Full Text


E-mail, print, or save the full text.




Full-text source: PsycARTICLES_FT

Maternal Sensitivity During Infancy and Subsequent Life Events Relate to Attachment Representation at Early Adulthood

Author: Beckwith, Leila; Cohen, Sarale E. Hamilton, Claire E. Source: Developmental Psychology May 1999 Vol. 35, No. 3, 693-700 ISSN: 0012-1649 Number: dev353693 Copyright: For personal use only-not for distribution

Thepresent study examines the issue of whether early experience, specifically maternal sensitivity to infant signals, is linked to adult understanding of attachment relationships. Does adult mental representation of attachment reflect actual infancy and childhood experiences (Fox, 1995)? Or is attachment representation a reconstruction derived from later, more concurrent experiences (Lamb, Thompson, Gardner, & Charnov, 1985; Seifer & Schiller, 1995)? To date, data are lacking. Whereas evidence exists for cross-generational transmission, that is, adults" understanding of attachment has been shown to affect their sensitivity to their offspring and their offspring"s own security of attachment (van IJzendoorn, 1995), evidence for historical continuity within an individual from experienced maternal sensitivity in infancy to adult representation of attachment is absent.

Maternal sensitivity to infant signals has been shown to influence a range of children"s developing cognitive and social-emotional competencies, including biobehavioral regulation and cortisol levels in infancy (Spangler, Schieche, Ilg, & Maier, 1994); faster habituation and greater novelty preference in infancy (Bornstein & Tamis-LeMonda, 1989); better perceptual performance at age 4 years (Lewis, 1993); faster discrimination learning and increased IQ scores also at age 4 years, even when infant skill in information processing and maternal noncontingent attentiveness were partialed out (Bornstein & Tamis-LeMonda, 1989); and school achievement at 7 years of age (Bradley, 1989). Furthermore, there is much empirical evidence that indicates that sensitive maternal care during infancy is a significant causal factor in infant security (Ainsworth, Blehar, Waters, & Wall, 1978; Egeland & Farber, 1984; Isabella, 1993; Smith & Pederson, 1988; Ward & Carlson, 1995; Zeanah et al., 1993). Moreover, maternal insensitivity has been linked to later behavioral problems. In children from low-income families, insensitivity to male toddlers-particularly those who were persistent in attention seeking, aggressive acts, and noncompliance-resulted in increased disruptive and aggressive behavior at age 3 (Shaw, Keenan, & Vondra, 1994). Conversely, maternal sensitivity with infants born prematurely was correlated with fewer teacher-reported behavior problems, and the children chose more prosocial and less aggressive responses on a social problem-solving measure (Goldberg, Lojkasek, Gartner, & Corter, 1989).

However, the continuity from infancy experience to adult representation would not be expected to be complete. Bowlby (1973) proposed that the organization of attachment, begun during infancy, continues to evolve in light of attachment experiences during the first 5 years of childhood and remains sensitive to changes even into adolescence. To date, Bowlby"s hypotheses about stability and lawful discontinuities of attachment organization are still speculative in several ways. Not only has the role of the early years in shaping adult representations of attachment just begun to be demonstrated, but lawful discontinuities and the question of what factors, under what conditions, disrupt the link between early and later attachment quality has also just begun to be studied (Hamilton, 1995; van IJzendoorn, 1995; Waters, Treboux, Crowell, Merrick, & Albersheim, 1995).

A prospective longitudinal research study from birth to 18 years of age of children born prematurely (Cohen, 1995) provided the opportunity to investigate linkages between maternal sensitivity during infancy, maternal sensitivity during adolescence, attachment representation at early adulthood, and discontinuity as affected by negative life events. Infants born prematurely have more difficulties in regulating emotion and in initiating and maintaining pleasurable social interactions than do full-term and healthy infants (Field, 1978). In this study, we reasoned that preterm infants would challenge individual differences in maternal sensitivity, particularly in the first months of life, in the modulation of the young infants" negative states of arousal (Goldberg & DiVitto, 1983). Thus, long-term consequences of differences in maternal sensitivity might be highlighted. By the end of the 1st year, few if any differences between full-term and preterm children have been found in patterns of infant-mother interaction and attachment security (Pederson & Moran, 1995; van IJzendoorn, Goldberg, Kroonenberg, & Frenkel, 1992). Therefore, we expected that patterns of attachment representation at early adulthood would be similar to those of normative populations.

We examined three hypotheses. First, the historical configuration of the infant-mother relationship, with specific regard to maternal sensitivity, would be significantly associated with attachment representation at early adulthood. Second, maternal sensitivity during early adolescence would be significantly associated with attachment representation at early adulthood. Third, childhood adverse life events related to attachment would moderate the link between maternal sensitivity during infancy and adult representation. Thus, the experience of sensitive maternal care with many adverse life events would yield a lower proportion of secure attachment representations than the combination of maternal sensitivity and the absence of adverse events. Similarly, the combination of early maternal insensitivity coupled with the absence of adverse life events would yield a higher proportion of secure individuals than among those who experienced maternal insensitivity with many adverse life events.



All children born prematurely (gestation equal to or less than 37 weeks and birth weight of 2,500 g or less) at the University of California at Los Angeles Medical Center from 1972 to 1974, with the exception of those with congenital anomalies or severe sensory or physical disabilities and their parents were recruited for the initial longitudinal study (Parmelee, Kopp, & Sigman, 1976). At age 18, 105 of the 126 children who had been studied intensively from birth to age 2 years were evaluated. Of these 105 participants, the Adult Attachment Interview (AAI; Main & Goldwyn, 1993) was not available for 16 participants because of lack of time, failure of audio recordings, or failure of out-of-town participants who completed questionnaires to return to the University of California, Los Angeles for testing, and observations of mother-child interaction during infancy were not complete for an additional 3 participants.

Therefore, the participants of this study became 86 eighteen-year-olds, 33 women and 53 men. Their mean gestational age had been 33 weeks (range = 25-37 weeks), and their mean birth weight was 1,878.5 g (range = 800-2,500 g). Perinatal problems ranged from minor to severe, and days of hospitalization ranged from 2 to 88, with a mean of 23.2 days. Families were heterogeneous in ethnic group, socioeconomic status, and parental education. There were 56 (65%) participants who had been born to English-speaking families, 24 (28%) born to Spanish-speaking families, and 6 (7%) from other cultures such as Korea or India. Years of maternal education ranged from 1 to 17, with a mean of 11.4 years, and socioeconomic status ranged from 8 to 66, with a mean of 36.8 on the Four-Factor Index of Social Status (Hollingshead, 1975). At the time of recruitment, most infants were born into intact families, defined as those in which the two biological parents lived together, whether married or not. Exceptions were 4 participants born to single mothers who did not live with the fathers of the infants and 1 participant who was adopted after placement in a foster home. By age 18, 37 (46%) participants of those born into families in which both biological parents resided had families who had been reorganized by separation and divorce.

We used Hotelling T 2 and chi-square analyses, as appropriate, to compare the 86 participants of this study with the 40 participants of the original infancy sample who were not included in this study. The two groups were not significantly different in terms of gender, birth weight, gestational age, length of hospitalization, or Gesell and Amatruda (1947) test scores at age 2. The two groups were different in terms of ethnic group membership, maternal education, and social class. The participants of this study had a smaller percentage of Spanish-speaking parents, more years of maternal education, and higher socioeconomic class than the attrition group.


Maternal sensitivity during infancy (Beckwith & Cohen, 1984).

Naturalistic home observations were conducted by Leila Beckwith and Sarale E. Cohen when the children were 1, 8, and 24 months of age, corrected for prematurity. At 1 month, the infants were observed through an awake cycle, that is, awakening from sleep, feeding, diapering, playing, soothing, and all other activities until they were asleep again for at least 30 min. At 8 months, observations were of 50 min of awake time, plus a feeding; at 24 months, 50 min of awake play time were observed.

At 1 month and 8 months, the observer used a precoded checklist and every 15 s checked the occurrence of infant vocal, gaze, and motor behaviors; mother vocal, gaze, and touching-holding behaviors to the infant; and maternal response to infant attachment signals of crying, vocalizing, gazing, and smiling. At 24 months, event sampling was used. Each event between mother and child was scored as to the initiator, the action, and the response or lack of response of the recipient.

At the time of observation, rather than using the entire interactive relationship between infant and adult, an a priori selection of relevant indicators of sensitivity was used for each age, emphasizing appropriate and prompt responsiveness to infant cues (Bowlby, 1973). Because developmental appropriateness had to be taken into account, the selection was guided by developmental changes in the meaning and frequency of attachment behaviors as well as the increasing control by the infant of the interactions. At 1 month, the score summed maternal positive attentiveness to the infant, talk to the infant, holding of the infant, contingency to infant distress, and mutual gaze. Only promptness of effort in response to infant distress was coded, not success in maintaining optimal state, in order to reduce the contribution of infant variability in emotion regulation. The 8-month measure summed maternal positive attentiveness to the infant, talk to the infant, contingency response to nondistress vocalizations, mutual gaze, and floor freedom. The latter scored maternal encouragement or interference with emerging exploration. At 24 months, positive attentiveness of the mother and reciprocal interactions between mother and toddler, whether vocal, verbal, or behavioral, were summed. Maternal responsiveness to the child"s bids, the child"s responsiveness to the mother"s initiations, and each one"s cooperative acts in maintaining an interaction were coded. Thus, the bidirectional nature and the joint control of interaction, not simply maternal behavior (Nicholls & Kirkland, 1996), were explicitly included in the coding. Interrater reliability was analyzed for 10 infants, across the three ages, who were simultaneously observed and coded by two judges. The interrater correlations for each behavior ranged from .80 to .95.

A pilot sample of 25 full-term infants was used to devise standardized scores for the number of such indicators that occurred at each age. The average composite score for the pilot sample was arbitrarily set at 100, with a standard deviation of 15. It was discovered later that the average score, particularly at 1 month, was significantly higher for the preterm sample as compared with the standardization full-term sample (Beckwith & Cohen, 1984).

The quality of attachment of the infant to the mother was not measured. It is not known, therefore, whether the maternal sensitivity measure would have predicted infant attachment security. However, our measure of sensitivity was similar to another measure also derived from behavioral counts (Isabella, Belsky, & von Eye, 1989) that did find a link between maternal sensitivity and infant attachment.

12-year maternal sensitivity (Beckwith, Rodning, & Cohen, 1992).

Maternal responsiveness was assessed from videotaped 20-min interactions in two laboratory tasks that required the cooperative participation of both mother and child. The first task required the mother and the child to agree on adjectival descriptions of an ideal person. The second task required the mother and the child to use an Etch-a-Sketch to copy a design. The videotapes were rated through use of a Q sort by a coder who had no knowledge of infancy or 18-year family or child data. The Q sort used 31 items that were distributed by the rater into a forced-choice, nine-step profile. Correlation with a prototype was chosen as the unit of analysis as a way to reduce data and thereby to reduce Type I error and to focus on the pattern of behavior rather than just discrete behaviors. The criterion sort for responsiveness was developed by averaging the judgments of five developmental psychologists. The Q-sort prototype described the most characteristic qualities of the responsive parent as being responsive to the child, showing clarity and consistency in the parental role, giving explanations for parental decisions, as well as encouraging independence in the child. The least characteristic qualities were showing criticism, hostility, and rejection toward the child; surrendering control of the situation to the child; and being competitive with the child. Interrater reliability analyses for 9 participants resulted in a mean item-total correlation of .84. Construct validity was demonstrated by more responsive mothers having children with higher school achievement and greater self-esteem and by their teachers reporting fewer behavioral and social problems (Beckwith et al., 1992).

Adult Attachment Interview.

When the participants were 18 years of age, the AAI (Main & Goldwyn, 1993) was administered to them by an interviewer who was unaware of the infancy or 12-year observations. The interviews were audiotaped and transcribed. Claire E. Hamilton, who was trained by Mary Main and Erik Hess, reliable with them on over 30 cases separate from this study, and naive to all other information about the participants, coded and classified the transcripts. A subset of 22 cases drawn from this study (n = 10) and from comparable high-risk groups (n = 12) was used to establish interrater reliability for this study between Claire E. Hamilton and two other coders. Overall agreement across the three scoring categories of secure, dismissing, and preoccupied was 80% (κ = .67).

Numerous studies have demonstrated construct validity through substantial predictive, concurrent, and retrospective correspondence between parents" mental representation of attachment as assessed by the AAI and their infants" attachment security as assessed in the Strange Situation (Fonagy, Steele, & Steele, 1991; Main, Kaplan, & Cassidy, 1985; van IJzendoorn, 1995). Discriminant validity has been shown by lack of correspondence between AAI classification and cognitive and memory tests, narrative styles, IQ, a variety of personality traits, and social desirability (Bakermans-Kranenburg & van IJzendoorn, 1993). Reliability over time, even with different interviewers, has been established (Sagi et al., 1994).

The AAI assesses an individual"s generalized representation of attachment by means of a structured interview to elicit information about childhood experiences and the individual"s ideas about the influence of those experiences on his or her personality. The interview is scored from a verbatim transcript by using scales that characterize discourse style (idealization, lack of memory, derogation, anger, passivity, and coherence). The style of discourse, not the quality of experience reported, is used to assign the adult to one of three major classifications: secure autonomous, insecure-dismissing, or insecure-preoccupied.

Individuals classified as secure easily describe diverse childhood experiences, good or bad; value attachment relationships; and view their experiences as influential to their development. Adults with insecure classifications present incoherent descriptions and a lack of integration of their experiences and their influence. Those classified as dismissing deny or devalue the influence of attachment relationships, have difficulty recalling events, and often idealize or normalize their experiences with their parents. Adults classified as preoccupied display confusion in their integration of past experiences and seem either passively or angrily currently entangled in those experiences. One additional insecure classification, unresolved, either as to the death of an attachment person or as to physical and sexual abuse, was not included in the present study because it occurred infrequently. The 10 unresolved participants were force classified into the other three classifications as suggested by Main and Goldwyn (1993).

Negative life events.

A core set of life events, suggested by Bowlby (1969) as directly interfering with a family"s ability to care for a child and thus to maintain attachment security, was identified. These events included never establishing a family, as in foster care; family group intact but not functioning effectively because of chronic severe illness of parent or child; family group intact but parent incapacitated by psychiatric disorder; family group dissolved by death of a parent, separation, or divorce; and physical or sexual abuse of the child.

Parents were interviewed when the children were 2, 5, 8, 12, and 18 years of age and questioned about the occurrence of these events, including parental divorce, psychiatric illness, death, serious illness of parent or child, and physical abuse or sexual abuse of the child. On the basis of parental report, the negative life events were scored as being present or absent for each participant.


The order of presentation of analyses is as follows: Preliminary analyses comparing the distribution of AAI classifications with other nonclinical samples and the association with demographic factors of gender, maternal education, socioeconomic class, and immigrant status are described first. Next, the first hypothesis, the link between maternal sensitivity during infancy and attachment representation at early adulthood, is examined by using analysis of variance (ANOVA) procedures to test differences in level of maternal sensitivity among attachment classifications. Then, the second hypothesis, the association between maternal sensitivity during adolescence and attachment representation, is analyzed similarly. The cumulative influence on attachment representation of maternal sensitivity during both infancy and adolescence is analyzed by using a multidimensional contingency table that compares the distribution of AAI classifications under combined conditions of sensitivity-insensitivity in infancy and at 12 years of age. Finally, the third hypothesis, the moderating influence of adverse childhood life events on the link between maternal sensitivity and adult attachment representation, is tested by using chi-square analyses, comparing the AAI classifications as to presence or absence of adverse life events during childhood, and a multidimensional contingency table that compares the distribution of AAI classifications under both conditions of sensitivity-insensitivity and presence-absence of negative life events.

Distribution of Adult Attachment Interview Classifications

The overall distribution of AAI classifications-42% secure, 17% preoccupied, and 41% dismissing-deviated somewhat from that of other nonclinical samples (van IJzendoorn & Bakermans-Kranenburg, 1996; van IJzendoorn et al., 1992) by an increase in dismissing participants and a slight decrease in secure participants. The deviation was statistically significant, χ2(2, N = 86) = 7.03, p = .03.

Chi-square analyses showed no evidence that the attachment classifications were associated with socioeconomic class or maternal education. However, there was a gender difference, with men being significantly more likely to be dismissing than women. Furthermore, children who grew up with parents who were nonnative English speakers were more likely to be dismissing and less likely to be secure than were children of native English speakers. The distribution of AAI classifications for the children of native English speakers was more comparable to that of the normative sample, with 28% dismissing and 48% secure, whereas the nonnative English-speaking group with 63% dismissing and 30% secure was clearly deviant from the norm of 58% secure and 24% dismissing (van IJzendoorn & Bakermans-Kranenburg, 1996).

Early Maternal Sensitivity and Adult Attachment Interview Classifications

Antecedents to the AAI classifications were analyzed by a 3 (attachment classification) × 3 (time) repeated measures ANOVA, with the 1-, 8-, and 24-month home-observation standardized composite scores as the dependent variables. Image Table 1 lists means and standard deviations for the analyses. As shown in Image Figure 1, there was a significant main effect for attachment classification, F(2, 83) = 4.76, p < .01. Tukey"s pairwise comparisons indicated that dismissing participants received lower maternal sensitivity scores than did the other two attachment groups, which did not differ from each other.

In addition, there was a significant main effect for time, F(2, 166) = 9.26, p < .001, and a significant interaction between time and attachment classification, F(4, 166) = 2.94, p < .02. Mothers of secure and preoccupied participants engaged in more sensitive, responsive acts when their infants were 1 month and 24 months old than when they were 8 months old, F(2, 166) = 9.86, p < .0001, and F(2, 166) = 4.46, p < .01, for secure and preoccupied participants, respectively. In contrast, mothers of dismissing participants maintained a low frequency of sensitive acts at all observations, without any changes dependent on child"s age, F(2, 166) = 0.01, p > .98. Furthermore, the classification groups differed in level of maternal sensitivity received at 1 month, F(2, 167) = 8.67, p < .0003, with dismissing significantly lower than secure or preoccupied, as shown by Tukey"s pairwise comparisons. There were no differences at 8 months, F(2, 167) = 0.32, p > .7. At 24 months, the groups again differed, F(2, 167) = 2.96, p < .05, with dismissing significantly different from preoccupied, as shown by Tukey"s pairwise comparisons. Despite the significance of the association between early maternal sensitivity and later AAI classification, the strength of the relation was very modest, accounting for 10% of the variability in classification.

Antecedents to the attachment classifications were not different for boys as compared with girls or for children of nonnative English speakers as compared with children of native English speakers, when these variables were included in the analyses. Mothers tended to be more sensitive to infant girls than to infant boys, but the level and pattern of sensitivity during infancy was not different for dismissing boys and girls, who received lower sensitivity scores than the other two attachment groups. Similarly, nonnative English speakers received lower sensitivity scores than native English speakers. Level and pattern of sensitivity during infancy was not different for dismissing participants reared by English-speaking or non-English-speaking mothers.

Maternal Sensitivity at 12 Years of Age and Adult Attachment Interview Classifications

A one-way ANOVA, with attachment classification as the grouping variable and maternal sensitivity at age 12 years as the dependent variable, indicated that attachment classification was unrelated to maternal sensitivity in early adolescence. Furthermore, the measure of maternal sensitivity at age 12 years was unrelated to the measures of maternal sensitivity during infancy, r(84) = .18, .03, and .05 (ps > .10), respectively, for 1-, 8-, and 24-month measures.

Maternal Sensitivity During Infancy and at Age 12 Years and Adult Attachment Interview Classifications

To determine if there were additive relations between maternal sensitivity during the two periods and attachment classifications, four groups were constructed: mothers high in responsiveness during both periods, mothers low in responsiveness during both periods, mothers high in responsiveness early and low later, and mothers low in responsiveness early and high later. High and low sensitivity during infancy was determined by dividing the participants at the median on the 1-month measure, selected because the 1-month measure showed the strongest link to the AAI classifications. Similarly, high and low sensitivity at age 12 was determined by dividing the participants at the median; mothers whose scores correlated .40 or higher with the prototype of sensitivity were considered to be high in sensitivity. The chi-square derived from the multidimensional contingency table was significant, χ2(3, N = 60) = 9.41, p = .02. There were significantly more secure and fewer dismissing individuals in the group that had received increased maternal sensitivity during both infancy and adolescence, whereas there were significantly fewer secure and more dismissing individuals in the group that had received consistently decreased maternal sensitivity.

Negative Life Events and Adult Attachment Interview Classifications

Preliminary analyses indicated that the demographic variables of gender and immigrant status were not associated with an increase in negative life events, and, therefore, these variables were not included in further analyses. Image Table 2 indicates the number of participants within each attachment classification who did and did not experience negative life events, including foster care, physical or sexual abuse, serious chronic illness, parental death, or parental divorce by age 12. Approximately one third of those with secure or dismissing representations of attachment had suffered a negative life event, whereas almost all preoccupied participants had. Thus, adverse life events were linked to insecurity of the preoccupied kind.

The most common negative life event was parental divorce. Family dissolution caused by divorce or separation occurred throughout childhood for children in all attachment classifications. However, the age of occurrence and the percentage affected differed among the attachment groups. For those who were later classified as secure, the majority lived with both biological parents through age 12, that is, 74% lived with both parents at age 5, 69% at age 12, and 45% at age 18. Similarly, for those who were later classified as dismissing, 88% were in intact families at age 5, 71% at age 12, and 62% at age 18. In contrast, those who were later classified as preoccupied experienced significantly more family dissolution and at an earlier age. By age 5, only a minority of individuals, 41%, were living with both parents, decreasing to 33% by age 12 and to 25% by age 18. Chi-square analyses of attachment classification and parental divorce showed significant differences between the preoccupied group and the other two groups beginning at age 5, χ2(2, N = 86) = 10.35, p < .006.

Early Maternal Sensitivity, Negative Life Events, and Adult Attachment Interview Classifications

As shown in Image Table 3, to determine the moderating influence of adverse life events on the association between maternal sensitivity during infancy and later attachment representation (Baron & Kenny, 1986), participants who had received more sensitive care early on were grouped as to absence or presence of negative life events, and participants who had experienced less sensitive care were grouped as to absence or presence of negative life events. A 4 × 3 multidimensional contingency analysis resulted in χ2(6, N = 86) = 23.50, p < .0007.

The major contribution to the chi-square was found in three cells. The number of secure individuals was significantly increased among those who had experienced more sensitive care during infancy and no adverse life events during childhood. (Seventy-one percent of that cell was classified as secure, as compared with approximately one third of the other three cells.) The number of dismissing individuals was significantly increased among those who had experienced less sensitive care in the absence of negative life events. (Seventy percent of that cell was classified as dismissing, as compared with 48% dismissing in those who had experienced less sensitive care and negative life events and approximately 25% dismissing in those who had experienced more sensitive care with or without the occurrence of negative life events.) Preoccupied individuals were rare in the groups without negative life events. (There were 5% in the more sensitive care group and 0% in the less sensitive group.) Their occurrence was significantly increased in one group; they represented 39% of those who had experienced sensitive care and negative life events.

There was no link between maternal sensitivity during infancy and the likelihood of later adverse life events. As seen in Image Table 3, 17 of those individuals who experienced higher maternal sensitivity grew up in benign life circumstances, but 26 did not. Of those who experienced less sensitive caregiving, 20 grew up in benign conditions and 23 under adversity.


Attachment theory proposes that there is continuity from infancy experiences to state of mind about attachment during adulthood but that attachment representations remain open to experiences during childhood, past infancy (Bowlby, 1969, 1973, 1988). This study provides preliminary evidence for both propositions. The data are theoretically consistent with other research that has shown predictive continuity of attachment status within the individual from infancy behavior, in the Strange Situation, to discourse during young adulthood, in the AAI (Hamilton, 1995; Waters et al., 1995).

The present study adds to such findings by showing that actual experiences with the mother during infancy are associated with later adult representations. Continuity from infancy experiences to representation of attachment at age 18 was demonstrated for young adults whose representation of attachment was dismissing. Those young adults classified as dismissing were those who could not recall childhood events, or depreciated their importance, or normalized or presented an idealized portrait of their parents, unsupported by memories. They had been observed as infants, 16-17 years earlier, to have mothers who were significantly less sensitive and responsive to their signals than all other mothers.

Infancy experience, however, as we measured it, did not differentiate those with secure representations, who could coherently describe diverse childhood events, from those who were preoccupied, confused in their integration and passively or angrily still entangled in their childhood experiences. Moreover, the strength of the association between early sensitivity and later attachment representation was consistent with the literature finding that maternal sensitivity contributes to only a small portion of variability in attachment classification, even when assessed concurrently (van IJzendoorn, 1995).

Attachment experiences beyond infancy, within the relationship between the mother and the child and within the family, moderated the association between early sensitivity and later state of mind about attachment. Within the mother-child relationship, experiences during infancy appeared to be more salient for later attachment representations than were adolescent experiences, but there also appeared to be an additive influence. Higher maternal sensitivity during both periods was associated with significantly more secure and fewer dismissing individuals, whereas consistently lower maternal sensitivity was associated with more dismissing and fewer secure individuals.

The data are not consistent with the theoretical position that asserts that continuity between infancy experience and adult state of mind with respect to attachment derives from concurrent experience with attachment figures, often correlated with earlier experience (Lamb et al., 1985). Although our measure of the mother-child relationship at early adolescence had been shown in prior research to be associated with concurrent school achievement, IQ, self-esteem, and less behavior problems reported by teachers (Beckwith et al., 1992), it did not correlate with prior maternal sensitivity, and by itself did not contribute to later mental representation of attachment. Although it may be that the adolescent measure was less successful than the infancy measure in capturing the specific attachment aspects of the relationship, little continuity in maternal behavior over time, particularly under different observation conditions, has been noted in other studies (Goldberg et al., 1989). Under conditions of discontinuity in maternal behavior, we found infancy experiences rather than more concurrent experiences to be more salient to state of mind about attachment.

Over and above the mother-child relationship, the presence of severe negative attachment events during childhood potentiated insecure states of mind with respect to attachment. This was particularly true for preoccupied representations. The moderating influence of negative life events is consistent with the formulation of Bowlby (1988) that attachment organization remains open, past infancy, to childhood and adolescent attachment experiences. The findings are also consistent with those of Cummings and Davies (1996), who posited that insecurity based in the marital dyad tends to increase a sense of insecurity in the child, over and above the specific parent-child relationship. Divorce of parents was the most common negative life event for our participants. Although some participants in all categories experienced parental divorce, it occurred at an earlier age and for a larger proportion of the individuals in the preoccupied group than in either the secure group or the dismissing group. Moreover, individuals classified as preoccupied also experienced a disproportionately higher occurrence of other negative life events, including physical and sexual abuse, serious physical illness, and death of a parent. Such events were exceedingly rare in the lives of secure individuals, were more frequent but still rare in the lives of dismissing adults, but had occurred in the lives of half of the preoccupied young adults.

Life events was a moderating variable that specified the conditions under which maternal sensitivity operated. A high probability of a secure state of mind with regard to attachment occurred under the combined conditions of early maternal sensitivity and benign family life events. Early maternal sensitivity did not decrease the likelihood of the occurrence of later adversities nor did it increase the likelihood of secure representations in the face of adversity. Under conditions of adversity, secure attachment representations were decreased, and preoccupied representations were increased, despite early maternal sensitivity. Benign life events, however, in the absence of maternal sensitivity, did not increase the proportion of secure individuals but rather significantly increased the likelihood of a dismissing state of mind.

The attachment system is complex. On the one hand, secure and preoccupied individuals, in contrast to dismissing individuals, share acknowledgment of negative as well as positive experiences in attachment relationships and affirmation of the influence of those relationships on themselves. On the other hand, secure and dismissing individuals, in contrast to preoccupied individuals, share a lack of passive or angry entanglement in experiences carried from the past into the present. Given the complexity, it is reasonable that dismissing models are set apart from all others under some contingencies and that other contingencies differentiate preoccupied from other models.

Generalizability from the present study is limited by the measures of sensitivity used during infancy and adolescence. The differentiation of the secure from the preoccupied individuals on the basis of their infancy experiences may have been restricted by the relatively short and infrequent observations, masking the inconsistency of maternal sensitivity that is the most salient factor associated with the infancy analogue of a preoccupied state of mind (Cassidy & Berlin, 1994). A different measure that allowed for closer grained analyses of pacing and included maternal and child affect (Martin, 1989) might have sharpened the differentiation. Similarly, constructing the adolescent measure to be more evocative of attachment issues might have increased its salience.

Additional limitations of the present study include the omission of measures of the infants" characteristics and the nature of the sample. Several theorists (Fox, 1995; Seifer & Schiller, 1995; van IJzendoorn, 1995; Vaughn et al., 1992) have suggested that other variables, specifically child temperament, affect attachment organization. Because the effects of maternal sensitivity and negative life events were not uniform across the young adults in this study, we wonder if temperamental characteristics acted as a buffer for negative life events or whether some individuals, dependent on complications attendant to prematurity, may have been more vulnerable to adversities. Furthermore, the characteristic difficulty of preterm infants in emotion regulation in the first few months may have increased the salience of the 1-month measure of maternal sensitivity above other periods of infancy.

Our findings are further constrained by the age of administration of the AAI. Early adulthood-late adolescence is a period of psychological flux. Later in adulthood, many individuals are found who have been able to integrate their past adversities and to become "earned secure" (Main & Goldwyn, 1993). Such individuals coherently describe their negative experiences, view those experiences as affecting their present personalities, and yet value attachment relationships. Little is known about the conditions under which "earned secure" occurs. Additional studies would be very helpful.

Cultural prescriptions, independently and in interaction with immigrant status, influence adult responses to infant attachment behaviors and may cause differences in mother-infant relationships (Harwood, 1992). Although children of non-English-speaking immigrants in the present study were more likely to be insecure, we consider it unwise to conclude cultural differences from such a small sample, particularly when little other information is available in the literature and when culture, immigrant status, and very low social class co-occurred.

Male participants experienced less sensitive caregiving during infancy and were more likely than female participants to be classified as dismissing. Gender differences in attachment distribution either in infancy or in adulthood have not been reported in the literature (van IJzendoorn & Bakermans-Kranenburg, 1996), although a study of late adolescents did find that male adolescents used more dismissing strategies than did female adolescents (Kobak, Cole, Ferenz-Gillies, & Fleming, 1993). More examination of gender differences is warranted.

In conclusion, the finding of lawful continuity of attachment representation in early adulthood from objectively observed maternal sensitivity during infancy and early adolescence underlines the importance of maternal sensitivity for children"s development. The finding of lawful discontinuity, under conditions of adverse life events within the family, points to the necessity of an integrative approach of family ecology with individual experience (Thompson, in press).


Ainsworth M. D. S., Blehar M. C., Waters E., Wall S. Patterns of attachment., Hillsdale, NJ: Erlbaum., 1978

Bakermans-Kranenburg M. J., van IJzendoorn M. H. A psychometric study of the Adult Attachment Interview: Reliability and discriminant validity., Developmental Psychology, Vol 29, 1993, 870-880

Baron R. M., Kenny D. A. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations., Journal of Personality and Social Psychology, Vol 51, 1986, 1173-1182

Beckwith L., Cohen S. E. Home environment and cognitive competence in preterm children during the first 5 years., In A. W. Gottfried (Ed.), Home environment and early cognitive development: Longitudinal research(pp. 235-271). Orlando, FL: Academic Press., 1984

Beckwith L., Rodning C., Cohen S. Preterm children at early adolescence and continuity and discontinuity in maternal responsiveness from infancy., Child Development, Vol 63, 1992, 1198-1208

Bornstein M. H., Tamis-LeMonda C. S. Maternal responsiveness and cognitive development in children., New Directions for Child Development, Vol 43, 1989, 49-61

Bowlby J. Attachment and loss., New York: Basic Books., 1969

Bowlby J. Separation., New York: Basic Books., 1973

Bowlby J. A secure base: Clinical applications of attachment theory., London: Routledge., 1988

Bradley R. H. HOME measurement of maternal responsiveness., New Directions for Child Development, Vol 43, 1989, 63-74

Cassidy J., Berlin L. J. The insecure/ambivalent pattern of attachment: Theory and research., Child Development, Vol 65, 1994, 971-981

Cohen S. E. Biosocial factors in early infancy as predictors of competence in adolescents who were born prematurely., Journal of Developmental & Behavioral Pediatrics, Vol 16, 1995, 36-41

Cummings E. M., Davies P. Emotional security as a regulatory process in normal development and the development of psychopathology., Development and Psychopathology, Vol 8, 1996, 123-139

Egeland B., Farber A. Infant-mother attachment: Factors related to its development and changes over time., Child Development, Vol 55, 1984, 753-771

Field T. The three Rs of infant-adult interaction: Rhythms, repertoires, and responsivity., Journal of Pediatric Psychology, Vol 3, 1978, 131-136

Fonagy P., Steele H., Steele M. Maternal representations of attachment during pregnancy predict the organization of infant-mother attachment at one year of age., Child Development, Vol 62, 1991, 891-905

Fox N. A. Of the way we were: Adult memories about attachment experiences and their role in determining infant-parent relationships: A commentary on van IJzendoorn (1995)., Psychological Bulletin, Vol 117, 1995, 404-410

Gesell A., Amatruda C. S. Developmental diagnosis (2nd ed., ). New York: Hoeber., 1947

Goldberg S., DiVitto B. A. Born too soon., San Francisco: Freeman., 1983

Goldberg S., Lojkasek M., Gartner G., Corter C. Maternal responsiveness and social development in preterm infants., New Directions for Child Development, Vol 43, 1989, 89-104

Hamilton C. E. Continuity and discontinuity of attachment from infancy through adolescence., Paper presented at the biennial meeting of the Society for Research in Child Development, Indianapolis, IN.

Harwood R. L. The influence of culturally derived values on Anglo and Puerto Rican mothers" perceptions of attachment behavior., Child Development, Vol 63, 1992, 822-839

Hollingshead A. The Four-Factor Index of Social Status., Unpublished manuscript., 1975

Isabella R. A. Origins of attachment: Maternal interactive behavior across the first year., Child Development, Vol 64, 1993, 605-662

Isabella R. A., Belsky J., von Eye A. Origins of infant-mother attachment: An examination of interactional synchrony during the infant"s first year., Developmental Psychology, Vol 25, 1989, 12-21

Kobak R. R., Cole H. E., Ferenz-Gillies R., Fleming W. S. Attachment and emotion regulation during mother-teen problem solving: A control theory analysis., Child Development, Vol 64, 1993, 231-245

Lamb M. E., Thompson R. A., Gardner W. P., Charnov E. L. Infant-mother attachment., Hillsdale, NJ: Erlbaum., 1985

Lewis M. D. Early socioemotional predictors of cognitive competency at 4 years., Developmental Psychology, Vol 29, 1993, 1036-1045

Main M., Goldwyn R. Adult Attachment Classification System., Unpublished manuscript, University of California, Berkeley., 1993

Main M., Kaplan N., Cassidy J. Security in infancy, childhood and adulthood: A move to the level of representation., Monographs of the Society for Research in Child Development, 50(1-2, Serial No. 209)., 1985

Martin J. A. Personal and interpersonal components of responsiveness., New Directions for Child Development, Vol 43, 1989, 5-14

Nicholls A., Kirkland J. Maternal sensitivity: A review of attachment literature and definitions., Early Child Development and Care, Vol 20, 1996, 55-65

Parmelee A. H., Kopp C. B., Sigman M. Selection of developmental assessment techniques for infants at risk., Merrill-Palmer Quarterly of Development and Behavior, Vol 22, 1976, 177-199

Pederson D. R., Moran G. A categorical description of infant-mother relationships in the home and its relation to Q-sort measures of infant-mother interaction., Monographs of the Society for Research in Child Development, 60(2-3, Serial No. 244)., 1995

Sagi A., van IJzendoorn M. H., Scharf M., Koren-Karie N., Joels T., Mayseless O. Stability and discriminant validity of the Adult Attachment Interview: A psychometric study in young Israeli adults., Developmental Psychology, Vol 30, 1994, 988-1000

Seifer R., Schiller M. The role of parenting sensitivity, infant temperament, and dyadic interaction in attachment theory and assessment., Monographs of the Society for Research in Child Development, 60(2-3, Serial No. 244)., 1995

Shaw D. S., Keenan K., Vondra J. I. Developmental precursors of externalizing behavior: Ages 1 to 3., Developmental Psychology, Vol 30, 1994, 355-364

Smith P. B., Pederson D. R. Maternal sensitivity and patterns of infant-mother attachment., Child Development, Vol 59, 1988, 1097-1101

Spangler G., Schieche M., Ilg U., Maier U. Maternal sensitivity as an external organizer for biobehavioral regulation in infancy., Developmental Psychobiology, Vol 27, 1994, 425-437

Thompson R. A. Early sociopersonality development., Social, emotional, and personality development (5th ed.), In W. Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child psychology: Vol. 3.New York: Wiley.

van IJzendoorn M. H. Adult attachment representations, parental responsiveness, and infant attachment: A meta-analysis on the predictive validity of the Adult Attachment Interview., Psychological Bulletin, Vol 117, 1995, 387-403

van IJzendoorn M. H., Bakermans-Kranenburg M. J. Attachment representations in mothers, fathers, adolescents, and clinical groups: A meta-analytic search for normative data., Journal of Consulting and Clinical Psychology, Vol 64, 1996, 8-21

van IJzendoorn M. H., Goldberg S., Kroonenberg P. M., Frenkel O. J. The relative effects of maternal and child problems on the quality of attachment: A meta-analysis of attachment in clinical samples., Child Development, Vol 63, 1992, 840-858

Vaughn B. E., Stevenson-Hinde J., Waters E., Kotsaftis A., Lefever G. B., Shouldice A., Trudel M., Belsky J. Attachment security and temperament in infancy and early childhood: Some conceptual clarifications., Developmental Psychology, Vol 28, 1992, 463-473

Ward M. J., Carlson E. A. Associations among adult attachment representations, maternal sensitivity, and infant-mother attachment in a sample of adolescent mothers., Child Development, Vol 66, 1995, 69-79

Waters E., Treboux D., Crowell J., Merrick S., Albersheim L. From the Strange Situation to the Adult Attachment Interview: A 20-year longitudinal study of attachment security in infancy and early adulthood., Paper presented at the biennial meeting of the Society for Research in Child Development, Indianapolis, IN.

Zeanah C. H., Benoit D., Barton M., Regan C., Hirshberg L. M., Lipsitt L. Representations of attachment in mothers and their one-year-old infants., Journal of the American Academy of Child and Adolescent Psychiatry, Vol 32, 1993, 278-286















































PSYCARTCLES Full Text results for: 'kw: attachment and kw: infancy and ft: fulltext'. Record 1 of 1



English  |  Español  |  Français  |  Japanese |  Chinese (Traditional) |  Chinese (Simplified) |  Options  |  Comments  |  Exit


© 1992-2003 OCLC

© 1992-2003 OCLC
Terms & Conditions