I agree with you, but thought it might be interesting to use an alternative set of the "given" variables in order to stimulate more ideas.  And I also agree that it is within the NLP map, which seems to be the game we're all playing here.  I would be more than happy to suggest or take suggestions about alternative epistemologies, as it seems much of NLP continues to gravitate around Bateson/Korzybski there are others that would be useful, that might imply, at least a reshuffling of the methodologies.  First off I'd say Gille Deleuze and Felix Guattari. As far as modeling goes, as a musician I learned clearly how to differentiate minute differences between one player and another, and then how to teach others how to imitate and keep firm contextual boundaries between one person's style and anothers.  It is the discipline, for a rock musician, since there isn't the same rigor that there is for classical guitarists, that makes the difference.  You know who buzzes the strings, when and for how long, what shapes the phrases are, and what edge of the pick you use and how close to the bridge.  You also imitate the facial musculature of each musician, their posture, and once you've listened to interviews, can begin to hypothesize what they're hearing inside there heads.  Most recently I've been teaching creative writing, and I try to teach writing along the same lines, as well.  One of the ways you teach poets, for instance, is to have them imitate other poets.  Once you begin to understand the differences in phrasing, rythm, concentration of particular consonants and vowels, for instance, and their distribution among other variables, such as types of language usually associated with particular contexts (ie technical vs personal, antiquated vs current conversational usage) you set up exercises, similar to NLP patterns, and have your students hammer them in unti they begin to feel as though they really are that particular poet.  Sure, this is different from R&J; watching video tapes of ME, and for the most part much less ambitious (for the most part), but per the context, we have a very fuzzy overlapping field. best, C