Loading Quotes...

The Circle of Security Project

The Circle of Security Project—An Attachment Theory-based Intervention

Used by permission
Permission given by

Robert Marvin, Ph.D.—Director
The Mary D. Ainsworth Child-Parent Attachment Clinic
University of Virginia Medical Center

Please do not make copies of this document
without first receiving permission from the author

Editorial comments by the Foundation are shown in italics


The following excerpt was taken from the narrative of a grant proposal submitted to the Foundation by Robert Marvin, Ph.D., in support of the Circle of Security Project. Frederick Leonhardt, the Foundation’s president, in reviewing this grant request, was impressed by the information this narrative contained on taking a cautious stance towards treatment modalities that go by such names as “Rebirthing,” “Rage Reduction,” “Attachment Therapy,” or “Holding Therapy.” The president felt that this narrative contained information that should be readily available to the general public. The president asked Dr. Marvin for permission to place this information on the Foundation’s web site as a web document. Dr. Marvin graciously gave his permission for the Foundation to do so. The Foundation extends it’s thanks to Dr. Marvin for sharing this information with the general public. For more on these types of treatment modalities, please see the book “Attachment Therapy on Trial” by Jean Mercer and her colleagues (please see the Reference area under the Resource menu). The FHL Foundation ascribes to the Bowlby-Ainsworth theory of attachment, and, as such, uses this theoretical orientation as a backdrop when evaluating grant requests and proposals.


Robert S. Marvin, Ph.D., Director
The Mary D. Ainsworth Child-Parent Attachment Clinic
University of Virginia Medical Center


The preschool period, between the ages of 18 or 20 months and about 5 years, is an especially important period in the development of children’s attachment relationships, and their relationship skills generally. Having consolidated their differential attachment(s) during infancy, it is during this latter period that they begin increasingly to organize their use of their attachment figures as Secure Bases and Havens of Safety on the basis of inferences they draw about their attachment figures’ internal states, and their emerging abilities to negotiate and compromise (e.g., Bowlby, 1969/82, Chap. 17; Marvin, 1977; Marvin & Britner, 1999). Between their third and fourth birthdays, for example, children become able to think about their mothers’ thoughts, feelings, and desires, are able to differentiate mother’s mental states from their own, and in a purposeful way are able to construct and interact on the basis of joint goals and plans with their mothers (Marvin & Greenberg, 1982). Bowlby (1969/82) labeled this fourth phase in the development of attachment the “goal-corrected partnership.” This set of developmental and relationship skills have, over the past 30 years, been labeled variously as perspective-taking (e.g., Flavell, 1985), “theory of mind” (e.g., Wellman, 1990), and more recently “mentalization” (e.g., Fonagy, Gergely, Jurist, & Target, 2002).

In the field of attachment research, this mentalizing activity has received most attention in the context of the caregiver’s mentalizing rather than the child’s. This is not too inappropriate, because of the crucial roles that the caregiver’s thoughts, feelings, and behavior toward the child play in the pattern and security of the child’s attachment (Bretherton & Munholland, 1999; Main & Hesse, 1990). Ainsworth herself (e.g., Ainsworth, Blehar, Waters & Wall, 1978) insisted that a caregiver’s ability to use the child’s cues to make accurate inferences about his or her internal state and needs is a crucial component of “maternal sensitivity.” And certainly the preschool child’s ability to make accurate inferences about mother’s internal states—the child’s mentalizing about the caregiver—is based in large part on the quality, accuracy, and patterns of the caregiver’s mentalizing in relationship with the child. This suggests that intervening in high-risk or problematic caregiver-preschool child attachments should include a significant focus on improving the parent’s inferences about the child (and herself!), in order to improve the attachment interactions between the two of them—with those attachment interactions themselves now being organized so extensively in terms of sometimes-different-sometimes-shared goals, plans, and feelings.

There is also increasing evidence from attachment research that individual differences in patterns of attachment-caregiving interactions tend, during the preschool years as well as during infancy, to be shared by parent and child in the form of a reciprocal behavioral “dance” (see Ainsworth et al., 1978; Cassidy, 1999; Cassidy & Marvin, 1992; Britner, Marvin & Pianta, in press). Each child and caregiver pattern can be identified reliably in terms of specific organizations of behaviors and of internal working models of both partners. At least four distinct dyadic patterns have been identified for older toddlers and preschool children (e.g., Cassidy & Marvin, 1992; Boris et. al., 2004): Secure; Insecure-Avoidant, Insecure-Ambivalent (Resistant); Disorganized Role-reversed; and Disorganized Insecure-Other (a group consisting of especially problematic patterns reflecting severe and chronic trauma, deprivation, neglect, and other forms of maltreatment).

Attachment Interventions

While there are a number of early intervention projects being developed and tested that focus on the security of attachment-caregiving relationships in high-risk parents and their infants (e.g., (Berlin, Ziv, Amaya-Jackson, & Greenberg, in press; Lieberman, Weston, & Pawl (1991); Van den Boom, 1995), there are very few evidence-based programs that are focused on older toddlers and preschool children. Unfortunately, those programs for preschoolers and older children that do exist tend to take one or a few ideas from attachment theory and research, incorporate them in an oversimplified manner into another, usually incompatible, theoretical framework, and apply the resulting programs without adequately testing their effectiveness.

Specifically, these programs, some of which are known as Rebirthing, Rage Reduction, Attachment, or Holding Therapy, and referred to in this article as “the holding therapies,” have been published privately, in edited books, and on the Internet (e.g., Cline & Fay, 1990; Levy & Orlans, 1998; Thomas, 1997; c.f., Mercer, Sarner, & Rosa, 2003). The designers of these programs draw from Bowlby’s theory a focus on the importance during infancy of close bodily contact, security, and trust in the child’s attachment figure. However, these “holding therapies” tend to have taken those constructs and integrated them, without appropriate rules of theory construction, into intervention models based either: a) on a distorted version of the classical psychoanalytic theory of regression (Rebirthing and Rage Reduction therapies); and/or on behaviorism and principles of behavior modification designed to reduce or eliminate undesirable behaviors (Attachment and Holding therapies). Clearly, these are not really attachment theory-based interventions, and they are not appropriately evidence-based.

Characteristics of the holding therapies have led to the proliferation of interventions that, while they are administered by therapists with the best of intentions, are at worst abusive and potentially life-threatening (e.g., the Evergreen Rage Reduction program), and at best are neither based on attachment theory, nor are empirically tested regarding treatment effectiveness (the less radical versions of “Holding Therapy”). Unfortunately, many of these programs claim to be based on the Bowlby-Ainsworth theory of attachment, and there are literally thousands of therapists in communities around the U.S., Canada, Europe, and Australia who use them in their daily practice. In defense of these therapists, it should also be noted that until very recently there have been few if any evidence-based interventions for preschool and older children, truly developed from attachment research, available to them.

Interventions that are evidence-based and derived from attachment theory will have many differences from the holding therapies. Three that are especially pertinent to evidence-based practice are:

1. Holding therapies tend to use a diagnostic system that is non-standardized, non-validated, and not able to differentiate among the many psychological disorders found in childhood. In contrast, attachment research-based interventions should utilize a “diagnostic” system that is based specifically on thought and behavior patterns from attachment theory and research, and on research in other areas of parent-child interaction and relationships. This system should be scientifically and clinically reliable and validated. It should be sufficiently limited in scope both to identify specific attachment problems, and at the same time to differentiate these problems from other childhood and relationship difficulties as well as from formal diagnostic groups such as ADHD, Oppositional Defiant Disorder, Conduct Disorder, Bi-Polar Disorder, and other disorders so often inappropriately associated in clinical practice with “Attachment Disorders.”

2. Holding therapies tend to direct the intervention to the child him or herself. In contrast, all attachment research-based interventions with which we are familiar, including the Circle of Security (described below), are designed to impact the child’s attachment pattern indirectly by directly focusing on caregiver patterns of behavior and thinking. This practice follows directly from Ainsworth’s focus on caregiver sensitivity to the child’s cues as perhaps the major variable in predicting security of attachment.

3. Holding therapies tend to view the behavior- and emotion-regulation problems of children with attachment difficulties as unacceptable behaviors, and often make inferences about negative intent on the part of the child. Most holding therapies attempt to eliminate or extinguish these intentions and behaviors through consistent, strong, and often negative, consequences that are based either on aversive behaviors toward the child, or on temporary separation, or interruption of the interaction, between the child and caregiver. In contrast, interventions based on attachment theory are likely to view the behavior- and emotion-regulation problems associated with attachment difficulties as patterns of thinking and behavior that are both an understandable adaptation to the child’s unique developmental and relationship history, and are reflections of the child’s inability, at the moment, adequately to self-regulate his or her affect and behavior. Even if the child is acting in an aggressive, controlling manner, attachment-based interventions will, in most but not all situations, view the child as anxious, vulnerable, “in over his head,” out of control, and needing the parent’s empathy, soothing, and management.

The Circle of Security (COS) Intervention Protocol

To our knowledge the only standardized, attachment theory-based protocol designed to intervene in the attachment-caregiving relationships of both toddlers and preschool children is the “Circle of Security” protocol. This is a protocol developed by Kent Hoffman, Bert Powell, and Glen Cooper of the Marycliff Institute, Spokane, WA, and this author. The following is an abbreviated outline; the reader is urged to read Marvin, Cooper, Hoffman & Powell (2002), and Cooper, Hoffman, Powell & Marvin (in press) for more detailed presentations of the protocol.

The protocol uses established attachment research procedures to diagnose and design a specific, individualized, treatment goal for each dyad. The actual intervention process, a 20-session weekly group-based protocol, is also derived from attachment theory and research, with specific exercises to: help the parent be more reflective about her child and herself; develop a practical, “user-friendly” understanding of attachment theory; and become more accurate and empathic in reading the child’s cues and miscues. The specific intervention protocol is based on video-review of the caregiver’s interactions with her young child in order to help her reflect accurately on those interactions. The video-review is also used to “trigger” a co-reflective process between therapist and caregiver about the caregiver’s own attachment-history, about how that history is related to the feelings experienced in challenging interactions with the child, and about how the caregiver’s reactions to his or her own painful and confusing feelings in attachment-caregiving interactions can interfere with the strong desire to be sensitively responsive to the child. Finally, by assessing the pattern of attachment-caregiving interactions again immediately post-intervention—as well as one year later—we can scientifically evaluate the effectiveness of the protocol.

The Circle of Security protocol received the Year 2000 Washington State Council for Prevention of Child Abuse and Neglect: Governor’s Award for “Innovations in Prevention.” In 2003 it was placed on the DHHS list of “promising” intervention protocols. Versions of the protocol are currently being used in Spokane, WA; Virginia; New Orleans, LA: Washington, D.C.; and the women’s prison system in Baltimore, MD. Studies of the effectiveness of different versions of the protocol are currently in progress.


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

Berlin, L.J., Ziv, Y., Amaya-Jackson, L. M., & Greenberg, M. T. (in press). Enhancing early attachments: Theory, research, intervention, and policy. New York: Guilford Press.

Boris, N., Hinshaw-Fuselier, S., Smyke, A., Scheeringa, M., Heller, S., & Zeanah, C. (2004). Comparing Criteria for Attachment Disorders: Establishing Reliability and Validity in High-Risk Samples. J . Am. Acad. Child & Adolescent Psychiatry, 43:5, 568-577.

Bowlby, J. (1969/1982). Attachment and loss. Vol. I: Attachment (2nd ed.). New York: Basic Books.

Bretherton, I., & Munholland, K. (1999). Internal Working Models in Attachment Relationships: A construct Revisited. In Cassidy, J. & Shaver, P. R., (Eds.), Handbook of attachment: Theory, research, and clinical applications. New York: Guilford.

Britner, P, Marvin, R, & Pianta, R. (In press). Parental Caregiving Behavior and Child Attachment: Classifications in the Preschool Strange Situation. Attachment & Human Development.

Cassidy, J. & Shaver, P. R., (Eds.) (1999). Handbook of attachment: Theory, research, and clinical applications. New York: Guilford.

Cassidy, J., & Marvin, R. S., with the MacArthur Working Group (1992). Attachment organization in preschool children: Procedures and coding manual (4th ed.). Unpublished manuscript, University of Virginia.

Cline, F. & Fay, J. (1990). Parenting with Love & Logic, Navpress.

Cooper, G., Hoffman, K., Powell, B., & Marvin, R. (in press). The Circle of Security intervention: Differential diagnosis and differential treatment. In Berlin, L.J., Ziv, Y., Amaya-Jackson, L. M., & Greenberg, M. T., Enhancing early attachments: Theory, research, intervention, and policy. New York: Guilford Press.

Flavell, J. (1990). Perspectives on Perspective-Taking. Paper presented at the 20th Annual Symposium of the Jean Piaget Society. Philadelphia, June 2.

Fonagy, P., G. Gergely, E. Jurist & M. Target. (2002). Affect Regulation, Mentalization and the Development of the Self. New York: Other Press.

Levy, T. & Orlans, M. (1998). Attachment, Trauma, and Healing: Understanding and Treating Attachment Disorder in Children and Families, Terry M. Levy and Michael Orlans, CWLA.

Lieberman, A.F., Weston, D., & Pawl, J. (1991). Preventive intervention and outcome with Anxiously attached dyads. Child Development, 62, 199–209.

Main, M. & Hesse, E. (1990). Parents’ unresolved traumatic experiences are related to infant disorganized attachment status: Is frightened and/or frightening parental behavior the linking mechanism? In In Greenberg, M., Cicchetti, D., and Cummings, M. (Eds.), Attachment In The Preschool Years: Theory, Research, and Intervention. Chicago: University of Chicago Press.

Marvin, R. S. (1977). An ethological-cognitive model for the attenuation of mother-child attachment behavior. In T. Alloway, P. Pliner, & L. Kramer (Eds.), Attachment behavior (pp. 25-60). New York: Plenum.

Marvin, R. S., & Britner, P. A. (1999). Normative development: The ontogeny of attachment. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 44-67). New York: Guilford.

Marvin, R.S., and Greenberg, M.T.(1982). Preschoolers’ changing conceptions of their mothers: a social-cognitive study of mother-child attachment. In D. Forbes and M.T. Greenberg (eds.), New Directors for Child Development: Children’s Planning Strategies, no. 18. San Francisco: Jossey-Bass.

Marvin, R. S., Cooper, G., Hoffman, K., & Powell, B. (2002). The Circle of Security project: Attachment-based intervention with caregiver-pre-school child dyads. Attachment & Human Development, 4, 107-124.

Mercer, J., Sarner, L., & Rosa, L. (2003). Attachment therapy on trial: the torture and death of Candace Newmaker. Greenwood Publ. Group.

Thomas, N. (1997). When Love is Not Enough: A Guide to Parenting Children with RAD.

van den Boom, D. C. (1995). Do first-year intervention effects endure? Follow-up during toddlerhood of a sample of Dutch irritable infants. Child Development, 66, 1798-18.

Wellman, H. M. (1990). The Child’s Theory of Mind. Cambridge, MA: MIT Press.