Families in Psychiatry

John Bowlby and Heinz Kohut, meet Karl Friston

Attachment theory and object relations theory, meet neuroscience


The bonding process between individuals is essential to human life as we know it. The depth and strength of emotional bonds that we develop with other individuals are among the cardinal aspects of being human.

The strengths of the bonds that we form with others often are the most enduring aspects of our lives and can stretch back in time to include family members and friends who are no longer alive. The brain processes and neurochemicals involved in the bonding process are related to psychological processes using Bayesian theory. Abnormalities in attachment are described in terms of abnormal predictive models.

John Bowlby MD, and colleagues developed a psychological theory of early childhood attachment that shows the impact of different childhood experiences.1 Over many decades, they conducted painstaking observations, trials, and retrials that elucidated specific types of attachment that reflect the child’s internal representations of the parent.

Dr. Bowlby described the mental representations of attachment as “the internal working models of expectations” about the maternal-child relationship. This internal model guides the child’s behaviors, attitudes, and expectations, and is constantly being revised and expanded as the child matures in order to adapt to increasingly complex relationships. More current research has shown that adults who have experienced extreme early childhood adversity develop odd or deficient descriptions of primary attachment relationships, now refined into a low-coherence CC category.2 These children experience attachments that are characterized by feelings of emptiness, inconsistency, and fragmentation.

baby crying in crib ©a-fitz/iStockphoto.com

Object relations theory was created by Heinz Kohut, MD, to explain our internal models of others in psychoanalytic language.3 Dr. Kohut expands the concept of transference describing three new transferential relationships: mirroring, idealized, and twinship transferences. A mirroring transference describes the psychological mechanism whereby “the other” serves as a mirror that reflects back to the person a sense of self-worth and value. A mirroring transference uses the affirming and positive responses of others so that a person then can see positive traits within themselves. The idealizing transference refers to a person’s need for “another” who will make them feel calm and comfortable. The other is idealized as somebody who is calm and soothing when the person cannot provide that on their own. The twinship/alter ego transference occurs when a person feels a sense of likeness with “an other.” These advances in psychoanalytic theory advance our understanding of how we internalize aspects of other people and use these internalizations in our own development.

The neuroscience of attachment was revealed in the study of the children abandoned in Romanian orphanages before the 1989 revolution. These children were placed in orphanages that lacked adequate staff, and they were subjected to profound psychosocial neglect. In the Bucharest Early Intervention Project, the effects of neglect were seen in children aged less than 30-96 months, in EEG data.4 Abnormalities were found in two aberrantly connected brain networks: a hyperconnected parieto-occipital network and a hypoconnected network between left temporal and distributed bilateral regions. This study provides the first evidence of the adverse effects of early psychosocial neglect on the wiring of the developing brain.

Neurochemistry has suggested the impact of oxytocin on the promotion of attachment. Oxytocin has been called the “cuddle hormone,” as it is stimulated in mothers to promote bonding with their infant. Many studies have examined the introduction of intranasal oxytocin increases bonding and trust with others, but recent meta-analyses challenge those findings.5,6 Nevertheless, the process of bonding and attachment is thought to engage the mesocorticolimbic, nigrostriatal dopaminergic, and oxytocinergic systems. Tops and colleagues proposed that oxytocin facilitates a shift from ventral striatal “novelty processing” toward dorsal striatal “familiarity processing.”7 This shift to familiarity is thought to create secure internal working models. In addition, brain activity in specific regions, such as the right caudate, is correlated with romantic love.8 In summary, specific pathways and neurotransmitters underlie the emotional process of bonding and attachment.

However, it is Karl Friston, MD, who suggests a model of brain functioning that describes how we actually process our experiences of the world.9 Dr. Friston uses Bayesian modeling to describe how we interact and understand our world, describing the way we process information. Dr. Friston states that the process of inference that we use to discern our world is a process that uses Bayes’ theorem to update the probability for a specific hypothesis about our world. As more evidence or information becomes available, we update our internal models of the world. Predictive modeling is the term used in Bayesian terms to describe this process of continuous revision of our internal model of the world. Any new sensory input is compared to our current internal model, and if they do not match, our internal model of the world is updated and revised. Dr. Friston states: “As surprise is resolved, Bayesian model evidence is increased. This means that every living organism behaves as if it is a little statistician, analyzing its sensory data in exactly the same way that scientists evaluate the evidence for their hypotheses about how experimental data were caused. In this view, minimizing surprise is, literally, the search for evidence for one’s own existence.” This process of predictive modeling also can be applied to our internal models of other people, meaning that our internal model of another person is an iterative process that changes as the other person changes.

Dr. Alison Heru

Dr. Alison Heru

When our experience of the world and others fall into predictive patterns, we expect the “usual” responses from our world and from the people in it. This predictive processing makes it easier for us to understand our world, so that we do not have to continually analyze and reanalyze each event as a new experience. For example, a young child examines the waves as they crash on the shore. For a while, he will watch the waves but soon begins to recognize that there is a predictable pattern. With each ocean he visits, he will examine wave action and soon develop an internal model of what happens when the water reaches the shore. Similarly, when a young child has a secure and loving parent, he will expect security and love in his future relationships. On the other hand, the child who has been subjected to abuse and neglect has an internal working model of the other as nonreciprocal and perhaps irrelevant to their well-being. They will, therefore, have extreme difficulty in understanding the purpose or value of relationships. Their predictive model of the other reflects a lack of input in the case of neglect, or perverted input in the case of abuse. Their internal predictive model will require a great deal of therapeutic healthy inputs to be able to evolve to allow a healthy relationship with the other. This is the work of therapy.

When thinking about relationships, predictive processing makes clinical sense. In a committed partnership, each partner has a predictive model of their partner’s response to events/speech utterances/emotional displays. The prediction of the partner’s response is based on prior experience. This predictive model reduces the need to repeatedly reinterpret the partner’s actions. The partner is understood through an internal predictive model. When one partner changes their behavior, the other partner experiences the world/environment as disrupted or different. The couple, through an iterative process, can incorporate change and grow together. If the patient has an internalized model of a loving relationship, then they are more likely to find a partner who resonates with this internalized model. Using predictive modeling, we can understand how two people with their own internal models of each other interact. As one person approaches the other person, each internal model of the other person finds a “fit” with the behaviors of the other person. When a child with emotional deprivation approaches the other, the child perceives the other person in a way that fits with their prior experience of deprivation. It is an easy fit that requires little internal work, and the patient falls into a familiar relational pattern. This model of predictive processing explains why people repeatedly fall into similar familiar relationships: They recognize the relationship patterns. There is less predictive processing required, less emotional and cognitive work, and therefore less adjustment.

Psychotherapy helps patients by improving their understanding of the link between prior and current experience. Each piece of behavior is analyzed and compared with the patient’s internal model of the event. What neuroscience now provides is a deeper scientific understanding of this process. A neuroscience-based visualization of our internal models of other people can help us better describe the iterative process that occurs in therapeutic progress.

In summary, the processes of attachment and bonding occur through oxytocin-mediated caudate–dorsal striatum pathways. Bayesian predictive modeling describes the processes by which our internal models of others are shaped and refined, and underscores the work of our psychotherapies. Neuroscience is beginning to delineate the “where” and “how” of attachment and bonding, thus advancing our understanding of the attachment process and giving us a new language to describe the work of psychotherapy.


1. Bowlby J. “A Secure Base: Parent-Child Attachment and Healthy Human Development.” New York: Basic Books, 1988.

2. Speranza AM et al. Attach Hum Dev. 2017 Dec;19(6):613-34.

3. Kohut H. “How Does Analysis Cure?” Chicago: University of Chicago Press, 1984.

4. Stamoulis C et al. J Neurophysiol. 2017 Oct 1;118(4):2275-88.

5. Leng G and Ludwig M.Biol Psychiatry. 2016 Feb 1;79(3):243-50.

6. Leppanen J et al. Neurosci Biobehav Rev. 2017 Jul;78:125-44.

7. Tops M et al. Pharmacol Biochem Behav. 2014 Apr;119:39-48.

8. Acevedo BP et al. Soc Cogn Affect Neurosci. 2012 Feb;7(2):145-59.

9. Friston K. Entropy (Basel). 2012 Nov;14(11):2100-21.

Next Article: