Culture and Development Article Critique

O N THE SHAPING OF A n A C H M E N T THEORY AND RESEARCH:

A N INTERVIEW WITH MARY D. S. AINSWORTH (FALL 1994)

Mary D. S. Ainsworth and Robert S. Marvin

I. O N THE ORIGINS OF THE CONCEPT OF THE SECURE BASE

Robert Marvin.—Your doctoral dissertation, written at the University of Toronto, focused on the security of adult relationships and was supervised by William Blatz—what was the core of Blatz's position, and was he influen- tial in shaping your own thinking about the concept of security?

Mary D. S. Ainsworth.—Very much so. Blatz focused on the warmth and comfort that a parent can give to the baby and on the importance of starting off by making the child feel secure. He believed that, given a background of feeling secure, the infant would interest himself in what goes on around

Editors' note.—This interview with Mary Ainsworth was conducted by Robert Marvin at our request. We are deeply grateful to Bob for making it possible to elicit the memories and thoughts of one who has been so fundamentally influential in shaping the direction of attachment theory and research; without Bob's readiness to collaborate, these comments would have remained unrecorded. To Mary Ainsworth, who acquiesced to our request with the generosity and grace that have always marked her behavior toward colleagues— and particularly toward her "extended family" of attachment researchers—our gratitude remains unbounded. The transcript of the conversation was cut and edited to fit the requirements of this written presentation; however, all Ainsworth's points are reflected faithfully in what follows, and we have tried to preserve as much of her conversational style as possible.

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

him and want to move out to explore his world. And, should such explora- tion get him into more than that infant was ready to cope with, it was crucially important that the parent be accessible—that the baby be able to retreat to his "secure base" for comfort and be, as it were, "recharged" before going off again on his own. I first encountered Blatz's thinking— which was brand new, quite original as far as I know, and certainly not like anything that Freud wrote—in a course that I took from him in my fourth undergraduate year, and it was his notion of this nice "back-and-forthing" of the baby, the idea of building a secure base from which one can explore, that struck me most and attracted me to his thinking.

RM.—Did your early exposure to Blatz's ideas influence your subse- quent receptiveness to Bowlby's ideas?

MDSA.—Yes, I saw the thinking of both as going in very similar direc- tions. Blatz and Bowlby had no opportunity to influence each other—they were each working quite independently on how early relationships get estab- lished and how this then affects the child's further development, and I think they both happened to be observing the same kind of phenomena and that this led to their theories being so compatible.

RM.—When was it that you first became aware of John Bowlby's think- ing, and when and how did you come to know him?

MDSA.—I wasn't aware of any of his thinking before 1 met him. That happened in 1950: I arrived in London in the fall of that year without a job (despite previous efforts to obtain one through correspondence) and looked up a friend, Edith Mercer, whom I got to know during the war when we were serving as "opposite numbers" in our respective armed services. She soon phoned me about a position that was being advertised in the London Times and that she thought might interest me. The ad sought someone who was an expert in child development and in projective techniques to work on a project investigating the effects of a young child's prolonged separa- tions from the mother on personality development. I applied, and I was promptly interviewed by John Bowlby for that position. [With a hearty laugh.] What can I say?... We liked each other! We both liked the same researchers and clinicians who were working on similar issues at the time, and we found we were thinking along the same lines. I got the job, and, as in fairy tales, "they lived happily ever after!" Over the 3 years that I worked there, I continued to be struck by how much Bowlby's and my thinking moved in the same directions.

RM.—What kind of work did this job involve? MDSA.—At first, John had me read all the existing literature on separa-

tion—I spent about year doing that and also becoming familiar with the two projects that he had going on at the time. One of these was a follow-up

NEW GROV/ING POINTS OF AHACHMENT

Study of school-aged children who, when they were quite young, had spent long periods separated from their parents in a TB sanatorium, and the other was Jimmy Robertson's study of a particular group of young tubercu- lar children. This was an intensive, clinical study that involved frequent visits to the homes—and it turned out to be just my sort of stuff!

1 was placed on Jimmy's project after that first year to use my talents in looking for individual differences; originally, John had wanted me to do Rorschachs on this group, but that turned out not to be practical. The materials with which I worked were social workers' handwritten reports of the home visits they had conducted, and I began to notice several patterns in the way in which these children's "personalities" were organized that seemed pertinent to the experiences they had had.

So, as you can see, the approach and methods that I went on to use in Uganda and in Baltimore really had their beginnings in my work with Blatz and then with John Bowlby and Jimmy Robertson . . . though, as I think about it, what is really more correct to say is that the interaction among John, Jimmy, and myself completely changed John's original research proj- ect and eventually led to Infancy in Uganda as well as all three volumes of Attachment and Loss.

RM.—You told me once that, when you went to Uganda, you were not as yet convinced of the validity of Bowlby's thinking, that it was your observations of the Ugandan infants and mothers that you found so compel- ling. First, was it Bowlby who encouraged you to do direct observation (perhaps as Leaky had encouraged Jane Goodall), or was your focus on naturalistic observation inspired by some other source?

MDSA.—Actually, it came from having familiarized myself so thor- oughly with the work of Jimmy Robertson. All of his data on mother-child interactions and relationships—which were very compelling in indexing the effects of prolonged early separations—came from visiting the children in their homes and separation environments and observing as well as listening to what transpired. The power of his observations in reflecting what was going on and communicating it effectively to others impressed me very much, so I took direct observation as my own model.

RM.—Was there any particular observation, or set of observations, that you made in your Ugandan work that became particularly influential in your subsequent thinking?

MDSA.—In seeing these babies on numerous occasions as they inter- acted with their mothers within the home, 1 think the thing that struck me most was how active babies are and how much it is they who take the initiative. They are not passive little things to whom you do things; in fact, in many ways they are the initiators of what happens to them. The picture that you

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

got in those days from the literature was one of a passive infant who merely reacted to whatever the environment did to him, and that was the notion with which I first arrived in Uganda.

Another common belief that I had learned from reading Freud, as well as various other writers, was that what underlies the baby's tie to his mother is the fact that it is she who feeds him—the infant's pleasure in being fed gives preeminence to the figure of the food provider and becomes the basis for forming an attachment bond. The idea that infants' attachments could develop for any other reason was almost unheard of. It took the writings of Bowlby—who had been influenced by the work of ethologists, particularly Lorenz's work on imprinting—to open up people's minds to the notion that other mechanisms might in fact be responsible. My observations in Uganda gave me firsthand evidence that what I had been taught could not be sup- ported—that feeding could not be properly conceived as being the "prime mover" of attachment bonds.

RM.—What particular observations were most fundamental in origi- nally shaping your distinction between secure and insecure patterns of at- tachment?

MDSA.—My observation protocols took special note of instances of cry- ing and of the conditions under which it occurred, and it became evident to me as I went on that both the amount of crying and the circumstances in which the crying took place provided a central behavioral distinction. Babies who would be identified as securely attached did relatively little cry- ing except, for instance, if they woke up and the mother wasn't there or if the circumstances were such that they had reason to believe that mother was going to leave them. The ones that seemed insecure in their attachment fussed a lot of the time; anything could set them off, and there wasn't this clear picture that it was the distance from the caregiver that was responsible for the crying.

RM.—These are patterns you could never have discerned had it not been for repeated observations?

MDSA.—Oh, you're absolutely right—you can't really learn much about the relationship between a baby and the mother if you only see them once. You might say here, "Well, how about the Strange Situation, which is designed to elicit artificially what it might take a long period of observa- tion in the home to notice?" Yes, you can indeed learn a lot from brief ob- servation under such conditions, but, if the assessment hadn't been built on the basis of a lot of expectations derived from observations in the nat- ural environment, there would have been no validity to the laboratory situation.

RM.—I understand that you were not much in touch with Bowlby during your stay in Uganda but that, when he visited you after your return

NEW GROWING POINTS OF AHACHMENT

to Baltimore and became aware of your views concerning the secure- base phenomenon, he described the similarities to his thinking that he saw in yours as "a pleasing and encouraging convergence of ideas." Is that so?

MDSA.—Yes, he did say it, I think in one of his talks. [With a chuckle.] I do remember tbat, at one point during my stay in Uganda, I became quite concerned that John's use of Lorenz's work to support arguments against the preeminence of feeding as a mechanism for forming attachments might come to hurt him—that using baby chicks' behavior to argue against an entrenched psychoanalytic position might ruin his reputation! I even wrote him a gently worded letter in this regard—but, as it turned out, I needn't have worried.

RM.—It seems as if the concept of secure base as you first conceived it in your Ugandan work has proved robust enough to serve virtually un- changed throughout your career. Is this correct, or have your ideas in fact changed in some significant way?

MDSA.—The core concept of using the mother as a secure base from which to explore the world has not changed; what has evolved are elabora- tions of how different patterns of relationships are built in the course of the child's particular experiences in seeking to use the mother as this secure base. At least as a starting point for observation and research, I have never seen any need to change what was already partly incorporated in the theo- ries of Blatz and Bowlby.

RM.—You once told me that, while doing the observational work in Uganda, you experienced a real paradigm shift. Can you say something about it here?

MDSA.—The transition I made from thinking in psychoanalytic terms to thinking in terms of ethology felt very much like what I later read Kuhn describe as a "paradigm shift." It was a stidden, total, and permanent change in perspective—I simply couldn't conceive of viewing what I observed in any other way. [With a chuckle.] In my many disagreements with behaviorists, I finally came to realize that, since they had never experienced this new way of looking at things, I really couldn't expect them to understand what I was talking about—and, although it made our quarrels useless, it did enable me to feel vastly superior!

II. O N UNIVERSAtlTY A N D CROSS-CUtTURAt DIFFERENCES

RM.—Turning to another topic, attachment theory conceives of the attachment control system as being part of our evolutionary endowment.

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

Does this mean we should expect to see infants organize their secure-base behavior in a similar fashion across all cultures, or should we expect to see this organization differ somewhat in different ecological settings?

MDSA.—Both. At the most fundamental level, there is a biological basis that determines the emergence and general shape of the behavioral organization—I think that environmental influences play no significant role in the infant's basic need for an attachment figure who can be trusted. But culture-related differences in ecologies and expectations will certainly affect how some specific aspects of that organization are expressed under particular conditions—I am thinking here, for instance, of proximity seeking.

RM.—As an example of what you are saying, are you thinking of the work that the Levines and I did with the Hausa, where living huts contained open fire pits as well as potentially dangerous tools and animals and babies did not explore under their own locomotion but were always carried around and handed from person to person—and yet we found that, despite this constant physical contact with numerous people, a distressed baby would want contact with an attachment figure, and a very distressed one would want only his or her primary attachment figure?

MDSA.—Yes, that's a very good example of part of what I mean. But I'm also thinking here of the atypically high incidence of insecure attach- ments that characterized Karin Grossmann's sample drawn in north (as opposed to south) Germany. Here, after behaving in ways resembling the typical interactions of securely attached mother-child pairs, cultural mores seemed to dictate that, some 3 months earlier than in other samples, the mother say, "This is enough—it's time that my child learn to be indepen- dent, to look after himself, and not to need me anymore." Even if done gently, this is very rough on a baby who has learned the definite expectation that "of course mother will pick me up! I've given her all the signals, she knows that I want her . . . and, hey, she's not doing it! and deliberately not doing it, by God!"

Some of the mothers evidently behaved in ways that fostered insecurity from the outset, but I do believe that this sudden and (from the baby's point of view) totally unexplainable shift in maternal behavior acted to make some previously secure attachments change into insecure ones. I do think that the two patterns of insecurity—those of babies who had an earlier "security- fostering" relationship and those of babies who did not—are likely to differ, but I am not sure exactly how, and I certainly think it's something very worthwhile pursuing.

My intuition is that mothers who come to reject their child's proximity bids as a matter of principle—"because this is the way to do things, it's

NEW GROV/ING POINTS OF AHACHMENT

the way my mother taught me"—may not be as consistent in rejecting the baby's requests and desires. That surely must have some effect, but I don't have a sense in exactly what direction. In any event, I have always believed that it was essential that we observe infants in more than one culture—that's what took me to Uganda. There are probably far, far more similarities across cultures during this early period than has ever been acknowledged.

RM.—Extending this line of thinking, let me ask you about work on species other than humans. First, how influential do you think this kind of work was on Bowlby's thinking, and, second, how much has it influenced your own formulations?

MDSA.—1 think that Lorenz's work on imprinting was very important to John Bowlby in pointing out the direction that his thinking would take. And Harlow's studies, as well as Robert Hinde's work, were very influential in extending his perception of the similarities between humans' and other species' behavior—in a sense, he saw it as verification of the validity of his evolutionary perspective on human development. As for me . . . many of my best friends have been primatologists and ethologists, and I have always been fascinated by their work. In fact, I have been reading a lot about wolves lately, and learning about their complex social organization certainly enriches one's life as well as enables one to empathize with their behavior. Mainly, however, findings obtained with other species have made me feel that I have been on the right track rather than helping me understand the specifics of human babies' behavior.

RM.—It seems that, in whatever form, cross-fertilization between at- tachment researchers and those working on nonhuman species has fallen off in recent years. Do you think this is indeed true, and, if so, why did it come about?

MDSA.—I do think it's true, but I'm not sure why it happened. And I do think it's too bad—contact between these two very different branches of science was good for the young scientists on both sides. Perhaps it has been partly because most, if not all, attachment research has focused on procedures aimed at distinguishing between secure and insecure patterns as well as among different patterns of both security and insecurity—and I don't think that most primatologists are interested in this kind of thing. I think they are far more concerned with getting more and better field data and in proposing cause-and-effect relations in an observational way. Re- search on human mothers and babies has extended in so many different directions: the age at which the attachment relationship—or its representa- tion—is assessed, the figures in the relationship, etc. I think that this has moved attachment researchers much further than primatologists in study-

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

ing individual differences . . . but, then, I think that's in the nature of the beast.

RM.—What do you mean? MDSA.—Well, it's obviously so much more difficult to observe the

development of interactional patterns between, say, a mother bear and her cub than it is to follow a human mother and infant. And when 1 think of what it takes to study porpoises and their offspring! . . . In any event, I think that interest in fieldwork—which is one of the things that had connected the two fields—has been largely lost among human attach- ment researchers. And this is something that I do regret—the fact that our emphasis has increasingly focused on measurement rather than assessment.

III. O N CONSTRUCTING ASSESSMENTS A N D RESEARCH DESIGNS

RM.—What is this distinction you make between measurement and as- sessment?

MDSA.—To me, measurement implies assigning numbers that reflect a precise amount or an equal interval ordering. In my own work 1 have focused on classifications of patterns or on matching to behavioral descrip- tions—and that gives you assessments.

RM.—Let me ask some questions about how your systems of assessment came into being. First, I have heard it said that the interactive scales we use to rate Strange Situation observations were developed by cutting up actual transcripts and ranking the relevant pieces—is that true?

MDSA.—My first reaction is to say, "Preposterous!" But in fact I do dimly recall something of this sort. Remember that at the time we had no technologies that could allow us to move information around, so shuffling about printed excerpts of specific behavioral observations was a very effec- tive way to get good definitions of scale points. In fact, it's very similar to what one still does when using a Q-sort deck!

RM.—How about the scales of maternal behavior? How did these de- velop?

MDSA.—Our narrative transcripts of the home visits were very lengthy, and they didn't focus on detailed descriptions of specific sequences of behav- ior, so the "chop them up" approach would not have helped. No, here I started as I had in Uganda, by asking, "What is there about the behavior of this mother that is important in making a difference in how the baby behaves?" And that led to the scale of maternal sensitivity, to differences in

10

NEW GROWING POINTS OF AHACHMENT

how observant and responsive the mother is in dealing with the baby. In some cases, I started with the negative pole of what became a scale because the effect of negative qualities on the interaction often can be seen more clearly. For instance, you see the mother sort of buffet the baby around, interfere with what he is doing—and it's very clear that the baby doesn't like it. So I felt this was another dimension we should explore, and, after looking at the whole range of observations we had made, I decided that cooperation was the most appropriate opposite end on this particular contin- uum. At one point I struggled hard to devise a scale of maternal warmth, but that never worked out.

RM.—Why didn't it, do you think? MDSA.—Various reasons. For one, it is very difficult to be precise

in defining behaviors that could be criterial in rating maternal warmth. Russel Tracy tried to explore this in our narratives by doing a very detailed analysis of all affectionate behaviors, and it became clear that something as obviously a candidate for "warmth" as kissing could in fact be a very unaffectionate peck on the cheek or forehead. In a similar way, the Ugandan mothers' habit of sitting the baby in their laps facing out rather than en face can look like lack of warmth, but in fact it has to do with their concern to create opportunities for exploration for the child, and the physical contact between the bodies does create an af- fectional warmth—think of how many 3-year-olds like to sit on a parent's lap!

Another reason I gave up is that maternal warmth did not seem to make much difference in itself—two mothers can be equally warm and yet have very different effects on their babies. There is a great difference be- tween maternal warmth and maternal sensitivity, and it took me a long time to appreciate it.

RM.—Is it perhaps that sensitivity is a response to the baby's initiative and warmth is more of a general characteristic of the mother?

MDSA.—Absolutely—and this focus on how the attachment figure supports or interferes with what the baby initiates has served me well all along.

RM.—So, in characterizing maternal sensitivity, we should make sure not to forget that it is sensitivity to the baby's initiative that is at issue. In this context, what do you think enabled you to find such robust rela- tions between maternal sensitivity and infant security in your Baltimore study?

MDSA.—The lengthy and repeated home observations on which we based our maternal ratings certainly had a lot to do with it. But 1 think it was also due to the fact that such a wide range of secure and insecure

11

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

patterns was represented in the sample—quite a few studies report an absence of C classifications in their groups. It could have been luck, or it could have been that the pediatricians who recruited potential partici- pants for us tended to select women who interested them—"this one is a charmer, that one puzzles me, I wonder how motherhood will work out for this one"—and that this led to our getting a particularly diverse group.

RM.—Moving on to your classificatory system, why is it that you built the system for classification of attachment patterns around behavior ob- served in the laboratory rather than in the home?

MDSA.—Although the classifications are made from behavior in the Strange Situation, the rationale for assigning particular meanings to these specific behaviors and their patterning was developed from the significance of the different interactional patterns that we had observed in the home—so, in the most fundamental sense, the classification is based on what we learned in the home. In fact, I have been quite disappointed that so many attachment researchers have gone on to do research with the Strange Situation rather than looking at what happens in the home or in other natural settings—like I said before, it marks a turning away from "field- work," and I don't think it's wise.

RM.—What is it, do you think, that has led to this focus on Strange Situation data rather than on home observation?

MDSA.—Partly it's the problem of getting funding—doing observa- tional work takes an enormous amount of time, especially when it involves repeated visits to the home. Pardy it has to do with the responsibilities one feels to one's students—getting them involved in research that can be completed within the time that they have available to them. And in part it has to do with the "publish or perish" realities of academic life!

RM.—Can you say a bit more about just how you went about con- structing the Strange Situation and the classifications?

MDSA.—I had seen a lot of separations and reunions in the homes, a lot of exploration, a lot of proximity seeking, and a lot of differences in how the baby and the mother behaved in these situations. So con- structing the episodes of the Strange Situation wasn't hard at all; as I recall, it took just about half an hour of talking with Barbara Wittig to decide on the episodes and their sequence—it just came naturally. I was very interested to see how the baby would handle these various situations that he encountered in the lab, and the thing that made me feel that I was really on the right track was that I turned out to be so good at anticipating what each specific baby would do—that it all was so predict- able, it hung together.

12

NEW GROWING POINTS OF ATTACHMENT

RM.—Was it the separations and reunions that you saw at home and in the lab that hung so well together?

MDSA.—They did—but it was also a matter of how the mother behaved at home in many other situations; how she responded to cries, to other bids, to feeding times.

RM.—Do you think that you drew a lot on the observations you made at the end of the year, when the babies could crawl and you saw them move away and back to the mother and using her as a secure base, just as you had seen it in Uganda?

MDSA.—Yes, these were certainly important. They also made me real- ize that you have to be very careful to think about the child's age, his devel- opmental status—we had some 12-month-olds who couldn't creep, so they might just sit there, cry, and hold out their hands.

RM.—Might such underdeveloped locomotion shift a baby's classifica- tion, perhaps even in some major way?

MDSA.—Oh, I think that's a real concern—there has been so much variation in the ages of different samples that sometimes they are not babies anymore. At this point, all you can do is extrapolate; you go by principles you have devised or acquired without being quite sure how, but you don't follow the actual rules of the system. I've done a lot of extrapolating myself in my life, and it doesn't trouble me too much—but then, do other people do it in the same way as I? Of course, never for a minute do I think that / could be wrong! [Laughter.]

RM.—Turning to another topic, that of research strategies, why do you think there has been so little work examining patterns of maternal behavior and infant secure-base behavior as they co-occur in actual inter- action? Most research seems to focus on prediction from one of these two to the other.

MDSA.—I think this is partly because many researchers think that ask- ing, "When mother does this, what does baby do?" gets you nothing but normative description and that that's not very exciting. It's wrong, of course: all the different patterns of attachment behaviors that we established in the Baltimore study, and then the concept that these patterns represent the baby's "strategies"—all these proved to be theoretically very exciting, and all came about from looking at, "When baby does this, his mother does that," over and over again. But it takes a lot of time before this kind of work comes together, and it's surely easier to publish a lot of papers if you just look for "significant" correlations! Of course, we did also use the data for some "predictive" analyses—like Silvia Bell's work on how maternal re- sponse to crying in an earlier quarter of the baby's first year related to the baby's crying in a later quarter—and this also was very useful in supporting

13

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

the ideas we were developing about how different patterns of attachment emerge.

RM.—So are you saying that some attachment research—not all, of course, but some significant portion—has shifted from a focus on descrip- tion and understanding toward simple prediction?

MDSA.—I am sorry to say it, but I do think this has been the case.

IV. O N INDIVIDUAL DIFFERENCES A N D AAATERNAL SUPPORT

RM.—The idea that attachment has evolutionary underpinnings sug- gests that there must be some limits on how it is organized behaviorally. Suppose that someone observed some pattern of Strange Situation behavior that departs radically from that of the secure, B babies and also demon- strated that, in terms of all other significant data, babies classified in this "new" pattern behave just as those who are currently classified as secure. Would that pose a problem for attachment theory?

MDSA.—Yes, I'd say it would. But finding that everything in the mother-child interaction indicates security except for the child's behavior in the Strange Situation seems so very, very unlikely given all the masses of data we have accumulated that the question strikes me as being somewhat nonsensical.

RM.—What if someone discovered a new pattern of insecure behavior? MDSA.—That would not be all that surprising; there are many more

possible ways of organizing behavioral strategies for coping with an insecure relationship.

RM.—It seems that infants become attached to their primary caregiver almost regardless of the care they receive, as long as the relationship doesn't involve multiple major separations or grave threats to the child's health. Does this surprise you?

MDSA.—Not at all. It's the presence of the caregiving figure rather than the caregiving behavior that is essential for the attachment to develop. It's a bit like Lorenz's work with ducklings—it didn't really matter whether the mother duck fussed with the little creatures or just did her own thing; if she would start swimming away, the ducklings would follow. It was her presence that was the essential thing, and I think this phenomenon of im- printing was really important to Bowlby's original formulation of human attachment.

RM.—So does this imply that the development of human attachment is a fairly close analogue to imprinting?

14

NEW GROWING POINTS OF AHACHMENT

MDSA.—To a certain extent, yes. It differs in the human case in that it needs more time to become effective—it's not that the newborn comes out of the womb, senses the mother, and hah! that's it! It's something that stretches out over a much longer period of time.

RM.—Do you think that there may be some particular period in the first year of life—maybe one in which a lot of different developmental milestones in locomotion and communication are being achieved—that may be particularly important in the consolidation of attachment?

MDSA.—I have a sense that the consolidation is more likely to be more evenly spread out, but it may well be that, if we did some intensive research on this extended early period, we would find evidence that some particular dme is particularly crucial. At this point we just don't know.

RM.—It seems nearly paradoxical that, even though you focused so much of your work on maternal sensitivity, you are saying that it is not the mother's sensitivity but just her presence that determines whether the baby becomes attached.

MDSA.—It is not attachment but the security of attachment that is af- fected by the mother's sensitivity.

RM.—I wonder whether this is widely understood—whether some peo- ple may not think that your work had always implied that babies become attached to the person who is the most sensitive to them and that sensitivity is more important than presence.

MDSA.—If two potential attachment figures are concurrendy involved, I can't help but think that their relative sensitivity to the baby might well affect the growth of the child's attachment—that it would indeed be the more sensitive of the two who first starts the attachment process going. But I'm not sure that this would necessarily be the case.

RM.—Given what you've been saying, what do you think of what is so often said to parents—that it's the quality of the time you spend with your child, not the quantity, that is important?

MDSA.—There are two issues here. One has to do with the nature of the relationship, and here you would expect that devoting "high-quality time" to the child would tend to evolve into a secure relationship. The mere continued presence of the adult figure has little, if anything, to do with the quality of the relationship—only with the emergence of some form of bond.

RM.—In a related vein, if a child is placed in day care for the full day and from a very early age (say 6 weeks) and is with the mother only in the evenings and over the weekends, do you think that baby's mother is still likely to be his primary attachment figure?

MDSA.—Yes. Group-care situations nearly always include multiple

15

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

caregivers, and it is much more difficult to form an attachment with one individual out of the several who look after the baby and whose presence is likely to vary day by day. But, if there should be a long-continuing, unin- terrupted relationship with some caregiver other than the mother, then, yes indeed, I think you could see the same kind of relationship emerge with that figure as you ordinarily see with the mother. The crucial point is that the relationship be continuing and pretty much uninterrupted—under such conditions, the nonmaternal caregiver can well become an attachment figure closely resembling that which usually evolves in the relationship with the mother.

RM.—In the early stages of your work—like when you first went to Uganda—were you primarily interested in normative questions as opposed to questions of individual differences?

MDSA.—I think my primary interests were normative. I was very interested in what "popped in" in terms of behavior, what new acquisi- tions emerged at what ages, and how well the normative patterns that I saw replicated what had been observed in the United States and England—I was very sure that normative similarities would be there. But that doesn't mean that I didn't expect to see individual differences as well. . . . I wasn't surprised by them, and I was very curious to see what circumstances made for these differences, for the departures from normative patterns.

RM.—The first reports of Strange Situation behavior that emerged from the Baltimore study focused on the "average" baby's typical responses to a new environment and to brief separations. What made you then shift the focus to individual differences—was it differences among babies or among mothers that caught your eye?

MDSA.—As I said earlier, the home visits led me to have very def- inite expectations about how individual babies would respond to the Strange Situation, so obviously differences in babies' behavior did cai;ch my eye. But these expectations were built not only on the baby's behavior but also on the mother's—it was the qualities of the many interactions that I had seen between the two that were behind my perceptions of differ- ences.

RM.—So, as it has at various points in our conversation, it appears that it was the repeated home visits, the gathering of a lot of information over a number of months of observation—that this was the key to under- standing what was going on in both your Uganda and in your Baltimore samples.

MDSA.—Indeed it was. RM.—To move on to debates concerning the determinants of individ-

16

NEW GROWING POINTS OF AnACHMENT

ual differences: arguments concerning effects of maternal sensitivity versus those of temperament have occupied the field for a long time. Do you think there is something particular about these two constructs that has made them so difficult to work with for so long?

MDSA.—I think that what has made them difficult is that we have tried to force them into an either/or dichotomy, whereas I believe that in fact there are effects from both. I happened to have focused on maternal sensi- tivity, on how babies react to the adult's responsiveness, because to assume that how the baby is treated, that how sensitive or insensitive the mother is in responding to her child, has no effect—that's impossible, especially if you have seen a lot of different interactions. But that's not to say that temperamental differences have no influence on behavior. I think they do, but, since I also think that the kinds of differences that emerge as a result of mother-child interaction have more to do with shaping the nature of the child's later relationships than the child's temperamental qualities, 1 have given them more emphasis.

RM.—It seems that, when Bowlby and you first started talking about attachment theory, you were perceived as being radically biologically deter- ministic; that, of course, was in the context of the sway that social learning theory held at that time. Now you are being seen by some as placing too much emphasis on environmental determinants by focusing so much on maternal sensitivity. My own impression is that you have in fact stood stock still over the years and that it's the field that has been swinging like a pendu- lum around you. [A hearty laugh from Mary.] But to return to the notion of temperament—what do you perceive it to be?

MDSA.—Well, I remember reading Sheldon's book on the varieties of temperament many years ago and finding it very interesting. I did know some people who seemed to absolutely personify some of Sheldon's types. For instance, I had a professor once who was tall and very skinny, had a very soft voice, made practically no noise in moving about, and was highly intellectual—he seemed to have all the qualities that Sheldon ascribed to an ectomorph and that he thought were genetically determined. And it may well be that such aspects of personality may have a significant effect on a person's behavior in many circumstances, but I don't think that they have any truly basic influence on the kind of relationships that the person will construct with other people.

RM.—So you think that, although temperamental differences can have a big impact on how a baby behaves, they are not the major determinant of whether the child becomes securely or insecurely attached?

MDSA.—Yes, I think it is how the mother responds to the particular kind of behavior shown by her baby that plays by far the major role.

17

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

V. O N CONTINUITIES INTO LATER YEARS

RM.—The notion that early relationships serve as prototypes for later ones is viewed by some as being the key concept in attachment theory. What are your thoughts on this matter?

MDSA.—I don't know that I would call it the key concept because I see a number of the ideas that I talked about at the beginning of our conversa- tion as also being "key" in defining what attachment theory is all about. I'm also not sure that I like the term prototype. It can be taken to imply that we expect a persistence of the same behavioral patterns, and that's clearly nonsense: a securely attached baby will cry wben mother leaves him alone in the Strange Situation, but we don't necessarily expect him to do so when he is a preschooler, let alone a teenager!

RM.—But what if we take it in another sense—as perhaps setting a pathway for later relationships or predicting what they are likely to be?

MDSA.—In that sense, yes. But I think we should limit it to intimate relationships. We do have some evidence that securely attached youngsters tend to interact more easily with other children and with adults than in- securely attached children, but such differences in social behavior that doesn't involve close ties can come about for a variety of reasons. The dif- ferences may well be related to differences in how sensitive a mother is in introducing her baby to unfamiliar people or to how secure she makes the child feel in exploring his environment, but I don't think that they are in any way a necessary consequence of differences in the mother-child attach- ment bond. Work on continuities in the structure of attachment bonds, on figuring out the "rules" by which transformations to qualities of later relationships come about, should focus on relationships that are close— those in which the other can be assumed to serve as some form of an attachment figure.

RM.—How about within the period of infancy? Suppose a baby is attached to one figure, and another caregiver is then added to his life. Do you think that the quality of the second relationship would be much influ- enced by the experience of the original caregiver?

MDSA.—I think it would certainly be influenced at its beginning be- cause the child's expectations of a caregiver will have been shaped by his interactions with the first figure. But, if the new relationship remains con- tinuous over a long period of time, I don't think this would remain a lasting effect on the nature of the attachment bond formed with the new figure.

RM.—So you think that, to the extent that the child's interactions with the new figure are different, the baby can develop a working model of

18

NEW GROWING POINTS OF AHACHMENT

the attachment relationship that differs from the one he had constructed before?

MDSA.—I keep thinking here of some Adult Attachment Interview [AAI] transcripts where it becomes clear that the child's relationship with the parents was essentially insecure but that a secure relationship with some other important figure—a grandparent, a teacher, a coach, someone who was close and understanding—had functioned to make up for the life of insecurity with the parents. It's hard to point to such instances within in- fancy itself, but having adults look back on their experiences makes you realize it is possible.

RM.—It sounds like you are saying that two processes—actual interac- tions and working models—are involved here: the baby initially comes to interaction with the new figure with the same attachment-related expecta- tions that he had developed with the first figure, but, to the extent that the second relationship develops along different lines, the earlier working model will gradually change to become something perhaps quite radically different?

MDSA.—Yes, that's very much what I am implying. RM.—Thinking about work that has been relevant to this issue, do you

think that the maternal deprivation literature can be taken to demonstrate the existence of at least a predictive relation between early attachment and later relationships?

MDSA.—One of the things it certainly demonstrates is that the attach- ment bond is not simply Cod given but—as I have said before—that it has to have an opportunity to develop. In Jimmy Robertson's observations of tubercular cbildren who had to spend many months in a sanatorium, the indifference with which the child treated the parent when the mother or father came to visit was remarkable. It was as if the child simply didn't care; it wasn't at all important. And I think it is fair to say that this is by far the most likely way these children would later treat other people—that what is set under such conditions is indeed a prototype for all subsequent affective relationships.

RM.—One of the most important findings in recent attachment re- search has been Mary Main's demonstration that classifications obtained by mothers on the Adult Attachment Interview are strongly related to their infants' Strange Situation classifications. What do you think has made this finding so influential?

MDSA.—Because it establishes a significant step in what is needed to show that there is a meaningful developmental link between the attachment- related experiences of the child and how that child behaves in attachment relationships when he grows up to be an adult.

19

WATERS, VAUGHN, POSADA, AND KONDO-IKEMURA, EDS.

RM.—Do you think that studies comparing infants' early Strange Situa- tion classifications with their own subsequent AAI classifications will support this even further by showing a similar predictability?

MDSA.—I do. And, taken all together, it will be eloquent evidence of meaningful developmental links—that just as attachment theory postulates, development follows a logical course and is not just a matter of happen- stance.

RM.—Let me push this even further by asking a hypothetical question: What if these data had started by showing the opposite, that is, a lack of continuities?

MDSA.—I think that, as developmentalists, it is safer for us to cling . . . yes, I'll stick by that word . . . cling to the belief that the processes of development are patterned. And, even though we have an incomplete un- derstanding of what rules underlie these patternings—what processes they involve and how these processes involve behavior—it is more sensible to believe that such rules exist than to believe that they don't. So I would look very carefully at such contradictory data, try to see what might account for the anomaly, and then try to get the study replicated using an intelligent and impeccable methodology. And, since we're in a hypothetical mode any- way, I'll bet you anything that the results would turn out to support presence of logical continuities!

RM.—Have you ever been surprised by the nature of any of the conti- nuities that you saw in your Baltimore data or that have been reported by Main?

MDSA.—I had anticipated most of what turned out from the Baltimore study, with one exception. What came as a real surprise was to see that mothers who ignored their infants had babies who we thought were clearly ignoring their mothers. This finding has now also been supported by Mary Main's finding that infants of mothers classified on the AAI as dismissing of attachment are classified as avoidant in the Strange Situation. But, origi- nally, I had expected that babies whom mother ignores would demand her attention more and more in trying to get her involved and wind up being fussy babies. I expected to see something more openly grieving, and maybe these babies did go briefly through some such phase. But in the end they wound up becoming avoidant.

RM.—So they ended up looking like the deprived child who is beyond the phase of protest?

MDSA.—Yes. Although not as dramatically as the children described in the literature, these babies were deprived: the mother did live in the same house, but the baby was severed from her by the closed nursery door and by the mother's indifference to sounds that emanated from behind that door.

20

NEW GROWING POINTS OF AHACHMENT

RM.—As researchers have moved to study attachment beyond infancy, there seems to have been a tendency to focus more and more on cognition— working models, thoughts, internal events—and less on observable interac- tions between the individual and his loved ones. How do you feel about this trend?

MDSA.—As I think I've already stressed at various points in our conver- sation, I believe that we must continue to rely on direct observation—on what I've called "fieldwork"—in gathering more information on what actu- ally happens in different contexts and in different relationships. But moving to also study internal attachment-related events is certainly appropriate and necessary to get at the "rules" and processes 1 mentioned earlier. As long as this trend doesn't become an "either/or" thing—either you observe, or you probe for internal events—but remains rather a matter of balance be- tween the two, I think it will serve us well.

RM.—Well, we've ranged over many topics in this conversation, and now the time has come to bring it to a close. Is there anything—maybe something personal—that you may wish to add?

MDSA.—Perhaps the most extraordinary thing in my life has been seeing so many people become interested in the concept of attachment and dedicating themselves to developing it further. They've done so much to contribute to its growth. . . . I think of them all as my "ex- tended family," and I send them my love, my thanks, and my very best wishes.

21