Hi Loren, What a great post - very clear, good sensory based examples and the best exposition so far for the argument that meta-programs are more content distinctions and rather less process. I have not really attempted to use metaprograms for modelling per se. What I have done is, as you say, use them as attention filters to get the appropriate information across to my patients in a way that gets them to act on it (face to face "in the moment"). Having learned the distinctions I have "forgotten" them and simply use whatever comes up spontaneously at the time based on calibration etc. You state for metaprograms that "3) Non-verbal elements of the skill are irrelevant" However, when I have used a "metaprograms filter", there are quite definite non-verbal cues accompanying the various distinctions. And it is also possible to get a submodality representation that is fairly typical for each distinction. Whether this is useful in modelling is another question. You wrote; "As far as I can tell WITW proposes a model that basically says I don't have a clue about what's important (how could I?), I'm going to put aside everything else I know and just mimic until I can do the same things and produce the same results as the exemplar.  Once this is stable in the neurology, and only then (if ever) will I even begin to try to apply any kinds of f2 filters to this in order to build a transferrable model." I agree that this is one way to do modelling - is it the only way? - very few seem to be able to do it this way. For example I am going to train with Steve Andreas in 2 weeks. He has "modelled" self-concept. Having read his book he has mainly used submodality distinctions to code how people build generalisations about themselves, how to enhance this when it is useful - and how to change it when not. He has made several improvements and added distinctions to the old visual squash model in this way. I made several shifts in my own self-concept simply by reading the material and am already using the material to elicit and change others. Steve says that he has produced a model of the main elements of self-concept. Now, does what he has done, useful as it is, qualify as modelling, or is it simply moving content around and feeling "different"? To give another example. You and I have both trained at one or other time with Carmine Baffa. He has a process for modelling which he claims accelerates learning and skills acquisition which runs along the following lines. 1. Get an example of the skillset your exemplar is demonstrating. 2. Run a video over and over in your mind's eye and ask your unconscious to identify with precision those times in the past when you have used the same gestures, posture, voice tone etc across all life contexts. Ask your unconscious to "build a collage" of this, and to signal you when it has done. 3.Then "morph" your image onto that of the exemplar so that it is "you" doing the skill. 4. Elicit submodalities of times in the past when you were displaying other skills very well. Map across the new skill. 5. Identify future times and contexts to use the new skill and install. This is a different process to the one described in Whispering and uses the products of what has already been modelled (accessing cues, time lines, mapping etc) to model. In your view is it therefore legitimate to call this modelling? I have to saythat personally, for all the NLP training I have done, I'm not at all sure whether its helped me rapidly acquire skills or if the fact that I'm enthusiastic and keen about what I'm going for is really the main catalyst! I suppose that says that the "modelling state" is really the prime concern! Look forward to your comments. Regards, Lewis.