A Response to Shawn Carlson’s blog post Patterns Within Patterns


I’d like to start by completely agreeing with Shawn’s description of a lovely piece of reorienting in time that can be used to consolidate any change, however made: An additional temporal step is added by inviting the client to step into the future and, looking back, see the changes she has made within the ‘past-future’ visible from the farther future, “step into the future, and looking back toward now, realize how far this change has taken you.”       

However, this great process doesn’t appear to me to fit the steps of the Meta Pattern, which assumes integration (“collapsing”) anchors, rather than some other way of combining experiences, such as chaining. The beginning state is a solution state, rather than a problem state, so when you ask a client to “step into the future,” that dissociates from a solution, and associates into a “farther future” that is a resource in itself, (rather than simply a “break state” from which to choose a resource state). While associated into this farther future, the client views scenarios from the intervening time (the past from the point of view of being associated into the farther future).

This process is an example of a general pattern that Erickson called “reorientation in time.” It is also an example of nesting one experience inside another one, rather than integrating two experiences, or chaining them together sequentially in time. It is structurally identical to the phobia cure, in which the client associates into the movie theater, and nested within that dissociates from the past phobic memory. It’s also an example of utilization. If someone has a phobia, we know that they are skilled at associating into a memory, so we ask them to associate into the movie theater visualization to provide a context for dissociating from the troubling memory. This nesting of experiences is a different experiential process (there are others) that doesn’t fit the steps of the Meta Pattern.


I’m getting a little tired of playing “whac-a-mole” responding to misrepresentations of what I have written. For instance, Shawn writes, This is exactly a formulation of the HNLP Meta Pattern (so I am a little surprised that Steve is arguing there is no such thing!).”

I never said there is “no such thing”; if that were true, we wouldn’t be able to talk about it! What I said was that this meta-pattern was a useful, but somewhat crude early generalization, with a lot of exceptions. Connirae’s and my statement from Change Your Mind (1987) starts with. “One very broad general formulation of change work . . .” The word “One” clearly implies that there are others, and stating that it was “very broad and general” indicates that it lacks a lot of detail, and likely has exceptions or counterexamples. It was also written 30 years ago; luckily we know more now than we knew then.

What I disagree with is Shawn’s statement that, “Perhaps if I demonstrate how the Meta Pattern is the basis of all change work enough times, the universal truth of the Meta Pattern will become unarguable!”

         There are a number of problems with this statement. First, it’s not possible to prove universality by providing “enough examples.” [just as Finding 100,000 apples, or even a million apples, wouldn’t “prove” that other types of fruit don’t exist — and I have already provided some oranges and bananas. The Meta Pattern is useful, but there are many significant examples of change work that don’t fit the Meta-Pattern.

Setting aside the question of whether or not there is any such thing as “universal truth,” let’s take a look at Shawn’s description of the visual squash, which he presents as an example of the Meta Pattern:

“Similarly, in the visual squash, you separate and externalize the parts that are creating the inner conflict. You place one as an image on the client’s left hand and the other on their right. We are now at step 2 of the Meta Pattern, we have dissociated the client from their problem (the internal conflict).” I agree with Shawn fully up to this point.

Shawn goes on to write: “On to step three of the Meta Pattern, associating into the resource. First let’s help your client to find the appropriate resource. We do this by chunking up on the ‘intentions’ of each part until we find a shared value, or at least values that sufficiently overlap. This shared valued is the resource. At the same time we build rapport between the parts by asking what each can learn from the other.”

It’s important to realize that the intention of each part — as well as any shared intention — is still dissociated, not associated. The two parts (and their intentions) are still represented as separate images in the two hands, and also separate from the client.

Shawn follows with, “We then associate the client into this resource, typically by internalizing it “bring that back inside your body.” Shawn has omitted a very crucial step, namely bringing the two hands, and their associated images together to “collapse” them, before bringing the resulting integrated image into the body. The Meta Pattern specifies “collapsing” two associated states (problem and resource). But in the visual squash, two dissociated images are “collapsed,” followed by association.

Furthermore, neither image in the visual squash is either a problem state or a resource state. Neither one is a problem in itself, the problem is the conflict between the two. The resource is the result of combining them.

         In the discussion above, and in previous posts, I have set forth my understanding of the Meta Pattern, and Shawn has set forth his. Readers will need to come to their own conclusions.


Shawn writes, “In collapsing anchors, as Steve rightly says, there have to be two states or two anchors to collapse. Steve then for some reason says that this requirement is not included in the Meta Pattern; this is incorrect.”

In the Meta Pattern diagram there is no mention of any anchoring — none whatsoever. There is also no mention of a cue for the problem state, or the context of the problem state. This kind of omission is inevitable in any vague general outline that purports to be universal. I prefer more detail and specific steps, to be sure that nothing essential is left out.

One of the sources of much of my disagreements with Shawn is that we are describing very different levels of detail. This can be illustrated by different responses to the question, “Where is Times Square?” One answer is, “It’s in North America.” Another is, “It’s in Midtown Manhattan, New York City, at the junction of Broadway and Seventh Avenue.” Both answers are true, but the second provides much more specific and detailed information.

A car that is running poorly could be described as being in a very bad “state,” but that wouldn’t indicate much about how to fix the problem. A good mechanic might be able to listen to it and determine that it only needs a small adjustment in timing. Likewise, someone’s “problem state” is a very general description, while “see a movie of yourself in that situation” is a very specific small intervention — a change in point of view that leaves all the other aspects of a “problem state” (content, sounds, context, etc.) undisturbed.

When talking about general states it is easy to ignore differences, and see similarities. I keep making distinctions, and Shawn keeps saying they aren’t important, and that they all fit the Meta Pattern, which is based on integration of a resource with a problem state.


One principle that I have found very useful in making changes (whether in NLP or in other contexts) is to make the smallest change necessary to get the desired outcome. There are at least two important reasons for this. One is that it is much easier to make a small adjustment, such as point of view, the location of an image, or in some other submodality. As Milton Erickson said, “Your task is that of adjusting, not abolishing.” The other reason is that a small adjustment is much less likely to interfere with the rest of the client’s functioning than a large one — what is usually called “ecology” but which would be more accurately called congruence.

For instance, imagine that you notice some irregularities and palpitations in the beating of your heart, so you consult a doctor. After she listens to your heart, takes an EKG, and asks a few questions, she says, “Well, some simple life-style changes will resolve those symptoms, but a heart transplant would also work.” I hope it’s obvious that a heart transplant would be a lot more difficult, and have significantly more risk of problematic consequences. Unfortunately working with “states” (especially ones like “awesomely confident”) is a lot more like a heart transplant, changing much more than is necessary or useful.


         Shawn writes that, “The problem and resource then ‘duke it out’ within the client’s physiology. This is repeated until the resource state (or a resource state) emerges the winner.” That is a description of what happens if you integrate two very different “large chunk” states, an indication of trying to make a “heart transplant” type of change, in contrast to a small adjustment. Unfortunately a client’s responses to a “heart transplant” change is often quite intense, leading both client and therapist to be unduly impressed, rather than realize it’s only a result of sloppy work. Years ago, a boss I had frequently said, “Do not confuse motion with progress.” The best changes don’t usually involve pyrotechnics; they are usually exemplified by the classic, “Hmn, I never thought about it that way.”

The idea that states have to “duke it out” until one “emerges a winner” usually results from thinking about the emotional intensity of a state, rather than its appropriateness for the problem. This leads to an assumption that I have called a “the mathematical fallacy of states”: that if you have a negative state rated at a 7, you need a positive state rated at least 7 or 8 to “overcome” it.

This is obviously false in the case of the phobia process. A phobia is a very emotionally intense state, let’s say somewhere in the range of 7-10; Dissociation is a much less intense state, perhaps somewhere in the range of 1-3. There is no “duking it out,” nor any “winner” when using the phobia process; there is no conflict at all. The change in point of view is a specific and precise adjustment that is makes it easy to view a troubling memory comfortably.


Shawn follows with, “The Meta Pattern is then repeated on an iterative basis by running the client through other triggers, and other contexts, until the client is not able to identify any more examples of the problem. This is typically fairly quick as the brain is an amazing generalization-machine!”

Yes, the brain is an amazing generalization machine, and you can do change work by repeating the same change on different contents and contexts to help it generalize, a Shawn describes. However, it is far easier and simpler to begin by asking the client to choose the most intense example of a problem response. This is identifying what cognitive linguists call a “prototype” experience that represents the entire category of problem experiences. If you do this before intervening, a change in the prototype will automatically generalize to all the other examples, in a kind of “domino” effect, so you only have to do the change once, which is more efficient. Of course it’s important to check other examples to make sure that generalization has occurred, but you don’t have to keep repeating the change work. Checking three different past examples, followed by three future examples, is usually sufficient to confirm generalization.


Shawn goes on to write, “By the way, when this happens the Meta-Pattern-TOTE-strategy ends (to quickly address another of Steve’s objections).” I assume that is how Shawn works; my objection (yet again) is that there is no mention in the Meta Pattern diagram of a TOTE, an exit, or a criterion for stopping. The arrows go around clockwise between the four circles, and Shawn has emphasized that you can start anywhere on the circle.


Shawn writes, “I agree with Steve that dissociation is part of the resource state for a phobia.” Again that is not what I wrote; I wrote that dissociation is the resource state (not “part” of it). The “laughter; and ‘safe-to-safe’ experience” that Shawn mentions are embellishments (and there are many others) that be used to support dissociation, but they aren’t usually required.


Finally, I completely disagree with Shawn’s statement that, “Steve’s modal collapse, ‘How is it possible that you should be able to do something you can’t do?’ is an example of what Robert Dilts calls ‘apply to self.’ ” I’m pretty sure Robert would agree with me, and I would think that John Overdurf might also agree. Can you send this section to John and ask him if he would be willing to comment?

The sentences, “I can’t do it” and “I should do it” both apply to the self of the person speaking, but neither refers to the sentence itself, which is what the phrase “apply to self” refers to.

A trivial, but instructive, example of “apply to self” is “This sentence is true,” because the sentence describes itself. A more useful example is, “Everything I do is stupid.” Since saying that sentence is one of the things that someone does, the sentence refers to itself, and therefore the sentence must also be stupid. Pointing this out can weaken or break the problematic generalization.

What happens in what Shawn is calling a “modal collapse” is that an unrecognized contradiction becomes apparent. “I should do it” presupposes that I can do it, while “I can’t do it” presupposes the opposite. By juxtaposing these two statements, the contradiction (“speak against”) becomes apparent, and something has to give. If there is good evidence that the “it” is not something I can actually do, the “should” vanishes. If the evidence is less robust, the contradiction can be resolved by reconsidering the “can’t” and exploring alternatives that I might actually be able to do.

I think the explanation above is a lot simpler and clearer than Shawn’s description: “The modals ‘can’t’ and ‘should’ exist in different ‘modal spaces’ (we will call them ‘frames’): ‘can’t’ generally inhabits either the epistemic (personal knowledge) frame, or the doxastic (personal beliefs, without knowledge) frame. ‘Should’ generally inhabits either the deontic (duty), or axiologic (cultural) frames. In any case for brevity I will only address the can’t modal here using the doxastic (belief) space.”

In volume II of my book, Six Blind Elephants, chapters 4, 5, & 6 go into much more detail about modal operators, self-reference, and self-contradiction and how to change them in useful ways.

Notice that Shawn uses the same word “collapse” to “apply to self” and “modal collapse (contradiction), ignoring the distinctions I have made above — and he also uses the same word to nesting experience (discussed in the first paragraph of this post). In earlier posts he uses “collapse” for both integration and chaining. There is a further possible alternative, combining two experiences into a hierarchy of relative importance.

Using the same word “collapse” for all these very different experiences, is like saying that Times Square is in North America. It’s not false, but failing to recognize the different kinds of “collapse” makes it impossible to do really elegant work that makes the smallest change necessary in order to get the client’s outcome.


P.S. I was pleased to see Shawn describe discussion of dopamine as a “red herring,” and then disappointed that he went on to discuss it as if it weren’t.

I think the finding that news of a loss results in a dopamine spike may be a result of the “gambler’s fallacy” in which a gambler believes (erroneously) that a loss on one round signals a greater probability of a win on the next one. If the news of a loss is immediately followed by an image of a future win, the latter, not the former, would be what actually elicits the dopamine spike. That would fit better with the overall finding that dopamine is released in response to an expectation of reward. Does anyone know a compulsive gambler they could model to explore this possibility? Even if it’s irrelevant to NLP change work, it’s an interesting puzzle.

Since Shawn describes the “Tree of life” process as not NLP, I won’t comment on that, except to respond to Shawn’s statement, “I find it to be the simplest model that captures any and all human experience.” It seems extremely unlikely to me that any simple model could do that.