Testing NLP

Neuro-Linguistic Programming (NLP) intrigues me since it tries to be concise about psychological facts. It seems to have the potential to become a science in the long run. I read several books on NLP, but have no personal experience with it (lack of time - too many other interests demand their share). But this will change sooner or later.

There are two ways to get statistics: One by averaging over a large collection of individuals; the other by studying a large collection of moments in single individuals. The second is the one appropriate for good psychological studies, and the structure of NLP seems particularly adapted to this. I always found statistical experiments ridiculous when they tried to study complex phenomena that work only when the persons involved are really interested (something impossible to achieve in a normed set-up), when what they do makes a real difference to them. Studying single persons (by video) in situations that matter is the only way to study them realistically. Well, now the difficulties begin... But it is not hopeless.

The type of questions that can be answered is not of the type `does NLP cure (better) this kind of problem' but `is this or that NLP statement an objective fact'. An important part of this is to have a well-defined vocabulary on factual relations between words and real-life observations.

``The therapist builds a precise model of the cognitive structure of the client's problem. S/he builds it through minimal nonverbal = cues, certain language patterns, and certain eye movements. There is no reliance on interpretations of words like "listless." Two well-trained NLP practitioners should reach identical diagnosis, and be able to tell you exactly how they reached it, down to the eye movement.''
Is this `should' fact or fiction? This can be investigated; the fact that you use the subjunctive form expresses your doubt that it is fact. In math or physics, one could delete this qualifying word. Indeed, natural sciences make progress to the extent their language becomes truly objective so that after training interpretations of reality are identical.

``...to correlate observable externals with internal states relies heavily on self-description. Self-description is one of the least = accurate gauges of mental state I think I've ever seen.''
Indeed. So how does NLP determine internal states? Or how does one notice the lack of reliability? There are cues for checking this, and this would have to be investigated. How can one assess the reliability with which people know to describe their internal states? Quite likely there are criteria for it, and having these would enable one to make more precise statements. What can be asserted scientifically is always how reliable such associations are. Since we all make our internal maps of other people by objective cues of one kind or another (though the rules for turning cues into maps may be very subjective), and some of us have much more insight into other people than the average, there should be enough relations concept - reality that can be tested scientifically.

To broaden the view, one might have to make conditional statements: under this and this condition, that is connected to this and this. Again it is important to be specific enough in the description of conditions, so that significant results turn up.

Scientific investigations of applications are more difficult to conduct and perhaps impossible. There are also no scientific studies of the usefulness of physics for engineering, say. Natural sciences are always concerned with the tools used in the applications; how the tools are used is, if at all, a matter of social sciences (with the associated difficulties). And studying the tools means telling how the tools affect the material they are used on locally, not how they are used to build a dome (or whatever else). One needs to find the `atoms' in terms of which one can correctly describe the relationships under study, without ambiguity. The (nearly complete) lack of ambiguity is the hallmark of science, and good science means that the body of unambiguous and true statements is large and well-organized.

I think a scientific investigation of various NLP concepts and tenets would be very valuable for those doing the investigation, if they are self-critical and willing to admit how little they know. Usually, before we act as a devil's advocate against our beliefs, we know a lot, but in a very vague way, and when we start to look at what we know critically, we discover in how bad a shape our knowledge is. If it works in practice, it is usually more intuition than knowledge, and all the words are like a cloud around the real thing. It takes years to make conform what one truly knows in an area to what one thinks one knows. Scientific investigations are the most important methodical way to systematically expose ones vagueness and to gradually replace it by clarity.

This are some of my ideas, but with more input from your side I could probably say much more. What would you like to investigate? Why are you interested in "Proving" NLP? Where do you feel scientific answers would be most easily obtained? most badly needed? Can one map out the area of NLP into stuff everybody agrees to, and controversial stuff where crucial deciding experiments are desirable?

Whatever you may want to say to these questions or from your own concern about the topic, I am looking forward to your reply.

Know Yourself
A word of caution...
my home page (http://www.mat.univie.ac.at/~neum)

Arnold Neumaier (Arnold.Neumaier@univie.ac.at)