The Adult Attachment Interview has focused attention on the roles mental representations play in adolescent and adult relationships. It has also made possible tests of a wide ranging hypotheses about the long term stability of attachment security, cross-generational transmission of individual differences, and the importance of attachment security in marriage.

Although conceptualized in terms of "attachment working models" and "states of mind with respect to attachment", the AAI provides only a sample of verbal behavior from which the coder makes inferences about the organization of underlying representations. Neither the architecture nore the content of the underlying representation is specified or directly measured.

This project begins with an explicit definition of a cognitive structure called the "secure base script". The elements of the script are: (1) Parent and child (or two adults) are interacting constructively, (2) something interferes or interrupts the interaction, (3) the child (or one partner) is distressed, (4) help is requested, (5) help is offered, (6) the offer of help is accepted, (7) the help is effective, (8) distress is relieved, and (9) constructive interaction is re-established. This script summarizes the common temporal-causal sequences experienced in secure base interactions throughout infancy, childhood, and beyond.

Dr. Waters has adapted a method she developed for research on the development of prose production skills to assess adolescents' and adults knowledge of the secure base script. Subjects are presented with a set 10-14 prompt words that suggest a secure base story line. Subjects are asked to build a brief story around the prompt words. Those for whom the secure base script is familiar and accessible immediately (if not necessarily consciously) assimilate the prompt word set to this familiar cognitive structure and tell a story organized around the secure base script. E.g. the prompt words mother, baby, blanket, play, bump, cry, etc. suggests a story in which mother and baby are playing on a blanket; baby somehow bumps something and begins to cry, etc. In contrast, subjects who do not know the secure base script might tell a story in which a mother takes her baby to the dry cleaning store to pick up a blanket; on the way they pass some children playing; the mother watches them instead of the road and bumps into the car ahead of her, etc.

This method is extremely easy to administer and score and provides highly reliable scores on the subject's knowledge of the secure base script. In addition, it can be adapted to assess generalized (a mother and her child) or specific (my mother and me) representations. Data comparing script knowledge using mother-child and adult-adult prompt word sets has shown that infant-parent and adult-adult relationships access a single secure base representation, not different representations for the two types of relationship. This is of considerable theoretical importance in research on the notion that infant-mother relationships are protypes for later love relationships.

.

One of the key limitations of the AAI it its failure to define a particular architecture for adult attachment repre-sentations. Our work suggests that important aspects of early attachment experience are represented as a script-like structure. Experimental psychologists have demonstrated that script-like structure has a number of distinctive effects on learning and memory.

We are using a variety of experimental paradigms including concept learning, prose production, free recall, and false recognition to validate the notion that attachment representations are literally (not metaphorically) script-like. We predict that subjects whose attachment experiences have been organized in terms of a secure base script will (a) learn more quickly to identify secure base versus non-secure base stories in a concept learning paradigm, (b) be much slower to produce passages from secure base prompt word sets presented in non-standard order, (c) have better recall for script-related propositions encountered in passages about close relationships, and (d) will be more likely to falsely recognize script relevant propositions in a recognition memory task.

Positive results in these studies would be strong evidence for the notion that the architecture of attachment representations is substantially organized around script-like temporal-causal structures. This would have important implications for the kinds of learning and memory processes involved in the development of attachment representations and for the mechanisms through which they effect behavior.

.

The promt word secure base script assessment is among the strongest correlates of the AAI. However, this alone does not imply that the measures are interchangable. It is entirely possible for two measures to be highly correlated and yet have very different patterns of correlations with other variables.

The close association of a mother's AAI classification with her infant's Strange Situation classification is one of the most influential findings in adult attachment research. We are currently conducting a cross-sectional study to determine whether, like maternal AAI classifications, mothers' knowledge and access to the secure base script is associated with their infant's Strange Situation classifications. Preliminary analyses conducted for SRCD 2003 indicate that the script based measure and the Strange Situation parallel those of the AAI. This is important evidence that the notion of a secure base script is relevant to understanding the AAI. It also supports the use of the script assessment in research that focuses on the secure vs. insecure distinction and requires greater specificity and economy than the AAI can provide.

Previous research has shown that attachment self report measures are not substantially or consistently related to the AAI. Nor are they, in our data, related to offspring's Strange Situation classification. This is further evidence that the AAI and the self report measures assess different constructs.

.

 The cross-cultural gnerality of attachment theory is an enduring topic of discussion and controversy. Bowlby described the capacity to form secure base relationships as a species specific, part of every human infant's genetic endowment. At the same time, there is ample evidence that culture is an important factor in many aspects of parental care and adult-adult relationships. Early research on the cross-cultural relevance of Bowlby's theory focused almost exclusively on the Strange Situation. Some argued saw evidence for generality in the fact that the procedure seemed "scorable" in diverse cultures. Others argued that rates of secure, avoidant, and resistant classifications were importantly different across cultures.

In fact, Bowlby's theory neither predicts nor requires that the Strange Situation be scorable or valid across cultures. Nor does it require simlar rates of Strange Situation classifications across cultures. It does require that infants in every culture have the capacity to organize their behavior along the lines defined by the secure base concept. It also requires that the link between sensitivity, cooperation, etc. and coherent secure base behavior, demonstrated by Ainsworth, apply in all cultures. Posada (SRCD Monograph, 1985); Developmental Psychology, 1999, 2002) has provided strong positive support on both points.

Bowlby's theory also requires that individuals in every culture construct mental representations of their secure base experiences. Ana Zevallos and Harriet Waters are using the prompt-word secure base script assessment to test this hypothesis. The method is easily adapted for cross-cultural use and easily scored without knowledge of the culture involved. So far they have collected data from, Peru, Zimbabwe, Switzerland, The United Arab Emirates, Russia, and Turkey. In every case, it is quite clear that adults organize their early secure base experience in terms of the secure base script mentioned above. This is a useful support for an important prediction about the cross-cultural generality of Bowlby's attachment theory.

.

The secure bas ephenomenon is one of the cornerstones of attachment theory. A great deal of research has examined links between maternal care and the ability in infancy and childhood to explore and master the physical environment. The present research extends this work by examining the caregiver's role in support of explorations into the realm of fantasy. Children and mothrs are observed playing with a variety of toys that lend themselves to pretend play. We have developed a scoring system that evaluates both the child's interest and enthusiasm in delving into pretend play and the mother's support for such explorations. The term play partnership refers to dyadic interaction in which the mother is facilitative but not intrusive and the child is curious and engaged and understands the conventions of pretend play. Although pretend play depends on the maturation of certain cognitive skills, it is also substantially a product of parent-child interaction. The lessons underpinning fluent and extensive pretend play include (a) knowledge of reciprocal roles and turn taking, (b) understanding that you can pretend behavior that you wouldn't really do, (c) the expectation that you can select and develop your own lines of play without intrusions or interference, (d) trust that mother will accept pretended behavior as playful and beyond sanction, and (e) confidence that mother will be available and supportive if the play takes an uncertain or frightening turn. We are currently looking at the links between maternal attachment representations and the quality of the mother-child play partnership. We hope this work will cast light on an important new facet of the secure base phenomenon and also help us understand individual differences in mothers' ability to serve effectivley as a co-therapist.

In addition to participating in a number of the studies above, Dr. Waters has recently worked with a number of Consortium members to prepare commentaries on current issues in attachment assessment. One such commentary (Attachment and Human Development, September 2002) follows up discussions from the Consortium's Summer 2002 Adult Attachment conference. Issues discussed include (a) whether a developmental perspective is integral to adult attachment theory in the Bowlby tradition, (b) whether attachment theory should be focused on secure base related phenomena or strive to be a general theory of personality, relationships, emotion, and psychopathology, (c) lack of coherence across different modes of assessment, (d) trait and relationship conceptualizations of attachment individual differences, and (e) differing views of attachment and emotion.

A second commentary (Developmental Psychology, 2003 in press), discussed the theoretical and empirical underpinnings for the "patterns of attachment" concept. As experimental psychologists have demonstrated, humans are prone to see discrete types and patterns where there are none and to simplify experience for storage by reducing complex arrays to a few key features. Accordingly, classification schemes should always be suspect. One of the most useful lines of argument in support of a proposed taxonomy is a detailed discussion of mechanisms that could produce such structure. Attachment theorists have rarely considered whether there are mechanisms that could plausibly explain discrete patterns of attachment. In any event, nothing in the logic of Bowlby's attachment theory predicts or requires that individual differences in attachment security organize into distinct categories. Both dimensional and categorical assessments have their uses. The choice in any particular context has more to do with psychometric and design issues than with attachment theory.

Another aspect of my research focuses on the development of child-parent relationships after infancy. I am currently investigating the role of secure base support in maintaining the organization of secure-base behavior in early childhood. Also, I am studying the relations between child secure base behavior and the organization of information about secure base relationships in both children and their mothers from a cross-cultural perspective. Finally, another aspect of my research focuses on methodological issues in the study of child-mother attachment relationships. I am interested in the development of tools and strategies to facilitate observations of child-mother interactions in naturalistic contexts (i.e., home, parks, and hospitals).

Following attachment theory, one aspect of the knowledge structure about close relationships includes knowledge about a partner’s likely future supportiveness or responsiveness. A knowledgeable person should recognize cues indicating that the partner will be there for support at times of distress. Therefore, we developed a measure that could be used to assess this aspect of a person’s knowledge. First, a large number of possible indicators were obtained from a sample of respondents. These indicators were then evaluated by another group. The group consensus was used to evaluate the quality of each indicator. Analyses strongly suggested that this knowledge has a prototype structure. If this is the case, the prototype structure should affect information processing in this domain. To test this hypothesis, a false recognition study is being conducted.

The cues that were rated highest corresponded well to those predicted by attachment theory. This hypothesis was also formally tested asking participants to rate how similar each cue is to the caregiving styles proposed to predict secure attachment of a baby (responsive caregiving that consists of sensitivity, acceptance, respect for autonomy, and cooperation; Ainsworth, Blehar, Waters, & Wall, 1978). Participants were also asked to rate how desirable each cue is in a relationship. The ratings for similarity to responsive caregiving and informativeness of trustworthiness produced very similar clusters (cluster analysis) and a very similar factor structure (factor analysis). However, the cluster and factor analyses produced a completely different picture for ratings of “desirability in a relationship”, suggesting that our findings are not solely due to how positive and desirable each item is viewed.

Using the items obtained, a new measure was developed assessing how well each participant discriminates good from poor cues. To validate this “cues questionnaire a lab experiment was used, where participants listened to a confederate talk about a problem. During this interaction, the confederate mentioned indicators suggesting that the confederate’s roommate’s boyfriend is not trustworthy. Participants who discriminated well on the “cues questionnaire” understood significantly better that the boyfriend is not trustworthy.
If knowledge about cues of trustworthiness is part of a larger knowledge structure about trust in attachment relationships, then participants scoring high on the cues questionnaire should also be more familiar with other aspects of this knowledge structure. Therefore, a person with a highly available knowledge structure should know the secure base script well and should be able to apply it to diverse situations, including abstract situations involving non-human interactants. Therefore, a laboratory task was devised to assess the ability to recognize the secure base-script, even in stories involving cellular organisms and atoms. We found that performance on this task was related to the ability to discriminate good and poor cues of trustworthiness. We are now conducting a new study to test the same hypothesis with animations depicting geometric figures acting out the secure base script. To summarize, this research suggests that individual differences in knowledge about trust seem to have reliable and meaningful correlates and consequences.