That’s a nice distinction. I think we could further divide it into the:

1. Conditions which are being assessed (I assume this is what you mean by the ‘target’).
2. Criteria.
3. Process of comparing #2 with #1.
4. Representation of the result of #3.

On top of that there is another assessment process, the calibration of whether the result is ‘good enough’, ‘frequent enough’, etc.

To take an example:

1. A facilitator engaged in a Clean certification (actually, the assessor’s observation of the facilitator’s behaviour).
2. Published criteria.
3. Assessor’s comparison of that facilitator’s behaviour in relation to the criteria.
4. Assessors representations of the results of #3 e.g. internal (VAK) and external (ticks on a sheet, verbal and written feedback).

After all this, there is a still a process to go through to say whether the facilitator has demonstrated enough of each criteria individually, and all the criteria taken together. However carefully we define the criteria this will always be a subjective assessment because in taking everything into account the relative weighting of the criteria are so complex and nuanced. That’s why we have two expert assessors making the decision.

For example, “How do you know the piano is in tune?”

A piano “in tune” meets certain criteria.  Granted these have a degree of  subjective variation — to my ear, a piano may be “in tune” but to the ear of a piano tuner, it may be off.  But nonetheless, a “tuned” piano has a certain degree of objective meaning.  I could specify this in terms of frequencies of various strings, etc.  In this sense, the “tuned-ness” could be said to be (at least somewhat) “in the piano”.

Then there is a second question: “How do you determine and consciously represent whether or not the target meets those criteria?”

To this I might answer: “I get this feeling in my solar plexus”.

The matching of object and criteria may occur at a deep level (as in deep structure), while the representation of the determination may occur at a surface layer (surface structure).

So suppose I’m trying to model an examplar, and asking them “how do you know when it is the right time to provide verbal and non-verbal encouragement to the learner?” – I could get answers on either the deep level or the surface level.

In terms of the surface level, they could say “I get a feeling in my center, I mark it in my memory, I go back to it a few moments later, and if it still feels right in my center, then I go with it”.

In terms of the deep structure, they could say “I hadn’t actually thought about it consciously before, but now that you mention it, I do realize that I seem to have a series of factors that I consider when determining when a person can benefit from encouraging input. I have these names (X,Y,Z) for those factors, and I have a series of reference experiences of how these factors correlate with responses to the offer of encouraging input.”

Eliciting the exemplar’s strategy for representing to themselves their surface structure of “knowing” could be called “making the internal external (the implicit explicit)”

Eliciting the exemplar’s criteria for making the assessment/determination could be called “making the unconscious conscious”.

In clip 8 of your modelling of Dilts, at 1:47, you use the wording “and what let’s you know that you can do that in that movie…” – there is a deep level “knowing” that “I can do that” (based on some criteria and references), and there is the surface self-representation (what let’s you know …)

My point:  it seems to me that the latter (the unconscious criteria for assessment) rather than the former (the idiosyncratic method of self-representation) is the more significant aspect of the modelling project (but both are important). 

Once the modeller elecits the exemplar’s unconscious criteria, they could then apply that set of criteria in their own deep structure, and use their own surface self-representational strategy quite different from that of the exemplar (e.g., they have a feeling in their solar plexus, the modeller sees a flashing light in their upper outer visual field).

I like your point and I agree people can use different signals that the same criteria are or are not being met and achieve similar results. And I think there are other factors to take into account.First, consider if you were able to acquire the signal for when a criterion is triggered without knowing (consciously) the criterion. Then you would be able to get similar results to the exemplar, you just wouldn’t know know why or how. (This is the basis of John Grinder’s unconscious uptake modelling.) Seeing as most exemplars don’t know how they do what they do so expertly, it wont matter that you don’t know, as long as you only want the acquisition for yourself. But if you want others to acquire it, then you will need to make the model explicit.

Second, I think people are more coherent than your description allows for. The ‘surface’ signal will likely not be arbitrary but have some relationship to the ‘deep’ criteria, and vice versa. They will both carry information about the other. This is a big help when modelling because it means each piece you get helps you to understand every other piece. For example, Robert’s “radar” signal tells us a lot about the kinds of things that can be detected. Imagine instead he had used the metaphor of a ‘sieve’, we would expect him to be noticing different kinds of criteria.

I think of it like co-evolution. Which came first, bright coloured flowers that need pollinating, or bees that can detect bright colours? The answer is neither, they co-evolved. Given that most experts have been doing what they are doing for at least ten years, they have had plenty of time for their system to hone signal-to-criteria and vice versa, and in so doing they will have increased their efficiency.

Good question. The Clean Language questions (CLQ) which might get to this kind of information would be:

• And what determines when [signal is triggered]?
• And where does [signal] come from
• And how does [signal] know when to [be triggered]?

It is worth remembering that most of the information you get will not come in direct answers to specific questions. It will come in the exemplar’s meanderings, metaphors, musings etc. The average number of CLQs asked in an interview is about 1/minute. When you go back over the tapes you find you have gathered much more information than you thought at the time.

As an example, compare the diagram of Penny Tompkins and my first-pass model of Robert Dilts drawn immediately after the interview, with our ‘final’ model completed after lots of reviewing of the data.

And remember it all came from one 45-minute interview and just 35 questions! (And of course observing Robert modelling on the previous day, but you take my point).

This is why you must record your exemplars. It’s also why the main job of a modeller when gathering information is to keep the exemplar’s attention in the right area. Then almost everything they say and do will be relevant.

Dilts talks about “fit” and “fitting”. Again, there are likely to be both surface and deep structures. The surface structure is “the feeling that this fits with that”. The deep structure is “this thing, in relation to that thing, meets these unconscious criteria for fit”.

It seems to me that much of what you are calling “strategies” are mostly surface structures – Dilts gets a feeling that things fit, Mozart that notes love each other, Michelangelo sees the statue in the block of stone and just removes the excess around it. Eliciting those surface strategies may be helpful to neophytes (possibly a lot, possible only a little). Eliciting the deeper criteria may be more important.  Thus, I suspect that the key aspect of learning to compose “like Mozart” has more do with “what does it mean when 2 notes are in love with each other” than with the strategy of getting the feeling in the solar plexus when the specific notes are in fact in love with each other.

You make another important point about ‘strategies’. In their barest form they leave a lot to the imagination.

Regarding your ‘learning to compose has more to do with what it means for 2 notes to love each other’, I wonder, since two notes don’t actually love each other except in Mozart’s mind-body. That’s why computers cannot produce music like he did. And I suspect if you asked Mozart what it “means” for two notes to love each other he might struggle with an answer (e.g. “It doesn’t mean anything, it just is” or “it means God is happy” – I’m obviously making these up but I’ve had plenty of answers like these).

Believe me, just because an exemplar tells you what they think their “unconscious criteria” are, it doesn’t mean that’s what actually determines whether they get the results they do. The modeller has to decide (later) which internal processes best achieve the results. This often isn’t discovered until they start testing their model.

I agree and since the exemplar may not be able to give this information directly, it is often more fruitful to start with what they do know. Often the criteria can be inferred from the signal or from observing the pattern of behaviour which gives rise to the signal.

In this sense (as you say) the pattern can be said to be ‘in the object’ – but that’s a shorthand for saying it’s in the how the perceiver perceives the object.

Note: There is a big difference between asking an exemplar ‘How do you know X?’ compared to asking ‘How do you do X?’. Except in rare cases we have to have a way of knowing how we know something, but it is common for us not to know how we get to know. Speaking is a good example, we know what we say, but we have next to no idea how the words get chosen and formed into the sentences we utter. Usually ‘How do you do X?’ is too complex a question for the exemplar to answer. If they could we wouldn’t need modellers. In a CL interview we tend to start from what the exemplar does know, the what happens, and delve from there.

My guess is the question will morph into dozens of other questions :o)

And remember, an exemplar will have hundreds if not thousands of criteria. Before figuring out how to get to a criterion, the modeller has to figure out which ones are essential. And that depends on what you are modelling.

These are brilliant questions to ask. In my opinion, it’s not so important to have answers (especially this early in your learning and before embarking on your modelling project). Your aim at this stage is to acquire what Penny and I call, a modellers mind – and that’s what you are doing admirably.

If you are particularly interested in criteria and eliciting them I would recommend an excellent book by David Gordon & Graham Dawes, Expanding Your World: Modeling the structure of experience. In particular, see their Belief Template.