haut de page




















haut de page























haut de page







mise à jour du
26 sptembre 2002
 Archives Italiennes de biologie
1999; 137; 85-100
 Resonance behaviors and mirror neurons
G Rizzolatti, L Fagiga, L Fogassi, V Gallese
Instituto di fisiologia Universita di Parma, Via Volturno 39, 43100 Parma, Italy
télécharger l'intégralité de ce texte au format pdf
Neurophysiological mechanisms underlying the understanding and imitation of action sur le site Nature Reviews Neuroscience
Cortical mechanisms of human imitation M Iacoboni, G Rizzotatti
Contagious yawning: the role of self-awareness and mental state attribution Platek SMet al
The perception-behavior expressway:automatic effects of social perception on social behavior


Summary : This article is subdivided into two parts. In the first part we review the properties of a particular class of premotor neurons, the "mirror" neurons. With this term we define neurons that discharge both when the monkey makes a particular action and when it observes another individual (monkey or human) making a similar action. The second part is an attempt to give a neurophysiological account of the mechanisms underlying behaviors where an individual reproduces, overtly or internally, movenients or actions made by another individual. We will refer to these behaviors as "resonance behaviors". We distinguish two types of resonance behavior. The first type is characterized by imitation, immediate or with delay, of movements made by other individuals. Examples of resonance behavior of this type are the "imitative" beliaviors observed in birds, young infants and patients with frontal lesions. The second type of resonance behavior is characterized by the occurrence, at the observation of an action, of a neural pattern, which, when internally generated. determines the making of the observed action. In this type of resonance behavior the observed action is, typically, not repeated (overtly). We argue that resonance beliavior of the second type is at the basis of the understanding of actions made by others. At the end of the article we review evidence of mirror mechanisms in humans and discuss their anatomical localizations.


There are many behaviors for which a "mirror" mechanism, similar to that described for F5 neurons, could represent the simplest (and most plausible) neural mechanism. We will refer to these behaviors as "resonance behaviors". We will posit that in resonance behavior a neural activity that is spontaneously generated during movements, gestures, or actions is also elicited when the individual observes another individual making similar movements, gestures and actions.

We will discuss two types of resonance behavior. One is that in which an individual repeats overtly, in a quasi automatic way, a movement made by another individual. The second is that in which an individual uses an internal motor template to repeat internally the actions made by others. This internal resonance may be used for an overt action, but its fundamental role is that of recognizing the observed action.

The two resonance behaviors may interact. For example, an individual may understand the goal of an action and try to achieve that goal. This can be obtained by repeating (or trying to repeat) the same movements that the actor of the action made or by making other movements different from those employed by the actor. In the present article we will not deal with these more complex cases, but we will focus on the two resonance behaviors defined above. We will refer to them as resonance behavior of the first type and resonance behavior of second type, respectively.

Resonance behavior of the first type

We define as resonance behavior of the first type the tendency that individuals have to reproduce, immediately or with some delay, movements, gestures or actions made by another individual. The repeated actions and the conditions in which they are repeated vary very much. Yet, in spite of this heterogeneity, we propose that in all of them the basic mechanism is an activation of neurons that generate motor actions identical to those observed.

A typical example of a resonance behavior of the first type is the "imitative" behavior observed in many species of animals when one or a few of them start an action. One of the best studied examples is, probably, the behavior displayed by shore birds when alarmed. Typically, one or few birds start wing flapping, then others repeat it and, eventually, the whole flock turns in flight (42, 43). This "contagious" behavior does not require, necessarily, an "understanding" of the action. What is important here is only that the action emitted by the first bird could act as a "release" signal (43). A "resonance" of the motor system of the observing individual would be a simple and very advantageous mechanism for implementing this behavior.

Another example of the resonance behavior of the first type is represented by the capacity that some birds have to repeat the songs of conspecifics. There is convincing evidence that the neural mechanism at the basis of it is represented by neurons that discharge both when the bird produces a song and when it hears it (see 17 for review).

Resonance behavior of the first type appears to be present also in humans, in infants where it plays a fundamental role in establishing a communication with adults, and in adults as well.

An example of resonance behavior in infants is the capacity that even very young infants have of imitating buccal and manual gestures (25). It is hard to think that, at this early age, there is an understanding of the meaning of the observed gesture and a subsequent conscious desire to repeat it. A "resonance" mechanism of the first type appears to be the mechanism most likely underlying the phenomenon. This explanation appears particularly convincing in the case of facial gestures that the infant is able to imitate, in spite of the tact that it has never seen its own face.

There is, however, an important difference between the infant behavior and the "released" behavior of birds described above. As shown by Meltzoff and Moore (25; see also 26), when the infant response is artificially delayed, the behavior does not disappear as it should if it was simply a matter of response release, but is emitted subsequently when the response becomes possible. This difference is probably related to the presence in humans (as well as in most evolved species of animals) of mechanisms (see below) storing the externally evoked response and controlling its emission. Although these control mechanisms are not mature in infants - typically adults do not repeat overtly the observed gestures - still they appear to be already present in infants and allow a storage of the response and its delayed repetition.

As far as the adults are concerned, one has to distinguish actions related to emotional and vegetative life ("hot actions") from actions in which these components are limited or absent ("cold actions"). While adults usually do not repeat cold actions, an "imitation" frequently occurs in the case of hot actions. Smiling produces a tendency to respond with smiling. Similarly, adults and children alike respond to the sight of an individual yawning by yawning themselves. Laughing is contagious. For all these actions there is no need to postulate a comprehension of the observed actions. The observed action simply releases in the observers the seen action. The term "response facilitation", proposed by Byrne (3) describes this behavior very well.

In contrast to normal adults, the repetition of "cold" action occurs in some severely demented patients. This behavior was named echopraxia. It is described as follows: (echopraxia) "is an impulsive or automatic imitation of other's people gestures, an imitation which is performed immediately with abruptness and speed of a reflex action. Irrespective of whether the gesture is natural or bizarre, helpful or dangerous, it is invariably reproduced" (7, 40).

It is likely that echopractic behavior represent a "release" of a covert resonance phenomenon of the first type present also in normal subjects, but inhibited in its expression by frontal and mesial cortical areas (21). Evidence that resonance phenomenon of the first type is present in normal subjects is provided by experiments in which evoked potentials were recorded from various arm muscles in normal subjects while they were observing hand and arm movements performed by an experimenter in front of them (8). The results showed a selective increase of motor evoked potentials in those muscles that the subjects normally use for producing the observed movements. The resonance phenomenon was present not only during observation of goal directed hand movements, but also during the observation of meaningless arm movements. These findings clearly show that the motor system "resonate" also in adult normal subjects, although the resonance is not sufficient to produce overt movements.

Another phenomenon due to the release of a cortical inhibition is the imitation behavior (21). This phenomenon, which is observed in frontal patients, especially in the case of fronto-orbital lesions, does not appear, however, to belong to resonance behavior of the first type. Unlike in echopraxia, patients with imitation behavior do not imitate the movements of the acting individual, but rather perform an action identical to the observed one. It is the goal rather than movements that is imitated in this pathology.

Resonance behavior of the second type

We define resonance behavior of the second type the activation of neurons coding motor actions during observation of similar actions made by others. While resonance behavior of the first type may be easily observed in humans and animals, the existence of a resonance phenomenon of the second type was the unexpected result of neurophysiological studies of area F5 of the monkey. The resonance behavior of the second type, although based as the first type on the activation of motor system in response to an observed action, differs from it for many important aspects.

First, unlike in the resonance behavior of the first type, the effect of the neural activity elicited by the observation of an action is not that of generating an overt motor response. A monkey looks at the action, and while looking at it, in its brain there is a motor replica of it. Yet the monkey does not repeat the seen action.

In an experiment we placed a second monkey in front of that from which action potentials were recorded. We then gave food to the newcomer taking it from a container. In this condition there was no obvious reason for the observing monkey (that from which we recorded) to repeat the gestures of the newcomer. The appropriate response was to jump on it and chase it away. Yet, the F5 mirror neurons fired any time the newcomer grasped the food (34).

Second, if the aim of the resonance phenomenon in F5 were that of allowing the observing individual to repeat the observed gesture, there ought to be a good motor matching between the observed action and the one to be repeated. The infants studied by Meltzoff and Moore protrude their tongue in the same direction as the experimenter does (25). The same is true for other instances of resonance behavior of the first type. In contrast, in F5 action generalization characterizes the visual responses of most neurons. Neurons respond regardless of whether the hand that grasps food is seen moving toward the monkey or away to it, of whether the movement is made from above the object or from below. In some neurons even grasping with the mouth is effective. From a motor point of view all these movements are different, but in terms of meaning they all represent the same action, "grasping".

Third, in many neurons the congruence between the effective observed action and the effective executed action is broad. A neuron that discharges during a specific hand action made by the monkey, e.g. a precision grip, fires not only when the monkey observes the experimenter grasping an object using a precision grip, but also when it observes the experimenter grasping a larger object using ail fingers (11). Neurons with these properties hardly could bc the basis for imitation behavior.

On the basis of these considerations we suggested that F5 mirror neurons are involved not in "imitation" but in action understanding (11, 34). The logic is the following. An individual that emits a movement typically "knows" (predicts) its consequences. This knowledge probably results from an association between the representation of the motor action, coded in F5 and in other motor centers, and the consequences of the action. The "resonance" mechanism in F5 does not determine the appearance of a motor response, but evokes a neural activity that corresponds to that which, when internally generated, represents a certain action. The meaning of an action can be therefore recognized, because of the similarity between the two representations.

This interpretation implies that, unlike the resonance activity of the first type the purpose of which is to determine overt movements, the purpose of the "resonance" in the mirror system is to generate a representation of the goal of an action. The properties of F5, or at least of some neurons of F5 (sec above), have precisely these characteristics. Note that the capacity to generate a goal-directed representation of illovement is present not only in F5 mirror neurons, but also in F5 canonical neurons. This indicates an evolution of the monkey motor cortex from a purely executive system. in which sensory input is hooked up directly to the output systems, to a systern in which part of it acts as a buffer storing the possible actions evoked by the external stimuli.

It is important to make it clear that we do not claim that F5 mirror neurons are exclusively involved in a resonance behavior of the second type. It may well be that in monkey the two resonance levels are not anatomically segregated. It could be, for example, that the mirror neurons that we classified as highly congruent, i.e. those that resonate only when the observed action coincide with the emitted one, underlie resonance activity of the first type, while those broadly tuned are responsible for action comprchension. Alternatively, il may bc that motor areas differeni from F5 are responsible for resonance activity of the first type. Finally, one cannot exclude that resonance activity of the first type concerns only socially relevant behavior. Since monkeys do not communicate using hands, it is possible that this type of behavior is limited in this species to facial or body movements and therefore does not concern hand movements.

....... télécharger l'intégralité de ce texte