In 1992, Giaccomo Rizzolatti and his team at Parma were investigating an area called F5 in the macaque brain. F5 is a premotor area; that means it is involved in the organisation of behavior, though the activity you find there is not directly translated into motor commands (the motor cortex does that). The Parma neuroscientists were measuring, with electrodes stuck in the macaque's brain, the electrical activity of F5 neurons. The activity in a premotor area correlates with the animal's gesture, so when the animal picks a nut, neurons in F5 fire (they send an electric charge to other neurons).

But some neurons in F5 fired not only when the monkey made a gesture, but also when he saw a gesture. This was very surprising, because neurons in premotor areas normally don't react to visual stimuli (other areas more on the rear do that). They were visuo-motor neurons. But "visuo-motor neurons" is not such a sexy label, is it? Rizzolatti decided to cal them "Mirror-Neurons". Here's why:

Rizzolatti's assistants had remarked that, among the "Mirror-Neurons", some (but not all; a third approximately) responded to a very precise kind of gesture. For example, the neuron fired when the monkey picked a nut, and when he saw someone pick a nut, and nothing else could make it fire. The Parma group hypothesized that the monkey was mentally repeating the gesture he saw, that he was involved in a covert form of imitation. The term "Mirror-Neurons" is a nice label for that hypothesis, but mind you, it is still an hypothesis. No one really knows whether monkeys are really mirroring anything. They could just be sorting the gesturezs they see into categories ("now, that's a hand grasping a peanut").

Later, the same Parma team showed that some mirror-neurons were responding precisely to targeted-gestures, that is, gestures with an object. A hand grasping a nut, a hand tearing paper, sets the Mirror-Neurons to fire, but the same hand performing the same gestures in a void provokes no reaction whatsoever. One could sum up this finding by saying that Mirror-Neurons respond to action, not to mere movement. Still in macaques, mirror-neurons were evidenced that also reacted to the sound of an action; they were called "audiovisual" Mirror-Neurons. To be exhaustive, one must cite canonical neurons: these neurons fire both when a monkey grasps an object, and when he sees the same graspable object. In jargon, the things you can do with an object are called the affordances of that object; one might say that canonical neurons react to affordances.

F5 is thought by some scientists to be the evolutionary homologous of the premotor ventral cortex in humans. It means that our last common ancestor with the macaque had a brain area that evolved to become F5 in macaques and the premotor ventral cortex in us. This possibility raised enthusiasm among many researchers, because the premotor ventral cortex (most often the left one) hoists Broca's area. Broca's area is known since the XIXth century to be a key area for speech and language. Some scientists have advanced the hypothesis that language might have evolved from something that resembled mirror-neurons, though no one can really say what.

For obvious ethical reasons, we can't open a human skull to stick an electrode in it, so no mirror-neuron has ever been directly measured in humans. Nevertheless, ventral premotor cortex activity, as measured with fMRI scanners and electroencephalography, seems to react in many ways similarly to F5. It has the same mirror and canonical properties. There is a good chance that the premotor ventral shelters mirror and canonical neurons.

Many a form of human behavior has been attributed to mirror-neurons: imitation (though macaques are very very poor at imitating), empathy, social understanding, Theory of Mind, the human capacity for culture, and even, recently, the widespread human taste for porn. The fact that mirror-neuron activity is impaired in autistic children fueled the speculation about the importance of mirror-neurons for social cognition.

It is not possible that such a small bunch of neurons be involved in all this; moreover, one must remember that Mirror-Neurons are possessed by macaques, who communicate poorly, if ever, hardly imitate each other, and are not especially keen on sharing the pain of others. Deflationist interpretation of Mirror-Neurons have recently been put forward; their main spokesman is Hungarian developmental psychologist Gergely Csibra. He thinks Mirror-Neurons are involved in preparing one's actions and anticipating the actions of others, not in mirroring them (which, when you think about it, is useless if you have no intention of imitating them, and if you have no interest in the precise workings of other people's gestures - and that is most often the case among macaques).

Links:

A very entertaining 10-minutes movie on the PBS website

V.S.Ramachandran explains why he thinks Mirror-Neurons are the greatest single scientific discovery of the late XXth century here.

An on-line colloquium discussing the relevance of Mirror-Neurons to philosophy and the social sciences: What do Mirror-Neurons mean?