haut de page

mise à jour du
30 septembre 2002
1977; 198; 4312; 74-78
Imitation of Facial and Manual Gestures by Human Neonates
Meltzoff AN, Moore MK
Department of experimental Psychology, Oxford University, England
The perception-behavior expressway:automatic effects of social perception on social behavior


Abstract : Infants between 12 and 21 days of age can imitate both facial and manual gestures; this behavior cannot be explained in terms of either conditioning or innate releasing mechanisms. Such imitation implies that human neonales can equate their own unseen behaviors with gestures they see others perforin.

Piaget and other students of developmental psychology consider the imitation of facial gestures to be a landmark achievement in infant development. Infants are thought to pass this milestone at approximately 8 to 12 months of age. Infants younger than this have been postulated to lack the perceptual-cognitive sophistication necessary to, match a gesture they see with a gesture of their own which they cannot see. The experiments we report show that the infant's imitative competence has been underestimated. We find that 12- to 21-day-old infants can imitate both facial and manual gestures (Fig). This result has implications for our conception of innate human abilities and for theories of social and cognitive development.


An experimental evaluation of the neonate's imitative competence raises several methodological difficulties. One consists of distinguishing true imitation from a global arousal response. For example, one can conclude nothing about imitation if an infant produces more tongue protrusions in response to a tongue protrusion demonstration than he does to the presentation of a neutral facial expression. It would be more parsimonious simply to conclude that a moving, human face is arousing for the infant and that increased oral activity is part of the infant's arousal response. A second issue involves , controlling interactions between aduit and infant that might shape the imitative response. We found that if parents were informed of the imitative tasks we planned to examine, they practiced these gestures with their infants before coming into the laboratory so that their baby "would do well on the test." In reviewing films of preliminary work, we also noticed that the examiner tended to alter the rhythm of his tongue protrusion as a function of the response of the infant. These kinds of interactions would expose findings of imitation to a variety of explanations, including the possibility that the infants were merely being conditioned to imitate tongue protrusion. A third issue concerns the scoring of the infant's responses. The movements tested were, not generally produced in a discrete, unambiguous fashion and not surprisingly, there were'gross differences in the scoring as a function of whether or not the observer knew which gesture had been demonstrated to the infant.

In the experiments we now report, these three issues are addressed as follows. Each infant's response to one gesture is compared to his response to another similar gesture demonstrated by the same adult, at the same distance from the infant, and at the same rate of movement. For instance, we test whether infants produce more tongue protrusions after an adult demonstrates tongue protrusion than after the same adult demonstrates mouth opening, and vice versa. If differential imitation occurs, it cannot be attributed to a mere arousal of oral activity by a dynamic, human face. Parents were not told that we were examining imitation until after the studies were completed; moreover, the experiments were designed to preclude the possibility that the experimenter might alter the rhythm of his demonstration as a function of the infant's response. The infant's reactions were videotaped and then scored by observers who were uninformed of the gesture shown to the infant they were scoring. [...]

At least three different mechanisms could potentially underlie the imitation we report.

1) It could be argued that the imitation is based on reinforcement administered by either the experimenter or the parents. In order to prevent the experimenter from shaping the infant's imitative responding, the procedure directed that he maintain an unreactive, neutral face during the response period. The experimenter's face was videotaped throughout both experiments in order to evaluate whether this procedure was followed. The videotaped segments were shown to observers whose task it was to score any reinforcements that the experimenter administered. No smiles or vocalizations were noted in any trial. Indeed, the only changes from the passive face occurred in three trials in experiment 1, when the experimenter was judged to "blink extremely rapidly." Considering only experiment 2, then, the experimental procedure does not appear to have been violated, and therefore, differential shaping of the mouth-opening and tongue-protrusion responses during the successive 150-second response periods is an unlikely source of the effects obtained. Since none of the parents were informed about the nature of the study, special practice on imitative tasks at home in preparation for the experiment was avoided. Further, informal questioning revealed that no parent was aware of ever having seen babies imitating in the first 21 days of life; indeed, most were astonished at the idea. Thus, a history of parental reinforcement seems an improbable basis for imitation at this very early age.

2) This early imitation might be based on an innate releasing mechanism such as that described by Lorenz and Tinbergen (6). This view would hold that tongue protrusion, mouth opening, lip protrusion, and sequential finger movement are each fixed-action patterns and that each is released by the corresponding adult gesture (sign stimulus). The overall organization of the infant's imitative response, particularly its lack of stereotypy, does not favor this interpretation. In addition, the fact that infants imitate not one, but four different gestures, renders this approach unwieldy.

3) The hypothesis we favor is that this imitation is based on the neonate's capacity to represent visually and proprioceptively perceived information in a form common to both modalities. The infant could thus compare the sensory information from his own unseen motor behavior to a "supramodal" representation of the visually perceived gesture and construct the match required. In brief, we hypothesize that the imitative responses observed are not innately organized and "released," but are accomplished through an active matching process and mediated by an abstract representational system. Our recent observations of facial imitation in six newborns-one only 60 minutes old-suggest to us that the ability to use intermodal equivalences is an innate ability of humans. If this is so, we must revise our current conceptions of infancy; which hold that such a capacity is the product of many months of postnatal development. The ability to act on the basis of an abstract representation of a perceptually absent stimulus becomes the starting point for psychological development in infancy and not its culmination.