How to do a Modelling Project - Section 6
Stage 3: Constructing Your Model
When modelling multiple exemplars for a class of experience, one
process for constructing your general model is to:
1. Describe from their perspective and in their
words how each exemplar does what they do
to get the required results; i.e. construct models for each exemplar using their descriptions.
2. Evaluate each model for:
Completeness - It has all necessary
distinctions/components. It answers 'what else?' questions with "nothing".
Coherency - The relationships between components adhere to
an internal logic. It answers 'why?' questions from within its own logic.
Consistency - It will get similar results even when circumstances change. It can answer 'what if?' questions.
* You can evaluate the above by the degree to which your model
shows 'operational closure'–
- When no new components or patterns emerge and the exemplar's
descriptions add no further information about how that operational
- When new components or examples continue to appear but they
are isomorphic (have the same function or organisation) as
previously identified patterns.
- When the logic of the exemplar's description encompasses an
entire configuration, a complete sequence or a coherent set of
premises (with no logical gaps).
- When the model enables you to predict ways of dealing with
unexpected situations, difficulties, interference or distractions
that have yet to be mentioned by the exemplar.
- When you repeat or demonstrate the operational unit to the
exemplar, and they acknowledge 'that's it, you got it'.
3. Compare and contrast individual models
component-by-component, step-by-step and function-by-function.
At this point you start to separate the information gathered from the
exemplar: It is no longer their model, it becomes your model because
you will represent the information in a different way to meet your modelling outomes.
4. Design your own model using one or more of the following
a. Identify similarities across exemplars and
construct a composite model based on similarities.
b. Use one of the models as a prototype and improve it by
adding/substituting distinctions/components/steps from the other
c. Deconstruct the individual models into the function of each
component/stage and construct a new model from the bottom-up.
d. Adapt existing models from other contexts that are
compatible with the model you are constructing, and use them as the
framework for your model (e.g. 'transformational grammar' was the
basis for the Meta Model, and 'self-organising systems theory' formed
the framework for Symbolic Modelling).
5. Evaluate and improve your model based on the degree to
which it is:
Effective - It gets similar results to the
Efficient - It requires the least number of
steps/components (use Occam's Razor to make it "as simple as possible, but no
Elegant - It is code congruent, i.e. the content of the model, the manner in which it is presented/coded and the means of getting the results are congruent.
Ethical - The effects are aligned with your and others' existing or desired values.
[NOTE: We borrowed the first three E's from John McWhirter and added the fourth ourselves]
And, evaluate whether distinctions/components are necessary by the
degree to which each is:
Effective - contributes to the overall outcome
of the model.
Efficient - serves multiple functions.
Elegant - fits into the overall coherency (internal code
congruency) and enhances the consistency (external code congruency)
of the model. It is compatible and aligned with:
The exemplars (Stage 2)
Itself (Stage 3)
The context where it will be tested (Stage 4)
The acquirers (Stage 5)
6. Test, get feedback, adjust model; test again, get
feedback, adjust; etc. ...
More on Model Construction
Exemplar's cannot not do their patterns of excellence. A key aspect of modelling is to determine how an
exemplar keeps achieving the same results even under changing circumstances. How is it that they cannot
not do it? How come they don't forget to do it? How do they adjust
for unfavourable circumstances and still get consistently
excellent results? In other words, how come it's habitual? This
information will not be in any of the components, but in the pattern
of relationships between perceptual components. It will likely be the
circular chains of relationships (Bateson) that keep the pattern
repeating. And if so, your model needs to have comparable circular chains.
You can consider 'Is there any way I can I run
this model and do something else?' and 'Under what circumstances
would I not get the required results?'. Adapting your model to
take these circumstances into account will make it more robust, and more
Except when, under inappropriate or extreme conditions, the pattern breaks down. At these times a values or ethical threshold is often involved. What are those conditions and
what do exemplars do then? Bateson warned that any behaviour taken to extremes will become toxic. What are you own values and ethical limits with regard to using this model? These are non-trivial questions that we believe need to be openly and honestly faced.