From: firstname.lastname@example.org (Don Steiny)
Subject: Re: Science and NLP (modeling)
Date: 21 Dec 93 19:45:12 GMT
email@example.com (Andrew Moreno) writes:
>Hi. I was reading a book about econometric modeling and how economists
>even with their big econometric models, still can't make useful
>predictions. Even with their computer models that link thousands of
>formulas, they still are off base.
>Is there a connection to NLP here and the type of modeling in NLP?
I have discussed modeling at length with my friend, John Grinder, John Grinder founded NLP with Richard Bandler. I have worked for and with John since about 1979 and have run the essence of this posting by him and got his agreement that it is a reasonably accurate.
First off, what is "modeling?" I have read some fairly confusing things about it, but I can tell you 100% where it came from and what it is supposed to mean.
In the early part of this century the American psychologist William James said that: "between stimulus and response there is a mental process." His idea is that human beings "have instinctive reactions once, after which they are modified by experience." James attempted to study these mental processes and felt that these mental processes were within the domain of scientific study.
In the 20's and 30's, there were exciting developments in the relationship between mathematics and science. It is hard to imagine this today, but until the mathematics was was considered spiritual not scientific. Logic was in the domain of philosophy, not science. There is a fascinating story of how, in the late 19th century an Indian invented complex theorems with no training in mathematics and people saw nothing strange about this. Mathematics was seen as revelations of divine order, and had been since the time of Pythagoras (see, for instance, Russell's History of Western Philosophy). The exciting development was that it was shown that mathematics was logical (Principia Mathematica - Russell & Whitehead).
Many philosophers believed for a time that everything was logical. They called this belief "logical positivism." This philosophical belief was applied to science and to psychology. The thread ran like this. Words must refer to something (a belief that has subsequently been pretty much abandoned by linguists and philosophers). Therefore we needed to find the things that the words referred to and study those things. Philosphical problems were caused by words that did not refer to things that we could observe emperically. Words like: "good" and "justice."
The ramifications of this for scientists were that science should restrict itself to things that can be emperically observed. Therefore "things" like James's "mental processes" were not part of the domain of the emperically observable and thus not part of the domain of scientific study. A school of psychology developed, called "behaviorism," its most famous proponents being Watson and Skinner.
The idea of behaviorism was that the only things that psychologists could study scientifically were things that people actually did. You could measure the stimulus and you could measure the response, but you could not observe or measure any processing that went on in between the two.
Skinner took this belief in interesting philosophical directions, and in his book "Language and Human Behavior," he basically said that there is no such thing as language. His argument, greatly simplified, is that there is no emperical basis to claim that Chinese is somehow the same as English, when then observable inputs and outputs are so different.
Most people have trouble with that idea, because it is intuitively obvious that there is language, and though the words are different people who are speaking Chinese and people who are speaking English are really doing the same things.
Noam Chomsky wrote a paper which demonstrated that the stimulus response model could not account for human language. In addition it can be shown that there must be some mental process going on in human language. In the case of English, no English speaker would accept the sentence:
ball bat boy hit the the the with
which is the sentence:
the boy hit the ball with the bat
with the words alphabetized.
If you think about it you can see that what is going on in a person's mind is not directly observable, you can make predictions about the legal orders of words and that these predictions can be empirically verified by asking a speaker of the language if sentences are good or not.
Chomsky got the idea that in our brain are sets of rules that allow us to know which sentences are good and create and understand language. We cannot directly observe these rules, but nonetheless these rules seem to exist. He called these rules the "grammar" of the language. Linguists strive to discover the grammar of a language, or to put down on paper the rules that a person is using to generate and understand langauge. This paper representation of the internal grammar is a model of the grammar that is in our heads.
I know what I am saying is a quite abstract, so I will give a very contrived simple example and I hope that any linguists that are reading this will not pounce on me (I have a BA in linguistics, and could come up with mind numbing examples). A linguist might say that in English there are "units of meaning" called "phonemes." There are rules that transfer these internal units of meaning into acutal speech. The actual "realization" of these units of meaning might change, but even though they come out differently in acutal speech, how they will come out is completely predictable and thus carries no information. For example, ask yourself the question "how do we form the plural in English." I have asked dozens of people this question and almost always they say "add an s".
However, if you actually listen to the words:
cats dogs trusses
you will hear that "cats" ends with an "s" sound, "dog" with a "z" sound and and trus with an "ez" sound. But your intuition tells you that that distinction is unimportant. Indeed, the "t" at the end of cats is unvoiced, the "g" in dogs is voiced. The difference between "s" and "z" is that "z" is voiced and "s" is not. Thus linguists might say that the "z" at the end of dogs is simply the voicing of the "g" carrying over to the "s" and does not change the meaning. Likewise, the extra "e" sound is added to words that end with an "s" sound or the added "s" would blur into the word and you would not hear it. Linguists can write a rule that predicts when the plural will be an "s" "z" or "ez" That rule will be one of the rules in a grammar of the language.
John Grinder came from a poor family in Detroit. He was an exceptional student. He joined the military and was a Green Beret. He had an exceptional capacity to learn languages, and spend several years as a field agent for the CIA. His sensibilities changed with the Vietnam war, like many people, and he left the CIA and went back to college to get a Phd in linguistics. He made significant contributions in that field. He eventually got a job at UCSC as an associate professor (non-tenured).
John told me that Richard Bandler approached him and said that linguists had elegant models of the unconscious process of understanding and generating language, and that perhaps they should try making models of other unconscious processes. The idea would be to identify units of communication other than linguistic components and to work out a "grammar" of communication.
Note that this did not occur in isolation. At UCSC was the philosopher/ scientist, Gregory Bateson. He and others, Jay Haley, Virginia Satir, Paul Watslawick, and others had been observing the interactions among families and had come to the conclusion that pathological human behavior needed to be understood in terms of the total human relationships within a family or broader group in which the individual existed. They were using results and terminology from cybernetics and systems theory. Again, like linguists, they were trying to come up with a way of describing the relationships. These descriptions are "models."
John and Richard used the results obtained by the group including Bateson as well as their own intuitions to come up with models of human communication that included extra-linguistic factors, like tone of voice, body posture, and other such things. From early on they achieved rather spectacular results.
It is worthy of note that the same zeitgeist that caused John and Richard to break free of the positivist view of behavior was also occuring in the scientific community. When I started college in 1970, virtually all psychology courses were behaviorist and now, 20 some years later, they are all virtually cognitive. Cognitive psychology deals in mental processes and models of mental processes.
The models in NLP Volume 1 are predicated on the assumption that there are behaviorial units called "accessing cues." That a person who is say, spelling, will have a "strategy" of accessing auditory units and translating the auditory information into kinesthetic information. A careful observer can detect this because of eye movements. The person will exhibit side to side eye movements and then will look down and to the right, showing that to spell a word they will start with the sound and then translate it into action. This is a bad spelling strategy, and the trained NLP person will observe this and make a note of it. A good speller will look up and then down, showning that they are first accessing visual information and then translating that visual information into action. Thus they see what the word looks like in their mind's eye, which is effective, but if they start with what it sounds like, the rules for translating sounds into characters in English are ambigious enough that error may occur.
As for economics. Economics is a human science like psychology. It involves human behavior. Economists are doing the same thing that psychologists or NLPers are doing. They try to identify meaningful units of economic behavior and construct the rules that relate these things together. When they make predictions, they are saying "according to my model, when such and such occurs, so and so should result." If that does not happen, they refine their models.
About science and NLP: Both cognitive psychology and NLP work with models of mental processes. Frankly, these days, I have been finding some of the work done with cognitive psychology more interesting and useful than NLP. John paints with a larger brush than scientists. Scientists have other concerns. They have to exist with in a paradigm of keeping their jobs at unversities, publishing, and so on. According to Kuhn, a paradigm is the rules that govern this type of behavior. John chose to simply let his own personal observations and descriptions be OK without constraints. He reasoned that he was exceptionally good at observing human behavior and that his observations and descriptions would generally be useful. Were he to constrain these observations within the paradigm of cognitive psychology, he could spend his whole life on one aspect. He preferred to make more observations less carefully, and to get immediate feedback on his models.
From the earliest days of NLP the axiom was "there is no failure, only feedback." With that axiom, how can you ever be wrong? I have met many many people who have spent lots of money on NLP and had poor results. Needless to say, they are not the cases that they write about in their books. On the other hand I have met many people who have had excellent results. My model of why it works is a bit different that theirs too.
Models are no big deal. We have models of everything. They are just theories. None of them work all the time, but they are useful. They descriptions of something that do not go into overwhelming detail. If it is something like the weather where the detail is needed, then the models are of limited utility. NLP is descriptions of human behavior based on the insights of several people with exceptional intuition. That is really all it is, it is worthwhile for that reason, and it is limited for that reason.
Don Steiny - firstname.lastname@example.org
Don Steiny Software - 214 Calvin Place - Santa Cruz, CA 95060 (408) 425-0382 "success is 99% failure" - Honda