Empirical Article
Constraints on the Timing of Infant Cognitive Change: Domain-Specific or Domain-General? Annette Karmiloff-Smith, Gisa Aschersleben, Scania de Schonen, Mayada Elsabbagh, Annette Hohenberger, and Josette Serres Most studies of infant cognition focus on group data from single domains. Yet, without the multidomain testing of the same infants longitudinally, such data cannot be used to evaluate whether the timing of cognitive change occurs in a domain-general or a domain-specific way. We present the results of a longitudinal study pooling data from three European laboratories set up identically. Over 100 healthy, monolingual infants each underwent multi-domain testing at 6 and again at 10 months in six experimental tasks (speech processing, face processing, and action/event processing), as well as a videotaped 3-minute recording of mother/infant dyads in a play session with an identical set of toys. Previous research examined the effects of maternal sensitivity only on general intelligence measures, but our approach is novel in that it assessed dyadic effects on specific cognitive domains, attempting to pinpoint in finer detail the effects of mother-infant dyadic interaction on the timing of cognitive change. Our findings highlight the importance of a multi-domain approach, in that unlike the assumptions drawn from cross-sectional data, our longitudinal study yielded different developmental timing across domains within the same infants. Our results also highlight a crucial difference: at the group level 6- and 10-month-olds display the expected effects found in previous research, but when re-analysed according to mother-child interaction ratings, the quality of dyadic interaction style turned out to subtly foster or delay development in domainspecific and age-specific ways, contributing to the range of individual differences in timing that we observe in cognitive development over the first year of life. Keywords: infancy, cognitive change, domain-specific, domain-general, mother-child interaction
Introduction
Numerous cross-sectional studies of infant development reveal significant differences in group data at different ages. Yet often camouflaged in these group data are substantial individual differences. For instance, studies may reveal that 6-month olds fail and 10-month olds succeed on a particular task at the group level (e.g., Behne, Carpenter, Call, & Tomasello, 2005; Branon, Lutz, & Cordes, 2006; Lipton & Spelke, 2004; Wood & Spelke, 2005). However, if one inspects individual data trends for a given task, then often some of the infants in the younger group already succeed while some of the older infants continue to fail. And even when 6 month-olds succeed as a group, there are still some infants who failed to display discrimination (Csibra, Gergely, Biro, Koos, & Brockbank, 1999). What explains these individual differences in the timing of infant cognitive change? European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
32
A. Karmiloff-Smith et al.
The main hypothesis that we test in six experimental tasks across three different cognitive domains (speech processing, face processing and action/event processing) is that individual differences hidden in infant group data can in part be explained by exogenous constraints, e.g., by differences in the quality of mother-infant dyadic interaction. In other words, we ask whether cognitive domains are endogenously constrained following a fairly rigid maturational course in infancy, or whether they are differentially affected by exogenous factors. Moreover, previous research has only examined the influence of maternal sensitivity on very general intelligence measures, whereas our approach is novel in that it examines dyadic effects longitudinally on several specific cognitive domains, attempting to pinpoint in finer detail the effects of mother-infant dyadic interaction on the timing of cognitive development. The effects of mother/child interaction on general cognitive measures like the Bayley Scales of Infant Development (Bayley, 1993) have been studied mainly in the area of maternal or infant psychopathology (Murray, Fiori-Cowley, & Hooper, 1996; Scarborough, 1990; Streissquth & Bee, 1972) or its effects on single domains (e.g., mother’s face recognition: Thompson & Trevathan, 2006). However, to our knowledge this is the first time in which the same healthy infants have undergone longitudinal, multi-domain testing, under the perspective of the quality of dyadic interaction. Method
Over 100 infants participated in a longitudinal design (at 6 and again at 10 months) from three European laboratories set up identically (London, Munich and Paris). On both visits, infants were tested on six experimental tasks and were observed in a 3-minute videotaped mother/child play session with a structured set of toys, identical across the three laboratories. Parents also filled out questionnaires with respect to socio-economic status and education, as well as the Kent Inventory of Developmental Skills (KIDS, Reuter & Bickett, 1985; Reuter, Katoff, & Gruber, 2001). KIDS is a caregiver report that assesses infants from 3 to 12 months of age, in terms of cognitive, motor, language, self-help, and social development. It comprises 252 items in total, e.g., “drinks from a cup”, “remembers where things are kept in the house”, or “shows jealousy”. Caregivers have three answer options: “yes, can do it”, “used to do it but outgrew it”, and “no, cannot do it yet”. In terms of validity, the KIDS correlate well with the Bayley Scales of Infant Development (r = .059-.072) (Aylward, 1994, 42f.; Bayley, 1993). Participants
After obtaining ethics permission from the universities of each of the centres involved, healthy, monolingual 6-month-olds were recruited from the infant subject pools in London, Munich and Paris. Mothers register their infants on each of the lab’s databases and European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
33
Infant Cognitive Change
those whose babies were close to 6 months of age were contacted by telephone to see whether they were willing to undertake the commitment of a longitudinal study, i.e., coming in when their babies were 6 months old and again when their babies reached 10 months of age. The infants were selected randomly from the databases, according to their date of birth. We also were obliged to restrict recruitment to Caucasian infants because of the design of the face processing tasks. Mothers and infants visited the laboratories for one half-day when the infants were 6 months of age and again four months later for another half-day when the same infants had reached 10 months of age. Infants were tested between 15 days prior and 15 days post reaching 6 months and between 15 days and 20 days prior and post reaching 10 months. Exclusion criteria were: any known neurological problems, prematurity, uncorrected visual or auditory deficits, and bilingualism. The three laboratories pooled their data, with each laboratory subsequently analyzing the data for two of the experimental tasks for the overall data from all three labs (London: the two speech processing tasks, Paris: the two face processing tasks, and Munich: the action/event processing tasks), and subsequently assessing the effects of dyadic interaction ratings on the results for each domain. Although each laboratory targeted 60 infants (to reach a total of 180 in all), there was considerable drop-out due to the normal attrition rate of longitudinal studies (Wells, 1985) with failure to return at 10 months, as well as to recruitment errors (e.g., bilingual children; wrong age), illness, sleep, fussiness, and experimenter error. Table 1 provides the participant characteristics for those retained overall, Table 2 the participant numbers retained for each domain (excluding experimenter errors, fussiness etc.), Table 3 the socio-economic data of these participants, and Table 4 the full KIDS scores at six months, together with the breakdown for the Cognitive, Communication and Motor components of these scores. Table 1. Participant Characteristics. Group Retained for analysis Mean age at testing Gender: male Singletons
Table 2.
N
London 6 mo.
Paris 6 mo.
Munich 6 mo.
Total N 6 mo
London 10 mo.
Paris 10 mo.
Munich 10 mo.
Total N 10 mo.
72 6.14 (+/- 28d) 59% 74%
39 6.21 (+/- 37d) 48% 61%
62 6.05 (+/- 24d) 50% 82%
173 6.13
62 10.22 (+/- 45d) 58% 74%
33 10.0 (+/- 40d) 46% 61%
60 10.02 (+/- 40d) 51% 82%
155 10.07
55% 72%
56% 72%
Participants remaining in analyses for each domain.
6 months Action Perception 57
Speech Perception 76
Face Perception 98
10 months Action Perception 55
Speech Perception 76
European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
Face Perception 81
34
A. Karmiloff-Smith et al.
Table 3. Socio-economic Data. Group Crèche - 6 months - 10 months
London
Paris
Munich
All
8% 21%
19% 41%
0.02% 4%
9% 33%
Mother’s level of schooling College Secondary school
79% 21%
70% 30%
76% 24%
75% 25%
Table 4. Scores on Kent Inventory of Developmental Skills (KIDS) Mean Overall Cognitive Communication Motor
London 119 26.43 16.43 35.57
Paris 105 23.4 14.0 29.8
Munich 106 23.4 11.6 36.8
All 110 24.4 14.0 34.1
As explained above, while there was shortfall from the targeted 180 infants, Tables 3 and 4 show that the participants who were retained for analysis were well balanced across the three labs in terms of gender, mothers’ education level and infants’ developmental skills. An unavoidable difference emerged with respect to those infants in crèche (Paris high) and those whose mothers stayed home (Munich high), but otherwise the home environments, i.e., level of mothers’ education and infants’ developmental skills, were closely matched across the three European centres. Procedure
Every infant from London, Munich and Paris was tested on all six experimental tasks and the mother-infant play session at both 6 months of age and again at 10 months of age. All stimuli for the six experiments were presented using a wide screen with speakers on each side. An observer, blind to the items tested, monitored the infant’s looking behaviour on a closed-circuit TV system outside the testing booth. Order of the six tasks was counterbalanced across infants in the three laboratories, and ample time was given for breaks. The three domains were selected because they represent sufficiently different aspects of infant development which have hitherto been studied in isolation and for which we aimed to collect cross-domain data from the same infants, to ascertain whether the timing of cognitive change was domain-specific or domain-general. The six particular tasks were chosen because at the group level they had previously been shown and replicated to yield important developmental changes between the ages of six and ten months (Best et al., 1995; Kiraly et al., 2003; Kotovsky & Baillargeon, 1994, 1998; Sangrigoli & de Schonen, 2004a, 2004b; Werker et al., 1981; Woodward, 1999). European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
Infant Cognitive Change
35
Experiments 1 and 2: Processing of human goal-directed actions and causal physical events. In the first year of life, infants learn to perceive human actions as being goaldirected. They also develop a model of physical causality such as the representation of collision events. Two experimental tasks were carried out in all three laboratories with two aims: (1) to replicate and extend findings of infants’ developing understanding of goal-directed human actions and of physical causal events, and (2) to assess the impact of dyadic interaction style on cognitive change on each of these tasks. An infant-dependent familiarization procedure was used. For the human goal-directed action, a modification of Woodward’s (1999) “Back-of-hand” study was employed, in which a salient action effect (relocating an object) was added to the unfamiliar action of lowering the back of the hand in front of an object, in order to enhance the perception of goal-directedness (Kiraly, Jovanovic, Prinz, Aschersleben, & Gergely, 2003). In four familiarization trials, the hand aimed at one of two objects on the screen. The positions of the objects were then swapped. In the test phase, the human agent’s hand alternately aimed at the “new” object while carrying out the same movement path, or at the “old” object while carrying out a new movement path. For the physical causality task, a modification of Kotovsky and Baillargeon’s (1994, 1998) collision study was used with the same infants. In four familiarization trials, the infants were familiarized with a collision event between two medium-sized balls, providing a calibration point (mid-stage) for their expectation of how far a ball of this size would dislocate a target ball of equal size. In the two test trials, the infants saw alternately a “possible” event, namely a big ball pushing the target ball to the end of the screen, or an “impossible” event, namely a tiny ball moving the target ball to the same distance. The two tasks were separated in time and order was counterbalanced across infants. Experiments 3 and 4: Processing of native and non-native speech contrasts. During the first year of life, infants start out by being able to discriminate speech sounds from all human languages, but as their brains become increasingly specialized for the sound patterns of their mother tongue, they become unable to discriminate between non-native sounds. Infant discrimination of syllables from the phonology of their native tongue was compared to discrimination of non-native phonemes. The experiments had two aims: (1) to replicate the finding of reduction in the discrimination of non-native phonemes between 6- and 10-months of age in English, as well as testing this for the first time in French and German, and (2) to explore the relationship between this phonological discrimination ability and dyadic interaction ratings. Each infant participated in two syllable discrimination tasks, one in their native language (English, French or German) and one from a different language family, namely Hindi. The experiments were separated in time, and order was counterbalanced across the infants. The syllable pairs were ‘ba’ versus ‘da’ in the native experiment, and dental vs. retroflex contrasts (voiceless, non-aspirated) in the non-native experiment (based on Werker et al., 1981). Various tokens of the syllables were recorded by native female speakers of each of the four languages. While Hindi speakers would hear the two retroflex and dental stop consonants as two distinct European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
36
A. Karmiloff-Smith et al.
phonemes, naïve adult listeners of English, French or German would hear the two Hindi consonants as a single, identical phoneme: alveolar- [t]. With the exception of the native vs. non-native syllables used, the two experiments were otherwise identical. A familiarization-preference procedure was used to assess discrimination (Bosch & Sebastian Galles, 2001). In the familiarization, half of the infants was familiarized with one syllable and the other half with the other syllable, until they accumulated 1.5 minutes of sustained attention to this material. In the test phase, four trials were presented quasi randomly, two in which new tokens of the same category as the one in the familiarization phase were presented (same trials), and two in which new tokens of the contrastive category were heard (switch trials). Familiarization and test trials started with an image on the centre screen to capture the infant’s attention. As soon as the infant began to look at it, the image disappeared and a different image appeared quasi-randomly on one of the two sides of the screen. When the infant looked in that direction, the presentation of the test stimuli began and continued until its completion (25 seconds) or until the infant ceased to look for more than two consecutive seconds. Experiments 5 and 6: Processing of native and non-native faces. Like the progressive perceptual narrowing for native speech processing, research has shown that infants are initially able to discriminate faces from any ethnic group, but with time they fail to discriminate faces from other groups because of their growing specialization with faces from their own ethic group. This subsequent failure, albeit a positive development, is known as the “Other Race Effect” (Sangrigoli & de Schonen, 2004a). The present tasks had two aims: (1) to ascertain whether the Other Race Effect becomes increasingly pronounced by 10 months compared to 6 months; and (2) to assess the impact of dyadic interaction styles on progressive face expertise. A visual paired comparison after habituation to criterion was used to assess recognition of individual faces, inferred from the infants’ visual novelty preference. Each infant from the three laboratories participated in two face recognition experimental tasks, one with Caucasian and one with Asiatic female faces as stimuli. The experiments were separated in time, and order of the experiments was counterbalanced across the infants. Black and white photographs of the faces of 5 Caucasian (aged 23-32 years) and 5 Asiatic women (same age range) were used as stimuli. All faces were photographed from frontal view with a neutral expression, under the same lighting and against a grey background. The same white shower-cap, pasted during editing of the photographs, masked hair and hairline. An Other Race Effect had previously been observed with these same stimuli in both children and adults (Sangrigoli & de Schonen, 2004a). Each face of a pair was used for an equal number of infants for familiarization and as a novel face. Infants were first habituated with one face and then presented with a pair of faces, the familiar and a new face of the same ethnic morphology. During habituation, the presentation of the stimulus started when the infant looked at the central fixation point. A look-away criterion of two seconds was used to determine the end of each fixation. The stimulus was represented as soon as the infant European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
Infant Cognitive Change
37
looked again at the fixation spot. The habituation criterion was reached either after a non-stop 25-second fixation duration, or when the mean fixation duration across three successive trials had decreased by 50% under the mean duration of the initial first three trials. Recognition was assessed immediately after habituation by a novelty visual preference test during which visual fixation on each stimulus was measured. There were two successive test trials of fixed duration (10 seconds each, with right left reversal of the two face positions), which started when the infant looked at the central fixation point, the latter disappearing when the stimulus pair was presented. Assessing mother-child interaction. Inserted into one break in the run of the six experimental tasks was a mother/infant dyadic interaction session, using a structured set of toys which was identical across the three laboratories. Two measures were used to rate the quality of mother-child interaction (Chatoor et al., 2004; Claussen & Crittenden, 2000; Crittenden, 1979-2004, 2006). These were chosen because, unlike many measures of dyadic interaction, they can be used in the first year of life. The Chatoor Play Scale is an instrument for the global rating of mother-child interaction in a free play situation. The mother is rated for the frequency of occurrence of affect, verbal and non-verbal communication and sensitivity to the infant on a 4-point scale (none, little, pretty much, very much); the infant is rated for affect, vocalizations, non-verbal communication, and play behavior in the same way. There are 4 subscales, namely (1) Dyadic Reciprocity, which captures positive exchanges between mother and child, (2) Maternal Unresponsiveness, which captures inappropriate maternal behaviors during play, (3) Dyadic Conflict, which captures mother’s and infant’s distress and/or anger, as when the mother makes critical remarks about the infant or her play, and (4) Maternal Intrusion, which captures mother’s directing or controlling the play without regard for her infant’s cues. A maternal dyadic reciprocity item is, e.g., “Talks to infant”; an infant reciprocity item is “Vocalizes to mother”. The Child Adult RElation (CARE)-Index (Crittenden, 2006) is an attachment instrument that can be applied in the first year of life for the purpose of screening, intervention, and research. It provides a qualitative and quantitative measure of dyadic interaction during play. Mothers and infants are rated for a 3-minute period of free play. Mothers are rated on 3 qualitative scales, namely, sensitive (main scale), controlling, and unresponsive; infants are rated on 4 scales, namely cooperative, compulsive-compliant, difficult, and passive. These qualities are scored on 7 expressive channels: facial expression, vocal expression, position/body contact, affection, turn-taking, control, and choice of activity. A maximum score of 14 (2 points x 7 scales) can be allocated to each interactional quality, but mixed scoring among them is also possible. So, for example, a maximally sensitive mother would receive a score of 14 on the main sensitivity scale; a mixed sensitive-controlling mother would receive a score of 7 for sensitivity and 7 for control. The results of the Chatoor turned out to be very similar to those of the CARE-Index, allowing us to feel confident that these instruments capture the nature of early mother-infant dyadic interaction, so only the CARE-Index data are reported here. European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
38
A. Karmiloff-Smith et al.
The mother was instructed to play with her infant as she usually does. A 3-minute video-recorded sequence of the play was subsequently evaluated by trained coders. Members of each of the three European laboratories were sent for training in the use and analysis of the CARE-Index. Inter-rater reliability for CARE-Index: Inter-rater reliability was established between two coders. Both coders first rated n = 10 participants independently. After reaching agreement on divergent scores, participants’ behaviours were re-coded. Reliability within each of the three labs was very high (above r = .9 for the sensitive/ cooperative scale, the control/compliant scale and the unresponsiveness/passivity scale. Inter-rater reliability across labs was obviously lower since no discussion to reach agreement on divergent scores could be made at a distance, although these had been agreed by two coders locally. These interlab differences may reflect cross-cultural variations in how the quality of mother-child dyadic interactions are perceived. Importantly, however, after initial analyses based solely on age groups, each lab then used the results from the infants from all three labs in order to divide the population into CARE-Index subgroups for the second analysis. Results
All children retained for analysis fell in the normal range on the KIDS, and there were no significant differences between the three European centres on developmental levels, socio-economic criteria or mother’s education, so all data were pooled for each task. Individual data points were first checked for outliers, which were removed if outside typical statistical criteria. For each domain, we first report the group data and then re-analyze the findings on the basis of mother-child interaction ratings. Experiments 1 and 2. Inter-rater reliability for action/event processing was above r = .95 for both tasks. As expected, at the group level, 6-month-old infants did not discriminate different human goal-oriented actions (F(1,48)=.296, p = .593), but the group of 10month-olds did display the expected effect (F(1,58)=17.25, p < .001). However, when the data were re-analyzed according to dyadic interaction ratings, then 6-month-old infants of mothers with low sensitivity were in advance, already discriminating between human goal-directed actions, whereas the 6-month-old infants of mothers with high sensitivity did not. Breaking down the low-sensitivity measure further into “moderately controlling mothers” and “unresponsive mothers”, it turned out that it was only the sub-group of compliant 6-month-olds with controlling mothers who displayed advanced discrimination of human goal-directed actions (F(1,26)= 4, p < .05). With respect to the processing of physical events, as expected at the group level 6-month-old infants did not discriminate between possible and impossible physical events (F(1,63)= .118, p = .73), but at 10 months they did (F(1,50)= 3.02, p < .05). However, when the data were reanalyzed taking dyadic interaction ratings into account, there European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
Infant Cognitive Change
39
was no effect at 6 months, but at 10 months it was only the infants of mothers with high sensitivity who exhibited successful discrimination of physical events (F(1,18)=16.64, p < .001). Thus, in these first two experiments the quality of dyadic interaction was shown to have differential developmental effects on the processing of human goal-directed actions versus physical causality events at the two ages tested. Experiments 3 and 4. Inter-rater reliability for speech processing was high (r = .92) for both tasks. At the group level 6-month-old infants were, as expected, able to discriminate syllables both from their own language (F(1,74)= 19.2, p < .001) and from Hindi (F(1,64)= 12.2, p = .001). At 10 months, by contrast, infants had by then narrowed their perceptual range in favour of the sound variations of their mother tongue, at the group level showing discrimination only for native syllables (F(1,73)= 10.5, p = .002), but no longer for non-native syllables (F(1,62)=.09, p =.76). However, when discrimination was assessed on the basis of dyadic interaction ratings, the 6-month old infants of mothers with high sensitivity showed earlier specialization for the sounds of their native tongue, i.e., revealing a positive outcome in no longer being able to discriminate between non-native contrasts already by 6 months of age (F(1,37) = 3.2, p = 0.08). Hence, although the vast majority of infants displayed the expected developmental pattern by 10 months, 6- month-olds of mothers with high sensitivity showed earlier specialization by narrowing their perceptual discriminations to the speech sounds of their mother tongue. Experiments 5 and 6. Inter-rater reliability for face processing was high (above r = .9) for both tasks. Interestingly, the results for face processing differed considerably from those of the other two domains tested. As a group, the 6-month-olds showed recognition of faces from their own ethnic group (F(1,94) = 8.349, p < .005), but not for faces from the different ethnic group (F(1,94) = .001, p >.10). In other words, the 6-month-olds already displayed the Other Race Effect. At 10 months, the group data yielded no novelty preference either for native faces (F(1,82) = .182, p > .10), or for non-native faces (F(1,75) = 1.854, p > .10). Importantly, and in contrast to the other two domains, there were no significant effects of dyadic interaction style on face processing at either age. In summary, at the group level, all the findings were in line with previous studies. However, when dyadic interaction was taken into account in the analyses, this gave rise to novel findings, affecting the timing of infant cognition differently across ages and across domains. When mothers display highly sensitive contingent behaviour, this gives rise to earlier specialization in their infants for native speech processing at 6 months and for physical event processing at 10 months. For human action processing, by contrast, it is the moderately controlling mothers whose compliant infants show earlier specialization at 6 months. Finally, for face processing, dyadic interaction styles yielded no effects at the ages studied. European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
40
A. Karmiloff-Smith et al.
Discussion
One might have expected that it would be dyads high on the sensitivity rating whose development in terms of the timing of specialization would be influenced across all domains, i.e., that the positive effects of highly contingent mother-infant interaction would be domain-general. But, interestingly, this was not the case. For the domains of both physical event processing and speech processing, the infants from dyads with high sensitivity were indeed in advance of their peers, i.e., contingent dyadic interaction fostered earlier specialization, at 6 months for speech processing and at 10 months for event processing. By contrast, the opposite held for the processing of human goaldirected actions. In the latter case, it was the compliant infants of moderately controlling mothers who displayed earlier specialization. Finally, face processing turned out not to be open to influences from dyadic interaction at the ages studied, since in the face-processing domain no relations emerged between infant behaviour and motherinfant sensitivity scales. What are these differences due to? Let us first take the case of human goal-directed actions and their relation to motherinfant dyadic interaction ratings. Why are the compliant infants of moderately controlling mothers the ones who demonstrate earlier success than their peers? Our analyses of the fine details of dyadic interactions revealed that moderately controlling mothers tended to impose their own choice of toy on their infants and to rapidly change the toys, without showing sensitivity to the infant’s current focus of attention. Such maternal behaviours leave little room for their child’s spontaneous initiatives. In this way, and unlike the mothers with high sensitivity, the controlling mothers place a burden on their infants who need to become constantly sensitized to their mother’s goals rather than focusing on their own. In other words, in their environments these infants actually need frequently to process human goal-directed actions in order to comply with their mother’s goals. By contrast, mothers with high sensitivity, while helping their infant initiate activities with a toy, tended to place equal emphasis on anticipating the goals of their infant, letting their infant initiate activities and altering their own goals to follow their infant’s goals. These very basic yet subtle variations in dyadic interaction style place different processing demands on infants and contribute, we argue, to the advance of the compliant 6-montholds with controlling mothers in the human action processing domain. The fact that it is infants from highly sensitive dyads who show earlier specialization in the domains of both physical event processing and speech processing may be explained by the fact that such mothers provide their offspring with a much greater amount of appropriate-level input, anticipating and updating the changing needs of their children as they explore objects in the physical world and listen to the varied sounds of language. Controlling mothers, by contrast, speak less to their infants or talk with less variation in style, and do so without taking into account the progressively changing nature of their infant’s vocalizations. Our findings thus indicate that the quality of dyadic interaction is one important determinant in early speech development. Much the same applies to physical event processing. The less sensitive mothEuropean Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
Infant Cognitive Change
41
ers tend to interrupt their infants’ exploration of objects and their infants comply with the interruption, whereas the more sensitive mothers leave their cooperative infants time to explore more fully the properties of objects, which builds up their knowledge of the physical world. Numerous infancy studies have shown that once infants build up knowledge of the physical world, they are able to show sensitivity to violations of that knowledge when it is presented in experimental designs (e.g., Kotovsky & Baillargeon, 1994, 1998; Spelke & Kinzler, 2007). With respect to faces, our results indicate that, unlike the domains of action/event processing and speech processing, individual differences in the timing of face processing are not explained by dyadic interaction ratings at either 6 or 10 months of age. This points to a course of development that is more tied to maturation and experience with proprietary inputs of face stimuli and less to other exogenous influences such as the quality of dyadic interaction. Obviously even sensitive mothers cannot vary the complexity of faces that their infants see, yet they can and do choose to gradually increase the complexity of the speech sounds they hear or the objects that they play with. Another important finding in this study was that, unlike the assumptions drawn from cross-sectional data suggesting that non-native speech processing and non-native face processing follow a similar developmental trajectory in the perceptual narrowing to native phonemes and faces, our longitudinal cross-domain approach yielded different developmental trajectories across these two domains. This concords with research on unilateral brain lesion in early childhood, in which the brain’s recovery turns out to be less plastic for face processing than for speech processing (de Schonen et al., 2005). This indicates that face processing is less open to cerebral reorganization and external influences than speech processing. Our study also suggests that there are greater endogenous constraints on face processing, but greater exogenous constraints on speech processing and action/event processing. As with any ambitious, large-scale study of infant cognition, there are both limitations and advantages to the current research. The study’s limitations involve the uneven numbers of infants retained from each of the labs and the cultural differences that obtain even when comparing relatively closely related European countries. This was particularly the case across the three labs with respect to the number of infants in crèche. Yet to counterbalance this, the levels of maternal schooling and infant developmental skills were actually well matched across the three labs. Analyses of motherinfant dyadic interaction are necessarily somewhat subjective. Nonetheless, within lab inter-rater reliability was extremely high on the CARE-Index, with cross-lab reliability being lower. Inter-rater reliability for the experimental tasks was very high. One of the advantages of the present study was that our total infant numbers were large compared to most studies of infant development, making our findings more robust. Moreover, rather than focusing on isolated domains cross-sectionally, we examined multi-domain cognition longitudinally between 6 and 10 months. Finally, whilst we replicated previous findings in each of the domains at the group level, our subtle sub-groupings according to dyadic interaction allowed us to reveal new findEuropean Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
42
A. Karmiloff-Smith et al.
ings, pinpointing the effects of mother/infant interaction, not as a domain-general phenomenon, but as a function of domain and age. One of the original purposes of our study on typically developing children was to lay the foundations for the subsequent study of early atypical development. For example, children and adults with Williams syndrome, a rare neurodevelopmental condition (Donnai & Karmiloff-Smith, 2000), are proficient at face and language processing but very poor in other domains, but little is known at the infant level regarding their capacities in speech processing (but see Nazzi, Paterson, & Karmiloff-Smith, 2003), face processing or in the domain of action/event processing. Moreover, once a parent knows that their child has a genetic disorder of any kind, their behaviour changes subtly in response to their infant (Karmiloff-Smith, 2009). So, how does mother/infant dyadic interaction affect infant development when the child is developing atypically? And does developmental change occur in the same age-specific and domain-specific ways in atypical development even if it is delayed? These are clearly important questions for future research that have relevance for the planning of early intervention programmes with parental participation. Of course other factors also affect the timing of cognitive change (Kuhl et al., 2003). One factor might lie in allelic differences on certain genes, but this was not at the time part of our experimental design. Another possibility affecting the findings in the current study is infant temperament. Fortunately, some aspects of infant temperament (cooperation, ease of recovery from distress, compliance, passivity) and their relation to maternal temperament are taken into account when using the CARE-Index. Another influence might come from whether the infants were singletons or not, but as shown earlier, the three labs were relatively well matched on this dimension. Other factors include various other environmental differences such as attendance at crèche although, as noted above, levels of maternal schooling and infant intelligence were well matched across the three European labs. To reiterate, dyadic interaction is not the only factor affecting cognitive change in early infancy, but our study shows that it is clearly one of the factors. Whether differences in dyadic interaction style continue to influence developmental timing beyond the first 10 months of life remains of course an open question. Our study also highlights the important fact that dyadic interaction doesn’t have an across-theboard, domain-general enabling or inhibitory effect on the timing of infant cognitive change, because our findings indicate that differences in dyadic style can subtly foster or delay development in domain-specific and age-specific ways. Our multidomain, longitudinal study allows us to conclude that the timing of early developmental change is indeed open to exogenous influences like the quality of dyadic interaction and the presence of appropriate-level stimuli in the environment. While cross-sectional, single-domain group data can camouflage these effects, our study reveals that these exogenous influences interact with endogenous constraints to contribute to the individual differences in timing that we observe in cognitive development over the first year of life. European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
Infant Cognitive Change
43
References Ainsworth, M.D., BBlehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the Strange Situation. Hillsdale, N.J.: Erlbaum. Aylward, G. P. (1994). Practitioner’s guide to developmental and psychological testing. New York: Plenum Publishing Corporation. Bayley, N. (1993). Bayley Scales of Infant Development. Second Edition. San Antonio, TX: The Psychological Corporation. Behne, T., Carpenter, M., Call, J., & Tomasello, M. (2005). Unwilling vs. unable: Infants’ understanding of intentional action. Developmental Psychology, 41, 328-337. Best, C. T., McRoberts, G. W., LaFleur, R., & Silver-Isenstadt, J. (1995). Divergent developmental patterns for infants’ perception of two nonnative consonant contrasts. Infant Behavior & Development, 18, 339-350. Chatoor, I., Surles, J., Ganiban, J., Beker, L., McWade Paez, L., & Kerzner, B. (2004). Failure to thrive and cognitive development in toddlers with infantile anorexia. Pediatrics, Vol. 113, no. 5, 5 May 2004, e440-e447. Claussen, A.H., & Crittenden, P.M. (2000). Maternal sensitivity. In P. M. Crittenden & A. H. Claussen (Eds.), The organization of attachment relationships: Maturation, culture, and context (pp.115-124). New York: Cambridge University Press. Crittenden, P. M. (1979-2004). CARE-Index: Coding Manual. Unpublished manuscript, Miami, FL. Available from the author at http://www.patcrittenden.com/ Crittenden, P.M. (2006). CARE-Index. Coding manual. Family Relations Inst., Miami, FL, USA. Csibra, G., Gergely, G., Biro, S., Koos, O., & Brockbank, M. (1999) Goal attribution without agency cues: The perception of ’pure’ reason in infancy. Cognition, 72, 237-267. de Schonen, S, Mancini, J., Camps, R., Maes, E., & Laurent, A. (2005). Early brain lesions and face-processing development. Developmental Psychobiology, 46, 184-208. Donnai, D., & Karmiloff-Smith, A. (2000) Williams syndrome: From genotype through to the cognitive phenotype. American Journal of Medical Genetics: Seminars in Medical Genetics, 97, 164-171. Karmiloff-Smith, A. (2009) Nativism vs. neuroconstructivism: Rethinking developmental disorders. Special issue on the interplay of biology and environment. Developmental Psychology, 45, 56-63. Király, I., Jovanovic, B., Prinz, W., Aschersleben, G., & Gergely, G. (2003). The early origins of goal attribution in infancy. Consciousness & Cognition, 12, 732-751. Kotovsky, L., & Baillargeon, R. (1994). Calibration-based reasoning about collision events in 11-month-old infants. Cognition, 51, 107-129. Kotovsky, L., & Baillargeon, R. (1998). The development of calibration-based reasoning about collision events in young infants. Cognition, 67, 311-351. Kuhl, P. K., Tsao. F.-M., & Liu, H.-M. (2003). Foreign-language experience in infancy: Effects of short-term exposure and social interaction on phonetic learning. Proceedings of the National Academy of Sciences, 100, 9096-9101. Lipton, J.S., & Spelke, E.S. (2004). Discrimination of large and small numerosities by human infants. Infancy, 5, 271-290. Murray, L., Fiori-Cowley, A., & Hooper, R. (1996). The impact of postnatal depression and associated adversity on early mother-infant interactions and later infant outcome. Child Development, 67, 2512-2526. European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
44
A. Karmiloff-Smith et al.
Nazzi, T., Paterson, S., & Karmiloff-Smith, A. (2003). Early word segmentation by infants and toddlers with Williams syndrome. Infancy, 4, 251-271. Reuter, J., & Bickett, L. (1985). Kent Infant Development Scale Manual, 2nd ed. Kent, OH: Kent Developmental Metrics. Reuter, J., Katoff, L., & Gruber, C. (2001). Kent Inventory of Developmental Skills (KIDS). Los Angeles, CA: |Western Psychological Services. Sangrigoli, S., & de Schonen, S. (2004a). Recognition of “own-Race” and “other-Race” faces by three-month-old infants. Journal of Child Psychology and Psychiatry, 45, 1-9. Sangrigoli, S., & de Schonen, S. (2004b). Effect of visual experience on face processing: A developmental study of inversion and non-native effects. Developmental Science, 7, 74-87. Spelke, E.S. & Kinzler, C.D. (2007). Core knowledge. Developmental Science, 10, 89-96. Streissquth, A.P., & Bee, H.L. (1972). Mother-child interactions and cognitive development in children. Young Children, 27, 154-173. Thompson, L.A., & Trevathan, W.R. (2006). Cortisol, mother-infant dynchrony, and learning in 6-month-olds. Poster presented at the International Conference for Infant Studies, Kyoto, Japan. Werker, J. F., Gilbert, J. H., Humphrey, K., & Tees, R. C. (1981). Developmental aspects of cross-language speech perception. Child Development, 52, 349-355. Wells, G. (1985). Language development in the pre-school years. Cambridge: CUP. Wood, J.N., & Spelke, E.S. (2005). Infants’ enumeration of actions: Numerical discrimination and its signature limits. Developmental Science, 8, 173-181. Woodward, A. L. (1999). Infants’ ability to distinguish between purposeful and non-purposeful behaviors. Infant Behavior and Development, 22, 145-160. Annette Karmiloff-Smith and Mayada Elsabbagh, Birkbeck Centre for Brain & Cognitive Development, University of London, UK; Gisa Aschersleben, Department of Psychology, Saarland University, Saarbrücken, Germany; Scania de Schonen and Josette Serres, Neurocognitive Development Group, CNRS, Paris, France; Annette Hohenberger, Middle East Technical University, METU, Ankara, Turkey. Author Note: This study was made possible by a grant to AK-S from Procter & Gamble International, Geneva, Switzerland, from which was formed the Pampers European Research Consortium (PERC). PERC comprised the six authors of this paper as well as the following individuals, who assisted part-time with participant recruitment, data collection, data coding and data analysis: Ruth Campos, Elisa Cunial, Joke van Herwegen, and Elizabeth Knight in London; Tanja Hofer, Andrea Walter, Sabine Gasser, Kathrin Muehlbauer, and Amory Faber in Munich; Edith Rosset and Sandy Sangrigoli in Paris. We also wish to thank Laura Bosch and Núria Sebastián-Gallés for very helpful suggestions regarding methodology. The data from the German infants were collected while GA and AH were at the Max Planck Institute for Human Cognitive and Brain Sciences, Munich, Germany. The data from the English infants were collected when AK-S and ME were at the Institute of Child Health, London. The data were fully analysed and written up from the authors’ new institutions. Correspondence concerning this article should be addressed to Professor A. Karmiloff-Smith, CBCD, Birkbeck, 32 Torrington Square, London, WC1E 7HX, U.K. or preferably by email: a.karmiloff
[email protected] Annette Karmiloff-Smith received her doctorate at Geneva University where she worked with Piaget. She is now Professorial Research Fellow at the Birkbeck Centre for Brain and Cognitive Development, University of London. Her research interests include tracing cognitive-level and neural-level deficits in genetic disorders back to their origins in infancy, genotype/phenotype relations, animal models, and building theories of human ontogeny. Gisa Aschersleben received her PhD and Habilitation at the University of Munich. From 2000-2006 she headed the “Infant Cognition and Action” Research Group at the Max Planck Institute for Human Cognitive & Brain Sciences in Munich, from where she moved to a full professorship in developmental psychology at Saarland University. Her research interests include early cognitive development, mother-child interaction, and Theory of Mind. Scania de Schonen started as a researcher at the National Centre for Scientific Research (CNRS) in Paris, focusing on memory and number in normal adults and those with brain damage, before moving to the Institute of Neurophysiology in Marseille. She has returned to Paris as Emeritus Director of Research at the CNRS. Her research interests include
European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
Infant Cognitive Change
45
functional plasticity of visual perception during normal and abnormal brain development (in particular, the development of face processing). Mayada Elsabbagh, PhD, is a fellow at the Centre for Brain and Cognitive Development in London. Her research focuses on understanding the brain basis of behavioral genetic disorders, combining different methodologies to study brain development, including EEG and eye-tracking with several clinical assessment tools. She is also active in applying scientific research to evidence-based clinical and educational practice. Annette Hohenberger obtained her PhD in Linguistics, subsequently researching various aspects of the psycholinguistics of spoken and sign language in German. A postdoctoral fellow at the Max Planck Institute for Human Cognitive and Brain Sciences, she is now Assistant Professor in cognitive science at the Middle East Technical University, in Ankara, Turkey. Her research interests include spoken and sign language acquisition and cognitive development, as well as the psychological aspects of attitudes to climate change. Josette Serres is a Research Engineer and Psychologist at the Laboratory of the Psychology of Perception in Paris. Her main expertise is in the field of mother-infant interaction. Received: July 2008 - Revision received: April 2010 - Accepted: April 2010
European Journal of Developmental Science [EJDS]. 2010, Vol. 4, No. 1, 31–45 © Vandenhoeck & Ruprecht 2010, ISSN 1863-3811
