sounds of tools that were manipulated by hands activated the parieto-frontal mirror circuit 53 . Similarly, it was shown that listening to the sound of hand and mouth motor acts activated the parieto-frontal mirror network 54 . This activation was somatotopically organ – ized in the left premotor cortex and was congruent with the motor somatotopy of hand and mouth actions. u nlike in monkeys, the parieto-frontal mirror circuit of humans also becomes active during the observation of individual movements 55–56 . The initial evidence for this mechanism was based on TMS experiments which indi – cated that the observation of the movements of others results in an activation of the muscles involved in the execution of those movements 32–36 . Additional support comes from ee G and M e G studies showing that the observation of movements without a goal desynchronizes the rhythms recorded from motor areas 5 7 –64 . Recently, it was shown that mirror coding might depend on the content of the observed behaviour. Motor evoked potentials (M ep s) in response to TMS were recorded from the right opponens pollicis (O p ) muscle in participants observing an experimenter either open – ing and closing normal and reverse pliers or using them to grasp objects 65 . The observation of tool movements (that is, opening and closing the pliers without grasping anything) activated a cortical representation of the hand movements involved in the observed motor behaviour. By contrast, the observation of the tool grasping action activated a cortical representation of the observed motor goal , irrespective of the individual movements and the order of movements required to achieve it. Together, these findings show that the human parieto-frontal mirror network encodes both motor acts and movements. Understanding the actions of others Cognitive functions of the parieto-frontal network: evidence and criticisms. w hy should the motor sys – tem encode the goal of actions performed by others? From the discovery of mirror neurons, the interpreta – tion of this finding was that they allow the observer to understand directly the goal of the actions of others 1–3 : observing actions performed by another individual elic – its a motor activation in the brain of the observer similar to that which occurs when the observer plans their own actions, and the similarity between these two activations allows the observer to understand the actions of others without needing inferential processing 43–44 . In support of this view, two studies showed that the meaning of the motor acts of other individuals could be understood in the absence of visual information describing them. In one study, monkeys heard the sounds of a motor act (such as ripping a piece of paper) without seeing it 66 ; in the other study, the monkeys knew that behind a screen was an object and saw the experimenter’s hand disappear behind the screen, but they could not see any hand–object interaction 67 . The results showed that in both experiments F5 mirror neu – rons in the monkeys fired in the absence of visual infor – mation describing the motor act of the experimenter. The neuronal activation therefore underpinned the comprehension of the goal of the motor act of the other individual, regardless of the sensory information that described that motor act. This interpretation of the function of the parieto-frontal mirror mechanism has been challenged with objections and alternative proposals 68–71 . A key criticism has been advanced by Csibra 69 . He argued that the interpretation of mirror neuron function in terms of action understanding contains a “tension” between “the claim that the mirror mechanism reflects nothing else but faithful duplication of the observed action” and “the claim that mirroring rep – resents high-level interpretation of the observed action”. In other words, if mirror activity represents a copy of the observed motor act, it is not sufficiently general to capture the goal of that motor act; conversely, if it is sufficiently general for goal understanding, it cannot be interpreted in terms of a direct matching mechanism between sensory and motor representations (see also R EFS 70,71 ). In the earlier studies on the mirror mechanism, it was indeed not clearly specified that the parieto-frontal mirror mechanism in humans is involved in two kinds of sensory–motor transformation — one mapping the observed movements onto the observer’s own motor representation of those movements (movement mirror – ing), the other mapping the goal of the observed motor act onto the observer’s own motor representation of that motor act (goal mirroring), as described above. By match – ing individual movements, mirror processing provides a representation of body part movements that might serve various functions (for example, imitation), but is devoid of any specific cognitive importance per se . By contrast, through matching the goal of the observed motor act with a motor act that has the same goal, the observer is able to understand what the agent is doing. This is true not only for the mirror neurons that are broadly congru – ent but also for those that are strictly congruent, because these neurons also do not encode the elementary aspects of a movement (for example, its kinematics), but respond to the goal of the observed motor acts 44,56 . Typically, authors who play down or even deny the importance of the motor system for cognitive functions suggest that goal understanding is primarily due to cortical activation in the STS. This region, as described in a series of fundamental studies in monkeys 72,73 , is involved in the visual analysis of the actions of others. Several fMRI studies showed a similar role for the STS in humans (see R EFS 74,75 for a review). There is little doubt that STS neurons have an impor – tant role in encoding the behaviour of others. However, it is unlikely that the STS by itself mediates the processing of action understanding, relegating the parieto-frontal mir – ror network to an ancillary role in this function 65 : among the neurons in various areas that become active during action observation, only those that can encode the goal of the motor behaviour of another individual with the great – est degree of generality can be considered to be crucial for action understanding, and the available evidence shows that this capacity for generalization characterizes the parieto- frontal mirror neurons rather than STS cells. Indeed, pari – eto-frontal mirror neurons encode the goal of observed motor acts regardless of whether they are performed with the mouth, the hand or even with tools. Although STS neurons may encode some types of motor act, goal gener – alization such as is achieved by the parieto-frontal mirror neurons seems to be absent in the STS 72,73 . Most importantly, there are theoretical reasons why STS neurons are unlikely to encode actions with the same degree of generality as parieto-frontal mirror neurons. If an STS neuron selectively encodes the visual features of a given hand action (for example, grasping), it is unclear how this neuron would also be able to encode selectively the visual features of a mouth performing the same motor act. One could postulate an associa – tion process similar to that described for the temporal lobe 76,77 . However, in the STS, the association would be between spatio-temporally adjacent visual representa – tions of body part movements and not between visual representations of the same motor goal achieved by different effectors. By contrast, parieto-frontal mirror neurons — owing to their motor nature and the fact that they encode the goal of motor acts — can be trig – gered by different visual stimuli (for example, hand and mouth actions) that have a common goal (for example, grasping). Only the presence of a ‘motor scaffold’ that provides the goal-related aspects of observed actions can allow this generalization; such generalization cannot be achieved by mere visual association. A recent study provides empirical evidence in favour of this point 78 . The study was based on a TMS adaptation paradigm 79 . p articipants were presented with ‘adapta – tion-inducing’ movies of a hand or foot acting on vari – ous objects and asked to respond as quickly as possible to a picture of a motor act similar to that of the movie. TMS pulses were delivered over the ventral premotor cortex bilaterally, over the left I pl and over the left STS. The results showed that the delivery of TMS over both premotor and I pl cortices shortened the reaction times to ‘adapted’ motor acts regardless of which effector performed the observed motor act; by contrast, TMS stimulation of the STS shortened the reaction times to ‘adapted’ motor acts only if the same effector executed the act in the movie and in the test picture. Understanding actions from the inside. Another argu – ment against the role of mirror neurons in action under – standing is that there are several behavioural instances in which individuals understand the actions of others even if they are unable to perform them. For example, macaques can react to the observation of humans mak – ing the gesture of throwing objects overhand towards them 80 . It was proposed that, although monkeys never throw objects overhand, they could nevertheless under – stand the action they saw because they analysed the vari – ous visual elements of the observed actions and applied some form of inferential reasoning . However, this argument would only be valid if the parieto-frontal mirror mechanism consisted solely of strictly congruent mirror neurons. As the authors of the study themselves recognize 80 , the capacity of broadly congruent mirror neurons to generalize the goal of motor acts might account for the observed phenome – non. Given that broadly congruent mirror neurons may generalize from a hand action to actions performed with tools, even when they are as bizarre as reverse pliers, it is plausible that they could equally generalize from one type of throwing to another. There is no doubt that, in some cases, understanding the motor behaviour of others might require a mechanism different from mirroring. A typical example is the capacity of humans to recognize the actions of animals that do not belong to the human motor repertoire and cannot be captured by a motor generalization. e vidence for a non- mirror mechanism in action recognition was provided by an fMRI study in which volunteers were presented with video clips showing motor acts that did or did not belong to the human motor repertoire 81 . Although all volunteers recognized the observed motor acts regardless of whether or not they belonged to their own motor repertoire, no activation of parieto-frontal mirror areas was found in response to acts that did not belong to their motor reper – toire (for example, a dog barking). The areas that became active in such cases were occipital visual and STS areas. By contrast, the sight of motor acts that were within the motor repertoire of the observer (for example, a dog biting) recruited the parieto-frontal mirror network. These data indicate that the recognition of the motor behaviour of others can rely on the mere processing of its visual aspects. This processing is similar to that performed by the ‘ventral stream’ areas for the recogni – tion of inanimate objects. It allows the labelling of the observed behaviour, but does not provide the observer with cues that are necessary for a real understanding of the conveyed message (for example, the communica – tive intent of the barking dog). By contrast, when the observed action impinges on the motor system through the mirror mechanism, that action is not only visu – ally labelled but also understood, because the motor epresentation of its goal is shared by the observer and the agent. In other words, the observed action is under – stood from the inside as a motor possibility and not just from the outside as a mere visual experience (BOX 2) . Understanding motor intentions of others From motor goals to motor intentions. The properties of parieto-frontal mirror neurons described above indicate that their activity reflects what is going on in the ‘here and now’. However, there is evidence that parietal and frontal mirror neurons are involved in encoding not only the observed motor acts but also the entire action of which the observed motor act is part. Monkeys were trained to grasp objects with two different motor inten – tions: to place them into a container or to bring them to their mouth 11 . After training, motor neurons in the I pl that encode grasping were studied in the two set-ups. The results showed that the majority of these neurons discharged with an intensity that varied according to the action in which the motor act was embedded (‘action- constrained motor neurons’). This finding implies that the I pl contains ‘chains’ of neurons in which each neuron encodes a given motor act and is linked to oth – ers that are selective for another specific motor act. Together, they encode a specific action (for example, grasping for eating). A striking result of this study was that many of these action-constrained motor neurons have mirror proper – ties. w hen tested in the two set-ups described above, the majority of these neurons were differently activated depending on the action to which the observed motor act belonged (‘action-constrained mirror neurons’). This finding indicates that, in addition to describing what the observed individual is doing (for example, grasping), I pl mirror neurons also help the observer to explain why the individual is performing the action, owing to chained organization in the I pl . That is, I pl mirror neurons ena – ble the observer to recognize the agent’s motor intention. A recent study demonstrated that action-constrained neurons are also present in area F5 ( REF . 82) . The compar – ison of F5 and I pl (specifically area p FG) mirror neuron properties revealed no clear differences in their capacity to encode the motor intentions of others. e vidence that the parieto-frontal mirror circuit in humans is also involved in intention encoding was first provided by an fMRI experiment consisting of three conditions 83 . In the first (the ‘context condition’) the vol – unteers saw a photo of some objects arranged as for an ongoing breakfast or arranged as though the breakfast had just finished; in the second (the ‘action condition’), the volunteers saw a photo of a hand grasping a mug without any context; in the third (the ‘intention condition’) they saw photos showing the same hand actions within the two contexts. In this condition, the context provided clues for understanding the intention of the motor act. The results showed that the intention condition induced a stronger activation than the other two conditions in the caudal inferior frontal gyrus of the right hemisphere. An activation of the right parieto-frontal mirror cir – cuit during intention understanding was also described in a repetition–suppression fMRI experiment 84 . p articipants were presented with movies showing motor actions (for example, pushing or pulling a lid) that could lead to the same or to different outcomes (for example, opening or closing a box). The results showed that the responses in the right I pl and right inferior frontal cortex were ‘suppressed’ when participants saw movies of motor actions that had the same outcome, regard – less of the individual movements involved. Responses in these regions were not influenced by the kinematics parameters of the observed motor action. Brain imaging experiments allow the cortical sub – strate of a given function to be located, but they do not give information about the mechanism underlying the function. Cattaneo and colleagues tested whether the understanding of motor intention in humans might be based on the ‘chain mechanism’ described in the monkey 85 . p articipants were asked to grasp a piece of food and eat it or to grasp a piece of food and place it in a container. In another condition, they had to observe an experimenter performing the same actions. In both the execution and the observation condition, the electromyographic activity of the mylohyoid muscle — a muscle involved in mouth opening — was recorded. Both the execution and the observation of the eating action produced a marked increase of mylohyoid muscle activity as early as the ‘reaching’ phase, whereas no mylohyoid muscle activ – ity was recorded during the execution and the observa – tion of the placing action. This indicates that, as soon as the action starts, the entire motor programme for a given action is activated. Interestingly, the observers also seem to have a motor copy of this programme. This ‘intrusion’ allows them to predict what action the agent is going to execute from the first observed motor act and thus to understand the agent’s motor intention. Finally, there is evidence that the mirror mechanism, possibly located in this case in the fronto-mesial areas, also has a role in setting up an anticipatory representation of the motor behaviour of another individual. It has been shown that the ‘Bereitschaftspotential’, an electrophysio – logical marker of the readiness to act 86 , occurs not only when an individual actively performs a motor act, but also when the nature and the onset time of an upcoming action performed by another individual is predictable on the basis of a visual cue 87 . Mirroring intentions and inferring reasons. The studies reviewed above indicate that the parieto-frontal mirror network may subserve the understanding of the motor intention underlying the actions of others. This capacity represents a functional property of the parieto-frontal mirror network that further distinguishes it from those of visual areas. Indeed, it is difficult to imagine how motor intention understanding could be based on visual processing alone, including visual processing that is car – ried out in higher-order visual areas such as the STS. It is true that some STS neurons are selective for a sequence of stimuli. For example, in contrast to classical visual neu – rons that respond to a specific static stimulus, some STS neurons respond to the static view of a body only when this stimulus occurs after a certain movement (for exam – ple, walk and stop) 88 . However, despite this fascinating property, these neurons do not give information about the agent’s motor intention: they describe a given motor act according to a previous motor behaviour, but they do not provide information about the motor intention underlying that motor act. This does not mean that the parieto-frontal mirror mechanism mediates all varieties of intention under – standing. Intention understanding is a multi-layer process involving different levels of action representation, from the motor intention that drives a given chain of motor acts to the propositional attitudes (beliefs, desires and so on) that — at least in humans — can be assumed to explain the observed behaviour in terms of its plausible psychological reasons. w e provide an example to clarify this point. Mary is interacting with an object (for example, a cup). According to how she is grasping the cup, we can understand why she is doing it (for example, to drink from it or to move it). This kind of understanding can be mediated by the parieto-frontal mirror mechanism by virtue of its motor chain organization. However, the mirror mechanism is not able to provide us with the reasons that might underlie the motor intention of Mary (for example, she grasped the cup to drink from it because she was thirsty or because she wanted some caffeine, or she did it to please her friends). u nderstanding the reasons behind an agent’s motor inten – tion requires additional inferential processes 89–91 . Recent empirical data confirmed these considera – tions. They showed that, although the parieto-frontal mirror mechanism is active in all conditions in which the motor task has to be directly understood, when vol – unteers were required to judge the reasons behind the observed actions, there was an activation of a sector of the anterior cingulate cortex and of other areas of the so-called ‘mentalizing network’ 92 . Activation of the same network was also shown in a study that investigated unu – sual actions performed in implausible versus plausible contexts 93 , as well as in a study on the neural basis of reason inference in non-stereotypical actions 94 . As there are different levels of action representation, there should be diverse neural mechanisms subserv – ing these different levels of intention understanding. u nderstanding motor intention relies on the parieto- frontal mirror mechanism and the motor chain organi – zation of the cortical motor system. u nderstanding the reason behind motor intention seems to be localized in cortical areas — the temporal parietal junction and a part of the anterior cingulate gyrus — that have not as yet been shown to have mirror properties. There have been theoretical attempts to integrate these two ways of understanding the intentions of others 95–96 . n onetheless, unlike for the mirror mechanism, there are currently no neurophysiological data that can explain how the ‘mental – izing network’ might work. Conclusions The mirror mechanism is a neurophysiological find – ing that has raised considerable interest over the past few years. It provides a basic mechanism that unifies action production and action observation, allowing the understanding of the actions of others from the inside. Such motor-based understanding seems to be a pri – mary way in which individuals relate to one another, as shown by its presence not only in humans and mon – keys, but also in evolutionarily distant species, such as swamp sparrows 4 and zebra finches 5 . Furthermore, this mechanism indicates the existence of a profound natural link between individuals that is crucial for establishing inter-individual interactions. Finally, preliminary evidence suggests that the impairment of this natural link may be one of the causes of the strik – ing inability of people with autism to relate to other individuals (BOX 3) .