Article Categories
[ Show ] All [ Hide ]
Clean Language
Article Selections
[ Show ] All [ Hide ]
5. Third iteration: Modelling from recordings

Having observed Robert Dilts do what he does, and having used Symbolic Modelling to interview him about how he does it, our modelling went through a third iteration. We used the source material we had gathered during the Northern School workshop to complete our modelling and produce a formal model. This required four steps:

We watched the video of our interview with Robert and reviewed what we had identified during the first and second iterations.

We produced a verbatim transcript (including notable nonverbals) and went through it several times filtering for different kinds of information, e.g. metaphors; internal and external behaviour; desired outcomes; evidence criteria, etc. These were marked on the transcript using different coloured highlight pens.

We organised the patterns identified in the transcript by:
  • Removing our questions and some duplication
  • Removing irrelevant and less-significant information
  • Bringing together (cut & paste) and organising the information relating to each of Robert's four phases
  • Identifying chunks of process within each phase and organising information into a logical sequence.
[You can see an example of this step in the Appendix, Section 12.]

We transferred information into location and systemic diagrams, continuing to prune and organise the remaining essential elements into a coherent whole, presented below.

Our Model for 'Selecting what is Essential'

We started with an outcome: To model how Robert Dilts 'selects what is essential' when he is modelling. We ended with a model that has four phases. Phases I-III occur while interviewing the exemplar, Phase IV happens after:

I.    Select what is significant
II.   Fit parts together
III.  Create associated/dissociated movie
IV.   Arrange what is essential into a model

At first sight it might appear that we accomplished our aim with Phase I and that we could have stopped there. However a number of factors meant we continued:
  • As a result of modelling Robert we now make a distinction between 'selecting what is significant' and 'selecting what is essential'. 'Significance' is required for something to be 'essential', but it is not the whole story. Essential only selects those significant events that constitute a minimum requirement without which the process would not work. Essential means no frills, even if those frills add value. Co-opting Robert's language, 'essential' is both "necessary and sufficient" – but no more.
  • Robert's selecting process does not just happen at the beginning. He is continually selecting and testing his selections with each new piece of information supplied, and with the congruence of his own imaginings. In each phase he selects for different things:
Phase I   - Selects individual parts
Phase II  - Selects pairs that fit together
Phase III - Selects coherent chunks that meet a high-level goal
Phase IV - Selects a minimal description from Phases I, II and III.
  • Also, given that Robert is one of the most prolific modellers in the field of NLP and this might be our only chance to interview him, we wanted to grab the opportunity with both hands and get as much as we could.
Our models of Phases I, II and III each contain two diagrams: Location and Process. These are two descriptions of similar information and are best 'read' together since the where and the how of experience are complementary.

It should be clear from the video clips that modelling for Robert is highly embodied. This is indicated by both his gestures and his embodied metaphors. The 'location diagram' shows where Robert experiences the various types of information in his internal perceptual space. Some of this is inside his physical body and some is in his mind-space but outside his body (see our article, When Where Matters).

The 'process diagram' depicts a systemic, circular chain of events. We decided to represent Robert's processes with systems diagrams to show his use of iterative feedback loops. An iteration applies the same process to the output of the previous iteration, over and over. In Robert's case he uses an iterative process for gathering, selecting, marking, fitting and organising information. Phases I, II and III are not only sequential, they are also cumulative. They enable Robert to home in on what is essential and settle on a model that he knows he can enact.

Our model of Phase IV is incomplete since we didn't observe Robert constructing his final model and it would have required more interview time to get to the internal processes behind his conceptual labels. Having said that, we included what we did get since it contains some useful information and gives a sense of Robert's modelling process from beginning to end.

To save repeating ourselves, the annotation accompanying each diagram assumes you have read our notes accompanying the video clips in Section 4.


So far, to take on Robert-as-modeller you copied him by saying and doing what he actually said and did while interviewing Martin. Then you imagined you were saying and doing what he described he was saying and doing while he was an exemplar being interviewed. Now if you want to you can take on our model.

To take on Robert's strategies you need to have a sense of how he makes use of his "somatic mind" and "cognitive mind". The location and process diagrams below will help you do this. Remember the phases are cumulative: when you transition to Phase II you are still running Phase I; and in Phase III you are running both Phases I and II, just more in the background.

You may find it helpful to have someone assist you to acquire the following models. Start by using the Phase I location diagram to help you associate into Robert's interior perspective. Then your assistant can step you round the process diagram. You will need to go round the loop several times until you sense a signal to move on to the next Phase. Your aim is, without knowing why, to do it Robert's way as best you can and to notice what happens. Your assistant's job will be to keep to the model and to not add in their own words, ideas or suggestions.


Consider this ...

When Mozart composed music he said that notes would come to him and sometimes he would get a feeling from the tone, and if he got the feeling he would hum those notes. Of all the notes that were coming the ones he would hum were the ones marked as significant. That’s how he selected notes. Mozart would put the notes he had marked into his bag of memory. Later he would pull them back out to test them. Are they still there? Are they still significant? Do they feel resonant? Sometimes they would feel more resonant than before.

Then, there’s a phase where things start fitting. In the beginning the notes are just coming and some of them he hums, and after he marks enough of them they start arranging themselves. Mozart was constantly looking for two notes that loved each other. Do these two notes love each other? Yes, all right, we’ll take those. Do these two notes love each other? Yes, we’ll take those. Do these two? OK. Then he'd wonder: How do all of these notes fit together? That’s different from selecting the notes and it starts to involve more of a cognitive mind because now he’s organizing them.

When he had collected enough notes, all of a sudden he would start to apply rules of point and counterpoint. What notes am I going to use? He didn’t apply these rules at the beginning. Not until he’d got enough would he go: that’s got to go there and that’s got to go there. Sometimes it would feel like he’d have to change something to really capture it. That’s even more of a cognitive process. Of course Mozart was trained in the structure of music but he also had intuitions about the basic feel of music.

And finally, once Mozart got the sound organised he would say we’re going to use this instrument to play that sound.

Phase I – Select what is significant

Our two diagrams in Phase I depict where and how Robert selects, marks and tests what is significant.

Phase I provides the building blocks which get organised in Phases II, III and IV. Phase I happens continuously throughout the modelling because the exemplar might add something significant at any moment. Similarly, regularly testing against an internal criteria of significance is necessary because the model you are creating will continuously evolve with each updating. You can consider the repeating, backtracking and testing as a form of quality control. If something is still registering as significant after several testings the more likely it is an essential part of the exemplar's process.

One of the challenges of taking on this model is getting a sense of what Robert is selecting for. He emphasised several times during the interview that although he is selecting, marking and testing words used by the exemplar (i.e. content), these are only labels or cues for bits of process or structure. The art is to remember to stay at this level.

We have already commented on the three-step nature of selecting what's significant: attending to center, noticing when a feeling of significance is activated, then marking the triggering content in memory and making it part of you. The process diagram shows how systemic this is. It also shows the importance backtracking plays in keeping the process going.

The feeling of what is significant – the radar signal – guides the direction of your questions. This, along with your goal for modelling clearly in mind throughout, means information gathering is not a random search. Although you cannot know in advance where you are going, the guidance system gives directionality to the search and increases your hit rate when sorting the wheat from the chaff.

Phase II – Fit parts together

The process Robert uses to fit together what he is marking as significant is depicted in our two Phase II diagrams.

Tompkins & Lawley's model of 
Robert Dilts modelling - Phase II Process diagram

While Phase II "is a different information gathering process" to Phase I, you can use a similar three-step strategy:

- Attend to your center
- Notice when there is a resonance between things marked as significant
- Capture those parts that fit together in a picture.

Robert describes how things "start to arrange themselves" and "a field is created". It seems he doesn't use his "cognitive mind" until the out-of-awareness arranging has progressed enough that it is ready to become conscious. Once this happens he starts to "explore a direction" which, we'd guess, enables him to "fill in" more and to test the robustness of the fits in another iterative loop.

Notice how the metaphors of "radar" and "guided" from Phase I, and "direction" and "explore" from Phase II work together as a coherent method of mapping a new territory.

Phase III – Create associated/dissociated movie

Once Robert has identified some of the significant parts of the exemplar's process and started to fit them together he can transition to arranging everything into a movie.

Tompkins & Lawley's model of Robert Dilts - Phase III  location diagram

We are not sure if in Phase II Robert creates one picture which contains all the significant parts that fit together or he creates a number of pictures – a "storyboard" – which naturally become a movie when there are enough frames. Either way, then he can "step in" to the exemplar in the movie.

In Section 4 we commented extensively on Robert's ability to associate into and dissociate from the internal movie he creates. By creating an inner movie and associating into the position of the exemplar in the situation where they apply their skills, Robert is "installing" the exemplar's process into himself. You will note that in Phase I the significant parts were already "becoming part of me". We suspect that when pairs of significant things are "registered" in Phase II that too has the effect of installing them. If so, installation is another aspect of Robert's strategy that happens throughout his modelling.

Although Robert mentioned "exploring a direction" in relation to Phase II, from observing him it is clear that in Phase III his questions have a definite directionality too. It seems that when bits are missing or feel vague he pursues a line of questioning around that chunk of the exemplar's process. He continues to "figure out" and "fill in" until a feeling of congruence lets him know he can do it. Then he's done.

Thus Phase III is essentially an extension of Phase II with three additions: sequencing of events; an associated/dissociated perspective; and an exit strategy.

Phase IV – Arranging what is essential into a model

Our model of Phase IV is less complete than Phases I, II and III because we didn't observe Robert produce his model and we didn't have much time to explore what he does internally during that process. This is why our description is more conceptual than the previous three phases. However we thought what we did get was useful and below we present an outline of a model.

Just as at the beginning of modelling, in Phase IV Robert is strongly focussed on his outcome: To organise what is significant in a way that is meaningful and useful:

Meaningful has to do with deeper desires (for you and the exemplar).
It is felt in the center
Useful is facilitating the meaningful to happen
Arranged outside (physically and perceptually)

As with previous Phases, Phase IV starts outside of Robert's awareness. We think it is highly likely that Phase IV processes have been operating in the background during the earlier phases. Robert becomes the recipient of this knowledge when there are enough significant things gathered and fitted together (in Phases I, II and III) that they start to fit into coherent structures; they start fitting together as a unit.

In locational terms, Robert arranges the parts of his model on a mental "workbench" in front of him, whereas his knowing that he has identified a deep structure is a felt-sense inside his body. As in the other phases his "cognitive" and "somatic" minds work in tandem.

In the most general of process terms, to construct a formal model once he has interviewed the exemplar Robert reviews his Phase III movie and his written notes, and:

Starts with general connections
Applies known rules/structures to fit things together as a unit
Gets detail about activities.

Given that Robert uses iteration in Phases I, II and III we can be reasonably sure that he does the same in Phase IV. If so these processes are not to be seen as linear procedure but more as a systemic wheels within wheels process.

Let's take a look at these processes in a little more detail:

Start with general connections
Identify connections – this is about that
Relate things by kinds of information, e.g. goals, activities
Find links between significant things
Use visual and auditory perspectives to find:
  • Nice fits – what goes with what
  • Words that fit together form themselves into where they belong in a process
  • Clusters of words which are clues to deeper structure
  • What makes sense
Feel inside for connections and relationships
(Remember, words are surface structure. They are cues/labels about a deeper process.)

Apply known rules/structures to fit things together as a unit
It is like using a workbench
Process rules
Cognitive structures, e.g. TOTE
Feel the deep structure of the process so that it flows through the whole model.

Get details about activities
Identify details of how to do each activity
Ask yourself: What is each activity trying to make happen?

When you look at Robert's model of Martin in the Appendix you will see that it has been structured in a way that is congruent with the above:

General comments on Phases I, II, III and IV

Mozart analogy
After the interview Robert said the most surprising thing to him was how much he referred to Mozart for analogies of his modelling process. These helped Robert explain what he does to himself as much as to us. We did not specifically reference Mozart in our diagrams. Instead we attempted to retain the value of analogy by putting all of Robert's references to Mozart into one preparatory story that replicates the four phases of his modelling.

Questions Robert asks himself
Central to Robert's methodology are the questions he asks himself. We counted about 40 in the transcript of our interview – that's over one per minute. Not only is the frequency important, so is the quality of the questions. They are 'pure' modelling questions which neatly dovetail with his outcome orientation.

We recommend you read through the Appendix, Section 12 and pick out the questions Robert asks himself. (We have made this easy for you by indenting them and putting them in italics.) When you look for the pattern in these questions (and the questions he imagines Mozart asks himself) you will notice that they are remarkably 'clean'. That is, they are short, to the point and they only ask for information about what is happening with minimal presupposition outside of the context. To answer his own questions Robert has to keep modelling. Searching for the answer to each question naturally takes him towards his outcome.

Bottom-up (parts to whole) and Top-down (whole to parts)
Robert is modelling bottom-up from specific examples provided by the exemplar to create a top-down model for an acquirer (see our article, Modelling Top-down and Bottom-up):

In Phase I bits of process are selected; in Phase II he finds pairs that fit together; and in Phase III he is looking for how they fit together into a movie which can be represented as a unit in Phase IV. Each phase, to use Ken Wilbur's term, "transcends and includes" the previous phases.

In Phase IV however Robert does something different. He finds general connections, applies known rules and then identifies the detailed 'how to' of each activity. This culminates in a physical representation of his model. So Phase IV is more of a top-down methodology. It starts at a high level and works its way down to a specific representation.

We characterise this bottom-up and then top-down process in the following diagram.

Levels of "fit"
The transcript and video clips demonstrate another common phenomenon in modelling:
homonymy – when the same word means different things. During the interview Robert uses the word "fit" 20 times, but not always in the same way. It took a diligent analysis to differentiate the meanings. We think there is a difference between the "fit" in Phase II, the "fit" in Phase III, and the "fit" in Phase IV. "Fit" in Phase II and Phase III means fitting parts together and then fitting the fitted parts together. These form the basis for the "fit" in Phase IV which is at least one logical level higher, at the whole-model level:

Phase II fits parts together
Phase III fits those fitted parts into a movie
Phase IV ensures a fit with the structure of the whole model.

In this way Robert covers several fundamental ways of organising experience: functional relationship between attributes; temporal relationship between events; and part-whole relationships.

Agent and recipient
While modelling is an active process requiring a large degree of agency on behalf of the modeller, much of what Robert does has a more passive 'it's just happening' feel to it. It appears Robert is as much a recipient of signals as he is an active agent. He experiences these signals or cues as feelings, embodied fits, intuitions, thinking "that's it", and congruence. These are not emotions; rather they are felt-senses or kinesthetic representations or embodied knowledge.

On one hand, Robert leaves the identification of what is significant to the activation of his radar signal and is guided by that feeling. Then things fit together and at some point start to arrange themselves. This creates a field and the words start to form themselves into where they belong in the structure. On the other hand, Robert actively gets involved in setting his goals and intent, listening to his center, and doing external behaviours such as note taking, backtracking and asking questions. Later he actively tries to capture what has been brought to his attention first in a picture, and then in a movie – if necessary filling in when something feels vague or is missing.

By the time he gets to Phase IV, Robert is mostly an active agent: relating things; finding links; using visual and auditory perspectives; using a workbench; and applying rules, cognitive structures, principles, training and experience.

This dual agent/recipient function can be seen by an analysis of Robert's metaphors:

Active Agent
Passive Recipient
I’ve got to find out what’s essential to create something [I'm] guided by the feeling of what’s essential
I listen a lot to my center In significant times the center becomes activated
I’m backtracking and pulling out those things
that have been marked
It’s like a radar signal that goes beep, beep, beep
Looking for what’s useful and what’s meaningful The radar is going to go ‘this thing is significant'
I’m trying to get a picture There’s some kind of a field created by these different things
I’m trying to fill in Do these two notes love each other?
To really capture what it is They start arranging themselves
I’m trying to construct a movie A phase where things start fitting
I’m already installing it Lots of data that comes
[I] figure out, where does this movie stop?
Where does it feel vague?
Words start forming themselves into where they belong
I’m going to register that Something will register
Mozart would start to apply rules Things start to fit into a structure

In a parallel process the relationship between Robert's "somatic mind" and his "cognitive mind" changes during the modelling. In Phase I, Robert's modelling mainly involves somatic mind with little or no cognitive mind. In Phase II and III there is a "interplay" between somatic and cognitive minds. By the time he gets to Phase IV his process is mostly cognitive, organising what he has previously identified somatically.


Now you have an overall sense of what Robert does you can rehearse his behaviours with more of a context for how they work together to produce excellent modelling. You can also return to Sections 3 and 4 (and the transcripts in the Appendix) and review the information more cognitively. Since we do not have a complete model of Robert’s method of modelling you will have to fill in bits – just like he does!

After you have acquired a model as close as you can to the way Robert does it, you can consider adjusting some of the elements to make better use of your existing resources. For example, your significance signal might be located somewhere else and it may not be like the beep, beep, beep, radar signal Robert uses. You can substitute your own location and metaphor as long as it has enough of the same characteristics that it performs the same function as in the model.

What now?

We have completed the tour through our model of Robert modelling. That journey involved three iterations, each one visiting the same material with fresh eyes but with an accumulated knowledge. Next we turn the spotlight on our methodology.

 »  Home  »  Symbolic Modelling  »  SyM and NLP  »  Modelling Robert Dilts Modelling
 »  Home  »  Modelling  »  How to Model  »  Modelling Robert Dilts Modelling
Article Options
a new book
1 Aug 2017

Special launch price
free reference card with every order

Insights in Space:
How to use Clean Space to solve problems, generate ideas and spark creativity

James Lawley & Marian Way

order here
view all featured events