First published in
ReSource Magazine Issue 8, April 2006
What is Therapeutic Modelling?
A dialogue with James Lawley and Penny Tompkins
This article has been written in dialogue format. Some
of these dialogues have actually occurred, although most of
the questions are composites of those we have been asked
over the years.
There has been an upsurge of interest in modelling recently.
I've been reading the debate stirred up by John Grinder, Carmen
Bostic St. Clair and Robert Dilts about different kinds of modelling
(Ref 1). I'm interested in understanding the difference between
modelling as it has been traditionally thought of in NLP, and what
you call 'therapeutic modelling'.
Penny: It seems appropriate that there should
be an upsurge of interest in modelling on the 30th anniversary of the
founding of NLP (Ref 2). Strangely, for a subject that prides itself
on making the structure of subjective experience explicit, much of
what is known about NLP modelling is implicit. John Grinder and
Richard Bandler hadn't ever written up how they did their early
modelling. James and I had to 'reverse engineer' how they did it and
then apply that to modelling David Grove.
In NLP, modelling is typically considered a 5-stage process
whereby a modeller first identifies an exemplar (a person, or people
who exemplify some desired behaviour or skill); the modeller then
gathers information about what the exemplar does; constructs a model
of how they do that; and tests whether using the model gets similar
results to the exemplar. The modeller then goes on to use the model
themselves, or facilitates others, acquirers, to learn how to apply
the model (Ref 3).
James: John McWhirter calls this "product modelling"
because the output is a physical product (representation) which if
followed should produce a specific result (Ref 4). In fact, NLP as
commonly taught is a 'product' of John and Richard's modelling - NLP
didn't even exist as a concept at the time that they were modelling
the original exemplars (Perls, Satir and Erickson). John and
Richard's first five books were the product of their modelling. Some,
but by no means all, NLP Master Practitioner programmes still train
product modelling in the form of a 'modelling project', and many use
David Gordon and Graham Dawes "Experiential Array" as the framework
for modelling (Ref 5).
This is obviously not what happens in therapy, so what is
Penny: Modelling in a therapeutic or any
facilitatory context (such as coaching) uses what we call
'modelling-in-the-moment' (or what Phil Swallow calls "modelling for
the moment"). In this kind of modelling the therapist does not
construct a formal model so there is no 'product' in the sense of a
physical representation. Sure the facilitator may take notes but these do not
constitute a consistent, coherent and complete model. The reason? The
therapist doesn't have time to do this. Psychotherapy is a dynamic
process. The client is always producing new information and the
therapist has to constantly update his or her model of the client's
model of the world.
James: Therapeutic modelling is a different kind of
modelling in part because it is used for a different purpose. The
whole point of Neurolinguistic Psychotherapy (NLPt) is for the client to
change their existing model of the world, and as soon as that
happens, any formal model the therapist had constructed would be out
In therapeutic modelling the therapist uses the client's patterns
of behaviour (movement of their body, vocal qualities and the words
they speak) to construct a model of the client's internal processes.
The therapist attempts to figure out how the structure of the
client's subjective experience so consistently gets them the results
they get. The therapist holds this model in their mind and body -
part consciously and part out-of-awareness. They continually update
this model and use it to decide how to respond to the client, where
to direct the client's attention, what question to ask, and how to
use their own body and voice for maximum therapeutic effect
If there is no 'product' to therapeutic modelling, then what
does it produce?
James: Therapeutic modelling does not produce a
physical 'product', it results in a dynamic 'process'; and more
explicitly, a series of interactions which aim to enable the client
to achieve (or be in a position to achieve) their desired outcome.
Ultimately the purpose of therapeutic modelling is for the client to
achieve a desired change. However change is an effect of therapeutic
modelling. The immediate 'output' of modelling-in-the-moment is the
behaviour of the therapist. The therapist has to gather information,
update their incomplete model, make decisions, and respond using
their model of the client's model - all within a few seconds!
Penny: Richard Bandler has said NLP is 95% information
gathering and 5% change work (Ref 11) Taking that literally means modelling
for 57 minutes of each therapeutic hour. We've found that with most
clients if we, and they, modelled their system for 57 minutes they
had already started to change. So we just kept modelling how they
James: And if they hadn't started changing, we modelled how
they were managing to stay the same.
So therapeutic modelling 'produces' the behaviour of the
therapist. If in product modelling there is an 'exemplar' and an
'acquirer', who plays those roles in therapeutic modelling?
Penny: We suggest that the client is playing
both roles simultaneously - and this is an ever-present conundrum of
psychotherapy. Think of it this way. Every client is the exemplar of
getting the (unwanted) outcomes they so consistently get. Each client
is excellent at their form of 'depression' or 'anxiety' or whatever.
At the same time clients want to acquire their own desired outcome.
They know what a happier or calmer life is like, they just don't know
how to be that kind of person. Acquiring is a different metaphor to
those commonly used in NLP which involve a formula (Present State + Resources = Desired State) or a
journey (PS --> DS). Both of these are useful, and I'm suggesting we
can also think of psychotherapy as a process of the client's system
learning from itself.
James: In other words the client is also a modeller. They
are modelling their own system - self-modelling - and as a therapist
I'm doing everything I can to facilitate them to self-model those
parts of their system that are most relevant to their desired
outcome. In this way a client doesn't just solve a problem, they
learn to develop.
Are there any other ways to distinguish between product and
Penny: I can think of two. A fundamental
difference is: Who owns the outcome? In product modelling it is the
modeller who decides what they want to model - that's their desired
outcome. The exemplar agrees to be part of the process but rarely has
any outcome beyond being helpful. Conversely, in therapeutic
modelling, it is the client who owns the outcome and the NLP
therapist agrees to set aside their own agenda to work towards the
A second difference is that I consider it unethical for an NLP
product modeller to have an intention for the exemplar's model to
change - while psychotherapists are paid to have just such an
James: And a third difference is that a product modeller
wants to take on their exemplar's states and strategies. However,
because clients are exemplars of doing behaviours which often cause
them great pain, a therapist has no desire to take on their clients'
states and strategies for depression, addiction, etc. A therapeutic
modeller needs to be careful to model the structure of the client's
experience, but not acquire it for themselves.
If I'm matching the client's body movement, voice qualities and
words, how do I model and not take on the client's 'bad' states?
Penny: You seem to be describing what is
commonly known as 'second position modelling'. And in therapy it can
be used to get an embodied sense of a client's experience. However if
second position modelling is overused it can have an adverse effect
on the therapist. Modelling a client's "lost and confused" state is
one thing, being lost and confused along with your client will likely
reduce your effectiveness. There are other ways to model which do not
have the side-effect about which you are rightly concerned.
James: For example, if you watch old video tapes of Richard
Bandler working with a client, you'll see he is often not body
matching, although he is definitely modelling. Similarly, David
Grove, one of the finest modellers I've ever seen, almost never body
matches but he still has exquisite rapport and modelling skills. When
I first worked with him as a client, he seemed to understand me so
well that I wondered if he was in my mind with me!
Penny: Apart from second position modelling you can model
from a purely "third" or "observer position". Or you can model by
moving your attention in and out of each of the three perceptual
positions. And there is also a place from which to model that
we would characterise as neither first, second nor third position. It
can be considered an amalgam of all three or, as we prefer to
envision it, a middle position between all three.
[NOTE ADDED 2012: See our article, Pointing to a New Modelling Perspective]
James: One particularly effective way to therapeutically
model and not 'take on' the client's stuff is to construct your model
around the client's body and in their perceptual space. This way of
modelling has a number of benefits. First it will keep your attention
firmly on the client. Second, it makes it easier to keep a clear
distinction between the client's information and your inferences.
Third, when a client walks out of your office at the end of a
session, they 'take away' their model and you can start fresh with
the next client.
You imply there is more than one way to do therapeutic
modelling. Can you say more about this?
Penny: Yes, there are lots of ways. As an
example, let's look at the difference between 'top-down' and
'bottom-up' modelling. In top-down modelling you start with a process
or structure for organising the client's information and you fit your
observations into that pre-chosen model. Although the content will be
new, it is organised in a predetermined way. For example, using the
Myers-Briggs categories to model is extremely top-down. Regardless of
the individual, we know their answers will fit somewhere on the MBTI
pre-given scale - the only question is where.
Many NLP therapists use what we call 'middle-down' modelling. For
example, they adapt Robert Dilts and Todd Epstein's timeline,
perceptual positions, neurological levels frameworks for modelling
the client's patterns (Ref 6). These models are closer to clients'
natural descriptions than MBTI, astrological houses or Jung's
archetypes. However, since they do not start with an individual
client's verbal and nonverbal behaviour they may miss the
idiosyncratic aspect of that client's particular experience.
James: When a therapist is bottom-up modelling they create
a model out of their observations of what is presented by the client.
The resultant model emerges out of the patterns and logic in the
client's information and stays close to the client's description.
This means that in the beginning, the modeller has no idea where they
are going or what the model will look like when they have
finished. They do not start with a preconception that the
client organises their experience into a 'timeline' or 'perceptual
positions' or the 'neurological levels' etc. Although the therapist
is likely to presume that each client has some way to organise time,
perspective and hierarchy, they will not model these unless the
client's information indicates they are relevant to their desired
Penny: Almost all the great psychotherapists say the same
thing: You have to start afresh with every new client. Well that's
what bottom-up modelling does. Bottom-up therapeutic modelling starts
from the premise that every client constructs their 'reality' in a
unique way. Their logic, motivation, desires, values all combine in
an idiosyncratic complex. Bottom-up modelling is a way to identify
the patterns, structures and organisation that means that client is
sitting in front of you with that issue and that unfulfilled desire
Are you saying an NLP therapist should only use bottom-up
James: No, we're not saying bottom-up modelling
is always the best way to model, even in a therapeutic setting. We
are saying that an experienced Neurolinguistic Psychotherapist should
have a choice whether to use this approach or not. To have that
choice the therapist needs to be able to hold off long enough (from
jumping to a conclusion, using a technique, etc.) to allow patterns
to emerge from observing the client's verbal and nonverbal behaviour.
If I'm modelling, be it top-down or bottom-up, what happens to
all the NLP techniques I've learned?
Penny: There are times when a technique works
just fine and there are circumstances when it can be
counterproductive. The advantage of a well-researched technique is
that it's a shortcut that can be effective most of the time for most
people. Generally, the more specific and standard the problem, the
more likely a technique will be useful. Of course you have to decide
which technique to use and then you're back to the need for
modelling! A client once described to me how a previous NLP therapist
had tried to use submodality techniques to help rid her of her
feelings of worthlessness. She said that although she hated those
feelings, as soon as the therapist started to try to get rid of them,
"I just played along and went though the motions." Why? Because those
feelings were all she had, and if she lost them, she didn't know what
would be left of her. If the therapist had spent more time modelling
the structure of this particular client's subjective experience, they
might have thought twice about rushing in with a standard technique.
James: Learning a technique is a good place to start as
long as you realise it's not the destination. On most NLP training
courses people only ever see demonstrations of techniques. However,
if you ask an experienced NLP therapist about what happens in a
'live' session with a 'real' client they'll tell you it's not like a
Dreyfus and Dreyfus studied (I'd say modelled) the developmental
path of many different types of professionals - from chess players to
nurses (Ref 7). They found that as an individual progresses from
being a "novice" to becoming an "expert", the more they progress from
using rule-based techniques, through top-down modelling, to
modelling. By Dreyfus and Dreyfus' definition, being an expert is not
about just having more knowledge and more techniques, it's about
having a wider and deeper, more systemic perspective based on many
years of experience. This is what enables experts to have those
apparently magical intuitions.
The aim, as Ken Wilber would say, is to "transcend and include"
techniques. Every technique I've ever learned sits in the background
of my knowledge and informs the models I am constructing. Even after
15 years of practising Neurolinguistic Psychotherapy I am still
amazed how, if given the chance, clients devise strategies and
solutions for themselves that never occurred to me and which have an
idiosyncratic creativity that perfectly fits their way of being.
The Dreyfus and Dreyfus 'Novice to Expert' Model
I realise you are known for introducing Symbolic Modelling into
NLP, how does it fit in?
Penny: Symbolic Modelling is primarily a
bottom-up modelling methodology which has applications in
psychotherapy, education, health, organisations, etc. (Ref 8).
Symbolic Modelling makes explicit several additional ways to model.
For example, the facilitator or therapist models the client's
internal symbolic map, or metaphor landscape as David Grove called
it. To do this the therapist makes use of the client's exact verbal
and nonverbal metaphors as the raw material for the modelling
process. As James said earlier, a client can be facilitated to
SELF-model, and Symbolic Modelling encourages this to happen.
NLP has been defined as "The study of the structure of subjective
experience". Symbolic Modelling fits this definition perfectly. There
is nothing more subjective than a client's metaphor landscape - and
it definitely has a structure! In fact, the more you study subjective
experience, the more you realise just how metaphorically structured
Over the last 10 years, we have found that Symbolic Modelling is
especially suited to working with 'higher levels' of experience -
core beliefs, identity, sense of purpose, the spiritual - as well as
complex and seemingly intractable issues, binds and double binds,
that are not amenable to traditional techniques.
Well, I certainly have a lot more to think about, but are all
these distinctions really necessary?
James: I believe that many NLP therapists are
already modelling, it's just that they have never thought to code
their work in this way. Being clear about these distinctions has
helped me refine my understanding of modelling in a therapeutic
setting so that I am more flexible and responsive to my clients.
For me the distinguishing features of Neurolinguistic
Psychotherapy are the combination of our experiential constructivist
philosophy and the unique way we model verbal and nonverbal behaviour
as the basis for understanding the structure of clients' subjective
experience. For many years The Neurolinguistic Psychotherapy and
Counselling Association (NLPtCA) has had a policy that NLP therapists
who want to be registered with The United Kingdom Council for
Psychotherapy (UKCP) must demonstrate:
"a practical integrated knowledge and experience of
the application of NLP as a methodology of behavioural modelling in
their ongoing supervised practice of Neurolinguistic Psychotherapy
(NLPt)." (Ref 9)
Penny: I take "behavioural modelling" to mean that we use
the patterns of the client's external verbal and nonverbal behaviours
to infer how they organise their internal map of the world - both
conscious and unconscious.
To sum up, therapeutic modellers start from the NLP presupposition
that "People already have (or potentially have) all of the resources
they need to act effectively" (Ref 10), and continue with an attitude
of: 'We may not know where those resources are, how they are
organised, why they are not being used or how they can be activated -
but let's find out.'
© 2006 Penny Tompkins and James Lawley
Discussion Section - "A Proposed Distinction for NLP".
According to Robert Dilts and Judith DeLozier in the
Encyclopedia of Systemic NLP and NLP New Coding, page
849, NLP was "founded by Richard Bandler and John Grinder in
See "Introducing Modelling to Organisations" in
Rapport, Issue 40, 1998 and "How To Do a Modelling Project" both available at:
John McWhirter, "Re-modelling NLP: Part 14: Re-Modelling
Modelling" in Rapport, Issue 59, 2002 reproduced at
David Gordon and Graham Dawes, Expanding your World:
Modeling the Structure of Experience, 2005
Robert Dilts, Modeling with NLP, 1998
Hubert L. Dreyfus and Stuart E. Dreyfus, Mind over
James Lawley and Penny Tompkins, Metaphors in Mind:
Transformation through Symbolic Modelling, 2000
NLPtCA Policy and Procedure for Accreditation and
Re-Accreditation of Psychotherapists, 2004
Robert Dilts, Strategies of Genius, Vol. I, 1995
Quoted by John McWhirter, 'Re-Modelling NLP Part One: Models and Modelling', Rapport 43, Spring 1999, download from: www.sensorysystems.co.uk/RemodellingNLPPart1.pdf
James is the lead trainer for the new two-year
Neurolinguistic Psychotherapy Diploma, sponsored by the Northern
School of NLP (www.nlpand.co.uk). The programme uses
therapeutic modelling as its basic methodology and prepares people to apply for accreditation with NLPtCA and registration UKCP.
Penny and James also train the advanced modules with Wendy Sullivan on the Clean Change Company's eight-module Clean
Language and Symbolic Modelling certified programme