Development of face recognition in an infant gibbon (Hylobates agilis)

Transcription

1 Infant Behavior & Development 24 (2001) Development of face recognition in an infant gibbon (Hylobates agilis) Masako Myowa-Yamakoshi*, Masaki Tomonaga Department of Behavioral and Brain Science, Primate Research Institute, Kyoto University, Inuyama, Aichi , Japan Received 27 April 2000; received in revised form 19 March 2001; accepted 13 June 2001 Abstract The development of the ability to recognize faces was studied in a nursery-reared male infant gibbon (Hylobates agilis). We used traditional and modified head-turning procedures that measured the infant s eye- and head- tracking of moving stimuli. In Experiment 1, the infant was presented with face-like and nonface-like drawings. He showed a preference for face-like stimuli. Experiment 2a tested the infant s recognition of photographs of familiar and unfamiliar faces; by 4 weeks of age, the infant preferred looking at a familiar human face to unfamiliar faces. Experiment 2b investigated the infant s sensitivity in acquiring a preference for faces. The infant was more sensitive to the characteristics of a familiar human face than to those of unfamiliar faces. These findings suggest that there may be similarities between the early face recognition ability of humans and gibbons Elsevier Science Inc. All rights reserved. Keywords: Infant gibbon; Face recognition; CONSPEC/CONLERN; Linear Systems Model 1. Introduction There have been many reports on facial recognition in human infants. Most studies on the development of face/nonface discrimination in human infants, using preference test techniques, indicate that infants show a preference for face patterns at around two months of age (e.g., Maurer & Barrera, 1981). 2-month-old infants looked significantly longer at a proper face-like stimulus than at a scrambled-face, whereas one-month-old infants did not. How- * Corresponding author. Tel.: ; fax: address: myowa@pri.kyoto-u.ac.jp (M. Myowa-Yamakoshi) /01/$ see front matter 2001 Elsevier Science Inc. All rights reserved. PII: S (01)

2 216 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) ever, some studies have found evidence of discrimination between facial images in 1-weekold neonates (e.g., Goren, Sarty & Wu, 1975; Meltzoff & Moore, 1977, 1983; Field, Woodson, Greenberg & Cohen, 1982). Related to this controversy, Johnson, Dziurawiec, Ellis, and Morton (1991) suggested that newborn infants, only minutes old, preferentially track an intact schematic face over a scrambled schematic face. Johnson (1990) proposed that a subcortical visual pathway involving the superior colliculus controls the tracking of moving face stimuli in the first month. Johnson and Morton (1991) applied the term CONSPEC to this primary mechanism. CONSPEC operates from the moment of birth, and its functioning rapidly declines by 1 month. A second mechanism, which they named CONLERN, was thought to be acquired at around 2 months of age. Johnson and Morton (1991) proposed that there was a developmental shift in processing from the subcortical visual pathway to a second mechanism that appears in plastic cortical visual pathways. This mechanism is thought to enable the recognition of characteristic facial expressions. In contrast, the Johnson and Morton model was challenged by studies testing a Linear Systems Model (LSM; Banks & Salapatek, 1981; Easterbrook, Kisilevsky, Hains & Muir, 1999, Kleiner, 1987). The LSM is a mathematical representation of infant visual preference based on the amount of amplitude spectrum and complexity found in stimuli (Kleiner, 1987). According to the LSM, newborns younger than two months old will preferentially look at the most complex stimuli rather than the least complex. In other words, the LSM predicts that infants will prefer the stimulus with the greatest amplitude, regardless of whether the stimulus resembles a face or not. This controversy regarding the onset of the ability to distinguish the features of the face in human infants raises a question about the phylogenic origin of the capacity for facial recognition. Chimpanzees (Pan troglodytes), our closest evolutionary relatives, have much in common with humans, especially during their early stages of life. Some studies suggest that there are similarities between the early competence of human and chimpanzee neonates when measured by the same tests. For example, Myowa (1996) and Myowa-Yamakoshi, Tomonaga, Tanaka and Matsuzawa (submitted) found that chimpanzee neonates could imitate mouth opening, tongue protrusion, and lip protrusion immediately after birth. Bard, Platzman, Lester and Suomi (1992) suggested that neonatal chimpanzees have the capacity for sustained attention to visual stimuli, as measured by the Brazelton Neonatal Behavioral Assessment Scale (NBAS). From birth, chimpanzees responded at a significantly higher rate to social stimuli (e.g., a human face) than to nonsocial stimuli (e.g., a red ball) throughout the neonatal period. Lutz, Lockard, Gunderson and Grant (1998) found that 2.5 to 10.0-week-old (mean age of 6.2 weeks) pigtailed macaques (Macaca nemestrina) demonstrated longer fixation time towards normal conspecific adult faces than towards faces of increasing distortion. Swartz (1983) also found that 2 to 3-month-old pigtailed macaques could discriminate between the faces of three species of adult macaques. In this sense, the early ability to recognize faces is likely a characteristic common to both humans and nonhuman primates. However, few systematic reports have focused on the development of face recognition in nonhuman primates. This paper describes a developmental study concerned with the phylogeny of face

3 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) Fig. 1. The four stimuli used in Experiment 1. recognition. We observed a lesser ape, an infant gibbon (Hylobates agilis), and investigated it s ability to recognize and discriminate between various facial types (i.e., arranged naturally versus scrambled faces; familiar versus unfamiliar faces), thereby providing some evidence concerning the two hypothesized systems, CONSPEC/CONLERN & LSM, proposed to underlie human infant visual preferences. 2. Experiment Method Subject The subject was an infant male gibbon (Hylobates agilis) reared in a nursery from when he was less than two weeks old (13 days after birth) because his biological mother provided inadequate maternal care. He was placed in an incubator and given nursery care by human caregivers. Testing was conducted when he was between 15 to 22 days old, excepting days 19 and Stimuli The stimuli consisted of four schematic faces, shown in Fig. 1. The faces were identical to those used by Johnson and Morton (1991) and measured 13.0 cm tall 10.0 cm wide Procedure Two researchers, A B took part in the experiment. The subject was placed on its back on researcher A s lap on thick towels. To prevent experimenter bias, researcher A was blind to the hypotheses being tested. Researcher B presented the first stimulus, approximately cm from the subject s face, from behind researcher A who was holding the subject. The subject s head was aligned midline and the stimulus was positioned directly in front of his face. As soon as the subject fixated on the stimulus, the stimulus was then moved slowly to one side (right or left) at a rate of approximately nine degrees per second. This procedure was defined as one trial, and

4 218 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) Fig. 2. The percentage of preferential score for each of four stimuli for each day of age in Experiment 1. repeated six times for each side, regardless of whether the subject responded with a head or eye turn of approximately 60 degrees. Thus each session consisted of 12 trials for each stimulus. The infant s eye and head turning responses in pursuit of the stimulus were recorded on videotape for later analysis. We attached a small CCD camera (Sony, CCD-MC100) onto the side of the stimulus, so that the subject s performance could be accordingly recorded. The order of presentation of each stimulus was randomly selected each day Results For each stimulus, the subject s response was scored according to whether the subject fixed the stimulus with a head or eye turn of approximately 60 degrees. The maximum potential score for each stimulus was twelve. Fig. 2 shows the percentage preferential score for each of the four stimuli for each day of age tested. At the beginning of the test (15 days after birth), the subject gazed intensely at face and config. Preferences for inverse and linear were not as strong. Fig. 3 shows the mean percentage of the score for the subject gazing at each stimulus. Kruskal-Wallis s nonparametric ANOVA showed significant differences between the four stimuli, [H (3, N 24) 17.65, p 0.001]. Bonferroni multiple type nonparametric comparison tests revealed significant differences (ps 0.05) between face versus inverse, face versus linear, config versus inverse, and config versus linear. The subject was more attentive to face and config than to inverse and linear.

5 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) Fig. 3. The mean percentage of score (plus standard error) that the subject gazed at each four stimulus in Experiment Discussion These results suggest that a face-like pattern elicited tracking behavior in the infant gibbon subject to a greater extent than did a nonface-like pattern. The infant gibbon was able to discriminate between face-like and nonface-like patterns. These findings are consistent with the CONSPEC/CONLERN theory, and suggest that the infant gibbon may possess a CONSPEC mechanism enabling him to orient preferentially to faces. In contrast, the results of this study did not support the LSM. Although the filtered amplitudes of config and inverse were the same, the gibbon discriminated between the two stimuli. However, we should point out that there is a rule of thumb specifying a 3-to-1 ratio of development in terms of life history in gibbons and humans (Kelly, 1997). That is to say, a 1-month-old gibbon is like a 3-month-old human in terms of development. We began testing our gibbon subject at 15 days of age. This would correspond to less than 1.5 months of age in a human infant. On the other hand, CONSPEC functioning rapidly declines by the age of 1 month (Johnson & Morton, 1991). As our subject had already passed the neonatal stage, we were not able to test CONSPEC in Experiment 1 successfully. It is likely that our gibbon infant subject had already developed the ability to distinguish characteristics of individual faces. The next experiment explored the same subject s ability to discriminate between individual faces.

6 220 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) Experiment 2a Human infants ability to distinguish individual faces increases with age (Johnson, Dziurawiec, Bartrip & Morton, 1992). Could a gibbon reared from birth by humans learn, through extensive experience, to recognize individual facial characteristics of this different species? In order to determine whether the infant gibbon subject would also develop the ability to discriminate between familiar and unfamiliar faces, we conducted another experiment using a two-choice preferential looking paradigm. The subject was presented with two photographs, each showing a different face. The photographs were presented to the left and right of the midline of the subject s face. We speculated that if the subject was able to discriminate the two faces shown on the photographs, he would look preferentially at one photograph rather than the other Method Subject The subject was the same individual as for Experiment 1. We observed the subject from 4 to 5 weeks of age Stimuli The stimuli consisted of three gray-scale photographs: a familiar human (a caregiver), an unfamiliar human (a stranger), and an unfamiliar conspecific (an adult gibbon). The photographs were enlarged to approximately cm and mounted on cardboard; the faces were cropped so that they occupied most of the photograph Procedure A two-choice preferential-looking paradigm was used to determine whether the subject was sensitive to facial characteristics. One session was conducted for each pair of faces: (a) caregiver versus stranger, (b) caregiver versus gibbon, and (c) stranger versus gibbon. As for Experiment 1, there were two researchers. The subject was placed on his back on researcher A s lap, reclining on thick towels. To prevent experimenter bias, researcher A was blind to the hypotheses being tested. Researcher B picked up two of the 3 stimuli and presented them approximately cm from the infant s face, from behind researcher A who held the subject. The subject s head was aligned in the midline and the first pair of stimuli was positioned directly in front of his face. As soon as the subject fixated on the stimuli, one face was moved slowly to the subject s left side and the other to the subject s right side, both at a rate of approximately nine degrees per second. This procedure was repeated five times, regardless of whether the subject responded with a head or eye turn of approximately 60 degrees. The subject participated in 3 sessions 5 trials per day. The infant s eye movement and head turning responses in pursuit of the stimuli were recorded on videotape for later analysis. We attached a small CCD camera (Sony, CCD-MC100) onto the side of each stimulus, so that the subject s perfor-

7 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) mance could be accordingly recorded. The order of presentation of each set of stimuli was randomly selected each day. This test was performed on 4 days of each week Results For each stimulus, the subject s preference was scored according to whether the subject fixated on the stimulus with an eye or head turn of more than approximately 60 degrees. The maximum potential score for each stimulus was five. Fig. 4 shows the developmental change as a percentage of the total preference score for each set of stimuli for each day of age tested. There were no significant differences between the three sets of stimuli during the experimental period (4 to 5 weeks of age). Fig. 5 shows the mean percentage of the preference score for each of the 3 stimuli. The subject looked preferentially at a familiar caregiver rather than at a stranger (binomial test, N 40, p ) or a gibbon (binomial test, N 40, p ). Moreover, there was a significant difference in preference between the stranger and gibbon (binomial test, N 40, p 0.05). By 4 weeks of age, the subject looked preferentially at a familiar human face rather than an unfamiliar face. Additionally, the subject tended to prefer human faces to the gibbon face Discussion The results revealed that by 4 weeks of age, the subject could discriminate between images of a familiar face (caregiver) and an unfamiliar face (stranger and gibbon). Moreover, the subject also discriminated between different unfamiliar individuals faces: the stranger s face and the adult gibbon s face. In general, the human faces elicited greater tracking behavior than did the gibbon face. Maurer and Barrera (1981) suggest that 2-month-old human infants appear to prefer natural configuration of the features of a human face. This observation of human infants applies to the findings of our study on a gibbon infant. As we mentioned in Experiment 1, the 4-week-old gibbon corresponds in terms of development to a 3-month-old human infant. More noteworthy still is that the gibbon had learned to recognize human faces in only two weeks; the gibbon was raised by human caretakers from the age of 2-week-old, and Experiment 2a was conducted when he was 4 weeks old. The gibbon already possessed an ability to discriminate between different individual faces by at least 4 weeks of age. Experiment 2b examined the subject s ability to discriminate between different arrangements of the features of familiar and unfamiliar faces. We predicted that the subject would prove more sensitive to detailed characteristics of a familiar human face than to those of unfamiliar faces. 4. Experiment 2b Our previous results revealed that between the age of 4 5 weeks, the infant gibbon subject could discriminate between a familiar face and unfamiliar faces and between an unfamiliar

8 222 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) Fig. 4. The developmental change as a percentage of total number of preference score for each set of stimulus in each day of age in Experiment 2a. (a) caregiver versus stranger, (b) caregiver versus gibbon, and (c) stranger versus gibbon. human face and a gibbon face. This experiment tested the infant gibbon s sensitivity in acquiring a preference for individual faces. We presented the subject with intact and scrambled faces of the same individuals. The intact and scrambled faces contained the same amount of visual information. The only difference was the location of features within the

9 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) Fig. 5. The mean percentage of preference score (plus standard error) that the subject gazed at each of the three stimuli in Experiment 2a. faces. If the subject can discriminate between different facial characteristics, he should be able to distinguish the intact faces from the scrambled faces. Since the results of Experiment 2a suggest that the infant gibbon is strongly attracted to a familiar face, we would expect to find differences in the sensitivity of preference for the stimuli used in Experiment 2a Method Subject The subject was the same gibbon tested in Experiments 1 and 2a. We observed the subject this time from 5 to 6 weeks of age Stimuli The stimuli were identical to those used in Experiment 2a, i.e., caregiver, stranger, and gibbon, but additionally these stimuli were reproduced as scrambled stimuli (see Fig. 6) making a total of six photographs. Each pair of stimuli consisted of the intact and scrambled photographs of the same individual Procedure The procedure was identical to that used in Experiment 2a. One session was conducted for each pair of stimuli: (a) intact caregiver versus scrambled caregiver, (b) intact stranger versus scrambled stranger, and (c) intact gibbon versus scrambled gibbon. The data were analyzed as in the previous experiment.

10 224 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) Fig. 6. The sample stimuli of the caregiver s faces used in Experiment 2b. (a) intact face and (b) scrambled face Results Fig. 7 shows the mean percentage of the preference scores for the 3 sets of stimuli. In the case of the familiar caregiver s face, the subject looked preferentially at the intact face rather than at the scrambled face (binomial test, N 40, p 0.001). On the other hand, there was Fig. 7. The mean percentage of preference score (plus standard error) that the subject gazed at each of three stimuli in Experiment 2b.

11 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) no significant difference between the intact and scrambled faces of the stranger (binomial test, N 40, p 0.134, n.s.) and gibbon (binomial test, N 40, p 0.318, n.s.). These results revealed that the subject was only sensitive to the difference between a familiar caregiver s intact and scrambled face Discussion The results of this study showed that the male infant gibbon subject was more sensitive to a familiar human face than to unfamiliar faces. He did not show any preference for one unfamiliar face over another. How can we interpret these results? There are two possible interpretations of these results; firstly, although the gibbon subject was able to discriminate perfectly well between intact and scrambled faces, he showed no interest in displaying a preference for one unfamiliar face over another. Secondly, the gibbon subject demonstrated less sensitivity to detailed characteristics of unfamiliar individuals faces than to the familiar human face. In the case of the familiar human face, he showed a preference for the intact face and was able to distinguish detailed characteristics of the familiar human face. On the other hand, he showed no preference for the intact stimulus over the scrambled stimulus for either of the unfamiliar faces. It is likely that both intact and scrambled images of unfamiliar faces are equally novel for the gibbon. Consequently, both the intact and scrambled images of unfamiliar faces might have elicited the same response. There is room for argument on this point. Further research should focus to investigate whether the gibbon is able to discriminate between individual faces irrespective of whether or not they are familiar, by using another test procedure such as the habituation-dishabituation procedure. 5. General Discussion This series of experiments clearly demonstrated that an infant gibbon could distinguish face-like stimuli from nonface-like stimuli from at least 15 days after birth. Human infants are not unique among primates in their ability to pay far more attention to faces than to other stimuli from shortly after birth. This preference can be interpreted as an ability resulting from natural selection. This early adaptive capacity to recognize face-like stimuli enables human and gibbon infants to orient themselves preferentially to the caregiver s face. The results of Experiment 1 demonstrated significantly greater tracking responses towards the face and config stimuli relative to the inverse and linear stimuli. These results support the CONSPEC over the LSM model of infant face recognition. However, our subject was 2 weeks old when we started this series of studies. Moreover, we need to investigate the gibbons capacity for face recognition from very shortly after birth ( true neonates) to reach a true conclusion in respect of the controversy between CONSPEC/CONLERN and LSM theories. There is one other issue that is important in regard to this controversy. Among facial features, the eyes are a particularly significant area of focus for animals. Baron-Cohen (1995) proposed a cognitive and neural module called the eye direction detector (EDD), which

12 226 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) detects the eye of another organism. He suggested that the rapid detection of another organism s eyes has considerable adaptive significance in evolution. Myowa-Yamakoshi and Tomonaga (2001) tested an infant gibbon s sensitivity to the existence of eyes in a face. Fifteen days after birth, the infant gibbon subject looked at an image of a face possessing two eyes preferentially over an image with no-eyes (i.e., possessing only a mouth and a nose). The series of studies by Easterbrook et al. (1999) generally supported the LSM rather than the CONSPEC/CONLERN model of newborn face recognition. However, one of their studies did not support the LSM model. The exception (Study 4) supported the CONSPEC model, since newborn humans tracked the paireyes (not posessing a mouth and a nose) as much as face (containing a mouth and a nose), even though the face had the greatest amount of amplitude information. Future research should examine which facial features are most attractive to human and nonhuman primate infants from both phylogenetic and ontogenetic perspectives. Our results also have implications for the flexibility of early social learning of face recognition. The ability to recognize faces would seem to be an important social and communicative skill with adaptive benefits. From birth, humans engage in the face-to-face mode with their caregivers. Our study showed that a nursery-reared gibbon preferred human faces over a conspecific s face, regardless of whether the human faces were familiar or not. These results suggest that gibbons possess early human-type communicative ability. That is to say that gibbon infants may also engage in face-to-face social interactions. To learn more about the communicative function of early face recognition in gibbons will require longitudinal studies of joint visual-attention, facial imitation and social referencing. Although this study investigated only a single gibbon, it offers the key to an understanding of the missing phylogenic link between cognitive abilities of monkeys and great apes. Nothing is known about the lesser apes early cognitive development. A phylogenic comparison of early facial recognition in a greater diversity of species will help us to understand the meaning and mechanism of this ability found in human infants. Acknowledgments Masako Myowa-Yamakoshi and Masaki Tomonaga, Department of Behavioral and Brain Sciences, Primate Research Institute, Kyoto University, Inuyama, Japan. This research was financed by Grant to G, Hatano and to M. Tomonaga from the Ministry of Education, Science, Sports, and Culture, Japan. Preparation of the manuscript was supported in part by a Research Fellowship of the Japan Society for the Promotion of Science for Young Scientists (No. 3642) to M. Myowa-Yamakoshi. We gratefully acknowledge T. Matsuzawa, G. Hatano, D.M. Fragaszy, T. Waki, K. Chatani, M. Uchikoshi, Y. Kato, and G. Yamakoshi for helpful comments on an earlier draft of this article and generous guidance throughout the project. References Banks, M. S., & Salapatek, P. (1981). Infant pattern vision: A new approach based on the contrast sensitivity function. Journal of Experimental Child Psychology,, 31, 1 45.

13 M. Myowa-Yamakoshi, M. Tomonaga / Infant Behavior & Development 24 (2001) Bard, K. A., Platzman, K. A., Lester, B. M., & Suomi, S. J. (1992). Orientation to social and nonsocial stimuli in neonatal chimpanzees and humans. Infant Behavior and Development, 15, Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory of mind. Cambridge, MA: MIT Press. Easterbrook, M. A., Kisilevsky, B. S., Hains, S. M. J., & Muir, D. W. (1999). Faceness or complexity: Evidence from newborn visual tracking of facelike stimuli. Infant Behavior and Development, 22, Field, T. M., Woodson, R., Greenberg, R., & Cohen, D. (1982). Discrimination and imitation of facial expressions by neonates. Science, 218, Goren, C., Sarty, M., & Wu, P. (1975). Visual following and pattern discrimination of face-like stimuli by newborn infants. Pediatrics, 56, Johnson, M. H. (1990). Cortical maturation and the development of visual attention in early infancy. Journal of Cognitive Neuroscience, 2, Johnson, M. H., Dziurawiec, S., Bartrip, J., & Morton, J. (1992). The effect of movement of internal features on infants preferences for face-like stimuli. Infant Behavior and Development, 15, Johnson, M. H., Dziurawiec, S., Ellis, H., & Morton, J. (1991). Newborns preferential tracking of face-like stimuli and its subsequent decline. Cognition, 40, Johnson, M. H., & Morton, J. (1991). Biology and cognitive development: the case of face recognition. Cambridge: Blackwell. Kelly, J. (1997). Paleobiological and phylogenetic significance of life history in Miocene hominoids. In D. R.Begun, C. V.Ward, & Rose M. D. (Eds.), Function, phylogeny, and fossils. NewYork: Plenum. Kleiner, K. A. (1987). Amplitude and phase spectra as indicates of infants pattern preferences. Infant Behavior and Development, 10, Lutz, C. K., Lockard, J. S., Gunderson, V. M., & Grant, K. S. (1998). Infant monkeys visual responses to drawings of normal and distorted faces. American Journal of Primatology, 44, Maurer, D., & Barrera, M. (1981). Infants perception of natural and distorted arrangements of a schematic face. Child Development, 52, Meltzoff, A. M., & Moore, M. K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, Meltzoff, A. M., & Moore, M. K. (1983). Newborn infants imitate adult facial gestures. Child Development, 54, Myowa, M. (1996). Imitation of facial gestures by an infant chimpanzee. Primates, 37, Myowa-Yamakoshi, M., & Tomonaga, M. (2001). Perceiving eye gaze in an infant gibbon (Hylobates agilis). Psychologia, 44, Myowa-Yamakoshi, M., Tomonaga, M., Tanaka, M., & Matsuzawa, T. Neonatal imitation in chimpanzees (Pan troglodytes). Submitted. Swartz, K. B. (1983). Species discrimination in infant pigtail macaques with pictorial stimuli. Developmental Psychobiology, 16,