The Development of Mimicry

Parents of 162 infants ages 6 to 20 months modeled subsets of four of the same set of eight behaviors, each for a maximum of 3 min, and encouraged their infants to imitate. Proportions of infants producing each behavior (a) when it was modeled and (b) during modeling of a different behavior were compared to estimate the age at which infants mimicked each kind of behavior. No reproduction of these motor acts—that is, no mimicry—was observed at 6 months. Mimicry appeared to develop slowly through most of the 2nd year, emerging at different ages for different behaviors. The findings suggest that newborns’ behavioral matching may not be continuous with mimicry later in infancy. Imitation is probably not one behavioral competency with one underlying mechanism. It is more likely a category of different ways of combining and using different types of knowledge, some of which develop across the first 2 years of life. This article presents a cross-sectional study of the imitative behavior of 162 infants ranging in age from 6 to 20 months. In this study, each infant saw a small number of different behaviors repeated by a parent, each for as long as 3 min, and was encouraged to perform the same actions. The primary measure was the proportion of infants at each age level who matched each behavior. At issue are the questions of when and how the capacity to imitate develops. Imitation is currently a subject of interest to researchers in a wide range of fields, including neuroscience, cognitive science, animal behavior, and robotics, as well as developmental science. Understanding the origins of the imitative capacity, how it works, and how it changes over time would have broad theoretical implications for each of these fields, and possibly practical value for researchers attempting to model or reproduce the capacity to imitate in nonhuman devices. Despite its wide interest and relevance, the development of imitation has not been the subject of a great deal of research. This may be because it is widely accepted among developmentalists that the ability to imitate is innate, and that much of the underlying mechanism is inherited. The basis for this belief is a series of reports, beginning with one by Meltzoff and Moore (1977), that newborn human infants will imitate the behavior of an adult model (see Anisfeld, 1996, and Butterworth, 1999, for reviews). Neonates have been reported to imitate mouth opening, tongue protruding, pouting, sequential finger movements, and head turning (Butterworth, 1999). The evidence is surprisingly limited and has been challenged on both theoretical and empirical grounds (e.g., Abravanel & DeYong, 1997; Anisfeld, 1996, 2005; Anisfeld et al., 2001; Jones, 1996, 2005, 2006). Nonetheless, the claim that newborn infants can imitate is a foundational idea in many theories, both within and beyond psychology (Hurley & Chater, 2005). In studies of infants beyond the newborn period, imitation appears largely as deferred imitation, a measure of memory (e.g., Cuevas, Rovee-Collier, & Learmonth, 2006; Meltzoff, 1988). Deferred imitation differs from the kind of imitation claimed for newborns. Want and Harris (2002) offered the term mimicry for the reproduction of another person’s specific muscle movements, as in newborns’ tongue protruding or mouth opening. In deferred imitation, infants reproduce—after some delay—the effects of another person’s actions on an object, but may or may not use the same movements to produce the same effect. Want and Harris suggested the use of Tomasello’s (1998) term emulation for actions that reproduce the effect or outcome of another person’s behavior. The present study examined the emergence and subsequent development of infants’ ability to reproduce eight motor actions performed by their parents. In four cases, the behaviors modeled for the infants resulted in sounds, so reproduction of those actions might be considered emulation. In the other four cases, however, the specific movement made no sound. This study is unique in focusing on infants’ matching of a standard set of specific motor actions from the middle of the 1st year through most of the 2nd. The developmental course of mimicry beyond the newborn period has been the subject of only Address correspondence to Susan S. Jones, Psychological and Brain Sciences, Indiana University, 1101 E. 10th St., Bloomington, IN 47405, e-mail: [email protected]. PSYCHOLOGICAL SCIENCE Volume 18—Number 7 593 Copyright r 2007 Association for Psychological Science a handful of previous studies, and the focus has been on mimicry of preverbal sounds and words during social interactions (e.g., Kokkinaki, 2003; Kokkinaki & Kugiumutzakis, 2000; Masur & Eichorst, 2002; Papousek & Papousek, 1989). Most mimicry of sounds found in the 1st year has been produced not by infants, but by their adult social partners. For example, Kokkinaki and Kugiumutzakis (2000) recorded interactions between 15 parents and their infants every 2 weeks from 2 to 6 months postpartum. Parents matched their infants’ vocalizations two to three times in each 10-min session. Infants, however, matched their parents only about once in every 20 to 30 min of interaction. It seems likely that this low rate of sound matching by infants was due to chance. In a study supporting this inference, Siegel, Cooper, Morgan, and Brenneise-Sarshad (1990) examined vocal imitation by infants between 9 and 12 months of age and found ‘‘no indications that the children imitated the vocal patterns of their speaking partners’’ (p. 9). Pawlby (1977) studied mimicry of all kinds of behavior in 8 infants weekly from about 4 to 10 months of age. As Kokkinaki and Kugiumutzakis (2000) found, parents imitated their infants at quite high rates—on average about seven times in each 10min session. Infants matched parents’ behaviors of all kinds only once or twice in the same interval. Thus, again, infants’ behavioral matches were infrequent, and may have been chance occurrences rather than mimicry. Interestingly, infants in this study, like those in the study by Siegel et al. (1990), failed to match their mothers’ sounds. Instead, they matched their mothers’ actions on objects: that is, they emulated. Masur and Rodemaker (1999), studying mimicry at 10, 13, 17, and 21 months of age, similarly reported that their infant subjects mostly matched actions on objects until the age of 17 months. One longitudinal case study (Jones & Yoshida, 2006) has tracked the development of mimicry of all sorts in a female infant named Yo. Yo was videotaped weekly from 3 to 16 months of age in 10-min face-to-face interactions with her mother. No objects were used. Initially, the interactions were natural and unconstrained. Later in the study, the mother was encouraged to make direct attempts to get Yo to mimic her behaviors. The resulting developmental sequence was strikingly similar to that observed by Piaget (1945) in his own 3 infants. Specifically, at 3 months, Yo matched nothing—although her mother mimicked Yo a lot. Yo’s first behavioral matches were observed in the second half of her 1st year. The actions matched were behaviors (a) that Yo had spontaneously and repeatedly produced at home, on her own; (b) that Yo’s parents had imitated; and (c) that produced or were accompanied by sounds. For example, when Yo spontaneously raised both arms, her mother said, ‘‘Bonsai!’’ After several sessions, the mother could say, ‘‘Bonsai,’’ while raising her own arms, and Yo would raise her arms. However, for a long time, just the sight of the mother raising her arms was not by itself a sufficient cue for Yo to raise her arms. Thus, the sight of the behavior did not elicit mimicry, but the word cued behavior that could have been mistaken for mimicry. In brief, as Piaget described, sounds appeared to be important learned cues for Yo’s eventual matching of some of her mother’s behaviors. Certainly, Yo matched behaviors cued by sounds before she matched silent behaviors. And Yo matched behaviors she could see herself do long before behaviors she could not see herself do. Yo did not match her mother’s tongue protruding until 14 months of age. The study reported here addresses three questions raised by the observations of Yo’s slow acquisition of mimicry. First, when do infants in general begin to match other people’s behaviors? Because Kokkinaki and Kugiumutzakis (2000) found so little infant behavioral matching in the 1st half-year, the present investigation began with 6-month-olds. Second, what kinds of behaviors do babies match first? Findings from Pawlby (1977), Masur and Rodemaker (1999), and the literature on deferred imitation suggest that babies will emulate before they begin to mimic. However, the literature on newborn imitation puts mimicry first in development (Want & Harris, 2002). Third, is there a special role for sound in initially linking behaviors that infants perform with behaviors that they observe? Infants in previous studies have not mimicked specific sounds. However, Piaget’s (1945) writings suggest that behaviors that produce sounds are likely to be matched first, as the sounds become learned cues for infants’ production of behaviors that incidentally match the behaviors of a social partner.