Dr Richard Bolstad is Transformations Principal Trainer

Parts Integration And Psychotherapy

by Richard Bolstad.

A Neurological Model For Understanding The Task Of Therapy

Can you talk comfortably with non-NLP trained psychotherapists about what we are doing in NLP? Do you have a model which enables you to explain NLP processes to a neurologist or medical doctor? If you'd like to, this article is for you.

Firstly, I plan to present one NLP compatible framework for understanding what we in NLP term "parts integration". Parts Integration is just one example of an NLP change process. I trust that NLP Practitioners will generalise the framework I present and use it to think about other things we do, especially anchoring and strategy installation.

I will show how this framework for understanding parts integration also explains what most other psychotherapy is doing. This will be recognised by NLP Practitioners as an unashamed attempt to make connections across the arbitrary barriers of our field and suggest that all psychotherapy has the same highest intention. As NLP Practitioners, you will also understand that it really does not matter whether this is hypothetically "true". The question is, does it help our clients more for us to believe we are totally unique and unrelated to all other therapy, or to believe that we are the latest in a series of approximations to the goal of successful psychotherapy? The answer, as NLP Practitioners realise already, is "Yes!".

The Brain

Since the brain is so closely linked to our psyche, it seems a good enough place to begin in attempting to explain what happens when we heal that psyche in psychotherapy. The human brain itself is made up of about one hundred billion nerve cells or neurons. These cells organise themselves into networks to manage specific tasks. When any experience occurs in our life, new neural networks are laid down to record that event and its meaning. To create these networks, the neurons grow an array of new dendrites (connections to other neurons). Each neuron has up to 20,000 dendrites, connecting it simultaneously into perhaps hundreds of different neural networks.

Steven Rose (1992) gives an example from his research with new-hatched chicks. After eating silver beads with a bitter coating, the chicks learn to avoid such beads. One peck is enough to cause the learning. Rose demonstrated that the chicks' brain cells change instantly, growing 60% more dendrites in the next 15 minutes. These new connections occur in very specific areas -what we might call the "bitter bead neural networks". These neural networks now store an important new strategy. The strategy is triggered each time the chick sees an object the right shape and size to peck at. This is a visual strategy of course. The trigger (seeing a small round object) is Visual external (Ve) and the operation (checking the colour) is also Visual external (Ve). The chick then compares the colour of the object it has found with the colour of the horrible bitter beads from its visual recall (Ve/Vr) and based on that test either pecks the object or moves away from it (Ke). We would diagram this strategy, for NLP purposes: Ve.Ve.Ve/Vr.Ke.

Obviously, the more strategies we learn, the more neural networks will be set up in the brain. California researcher Dr Marion Diamond (1988) and her Illinois colleague Dr William Greenough (1992) have demonstrated that rats in "enriched" environments grow 25% more dendrite connections than usual, as they lay down hundreds of new strategies. Autopsy studies on humans confirm the process. Graduate students have 40% more dendrite connections than high school dropouts, and those students who challenged themselves more had even higher scores (Jacobs et alia, 1993).

How do messages get from one neuron to another in the brain? The transmission of impulses between neurons and dendrites occurs via hundreds of precise chemicals called "information substances"; substances such as dopamine, noradrenaline (norepinephrine), and acetylcholine. These chemical float from one cell to another, transmitting messages across the "synapse" or gap between them. Without these chemicals, the strategy stored in the neural network cannot run. These chemicals are also the basis for what we call an emotional state, and they infuse not just the nervous system but the entire body, altering every body system. A considerable amount of research suggests that strong emotional states are useful to learning new strategies. J. O'Keefe and L. Nadel found (Jensen, 1995, p 38) that emotions enhance the brain's ability to make cognitive maps of (understand and organise) new information. Dr James McGaugh, psychobiologist at UC Irvine notes that even injecting rats with a blend of emotion related hormones such as enkephalin and adrenaline means that the rats remember longer and better (Jensen, 1995, p 33-34). He says "We think these chemicals are memory fixatives(?)�&. They signal the brain, "This is important, keep this!"... emotions can and do enhance retention."

Neural Networks Are State Dependent

However there is another important effect of the emotional state on the strategies we run. The particular mixture of chemicals present when a neural network is laid down must be recreated for the neural network to be fully re-activated and for the strategy it holds to run as it originally did. If someone is angry, for example, when a particular new event happens, they have higher noradrenaline levels. Future events which result in higher noradrenaline levels will re-activate this neural network and the strategy they used then. As a result, the new event will be connected by dendrites to the previous one, and there will even be a tendency to confuse the new event with the previous one. If my childhood caregiver yelled at me and told me that I was stupid, I may have entered a state of fear, and stored that memory in a very important neural network. When someone else yells at me as an adult, if I access the same state of fear, I may feel as if I am re-experiencing the original event, and may even hear a voice telling me I'm stupid.

This is called "state dependent memory and learning" or SDML. Our memories and learnings, our strategies, are dependent on the state they are created in. "Neuronal networks may be defined in terms of the activation of specifically localised areas of neurons by information substances that reach them via diffusion through the extracellular fluid... In the simplest case, a 15-square mm neuronal network could be turned on or off by the presence or absence of a specific information substance. That is, the activity of this neuronal network would be "state-dependent" on the presence or absence of that information substance." (Rossi and Cheek, 1988, p 57). Actually, all learning is state dependent, and examples of this phenomenon have been understood for a long time. When someone is drunk, their body is flooded with alcohol and its by-products. All experiences encoded at that time are encoded in a very different state to normal. If the difference is severe enough, they may not be able to access those memories at all(?)�& until they get drunk again!

At times, the neural networks laid down in one experience or set of experiences can be quite "cut off" (due to their different neuro-chemical basis) from the rest of the person's brain. New brain scanning techniques begin to give us more realistic images of how this actually looks. Psychiatrist Don Condie and neurobiologist Guochuan Tsai used a fMRI scanner to study the brain patterns of a woman with "multiple personality disorder". In this disorder, the woman switched regularly between her normal personality and an alter ego called "Guardian". The two personalities had separate memory systems and quite different strategies. The fMRI brain scan showed that each of these two personalities used different neural networks (different areas of the brain lit up when each personality emerged). If the woman only pretended to be a separate person, her brain continued to use her usual neural networks, but as soon as the "Guardian" actually took over her consciousness, it activated precise, different areas of the hippocampus and surrounding temporal cortex (brain areas associated with memory and emotion).(Adler, 1999, p 29-30)

To give an example from another psychiatric condition, research on Post Traumatic Stress Disorder shows the state-dependent nature of its symptoms (van der Kolk et alia, 1996, p291-292). Sudden re-experiencing of a traumatic event (called a flashback) is one of the key problems in PTSD. Medications which stimulate body arousal (such as lactate, a by-product of physiological stress) will produce flashbacks in people with PTSD, but not in people without the problem (Rainey et alia, 1987; Southwick et alia, 1993). Other laboratory studies show that sensory stimuli which recreate some aspect of the original trauma (such as a sudden noise) will also cause full flashbacks in people with PTSD (van der Kolk, 1994). This phenomenon is Pavlov's "classical conditioning", also known in NLP as "anchoring". State dependent learning is the biological process behind classical conditioning. The results of such classical conditioning can be bizarre. Mice who have been given electric shocks while in a small box will actually voluntarily return to that box when they experience a subsequent physical stress (Mitchell et alia, 1985). This is not a very nice experiment, and it also sheds light on some bizarre behaviours that some humans engage in.

Freud based much of his approach to therapy on the idea of "repression" and an internal struggle for control of memory and thinking strategies. This explanation of the existence of "unconscious" memories and motivations ("complexes") can now be expanded by the state dependent learning hypothesis. No internal struggle is needed to account for any of the previously described phenomena. The "complex" (in Freudian terms) can be considered as simply a series of strategies being run from a neural network which is not activated by the person's usual chemical states. Rossi and Cheek note "This leads to the provocative insight that the entire history of depth psychology and psychoanalysis now can be understood as a prolonged clinical investigation of how dissociated or state-dependent memories remain active at unconscious levels, giving rise to the "complexes"... that are the source of psychological and psychosomatic problems." (Rossi and Cheek, 1988, p 57).

People come to psychotherapists to solve a variety of problems. Most of these are due to strategies which are run by state-dependent neural networks that are quite dramatically separated from the rest of the person's brain. This means that the person has all the skills they need to solve their own problem, but those skills are kept in neural networks which are not able to connect with the networks from which their problems are run. The task of NLP change agents is often to transfer skills from functional networks (networks that do things the person is pleased with) to less functional networks (networks that do things they are not happy about). One easy way to do this is to ensure that the neural networks are connected into one unit

Parts Integration In NLP

In describing trance, Milton Erickson makes the distinction between the "conscious mind" and the "unconscious mind" (Erickson, 1980, Vol III, p 611). This recognition of two separate "subpersonalities" within the brain is one specific example of a wider model, describing what are called in NLP "parts". For our purposes, the word "part" refers to any state dependent neural network with enough functional autonomy to run its strategies without control by the rest of the brain. The "conscious mind" is, in this sense, as much a part as the unconscious.

As early as 1976 Richard Bandler and John Grinder presented a variety of different ways to work with conflicting parts, including Satir's Parts Party and the Gestalt Empty Chair process. Their methods involved conceptually separating or "sorting" the "polarities" (another Gestalt term for opposing parts) and having the client assume a meta-position from which they can enable the parts to "reorganise themselves into a new, single representation which will include all of the paramessages of both, and will be itself greater than the sum of the two." (Grinder and Bandler, 1976, p 82).

One early model of parts work used in NLP was called six step reframing. Bandler and Grinder explain its rationale in an early demonstration by saying "This only makes sense if you have a belief system that says "Look. If he had conscious control over this behaviour it would have changed already." So some part of him which is not conscious is running this pattern of behaviour... I also make the assumption that the part of you that makes you X -even though you don't like that consciously- is doing something on your behalf, something that benefits you in some way." (Bandler and Grinder, 1979, p 139-140).

Six step reframing involves communicating with the part and asking the person's unconscious mind to think up some equally effective and more acceptable ways to meet the positive outcome of that part. This method has been shown to be as effective as relaxation training at curing psychosomatic physical symptoms (such as headaches; see Bacon, 1983). Six step reframing, unlike relaxation, does not need to be consciously applied in an ongoing way to keep the problem solved. The method has also been successfully used with addictive and compulsive behaviour patterns such as bulimia (Gl(?)ser, 1991) and alcoholism (Sterman, 1990).

The parts integration model itself is a development of an earlier process called the Visual Squash. Its advantage is that it connects two opposing parts together, where the six step reframe left the part that was of concern to generate new behaviours while still separate. Parts Integration is discussed by Robert Dilts in the book Beliefs (Dilts, Hallbom and Smith, 1990, p 101-126, p 165) where Dilts gives a transcript of its successful use to treat an allergy to cats, tested in the training room immediately after.

Since the development of Parts Integration, other NLP processes to deal with parts issues have continued to evolve, including Core Transformation (Andreas 1992) and Time Line Therapy TM (James and Woodsmall, 1988). Generally, these utilise the outcome chain process, where the person is asked for the part's "outcome" and then asked "If you get that fully and completely, what do you get, through having that, that's even more important?".

Parts Integration In Psychodynamic Therapies

Each model of psychotherapy has had to account for the activity of state dependent neural networks, and evolve some model of such "parts" of the mind. In each case, the structures are defined differently, and the precision of these definitions is often of importance to the development of techniques within that therapy. Respecting this importance, I will attempt to acknowledge carefully some of these distinctions here, while drawing attention to the universality of parts work in the field.

The notion of parts has its psychotherapeutic origins in the dynamic model of psychoanalysis. There, the core parts of the psyche spoken of are the id (which Freud says is based on "untamed passions"), the ego (the area based on "reason and circumspection") and the superego (which upholds socially required, moral "norms of behaviour"). Freud suggested that the psychoanalyst is allied with the ego to defend it from anxiety and extend its "territory". He says his aim is "to strengthen the ego, to make it more independent of the super- ego, to widen its field of vision, and so to extend its organisation that it can take over new portions of the id. Where id was, there shall ego be." (Freud, 1933).

Modern object-relations based psychoanalysis has paid more attention to certain splits within the ego itself; particularly splits which occur pathologically if too severe a challenge confronts the ego early in life. A healthy mature ego, says Otto Kernberg, can repress undesired material out of conscious awareness, without needing to split it off completely. He says "Splitting consists in dissociating or actively maintaining apart identification systems with opposite valences (conflicting identification systems) without regard to access to consciousness or to perceptual or motor control." (Kernberg, 1976, p 44). The result is severe personality disorders and what NLP calls "sequential incongruity" -doing and desiring one thing at one time, and the opposite at another time. Healing is a process of "learning of "managerial skills" in order to understand one's self, one's boundaries, one's internal needs, one's environment and one's life tasks." (Kernberg, 1976, p 265). This managerial or co-ordinating function of the healthy ego is what NLP refers to as parts work.

By contrast, the focus of Transactional Analysis has been on healthy functional divisions within the ego. It defines three core ego states called the Parent (using strategies copied from parents/caregivers), Adult (using strategies developed in adult life and as here-and-now responses) and Child (using strategies learned as a child), and some subdivisions of these. Each ego state is "a consistent pattern of feeling and experience directly related to a corresponding persistent pattern of behaviour" (Stewart and Joines, 1987, p 15). In other words, each ego state is a state-dependent set of strategies. The aim of TA work with these ego states is often to prevent one state "contaminating" another or "excluding" another from expression. Again, the aim is the development of a situation where each state has access to expression in ways that are useful to the entire person, and this is what we are doing in parts work in NLP.

Parts Integration In Transpersonal and Interpersonal Therapies

In these formulations we have been discussing, the assumption is made that the main "parts" to be found in a human being are largely similar from one person to another. With the theory of archetypes common to the collective unconscious of humanity, Carl Jung takes this a step further. He describes an archetype as "an inherited mode of psychic functioning... In other words it is a 'pattern of behaviour'." (quoted in Whitmont, 1991, p 104). He describes in this way such elements as the anima (inner feminine principle in a male) and animus (inner masculine principle in a woman), the persona (our adaptation to social expectations) and the Self (a suprapersonal summation of ones life and its meaning). Archetypes are the central elements of complexes. Jung explains that each complex "appears as an autonomous formation intruding on consciousness... While complexes owe their relative autonomy to their emotional nature, their expression is always dependent on a network of associations, grouped round a centre charged with affect." (in Whitmont, 1991, p 63-64). In other words, again, we have here a definition of state dependent strategies. Unlike Freud, however, Jung did not see it as the ego's task to replace such other complexes. He says "Conscious and unconscious do not make a whole when one of them is suppressed and injured by the other(?)�&. Both are aspects of life... It is the old game of hammer and anvil; between them the patient iron is forged into an indestructible whole, an "individual"." (in Whitmont, 1991, p 18).

The work of Roberto Assagioli describes parts work in a very similar frame to that used by NLP. The aim, in both cases, is the integration of separate parts into one unitary whole (a process called by Assagioli Psychosynthesis). Assagioli describes his aim as "Co-ordination and subordination of the various psychological energies and functions, the creation of a firm organisation of the personality." (Assagioli, 1976, p 29). In Psychosynthesis, as in NLP, this process is recognised as ultimately leading to transcendent states of pure awareness, joy, peace and love, and as extending beyond the "individual human being" as normally understood in the west (Assagioli, 1976, p 5).

The metaphor Jacob Moreno uses for parts is drawn from the theatre. By "role", he refers to "the functioning form the individual assumes in the specific moment he reacts to the specific situation in which other persons or objects are involved." (Moreno, 1977, p iv). Moreno points out that the role, with its strategies, its emotional state etc, evolves independently from the individual's interaction with the world. "Role playing is prior to the emergence of the self. Roles do not emerge from the self, but the self may emerge from roles... It is possible, as we see with infants and psychotics, for the individual to operate with several alter egos." (Moreno, 1977, p 153). Part of the task of therapy, then, is to help individuals to reduce role conflict, to harmonise the functioning of roles, and to allow the smooth transition from role to role as required by the situation. The ability to produce roles which respond adequately and effectively to unprecedented situations (spontaneity) is a highly valued aim of Moreno's work (Moreno, 1977, p 85-89), and parallels the aim of parts work in NLP. In Psychodrama, a role is said to have five components. These equate to some extent with the TOTE: so that Context = the Trigger, Behaviour = the Operation, Belief = the Test, and Feeling and Consequences = the Exit. (Williams, 1989, p 58).

Parts Integration In Body Therapies

Alexander Lowen describes the healthy individual after body-oriented psychotherapy in physiological terms. "He is aware of his contact with the ground and feels more rooted. He says he feels connected with his body, his sexuality and the ground. To be so connected is not an ideal of health; in my opinion it is the minimum of health." (Lowen, 1972, p 61-62). What stops this connectedness, Lowen argues, are patterns of chronic muscular tension which prevent contact between the various layers of body tissue. Each layer, he suggests holds a state and the strategies which express it. Body tension is the key way people inhibit the activation of what are literally important "parts" of their experience. "In effect, the area of the body that would be involved in the expression of the impulse is deadened, relatively speaking, by the chronic muscular tension that develops as consequence of the continual holding pattern." (Lowen, 1972, p 81). By enabling full flowing breathing and movement, the body therapist creates access between the various "parts" of the person.

Arthur Janov evolved a model of Primal Therapy which is closely based on brain research. He says that the aim of his cathartic therapy is to release pain impulses which have been locked in lower brain circuits. Once the original events are re-experienced consciously, connection between these memories and the conscious mind has a profound healing effect. He says "Unconsciousness represents a breakdown in the integrative capacities of the brain as it mediates bodily processes. It occurs when the system is overloaded so that impulses (resulting from action potentials mediated by synapses) which normally have specific innervations with the cerebral cortex to make us conscious, overwhelm the integrative faculties and become shunted into alternate cerebral pathways rendering us, in that sense, unconscious," (Janov, 1977, p 4-5). Janov says "It is consciousness which finally conceptualises and interprets these deep pains and makes sense of them." (Janov, 1977, p 35).

The concept of parts as used in NLP was present directly in the work of Fritz Perls. Perls explains "If some of our thoughts, feelings, are unacceptable to us, we want to disown them. Me wanting to kill you? So we disown the killing thought and say, "That's not me -that's a compulsion," or we remove the killing, or we repress and become blind to that. There are many of these kinds of ways to remain intact, but always only at the cost of disowning many, many valuable parts of ourselves... You do not allow yourself -or you are not allowed- to be totally yourself." (Perls, 1969, p 11). He adds, "So what we are trying to do in therapy is step-by-step to re-own the disowned parts of the personality until the person becomes strong enough to facilitate his own growth(?)�&." (Perls, 1969, p 38).

Parts Integration In Humanistic And Cognitive Therapies

The other models for early NLP also used an explicit "parts" theory. Virginia Satir's way of dealing with parts issues was through a psychodramatic method she called the "Parts Party". She explains "We all have a number of different parts, each with expectations of fulfillment. These parts often find it difficult to get along withy each other and may have inhibitory influences on one another(?)�&. The process of the parts party offers a person an opportunity to observe these parts and to learn how they can function more harmoniously when they co-operate rather than compete." (Satir and Baldwin, 1983, p 258).

Client centred therapy developer Carl Rogers is far more cautious about suggesting that there may be organised "structures" within the human psyche, but he makes it clear that he is working with what NLP would call parts, when he says: "From a descriptive clinical point of view(?)�& we may say that successful therapy seems to entail the bringing into awareness, in an adequately differentiated and accurately symbolised way, those experiences and feelings which are currently in contradiction to the client's concept of self." (Rogers, 1973, p 148-149). A central result of therapy, Rogers notes, is "Increased unification and integration of the personality" (Rogers, 1973, p 178).

Behavioural psychotherapy began by carefully avoiding any speculation about the inner "structuring" of the mind. In the 1980s, cognitive behaviourists evolved the concept of schemas to account for many of the phenomena which we are discussing here. A schema is defined in two ways: "On the one hand, the term schema has been used to refer to a hypothesised structure of cognition, such as a mental filter or template that guides the processing of information(?)�&. More regularly, however, we use the construct of >schema in reference to the content of fundamental, core beliefs: the basic rules that an individual uses to organise his or her perceptions of the world, self and future, and to adapt to life's challenges (Layden et alia, 1993, p 7). Such schemas can be restructured (creating a whole new way of responding), modified (altered so that there are times the schema need not apply, for example) or reinterpreted (reframed as having a positive meaning rather than a limiting one); all methods that clearly relate to NLP parts work (Layden et alia, 1993, p 11-12).


I began by examining the way that new learnings, new strategies, are laid down in neural networks in the brain. When strategies are laid down during a very unique emotional state, such networks can be poorly connected to the rest of the person's brain. Psychotherapy involves reconnecting such networks so that the person can function in an integrated way. In NLP, a variety of methods have been developed to meet this outcome, including finding new solutions for a part with six step reframing, and creating oneness between two parts through parts integration. Other methods of psychotherapy have their own metaphors for these processes, including:

This article represents one small contribution towards a mental map of the field of psychotherapy from NLP perspectives. My belief is that all psychotherapy has been a search for ways to heal the same challenges. Neurological research will continue to refine one metaphor of how that happens. Having flexibility in the ways we conceptualise such change enables us to communicate with other therapists, as well as to do a better job ourselves.


Dr Richard Bolstad is an NLP Master Practitioner and Trainer who has worked with clients individually and as a trainer of groups since 1990. He can be contacted at PO Box 35111, Browns Bay, Auckland, New Zealand, Phone/Fax: +64-9-478-4895 E-mail: learn@transformations.net.nz Website: http://www.transformations.net.nz