Symbolic Modelling: An overview

The article that introduced Symbolic Modelling to the world.
Print Friendly, PDF & Email
First published in Rapport magazine Issue 38
“The inability to distinguish either behaviourally or cognitively the consequences and applications of NLP (Neuro-Linguistic Programming) from core NLP itself (modelling of excellence) is extremely commonplace.”
John Grinder [ref. 1 page 41]

This article is about modelling: The process which gave birth to NLP back in the 1970’s. In the first half of the article we describe some of the fundamentals of modelling including: the difference between three types of modelling (Sensory, Conceptual and Symbolic); and three methods of modelling (First, Second and Third Position modelling).

The second half focusses on the most recent addition to the modelling toolbox, Symbolic Modelling. We describe what it involves, how it is being used in a therapeutic setting and a range of other applications. The information presented is an extension of our articles on Grovian Metaphor in Rapport 35 and 36 (ref. 2-3).

Before NLP

The importance of modelling to NLP can be discerned from the books that were written by Richard Bandler and John Grinder before the label ‘NLP’ was invented to describe the processes they were discovering. The first five books they published were all the result of modelling world famous psychotherapists (ref. 4- 8).

Then, in 1980 came NLP Volume 1 written with the help of Robert Dilts and Judith DeLozier (ref. 9). This was to be the first of a series of books on the modelling process itself. Volume 1 covered eliciting, designing, utilising and installing strategies. Unfortunately, Volume 2 never appeared. We find it intriguing that the last NLP book detailing the results from first-hand modelling of a single person was published in 1977 (ref. 8). Little wonder John Grinder has recently said:

My hope at the time was that there would arise a group of committed men and women who would recognise the meta level tools which we had either discovered or created, and go out and identify and create new models of excellence to offer the world. This has not happened and is very disappointing to me. NLP is popularly represented and commonly practised at least one logical level below what it was clearly understood to be at the time by Bandler and me. (ref. 1 page 40)

What is Modelling?

So what is this thing called modelling? Joseph O’Connor and John Seymour define modelling as “The process of discerning the sequence of ideas and behaviour that enable someone to accomplish a task.” (ref. 10, page 230). Robert Dilts has another definition, “The process of observing and mapping the successful behaviours of other people.” (ref. 11 Vol. 1, page 312)

If we combine these two definitions we can conclude that modelling is a process, i.e.. it is something that happens over a period of time and, at the very least, involves:

(a) Observing someone who is achieving something


(b) Establishing a map or sequence (a model) of what they are doing.

While this might be the bare bones of modelling, there is a lot more to it. To start with, there is more than one type of modelling. Second, there are a number of stages to the modelling process and third, a variety of skills are required to perform each stage. Some of the essential aspects of modelling are described below, albeit briefly.

To make life interesting you may note the word ‘model’ can be used three ways. There is ‘to model,’ which is the process defined above and for which we shall use the word ‘modelling.’ There is ‘a model’ who is the person from whom the information is being elicited. And finally there is ‘the model’ which is the end result of the modelling process.

Sensory Modelling

John and Richard examined the micro-behavioural and linguistic patterns of Fritz Perls, Milton Erickson, Virginia Satir and others in the hope of being able to reproduce what these wizards could do. They elicited a range of patterns which were the basis for the amazing results obtained by these therapists. John and Richard tested their codification of their model’s behaviour by using their newly acquired skills in real situations, and in so doing developed their ideas further. The results of these projects became known at the Meta Model, Representational Systems and the Milton Model.

Those modelled were later to remark that they were unaware of much of what John and Richard discovered they were doing! That is, they were unconsciously excellent. The next step for John and Richard was to develop these models so they could be passed on to others. Thus was devised the process for the ongoing dissemination of NLP.

The first five books were the result of studying people for observable behaviour patterns and the resulting models were descriptions of behaviour. For example, Richard and John suggested that people who learned to use the Meta Model would gain:

  1. A specific set of questioning techniques based on the client’s verbal communications
  2. How the use of particular non-verbal techniques may be indicated by verbal cues. (ref. 4 page 4)


We categorise this form of modelling as Sensory Modelling as it seeks to reproduce behaviours and language patterns that are directly perceivable through seeing, hearing and feeling.

Conceptual Modelling

Compare John and Richard’s approach to modelling with the one adopted by Robert Dilts to elicit the ‘Strategies of Genius’ of Aristotle, Sherlock Holmes, Walt Disney, Mozart, Einstein, Freud, da Vinci and Nikola Tesla. (ref. 11). What Robert identifies are mental strategies, in which:

You are specifically looking for a mental map that was used by the individual whom you are modelling in order to orchestrate or organise his or her activities to accomplish an effective result. (ref. 11 Vol. 1, page xxiv).

This is patently different from Sensory Modelling and thus requires a different approach. His modelling project does not specify linguistic and non-verbal behaviours. Instead it is a set of 20 conceptual rules such as:

  1. Develop special states and strategies for access to unconscious processes
  2. Encourage the use of ‘self-organising’ processes
  3. Acquire familiarity with necessary information through self-managed learning. (ref. 11 Vol. 3, page 395 )


Robert’s form of modelling can be called ‘conceptual modelling’ because the written narratives he used as input are mostly conceptual descriptions and his output was almost entirely couched in conceptual terms. It is clear that Robert Dilts is operating at a different level to that of Bandler and Grinder’s original work.

Another well-known conceptual model is that of the ‘Presuppositions of NLP.’ These are not a description of specific behaviours. They are principles and beliefs which guide behaviour. They were derived by implication rather than by direct observation. (ref 11)

Levels of Modelling

Dilts recognises that there are a number of different aspects or levels to modelling. He notes that:

We can look at where and when the person operated, the environment. We can examine what the person did, their behaviour. We may also look at the intellectual and cognitive strategies of the individual, their capabilities. We could further explore the beliefs and values that motivated and shaped their thinking strategies. We could look more deeply to the individual’s perception of the self or identity. We also might want to examine the way in which that identity manifested itself in relationship to the individual’s personal, social and ultimately spiritual larger systems. (ref. 11 Vol. 1, page xxvi)

Thus Robert Dilts uses his ‘Logical Levels of Experience’ delineation to categorise various aspects of modelling. While we recognise and value the differences between the six levels, we believe it is instructive also to recognise the similarities of the pairings: Environment with Behaviour, Capabilities with Beliefs and Identity with Spiritual.

The first two levels form a pair because they are observable through the five human senses of seeing, hearing, feeling, smelling and tasting; hence Sensory Modelling. The second two form a pair as they are mental processes which are perceivable only through inference and typically require conceptual descriptions; thus Conceptual Modelling. The final pair form a unity, as metaphor and symbol are the most common way of realising this experience; hence Symbolic Modelling.

Symbolic Modelling is a process for identifying how people represent their experience through metaphor and symbol. Before we explain more about this, let ‘s look at some of the fundamental skills required to enact the modelling process.

Figure 1 depicts the relationships between the Logical Levels, the types of modelling and some of the resultant models.

Metaphoric Modelling
Conceptual Modelling
Sensory Modelling
Figure 1: Relationship between Levels, Modelling and Models

Modelling The Modelling Process

Our purpose for defining three types of modelling is to distinguish between different approaches to model-building. However, the overall framework originally used by Bandler and Grinder is applicable to all types of modelling. This can be summarised, in its simplest form, as a five-stage process:

  1. Set outcome and identify models
  2. Gather information
  3. Construct model
  4. Test model by using it
  5. Modify model for transferring on (if required)


Although we describe the five stages as a linear process, it should be obvious that it is systemic as each stage feeds-forward to the following stages and feeds-back to the previous stages.

Each of the five stages requires a different set of skills. For example at Stage 2, a useful way to describe the skills required to gather information is by distinguishing between:

(A) Second Position Modelling
(B) Third Position Modelling
(C) First Position Modelling

(A): 2nd Position Modelling

One of the fundamental skills required at Stage 2, however the information is gathered, is the ability of the modeller to remain in a state of ‘not knowing’ for as long as possible. In other words, to not jump to premature conclusions which would inevitably be based on the modeller’s experience rather than what the other person was actually saying or doing. Judith DeLozier calls this the ‘nerk nerk’ state. (ref. 12)

In Second Position modelling, the modeller adopts the ‘nerk nerk’ state and accepts all sense impressions they receive without interpretation. This requires a significant conscious and unconscious commitment on behalf of the modeller.

Erickson’s Map

One way to undertake Second Position modelling is called ‘deep trance identification.’ It requires the modeller to set up appropriate ‘safety lines’ and ‘contextual markers’ so they can come back to themselves after each modelling session is complete. Charlotte Bretto says it took John Grinder six months to prepare himself sufficiently before he was ready to commit 100% of his neurology to ‘becoming Milton Erickson’. Until then he could not act with complete congruence.

An important distinction here is between ‘process’ and ‘content’. Modelling is about acquiring a working description of the map of the person being modelled not the person themselves – that is the territory. Apparently, one of the early NLP modellers took to a wheelchair in order to act like Erickson (who was partially paralysed due to Polio) until Erickson found out and reprimanded him. This is an example of confusing map and territory.

As an aside, we find it ironic that while most traditional education is about getting to a state of knowing as quickly as possible, the way to acquire expertise through modelling is to stay in the state of not-knowing for as long as possible. The only NLP book we know that addresses this skill directly is Turtles All The Way Down by Judith DeLozier and John Grinder (ref. 13).

Spiritual Healing

When James started learning the modelling process, he chose to model three spiritual healers “for the state they are in when they think healing is taking place.” Given that none of the healers used words during the healing sessions and none of them thought they were consciously controlling the process, this was a challenging first assignment! Much of the information required to build the model could only be accumulated via Second Positioning.

James discovered that in order to return to himself after ‘becoming each healer,’ he would have to have a strong sense of his own identity to come back to. At the time, a sense of his own identity was not available to his conscious mind. So he set out on a quest to ‘find’ his identity before he could fully commit to Second Positioning his models. This was quite a lengthy journey. What’s more, he discovered that each of these healers had a method for ‘setting aside the self’ so that they could become a conduit for ‘healing energy’. Thus no sooner had he discovered his identity than he had to learn to set it aside. He now sees this as ‘divine humour’!

A subtle distinction we make is between (a) going to Second Position with the person being modelled and what they are doing; and (b) going to Second Position with the information and the way the information is input, processed and output by the person being modelled. The first approach is typical of Sensory Modelling while the second approach is the preferred mode of Symbolic Modelling.

(B): 3rd Position Modelling

Third Position modelling occurs when the modeller takes a detached standpoint to observe and ask questions of the person being modelled. As a modeller is attempting to find out what happens, it is important that they can distinguish between what they are seeing and hearing, and their own interpretations. This is particularly true when asking questions, as all questions contain presuppositions and therefore influence.

The last thing you want is for what happens to change before you have modelled it! Especially as we have heard of modellers unwittingly “messing up” the productive strategy of their models with inappropriate questions.

While keeping the modeller’s presuppositions out of the process is central to all forms of modelling, it is paramount in Symbolic Modelling. Because of this, David Grove has developed a set of questions which contain minimal presupposition and are thus called Clean Language. The nine basic Clean Language questions were outlined in Rapport 35 (ref. 2).

Another form of third-position modelling was conducted by Robert Dilts when compiling his Strategies of Genius series. His only source of information was the written word in either autobiographical or biographical form. Therefore he had to work from historical descriptions about what these geniuses did. By necessity these accounts were once, twice and sometimes three times removed from the original behaviour.

(C): 1st Position Modelling

We are grateful to Michael Breen for pointing out to us that “All modelling is self modelling.” Thus what is being modelled is the modeller’s impression of the information gathered from studying the other person. In other words, we cannot get to someone else’s map without going through our own neurology.

A higher level description of the underlying patterns (i.e. a model) will start to emerge from the modeller’s unconscious mind when they have acquired sufficient sense-impressions. The process has now entered Stage 3 as the modeller is starting to use their own neurolinguistic systems to model themselves! We call this First Position modelling.

This is why it is so important in Stage 2 to remain in the ‘nerk nerk’ state for as long as possible. Otherwise the modeller ends up with a description of their own map! This may be valuable but there is a danger of believing “every magician I model thinks just like I do. Therefore I must be a magician!”

Four Steps to Learning

We recognise our description of modelling does not follow the traditional 4 steps of learning presented on many NLP programmes. These are:

(a) unconscious-incompetence
(b) conscious-incompetence
(c) conscious-competence
(d) unconscious-competence.

Our research shows that the modelling process (which is another name for learning) is more accurately represented by the sequence:

(a) unconscious-incompetence
(b) conscious-incompetence
(c) unconscious-competence
(d) conscious-competence.

The difference between the two formulations is that steps (c) and (d) have been transposed. In our experience of modelling, step (b) represents the ‘nerk-nerk’ state. As a result, the unconscious learns something new by step (c). Only now is it possible for the conscious mind to know. Thus step (d) denotes the meta-knowing of a true expert: one who consciously knows through experience.

Young children mostly learn by modelling. What separates their approach from that of adults is that they either miss out step (b) and go straight from unconscious-incompetence to unconscious-competence or they are completely comfortable with not knowing. Learning to welcome a state of conscious-incompetence as part of the modelling process is a stretching new idea for many people.

The Spelling Strategy

Having coded your model’s behaviour, the modelling process needs to move to stage 5 if you want to transfer the information to someone else. Your coding of the material may well have to be modified so that it can be learned by someone who was not involved in the original process. The NLP spelling strategy illustrates this perfectly.

Part of learning the spelling strategy is to spell the word being learned backwards (ref. 10, page 182). None of the original excellent spellers did this. So why is it is in the strategy?

When they first tried to teach the strategy to poor spellers, they found that even though they learned it, they did not believe this was enough to become a good speller. So someone had the bright idea of getting them to spell the words they were learning backwards on the basis that “If you can spell the word backwards, you know spelling it forwards will be easy.” So for the spelling strategy to be useful an extra ‘convincer’ step had to be added.



Relationship between Levels, Modelling and Models


Metaphoric Modelling
Conceptual Modelling
Sensory Modelling


Figure 1


Jubilee Clip

The following extract from a 4,000 word transcript illustrates how Symbolic Modelling works in practice. The client’s is a senior manager in a multi-national company. His outcome is, “Confidence in my own abilities.” His presenting problem is, “I got to senior level by accident and I now feel vulnerable.” We pick up the conversations half-way through the first session:

(C = client, T = therapist)
C: … An adult feeling of vulnerability.
The client’s conceptual description of the problem.
T: And an adult feeling of vulnerability. And is there anything else about that feeling of vulnerability?
Therapist repeats client’s words exactly to honour the experience. And asks for attributes of vulnerability.
C: Rapid, shallow breathing, and a tightness.
Client gives a sensory description.
T: And when rapid, shallow breathing, and a tightness, where could that tightness be?
Therapist asks for location (relative address) of the symptoms.
C: In the chest. In the upper chest.
T: And in the chest, in the upper chest. And when tightness in the upper chest, that’s tightness like what?
Therapist asks for a metaphorical description.
C: [Client grasps left forefinger with right hand] … It’s like a finger is grabbed and pressure is applied. … I have a picture … It’s of a jubilee clip tightening around a hose.
Now we have 2 symbols – a clip and a hose – and a relationship between them – tightening.
T: And a jubilee clip tightening around a hose. And where does that tightening around hose come from?
Asks for source of tightening.
C: A screwdriver.
Symbol number 3.
T: And a screwdriver. And what kind of screwdriver is that screwdriver which is tightening jubilee clip?
Asks for attributes of the symbol.
C: Flat-bladed with a yellow plastic handle.
Note the detail, a clue to the importance of this symbol.
T: And flat-bladed with a yellow plastic handle screwdriver. And what would screwdriver like to do?
Asks for desired outcome of the symbol.
C: Undo it. …
T: And undo it. And can screwdriver undo jubilee clip?
Testing power of symbol to achieve its outcome.
C: No. There’s a conflict.
The first description of a ‘relational bind’ which hold the unwanted pattern in place.
T: And there’s a conflict. And what kind of conflict could that conflict be when screwdriver wants to undo jubilee clip?
Asking for attributes of the relationship.
C: There’s a fear about undoing the clip … There’s an unknown risk.
The limiting (homeostatic) pattern begins to emerge. This is not just any risk, this is an ‘unknown risk’ which is producing a fear which is preventing the clip from being undone.

This small sample of the transcript demonstrates the value of Clean Language and how much quality information emerges in just a few questions. As a result the client was able to describe something that has limited him for over 30 years. What’s more, this piece of his jigsaw has dramatically come to life. The conclusion of this session is discussed below.

Pattern of Organisation

Once the patterns between and across symbols have emerged, the overall pattern of organisation of the relationships can be discovered and then described symbolically. This is necessary because, according to Fritjof Capra:

Patterns cannot be measured or weighed; they must be mapped. To understand a pattern, we must first map a configuration of relationships.(ref. 18, page 81)

In other words, there must be a context within which the pattern of organisation can emerge: a Metaphoric Landscape. For many of us, our patterns of organisation are so core and so familiar that they seem to be who we are!

Examples of unproductive (maladaptive) patterns of organisation we have modelled include:

  • Paradox
  • Dilemmas
  • Balancing
  • Conflicts
  • Permanent Incongruities
  • Double Binds
  • Circularities
  • Irrevocable Decisions
  • Predestined Consequences


You will notice that these are all conceptual descriptions which are, by definition, unsolvable. However, once ‘problematic’ patterns are represented symbolically they become amenable to resolution and evolution.

In summary, Symbolic Modelling uses Clean Language to identify three levels: Components, Relationships and Organisation. These are depicted in Figure 2 below and are defined as:

Components are the symbolic elements within the Metaphoric Landscape. They are the embodiment of the pattern of organisation resulting from the structural, process and functional relationships.

Relationships (among the symbols of the Metaphoric Landscape) determine the system’s Structure (form), Process (activity over time) and Function (attributes) that sustain and evolve the system.

Pattern of Organisation is the overall configuration of relationships (among the symbols of the Metaphoric Landscape) that determines the system’s essential characteristics – the pattern of the patterns.


Let’s take an example of the various levels of symbol for the client in the transcript above. The Symbolic components which originally comprised the client’s Metaphoric Landscape were: a boy, the look from his mother, an exam, a jubilee clip, a hose pipe and a screwdriver.

The key relationships for the client are: (a) the influence of the mother’s look on the boy when she found out he had failed his 11 plus exam; (b) the tightening of a jubilee clip around a hose by a screwdriver and (c) running round a track trying to overtake his ideal self – twice. Conceptually these patterns were described as ‘vulnerability,’ ‘helplessness’ and ‘no-win,’ respectively.

The symbol that emerges to represent the overall pattern of organisation is climbing a mountain that gets higher the more it is climbed. This metaphor operates at a meta-level and encapsulates the other three patterns. In addition to these (apparently) unproductive patterns the client’s Metaphoric Landscape also contained resource symbols; such as the Master from the TV programme Kung Fu and his red, mature heart.

When the pattern of organisation emerged the resource symbols in the Metaphoric Landscape could play their role. The end result was that this client realised: “For the last few years I’ve been asking myself, ‘Am I doing what I want to do or is it time to look for something new?’ But I kept getting blanks. Now I know what my mission is. … Helping others to undo their jubilee clips!”

This was a sacred moment. As so often happens, through the process of overcoming his own habitual responses he had found a way to serve others.

Other Applications

As well as applying Symbolic Modelling in a therapeutic context, we know of several people who are using it elsewhere. For example, it is being used in education by Caitlin Walker to help children identify their unproductive and productive learning strategies so they can transfer from one to the other.

Ana Robles uses student-generated metaphors in the teaching of English as a second language in five ways:

As a powerful tool to facilitate change; to link second language words with their inner representations; to glimpse into my student’s attitudes without their feeling threatened; to have fun; and to practise the four basic skills (reading, writing, speaking, listening). (ref. 20)

Rapport, Issue 37 includes an article by Simon Stanton explaining how he has tailored Symbolic Modelling to help people learn new computer skills (ref. 21).

Also in business, we use this approach extensively with our Executive Coaching clients. They are quick to make connections between their developing symbols and how they are responding to situations at work. One client who was on a final warning for his poor sales performance said “it is like I am operating with the hand brake on.” Once he had safely released the brake and developed his Metaphoric Landscape, within 6 months he became the second highest performing salesperson in the company.

We taught the rudiments of Symbolic Modelling to a group of nurses who specialise in treating patients with Multiple Sclerosis (MS). They found it difficult to relate to patients who used “weird ways of describing their symptoms.” For example, one nurse said, “What do you do when a patient says their MS feels like ‘cheese wire wrapped around my legs’ or ‘it’s like ants running all over my body’?” Using the Clean Language of David Grove they were able to build rapport, better relate to their patient’s symptoms and in so doing reassure the patients they were not going mad.

We also know of a medical practitioner who, when she has a spare 10 or 15 minutes with a patient, uses this approach to get a clearer description of their symptoms and explore how the patient perceives the process of healing.

Generative Modelling

We have had the privilege of working with clients who do not want to ‘solve problems’; instead they want to ‘find a purpose in my life’ or ‘have a clearer sense of my direction’ or ‘feel more connected with my higher self.’ These clients are not requesting remedial therapy, rather personal and spiritual development. In other words, they require a generative approach. Supporting clients to tap into their symbolic world is ideal for exploring and bringing into consciousness these ephemeral and yet vital parts of their experience. There they can uncover symbolic answers, questions, pointers, or whatever they need to discover. As James Hillman says in his latest book, The Soul’s Code:

What is lost in so many lives, and what must be recovered: [is] a sense of personal calling, that there is a reason I am alive.  (ref. 22, page 4)


In this article we have briefly described the NLP modelling process and shown how modelling client-generated metaphors enriches the methods traditionally used to encode excellence.

It is our belief that some human processes are either so complex or so core that everyday behaviour, words and concepts are insufficient descriptors. Thus the human mindbody has evolved the ability to process information in metaphor and symbol. This is not just another way of representing our experience, it is a fundamentally different way of making sense of the world.

Our perception of this sense-making is coded in symbolic representations which are a way of embodying abstract meaning. They give life to something that does not exist at the physical level of reality but only in the mindbody of their creator. In other words, they allow us to give emergent properties form.

Certain aspects of our subjective experience seem best suited to metaphoric and symbolic expression. To bring this type of experience to consciousness requires a method which is compatible with the nature of metaphor. This article was written as a step towards establishing such a methodology.


1 – “An Interview with Dr. John Grinder” in NLP World (Vol. 4 No 1, March 1997)

2 – Penny Tompkins and James Lawley, Less is More: The Art of Clean Language in Rapport (Issue 35, Autumn 1996)

3 – Penny Tompkins and James Lawley, “Meta, Milton and Metaphor: Models of Subject Experience” in Rapport (Issue 36, Summer 1997)

4 – Richard Bandler and John Grinder, The Structure of Magic Volume I (1975)

5 – Richard Bandler and John Grinder, Patterns of Milton H. Erickson Volume 1 (1975)

6 – John Grinder and Richard Bandler, The Structure of Magic Volume II (1976)

7 – Virginia Satir, John Grinder and Richard Bandler Changing With Families (1976)

8 – John Grinder, Judith DeLozier and Richard Bandler, Patterns of Milton H. Erickson Volume 2 (1977)

9 – Richard Bandler, John Grinder, Robert Dilts & Judith DeLozier, NLP Volume 1 (1980)

10 – Joseph O’Connor and John Seymour, Introducing NLP (1990)

11 – Robert Dilts, Strategies of Genius Volumes I, II and III (1994/1995)

12 – Judith DeLozier, “Mastery, New Coding and Systemic NLP” in NLP World (Vol. 2 No 1, March 1995)

13 – Judith DeLozier and John Grinder, Turtles All The Way Down (1987)

14 – Robert Dilts, “Darwin’s Thinking Path” in Anchor Point (Vol. 10 No. 11, Nov. 1996)

15 – Carl Jung, Man and His Symbols (1964)

16 – “And What Kind of a Man is David Grove?” in Rapport (Issue 33 Autumn 1996)

17 – Karl Pribram, “From Metaphors to Models” in Metaphors in the History of Psychology, edited by David Leary (1990)

18 – Fritjof Capra, The Web of Life (1996)

19 – Carl Jung, Memories, Dreams Reflections (1983)

20 – Ana Robles, “Metaphors in the Traditional Classroom” in SEAL Journal, Winter 1997/98

21 – Simon Stanton, “Using Metaphors in IT Training” in Rapport (Issue 37 Autumn 1997)

22 – James Hillman, The Soul’s Code (1996)

© 1997, Penny Tompkins & James Lawley

Print Friendly, PDF & Email
body * { color: inherit !important; }