Attachment classification and infant responsiveness during interactions

13
INFANT BEHAVIOR AND DEVELOPMENT 19, 13 l-l 43 (1996) Attachment Classification and Infant During Interactions Responsiveness GEORGE M. TARABULSY, R~JEAN TESSIER,JOCELYNE GAGNON, AND CHRISTIANE PICHB Universite’ Lava1 The relation between infant responsiveness during interactions and attachment classification was investigated. Sixty-two 12- to 16-month-old infants classified as secure, avoidant, and ambiva- lent in the Strange Situation were seen during a 9-min mother-infant teaching task. Infant responsiveness was defined as the difference between base rate occurrence of infant behaviors and the probability of infant behavior given a preceeding maternal behavior. Different patterns of infant responsiveness emerged for the three groups of dyads. The positive behavior of infants in secure dyads was elicited by, and negative behavior was suppressed by, maternal positive and helping behavior. The positive behavior of avoidant infants was elicited by positive maternal behavior and was suppressed by maternal positive affect. The positive behaviors of ambivalent infants were elicited by maternal positive, helping, and interfering-coercive behaviors. These results indicate that patterns of responsiveness, rather than absolute proportions of behaviors, may discriminate among infants in the three attachment classifications and may provide clues as to the relational objectives pursued by infants as a function of their interactive history. attachment interaction responsiveness contingency infancy One of the primary assumptions of attachment theory is that the security of the attachment rela- tionship is in large part determined by the quali- ty of interactions mothers and infants experi- ence over the course of the 1st year of the infant’s life (Ainsworth, Blehar, Waters, & Wall, 1978). One of the characteristics of inter- actions used to define their quality is the contin- gency between maternal and infant behaviors. A contingency is defined as the presence of a tem- poral relationship between the occurrence of two behaviors (Moran, Dumas, & Symons, 1992). Behavior B is said to be contingent upon the occurrence of behavior A when the proba- bility of occurrence of B, given the prior occur- rence of A, is greater than that of B without A (Moran et al., 1992; Watson, 1979). We thank the mothers and infants who participated in this study and those who helped gather data: Christine Bouffard, Julie Dionne, Audr6e Levesque, Brigitte Couture, Barbara L&Cillte, H61Bne LaRue, Danielle Nadeau. We also thank Jean Dumas for his support in learning INTERACT and in developing the coding scheme, David Pederson and Greg Moran for their comments on earlier versions of the manuscript. This study was supported by a research fellow- ship to G.M.T. from the Fonds FCAR (Quebec). Correspondence and requests for reprints should be sent to George Tarabulsy, dkpartment de psychologie, universitk Du QuCbec & Trois-Rivi&es, PO Box 500, Trois-Rivikres Quebec, Canada G9A 5H7. Descriptive and theoretical accounts of the interactive correlates and antecedents of the attachment relationship often call upon the notion of contingency to account for individual differences in outcome. Specifically, the nature and delay of maternal (i.e., the primary care- giver’s) responses to infant behaviors through- out the 1st year, critical elements in the defini- tion of maternal sensitivity, are believed to be among the major contributing factors to infant behavioral organization in different contexts, including the Ainsworth Strange Situation (SS), the classic measure of the attachment relation- ship (Ainsworth et al., 1978; Isabella & Belsky, 1991; Smith & Pederson, 1988; Sroufe & Waters, 1977). Through interactive contingen- cies, infants build expectations regarding maternal behavior in different situations and learn to organize their own adaptive behavior in a way that is coherent with their expectations. In the SS, a paradigm that standardizes mater- nal conduct and elicits infant adaptive behavior, such expectations give way to different patterns of behavioral organization. Within this stressful procedure, it is inferred that infant behavioral organization, especially during reunion episodes, is a reflection of their expectations regarding maternal behavior, learned from inter- action in more habitual contexts (Sroufe & Waters, 1977). 131

Transcript of Attachment classification and infant responsiveness during interactions

INFANT BEHAVIOR AND DEVELOPMENT 19, 13 l-l 43 (1996)

Attachment Classification and Infant During Interactions

Responsiveness

GEORGE M. TARABULSY, R~JEAN TESSIER, JOCELYNE GAGNON,

AND CHRISTIANE PICHB Universite’ Lava1

The relation between infant responsiveness during interactions and attachment classification was investigated. Sixty-two 12- to 16-month-old infants classified as secure, avoidant, and ambiva- lent in the Strange Situation were seen during a 9-min mother-infant teaching task. Infant responsiveness was defined as the difference between base rate occurrence of infant behaviors and the probability of infant behavior given a preceeding maternal behavior. Different patterns of infant responsiveness emerged for the three groups of dyads. The positive behavior of infants in secure dyads was elicited by, and negative behavior was suppressed by, maternal positive and helping behavior. The positive behavior of avoidant infants was elicited by positive maternal behavior and was suppressed by maternal positive affect. The positive behaviors of ambivalent infants were elicited by maternal positive, helping, and interfering-coercive behaviors. These results indicate that patterns of responsiveness, rather than absolute proportions of behaviors, may discriminate among infants in the three attachment classifications and may provide clues as to the relational objectives pursued by infants as a function of their interactive history.

attachment interaction responsiveness contingency infancy

One of the primary assumptions of attachment theory is that the security of the attachment rela- tionship is in large part determined by the quali- ty of interactions mothers and infants experi- ence over the course of the 1st year of the infant’s life (Ainsworth, Blehar, Waters, & Wall, 1978). One of the characteristics of inter- actions used to define their quality is the contin- gency between maternal and infant behaviors. A contingency is defined as the presence of a tem- poral relationship between the occurrence of two behaviors (Moran, Dumas, & Symons, 1992). Behavior B is said to be contingent upon the occurrence of behavior A when the proba- bility of occurrence of B, given the prior occur- rence of A, is greater than that of B without A (Moran et al., 1992; Watson, 1979).

We thank the mothers and infants who participated in this study and those who helped gather data: Christine Bouffard, Julie Dionne, Audr6e Levesque, Brigitte Couture, Barbara L&Cillte, H61Bne LaRue, Danielle Nadeau. We also thank Jean Dumas for his support in learning INTERACT and in developing the coding scheme, David Pederson and Greg Moran for their comments on earlier versions of the manuscript. This study was supported by a research fellow- ship to G.M.T. from the Fonds FCAR (Quebec).

Correspondence and requests for reprints should be sent to George Tarabulsy, dkpartment de psychologie, universitk Du QuCbec & Trois-Rivi&es, PO Box 500, Trois-Rivikres Quebec, Canada G9A 5H7.

Descriptive and theoretical accounts of the interactive correlates and antecedents of the attachment relationship often call upon the notion of contingency to account for individual differences in outcome. Specifically, the nature and delay of maternal (i.e., the primary care- giver’s) responses to infant behaviors through- out the 1st year, critical elements in the defini- tion of maternal sensitivity, are believed to be among the major contributing factors to infant behavioral organization in different contexts, including the Ainsworth Strange Situation (SS), the classic measure of the attachment relation- ship (Ainsworth et al., 1978; Isabella & Belsky, 1991; Smith & Pederson, 1988; Sroufe & Waters, 1977). Through interactive contingen- cies, infants build expectations regarding maternal behavior in different situations and learn to organize their own adaptive behavior in a way that is coherent with their expectations. In the SS, a paradigm that standardizes mater- nal conduct and elicits infant adaptive behavior, such expectations give way to different patterns of behavioral organization. Within this stressful procedure, it is inferred that infant behavioral organization, especially during reunion episodes, is a reflection of their expectations regarding maternal behavior, learned from inter- action in more habitual contexts (Sroufe & Waters, 1977).

131

132 T a r a b u ls y , T e s s i e r, G a g n o n, a n d P i c h 6

Much of attachment research has been aimed at providing empirical grounds for this interpretation of infant SS behavior by showing systematic links between maternal interactive behavior and attachment classification. Although there has been some concern with the regularity and consistency of findings obtained (Goldsmith & Alansky, 1987; Lamb, Thompson, Gardner, & Charnov, 1985; Schneider-Rosen & Rothbaum, 1993), there appears to be growing consensus regarding the relation between maternal behav- ior and attachment. Mothers in secure relation- ships (B) are portrayed as sensitive, responding promptly and appropriately to their child’s sig- nals and behaviors, and providing a supportive, predictable, and coherent context for infants to develop. Mothers in anxious-avoidant relation- ships (A) are generally perceived as being pre- dictably insensitive and perhaps stressful to their child, because of the intrusive or rejecting quality of their behaviors, whereas mothers in anxious-ambivalent dyads (C) have most often been seen as lacking in appropriateness and consistency of their caregiving behaviors (Ainsworth et al., 1978; Belsky, Rovine, & Taylor, 1984; Cassidy & Berlin, 1994; Egeland & Farber, 1984; Grossmann, Grossmann, Spangler, Suess, & Unzner, 1985; Isabella, 1993; Main & Weston, 1982; Maslin & Bates, 1983; Pederson & Moran, 1995, in press; Smith & Pederson, 1988). Thus, there appears to be support for the claim that maternal interactive behavior is meaningfully related to SS classifi- cation.

Two conceptual issues arise from research on the interactive antecedants and correlates of the attachment relationship. First, in spite of the importance attributed to contingency in inter- action, relatively little research has attempted to map out, in behavioral terms, the contingencies that are related to different attachment classifi- cations in contexts other than the SS. Those studies that have focused on behavioral indices of the quality of intemctions have generally dif- ferentiated SS classifications on the basis of variations in the occurrences of maternal rather than infant behavior (Belsky et al., 1984; Smith & Pederson, 1988). In many of these studies, the issue of the contingency of behavior during interactions is examined in general terms, where the quality or appropriateness of mater- nal behavior is interpreted as a function of infant behavior. Although these data provide

important indices as to the structure of the interactions that characterize dyads in the dif- ferent attachment categories, studies that have directly addressed the concept of interactive contingency are rare.

The second issue concerns the link between infant behavior in the SS and in other contexts. In the SS, the quality of the attachment rela- tionship is inferred from patterns of infant behavioral organization in a procedure where maternal behavior is standardized. In light of the success of researchers in using this proce- dure, it is somewhat surprising that more effort has not been focussed on identifying patterns of infant behavior in less distressful situations, where maternal behavior is not scripted (Cohn, Campbell, & Ross, 1991). There is an expecta- tion that infant behavior in the SS is coherent with that manifested during interactions with mother in other contexts (Pederson & Moran, 1995; see also, Ainsworth et al., 1978; Belsky et al., 1984; Grossmann et al., 1985). This rea- soning is partly behind the development of alternative assessments of the attachment rela- tionship, such as the Q-Sort procedure devel- oped by Waters and Deane (1985) and the Home Classification scheme recently intro- duced by Pederson and Moran (1995; in press). However, although evidence is convincingly accumulating regarding the role of maternal interactive behavior as an important precursor and correlate of SS classification (Isabella, 1993; Pederson & Moran, in press), thereby validating a crucial element of attachment theory, the manner in which infant behavioral organi- zation in this procedure is a reflection of inter- active behavior in other contexts is still rela- tively unclear. The demonstration of a link between SS classification and infant interactive behavior in other contexts, where maternal behavior is not scripted, may provide clues as to how infants organize their behavior during more natural interactions with their primary attachment figure.

Researchers who have attempted to draw a link between SS classification and either infant behavior during interactions or characteristics of infants that may contribute to the develop- ment of attachment have been faced with inconsistent findings. In general, antecedent and concurrent assessments of rates or frequency counts of infant behavior during interactions have shown either low or nonsignificant associ-

A t t a c h m e n t a n d In f a n t R e s p o n s iv e n e s s 1 3 3

ations with attachment classification (Belsky et al., 1984; Smith & Pederson, 1988), and studies that have attempted to examine infant contributions to attachment from the perspec- tive of temperament (e.g., Mangelsdorf, Gunnar, Kestenbaum, Lang, & Andreas, 1990) and bio- logical risk (e.g., van Izjendoom, Goldberg, Kroonenberg, & Frankel, 1992) have provided few converging results.

Such findings stand in contrast to results obtained by those who have examined the issue of infant responsiveness to events in both nonsocial and social circumstances. Here, authors have portrayed infants as being actively involved in organizing their behavior in mean- ingful and coherent ways as a function of either environmental events or maternal behaviors. This observation has been made by researchers who have used a variety of laboratory para- digms to address the issue of the contingency of infant responses to nonsocial stimulation (Lewis, Sullivan, Ramsay, & Alessandri, 1992; Wentworth & Haith, 1992) and by those who have looked at contingencies in mother-infant interactions involving standardized (Cohn & Tronick, 1988; Dunham, Dunham, Hurschman, & Alexander, 1989; Millar, 1988) and non- standardized maternal behavior (Dunham & Dunham, 1990). Although several workers in this area have suggested that the organization of infant responsiveness during interaction may be linked to socio-emotional outcome (Dunham & Dunham, 1990; Lewis et al., 1992), and that this particular aspect of infant interactive behavior may provide clues as to the coherence and predictability of interactions in the home, research with infants older than 6 months is rare, as are links with established indices of development, such as the SS.

Within the attachment literature, there are two groups of researchers who have addressed the issue of the contingency of behavior during interaction while taking into account some ele- ment of infant responsiveness. Isabella and Belsky (1991; Isabella, Belsky, & von Eye, 1989) conducted two longitudinal studies where they examined the frequency of synchro- nous mother-infant interactions in the home at 1, 3, and 9 months and its relation to the securi- ty of attachment at 12 months. Synchronous interactions are operationalized as “reciprocal and mutually rewarding behavioral exchanges between mother and infant” (Isabella & Belsky,

1991, p. 376). By their definition, synchronous interactions involve contingently organized behavior on the part of both interactants. In both studies, Isabella and his coworkers found that the frequency of synchronous interactions at 1 and 3 months, but not at 9 months, was sig- nificantly related to attachment security at 12 months. They also reported that the asynchro- nous interactions which characterized infants who had avoidant attachment relationships involved mothers who “intruded” on infant activity by acting out of turn, whereas mothers in C dyads did not respond as often as mothers in secure relationships to infant vocalizations (Isabella & Belsky, 1991; Isabella et al., 1989). All of these findings provide important support for the view that when mothers are sensitive to infant behaviors, and respond in a way that is contingent upon their occurrence, the develop- ment of a secure attachment relationship is favored. However, because the authors were primarily preoccupied with addressing interac- tive contingencies from the perspective of maternal sensitivity, little is said about infant responsiveness and its potential link to SS classification.

In another study, Cohn et al. (1991) exam- ined the relation between attachment classifica- tion at 12 months and infant reactions to exper- imentally provoked maternal violations of interactive contingencies (the “still face para- digm”) at 2, 4, and 6 months. This is the only study found where an element of infant respon- siveness to mother was addressed in relation to attachment. Results showed that when interac- tive contingencies are violated at 6 months (but not at 2 or 4 months), infant manifestations of positive behavior are related to secure attach- ment at 12 months. The absence of positive behavior is predictive of avoidant attachment (the low number of ambivalent dyads precluded their inclusion in the analyses). These results suggest that by 6 months, certain aspects of infant responsiveness, observed in the context of the violation of a contingency, may be mean- ingfully related to behavioral organization in the SS. However, parallel to the SS, the still face paradigm addresses infant behavior as a function of scripted maternal behavior. As the authors noted, their results indicate that one “probe” (p. 373) of attachment formation observed at 6 months is related to a measure of attachment at 12 months. Thus, although this

1 3 4 T a r a b u ls y , T e rs i e r, G a g n o n, a n d p i c h i

study provides provocative findings concerning the relation of infant interactive responsiveness to attachment, the possibility that under less structured circumstances varying patterns of responsiveness characterize different attach- ment relationships merits further investigation.

The purpose of this report is to address this issue by examining the concurrent relation between attachment classification and patterns of infant responsiveness of a sample of 12- to 16-month-old infants during a mother-infant teaching task where maternal behavior is not standardized. The approach used is inspired by the work of Dumas and LaFreniere (1993; Dumas, LaFreniere, & Serketich, 1995; LaFreniere & Dumas, 1992) who examined the patterns of mother-child interactions in differ- ent groups of toddlers and preschoolers. Two questions were asked to guide data analyses: (a) Do mothers and infants in the different attach- ment classifications differ with respect to the overall proportions of positive and negative behaviors and affect manifested during interac- tion? This question is asked because it enables the data to be explored in a manner similar to that in which it is often reported in behavioral accounts of interaction (e.g., Ainsworth et al., 1978; Belsky et al., 1984; Smith & Pederson, 1988), thereby allowing for possible compar- isons with previous studies. It does not, how- ever, directly speak to the issue of the contin- gency of infant responses, hence the purpose behind the second question. (b) Do different patterns of contingent responsiveness to mater- nal behavior emerge for infants in the different

attachment classifications during mother-infant teaching interactions (Dumas & LaFreniere, 1993; Moran et al., 1992)?

METHOD Participants

Mothers and their infants were recruited through advertise- ments in community centers and newspapers in the QuCbec City area as part of a larger study of infant socio-emotional development. Minimal inclusion criteria were the following for mothers: (a) Caucasian and French speaking; (b) current cohabitation with the infant’s biological father. Inclusion cri- teria for infants were: (a) single birth after 38 to 42 weeks of pregnancy; (b) minimal birth weight of 2,500 gms; (c) absence of birth complications, genetic, or chromosomal anomalies; (d) between 12- and 16-months-old. In total, 62 dyads were recruited. Table 1 presents the demographic vati- ables of the sample. Five additional dyads took part in the study but were excluded because they did not meet infant age requirements when they were seen at the university.

Procedure and Measures Mothers interested in participating called the university lab- oratory where a member of the research group described the study. If mothers agreed to take part in the project, two appomtments were made: one at the mother’s home to fur- ther describe the study and complete a battery of question- naires, and the second at the university, where the mother-infant dyad took part in several activities, among which were the interactions, followed by the Strange Situation. The total time spent by the dyad in the laboratory was approximately 90 min.

Laboratory Interaction Task and Observations

Infants sat in a high chair at the comer of a table with their mother, with a camera placed behind a one-way mirror, approximately 2 m away. All interactions were videotaped for later coding. Mothers were asked to teach their child

T A B L E 1 S u m m a ry D e s c rip t io n of S a m pl e C h a r a c t e r i s t i c s

M a t e rn a l C h a r a c t e r i s t i c s :

Marital Status (58 respondents) Common-law relationship: n = 24 (41.38%) Married: n = 34 (58.62%)

t!Zation: M=31.18years(SD=5.18) M = 1 5.6 7 years (SD = 2.45)

In f a n t C h a r a c t e r i s t i c s :

Sex: 36 boys, 26 girls Age: M = 1 3.3 4 months (SD = 1.41) Family Rank: 1 st born n = 39 (62.90%)

2nd born n = 14 (22.58%) 3rd born n = 9 (14.52%)

Experience of re 7

uior nonmaternal care: n = 3 9 (6 2.9 0 %) Average hours o nonmaternal care/week: 15.92 (SD = 15.84)

A t t a c h m e n t a n d In f a n t R e s p o n s iv e n e s s 1 3 5

how to play with a series of four toys for 3-min each. A member of the research team placed the first toy on the table between the mother and child, gave her three boxes labelled 2 to 4 which contained the other toys, and left the room. At the end of each 3-min period, a light knock on the door indi- cated to mothers that they should move on to the next toy.

The toys were selected during two different pilot stud- ies, involving 17 mother-infant dyads, with infants aged 10 to 20 months. We were looking for toys that elicited infant interest, and were not overly common, such that previous exposure would not be a major factor in the interactions. We also wanted toys for which infants appeared aware of the functional purpose, but for developmental reasons were unable to accomplish it without the guidance of mother. Appendix A lists examples of the kinds of behaviors which were indicative of this criteria and presents descriptions of the three toys that were finally used (9 min of interaction were coded for each dyad).’ The toys were presented to dyads in order of increasing difficulty. Prior to the begin- ning of the sequence, mothers were told by a member of the research team that the toys were a little difficult for their child, and that their role was to facilitate their child’s play in a way that would help the child accomplish at least part of the functional purpose of the toy. They were also told that it was not important that their child play with the toys in a perfect way, and as such, were instructed not to accom- plish the tasks in the place of the child. Finally, mothers were told that part of their role was to help their child main- tain interest in the task, should he or she decide to attend to something else.

The toys were selected with the purpose of encouraging teaching interactions between mother and child that would draw out responsive infant behavior, that is, behavior that was contingent on, and meaningfully related to, maternal behavior. Related strategies have been successfully used in studies that differentiated various groups of toddlers and preschoolers through child interactive behavior (e.g., Dumas & LaFreniere, 1993; Matas, Arend, & Sroufe, 1978). The presentation of toys of increasing difficulty and the instructions given to mothers are based on the idea that the most attachment-relevant behaviors are manifested by both mothers and infants in contexts where there are certain demands placed on one or both members of the dyad. This is part of the rationale behind the SS, and it has been used by Pederson and his colleagues (Pederson & Moran, 1995, in press; Smith & Pederson, 1988) who placed certain demands on mothers by dividing their attention between two tasks, and by Dumas and LaFreniere (1993) who exposed dyads to certain task objectives.

Videotapes of interactions were coded using a real- time, microcomputer coding program, INTERACT, devel- oped by Dumas (1987). The program divides the actors and events of an interaction into five different categories: (a) Actor 1, who initiates a behavior; (b) the behavior that is initiated; (c) Actor 2, the target of the behavior; (d) an adverb, describing the behavior; (e) valence of the behav- ior. The last two categories are optional and were not used

‘One of the toys, the yellow pegs and board from the Bayley Scales, was discarded because it was too easy to accomplish for many children in the sample, and in a num- ber of cases, the pegs kept rolling off the table such that much of maternal behavior was directed toward picking them up off the floor.

in this study. Interactions were coded from videotape using a Panasonic (model no. AG-6300) video cassette player and an IBM-PC-AT compatible microcomputer.

The codes used by Dumas and LaFreniere (1993; Dumas et al., 1995; LaFreniere & Dumas, 1992) were adapted for use with 12- to 16-month-olds in two ways, First, using videotapes from the two pilot studies, a listing of maternal and infant behaviors was compiled which char- acterized this age group in this particular context. Second, for both mothers and infants, discrete behaviors were grouped into meaningful clusters based on the purposes they accomplished in the interactions. This consideration emanates from Ainsworth’s writings stating that in the SS, behaviors are understood in the context of the goals they serve (Ainsworth et al., 1978; see also, Matas et al., 1978; Sroufe & Waters, 1977). Therefore, the particular grouping of behaviors used in the context of a mother-infant teach- ing task may not be appropriate in another situation, where they may serve to accomplish other mother and/or infant objectives. As with LaFreniere and Dumas, comparable clusters of mother and child behaviors were devised. For mothers, the behavior clusters were: (a) positive behavior, which consisted of encouragements, approval, and positive vocalizations; (b) positive affective behaviors, which were smiles, laughs, manifestations of empathy (e.g., “you find this game hard, don’t you?‘); (c) helping behavior, com- posed of commands or requests, structuring of toys to facil- itate infant play, asking focused questions to infant, descriptions of the toy; (d) negative behavior, made up of insults, criticism, harsh vocalizations, punishments, sar casm regarding infant behavior, impossible requests that were clearly out of developmental range; (e) negative affec- tive behaviors, which were manifestations of anger or dis- appointment; (f) interfering or coercive behavior, which was coded when maternal actions interrupted or disrupted the infant’s line of activity, when the mother physically forced the infant to perform a behavior, or when helping behaviors were inappropriate for the task.

Infant behavior clusters were the following: (a) positive behavior, composed of compliance to commands, focus on game and behavioral or vocal replies to maternal requests; (b) positive affective behaviors, which were smiling, laugh- ing; (c) help-seeking behavior, made up of those times when the infant wanted mother to participate in the interac- tion in a very specific way by bringing her hand to a toy, handing her a toy, looking up at her, and showing her a toy; (d) negative behaviors, made up of noncompliance to maternal requests or commands, repeatedly throwing toys either at mother or elsewhere against mother’s will, reject- ing the game in the face of maternal attempts to keep infant interested, hitting mother; (e) negative aflective behaviors, which were crying, complaining vocalizations, manifesta- tions of anger. In all cases, behaviors were coded at their onset, and the duration of behavior was not a consideration. Thus, for example, if a behavior was observed that lasted 3 s, it was only coded once, at the beginning of this interval. If a change in behavior was observed, a separate code was recorded even though this behavior may have been a part of the same cluster. For example, mother may be verbally describing the toy to the child for 5 s, then picks up the toy to demonstrate to the infant how it works for an additional 4 s. Each behavior would be coded at their onset as mater- nal helping behavior.

Two coders trained intensively for a 3-month period and established interrater agreement on 17 (27.4%) ran-

1 3 6 T a r a b u ls y , T e s s i e r, G a g n o n, a n d P i c h 6

domly selected dyads, 7 at the onset of coding and 10 more throughout coding. Agreement was counted when both observers coded the presence or absence of the same behavior within 5 s of each other. Mean percentages of agreement and Cohen’s kappas are reported in Table 2 for all behavior clusters. Lower kappa values are associated with lower mean occurrence of a behavior. The primary coder for interactions was not involved in the coding of attachment relationships and was responsible for the data for 44 of 62 (71.0%) dyads. Although the secondary coder had participated in SS observations, these had taken place 8 to 12 months prior to the coding of interactions.

Two types of measures resulted from this coding scheme. First, overall proportions for each maternal and infant behavior cluster were calculated. This score corre- sponded to the percentage of total behaviors accounted for by each behavior cluster. Second, conditional probabilities of the occurrence of infant behaviors, given the previous occurrence of maternal behaviors (within 15 s) were calcu- lated and compared to expected probabilities of infant behavior. Expected probabilities were calculated using a time-based procedure, described in Moran et al. (1992). Conditional probabilities were calculated in order to address the issue of infant responsiveness to maternal behaviors. In comparing them to expected probabilities, conditional probabilities answer the following question: Does the previous occurrence of a given maternal behavior render more, less, or equally likely the occurrence of a given infant behavior (Dumas, 1987)? Comparisons between conditional and expected probabilities were con- ducted using a z statistic, modified to control for autocorre- lation (Dumas, 1986). A significant, positive z statistic indi- cates a greater than expected conditional probability, whereas a significant, negative z statistic indicates a lesser than expected conditional probability. Both reflect forms of infant responsiveness in that behavior is either elicited or suppressed by maternal behavior.

The 15-s interval was selected to remain coherent with previously published research that has used the INTER- ACT analysis package, as well as the time-based (as opposed to interval- or event-based) method for calculating expected probabilities of behaviors (Dumas & LaFreniere, 1993; Dumas et al., 1995; LaFreniere & Dumas, 1992).

T A B L E 2 In t e r-R a t e r R e l i a b i l i t y

fo r a l l B e h a v io m l C lu s t e rs

% A gr e e m e n t K a p p a

M a t e rn a l B e h a v io rs Positive Helping Positive Affect Negative Negative Affect Interference-Coercion

In f a n t B e h a v io rs Positive Help Seeking Positive Affect Negative Negative Affect

.88 .75

.93 .69

.93 .69

.96 .62

.99 .66

.94 .61

.90 .79

.98 .70

.96 .63

.93 .72

.93 .68

Furthermore, two important observations emerged from qualitative examinations of interactions during the two pilot studies which preceeded this report: First, some of the behavior clusters, such as maternal helping behavior and infant positive behavior, were often of long duration. Had a shorter interval been selected, the infant’s response to long behaviors would have been missed. Second, infants often appeared to make more than one response to maternal behaviors. For example, if mother took a toy out of the infant’s hands (maternal interference), the infant might respond first by crying out (infant negative affect). then by trying to recover the toy from mother (infant positive behavior), followed by another angry fuss directed at moth- er (infant negative affect). The 15-s interval was selected as a conservative way of diminishing the impact of behaviors that were of long duration and to include behaviors that were contingently responsive, but not the first response emitted by the infant following a given maternal behavior.

Attachment The Strange Situation (Ainsworth et al., 1978) is a labora- tory procedure intended for 12. to 18-month-old infants, involving their mothers and a stranger. It is composed of a 30-s introduction and seven 3-min episodes where the child is either alone, with mother, with mother and the stranger, or alone with the stranger. Infant behavioral organization with respect to the mother in the reunion episodes are the focus of the coding procedures, which yield a standard, tripartite classification scheme for attachment relationships: secure (B), insecure-avoidant (A), and insecure-ambivalent (C).

The primary coder, GM. Tarabulsy, was trained in the standard A, B, C classification by David R. Pederson at the University of Western Ontario, during a 3-month practicum in 1993, and again during a l-week workshop in the spring of 1994. lnterrater reliability was established with Pederson on 15 (24.2%) of the 62 dyads following the first training period. One hundred percent agreement was obtained with respect to the major classification, and 86.7% (13/15) for subclassifications. Pederson participated in the coding of an additional 9 dyads where there was considerable doubt as to the primary classification. Thus, in total, independent coding of the sample accounted for 24 of 62 (38.7%) attachment classifications. Overall, 19 dyads were coded as avoidant (30.7%). 25 as secure (40.3%). and 18 as ambivalent (29.0%). Although both categories of insecure attachment are overrepresented. it is the number of ambivalent relation- ships that is especially conspicuous. Given that our sample was composed of predominantly educated mothers from middle-class families, we did not suspect that demographic variables could be responsible for these proportions. Nevertheless, we carried out a series of chi-square analyses to examine whether or not infant sex, exposure to nonmater- nal care, or parental marital status were related to attachment. A multivariate analysis of variance-(MANOVA) was per- formed with infant family rank, infant age, hours per week of exposure to nonmatemal care, maternal age, years of mater- nal education, and annual family income as dependent var- ables. As in past research, both sets of analyses revealed no relation between attachment and demographic variables. We then suspected that coding bias may have been responsible for the unusually elevated proportion of insecure dyads. Following the second training seminar, 13 (21.0%) dyads were selected for recoding in conference with at least 3 semi- nar participants. In 11 cases (84.6%), the second code agreed

A t t a c h m e n t a n d In f a n t R e s p o n s iv e n e s s

with the first with respect to both major and minor classifica- tions. In the two cases where there was disagreement, the category changed from B 1 to Al, and from B4 to C 1. In light of these findings, we do not believe that either demographic variables or coding variables are responsible for the propor- tions obtained in this study, leaving us to consider possible cultural or chance effects. However, both of these issues are not the focus of this study and will be dealt with elsewhere.

1 3 7

MANOVAs, one for maternal and one for infant behavior clusters. Independent, demo- graphic variables were: infant gender, exposure to nonmaternal care (yes/no), and parental mar- ital status (married/common law relationship). With regard to maternal behaviors, multivariate effects were not significant for sex, F(5, 50) = 1.06, p > .05, exposure to nonmaternal care, F(5, 50) = 0.87, p > .05, or marital status, F(5, 50) = 0.54, p > .05. The same was true when considering infant behavior clusters: sex, F(4, 51) = 1.15, p > .05; nonmaternal care, F(4, 51) = 1.41, p > .05; marital status, F(4, 51) = 1.01, p > .05. Thus, no demographic vari- ables were considered throughout the remain- der of the analyses.

RESULTS

Preliminary Analyses Proportions of all behavior clusters are present- ed in Table 3. As can be readily observed, maternal helping behaviors and infant positive behaviors make up the greater part of all behav- iors. Pearson correlations were calculated between proportions of maternal and infant behaviors and the following demographic vari- ables: infant age, infant family rank, hours per week of nonmaternal care, maternal age, mater- nal years of education, and annual family income. Of 66 correlations calculated, only 3 reached statistical significance, between mater- nal interference-coercion and maternal age, r = -.25, p = .05, interference-coercion and family income, r = -.30, p = .02, and between infant positive behavior and family rank, r = -.29, p = .02. Given the overall lack of rela- tion between these demographic variables and the behavioral clusters, and because the signifi- cant correlations can be considered as moder- ate, at best, none of these demographic vari- ables were covaried in subsequent analyses.

The influence of categorical demographic variables was assessed with two 2 x 2 x 2

T A B L E 3 R a t e s of O c c urr e n c e

of A l l B e h a v io r C lu s t e rs (%)

A t t a c h m e n t

T o t 4 A B C

M a t e rn a l B e h a v io r Positive Helping Positive Affect N e g a t i v e Interfering-Coercive Negative Affect

C h i ld B e h a v io r Positive Help Seeking Positive Affect Negative Negative Affect

.12 .lO

.60 .60

.12 .12

.03 .03

.lO .l 1

.02 .03

.61 .60

.04 .05

.05 .04

.18 .20

.12 .l 1

.14 .12

.60 59

.12 .14

.03 .04

.l 1 .09

.Ol .02

.63

.04

.06

.15

.l 1

.58

.04

.06

.18

.14

Attachment and Behavior Proportions To help focus analyses and results around meaningful behavior clusters, those that did not represent a minimum of 5% of all mother or infant behaviors were eliminated from further analyses. Four MANOVAs were performed to examine differences in proportions of behavior clusters among attachment categories. Attachment was the independent variable, con- sidered in its secure/insecure dichotomy and its A/B/C forms. Two of the MANOVAs involved maternal behaviors (positive, helping, positive affect, and interference-coercion) as dependent variables, and two focused on infant behaviors (positive behaviors, negative behaviors, mani- festations of negative affect). No significant multivariate effects were found (three of four ps > .25), although in its dichotomous form, attachment tended to be related to maternal positive behavior, Wilks’s Lambda = 86; F(4, 57) = 2.28, p = .07. Univariate analyses revealed that 14.13% of the behaviors of moth- ers in secure dyads were positive, whereas this was true for only 11.17% of mothers in inse- cure dyads, p < .05. Overall, however, there appears to be little or no relation between attachment classification and proportions of behaviors manifested by infants and mothers interacting in a teaching task (see Table 3).

Conditional Probabilities Results relating attachment classification to infant responsiveness are summarized in Tables 4 and 5. Table 4 reports probabilities of occur- rence of infant behaviors given the previous occurrence of maternal behaviors and compar-

1 3 8 T a r a b u ls y , T e s s i e r, G a g n o n, a n d P i c h e

T A B L E 4 C o n d i t io n a l P ro b a b i l i t i e s o f In f a n t R e s p o n s iv e n e s s

a s a F u n c t io n of A t t a c h m e n t C l a s s & c a t io n

P o s i t i v e /P o s i t i v e A B r

kodive/Negalive A

: P o s i t i v e /N e g a t i v e A ff e c t A

! H e lp in g/P o s i t i v e A

: H e lp in g/N e g a t iv e A

: H e lp in g/N e g a t iv e A ff e c t A

:

.8 5 .91*

.8 8 .93**

.8 6 .91 '

. A 5 .30"

.A1 .29"

.A6 .28" '

.30 .26

.30 .23*

.37 .29 '

.85 .86

.88 .90***

.86 .89" '

.A5 .AO" '

.A1 .37" '

.A6 .42" '

.30 .27**

.30 .25** '

.37 .32" '

P o s i t i v e A ff e c t/P o s i t i v e .8 4 .8 8 .8 6

P o s i t i v e A ff e c t/N e g a t iv e . A 5 .A1 .A6

Positive Affect/Negative Affect .30 .30 .37

Interference/Positive .86 .88 .86

Interference/Negative .A3 .41 .A6

Interference/Negative Affect .31 .30 .37

.74"

.85

.88

.33 ' "

.36

.37*

.26

.27

.31

.88

.89

.91****

.31" '

.37

.A1

.31

.23*

.25 '

aMaternal Behavior (M)/ lnfont Behavior (I) . *p < .05. **p < .Ol **p c .oOl . ****p c .06

isons to chance or expected probabilities of infant behaviors. Table 5 is a descriptive pre- sentation of the results presented in Table 4. The following patterns appeared to emerge.

Responsiveness to Positive Maternal Behavior. Positive infant behavior was elicited by posi- tive maternal behavior for the three groups of children, although the effect was somewhat more pronounced for infants in the B group. Positive maternal behavior suppressed the occurrence of negative infant behaviors for all three groups but suppressed negative affect only in B and C dyads.

Responsiveness to Helping Maternal Behavior. In all dyads, helping behavior served to sup- press the occurrence of negative behavior and affect. However, in B and C dyads, helping behavior elicited positive child behavior, where- as this was not the case for infants in A dyads.

Responsiveness to Positive Maternal Affect. Manifestations of positive maternal affect were unrelated to the behavior of infants in secure relationships but were somewhat related to the behavioral organization of infants in insecure

relationships. Infants in A dyads showed fewer incidences of positive behavior and negative behavior following positive maternal affect, indicating that when mothers smiled or laughed, avoidant infants would not focus on the toy in front of them, nor would they direct negative behavior towards mother. Under the same conditions, infants in C relationships manifested less negative behavior. Responsiveness to Maternal Interference- Coercion. Responsiveness to interference-coer- cion was different for infants in the three groups. Children in A relationships showed lower than chance manifestations of negative behavior, whereas infants in B dyads showed lower than chance probabilities of negative affect. For infants in the C group, maternal interference or coercion tended, p < .06, to elicit positive behavior, while suppressing neg- ative infant affect.

In summary, for infants in secure dyads, posi- tive and helping maternal behaviors elicit posi- tive infant behavior and suppress the occurrence of negative behavior and affect. Infants in A dyads show positive behavior only when moth-

A t t a c h m e n t a n d In f a n t R e s p o n s iv e n e s s 1 3 9

T A B L E 5 In f a n t R e s p o n s iv e n e s s a s a F u n c t io n of M a t e rn a l B e h a v io r a n d A t t a c h m e n t

A t t a c h m e n t

M a t e rn a l B e h a v io r

E l i c i t e d In f a n t R e s p o n s e Positive Helping Positive Affect Interference/Coercion

S u p pr e s s e d In f a n t R e s p o n s e Positive Helping Positive Affect Interference/Coercion

A

Positive

Neg??y&Aff Positive/PosAff

Negative

B

Positive Positive

Negative/NegAffo Negative/NegAff

NegAff

C

Positive Positive

Positive

Negative Negative/NegAff

Negative NegAff

oNegAff = negative affect.

ers first manifest positive behavior, and positive behaviors are less likely if positive maternal affect occurs. Manifestations of maternal posi- tive affect or interference-coercion serve to sup- press infant negative behavior in this attachment category. Infants classified as ambivalent in the SS showed a pattern of positive behavior similar to that of secure infants in that it was elicited by both positive and helping maternal behavior. Also, negative behavior and affect were sup- pressed by maternal helping behavior, as with B dyads. However, positive behavior was elicited by interfering or coercive maternal behavior, unlike infants in the B group, and, as with infants in the A group, negative behavior was made less likely by the occurrence of maternal smiles and laughs (Tables 4 and 5).

DISCUSSION The purpose of this report was to examine the relation between attachment classification and the contingent, responsive behavior of infants during interactions with their mother. The proce- dure used to assess the quality of the attachment relationship, the SS, is grounded in the notion that by standardizing maternal behavior in a para- digm that is distressing to the infant, infant behavioral organization around mother provides an indication of past interactive history. The results of this study suggest that infant behavior in the SS, upon which attachment classification is based, are linked to their responsiveness in a con- text where maternal behavior is not standardized. Qualitatively different patterns of interactive responsiveness appeared to emerge for infants in the three attachment classifications. As a caution-

ary note, it should be underlined that some of the differences between expected and conditional probabilities of behaviors were significant only at the p = .05 level and one at p < .06 (see Table 4), meaning that reported findings should be viewed as indicative of trends necessitating further atten- tion. The discussion is tempered accordingly.

Behavior Proportions Versus Behavioral Organization

The 9 min of mother-infant teaching tasks used in this study failed to discriminate between the three attachment classifications on the basis of the proportions of behaviors manifested by mothers and infants. This occurred in spite of grouping behaviors into meaningful clusters, as other teams of researchers have successfully accomplished in the past (Belsky et al., 1984), and using a series of tasks in a context intended to draw out attachment-relevant behavior (Matas et al., 1978). A number of reasons may account for the lack of significant findings, such as the short observation period (Isabella, 1993; Moran, Pederson, & Tarabulsy, in press, Pederson & Moran, 1995) the use of a labora- tory situation (Schneider-Rosen & Rothbaum, 1993), and the focus on behavioral, rather than global, measures of infant and mother function- ing (Lamb et al., 1985; Moran et al., in press). Another explanation may reside in the a priori, theoretically based clusters of behaviors used in this study. In the past, other studies which have found links between maternal behaviors and attachment have sometimes relied on post hoc groupings, based on factor or correlational analyses (Moran et al., in press). The use of

1 4 0 T a r a b u lry , T e s s i e r, G a g n o n, a n d P i c h C

predetermined behavior clusters has often been associated with weaker relations to attachment classifications (Lamb et al., 198.5; Schneider- Rosen & Rothbaum, 1993).

In spite of these findings, it is noteworthy that when conditional probabilities were used to draw out patterns of interactions, certain qualitative differences appeared to distinguish attachment categories. This suggests that there may be a link between the notion of contingent responsiveness and the concept of infant behavioral organization, which has been proposed by attachment researchers to be responsible for infant behavior in the SS (Ainsworth et al., 1978; Sroufe & Waters, 1977). Infants in the three attachment categories appeared to emit positive behaviors and negative behaviors and affect following dif- ferent types of maternal behaviors.

Infant Responsiveness as a Function of Attachment

The most coherent or straightforward patterns of infant responsiveness were found in the B dyads. Positive infant behavior followed either positive or helping maternal behaviors. When mothers in secure relationships encouraged their infants, asked or directed them to do something, or structured some aspect of the toy to facilitate play, infants more often responded by focusing on the toy and complying to mater- nal requests. Because the positive behavior cluster made up 61% of all infant behaviors, and positive and helping maternal clusters occupied 72% of all maternal behaviors, the temporal relations between these behaviors have particular significance in the context of these interactions. Furthermore, positive and helping maternal behaviors were also followed by fewer than chance occurrence of infant neg- ative behaviors (hitting mother, noncompli- ance, rejection of the toy) and affect (crying, complaining vocalizations, and anger). Thus, both positive behaviors and negative behaviors and affect appeared to be contingently and coherently organized around maternal conduct.

Some of the patterns of B infants were found in both A and C dyads, but there were also dif- ferences that require attention. For infants in avoidant dyads, Tables 4 and 5 show that posi- tive infant behaviors followed maternal positive behaviors, but not helping behaviors, and were suppressed by manifestations of positive affect. Two points need to be made here. First, infants

in the A group did not organize their most fre- quently occurring behavior, positive behaviors, around the most frequent maternal cluster, helping behaviors (60% of maternal behaviors), suggesting that infant focus on toys may have been somewhat independent of maternal attempts to facilitate learning, a pattern not unlike the exploratory behavior of avoidant infants during the SS.

Second, smiles and laughs by the mother inhibit the occurrence of positive infant behav- iors. Isabella (1993) has suggested that the behavioral organization of infants in avoidant dyads are characterized by the presence of defensive strategies, intended to buffer the effect of certain maternal behaviors, as well as by manifestations of the desire to experience a close relationship with mother. That positive infant behaviors are elicted by positive mater- nal behavior, and are suppressed by positive maternal affect, suggests one manner in which the pattern proposed by Isabella may be mani- fested during teaching interactions. It is also possible that within the home, maternal smiles and laughs are linked to the overstimulating and/or rejecting behaviors that have been viewed as characteristic of the interactions of such dyads (Belsky et al., 1984; Isabella & Belsky, 1991; Lewis & Feiring, 1989), and that the responses of infants may be a reflection of the way they usually cope with these types of maternal behaviors. Clearly, although the inter- pretation of such a behavioral profile requires replication, its presence is intriguing and may provide clues as to the manifestation of avoidant behavior in contexts other than the SS.

As with infants in secure relationships, infants in the C classification emitted positive behaviors contingently upon maternal positive and helping behaviors. However, these infants also tended to manifest positive behaviors following maternal interference or coercion. The strategy used by these infants to organize their behavior may carry the label “compliance at all cost,” as their posi- tive behavior was related to previous maternal interruption of ongoing activity or use of physical means to have the child attend to what the mother viewed as important. Contrary to infants in avoidant relationships, who show a general pat- tern of positive behavior reduction in response to maternal behaviors, infants in ambivalent dyads tend to increase their rates of positive behavior. In the past, these infants have been perceived as

A t t a c h m e n t a n d In f a n t R e s p o n s iv e n e s s 1 4 1

overly preoccupied with their mothers during the SS (Amsworth et al., 1978), perhaps because these mothers have been “minimally or inconsis- tently” responsive to infant behaviors and needs (Cassidy & Berlin, 1994, p. 982; Isabella, 1993). One interpretation of these data is that infants in these relationships adapt to such caregiving prac- tices by “overresponding” to maternal behaviors, perhaps to increase the likelihood that mother will behave in a way that is contingent on their response, in a way that may be analogous to their behavior during reunion episodes in the SS.

Three additional, general points are raised by the data. First, it is important to underline that the patterns of interactions observed, although apparently coherent with the body of attachment literature, must be interpreted with care given the exploratory (rather than confir- matory) nature of this study, the laboratory context and tasks presently used, short observa- tion period, and unusual proportions of attach- ment classifications obtained. Future work that addresses similar issues in a related manner must tailor coding procedures to ensure that the grouping of behavioral clusters is meaningful to that particular context.

Second, the patterns of interactions remind us that infant behaviors may carry different sig- nificance for children exposed to various devel- opmental environments. That the avoidant child reduces positive behavior following positive maternal affect and the ambivalent child increases positive behavior following maternal interference-coercion may indicate that positive behavior has a different meaning for different infants. For example, it is possible that patterns of behavior reflect infant attempts at exerting some control over maternal behaviors, accord- ing to the contingencies that have been set in the previous year. Therefore, the positive behavior code is a reflection of something more than just focus on the toy, as the authors had planned. It may reflect how, in a particular teaching context, some infants may be involved in influencing the type of behavior that their mother directs towards them. Thus, although behaviors were grouped according to the pur- poses they served within the context of teach- ing tasks, and consequently carry the same label for all mothers and infants involved, what is termed “positive behaviors” may serve a dif- ferent interactive goal for the three groups of children. Future research may address this issue

by devising coding strategies that specifically address, in an a priori fashion, the different goals of infants during interaction with their mother (Pederson & Moran, in press).

Finally, given that all observations were car- ried out at one time, certain questions require attention before the scope of the results can be adequately understood. These results point to differences in infant responsiveness as a func- tion of attachment classification, but the devel- opmental course of the former variable, apart from certain highly controlled studies in early infancy (Cohn & Tronick, 1988) is relatively uncharted. The longitudinal investigation of this phenomenon may allow for a clearer understanding between the development of infant responsiveness during interactions, the manner in which infants learn the contingencies to which they are exposed, and the develop- ment of different types of attachment.

REFERENCES Ainsworth, M.D.S., Blehar, M.C., Waters, E., & Wall, S.

(1978). Patterns of attachment: A psychological study of the Sfrange Situation. Hillsdale, NJ: Erlbaum.

Belsky, .I., Rovine, M.J., & Taylor, D.G. (1984). The Pennsylvania Infant and Family Development Project III: The origins of individual differences in infant-mother attachment: Maternal and infant con- tributions. Child Development, 55,718-728.

Cassidy, J., & Berlin, L.J. (1994). The insecure/ambivalent pattern of attachment: Theory and research. Child Development, 65,971-991.

Cohn, J.F., Campbell, S.B., & Ross, S. (1991). Infant response in the still-face paradigm at 6 months pre- dicts avoidant and secure attachment at 12 months. Development and Psychopathology, 3.367-376.

Cohn, J.F., & Tronick, E.Z. (1988). Mother-infant face-to- face interaction: Influence is bidirectional and unre- lated to periodic cycles in either partner’s behavior. Developmental Psychology, 24.386-392.

Dumas, J.E. (1986). Controlling for autocorrelation in social interaction analysis. Psychological Bulletin, 100, 125-127.

Dumas, J.E. (1987). INTERACT-A computer-based cod- ing and data management system to assess family interactions. In R.J. Prinz (Ed.), Advances in behav- ior-al assessment of children and families (Vol. 3). New York: JAI Press.

Dumas, J.E., & LaFreniere, P.J. (1993). Mother-child rela- tionships as sources of support or stress: A compari- son of competent, average, aggressive and anxious dyads. Child Development, 64, 1732-1754.

Dumas, J.E., LaFreniere, P.J., & Serketich, W.J. (1995). “Balance of power”: A transactional analysis of control in mother-child dyads involving socially competent, aggressive and anxious children. Journal of Abnormal Psychology, 104, 104-l 13.

Dunham, P., & Dunham, F. (1990). Effects of mother-infant social interactions on infants’ subse-

1 4 2 T a r a b u ls y , T e s s i e r, G a g n o n, a n d P i c h 6

quent contingency task performance. Child Devel- between quality of attachment and later competence. opment, 6/,785-793. Child Development, 49, 547-556.

Dunham, P., & Dunham, F., Hurschman, A., & Alexander, T. (1989). Social contingency effects on subsequent perceptual-cognitive tasks in young infants. Child Development, 60, 1486-1496.

Millar, W.S. (1988). Smiling, vocal and attentive behavior during social contingency learning in seven and ten- month-old infants. Merrill Palmer Quarterly, 34. 301-325.

Egeland, B., & Farber, E.A. (1984). Infant-mother attach- ment: Factors related to its development and changes over time. Child Development, 55.753-77 1.

Goldsmith, H.H., & Alansky, J.A. (1987). Maternal and infant temperamental predictors of attachment: A meta-analytic review. Journal of Consulting and Clinical Psychology, 55, 805-8 16.

Moran, G., Dumas, J.E., & Symons, D.K. (1992). Approaches to sequential analysis and the descrip- tion of contingency in behavioral interaction. Behavioral Assessment, 14.65-92.

Grossmann, K., Grossmann, K.E., Spangler, G., Suess, G.. & Unzner, L. (1985). Maternal sensitivity and new- borns’ orientation responses as related to quality of attachment in Northern Germany. In I. Bretherton & E. Waters (Eds.), Growing points in attachment theory and research. Monographs of the Society for Research in Child Development. 50 (l-2, Serial No. 209).

Moran, G., Pederson, D.R., & Tarabulsy, G.M. (in press). Le role de la thdorie d’attachement dans I’analyse des interactions mere-enfant durant la petite enfance: Descriptions precises et interpretations sig- nificatives [The role of attachment theory in the analysis of mother-infant interactions: Targeted description and meaningful interpretation]. In G.M. Tarabulsy & R. Tessier (Series and Volume Eds.), D' Enfance Vol. I : Families et m&es: Ie dbeloppe- ment social des enfants en contexte. Quebec: Les Presses de I’UniversitC du Quebec.

Isabella, R.A. (1993). Origins of attachment: Maternal interactive behavior across the first year. Child Development, 64, 605-62 I.

Isabella, R.A., & Belsky, J. (1991). Interactional synchrony and the origins of infant-mother attachment: A replication study. Child Development. 62, 373-384.

Isabella, R.A., Belsky, J., & von Eye, A. (1989). Origins of infant-mother attachment: An examination of inter- actional synchrony during the infant’s first year. Developmental Psychology, 25, 12-2 I.

Pederson, D.R., & Moran G. (1995). A categorical descrip- tion of attachment relationships in the home and its relation to q-sort measures of infant-mother interac- tion In E. Waters, BE. Vaughn, G. Posada & K. Kondo-Ikemura (Eds.), caregiving, cultural and cognitive perspectives on secure-base behavior and working models: New growing points of attachment theory and research. Monographs of the Society for Research in Child Development, 60 (Serial No. 244).

LaFreniere, P.J., & Dumas, J.E. (1992). A transactional analysis of early childhood anxiety and social withdrawal. Del,elopment and Psychopathology, 4, 385402.

Pederson, D.R., & Moran, G. (in press). Expressions of attachment outside of the Strange Situation. Chrld Development.

Lamb, M.E., Thompson, R.A., Gardner, W., & Chamov, E.L., (1985). Infant-mother attachment: The origins and developmental significunw of individual differ- ences in Strange-Situation hehuvior. Hillsdale, NJ: Erlbaum.

Schneider-Rosen, K., & Rothbaum, F. (1993). Quality of parental caregiving and security of attachment. Developmental Psychology, 29,358-367.

Smith, P.B., & Pederson, D.R. (1988). Maternal sensitivity and patterns of infant-mother attachment. Child Development, 59, 1097-l 101.

Lewis. M., & Feiring, C. (1989). Infant, mother, and mother- infant interaction behavior and subsequent attach- ment. Child Development, 60, 831-837.

Sroufe, L.A., & Waters, E. (1977). Attachment as an organizational construct. Child Development. 48. 1184-I 199.

Lewis, M., Sullivan, M.W., Ramsay, D.S., & Alessandri, S.M. (1992). Individual differences in anger and sad expressions during extinction: Antecedents and con- sequences Infant Behavior and Development, 15. 443452.

van IJzendoom, M.H., Goldberg, S., Kroonenberg, P.M.. & Frankel, O.J. (1992). The relative effects of mater- nal and child problems on the quality of attachment: A meta-analysis of attachment in clinical samples. Child Development, 63, 840-858.

Main, M., & Weston, D.R. (I 982). Avoidance of the attach- ment figure in infancy: Descriptions and interpreta- tions. In C.M. Parkes & J. Stevenson-Hinde (Eds.). The place of attac~hment in human hehal?or. New York: Basic Books.

Mangelsdorf, S., Gunnar, M., Kestenbaum, R., Lang, S., & Andreas, D. (1990). Infant proneness-to-distress temperament, maternal personality, and mother- infant attachment: Associations and goodness of fit. Child Development. 61, 820-83 1.

Maslin, C.A., & Bates, J.E. (1983, April). Precursors of arlxious and secure attachments.. A multivariate model at age 6 months. Paper presented at the meet- ing of the Society for Research in Child Develop- ment, Detroit, MI.

Waters, E., & Deane, K. (1985). Defining and assessing individual differences in attachment relationships: Q-methodology and the organization of behavior in infancy and early childhood. In I. Brethcrton & E. Waters (Eds.), Growing points ofattac~hment theoq and research. Monographs of the Society fo! Research in Child Dei,elopment. 50. (l-2. Serial No. 209).

Watson, J.S. (1979). Perception of contingency as a deter- minant of social responsiveness. In E.B. Thoman (Ed.), Origins of the infunt’s social ,.e.rponsi~,ene.ss. Hillsdale. NJ: Erlbaum.

Wentworth, N., & Haith, M.M. (1992). Event-specific expecytions of 2 and 3 month old infants. Develop mental Psycholog!. 28. 842-850.

Matas. L., Arend, R.A., & Sroufe, L.A. (1978). Continuity of adaptation in the cecond year. The relationship 02 December 1994; Revised 15 August 1995 H

A t t a c h m e n t a n d In f a n t R e s p o n s iv e n e s s 1 4 3

APPENDIX

Description of Toys Used

Toy I Geometric Shapes Puzzle: Six large- sized, brightly colored wooden shapes, with knobs in the middle to facilitate their handling by infants, were placed in their appropriate location on a 45 cm x 30 cm white wooden board. This toy was selected because infants regularly showed that they could take the pieces out of their location and tried to put them back in their place, but had difficulty without mother’s help. Additionally, mother sometimes had the infant remove more than one piece at a time, making the child confused as to the appropriate location.

Toy 2 Man in the Box: The toy was a red, trian- gular container, 10 cm x 10 cm x 8 cm, with a revolving door which the infant could open to take out a little man and close again to place the man back in through a hale on top of the toy. The toy was selected because infants were interested in getting the man out but seemed unsure of how to go about it. They tried the following: (a) to make the little man fall out of the hole; (b) to put their hand in the hole to take the man out (it was

much too small); (c) once they found a way to get the man out, they often tried to put him back into the container through the hole with the revolving door still open and blocking the hole. Toy 3 Red and Blue Ball With Geometric Shapes: This was perhaps the toy with which infants were most familiar (mothers sometimes reported having it at home), but it was clearly the most difficult. It was a red and blue ball, which had 10 holes of various geometric shapes. It was hollow and contained plastic pieces which were to be taken out of the ball and put back in by placing each in its appropri- ate hole. The ball had two diametrically opposed knobs. By pulling the two knobs apart, the ball opened up in the middle to let the pieces fall out. Once all pieces were out, infants had to put them back in. Although this toy was clearly above the skills of infants in this study, infants nevertheless showed a great deal of interest in it and tried to take the pieces out of the ball by either pulling on the knobs them- selves (only mother was able to do this) or putting their hands in the holes. Once the pieces were out, they tried to place them in the ball through any hole.