Thanks Michael for your response! What I didn't mention in my post is that I've consistently found that there are many states I and my clients have where the 'outside world' looks and sounds different. For example: the extremely depressed client who literrally had a blue hue over her visual perception of the world when she felt depressed. She didn't notice it until I got her to contrast a wonderfully happy state vs her depressed state. She told me that the happy state actually had a red hue. To me this matched with her langaugae of "feeling blue" and "why couldn't things just be rosy again?". So, by teaching her to have control over the hues she used, she could automatically control her mood and state. I find this kind of contrast in almost every kind of states/moods that are significantly distinct. Afterall, submodality distinctions are part of each representation, not only internal visualisations(I'm only focusing in Visuals here). To me these distinctions arn't 'illusions'. Illusions come from the presuposition that their is reality we can identify and use as a base comparison. To me that's an illusion :) I've been thinking about my question. I realise that I tend to be highly suspicious of linear models. (as is the F1-FA-F2 model). For example: I think F2 transforms could easily effect F1. After all, language is hypnotic, and hypnosis can definately effect processes at the molecular level. For example, the classic experiment where someone in trance is told that they are going to be touched with a hot coal. Ice is used instead. And a blister develops. So if our langauge can and does effect molecular processes therefore F2 transformations can effect F1. (or am I just in hypnosis mode and going rampent with my cause and effects!) So these VAK tranforms (as I call them) could happen at any and all levels. Maybe we should just go for what model works the best! And John and Carmen's model works well. I don't know what John and Carmen think but I'm not sure if it's profoundly useful to try and search for ways that to make NLP match up with scientific 'reality'. I know a lot of people (some in this forum) who want to find scientific evidence to back up the NLP models. But I don't think NLP is scientific. That's why it works so well (in my opinion)! So my attitude is that the VAK transforms should be included in the model (or not) depending on if it makes the model more effective. Sorry if I rambled in this post! Michael, thanks again for giving me your thoughtful reply and also for your explaination of the 'know-nothing' state. All the best, Michael Norman