Writings > Development of the Mind and Brain > Part II > Chapter 10

Chapter 10

Testing CMS Theory

Because we thought our typology subdivided normal populations as well as abnormal ones, we tested the theory in normal — that is, nonclinical — populations. We were fortunate to gain access to data from two longitudinal studies. The data that we used for our testing were from the archives of the Oakland Growth Study (OGS) and the Family Socialization Project (FSP),¹ two longitudinal studies done at the Institute of Human Development at the University of California Berkeley. These studies tracked 100 to 200 families over long periods of time. For a description of the OGS and FSP subjects and the kinds of data that we used, see Appendix B; Gibson, Malerstein, Ahern, and Jones (1989); Malerstein, Ahern, Pulos, and Arasteh (1995); and Malerstein, Ahern, and Pulos (2001).

Relative Frequency of CMS Types

The predominant CMS type in our two samples was Symbolic — 54 of the 78 OGS participants and 49 of the FSP mothers fell into this category.² We think that Symbolics and Intuitives are more common than Operationals. In our samples, Intuitives were probably underrepresented. Only 4 of the OGS participants and only 2 of the FSP mothers were Intuitives. We think that Intuitive subjects and Intuitive parents would not be particularly interested in participating in longitudinal studies.

Reliability

The first question that we wanted to answer was, How reliable are we in our recognition of the three CMS types and the Symbolic subtypes? That is, Can we independently agree on the CMS types of the individuals in the populations at percentages that are better than chance?³

Our answer was yes. We independently recognized CMS types in adolescents and in adults who were members of the two populations. In OGS participants — 78 men and women in their late 40s — Ahern’s and my independent assessments of CMS types agreed in 76 % of cases. In FSP subjects — 68 mothers and adolescents — we agreed in 71% and 55% of cases, respectively. In 30 of these mothers, research assistants who had limited training in our way of classifying people agreed with Ahern’s and my assessments in 77 % of cases. CMS subtypes were also independently recognizable. See Gibson et al., (1989) and Malerstein et al., (1995) for a full description of our testing methods.

Prediction

Our second question was, Could we predict CMS types from the kinds of caregiving settings that children experienced when they were in the Concrete Operational Period? The FSP had interviews of the children’s mothers conducted when the children were 9 (Mean age was 8 ½ in our sample.) and interviews of the children themselves conducted when they were 14 (Mean age was 15 in our sample). Presumably, 9-year-old children were in the Concrete Operational Period when their mothers were interviewed.⁴ From our assessment of the caregiving setting that was revealed in each mother’s interviews conducted, we predicted what that child’s CMS type would be when he was 15. Blind to which mother was the parent of which child, we categorized the CMS type of each child from his own interviews when he was 15. We could then see how well we had predicted the different CMS types of the children from the earlier caregiving settings.

I wish to emphasize that, although all data was archival, our predictions were true predictions. That is, they were prospective. We did not just find a correlation between two assessments.

Our predictions were successful. We predicted adolescents’ CMS types from their earlier caregiving setting in 56 out of 68 cases. Statistically this prediction is highly significant.⁵

An unanticipated, but theory-compatible finding was that the mother’s CMS type, based on her interviews when her child was 9, tended to be the same as her child’s CMS type when that child was 15. We expect a Symbolic to rear a Symbolic, an Intuitive to rear an Intuitive, and an Operational to rear an Operational, though not invariably. An Intuitive mother, whose focus is on what looks good for the moment and on getting and having, will be busy serving her own needs, not those of her child. A child reared under these conditions is likely to become an Intuitive. Similar reasoning applies to Operational mothers and to Symbolic mothers.

To rule out the possibility that descriptions of the children’s behavior — part of the content of the mothers’ interviews — might have influenced our predictions, we repeated the predictions, using interviews with the mothers conducted when the children were 4, 9, and 15. Each interview was subdivided by research assistants into two part interviews — one that contained no references to child behavior and one that was composed solely of descriptions of child behavior.

From the part interviews of the mother that were conducted when her child was 9, and that contained no descriptions of the child’s behavior, we were still able to predict the CMS type of the child when he was 15. Based on interviews of the mother conducted when her child was 4 or 9, the mother’s CMS type tended to be the same as that of her child when he was 15. Apparently, the caregiving setting at 9 and the CMS type of the mother when a child is 4 or 9 is related to CMS type of that child when he 15.

Our assessment of the 15-year-old’s CMS type, based on his own interviews, tended to agree with our assessment of the 15-year-old’s CMS type, based on his mother’s part interviews that were composed solely of descriptions of his behavior at 15. However, our assessment of the 15-year-old’s CMS type, based on his own interviews, did not tend to agree with our assessment of his CMS type, based on the part interviews of his mother composed solely of descriptions of his behavior when he was 4 or 9. This would suggest that, for our purposes, either mother is a poor describer of her child when he is 4 or 9, but a good describer of him when he is 15. Or that CMS type is sufficiently established that it may be recognized when the child is 15, but is not sufficiently established to be recognized when he is 4 or 9.

The main point to be made from our empirical testing is that, although all of our findings are modest, they are consistent. They all support our position that caregiving during the Concrete Operational Period influences formation of CMS types.

Empirical testing of psychodynamically-oriented psychological theories is uncommon. One such theory — Attachment theory — has been subject to extensive testing. As will be evident, a primary conflict between CMS theory and Attachment theory is that Attachment theory, like most other psychodynamically-oriented theories of development, propose that personality traits — read character structure — form during infancy, not as late as the Concrete Operational Period. Another difference between CMS theory and most other developmental theories, including Attachment theory, is that CMS theory is not linear, with primitive and pathological at one end of a single continuum and sophisticated and normal or healthy at the other end of the continuum. Instead CMS theory proposes three such continua that extend from pathological to normal. Nonetheless, Attachment theory and CMS theory have much in common.

Attachment Theory and CMS Theory Take Different Bites of the Apple

Both are psychodynamic theories that have their roots in clinical observations of adults and adolescents. Bowlby theorized that, based on their experience with their primary caregivers, children develop a working model — expectations and beliefs about the self, the world, and relationships (Bowlby, 1973; Waters, Weinfield, & Hamilton, 2000). CMS theory that early caregiving influences formation of CMS type — characterized by a basic style of social cognition and a related motivational investment — is similar to Bowlby’s theory.

Attachment theory grew out of Bowlby’s (1973) idea that the helpless infant needs proximity to protective figures in order to survive, and out of Ainsworth’s (1982) studies of children’s responses to the Strange Situation . Bowlby had proposed that beginning at about 1 year of age, humans automatically track their lifelong access to a protective figure (Main, 1987). He reasoned that this tendency was a genetic adaptation that protected the infant from predators when humans first roamed the savanna. With Bowlby’s proposal in mind, Ainsworth constructed the Strange Situation. In the Strange Situation, a mother brings her 1-year-old into a room that is furnished with objects that would interest a child. A stranger enters. The mother leaves, returns, then leaves, and returns once more — all within 20 minutes.

Depending on the baby’ responses, Ainsworth classified the babies as A, B, or C. The majority of babies were B babies — babies whom she regarded as having secure attachments . They continued to explore while their mothers were gone. When their mothers returned, they went to their mothers and then explored some more. A and C babies were regarded as having insecure attachments . A babies ignored their mothers’ absence and shunned their mothers when they returned. C babies were increasingly distressed when their mothers left, and could not be consoled when they returned. Generally, the baby’s classification did not change at 18 months, unless his mother’s life had changed. Ainsworth found that the mothers of B babies accepted attachment, and were sensitive and responsive; mothers of A babies rejected attachment; and mothers of C babies were inconsistent in their attitudes toward attachment.

Main (1987) reported that, at age 6, B children were better adjusted emotionally and socially than A children, who were aggressive, hostile, and detached, and C children, who were anxious and were often subjected to bullying. At this same time — when the children were 6 — the mother of a B child was fluent and coherent in her descriptions of her own parents and of her early life. The mother of an A child had difficulty remembering her relationship to her parents, and tended to idealize that relationship. Specific memories, however, suggested that her parents were unavailable or were not interested in her. The mother of a C child was preoccupied with the past. Her stories about her parents lacked coherence, and sometimes the past intruded into the present as she spoke.

Main reexamined these children’s Strange-Situation responses when they were first tested. Among those babies’ whose responses were difficult to score, she found a fourth type — D babies. D babies neither simply avoided nor clung to the mothers. For example, when a mother returned, one child hid under a table. Another stopped in his tracks and appeared dazed. At age 6, children who had been D babies tended to mother or to boss their mothers; they reversed roles with the caregiver. The mothers of these children were frightened or frightening.

Main and her colleagues developed the Adult Attachment Interview (AAI), which differentiates adults in terms similar to those that differentiated the mothers of the 6-year-olds. The AAI is a structured interview that involves a complex rating system. The interview centers on the person’s relationship to his or her parents or substitute caregivers. Ratings are based on a believable fit between memories and evaluations involving a mother’s (or father’s) attachment to her (or his) parents. A mother rated as autonomous secure is one who provides an unelaborated but clear picture of the relationship to her parents that is orderly and includes some details. When the person reports past loss or trauma, she is rated by how well she coped with it. For the person’s attachment to be rated as autonomous secure, she must clearly recognize — not dismiss — the negative event; must not dwell on it; must recognize that it was not her fault. Detailed descriptions of the AAI rating system have not been published. The foregoing description of the AAI was taken from Steele, Steele, and Johansson (2002).

Longitudinal studies have been used to see if babies’ responses to the Strange Situation agreed with their AAI ratings when they were adolescents or young adults — that is, to see if attachment style at 12 or 18 months remained constant until adolescence and adulthood. The findings of these studies have varied.

Prebirth AAI ratings of both prospective mothers and fathers predicted their children’s responses to the Strange Situation when the children were 12 and 18 months (Steele, Steele, & Johansson, 2002). These same AAI ratings of prospective mothers — the primary caregivers in this study — as autonomous-secure or insecure, predicted ratings of their 11-year-old children’s coherence or incoherence, respectively, in the resolutions that the children proposed, and in their attribution of thoughts and feelings to characters in line-drawn sequences of stressful situations — for example, pictures of a child being bullied by another child or being ignored by an adult. Prebirth AAI ratings of the fathers did not predict these children’s ratings at 11.

However, the Strange-Situation response of these children when they were infants did not correlate with these children’s ratings when they were 11. An early-experience-matters-most hypothesis will not account for these findings.

Two studies found continuity between infants’ responses to the Strange Situation and the s e same children’s AAI classification later, when they were either adolescents or adults (Waters et al., 1995; Hamilton, 2000). Two studies found no such continuity (Lewis, Feiring, & Rosenthal, 2000; Sroufe et al., 1999). In the studies of all four populations, it was found that negative life events — for example, separation of parents or loss of a parent — after infancy was related to insecure attachment on the AAI. In one instance, an insecure baby who developed a serious illness scored secure on the AAI. Presumably, the parents became more attentive to the child when he was ill.

A way of looking at these findings is that, if the caregiver remains basically available, interested and concerned or if the caregiver remains unavailable or uninterested and unconcerned about the baby as the baby becomes an older child, it will look as if the adult’s working model was first formed in infancy. If nothing changes in the quality of the caregiver’s availability, interest and concern or in the family circumstances, it will look as if attachment style continued from the first year into adulthood. The findings do not support the idea, which has been implicit to Attachment Theory, that an adult’s working model is formed from experience at age 1 or 2.

My way of looking at these findings is that some continuity between infancy and adolescence or adulthood holds some of the time. The question is, When does it hold and what causes it to hold? Perhaps, repeat measures of caregiver-child-interaction from infancy to adulthood could establish whether and what kind of caregiving at a particular time in development is critical for a working model to become stable.

Running through the attachment literature is the notion that throughout life, secure attachment is good and insecure attachment is bad — that is, pathological. This notion persists in spite of the fact that the initial insecure attachment categorizations were made in nonclinical populations. We propose that the CMS type of a person who had an early (though later than proposed by Attachment theory) caregiving experience similar to the caregiving that is correlated with insecure attachment, is not necessarily dysfunctional or abnormal. Further, depending on the circumstances, each of the three CMS types has its own advantages and disadvantages.

Another notion that runs through the attachment literature is the sense that attachment is an abiding force throughout everyone’s life, and that types of attachment are intergenerational — that is, like mother, like child. This second point is supported to some extent by our finding, cited earlier, that the mother’s CMS type when her child was 4 or 9 tended to be the same as her adolescent’s CMS type. We found, however, that the mother’s CMS type when her child was an adolescent did not tend to be the same as her adolescent’s CMS type. Our findings suggest that there are times when the mother’s CMS type influences the later development of the child’s CMS type — and that there is a time after which it does not.

As mentioned, Attachment theory and its typology have been extensively tested. Various correlations and predictions from the infant types appear to hold, until age 6 or so. After that, they do not hold up very well. An interesting example is Lyons-Ruth, Alpern , and Repacholi’s (1993) study. They found that disorganized attachment — the D baby response — at 18 months failed to correlate with ratings of aggressive behavior in children at 5. Yet of the 5-year-olds who were aggressive, 71 % had exhibited disorganized attachment at 18 months. Perhaps when a child exhibits exaggerated behavior, such as becoming disorganized whenever he is left with a new babysitter, his parents or other caregivers often take notice and make some changes in what they do with him.

There is some convergence between Attachment theory and CMS theory. For example, measured by the AAI, mothers of A children are like our Intuitives and Intuitive Symbolics, and mothers of B children are like our Operationals or Operational Symbolics.

We also know that Bowlby thought of his patients as fully recovered, even if they needed to stay in contact with him after they finished treatment. Apparently, he considered the need for attachment as normal.

We agree with Bowlby that, if we exclude schizoid persons and psychopaths, we all need contact with others. We also think that the most prevalent adult CMS type is the Symbolic — a person for whom attachment is central — and that most Symbolics are normal. However, two of our patient types did not need to continue their attachment to the therapist after the treatment ended. One might claim that, after treatment ended, those patients attached to other persons or to organizations when they ended their attachment to the psychotherapist. And that certainly happens. We would claim, however, that attachment is not a central focus of Intuitives and Operationals.

As emphasized, a major difference between CMS theory and Attachment theory is that CMS theory proposes a discontinuity between the infant’s constructs of the self and of the world, and the Concrete Operational child’s constructs of the self and of the world. Attachment theory proposes that secure attachment is a continuous function. It begins in infancy, but it is a protective and persisting construct that lasts into adulthood unless the child is subject to adverse experiences. Insecure attachment, which also begins in infancy, is a less-than-optimal, persistent construct that lasts into adulthood. Each type of attachment lasts unless the child is subject to some corrective experience (Sroufe et al., 1999; Aguilar et al., 2000).

Many people, lay and professional alike, continue to assume that experience during the first 2 or 3 years determines adult traits and adult pathology, although empirical studies have repeatedly failed to justify this assumption. Kagan (1984) called attention to this inconsistency. Rutter (1984, p. 62) stated that "correlations from the infancy period to maturity [are] near-zero"

We do not dismiss the importance of the first 3 years. But a 3-year-old’s tasks and comprehensions as she prepares to go to kindergarten differ from a preadolescent’s tasks and comprehensions as she moves into the greater society. For example, the 3-year-old learns effective gross motor control and adapts to the feeding, toilet habits, and language of the family, while the preadolescent develops refined skills and complex rules that begin to deal with an adult physical and social world.

I think that Bowlby’s clinical observations that certain types of caregiving were related to later types of behavior were sound and groundbreaking. I believe, however, that he reached back to a time too early in child development to explain his clinical work with adult patients and his studies of delinquents (Bowlby, 1944), which were the origins of his ideas.

The types of infant-caregiver interaction that he thought would correlate with an infant’s particular working model are similar, though not identical, to the types of child-caregiver interaction that we think influences formation of CMS types. As mentioned, we differ from Bowlby in our proposal that our CMS types are not formed until much later than he proposed that his working model was formed. We also differ in our proposal that none of our three CMS types or their past caregiving experiences are inherently abnormal. In fact, we think each CMS type is adaptive to particular social situations.

Much work remains to be done in order to understand development of CMS. The statistical findings that support CMS theory, although real and important, are not so strong that they exclude factors other than caregiving style that affect development of CMS. CMS theory should not be seen as a total explanation. Environmental and constitutional factors, other than caregiving, cannot be discounted. Our testing methods sample the effects of an abiding type of caregiving during the Concrete Operational Period. It is possible that one salient caregiving experience could determine a person’s CMS type. Further, our testing does not shed light on the mechanism of CMS formation — that is, whether it is construction or identification. Finally, to make our predictions, we worked with assessments of actual caregiving that was revealed in a mother’s interviews. We do not know how her child experienced his caregiving or his caregiver, which we think is instrumental in determining the CMS type that he manifests later. We assumed that, on-balance, the child’s experience correlated with actual caregiving.

Block (1971) attributed to Freud the statement that life is lived forward, but can be understood only backward. Generally speaking, this is so. However, Attachment theory — derived from Bowlby’s psychodynamic formulation — predicts certain types of behavior from infancy to age 6 or 7. And CMS theory — derived from an adaptation of Piaget’s work to understand a psychodynamic construction of character structure — predicts types of behavior in adolescence from caregiving during the period between 7 and 11. The predictions from Attachment theory and from CMS theory show that, at different periods of time, we can understand some aspects of life lived forward.

This chapter is a brief account of CMS theory and of empirical testing of the theory. Although the findings are modest, to my knowledge, no other clinically-derived theory of character structure formation has withstood such testing.

My primary effort in this chapter and in the earlier chapters has been to synthesize findings from different domains of knowledge. To do this, I proposed some unusual ideas or theories. These are listed in the Epilogue.

back to top

---

back to top

< Chapter 9 | Contents | Chapter 11 >

Do you have a comment or would like to open a discussion? The author would be pleased to hear from you. Contact AJ Malerstein here.

DevMindBrain.com

Copyright © 2006-2010 AJ Malerstein