- Email: [email protected]
Early declarative memory and self-concept
Infant Behavior & Development 28 (2005) 132–144
Early declarative memory and self-concept Nathalie Prudhomme ∗ Universit´e Ren´e Descartes – PARIS V, Institut de Psychologie, Laboratoire Cognition et D´eveloppement (CNRS UMR 8605), 71 Avenue Edouard Vaillant, F-92774 Boulogne Billancourt Cedex, France Received 1 February 2003; received in revised form 3 December 2004; accepted 3 January 2005
Abstract Infantile amnesia is a major issue in memory development. For a growing body of researchers, self is probably a key-concept for this enigma. The role of the cognitive self in early declarative memory as a function of perceptual manipulation between study- and test-phase was investigated. Two recall conditions were compared: either the same color props as in the modeling phase or different color props. Twenty-four 20-month-old children were observed in a 10-min deferred imitation task with two enabling sequences. An additional control group (N = 16) ensured that imitation scores did not rely on spontaneous production of the target behaviors at this age. Children in the experimental groups were classified as early or late recognizer depending on their success or failure to the mirror test. A significant interaction was observed between color props conditions and recognizer types: only late recognizers’ memory performance was affected by the perceptual manipulation. Results suggest that cognitive self could be one of the factors contributing to the differentiation between episodic and semantic subsystems in early declarative memory by the end of the second year. © 2005 Elsevier Inc. All rights reserved. Keywords: Declarative memory; Self; Toddler; Deferred imitation
1. Introduction A growing body of studies shows, mainly from differed imitation paradigms, that declarative memory emerges around 6 months of age. However, first autobiographic memories do not appear before the age of ∗
Tel.: +33 1 55 20 59 97; fax: +33 1 55 20 59 85. E-mail address: [email protected].
0163-6383/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.infbeh.2005.01.002
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
133
3 years. Several researchers consider self-awareness or cognitive self as probably one of the keys to the long delay in autobiographic memory appearance. The aim of this study is to explore the role of cognitive self in declarative memory development. Autobiographical memory allows retrieving experienced events as personally experienced (Bauer, Wenner, Dropik, & Wewerka, 2000; Howe & Courage, 1997; Nelson, 1993; Perner & Ruffman, 1995; Wheeler, Stuss, & Tulving, 1997). Such memory depends on episodic memory (Tulving, 2002; Tulving & Markowisch, 1998; Wheeler et al., 1997) and, like semantic memory, is a subsystem of the declarative memory system (Tulving, 1983, 2002). Episodic memory is considered as emerging later than semantic memory and stems from “the conjunction of three concepts—self, autonoetic awareness, and subjectively sensed time” (Tulving, 2002, p. 5). The sense of subjective time, possible only with autonoetic awareness, opens the ability to mentally travel in time from present to experienced past (and/or to the imagined future). Autonoetic awareness differs from noetic awareness because it allows encoding facts as personally experienced, and recollecting them, as they were experienced. The self is the traveler, existing in subjective time. In this perspective, it is a key concept for the emergence of autobiographic memory. For Howe and Courage (1997), the cognitive self or self-concept emerging by the end of the second year of life sets the lower limit for early autobiographical memories. The cognitive self is defined as a new knowledge structure that serves to organize memories of experiences that happened to me. These authors propose a memory model in which traces, integrated during the learning phase,1 consist of collections of primitive elements. Once the self is “viable”, its features can be sampled and encoded in traces following a probabilistic process: “as more features are added to the ‘urn’, the greater the likelihood that at least some self features are sampled and encoded in the functional trace for an event” (Howe & Courage, p. 513). A memory trace that includes some features of the self potentially could change into an autobiographical memory and, in turn, contribute to the development of cognitive self. In this model, the probabilistic process of sampling self-features is not specific to the cognitive self but also serves for sampling primitive elements from all knowledge structures integrated into a new memory trace. Thus, autobiographic memories differ from other memories because of the contribution of cognitive self and, consequently, depend upon the developmental level of the self. The development of the self has been of particular interest for many researches during the past 10 years. Two aspects of the self are generally distinguished: the ‘I’ as subjective or implicit sense of one’s own physical and social entity; and the ‘me’ as objective or cognitive sense of self. Recent research focuses on the subjective self considered as either innate or developing from birth through infants’ interactions with their physical and social world. For Rochat (2001) and Lewis (1999), it is not specifically human and it can be described as “an existential sense of self” (Rochat, 2001, p. 204). The cognitive self, assumed to emerge from the subjective self by the end of the second year, corresponds to the self as object of experience, knowledge, and imagination, i.e., me. This sense of self is probably specific to human beings (Lewis, 1999) and the mirror task can serve to evaluate its development in toddlers. Although limited to a particular experience and sensory modality (visual image), the mirror test is considered as “a valid instrument to assess self-knowledge at a conceptual and re-cognitory level” (Rochat, 2001, p. 206). It consists of applying surreptitiously an odorless red mark on the infant’s nose to see if he demonstrates a self-directed behavior in front of a mirror by touching his nose instead of the mirror. For Wheeler 1
And possibly disintegrated/reintegrated across retention intervals and testing phases.
134
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
et al. (1997), this behavior “involves an awareness of one’s physical appearance and the constancy of that appearance throughout time and space” (p. 344) and thus refers to the self as associated to episodic memory in Tulving’s theory. For many authors (Asendorpf & Baudonniere, 1993; Butterworth, 1990; Kagan, 1981; Lewis, 1999; Meltzoff, 1990; Neisser, 1993; Rochat, 2001) to pass the mirror test “marks an important developmental milestone in the second year of life” (Howe & Courage, 1997, p. 506) and would be linked to event memory by the end of the second year (Howe & Courage, 1997). To test this hypothesis, Harley and Reese (1999) computed the correlation between 19-month-olds’ performances in mirror task and deferred imitation. Indeed, they reported a small but significant positive correlation (r = .32, p < .05) between the recalled temporal information and self-recognition. This correlation questions the role of cognitive self in early declarative memory as evidenced by deferred imitation. The deferred imitation test is a non-verbal test of declarative memory (Bauer, 1995; Howe & Courage, 1997; Mandler, 1990; McDonough, Mandler, McKee, & Squire, 1995), used to study event memory in infancy. In this test, new props are used to show to the child specific new actions or action sequences. After a retention interval designed following the aim of the studies, the props are given to the child who then may reproduce the target actions previously demonstrated. In this procedure, children are not allowed to explore the props before the testing phase. For Wheeler (2000), young children’s deferred imitation performances are not evidence of episodic but only of a declarative memory that is probably semantic-like. Under 3–4 years of age, children “are without the capacity to recollect their past in the rich, personal way that comprises episodic retrieval” (Wheeler, 2000, p. 603) because of their insufficiently developed self-awareness. For Bauer (1997), the deferred imitation paradigm potentially taps on episodic memory for at least two reasons: (1) it allows observing the recall of specific, new, previously and briefly exposed events and (2) performances are flexible, e.g., not tied to a specific modality or learning context. Several studies, focusing on flexibility in deferred imitation with toddlers, revealed developmental changes in memory’s specificity over the second year of life; they also showed that, by the end of the second year, children easily generalize their event memories to different colors and/or forms’ props, at least considering number of target actions recalled (Barnat, Klein, & Meltzoff, 1996; Bauer & Dow, 1994; Hayne, Boniface, & Barr, 2000; Hayne, MacDonald, & Barr, 1997). In short, by the end of the second year deferred imitation paradigm allows young children to form and retrieve declarative event memories that would be either semantic-like or episodic-like. At this age, the emergence of the cognitive self, as a new knowledge structure that serves to mediate personally experienced events, may contribute to encode and recollect these event memories in a richer and more flexible way. Specifically, development of the cognitive self may be linked with memories’ flexibility around two years of age. To test this hypothesis, the mirror self-recognition test was used to assess the cognitive self-development with 20-month-old infants. This age was chosen so as to have as many children who pass the self-recognition test and as many who do not. Children were then tested in a deferred imitation task with a 10 min recall delay ensuring that recall tapped long-term memory rather than working memory (Bauer, Van Abbema, & de Haan, 1999). For half of the group, the color of the props was changed in the test phase to examine a possible interaction between color change and success or failure to the mirror task. If the cognitive self contribute to a greater flexibility of event memories, deferred imitation scores would be affected by perceptual manipulations between study and test phases only for the children who failed the mirror test.
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
135
2. Method 2.1. Participants Twenty-four children (12 boys) were recruited by mail from the birth records of the city hall of Paris. Mean age2 was 20 months (m = 601 days; S.D. = 10.9). A control group composed of 16 children (eight boys), mean age: 20 months (m = 600 days; S.D. = 8.7), drawn from the same population, ensured that performance observed in the experimental groups was based on the sequence representation, not on the spontaneous occurrence of the target behaviors. All children were observed individually in the laboratory for a 20-min session and received small gifts at the end of the session. 2.2. Tasks, procedure, and scoring After a brief warm-up period, the child and the experimenter sat around a child-size table. The child’s parent remained in the room, seated on a sofa where the child could see him/her, and was asked to refrain from acting or speaking for the rest of the session. On the floor, close to the child’s chair, several unrelated toys were available to the child. A 100 cm × 80 cm covered mirror, fixed on an easel 30 cm from the floor, was placed near the table. 2.2.1. Deferred imitation The children of the experimental groups were tested for a 10-min deferred recall test on two enabling three-steps sequences. These two sequences were chosen among a pool of sequences commonly used with 20-months-old: making the car go and making the rattle (see Appendix A). Each sequence had two sets of props differing only on the color. For all children, sequences were modeled with the same set of props. Children were placed in an incidental learning condition. The experimenter invited the child to a free play with unrelated toys of his/her choice. When the child was both sufficiently engaged in his/her play and ready to attend to sequence modeling, the experimenter modeled the two sequences without comments, twice in succession. During this modeling phase, the experimenter verified that the child was watching the demonstration. After each demonstration, the props were immediately removed from the child’s sight and the 10-min recall delay started immediately. Children were not permitted to reproduce the two sequences prior to this delay. Unrelated distracting toys helped to increase the child’s tolerance for the procedure. Half of the children were requested to recall the sequences with the color props used for the demonstrations (we refer to this condition as SC), while the other half had to recall the sequences with different color props (we refer to this condition as DC). The order of presentation of the two sequences was counterbalanced across children, with sequences introduced in the same order during modeling and test for each child. During the recall delay, the experimenter enrolled the child in unrelated free play and submitted the child to the mirror test of self-recognition. When the recall delay was over, the experimenter gave the props for each sequence in turn to the child with an explicit verbal instruction to reproduce what she had modeled 10 min before: “Look, I am lending you my game. Can you show me what I did with all this stuff?” Note that the experimenter never provided a verbal reminder such as the verbal label of the sequences, props or target actions. 2
See Table 1 for detailed mean age.
136
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
Children in the control group were enrolled in a free play session with unrelated toys. After 5–10 min of play, the experimenter gave to the child one set of props, for each sequence in turn. During this childcontrolled phase, the experimenter encouraged him/her to explore and manipulate the props by prompting such as: “Look, what can you do with all this stuff?”
2.2.2. Mirror test of self-recognition During the recall delay, the experimenter placed odorless red face paint on the child’s nose in the guise of wiping it with a paper handkerchief. After 2 min of play with unrelated toys, the experimenter drew the child’s attention to the covered mirror and removed the mirror cover. The criterion of selfrecognition was a child’s self-directed behavior such as nose or face touching (Lewis & Brooks-Gunn, 1979). The mirror test was considered as passed if the child touched his/her face within 2 cm of the paint (Harley & Reese, 1999). Children who passed the mirror test (N = 12) were classified as early recognizers (referred as R+). Children who failed the mirror test (N = 12) were classified as late recognizers (referred as R−). The random assignments of each child to one of the prop conditions (SC or DC) and to one order of presentation of the sequences were decided before knowing if he/she was an early or late recognizer. This procedure allowed getting equal numbers of girls and boys in each prop condition and equal numbers of children in each order of presentation. However, the four experimental groups resulting from crossing recognizers’ type (R+; R−) and prop conditions (SC; DC) were not equated for the number of children (R+/SC and R−/DC groups had N = 7; R+/DC and R−/SC had N = 5; see Table 1).
2.2.3. Data reduction All sessions were videotaped for later analysis. All target behaviors, including target actions in each sequence and self-directed behavior in the mirror test situation, were examined. A scorer, unaware of the hypotheses under investigation, coded all sessions. She was trained until she reached an agreement of 90% or more with the experimenter on three successive children from a previous corpus of deferred imitation data. Bauer, Hertsgaard, Dropik, and Daly’ (1998) method of scoring deferred imitation coding was applied. For each sequence, the total number of different target actions produced (max = 3 for each sequence) and the number of pairs of actions produced in the target order (max = 2 for each sequence) were computed for each child. Only the first occurrence of each target action was taken into account. For example, on Make a rattle, if a child produced all three components in the target order, he/she would receive credit for the three different target actions and for the two correctly ordered pairs of actions: put the cube in the half of the container (action 1), cover it with the other half (action 2), and shake the rattle (action 3). If a child only produced actions 1 and 3 in that order, one point would be credited for the number of ordered pairs of actions and two points for the target actions. If a child produced the string of actions 3, 1, 2, 3, he/she would still receive credit for the production of three different target actions. However, the child would only be credited with one correctly ordered pair of actions (i.e., 1–2) but would not be credited with the correctly ordered pair 2–3, because action 3 was performed first. This scoring procedure reduces the likelihood of a child receiving credit for correctly ordered pairs of actions by chance or by trial and error (Bauer et al., 1998). Finally, all children coded as showing a self-directed behavior in the mirror test were classified as R+ while others were classified as R−.
Sequences
Number of target actions (max = 3) Car
Number of pairs of actions in target order (max = 2) Number of target actions (max = 3)
Rattle
Number of pairs of actions in target order (max = 2)
Early recognizers (R+)
Late recognizers (R−)
SC (N = 7, 604a )
DC (N = 5, 599a )
SC (N = 5, 602a )
DC (N = 7, 598a )
m (σ) t-value m (σ) t-value
2.14 (0.69) t(21) = 4.09*** 1.00 (0.67) t(21) = 3.24**
2.60 (0.89) t(19) = 4.51*** 1.40 (0.89) t(19) = 4.34***
1.80 (0.45) t(19) = 2.85** 0.60 (0.55) t(19) = 1.84, ns
1.14 (0.90) t(21) = 1.07, ns 0.43 (0.53) t(21) = 1.20, ns
m (σ) t-value m (σ) t-value
2.71 (0.76) t(21) = 5.32*** 1.71 (0.76) t(21) = 5.50***
3.00 (0.00) t(19) = 5.82*** 2.00 (0.00) t(19) = 7.32***
3.00 (0.00) t(19) = 5.82*** 2.00 (0.00) t(19) = 7.32***
1.43 (1.51) t(21) = 1.49, ns 0.86 (1.07) t(21) = 2.01*
t values result from the comparison of experimental groups with the control group. a Mean age (in days). ∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.
Control group (N = 16, 600a ) 0.75 (0.69) 0.19 (0.40) 0.69 (0.87) 0.19 (0.54)
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
Table 1 Mean numbers of target actions recalled and pairs of actions produced in the target order and S.D.
137
138
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
3. Results 3.1. Preliminary analyses The overall mean recall delay for both sequences and for the four experimental groups was 11 min 43 s (range = 10 min 52 s to 12 min 56 s). Preliminary analyses revealed no significant group difference between sequences for the four experimental groups variables, no gender effect and no effect of order of presentation of the sequences on both dependent variables (mean number of target actions produced and mean number of correctly ordered pairs of actions). Thus, gender and order independent variables were excluded from subsequent analyses.
3.2. Deferred imitation and self-recognition 3.2.1. Comparisons between experimental and control groups Descriptive statistics for target actions and pairs of actions produced in the target order for all groups are presented in Table 1. Experimental groups’ performance as a whole was compared to the control group’s performance for both dependent variables with a 2 (conditions: experimental/control) by 2 (sequences: rattle/car) mixed ANOVA. For the mean numbers of target actions produced, a significant effect of condition was observed: F(1,38) = 33.52, p < .00001. Children in the experimental condition had significantly higher performances than those in the control condition. Neither sequence nor interaction effects were observed. For the mean number of correctly ordered pairs of actions produced, significant main effects of condition (F(1,38) = 30.67, p < .0001) and sequence (F(1,38) = 9.12, p < .01) were observed. Interaction between conditions and sequences was also significant: F(1,38) = 9.12, p < .01. Post hoc comparisons with Tukey HSD test showed that (1) children in experimental conditions produced significantly more correctly ordered pairs of actions across sequences than those in the control condition, and (2) in experimental conditions, children produced significantly more correctly ordered pairs of actions for the rattle than they did for the car. To investigate whether early and late recognizers were equally able to recall specific event sequences after a 10-min delay, depending on whether or not the props color was changed, we compared each experimental groups’ performances with the control group’s performances with a Student’s t-test for both dependent measures and each sequence (see Table 1). Considering the number of target actions, both early and late recognizers in the SC condition always performed significantly better than the control subjects. In the DC condition and for both sequences, only early recognizers produced significantly more target actions than the control subjects. Considering the number of correctly ordered pairs of actions, the children in the R+ groups always had significantly higher scores than the controls. In contrast, late recognizers did not show significantly higher scores than control subjects except in the SC condition and for the rattle sequence only. These results showed that early and late recognizers were not equally able to recall specific event sequences after a 10-min delay, depending on color props condition, sequences and dependent measures. Nevertheless, for both dependent measures and both sequences, children in the experimental groups as a whole always displayed significantly higher performances than did children in the control group, showing evidence of deferred recall. Further analyses concern experimental groups only.
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
139
Fig. 1. Interaction between recognizer type and color prop condition. ∗ p < .05; ** p < .01.
3.2.2. Effects of self-recognition and color prop condition To ensure that the two sequences show the same pattern of results and considering the number of target actions, a 2 (recognizer type: R−/R+) by 2 (color prop condition: SC/DC) mixed analysis of variance (ANOVA) was performed with sequences as a within-subjects variable. Two significant main effects were observed. Children recalled more target actions for the rattle than they did for the car sequence: F(1,20) = 11.63, p < .003. Early recognizers recalled more target actions than did late recognizers: F(1,20) = 6.5, p < .02. No main effect of color prop condition was observed but a significant interaction between recognizers’ type and color prop was found: F(1,20) = 6.05, p < .05. In sum, no significant interaction with sequences was observed. According to these results, Make a car go was always less recalled than Make a rattle. Analysis of individual data showed that most children had difficulties with the second target action of the car sequence (see Appendix A). As a consequence, this sequence was more difficult to encode and/or to recall. Nevertheless, the same pattern of results was observed for each sequence, since no reliable interaction was observed with sequences (see Fig. 1). Consequently, results for each sequence were collapsed (see Table 2) and a 2 (recognizer type: R+/R−) by 2 (color prop condition: SC/DC) mixed ANOVA was performed to examine the effect of self-recognition as a function of color prop condition. As previously, recognizers’ type shows a significant main effect (early recognizers recalled more target actions than did late recognizers: F(1,20) = 6.52, p < .05), no reliable main effect of color prop condition was observed (F(1,20) = 1.51, ns) and a significant interaction between recognizers’ type and color prop was found (F(1,20) = 6.05, p < .05). This interaction was analyzed with separate one-way
140
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
Table 2 Mean numbers of target actions recalled and pairs of actions produced in the target order for experimental groups when sequences are collapsed
Number of pairs of actions in target order (max = 4) t values ∗
Early recognizers (R+) m(σ) t value
Late recognizers (R−) m(σ) t value
All experimental groups m(σ)
Control group m(σ)
SC (N = 7)
S (N = 5)
(N = 24)
(N = 16)
2.42 (1.35)
0.38 (0.94)
DC (N = 5)
DC (N = 7)
2.71 (1.38) 3.40 (0.89) 2.60 (0.55) 1.29 (1.38) t(21) = 5.40*** t(19) = 7.78*** t(19) = 6.33*** t(21) = 2.10∗
p < .05; *** p < .01; *** p < .001.
ANOVA. Analyses of the main effect revealed that in the SC condition, performance of the early and late recognizers did not differ significantly (F(1,20) < 1). In contrast, in the DC condition, early recognizers recalled significantly more target actions than did the late recognizers: F(1,20) = 12.57, p <.01. Moreover, performance of the early recognizers did not differ significantly across color prop conditions (F(1,20) < 1), while the late recognizers recalled significantly more target actions in the SC than in the DC condition: F(1,20) = 6.81, p < .05. The late recognizers were thus more sensitive than the early recognizers to the perceptual characteristics of the props available during the testing phase, indicating that the latter can benefit from event memories more flexible than the former. Considering numbers of correctly ordered pairs of actions produced, and since this measure is not independent of the number of target actions recalled, we first examined the correlations between the two dependent variables, for each sequence and for the experimental groups as a whole. Indeed, the two measures were highly linked: r = .86 for the car and r = .98 for the rattle. Consequently, it was unnecessary to perform a 2 (recognizers’ type: R+/R−) by 2 (color prop conditions: SC/DC) mixed ANOVA for the recall of temporal order, which will mirror the results on mean numbers of target actions recalled (see Fig. 1). These high correlations also precluded a covariance analysis across recognizers’ type and color prop conditions using the absolute number of target actions recalled as covariate (see Barr & Hayne, 1996 for such an analysis). Furthermore, a formal test is usually applied to compare the observed- to the expected by chancerecall of temporal information (Bauer & Dow, 1994; Bauer & Fivush, 1992). This test must be used for each sequence separately. In our experiment, all children in the R−/SC and R+/DC groups produced a complete recall in the target order for the rattle. In the R−/DC group, only three children recalled at least two actions for each sequence and always did it in the correct order (see Table 3). These results preclude the use of the formal test such as indicated above. Data on the recall of temporal information could only be examined considering individual performances. Table 3 shows the number of children who recalled three or two actions in order, or one or no action, for each sequence and each experimental group. For the difficult sequence (i.e., car sequence), it appears that the few late recognizers who recalled two actions in the DC condition always did it in the correct order while the wrong order was observed in all three other groups (i.e., R−/SC, R+/SC, and R+/DC). For the rattle sequence and across recognizers’ type and color prop condition, all children recalling more than one action indeed recalled all the sequence in the target order. Therefore, when the to-be remembered event was easy (i.e., rattle sequence), all observed recalls were all-or-none retrievals; when the to-be remembered event was difficult (i.e., car sequence), only late
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
141
Table 3 Number of children by actions recalled as a function of sequence, recognizers’ type, and color prop condition The car (rAP = .86a ) R+
The rattle (rAP = .98a ) R−
R+
R−
SC
DC
SC
DC
SC
DC
SC
DC
Three actions recalled Three in order Two in order
2 2
4 3 1
0
0
6 6
5 5
5 5
3 3
Two actions recalled In order In disorder
4 3 1
0
4 2 2
3
0 3
0
0
0
One action recalled No action recalled
1 0
1 0
1 0
2 2
1 0
0 0
0 0
1 3
N total
7
5
5
7
7
5
5
7
a
Correlation between number of individual target actions recalled (A) and number of pairs of actions correctly ordered (P).
recognizers in the DC condition displayed all-or-none retrievals while all other children displayed partial retrievals.
4. Discussion The aim of this study was to assess the role of cognitive self in early declarative memory as a function of perceptual manipulation between study and test in a 10-min deferred imitation task. This question relies on Howe and Courage’s (1997) theory which argues that the cognitive self could act on early declarative memory, allowing more elaborative processing through personal features embedded in the memory traces and/or allowing more flexibility in memory. Two major conclusions can be drawn from the target actions recalled: (1) the different-color props condition did not disrupt nor alter early recognizers’ event memory retrievals; (2) it did so for more than half of the late recognizers whatever the difficulty of the to-be remembered event. In our experiment, early and late recognizers as a whole generalized their event memories to different color props, which is congruent with previous studies (Barnat et al., 1996; Bauer & Dow, 1994; Hayne et al., 1997). Moreover, early and late recognizers displayed equivalent performances when the color of props did not change between study and test. Nevertheless, when the color of props changed between study and test, a different pattern of results emerged for early and late recognizers: while most of the latter failed in generalizing their memories, the former always succeeded. These results argue for the role of cognitive self in early declarative memory. Cognitive self may prompt young children’s generalization of specific event memories to different color props over a 10-min delay. When the cognitive self is available, declarative memories could be more flexible and retrieved from identical- as well as different-perceptual cues. In the absence of cognitive self, retrieving could be more dependant on perceptual cues or fluency processing and be disrupted when the available cues perceptively differ from those previously encoded. Nevertheless, it is probably not the only way in which the cognitive self could act on early declarative memory.
142
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
The emergence of a cognitive self as measured by the mirror test may be associated with better representational abilities: across sequences and color prop conditions, early recognizers had better performances than late recognizers. When the cognitive self is available, children could form better organized and detailed declarative memories. This may be identified through the recall of temporal information, because critical information about the order in which the event should unfold is not perceptually cued at testing time: the available props do provide perceptual support to recall target actions but surely not to recall temporal information. To do this, infants have to encode order of the events during demonstration and retrieve it from a representation of the events and in absence of ongoing perceptual support. Unfortunately in the present experiment, sample size and high correlations observed between the two dependent measures do not allow appreciating recall of temporal information independently of recalled specific items. Finally, cognitive self could foster both representational abilities and memories’ flexibility. However, representational abilities and memories’ flexibility may develop jointly or independently. In our experiment, as we did not reach children’s representational abilities per se, it is not possible to decide whether cognitive self prompt general representational abilities, memories’ flexibility or both. In terms of specific items information recalled, the present findings show indeed that the cognitive self is linked to early declarative memory development at least through memories’ flexibility. Now the question is: how does the cognitive self contribute to a greater flexibility in early declarative memories? Following Howe and Courage’s theory (1997), children who succeeded the mirror test could form event memories organized not only as a function of context, categorization, spatial location, object knowledge and so on, but also as a function of me. Therefore, such declarative memories partly mediated by the self could benefit from a more detailed and organized encoding. With only few personal features available for embedding in the memory traces, processing could be more elaborative and memories would be episodic-like rather than semantic. As a consequence, children could recollect these memories in a richer and more personal way that makes them less sensitive to external perceptual features cueing and/or constraining the retrieval. Moreover, the explicit instruction to recall what was witnessed 10 min before could favor episodic retrieval as it referred to personally experienced events. Such recalls could be the first expressions of an emergent episodic subsystem in early declarative memory. In the same way, it could be argued that the majority of the children who failed the mirror test could not form episodic-like memories because they did not reach a “sufficiently viable” cognitive self (Howe & Courage, 1997) and lacked some personal features to be embedded in the memory traces. These children would reach less elaborated and/or more perceptual encoding of the to-be remembered events. Furthermore, if the available cues can instantaneously match encoded items, a priming process could underlie children’s retrieving. In adults, Rajaram (1993) showed that identical primes between study and test increased “know responses”, which are associated with semantic memory. Moreover, “fluency of processing may seem more plausible when it refers to a single previous encounter with one studied item” (Gardiner & Richardson-Klavehn, 2000, p. 235). In the present study, the to-be remembered events were encountered only once in an incidental study phase: children were not informed of the later testing phase. In these conditions, perceptual processing could be favored with some restriction and fostering in turn a priming process: when the to-be remembered event is easy, such a process could affect all children; when the to-be remembered event is difficult, it would affect specifically the late recognizers confronted with a perceptual challenge (e.g., a change of color between study and test). This is plausible since no memory task is process pure (Jacoby, 1991; Rajaram, 1993; Rovee-Collier, 1997). In sum and across sequences, late recognizers’ declarative memories would be semantic-like rather than episodic. This account is also supported by storage and/or retrieval failures hypothesis that Howe and Courage (1997)
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
143
and Bauer et al. (2000) stand for, especially when color of the props in testing phase differed from modeling. In sum, the present findings may show the role of cognitive self as a factor of differentiation between semantic and episodic subsystems in early declarative memory system. Nevertheless, a few children failed the mirror test but succeeded in recalling the events when the color of the props changed between study and test. Possible explanations are that the cognitive self-development is not an all-or-none achievement (Rochat, 2001), not the only factor that contribute to early declarative memory development and not the only one factor that the mirror test may measure (Harley & Reese, 1999; see Povinelli, 1995 for discussion). Some children could enter the adult-like declarative memory through other ways than selfrecognition achievement and linguistic means could prevail over self-recognition for late recognizers. To explore this possibility, the role of cognitive-self in early declarative memory needs to be investigated controlling general representational- and language-abilities in deferred imitation task. These conclusions also provide motivation for future research on larger samples and longer recall delays. The case of the few children who failed the mirror test and succeeded in the imitation task is of particular interest in order to precise the role of cognitive self in early declarative memory.
Appendix A Make a rattle (two half of a plastic barrel, one wooden cube): the experimenter modeled put the cube in one half of the barrel, cover with the other half (to close the barrel), shake the barrel. Make the car go (a wooden base, a wooden slope in two parts attached by a hinge, a small plastic car): the experimenter modeled unfold the slope, put the slope on the base (to form a hill), put the car at the top of the hill (resulting in the car rolling down).
References Asendorpf, J. B., & Baudonniere, P. M. (1993). Self-awareness and other-awareness: Mirror self-recognition and synchronic imitation among unfamiliar peers. Developmental Psychology, 29(1), 88–95. Barnat, S. B., Klein, P. J., & Meltzoff, A. N. (1996). Deferred imitation across changes in context and object: Memory and generalization in 14-month-old infants. Infant Behavior and Development, 19(2), 241–251. Barr, R., & Hayne, H. (1996). The effect of event structure on imitation in infancy: Practice makes perfect? Infant Behavior and Development, 19(2), 253–257. Bauer, P. J. (1995). Recalling past events: From infancy to early childhood. In R. Vasta (Ed.), Annals of child development: A research annual: Vol. 11 (pp. 25–71). London, England: Jessica Kingsley Publishers Ltd. Bauer, P. J. (1997). Development of memory in early childhood. In N. Cowan (Ed.), The development of memory in childhood. Studies in developmental psychology: (pp. 83–111). Howe, England: Psychology Press/Erlbaum (UK) Taylor & Francis. Bauer, P. J., & Dow, G. A. (1994). Episodic memory in 16- and 20-month-old children: Specifics are generalized but not forgotten. Developmental Psychology, 30(3), 403–417. Bauer, P. J., & Fivush, R. (1992). Constructing event representations: Building on a foundation of variation and enabling relations. Cognitive Development, 7(3), 381–401. Bauer, P. J., Hertsgaard, L. A., Dropik, P., & Daly, B. P. (1998). When even arbitrary order becomes important: Developments in reliable temporal sequencing of arbitrarily ordered events. Memory, 6(2), 165–198. Bauer, P. J., Van Abbema, D. L., & de Haan, M. (1999). In for the short haul: Immediate and short-term remembering and forgetting by 20-month-old children. Infant Behavior and Development, 22(3), 321–343.
144
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
Bauer, P. J., Wenner, J. A., Dropik, P. L., & Wewerka, S. S. (2000). Parameters of remembering and forgetting in the transition from infancy to early childhood. Monographs of the Society for Research in Child Development, 65(4), 1–204 (Serial No. 263). Butterworth, G. (1990). Self-perception in infancy. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to childhood. The John D. and Catherine T. MacArthur foundation series on mental health and development (pp. 119–137). Chicago, IL, USA: University of Chicago Press. Gardiner, J. M., & Richardson-Klavehn, A. (2000). Remembering and knowing. In E. Tulving & F. Craik (Eds.), The Oxford handbook of memory (pp. 229–244). Oxford: University Press. Harley, K., & Reese, E. (1999). Origins of autobiographical memory. Developmental Psychology, 35(5), 1338–1348. Hayne, H., Boniface, J., & Barr, R. (2000). The development of declarative memory in human infants: Age related changes in deferred imitation. Behavioral Neuroscience, 114(1), 77–83. Hayne, H., MacDonald, S., & Barr, R. (1997). Developmental changes in the specificity of memory over the second year of life. Infant Behavior and Development, 20(2), 233–245. Howe, M. L., & Courage, M. L. (1997). The emergence and early development of autobiographical memory. Psychological Review, 104(3), 499–523. Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30(5), 513–541. Kagan, J. (1981). The second year: the emergence of self-awareness. Cambridge, MA: Harvard University Press. Lewis, M. (1999). Social cognition and the self. In P. Rochat (Ed.), Early social cognition understanding others in the first months of life (pp. 81–100). Mahwah, NJ: Lawrence Erlbaum Associates. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum Press. Mandler, J. M. (1990). Recall of events of preverbal children. In A. Diamond (Ed.), The development and neural bases of higher cognitive functions: 608 (pp. 485–516). New York: Annals of the New York Academy of Sciences. McDonough, L., Mandler, J. M., McKee, R. D., & Squire, L. R. (1995). The deferred imitation task as a nonverbal measure of declarative memory. Proceedings of the National Academy of Sciences of the United States of America, 92, 7580–7584. Meltzoff, A. N. (1990). Towards a developmental cognitive science: The implications of cross-modal matching and imitation for the development of representation and memory in infancy. Annals of the New York Academy of Sciences, 608, 1–37. Neisser, U. (Ed.). (1993). The perceived self: Ecological and interpersonal sources of self-knowledge. New York, NY, USA: Cambridge University Press. Nelson, K. (1993). Explaining the emergence of autobiographical memory in early childhood. In A. F. Collins (Ed.), Theories of memory (pp. 355–385). Hove, UK: Lawrence Erlbaum Associates Inc. Perner, J., & Ruffman, T. (1995). Episodic memory and autonoetic consciousness: Developmental evidence and a theory of childhood amnesia. Journal of Experimental Child Psychology, 59(3), 516–548. Povinelli, D. J. (1995). The unduplicated self. In P. Rochat (Ed.), Advances in psychology: Vol. 112. The self in infancy: Theory and research (pp. 161–192). Amsterdam: North-Holland-Elsevier. Rajaram, S. (1993). Remembering and knowing: Two means of access to the personal past. Memory and Cognition, 21(1), 89–102. Rochat, P. (2001). Origins of self-concept. In G. Bremmer & A. Fogel (Eds.), Blackwell handbook of infant development. Handbooks of developmental psychology (pp. 191–212). Malden, USA: Blackwell Publishers. Rovee-Collier, C. (1997). Dissociations in infant memory: Rethinking the development of implicit and explicit memory. Psychological Review, 104(3), 467–498. Tulving, E. (1983). Elements of episodic memory. New York: Oxford University Press. Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of Psychology, 53(1), 1–25. Tulving, E., & Markowitsch, H. J. (1998). Episodic and declarative memory: Role of the hippocampus. Hippocampus, 8(3), 198–204. Wheeler, M. A. (2000). Episodic memory and autonoetic awareness. In E. Tulving & F. Craik (Eds.), The Oxford handbook of memory (pp. 597–608). Oxford: Oxford University Press. Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic memory: The frontal lobes and autonoetic consciousness. Psychological Bulletin, 121(3), 331–354.
Early declarative memory and self-concept Nathalie Prudhomme ∗ Universit´e Ren´e Descartes – PARIS V, Institut de Psychologie, Laboratoire Cognition et D´eveloppement (CNRS UMR 8605), 71 Avenue Edouard Vaillant, F-92774 Boulogne Billancourt Cedex, France Received 1 February 2003; received in revised form 3 December 2004; accepted 3 January 2005
Abstract Infantile amnesia is a major issue in memory development. For a growing body of researchers, self is probably a key-concept for this enigma. The role of the cognitive self in early declarative memory as a function of perceptual manipulation between study- and test-phase was investigated. Two recall conditions were compared: either the same color props as in the modeling phase or different color props. Twenty-four 20-month-old children were observed in a 10-min deferred imitation task with two enabling sequences. An additional control group (N = 16) ensured that imitation scores did not rely on spontaneous production of the target behaviors at this age. Children in the experimental groups were classified as early or late recognizer depending on their success or failure to the mirror test. A significant interaction was observed between color props conditions and recognizer types: only late recognizers’ memory performance was affected by the perceptual manipulation. Results suggest that cognitive self could be one of the factors contributing to the differentiation between episodic and semantic subsystems in early declarative memory by the end of the second year. © 2005 Elsevier Inc. All rights reserved. Keywords: Declarative memory; Self; Toddler; Deferred imitation
1. Introduction A growing body of studies shows, mainly from differed imitation paradigms, that declarative memory emerges around 6 months of age. However, first autobiographic memories do not appear before the age of ∗
Tel.: +33 1 55 20 59 97; fax: +33 1 55 20 59 85. E-mail address: [email protected].
0163-6383/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.infbeh.2005.01.002
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
133
3 years. Several researchers consider self-awareness or cognitive self as probably one of the keys to the long delay in autobiographic memory appearance. The aim of this study is to explore the role of cognitive self in declarative memory development. Autobiographical memory allows retrieving experienced events as personally experienced (Bauer, Wenner, Dropik, & Wewerka, 2000; Howe & Courage, 1997; Nelson, 1993; Perner & Ruffman, 1995; Wheeler, Stuss, & Tulving, 1997). Such memory depends on episodic memory (Tulving, 2002; Tulving & Markowisch, 1998; Wheeler et al., 1997) and, like semantic memory, is a subsystem of the declarative memory system (Tulving, 1983, 2002). Episodic memory is considered as emerging later than semantic memory and stems from “the conjunction of three concepts—self, autonoetic awareness, and subjectively sensed time” (Tulving, 2002, p. 5). The sense of subjective time, possible only with autonoetic awareness, opens the ability to mentally travel in time from present to experienced past (and/or to the imagined future). Autonoetic awareness differs from noetic awareness because it allows encoding facts as personally experienced, and recollecting them, as they were experienced. The self is the traveler, existing in subjective time. In this perspective, it is a key concept for the emergence of autobiographic memory. For Howe and Courage (1997), the cognitive self or self-concept emerging by the end of the second year of life sets the lower limit for early autobiographical memories. The cognitive self is defined as a new knowledge structure that serves to organize memories of experiences that happened to me. These authors propose a memory model in which traces, integrated during the learning phase,1 consist of collections of primitive elements. Once the self is “viable”, its features can be sampled and encoded in traces following a probabilistic process: “as more features are added to the ‘urn’, the greater the likelihood that at least some self features are sampled and encoded in the functional trace for an event” (Howe & Courage, p. 513). A memory trace that includes some features of the self potentially could change into an autobiographical memory and, in turn, contribute to the development of cognitive self. In this model, the probabilistic process of sampling self-features is not specific to the cognitive self but also serves for sampling primitive elements from all knowledge structures integrated into a new memory trace. Thus, autobiographic memories differ from other memories because of the contribution of cognitive self and, consequently, depend upon the developmental level of the self. The development of the self has been of particular interest for many researches during the past 10 years. Two aspects of the self are generally distinguished: the ‘I’ as subjective or implicit sense of one’s own physical and social entity; and the ‘me’ as objective or cognitive sense of self. Recent research focuses on the subjective self considered as either innate or developing from birth through infants’ interactions with their physical and social world. For Rochat (2001) and Lewis (1999), it is not specifically human and it can be described as “an existential sense of self” (Rochat, 2001, p. 204). The cognitive self, assumed to emerge from the subjective self by the end of the second year, corresponds to the self as object of experience, knowledge, and imagination, i.e., me. This sense of self is probably specific to human beings (Lewis, 1999) and the mirror task can serve to evaluate its development in toddlers. Although limited to a particular experience and sensory modality (visual image), the mirror test is considered as “a valid instrument to assess self-knowledge at a conceptual and re-cognitory level” (Rochat, 2001, p. 206). It consists of applying surreptitiously an odorless red mark on the infant’s nose to see if he demonstrates a self-directed behavior in front of a mirror by touching his nose instead of the mirror. For Wheeler 1
And possibly disintegrated/reintegrated across retention intervals and testing phases.
134
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
et al. (1997), this behavior “involves an awareness of one’s physical appearance and the constancy of that appearance throughout time and space” (p. 344) and thus refers to the self as associated to episodic memory in Tulving’s theory. For many authors (Asendorpf & Baudonniere, 1993; Butterworth, 1990; Kagan, 1981; Lewis, 1999; Meltzoff, 1990; Neisser, 1993; Rochat, 2001) to pass the mirror test “marks an important developmental milestone in the second year of life” (Howe & Courage, 1997, p. 506) and would be linked to event memory by the end of the second year (Howe & Courage, 1997). To test this hypothesis, Harley and Reese (1999) computed the correlation between 19-month-olds’ performances in mirror task and deferred imitation. Indeed, they reported a small but significant positive correlation (r = .32, p < .05) between the recalled temporal information and self-recognition. This correlation questions the role of cognitive self in early declarative memory as evidenced by deferred imitation. The deferred imitation test is a non-verbal test of declarative memory (Bauer, 1995; Howe & Courage, 1997; Mandler, 1990; McDonough, Mandler, McKee, & Squire, 1995), used to study event memory in infancy. In this test, new props are used to show to the child specific new actions or action sequences. After a retention interval designed following the aim of the studies, the props are given to the child who then may reproduce the target actions previously demonstrated. In this procedure, children are not allowed to explore the props before the testing phase. For Wheeler (2000), young children’s deferred imitation performances are not evidence of episodic but only of a declarative memory that is probably semantic-like. Under 3–4 years of age, children “are without the capacity to recollect their past in the rich, personal way that comprises episodic retrieval” (Wheeler, 2000, p. 603) because of their insufficiently developed self-awareness. For Bauer (1997), the deferred imitation paradigm potentially taps on episodic memory for at least two reasons: (1) it allows observing the recall of specific, new, previously and briefly exposed events and (2) performances are flexible, e.g., not tied to a specific modality or learning context. Several studies, focusing on flexibility in deferred imitation with toddlers, revealed developmental changes in memory’s specificity over the second year of life; they also showed that, by the end of the second year, children easily generalize their event memories to different colors and/or forms’ props, at least considering number of target actions recalled (Barnat, Klein, & Meltzoff, 1996; Bauer & Dow, 1994; Hayne, Boniface, & Barr, 2000; Hayne, MacDonald, & Barr, 1997). In short, by the end of the second year deferred imitation paradigm allows young children to form and retrieve declarative event memories that would be either semantic-like or episodic-like. At this age, the emergence of the cognitive self, as a new knowledge structure that serves to mediate personally experienced events, may contribute to encode and recollect these event memories in a richer and more flexible way. Specifically, development of the cognitive self may be linked with memories’ flexibility around two years of age. To test this hypothesis, the mirror self-recognition test was used to assess the cognitive self-development with 20-month-old infants. This age was chosen so as to have as many children who pass the self-recognition test and as many who do not. Children were then tested in a deferred imitation task with a 10 min recall delay ensuring that recall tapped long-term memory rather than working memory (Bauer, Van Abbema, & de Haan, 1999). For half of the group, the color of the props was changed in the test phase to examine a possible interaction between color change and success or failure to the mirror task. If the cognitive self contribute to a greater flexibility of event memories, deferred imitation scores would be affected by perceptual manipulations between study and test phases only for the children who failed the mirror test.
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
135
2. Method 2.1. Participants Twenty-four children (12 boys) were recruited by mail from the birth records of the city hall of Paris. Mean age2 was 20 months (m = 601 days; S.D. = 10.9). A control group composed of 16 children (eight boys), mean age: 20 months (m = 600 days; S.D. = 8.7), drawn from the same population, ensured that performance observed in the experimental groups was based on the sequence representation, not on the spontaneous occurrence of the target behaviors. All children were observed individually in the laboratory for a 20-min session and received small gifts at the end of the session. 2.2. Tasks, procedure, and scoring After a brief warm-up period, the child and the experimenter sat around a child-size table. The child’s parent remained in the room, seated on a sofa where the child could see him/her, and was asked to refrain from acting or speaking for the rest of the session. On the floor, close to the child’s chair, several unrelated toys were available to the child. A 100 cm × 80 cm covered mirror, fixed on an easel 30 cm from the floor, was placed near the table. 2.2.1. Deferred imitation The children of the experimental groups were tested for a 10-min deferred recall test on two enabling three-steps sequences. These two sequences were chosen among a pool of sequences commonly used with 20-months-old: making the car go and making the rattle (see Appendix A). Each sequence had two sets of props differing only on the color. For all children, sequences were modeled with the same set of props. Children were placed in an incidental learning condition. The experimenter invited the child to a free play with unrelated toys of his/her choice. When the child was both sufficiently engaged in his/her play and ready to attend to sequence modeling, the experimenter modeled the two sequences without comments, twice in succession. During this modeling phase, the experimenter verified that the child was watching the demonstration. After each demonstration, the props were immediately removed from the child’s sight and the 10-min recall delay started immediately. Children were not permitted to reproduce the two sequences prior to this delay. Unrelated distracting toys helped to increase the child’s tolerance for the procedure. Half of the children were requested to recall the sequences with the color props used for the demonstrations (we refer to this condition as SC), while the other half had to recall the sequences with different color props (we refer to this condition as DC). The order of presentation of the two sequences was counterbalanced across children, with sequences introduced in the same order during modeling and test for each child. During the recall delay, the experimenter enrolled the child in unrelated free play and submitted the child to the mirror test of self-recognition. When the recall delay was over, the experimenter gave the props for each sequence in turn to the child with an explicit verbal instruction to reproduce what she had modeled 10 min before: “Look, I am lending you my game. Can you show me what I did with all this stuff?” Note that the experimenter never provided a verbal reminder such as the verbal label of the sequences, props or target actions. 2
See Table 1 for detailed mean age.
136
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
Children in the control group were enrolled in a free play session with unrelated toys. After 5–10 min of play, the experimenter gave to the child one set of props, for each sequence in turn. During this childcontrolled phase, the experimenter encouraged him/her to explore and manipulate the props by prompting such as: “Look, what can you do with all this stuff?”
2.2.2. Mirror test of self-recognition During the recall delay, the experimenter placed odorless red face paint on the child’s nose in the guise of wiping it with a paper handkerchief. After 2 min of play with unrelated toys, the experimenter drew the child’s attention to the covered mirror and removed the mirror cover. The criterion of selfrecognition was a child’s self-directed behavior such as nose or face touching (Lewis & Brooks-Gunn, 1979). The mirror test was considered as passed if the child touched his/her face within 2 cm of the paint (Harley & Reese, 1999). Children who passed the mirror test (N = 12) were classified as early recognizers (referred as R+). Children who failed the mirror test (N = 12) were classified as late recognizers (referred as R−). The random assignments of each child to one of the prop conditions (SC or DC) and to one order of presentation of the sequences were decided before knowing if he/she was an early or late recognizer. This procedure allowed getting equal numbers of girls and boys in each prop condition and equal numbers of children in each order of presentation. However, the four experimental groups resulting from crossing recognizers’ type (R+; R−) and prop conditions (SC; DC) were not equated for the number of children (R+/SC and R−/DC groups had N = 7; R+/DC and R−/SC had N = 5; see Table 1).
2.2.3. Data reduction All sessions were videotaped for later analysis. All target behaviors, including target actions in each sequence and self-directed behavior in the mirror test situation, were examined. A scorer, unaware of the hypotheses under investigation, coded all sessions. She was trained until she reached an agreement of 90% or more with the experimenter on three successive children from a previous corpus of deferred imitation data. Bauer, Hertsgaard, Dropik, and Daly’ (1998) method of scoring deferred imitation coding was applied. For each sequence, the total number of different target actions produced (max = 3 for each sequence) and the number of pairs of actions produced in the target order (max = 2 for each sequence) were computed for each child. Only the first occurrence of each target action was taken into account. For example, on Make a rattle, if a child produced all three components in the target order, he/she would receive credit for the three different target actions and for the two correctly ordered pairs of actions: put the cube in the half of the container (action 1), cover it with the other half (action 2), and shake the rattle (action 3). If a child only produced actions 1 and 3 in that order, one point would be credited for the number of ordered pairs of actions and two points for the target actions. If a child produced the string of actions 3, 1, 2, 3, he/she would still receive credit for the production of three different target actions. However, the child would only be credited with one correctly ordered pair of actions (i.e., 1–2) but would not be credited with the correctly ordered pair 2–3, because action 3 was performed first. This scoring procedure reduces the likelihood of a child receiving credit for correctly ordered pairs of actions by chance or by trial and error (Bauer et al., 1998). Finally, all children coded as showing a self-directed behavior in the mirror test were classified as R+ while others were classified as R−.
Sequences
Number of target actions (max = 3) Car
Number of pairs of actions in target order (max = 2) Number of target actions (max = 3)
Rattle
Number of pairs of actions in target order (max = 2)
Early recognizers (R+)
Late recognizers (R−)
SC (N = 7, 604a )
DC (N = 5, 599a )
SC (N = 5, 602a )
DC (N = 7, 598a )
m (σ) t-value m (σ) t-value
2.14 (0.69) t(21) = 4.09*** 1.00 (0.67) t(21) = 3.24**
2.60 (0.89) t(19) = 4.51*** 1.40 (0.89) t(19) = 4.34***
1.80 (0.45) t(19) = 2.85** 0.60 (0.55) t(19) = 1.84, ns
1.14 (0.90) t(21) = 1.07, ns 0.43 (0.53) t(21) = 1.20, ns
m (σ) t-value m (σ) t-value
2.71 (0.76) t(21) = 5.32*** 1.71 (0.76) t(21) = 5.50***
3.00 (0.00) t(19) = 5.82*** 2.00 (0.00) t(19) = 7.32***
3.00 (0.00) t(19) = 5.82*** 2.00 (0.00) t(19) = 7.32***
1.43 (1.51) t(21) = 1.49, ns 0.86 (1.07) t(21) = 2.01*
t values result from the comparison of experimental groups with the control group. a Mean age (in days). ∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.
Control group (N = 16, 600a ) 0.75 (0.69) 0.19 (0.40) 0.69 (0.87) 0.19 (0.54)
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
Table 1 Mean numbers of target actions recalled and pairs of actions produced in the target order and S.D.
137
138
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
3. Results 3.1. Preliminary analyses The overall mean recall delay for both sequences and for the four experimental groups was 11 min 43 s (range = 10 min 52 s to 12 min 56 s). Preliminary analyses revealed no significant group difference between sequences for the four experimental groups variables, no gender effect and no effect of order of presentation of the sequences on both dependent variables (mean number of target actions produced and mean number of correctly ordered pairs of actions). Thus, gender and order independent variables were excluded from subsequent analyses.
3.2. Deferred imitation and self-recognition 3.2.1. Comparisons between experimental and control groups Descriptive statistics for target actions and pairs of actions produced in the target order for all groups are presented in Table 1. Experimental groups’ performance as a whole was compared to the control group’s performance for both dependent variables with a 2 (conditions: experimental/control) by 2 (sequences: rattle/car) mixed ANOVA. For the mean numbers of target actions produced, a significant effect of condition was observed: F(1,38) = 33.52, p < .00001. Children in the experimental condition had significantly higher performances than those in the control condition. Neither sequence nor interaction effects were observed. For the mean number of correctly ordered pairs of actions produced, significant main effects of condition (F(1,38) = 30.67, p < .0001) and sequence (F(1,38) = 9.12, p < .01) were observed. Interaction between conditions and sequences was also significant: F(1,38) = 9.12, p < .01. Post hoc comparisons with Tukey HSD test showed that (1) children in experimental conditions produced significantly more correctly ordered pairs of actions across sequences than those in the control condition, and (2) in experimental conditions, children produced significantly more correctly ordered pairs of actions for the rattle than they did for the car. To investigate whether early and late recognizers were equally able to recall specific event sequences after a 10-min delay, depending on whether or not the props color was changed, we compared each experimental groups’ performances with the control group’s performances with a Student’s t-test for both dependent measures and each sequence (see Table 1). Considering the number of target actions, both early and late recognizers in the SC condition always performed significantly better than the control subjects. In the DC condition and for both sequences, only early recognizers produced significantly more target actions than the control subjects. Considering the number of correctly ordered pairs of actions, the children in the R+ groups always had significantly higher scores than the controls. In contrast, late recognizers did not show significantly higher scores than control subjects except in the SC condition and for the rattle sequence only. These results showed that early and late recognizers were not equally able to recall specific event sequences after a 10-min delay, depending on color props condition, sequences and dependent measures. Nevertheless, for both dependent measures and both sequences, children in the experimental groups as a whole always displayed significantly higher performances than did children in the control group, showing evidence of deferred recall. Further analyses concern experimental groups only.
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
139
Fig. 1. Interaction between recognizer type and color prop condition. ∗ p < .05; ** p < .01.
3.2.2. Effects of self-recognition and color prop condition To ensure that the two sequences show the same pattern of results and considering the number of target actions, a 2 (recognizer type: R−/R+) by 2 (color prop condition: SC/DC) mixed analysis of variance (ANOVA) was performed with sequences as a within-subjects variable. Two significant main effects were observed. Children recalled more target actions for the rattle than they did for the car sequence: F(1,20) = 11.63, p < .003. Early recognizers recalled more target actions than did late recognizers: F(1,20) = 6.5, p < .02. No main effect of color prop condition was observed but a significant interaction between recognizers’ type and color prop was found: F(1,20) = 6.05, p < .05. In sum, no significant interaction with sequences was observed. According to these results, Make a car go was always less recalled than Make a rattle. Analysis of individual data showed that most children had difficulties with the second target action of the car sequence (see Appendix A). As a consequence, this sequence was more difficult to encode and/or to recall. Nevertheless, the same pattern of results was observed for each sequence, since no reliable interaction was observed with sequences (see Fig. 1). Consequently, results for each sequence were collapsed (see Table 2) and a 2 (recognizer type: R+/R−) by 2 (color prop condition: SC/DC) mixed ANOVA was performed to examine the effect of self-recognition as a function of color prop condition. As previously, recognizers’ type shows a significant main effect (early recognizers recalled more target actions than did late recognizers: F(1,20) = 6.52, p < .05), no reliable main effect of color prop condition was observed (F(1,20) = 1.51, ns) and a significant interaction between recognizers’ type and color prop was found (F(1,20) = 6.05, p < .05). This interaction was analyzed with separate one-way
140
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
Table 2 Mean numbers of target actions recalled and pairs of actions produced in the target order for experimental groups when sequences are collapsed
Number of pairs of actions in target order (max = 4) t values ∗
Early recognizers (R+) m(σ) t value
Late recognizers (R−) m(σ) t value
All experimental groups m(σ)
Control group m(σ)
SC (N = 7)
S (N = 5)
(N = 24)
(N = 16)
2.42 (1.35)
0.38 (0.94)
DC (N = 5)
DC (N = 7)
2.71 (1.38) 3.40 (0.89) 2.60 (0.55) 1.29 (1.38) t(21) = 5.40*** t(19) = 7.78*** t(19) = 6.33*** t(21) = 2.10∗
p < .05; *** p < .01; *** p < .001.
ANOVA. Analyses of the main effect revealed that in the SC condition, performance of the early and late recognizers did not differ significantly (F(1,20) < 1). In contrast, in the DC condition, early recognizers recalled significantly more target actions than did the late recognizers: F(1,20) = 12.57, p <.01. Moreover, performance of the early recognizers did not differ significantly across color prop conditions (F(1,20) < 1), while the late recognizers recalled significantly more target actions in the SC than in the DC condition: F(1,20) = 6.81, p < .05. The late recognizers were thus more sensitive than the early recognizers to the perceptual characteristics of the props available during the testing phase, indicating that the latter can benefit from event memories more flexible than the former. Considering numbers of correctly ordered pairs of actions produced, and since this measure is not independent of the number of target actions recalled, we first examined the correlations between the two dependent variables, for each sequence and for the experimental groups as a whole. Indeed, the two measures were highly linked: r = .86 for the car and r = .98 for the rattle. Consequently, it was unnecessary to perform a 2 (recognizers’ type: R+/R−) by 2 (color prop conditions: SC/DC) mixed ANOVA for the recall of temporal order, which will mirror the results on mean numbers of target actions recalled (see Fig. 1). These high correlations also precluded a covariance analysis across recognizers’ type and color prop conditions using the absolute number of target actions recalled as covariate (see Barr & Hayne, 1996 for such an analysis). Furthermore, a formal test is usually applied to compare the observed- to the expected by chancerecall of temporal information (Bauer & Dow, 1994; Bauer & Fivush, 1992). This test must be used for each sequence separately. In our experiment, all children in the R−/SC and R+/DC groups produced a complete recall in the target order for the rattle. In the R−/DC group, only three children recalled at least two actions for each sequence and always did it in the correct order (see Table 3). These results preclude the use of the formal test such as indicated above. Data on the recall of temporal information could only be examined considering individual performances. Table 3 shows the number of children who recalled three or two actions in order, or one or no action, for each sequence and each experimental group. For the difficult sequence (i.e., car sequence), it appears that the few late recognizers who recalled two actions in the DC condition always did it in the correct order while the wrong order was observed in all three other groups (i.e., R−/SC, R+/SC, and R+/DC). For the rattle sequence and across recognizers’ type and color prop condition, all children recalling more than one action indeed recalled all the sequence in the target order. Therefore, when the to-be remembered event was easy (i.e., rattle sequence), all observed recalls were all-or-none retrievals; when the to-be remembered event was difficult (i.e., car sequence), only late
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
141
Table 3 Number of children by actions recalled as a function of sequence, recognizers’ type, and color prop condition The car (rAP = .86a ) R+
The rattle (rAP = .98a ) R−
R+
R−
SC
DC
SC
DC
SC
DC
SC
DC
Three actions recalled Three in order Two in order
2 2
4 3 1
0
0
6 6
5 5
5 5
3 3
Two actions recalled In order In disorder
4 3 1
0
4 2 2
3
0 3
0
0
0
One action recalled No action recalled
1 0
1 0
1 0
2 2
1 0
0 0
0 0
1 3
N total
7
5
5
7
7
5
5
7
a
Correlation between number of individual target actions recalled (A) and number of pairs of actions correctly ordered (P).
recognizers in the DC condition displayed all-or-none retrievals while all other children displayed partial retrievals.
4. Discussion The aim of this study was to assess the role of cognitive self in early declarative memory as a function of perceptual manipulation between study and test in a 10-min deferred imitation task. This question relies on Howe and Courage’s (1997) theory which argues that the cognitive self could act on early declarative memory, allowing more elaborative processing through personal features embedded in the memory traces and/or allowing more flexibility in memory. Two major conclusions can be drawn from the target actions recalled: (1) the different-color props condition did not disrupt nor alter early recognizers’ event memory retrievals; (2) it did so for more than half of the late recognizers whatever the difficulty of the to-be remembered event. In our experiment, early and late recognizers as a whole generalized their event memories to different color props, which is congruent with previous studies (Barnat et al., 1996; Bauer & Dow, 1994; Hayne et al., 1997). Moreover, early and late recognizers displayed equivalent performances when the color of props did not change between study and test. Nevertheless, when the color of props changed between study and test, a different pattern of results emerged for early and late recognizers: while most of the latter failed in generalizing their memories, the former always succeeded. These results argue for the role of cognitive self in early declarative memory. Cognitive self may prompt young children’s generalization of specific event memories to different color props over a 10-min delay. When the cognitive self is available, declarative memories could be more flexible and retrieved from identical- as well as different-perceptual cues. In the absence of cognitive self, retrieving could be more dependant on perceptual cues or fluency processing and be disrupted when the available cues perceptively differ from those previously encoded. Nevertheless, it is probably not the only way in which the cognitive self could act on early declarative memory.
142
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
The emergence of a cognitive self as measured by the mirror test may be associated with better representational abilities: across sequences and color prop conditions, early recognizers had better performances than late recognizers. When the cognitive self is available, children could form better organized and detailed declarative memories. This may be identified through the recall of temporal information, because critical information about the order in which the event should unfold is not perceptually cued at testing time: the available props do provide perceptual support to recall target actions but surely not to recall temporal information. To do this, infants have to encode order of the events during demonstration and retrieve it from a representation of the events and in absence of ongoing perceptual support. Unfortunately in the present experiment, sample size and high correlations observed between the two dependent measures do not allow appreciating recall of temporal information independently of recalled specific items. Finally, cognitive self could foster both representational abilities and memories’ flexibility. However, representational abilities and memories’ flexibility may develop jointly or independently. In our experiment, as we did not reach children’s representational abilities per se, it is not possible to decide whether cognitive self prompt general representational abilities, memories’ flexibility or both. In terms of specific items information recalled, the present findings show indeed that the cognitive self is linked to early declarative memory development at least through memories’ flexibility. Now the question is: how does the cognitive self contribute to a greater flexibility in early declarative memories? Following Howe and Courage’s theory (1997), children who succeeded the mirror test could form event memories organized not only as a function of context, categorization, spatial location, object knowledge and so on, but also as a function of me. Therefore, such declarative memories partly mediated by the self could benefit from a more detailed and organized encoding. With only few personal features available for embedding in the memory traces, processing could be more elaborative and memories would be episodic-like rather than semantic. As a consequence, children could recollect these memories in a richer and more personal way that makes them less sensitive to external perceptual features cueing and/or constraining the retrieval. Moreover, the explicit instruction to recall what was witnessed 10 min before could favor episodic retrieval as it referred to personally experienced events. Such recalls could be the first expressions of an emergent episodic subsystem in early declarative memory. In the same way, it could be argued that the majority of the children who failed the mirror test could not form episodic-like memories because they did not reach a “sufficiently viable” cognitive self (Howe & Courage, 1997) and lacked some personal features to be embedded in the memory traces. These children would reach less elaborated and/or more perceptual encoding of the to-be remembered events. Furthermore, if the available cues can instantaneously match encoded items, a priming process could underlie children’s retrieving. In adults, Rajaram (1993) showed that identical primes between study and test increased “know responses”, which are associated with semantic memory. Moreover, “fluency of processing may seem more plausible when it refers to a single previous encounter with one studied item” (Gardiner & Richardson-Klavehn, 2000, p. 235). In the present study, the to-be remembered events were encountered only once in an incidental study phase: children were not informed of the later testing phase. In these conditions, perceptual processing could be favored with some restriction and fostering in turn a priming process: when the to-be remembered event is easy, such a process could affect all children; when the to-be remembered event is difficult, it would affect specifically the late recognizers confronted with a perceptual challenge (e.g., a change of color between study and test). This is plausible since no memory task is process pure (Jacoby, 1991; Rajaram, 1993; Rovee-Collier, 1997). In sum and across sequences, late recognizers’ declarative memories would be semantic-like rather than episodic. This account is also supported by storage and/or retrieval failures hypothesis that Howe and Courage (1997)
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
143
and Bauer et al. (2000) stand for, especially when color of the props in testing phase differed from modeling. In sum, the present findings may show the role of cognitive self as a factor of differentiation between semantic and episodic subsystems in early declarative memory system. Nevertheless, a few children failed the mirror test but succeeded in recalling the events when the color of the props changed between study and test. Possible explanations are that the cognitive self-development is not an all-or-none achievement (Rochat, 2001), not the only factor that contribute to early declarative memory development and not the only one factor that the mirror test may measure (Harley & Reese, 1999; see Povinelli, 1995 for discussion). Some children could enter the adult-like declarative memory through other ways than selfrecognition achievement and linguistic means could prevail over self-recognition for late recognizers. To explore this possibility, the role of cognitive-self in early declarative memory needs to be investigated controlling general representational- and language-abilities in deferred imitation task. These conclusions also provide motivation for future research on larger samples and longer recall delays. The case of the few children who failed the mirror test and succeeded in the imitation task is of particular interest in order to precise the role of cognitive self in early declarative memory.
Appendix A Make a rattle (two half of a plastic barrel, one wooden cube): the experimenter modeled put the cube in one half of the barrel, cover with the other half (to close the barrel), shake the barrel. Make the car go (a wooden base, a wooden slope in two parts attached by a hinge, a small plastic car): the experimenter modeled unfold the slope, put the slope on the base (to form a hill), put the car at the top of the hill (resulting in the car rolling down).
References Asendorpf, J. B., & Baudonniere, P. M. (1993). Self-awareness and other-awareness: Mirror self-recognition and synchronic imitation among unfamiliar peers. Developmental Psychology, 29(1), 88–95. Barnat, S. B., Klein, P. J., & Meltzoff, A. N. (1996). Deferred imitation across changes in context and object: Memory and generalization in 14-month-old infants. Infant Behavior and Development, 19(2), 241–251. Barr, R., & Hayne, H. (1996). The effect of event structure on imitation in infancy: Practice makes perfect? Infant Behavior and Development, 19(2), 253–257. Bauer, P. J. (1995). Recalling past events: From infancy to early childhood. In R. Vasta (Ed.), Annals of child development: A research annual: Vol. 11 (pp. 25–71). London, England: Jessica Kingsley Publishers Ltd. Bauer, P. J. (1997). Development of memory in early childhood. In N. Cowan (Ed.), The development of memory in childhood. Studies in developmental psychology: (pp. 83–111). Howe, England: Psychology Press/Erlbaum (UK) Taylor & Francis. Bauer, P. J., & Dow, G. A. (1994). Episodic memory in 16- and 20-month-old children: Specifics are generalized but not forgotten. Developmental Psychology, 30(3), 403–417. Bauer, P. J., & Fivush, R. (1992). Constructing event representations: Building on a foundation of variation and enabling relations. Cognitive Development, 7(3), 381–401. Bauer, P. J., Hertsgaard, L. A., Dropik, P., & Daly, B. P. (1998). When even arbitrary order becomes important: Developments in reliable temporal sequencing of arbitrarily ordered events. Memory, 6(2), 165–198. Bauer, P. J., Van Abbema, D. L., & de Haan, M. (1999). In for the short haul: Immediate and short-term remembering and forgetting by 20-month-old children. Infant Behavior and Development, 22(3), 321–343.
144
N. Prudhomme / Infant Behavior & Development 28 (2005) 132–144
Bauer, P. J., Wenner, J. A., Dropik, P. L., & Wewerka, S. S. (2000). Parameters of remembering and forgetting in the transition from infancy to early childhood. Monographs of the Society for Research in Child Development, 65(4), 1–204 (Serial No. 263). Butterworth, G. (1990). Self-perception in infancy. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to childhood. The John D. and Catherine T. MacArthur foundation series on mental health and development (pp. 119–137). Chicago, IL, USA: University of Chicago Press. Gardiner, J. M., & Richardson-Klavehn, A. (2000). Remembering and knowing. In E. Tulving & F. Craik (Eds.), The Oxford handbook of memory (pp. 229–244). Oxford: University Press. Harley, K., & Reese, E. (1999). Origins of autobiographical memory. Developmental Psychology, 35(5), 1338–1348. Hayne, H., Boniface, J., & Barr, R. (2000). The development of declarative memory in human infants: Age related changes in deferred imitation. Behavioral Neuroscience, 114(1), 77–83. Hayne, H., MacDonald, S., & Barr, R. (1997). Developmental changes in the specificity of memory over the second year of life. Infant Behavior and Development, 20(2), 233–245. Howe, M. L., & Courage, M. L. (1997). The emergence and early development of autobiographical memory. Psychological Review, 104(3), 499–523. Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30(5), 513–541. Kagan, J. (1981). The second year: the emergence of self-awareness. Cambridge, MA: Harvard University Press. Lewis, M. (1999). Social cognition and the self. In P. Rochat (Ed.), Early social cognition understanding others in the first months of life (pp. 81–100). Mahwah, NJ: Lawrence Erlbaum Associates. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum Press. Mandler, J. M. (1990). Recall of events of preverbal children. In A. Diamond (Ed.), The development and neural bases of higher cognitive functions: 608 (pp. 485–516). New York: Annals of the New York Academy of Sciences. McDonough, L., Mandler, J. M., McKee, R. D., & Squire, L. R. (1995). The deferred imitation task as a nonverbal measure of declarative memory. Proceedings of the National Academy of Sciences of the United States of America, 92, 7580–7584. Meltzoff, A. N. (1990). Towards a developmental cognitive science: The implications of cross-modal matching and imitation for the development of representation and memory in infancy. Annals of the New York Academy of Sciences, 608, 1–37. Neisser, U. (Ed.). (1993). The perceived self: Ecological and interpersonal sources of self-knowledge. New York, NY, USA: Cambridge University Press. Nelson, K. (1993). Explaining the emergence of autobiographical memory in early childhood. In A. F. Collins (Ed.), Theories of memory (pp. 355–385). Hove, UK: Lawrence Erlbaum Associates Inc. Perner, J., & Ruffman, T. (1995). Episodic memory and autonoetic consciousness: Developmental evidence and a theory of childhood amnesia. Journal of Experimental Child Psychology, 59(3), 516–548. Povinelli, D. J. (1995). The unduplicated self. In P. Rochat (Ed.), Advances in psychology: Vol. 112. The self in infancy: Theory and research (pp. 161–192). Amsterdam: North-Holland-Elsevier. Rajaram, S. (1993). Remembering and knowing: Two means of access to the personal past. Memory and Cognition, 21(1), 89–102. Rochat, P. (2001). Origins of self-concept. In G. Bremmer & A. Fogel (Eds.), Blackwell handbook of infant development. Handbooks of developmental psychology (pp. 191–212). Malden, USA: Blackwell Publishers. Rovee-Collier, C. (1997). Dissociations in infant memory: Rethinking the development of implicit and explicit memory. Psychological Review, 104(3), 467–498. Tulving, E. (1983). Elements of episodic memory. New York: Oxford University Press. Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of Psychology, 53(1), 1–25. Tulving, E., & Markowitsch, H. J. (1998). Episodic and declarative memory: Role of the hippocampus. Hippocampus, 8(3), 198–204. Wheeler, M. A. (2000). Episodic memory and autonoetic awareness. In E. Tulving & F. Craik (Eds.), The Oxford handbook of memory (pp. 597–608). Oxford: Oxford University Press. Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic memory: The frontal lobes and autonoetic consciousness. Psychological Bulletin, 121(3), 331–354.