PFF, EV, LF, and SJS conceived and designed the experiments. PFF, EV, and AR performed the experiments. PFF and AP analyzed the data. PFF, EV, and SJS wrote the paper.
The authors have declared that no competing interests exist.
The emergence of social behaviors early in life is likely crucial for the development of mother–infant relationships. Some of these behaviors, such as the capacity of neonates to imitate adult facial movements, were previously thought to be limited to humans and perhaps the ape lineage. Here we report the behavioral responses of infant rhesus macaques
This manuscript provides the first quantitative description of neonatal imitation in a nonhuman primate, indicating imitative capacities are not unique to the ape and human lineage, contrary to what was previously thought.
Matching one's own behavior with that of others allows individuals to detect contingencies in the social world. This process could allow an individual to synchronize its activity with those of its group members, to copy the behavior of other individuals, and to learn the context in which an activity should be performed [
To date, studies of early signs of this matching capacity have been largely limited to human infants. Almost 30 years ago, Meltzoff and Moore [
We know very little about the evolutionary origin of this capacity. Recently, Matsuzawa and colleagues studied neonatal imitation in two infant chimpanzees
Although neonatal imitation in chimpanzees, and especially in other ape species, requires further investigation, it seems that this phenomenon has features similar to human neonatal imitation, both in terms of timing and type of imitated gesture. This observation is congruent to the finding that humans and apes appear better endowed for imitation than are other primate species [
To pursue this goal, we investigated the presence of neonatal imitation in rhesus macaques, an Old World monkey species that diverged from the human lineage about 25 million y ago [
We tested 21 infant rhesus macaques at ages of 1, 3, 7, and 14 d. Infants were tested once a day in six different conditions (
Figures on the left represent stimuli during resting conditions and baseline. Figures on the right depict the stimuli when fully expressed. In the DISK condition, the disk was repeatedly rotated 90° clockwise and counterclockwise.
We videotaped the infants' behavior in each condition and analyzed whether their behavior matched the stimulus. We compared the frequency of matched behaviors in the baseline and stimulus periods and the frequency of the matched behaviors in each specific condition with the corresponding behavior in the biological control condition (EYE).
During both baseline and stimulus periods, the amount of attention paid to the experimenter face/stimulus during the presentation of biological (mouth, tongue, eyes, and hand) and nonbiological stimuli (disk) did not differ among conditions. In general, the infants looked more at the stimuli during the stimulus period than at the baseline (
Asterisks (*) indicate a significant increase in number of looks (stimulus versus baseline) for a specific stimulus (at least
Averaged scores are calculated as the difference between the frequency of the imitated behavior in the stimulus period and the baseline. The scores the infants obtained are reported in relation to age and to the different experimental conditions (MO, LPS, TP, and HO). Scores are ± standard error of the mean.
On day 1, the frequency of MOs made by infant macaques was very low or absent (mean number of MOs during stimulus period in the different conditions were 0.33 in MO, 0.41 in TP, 0.47 in LPS, 0.13 in HO, and 0.07 in EYE). In contrast, high rhythmic mouth openings/closures (defined as LPS) were frequent. On day 1, in the MO condition, the frequency of MOs in the stimulus period was not different from baseline, although the frequency of LPS (in the MO condition) was significantly higher (
On day 3, the frequency of TPs in the TP condition and of LPS in the LPS condition were significantly higher during the stimulus period than during baseline (
MO is shown on the left; TP on the right. Figures were taken from
Averaged scores are calculated as the difference between the frequency of the imitated behavior in the stimulus period and the baseline. The asterisk (*) indicates a significant difference of LPS score in the MO compared with TP and HO conditions. Pound symbols (#) indicate a significant difference of the behavioral score recorded in that condition compared with all the other conditions. Scores are ± standard error of the mean.
In the LPS condition, the frequencies of HOs, MOs, and TPs did not increase in the stimulus period compared to the baseline. In the TP condition, the frequencies of HOs, MOs, and LPSs did not increase in the stimulus period compared to the baseline.
On day 7, there was a tendency to perform more LPS in the LPS condition than in the EYE condition (
Our findings show that 3-d-old macaques imitate LPS and TP when seeing these gestures being performed by a human experimenter. On day 1, the MO stimulus elicited a significantly higher frequency of MOs in terms of lip smacks (repeated MO) but not in terms of the exactly matched behavior (a single MO). Thus, infants matched the type of behavior in the form but not in the pattern (repetition of mouth opening/closure). This finding could be attributed to several factors that are not mutually exclusive. First, the frequency of infants' spontaneous MOs was very low or virtually absent not only on day 1 but also on any other testing day. In contrast, as shown by our data, LPS is much more frequently displayed soon after birth than is MO, and it could be considered an easier behavior to match the MO of the model. Second, the visual system of the infant is not fully developed immediately after birth, and thus the model's MO might provide a much more visible and salient stimulus than LPS because, although both share some visible features, MO (contrary to LPS) involves a wide opening of the mouth. Thus, the infant could recognize the model's MO as a form of LPS behavior and, consequently, might respond to it.
Our findings cannot be interpreted in terms of a general, nonspecific arousal response of the infant to the observation of mouth or hand gestures for the following two reasons: (i) because the increase of a specific behavior was recorded only in the matching condition; i.e., TP increased only in the TP condition; and (ii) because we did not find increased frequencies of all the behaviors, regardless of whether they matched that performed by the model, as a generic arousal model will predict.
By day 7, neonatal imitation had largely disappeared, although some individuals (four out of 12) still matched the LPS. The infants were attentive to all the stimuli; nevertheless, only a few stimuli elicited neonatal imitation and only during the very first days of life. The other stimuli (i.e., hand, eyes, and disk) elicited the infants' interest but did not produce any specific change in the infants' responses. Thus, the mouth and the tongue appeared to be the only effective stimuli among those tested in producing an imitative response in these macaque infants. The lack of neonatal imitation of hand gestures reported here is in agreement with what has been found in chimpanzees [
Environmental rearing conditions and the unnatural source of stimuli might account for the limited number of gestures matched and the short time course in which neonatal imitation was observed. It is possible that infants that are separated from the mother at birth lack the rich social input required to adequately respond to gestures and to maintain such responses over time. Moreover, the biological stimuli provided by the experimenter were most likely less salient for monkeys than for those routinely provided by conspecifics (i.e., the mother or group members). These factors could have reduced the effectiveness of the stimulus and, consequently, the amplitude and time course of the imitative response. In addition, because the infant was not emotionally attached to the experimenter, the possible functional meaning of neonatal imitation might have been masked or could not emerge in its complexity.
Neonatal imitation in humans shows great interindividual variation [
One of the main differences between the conclusions of our study and the other primate studies is the temporal window in which neonatal imitation was observed. In contrast to humans and chimpanzees, our infant macaques showed the phenomenon for only a few days after birth. As mentioned above, some individuals still displayed imitation of LPS at day 7, but not beyond that. How can such species differences be explained? Motor and cognitive development in macaques is much more rapid in macaques than in humans and chimpanzees [
Another important aspect that emerged from our study was the marked interindividual variability in neonatal imitative abilities. Attention paid to the stimulus was not predictive of matching the gesture. Some infants consistently imitated the model's gestures, whereas others did not imitate at all at any age. Interindividual variability cannot be attributed to environmental factors, because the housing and timing of testing were identical for all subjects. Rather, it might be related to differences in temperament that predisposes the sensorimotor system to be differently sensitive and reactive to external social events. However, we cannot draw any conclusion from our data on possible relations between predispositions of the sensorimotor system and stable individual traits. Longitudinal long-term studies may help in clarifying this possible relation. Finally, infants who imitated one specific gesture were not necessarily the same individuals who imitated the other gestures. Thus, the capacity to respond to the model may not reflect a general imitative skill but rather a sensorimotor sensitivity tuned to specific facial features.
A traditional notion in primate behavior is that apes imitate, and monkeys do not [
Several hypotheses have been put forward to identify which mechanisms might underlie neonatal imitation [
Meltzoff and Moore [
Investigations of the ontogeny of communicative gestures in macaques showed that LPS begins to develop in the first few days of life [
Subjects were 21 infant rhesus macaques
All animals were provided with a 50:50 mixture of Similac (Ross Laboratories, Columbus, Ohio, United States) and Rimilac (Bio-Serv, Frenchtown, New Jersey, United States) formulas. They were hand-fed until they were old enough to feed independently, usually by day 4. Formula was administered ad libitum until 4 mo of age. Purina High Protein Monkey Chow (#5038) (Purina, St. Louis, Missouri, United States) and water were available ad libitum when nursery-reared animals reached 1 mo of age.
All testing was conducted in accordance with regulations governing the care and use of laboratory animals and had prior approval from the Institutional Animal Care and Use Committee of the National Institute of Child Health and Human Development.
Subjects were tested at ages of 1, 3, 7, 14, and 30 d or, due to experimental constraints, within 1 d before or after these days. Early in the study, we found that by day 30, infants were highly mobile and difficult to hold for more than few seconds. For this reason, we abandoned the day 30 testing. Seven infants were tested at all four remaining ages, seven at only three different ages, one at two different ages (days 1 and 3), and five at one age (
Three experimenters were involved in the data collection. One experimenter held the infant monkey in his/her hands, the second (the demonstrator) served as the source of stimuli, and the third videotaped the experiment and informed the demonstrator of the correct sequence of stimuli. Two video cameras (Panasonic VHS, Panasonic, Secausus, New Jersey, United States), and Sony digital, Sony, Tokyo, Japan; positioned 1.5 m lateral to the monkey) recorded the experiment. One video camera recorded both the experimenter and the infant in side view; the other recorded solely the subject's entire body from the other side (at about 120° angle from the other camera).
Each test session included six different conditions (
Most of the tapes (80%) were digitally analyzed by two coders not blind to the experimental condition using all occurrence sampling for all behaviors listed below. Reliability between the two coders was very high (Cohen's kappa = 0.95). The analysis was not blind, to allow the coders to score the infants' behavior in relation to the beginning of each period, which was aligned to the time point in which the stimulus appeared on the screen, started to move (stimulus period), and ceased to move (post-stimulus period). However, to ensure the reliability of this procedure, 20% of sessions were coded with the human model covered on the screen so that the scorer was blind to the experimental condition. Reliability between the two coders was still high (Cohen's kappa = 0.86). The outcomes of these sessions were compared with the analysis of the same sessions in which the scorer was not blind to the experimental condition. Consistency between the blind and nonblind coding was very high (Pearson correlation:
The following behaviors were scored for analysis: (i) Attention to the model (LOOK). The monkey orients and looks at the stimulus (neutral face during baseline and post-stimulus, stimulus during stimulus presentation). Looking could vary from brief scans to extended visual contact for several seconds. Each look at the model was counted as one occurrence of LOOK. (ii) LPS. The mouth is opened and closed quickly. The mouth is not opened to its full extent (but generally to one-third). LPS may be combined with TP. Each opening of the mouth was counted as one occurrence of LPS. Occurrences of TP were scored separately. (iii) MO. The mouth is opened for at least half of its total opening span, and usually only once. MO is performed more slowly than LPS, and the mouth is maintained open for a slightly longer period. Each opening of the mouth was counted as one occurrence of MO. TPs could occur in combination with MO. Occurrences of TP were scored separately. (iv) TP. Forward movements of the tongue so that it crosses the inner edge of the lower lip. Each thrust was scored as one occurrence of TP. (v) HO. Opening and closing of a hand without arm movements. Generally, fingers are tightened around support (usually fleece fabric) with a whole hand grip. Each opening and closing of one hand was scored as one occurrence of HO. (vi) Move arm and grasp (MOVE-HO). Grip is released from support, arm moves toward another area on support, and the support is gripped again. Each of these sequences was counted as one occurrence of MOVE-HO.
Wilcoxon paired tests were used to compare the amount of attention (LOOK) that the infant paid to the biological (face or hand) or to the nonbiological stimulus (disk) during the baseline and the stimulus periods. For each animal, scores obtained in the three facial conditions (LPS, TP, and MO) were averaged.
In each condition, we assessed whether the monkeys' behavior that matched the behavior provided by the experimenter (target behavior) was performed by the infant with higher frequency during stimulus periods than during baseline. For this purpose, we compared the frequency of each behavior displayed during the stimulus period with that displayed during baseline. The frequency of each behavior in the stimulus and baseline periods were compared with Wilcoxon paired tests.
To compare the frequency of the matched behavior in a condition with that scored in the control, we calculated for each infant the difference in frequency between the matched behavior displayed during the stimulus period and the baseline period. A negative score indicated that a behavior was observed more frequently during baseline; a positive score indicated that a behavior was observed more frequently during the stimulus period. Wilcoxon tests were used to compare the score for each matched behavior displayed in a specific condition with the score of the same behavior displayed in the control condition (EYE condition). We ran the same comparison between each stimulus period and the same behavior displayed in the nonbiological control (DISK condition). To exclude that the frequency of the matched behavior could increase as a consequence of neonate general arousal for seeing a specific mouth or hand movement, we compared the score of each matched behavior displayed in a specific condition with that obtained in the other conditions (Wilcoxon paired tests).
This video illustrates a 3-d-old infant male macaque responding to the experimenter mouth gesture.
(3.3 MB AVI)
This video illustrates a 3-d-old infant female macaque responding to the experimenter's TP.
(4.0 MB AVI)
This video was taken at the field station in Poolesville, Maryland, United States (National Institute of Child Health and Human Development). It depicts a face-to-face mother–infant LPS exchange with the mother initiating the interaction. The infant is less than 10 d old.
(2.5 MB AVI)
We thank Giacomo Rizzolatti, Vittorio Gallese, Arthur Glenberg, and Elsa Addessi for their valuable comments on an early draft of the manuscript.
eyes opening
hand opening
lip smacking
mouth opening
tongue protrusion