Hi Patrick, Your response has got me thinking all kinds of thoughts - confusing at present - but maybe some clarity will arise at some point. ;-) You wrote: “Level III is the epistemological level, the level where the philosophers of science are working on to decide (1) what is the methodology which should be applied for doing modeling at level II and (2) what are the distinctions and coding conventions. This is a meta-level: the discussion is about how to do modeling.” My dictionary defines Epistemology as “the theory of knowledge”.  In NLP terms I have heard it described as “How we know what we know.” It seems to me that this is a conscious mind application to work out the already present unconscious strategies and patterns of which we may be currently unaware. To my (fairly simple) mind that suggests that the set of all data which is unconscious is larger than, and includes, that which has been consciously mapped. I get a bit confused by the word “meta”. By this do you mean an ordering principle which is at a higher level of organisation and as such both includes and organises the data below? Or do you mean it a bit like Michael Hall’s definition when anything can be meta to anything at all? Is it possible to be meta to something and still be on the same logical level? ( I think so). You also wrote: “But to me, a level II modeling process could also be done entirely consciously: you videotape the model displaying the excellence, and then analyze everything one distinction at a time: e.g. first eye-movements, then meta-model distinctions, then meta-programs, etc.” This, to me at any rate, implies that there are two levels of modelling - unconscious and conscious. Most NLP modelling today is of the conscious application of tools that have already been modelled. For example a regular theme to Steve Andreas work is the elicitation of a submodality template onto which he maps what needs to be changed. He has done this with forgiveness, resolving grief, and more recently self-concept. This is not to denigrate Steve’s work - I will be training with him in 2 weeks - his NLP application is rigorous and done with integrity.  Analysing a videotape - something I too have done - implies that you either know what you are looking for and/or are using previously modelled categories of distinction (often initially modelled unconsciously). I am not certain that the conscious mind uptake gives accelerated unconscious skills acquisition. My own experience with this is that I can see intellectually what the model is doing, and can talk about it at length - I’m just not terribly good at rapidly acquiring the behaviour! In fact, I often think I get in my own way far more frequently using this approach. I also use a form of modelling everyday - and that is to map my patient’s worlds and find out what needs to be done with their mental circuitry to effect change. Some of that so far is simple pattern application. More of it nowadays is using the modelling tools to design some intervention specifically for them - usually a variation on a theme. Very little modelling is done the way John Grinder did it - unconsciously. I wonder why that is? Maybe it’s too difficult for most people to suspend one’s filters in the way he did. Maybe most people like me, want it all now  - both consciously and unconsciously - and are not prepared to wait for conscious explication. Yet again, maybe John’s methodology is actually an old one and needs updating! At any rate, because you can be unconsciously skilled without knowing consciously how you do it - or even your conscious idea of how you do it may be wrong - then that suggests to me that unconscious modelling is at a higher logical level than the conscious modelling you describe (using for example, meta-programs). Design is a slightly different thing and I wonder if that is part of what you are suggesting in your Level 3 - using what we already know in different ways, different combinations.  I think it’s perfectly possible to use the conscious mind to set up in sequence a set of unconscious processes to make a new and improved decision strategy for example, without modelling it from someone else. After all, many of the models have their own inherent problems which may drive their seemingly successful strategies, yet you might not want to take all that fully on board. Again though, this is the utilisation of existing modelling tools and processes, albeit in different ways, and seems to me to be subsumed in the larger set of unconscious modelling and thus at a lower logical level. This whole thread reminds of John McWhirter’s distinctions which went along the following lines: Level 1 - technology - WHAT you do. Level 2 - methodology - the HOW of the what. Level 3 - epistemology - the HOW of the how! He also lost me completely with his further detailed analysis of  the “what of the how”, “the why of the what” etc ad infinitum! I think I’ll stop there - my brain still hurts! Thanks for stimulating my thinking processes with your observations Regards, Lewis.