Steve Andreas writes:

My previous blog post contained a statement by myself, presenting exercises eliciting several very different experiences that could be described using the word “Meta,” stating that we need to be more specific about which of these experiences we refer to, and concluding that since the word “Meta” is so ambiguous, it would be better to abandon using “Meta” altogether. This was followed by a response from Michael Hall, who is the foremost proponent of using “Meta” as a way of understanding and changing experience.

There were eight comments in response to this post, which are reproduced below (in brown) for convenient reference.

Comments 1-3 are mostly thanks to both of us for the respectful discussion format we used, and encouraging others to discuss other topics in a similar way. Comments 4, 6 and 8 also urged us to continue the dialogue.

Comments 3, 5, 6, 7, and 8 all agree with the distinctions I made between very different kinds of “Meta,” the resulting ambiguity, my proposal to abandon the word “Meta” in favor of more specific experiential descriptions of subjective experience, and that this would avoid ambiguity and increase precision in our instructions, and our ability to predict the results of using a given intervention:

(From comment 6) (paragraph below in brown)
“These distinctions could improve the effectiveness of the whole “Meta State” approach by identifying which of the six (or more) possibilities the “Meta State” question was intended to create and what was actually created.”

(From comment 7) (paragraph below in brown)
“So I find Steve’s suggestion to be more specific a useful one, for it brings the discussion to a more practical ground, i.e. the relevance of “meta” for change and how to more effectively lead a client.”

(From comment 8) (paragraph below in brown)
“When you tell a client to do X, their intuitions about what that means and how to carry it out inform what they actually do. If they do something else that doesn’t work, or if they don’t know HOW to do what you asked in a useful way, no amount of trying harder will get the job done. This is where very specific instructions about EXACTLY what to do and how to do it really shine.”

  1. Bruce Grimley


What a respectful and educational format. I hope NLP as a group learn from this and adopt it when discussing differences in the future. Thank you, Steve and Michael, for your discussion and with best wishes, :-)

  1. Viktoria Ter-Nikoghosyan


Thank you Steve for posting this exchange of opinions! I learned a lot! Steve’s specific examples always teach me new ways of using my NLP knowledge skills. So, grateful!

  1. Marco Fida


Thank you for posting this very interesting discussion, and thank you both for offering your positions.
I find the proposed discussion format both respectful and useful.

For a next time, I also think that organizing the discussion along answers to agreed points or questions would facilitate to compare the positions.

I also appreciate Steve’s proposal of replacing the somewhat ambiguous instruction to “go meta” with a more specific description of the experience we want to elicit. I believe this will both enrich the NLP model and facilitate to guide our clients.

  1. Kelly Gerling


That appears to be round one. 

Steve, are you going to respond, to start another iteration of the discussion??

  1. Will Murray


If I had to translate this into simple English, would I be close if I said “‘meta’ may mean many different things?” And in NLP, which relies on accurate and specific meaning appropriate to the context, making the distinction about what “meta” means in the specific context would be important.


Steve Andreas writes:

Comment 6 below offers examples of additional uses of the word “Meta” giving further support for my assertion that the word is nonspecific and ambiguous, and potentially misleading. Some of these examples overlap with the examples I presented. For instance, 3. Above, is one of many possible changes in point of view, and the examples in 5. Reflexive / Evaluate Evaluations, and 6. Organise are all descriptions or categorizations of experience at a more abstract logical level.


  1. John McWhirter


Thank you both.

I wrote this as part of an Article on Re-Modelling NLP in 2003. It is a description of six different things that are described as being “Meta.”

Summary of types of Meta

Development of Type of use of ’META’
Examples related to NLP
1. After:
Original use: Not in official use in NLP:
It could be used for identify as use of conjunction and sequence
e.g. “You are feeling anxious and now can beginning to relax”.
Some types of ‘Meta States’

2. Beyond Another older usage: Not in official use:
It could be used for things like non-sequitor e .g. “You are getting angry at him and you have to get the food ready for the children tonight”.
Some types of ‘Meta States’

3. Above “Go Meta”

“ Now float up and look down and see how you and your wife are relating”.
Some types of ‘Meta States’

4. Contain Models that contain other models or parts of models.

“NLP Meta Model”
“NLP Milton Model”
Some types of ’Meta Theories’
Some types of ‘Meta States’ e.g.

5. Reflexive / Evaluate Evaluations require changing perceptual position to complete them (See article 13)

Work of Satir leading to Meta States e.g. “What do you feel about feeling that?”
Also potentially useful are:
“What do you think about feeling that?”
“What do you want to do about feeling that?”
“What do you feel about thinking that?”
“What do you think about thinking that?”
“What do you want to do about thinking that?”
“What do you feel about doing that?”
“What do you think about doing that?”
“What do you want to do about doing that?”

6. Organise
“Meta Programmes: organise sensory and language ‘programmes’”
“Batesonian Meta Learning and Meta Theories”
Some ‘Meta’ States, e.g. depression, joyfulness
The DBM® Levels of Modelling NLP (See Article 1)

One of the benefits of modelling is to identify the different ways that both the speaker and listener in any communication use words. For example Virginia Satir’s reflexive question could elicit all six of the uses of ‘meta’ (and probably more). These distinctions could improve the effectiveness of the whole “Meta State” approach by identifying which of the six (or more) possibilities the “Meta State” question was intended to create and what was actually created.


  1. Marco Fida


It seems to me that it is now clear that “meta” can mean refer to several experiences, the point would be to agree on what might be the conventionally agreed meaning and use of it as a distinction within the NLP model.

In my understanding, “meta” has been mainly used to indicate a state from which one can reflect upon or consider his own perspective on things (present state, desired state, resources etc.) in a less involved or objective way. Because of the spatialization of timeline, it was referred to as a “meta-position.”        

Because “meta-cognition” (thinking about thinking) can apply to any content or thought, it can refer to many specific experiences, such as Satir’s “How do you feel about feeling that way?”, V/K dissociation, observer position, etc.

In other words, in all these cases what counts, or what makes change possible, is how I relate to what I am experiencing, i.e. the process vs. content difference.

So I find Steve’s suggestion to be more specific a useful one, for it brings the discussion to a more practical ground, i.e. the relevance of “meta” for change and how to more effectively lead a client.

Finally, it seems to me that while submodalities in themselves can be usefully described as subdivisions within the sensory modalities, the ability to voluntarily change them seems to me to require to “dis-identify” from one’s own thoughts and “objectify” them (meta-cognition) in order to choose a more useful way of thinking.


  1. Joy Livingwell


Interesting discussion and comments!

Artful vagueness and sensory specificity both have their place. What are you trying to accomplish? How are you trying to accomplish it?

Change work relies on the client’s intuitions. When you tell a client to do X, their intuitions about what that means and how to carry it out inform what they actually do. If they do something else that doesn’t work, or if they don’t know HOW to do what you asked in a useful way, no amount of trying harder will get the job done.

This is where very specific instructions about EXACTLY what to do and how to do it really shine. If a client’s intuitions are similar enough to those of a technique’s developer, you don’t need detailed instructions; just give general ones and the client will follow them successfully. When the client’s intuitions differ a lot from those of the technique’s developer, and from yours, exact instructions help a lot, as do instructions about what to do with various client presentations, and this is something Steve excels at.

Other times artful vagueness is just what you need to keep the client from getting bogged down in or distracted by irrelevant details. IMO Michael is particularly good at artful generalizations and large-chunk, big-picture thinking and instructions. My experience is that his instructions as given work well for a lot of people, but not everybody. YMMV.

What interests me most here is the contrast in how two great minds generate intuitions, inferences, and generalizations; what they pay attention to; and what results. Such differences can make understanding and communication difficult. They can also be highly useful, pointing to things that one person alone would not have created, and generating useful synergies. I hope the NLP field will see far more of this sort of discussion and synergy in the near future.


Steve Andreas writes:

Finally, I received another comment from a colleague privately, which I think points out additional aspects of our dialogue.

“In reading your description of Meta, I felt no need to ask meta-model questions, you answered them all. You gave sufficient examples of each category so that I could generate other lateral chunks. It was well thought out and organized. Michael’s descriptions were so vague that even if I thought I knew what he meant, which occasionally I thought I did, I was not so sure. I had a hard time getting through it. The quotes seemed random and not connected. Did he know what he meant but did not sufficiently chunk down for it to be correctly understood? I am mind-reading that you are not at all surprised by his lack of specificity.”

As Michael and I exchanged earlier drafts of my article and his response, I repeatedly asked him to be more specific and experiential, and to respond to the distinctions I made, but he continued to respond with abstractions and quotations from pioneers such as Korzybski and Bateson. Eventually, I decided to post his final draft without any further comments that might prejudice the reader.

However, now I want to point out several (“Meta”) aspects of our exchange that I think are useful and important:

1. I presented exercises that elicited specific subjective experiences, while Michael’s responses are almost entirely linguistic abstractions that don’t clearly specify subjective experiences.

2. Accordingly our “dialogue” might better be described as a pair of monologues, since Hall didn’t respond directly to any of the distinctions and conclusions that I offered. This is an example of the informal fallacy known as ignoratio elenchi, “missing the point” identified by the Romans over 2,000 years ago. It can be summarized as, “If you don’t have a good response to an argument, talk about something else.”

3. A large part of Hall’s response consists of quotations from Korzybski and Bateson. A pithy quotation can be an effective way of presenting a point, particularly if both parties in a discussion have agreed that the person quoted is an appropriate authority. I respect the efforts of the pioneers who first explored the land of “Meta” with the limited conceptual and experiential tools at their disposal. However, we have learned a great deal since their early stumbling efforts to adequately describe that landscape. The use of many quotes is another informal fallacy known as “argument from authority,” also identified by the Romans over 2,000 years ago. This can be summarized as, “It’s true because a lot of authorities said it is.”

4. Lastly, I’d like to point out that all the different comments that readers made about our dialogue are “Meta” comments, and all my different responses to them (including this self-referential sentence) are also “Meta” comments. Assuming that Michael chooses to respond, his comments will also all be “Meta.” To lump all those different detailed responses into the general category “Meta” (as I have just done) gains nothing, and loses all the distinctions, arguments and conclusions that any of us made. That is one of the inevitable results of using highly abstract general categories like “Meta.”


Additional Comments on the Use of the term “meta”

Feb. 25, 2018

L. Michael Hall, Ph.D.

The Greek term “meta” has to do for lots of references. Like any multi-ordinal term (see Korzybski, or the Extended Meta-Model in Communication Magic or in Magic De-Mystified), the word means different things when used at different logical levels.

It would be great if we could reduce it to 6 distinctions and limit it to only 6. In that, I can sympathize with Steve’s desire. If only it was that simple! But sometimes such reductionism does not serve us well.

The simplest use of “meta” is a reflection— thinking about what you just experienced. This is the use of “meta” in early NLP History. We had meta-positions, meta-comment, meta-tactics, meta-comments, and on and on.

Things get more complex when the system under consideration operates systemically and not merely linearly. With reflexivity, you can track the meta move to the next higher logical level easily (well, if you are trained, fairly easily) a level or two. This gives the basic and most simple meta-states like joyful learning (joy about learning), respectful anger (respect about the person with whom you are angry), etc. The gestalt that occurs when you have two or more meta-levels interacting simultaneously creates a new level of complexity. The gestalt — the experience that becomes more than and different from the sum of the parts” — now cannot be explained linearly by reductionism.

Now the states (of thinking, feeling, somatizing, etc.) merge and mix so that the resultant state (as a gestalt experience) arises as a meta-stating of multiple variables. Now we have new resultant states like courage, seeing opportunity, magnanimity, forgiveness, etc. that cannot be explained by reducing things to their parts. When I describe this very specifically, Steve calls it vague. Well, yes it is “vague” if you are searching for the answer in a reductionistic way. It is not vague if you are thinking systemically. (For more about this, see my book, Systemic Coaching, 2012).

In chemistry (Steve will love this) when you bring sodium and chlorine together, the two poisons do not create a super-poison, but table salt. Something more than the sum of the parts, something different from the sum of the parts. You can’t explain table salt in terms of the parts. Now you have to think systemically.

The recursiveness that occurs when you reflect back upon your state repeatedly layers multiple layers, but it is not a hierarchy. In systems language, it is a holarchy. This is what Robert Dilts and I have been focusing on and emphasizing for years. I think John McWhirter has moved in that direction as well.

By going meta to your experiencing and layering awareness, beliefs, values, memories, imaginations, permissions, prohibitions, and many, many other meta-level frames (I have over 100 in the book, Neuro-Semantics, 2012), a rich system of recursiveness occurs and because it is all going on simultaneously (or nearly so), this generates new and higher and more complex “states.” Hence, that is what the Meta-States Model models— the self-reflexive consciousness of we humans. I have detailed that in Meta-States (2007).



Since Michael suggested that Robert Dilts and John McWhirter might support his position, I sent them the exchange above and invited them to contribute. Dilts didn’t respond, but McWhirter did.


John McWhirter responded:

A few further reflections on the “Meta” conversation. I want to add a few points which will hopefully clarify a few common confusions around “meta,” “levels,” “recursion,” and reductionism in general.

1. It is not reductionist to “specify” and enrich our understanding. It is reductionist to take six (or more) different processes and reduce them all to one vague “Meta.” If we don’t know that there are six (or more) different ways to respond, how can we decide which is most useful and make sure that the client responds usefully?

2. My original appreciation of feeling about feeling was from the work of Virginia Satir 35 years ago, where she would change the “feeling” until she got one that she wanted to work with. I also recognized that this could also be used negatively to move away from a feeling or to avoid exploring how a specific feeling was being producing. While it can be useful to move away from a feeling and go to a different one, to a “bigger” encompassing feeling, or a higher organized feeling, it is not universally useful. Indeed 40 years ago I enjoyed greatly the Gestalt Therapy awareness exercises that directed attention “down” from the language of feeling to the feeling of the feeling (I still recommend Steve’s book “Awareness” [John O. Stevens] and was pleased to read Robert Dilts praising it in a recent posting).

3. Linear progression vs recursion: This is another common confusion, that when something is repeated that this is recursive, for example, “How do you feel about feeling this?” it is only a repetition. Recursion necessitates the inclusion of the new product back into the previous one. I played with this many years ago and found that it was useful for developing specific reactions, understandings, and feelings. Instead of progressing in the Satir way of repeating, recursion can be used to stay with the original feeling and develop it. For example, Client feeling angry,
T: “How do you feel about feeling angry?”
C: “Disappointed!”
T: “Now this feeling angry, how does it feel knowing that you will feel disappointed?”
And so on doing the same to develop through recursion rather than moving away or “up.”

4. In the example of Sodium, Chlorine, and Salt: The difficulty in predicting runs both ways, not just from the joining but also from the separating. Equally there is no new need for systems thinking, there is the same benefit from systems thinking in both examples. We benefit from making more precise distinctions at all sizes, within each size and between sizes.  I could have used “levels” here instead of “sizes” but it is not the accurate label. There are no levels operating here. The overuse of levels when describing different sized structures is another common mistake, especially with the western cultural bias that “higher” is better. Instead of “sizes” I could have used “scope” which also covers other types of extended structures and functions, but I wanted to describe the specific one here.


Steve Andreas comments:

As before, Michael’s response is full of important-sounding abstract linguistic terms—“multi-ordinal,” “reductionism,” “systemically,” “reflexivity,” “gestalt experience,” “multiple variables,” “recursiveness,” “multiple layers,” “hierarchy,” “holarchy,” “more complex ‘states,’ ” “self-reflexive consciousness”—without a hint of what subjective (VAK) experience those terms might indicate. NLP is often described as “the study of the structure of subjective experience” (not “the study of the structure of linguistic terms”) Without subjective experience, there is no structure to study, nothing to go “Meta” to.

Knowledge in any field progresses in two interrelated ways:

  1. Accumulating specific detailed data/information through observation, experimentation, quantification, etc.
  2. Organizing the specific information available into abstract hypothetical/theoretical categories, relationships, cause-effects, etc.

Additional data often suggests new or revised ways to organize it, and organizing it often suggests new or revised ways to search for more specific data/information.

Data without organization is only an incomprehensible jungle of specific facts, and grand abstract theories are just as useless if they have no link to the real world of specific observations.

“Meta” is a very large abstraction that includes many different kinds of subjective experiences, three of which I described in my original statement, and John McWhirter’s comment identified several more. Using the same word for very different experiences creates ambiguity and confusion at best, and serious misunderstandings at worst. I’m going to focus on three paragraphs of Michael’s response, to show how easy it is to go astray with such a large and ambiguous generalization.



Michael wrote:

“In chemistry (Steve will love this) when you bring sodium and chlorine together, the two poisons do not create a super-poison, but table salt. Something more than the sum of the parts, something different from the sum of the parts. Now you have to think systemically.”


Steve writes:

I do love this example, because it nicely illustrates the limitations of large generalizations, and the importance of specific detail. Both sodium and chlorine are poisons because they are very chemically reactive, causing serious burns. The category “poison” is a large generalization based on their effects, which result from opposite and (equal) reasons. The reactivity of sodium results from the fact that its atoms have only one electron in its outer shell, which is loosely held and easily lost. The reactivity of chlorine results from the fact that its atoms have one missing electron in its outer shell, which strongly attracts an additional electron. As any high school chemistry student can tell you (I have a BS in Chemistry from Caltech) it’s “elementary” to predict that the poisonous effects of each are completely nullified when they react together to form table salt. You can “explain table salt in terms of the parts” quite easily.

The words “thinking systemically” may sound erudite, but it’s only “smoke and mirrors” that hide the ignorance of how chemicals react. Basic chemistry allows you to predict that sodium will react similarly with fluorine, bromine, iodine, and astatine, and that chlorine will react similarly with lithium, potassium, rubidium, cesium and francium. There will also be differences in their reactivity, and these are also predictable, given enough knowledge of atomic mass, isotopic composition, etc.

It’s very useful to recognize that whenever someone speaks of “emergent properties that couldn’t be predicted in advance,” that is only a statement about the speaker’s ignorance. When you understand anything in great detail, you can make all kinds of predictions, and that is also true in NLP. Thinking in vague abstract categories get vague and unpredictable results; detailed thinking gets specific and predictable results. That is why the core of the Meta-Model’s ability to reveal the structure of subjective experience is the word “specifically.” “Who specifically, how specifically, what specifically, when specifically, where specifically.” We need to apply the Meta-Model to the term “Meta” in order to know what kind of experience we’re talking about.



Michael wrote:

“By going meta to your experiencing and layering awareness, beliefs, values, memories, imaginations, permissions, prohibitions, and many, many other meta-level frames (I have over 100 in the book, Neuro-Semantics, 2012)”


Steve writes:

It’s not “reductionist” to continue to learn more and more about the underlying structure of how chemicals react, and it’s not reductionist to learn more and more about the different structures that underlie the generalization “Meta.”

Michael is apparently arguing that since he has “over 100 meta-level” frames, having only the three I proposed (or the six that McWhirter proposed) is reductionist. However, “awareness, beliefs, values, memories, imaginations, permissions, prohibitions,” etc., are all differences in content. No matter how useful those frames might be (like positive intention) the structure, “I X about Y,” is identical. The difference between content and process is a key distinction in NLP; that’s what allows us to quickly change any phobia, regardless of the content, because we know that the structure is the same.



Michael wrote:

“The recursiveness that occurs when you reflect back upon your state repeatedly layers multiple layers, but it is not a hierarchy. In systems language, it is a holarchy.”


Steve writes:

If I reflect on my own experience, that is indeed recursive, because it’s a categorization of my own experience, as in “I’m angry about feeling powerless,” because the subject and the object of the sentence are the same, creating self-reference. But if I reflect on someone else’s experience, as in “I’m angry about Fred’s passivity,” that is also a “Meta” statement, but it’s not recursive, because the subject and object are not the same. “Meta” “states” are not always recursive or reflexive as Michael seems to think.

“Layers multiple layers” is a metaphor that doesn’t specify what a “layer” (or “multiple layers”) is in subjective experience, or how it is represented (VAK). Does “layers” mean translucent images superimposed? If so, do you see one image through the other, or do the two images blend into each other to form a single image? Or something else?

Take a moment to notice the image that pops into your mind spontaneously when you read the word “surgeon”. . . and “butcher.” . . .      Now notice your image of the sentence, “The surgeon was a butcher,” . . . and compare that with your image of, “The butcher was a surgeon.” . . . The first sentence describes a person as a surgeon within the larger category butcher, while the second sentence does the reverse. The order of the nesting of categories makes a huge difference. Notice how it determines which context is preserved in the combined image, and what aspect of your image is amplified. These effects are not even hinted at by the metaphor “layer.”

Finally, a sequence of nested categories (of whatever length, and whether recursive or not) does create a hierarchy of logical level classification, which I described in my original statement as a third kind of “Meta.” (imaging “person,” “human,” “mammal,” “vertebrate,” etc.) A holarchy creates a very different kind of hierarchy of scope inclusion at the same logical level, which I described in my original statement as a second kind of “Meta” (seeing “leaves,” “branches.” “trees,” “forest,” etc.).

When someone describes a system as a holarchy, in which holons interact with each other in a non-linear way that is recursive or reflexive, that is a very abstract description that could apply to an organ, an organelle, an organism, an ecology of organisms, the components in an electronic amplifier, a solar system, or any number of other systems.

The universality of that general understanding is one of the key advantages of abstraction, but it also provides a false sense of understanding that is essentially meaningless unless you know the specific details of the particular system being described.

A further source of misunderstanding is when someone uses important-sounding words without really knowing what they actually mean. For instance, read the following words, notice your (VAK) understanding. . . .

Then check out the definitions of system, holarchy, non-linear, recursion, and reflexive, and compare your understanding with what you read. . . .

You probably did pretty well with “system,” because it is a common term used to describe familiar experiences (“nervous system,” “economic system,” etc.) But how accurate was your understanding of the other terms? And do you find it easy to apply those abstract terms usefully to the change work that you do with clients?

Yet another source of error in any large generalization is the omission of counterexamples. For instance, some interactions between holons in a holarchy are not non-linear, and many are not recursive or reflexive.

As an old saying goes, “The devil is in the details,”—but so are the angels!


This blog post was sent to Michael, offering him the opportunity to respond further. He replied, “My full training schedule, and other personal events, make it impossible for me to respond further at this time.” Any future response can become a further continuation of the dialogue.


UPDATE 7/10/2018: Michael’s response and the conclusion to our dialogue is here.