Hi Sarah Well, you certainly picked the key term in NLP application. Let's start with a loose definition and see how well we can tighten it up. Choice is a nominalization of the verb "to choose". To choose something (action X) implies that prior to the act of choosing there was an array of possible futures, from which the person choosing selected X. So far, so good. Now come the interesting considerations: 1. How do we know prior to selecting and executing X that we actually did have an array which contained X as well as other options? 2. Does the act of choosing an action from an array require some consciousness - either of the array (a subset of the array although not necessarily the entire array) or of the act of choosing from the array? Let's look at the question of how we know prior to actually choosing X, X was a member of a set orf possible action. Certainly, if we examine our own experience, it is not difficult prior to choosing and acting to form representations from 3rd position in all or some representional systems of ourselves performing differnt acts. This maps to well known formats such as the New Behavior Generator. We have the impression both by examining our our own experiences and observing (visually, auditorily and to some degree kinesthetically (2nd position shift)that of clients and students) that with a significant amount of practice this is indeed what occurs.  Critically for the second issue (whether consciousness is a required element of choice) this can occur without being aware of what the choices in the array that includes X are. There are plenty of epistemological challenges about how you might verify that the experience from 3rd of having multiple futures prior to selecing X is not simply self-delusional but passing these for the moment, we continue. Given the present state of development of NLP, we might propose the following. Suppose we observe some person, P, in some context, C, a number of times. Further suppose we note that at time, t1, the person does X, however at time t2, the person in the same context, C, does Y - some other action from presumably from the same array as X came from. Further at time, t3 in context C, P does Z. If such variations in behavior occurred a number of times with these different acts being displayed by P, we would be willing to attribute choice to their actions. Please note the extreme difficulty of knowing whether from the perceptual position of P, the context, C, is the same at each of the times, t1, t2, t3... In the above description, we are making a perceptual judgment about the context C. We are judging C at the various times t1, t2, t3... to be the same. As you are well aware, no doubt, a significant portion of how any non-trivial context is perceived is specified by the internal representations (mental maps) that deeply influence our perceptions through the filtering effects of feedforward. Especially in organisms as complex as humans, it is likely that having addressed and responded to what we from the outside perceive to be the same context, P will perceive C as different on subsequent exposures to C. This is, of course, called learning. Sarah, the way you presented the issue has some interesting presuppositions: "but that 'choice' was necessarily determined by some other set of programming" It seems to us that NLP application is precisely the art of living with an ever increasing array of choices and that NLP patterning when applied to therapeutic contexts consists in the identification of contexts in one's life (or the life of a client) where either: 1. we or the client do NOT experience choice 2. the choices we experience are not satisfactory (that is, do not lead to the quality of experience and consequences we seek) 3. the choices we experience are quite satisfactory but we recognize a meta-principle; namely that if we persist in exercising a small set of successful choices, we are exposing ourselves to the significant risk of repetition with its devastating consequence of falling completely asleep with respect to new experiences - the source of new patterning. Sarah, you have pointed at a whole complex of issues that touch on a sequence of traditionally intractable issues such as free will, the infinite regress of the homunculus (the little person inside who actually has what we call our experience... Your use of the double modal operators ("necessarily determined") begs the question. This, of course, raises the issue of appropriateness of accepting Cause-Effect relationships as appropriate descriptions of the actions of humans (or for that matter of any living organisms with colateral energy). We argue strongly in Whispering that it is entirely inappropriate and the consequences of the use of such a set of descriptions - namely, personal responsibility. Your comment that Bandler must have been the one to say, "people make the best choices given their model of the world" is not quite accurate. This dictum was developed as one of the so-called presuppositions of NLP - I am uncertain who coined it but it was used as a rule of thumb in the very early days of the creation of NLP for appreciating and outframing the actions of clients in a psychotherapeutic context and it suggests that the process of change in such a context is a movement from no choice or highly constrained choice on the part of a client to a position of choice with the attendant personal responsibilities that choice implies. As with Whispering itself, the comments offered here are designed to stimulate a conversation - they certainly do not constitute a definition as Sarah requested. Hopefully, they offer a well enough defined vocabulary to permit the exchange of ideas to refine the central and key notion of choice in NLP applications. Carmen and John