Hi Lewis. After reading your post, my first response was "Those are some really good questions I wish I knew the answers"... After thinking about it a bit I hope I can offer something useful. 1) What specific arrangements did I make?... What comes to mind is an NLP Study Group I used to belong to.  We practiced numerous times all the exercises from Turtles and John's Advanced Modelling audio tapes covering building lifelines, Stop the World states, multi-processing, etc...In the specific case of this project the person I was modelling is someone I trust a great deal.  The arrangement was not much more than, "Is it safe to go over there?"  "Yup" "Okay let's go."  2) How did I make it context specific? The original motivation for the modelling work was that I was working with a friend of mine in a business context.  We were working with a specific individual, call him Mike, on a business project.  My personal history with Mike had been less than satisfying and I had never before been able to elicit the kind of responses or cooperation from Mike that I was at all pleased with.  However in the meeting that so captured my attention my friend easily elicited wonderful responses quite quickly and a huge amount was accomplished. With that as background I ended up with opportunities to interact with Mike on projects outside of the presence of my friend.  Any context with Mike was the contextual marker.  It took very little time once I stepped into 2nd to elicit similar responses.  Then I just had to keep the responses stable, or rather myself stable in that position for a sustained period of time.  Once I could elicit the class of responses and cooperation from Mike that I wanted on a consistent basis, then I stepped back and compared other contexts in which I'd seen my friend demonstrate excellence where I had not yet succeeded.  Then I applied myself to those contexts until I was certain I could consistently generate similar responses. (3) I'm not sure if the "For example" bit is asking about a lifeline or how I personally make the shift to 2nd position. (4) As for the last bit about what wasn't useful... I think the first thing I did was that once the 2nd position was stable and I could elicit the classes of responses I wanted, I set out to make myself comfortable...I reduced the amount of muscular tension present in my body and relaxed and deepened my breathing to something much more comfortable and normal for me.  My experience was that this did not make me any less effective.  I quickly found that by adding back rapport skills that I normally use things got even better.  I guess technically speaking this wasn't a pure manipulation of the variables I found present in the exemplar, but... What happened next was that I began to notice chunks of behavior and I would get some internal dialog right after a piece of behavior saying, "Hey that was weird you don't usually do that."  As I identified chunks of behavior I guess I just set the intention to drop "everything" out and only use the pattern I'd just recognized.  Then once I played with that and noticed the results, I'd go back to the original model and just do that again.  Then I'd keep noticing new pieces of behavior and try them individually. Now that I've written this down I'm feeling a little lost in the process.  I know that I'm utilizing multiple behaviors I've borrowed simultaneously.  By observing myself from 3rd  what I'm doing seems much simpler and pleasant than the original model.  What I haven't done yet is attempt to write down or codify specifically what is happening.  I think I'm just putting off the coding phase. Is that helpful? Loren