In addition to these pre-suppositions, there are a number of tools aimed at providing more accurate definitions of the details to improve communication. Many of these were developed and refined by a second generation of experts in the field of NLP who, over thirty years, have significantly extended and defined the tools beyond the original materials.

One of the original NLP tools is the meta-model of language. This language tool uses standard models of linguistic well-formedness to determine whether and to what extent conscious formulations have been distorted from their initial sensory (non-verbal) representations. It also helps therapists and modelers to recover relevant aspects and distortions of that information (Bandler & Grinder, 1975; Bostic St. Clare & Grinder, 2001; Dilts et al., 1980; Grinder & Pucelik, 2012; Lewis & Pucelik, 1990).  [see also, for example, Chomsky, Aspects of the Theory of Syntax, Cambridge, Mass., 1965; Lakoff, “Presupposition and semantic well-formedness” in Semantics: An Interdisciplinary Reader in Philosophy, Linguistics and Psychology, Steinberg & Jakobovits, ED., Cambirdge Univ. Press, NY, 1971.]

In alignment with the therapeutic tenets of American Psychologist, Carl Rogers (1951), rapport skills, stand out as an important tool in NLP. Rapport can be established by matching breathing, posture, language patterns, tempo, and through other means. In NLP, Rapport refers to the ability to consciously enter unconscious synchrony with another person and to lead them in a new direction; it does not imply personal disclosure. While not foundational to NLP it is a strategy that arises directly from some of NLP’s basic postulates (Andreas & Andreas, 1989; Bandler & Grinder, 1979; Dilts et al., 1980; Lewis & Pucelik, 1990).

Another such tool involves understanding the effects of submodality differences in personal perception and their usefulness in changing perception.  As previously noted, submodalities are the dimensions of experience that code for intensity and valence. Generally, for all senses, they include size, distance, locus, intensity, complexity, movement, focus and others. Each sensory modality has certain submodalities that are specific to that sense.  Examples of the effects of submodality differences that have been observed and documented in hundreds of individuals include:  Something that is large and looming, is usually perceived as more important than something small and moving away.  Something in full color is perceived as more present than something in black and white.  Perceptions from within the body, associated, are felt to be more important than things perceived from outside the body, disassociated. Stereophonic sounds are perceived as being more real than mono-dimensional sounds.  Submodalities are fundamental to NLP (Andreas & Andreas, 1989; Bandler & Grinder, 1979; Dilts et al., 1980). Although not directly adopted from his writings, these elements with the VAKOG are nearly identical to Titchner’s view of the basic structures of mental experience (Titchner, 1908; James, 1950).

Previously, mention was made of Miller, Galanter and Pribram’s T.O.T.E. model[1]. This may be compared to a loop structure in computer programming. The recursive loop cycles though a certain sequence of actions over and over until a certain criterion is reached. When the criterion is reached, the loop ends. On the most basic level, we see this same pattern in the TOTE model. TOTE is an acronym, it refers to Test, Operate, Test, Exit and is derived from a seminal publication in cognitive linguistics by Miller, Galanter and Pribram (1960), Plans and the Structure of Behavior.

These authors set forth the idea that behaviors in complex systems that have no defined end (like do-this-N-times-and-stop) need to have some guiding process that allows them to know what to do and when to stop doing it. This means that behaviors in living organisms can be usefully compared to a computer program that sets a criterion, operates on the data, and tests to see whether the criterion has been met. If the criterion has not been met, the program loops through again.  If the criterion has been met, the program ends.

In the language of NLP we begin with an outcome. The first test in the TOTE strategy is the comparison of our present state to a desired state—an outcome.  If they fail to match, we perform some operation in hope of changing percepts, behaviors or the world in the direction of our stated outcome. The second test in the algorithm again compares the (newly-revised) present state to the outcome.  If the outcome criteria have been met, the process ends; we exit the process.  If, however, the outcome has not been met, we loop back through the test-operate procedure until it does (Dilts, 1983; Dilts & Delozier, 2000; Dilts, Grinder, Bandler & Delozier, 1980; Wake, 2010).

As it appears in NLP, the model usually specifies that testing happens in the sensory modality most relevant to the issue.  A carpenter hammering nails might use sight or feel—a hammer hitting a nail off-center feels very different from one that has hit the nail correctly (Dilts et al., 1980; Dilts & Delozier, 2000). While it can be used to describe simple behaviors, like hitting a nail with a hammer, it can also be used to build complex hierarchical models of much more complex behaviors.  If we imagine that a basic TOTE can be used to assemble a set of rudimentary skills that are necessary to a larger task, we can imagine that increasingly larger and more complex tasks can become unified wholes using the same model.

The process of NLP modeling, whether in a therapeutic context or as used to make some skill transferrable, begins with the identification of a target behavior.  The outcome is to be able to replicate the behavior and this is often expressed; ‘Can you teach me how to do that?’ Whether the aim is therapeutic or transformational, the core idea is to find the patterns that directly give rise to the behavior. As with Grounded Theory, context is a crucial piece of information: When does this behavior occur or when is it appropriate?

The process continues with the client providing the best description his/her current state that they can.  The practitioner, in response, encourages the client to be more and more specific as to the sequence of sensory and motor elements, using the meta-model (a series of linguistic patterns that allow him/her to challenge unspecified, ill-specified  or skipped steps thereby enabling more detailed analysis of the elements). Questions also target the submodality structure of the various perceptions: was the picture big or small, close or far away, moving or still? Was the sound source small or large, where did the voice come from, was it loud or soft, what was the tonality, whose voice was it, and so on for the other senses. Engaging in conversation about the process, the modeler, continuously checks his understanding with the client to determine whether she has gotten it right and to discover where there may be gaps or misinterpretations, and the process continues as a dialogue. In addition to the above, the client’s mood, intent and any immediate precursors to the behavior should be identified.

There are variants of the modeling procedure so that if the individual being modeled has written books or produced recordings, as with Grounded Theory, they may be used to inform the process. In some cases, notably Bandler’s (Grinder & Pucelik, 2012) modeling of Perls (who was by then dead), and Dilts’ modeling of the thinking patterns of some of the great geniuses of our time (1995) the enterprise may continue without access to the living model.  There are some objections to this approach, (Bostic St. Clair & Grinder, 2001; Hall, 2001). Nevertheless the practice is affirmed by other writers.

In the NLP modeling process described here, there is a recursive dynamic, that seeks structural information, tests its validity, seeks missing information, integrates it into a growing provisional model and proceeds until the description is sufficiently accurate that the behavior becomes replicable or there is enough information about a non-useful response so that the pattern can be modified and tested for its effect on the problem. When used on the level of an individual treatment or skill model, the process is completed when the description is sufficiently complete so that an intervention or skill/behavior transfer is possible (Bostic St. Clare & Grinder, 2001; Dilts, 1998).

In some cases, notably, the RTM-V/KD protocol, rapport patterns and the NLP Spelling strategy, the patterns elicited from multiple individuals may be aggregated to create a technique of general utility. These kinds of interventions, which have essentially been validated by their successful application to multiple similar cases, can be subjected to standard qualitative evaluations. The less formal techniques, individualized treatments for depression, anxiety, OCD, and others may be evaluated by aggregating patient satisfaction responses (On a scale of one to ten, how well was your outcome met?) and may be enhanced by pre and post testing where an appropriate symptom scale is presented. See the section Basis for Evaluative Research at the end of this article for a complete description of this topic.

Case study—example

Let us start by looking at the standard clinical approach to depression. A client comes in and complains of being depressed.  Any competent psychologist or clinician begins to probe in order to determine the nature of the depression. Is it clinical, sub-clinical, event related, post-partum, or something else? An effort is made to diagnose the nature of the complaint in accordance with an acceptable DSM Five or ICD-10 category.  In so-doing the clinician explores with the client the dimensions of the condition.  Unfortunately, in modern psychiatry and clinical psychology, most of these analyses stop when an appropriate category is identified at which point either an appropriate course of medication is provide or a fairly routinized course of therapy is prescribed, with or without medication.

NLP, on the other hand, begins with a similar probing but does not necessarily assume that the diagnostic category is the essential end point of the process. NLP reimagines pathology in terms of client-specific maps and patterns of affect and behavior that can be changed. Those patterns are often idiosyncratic but their details can be discovered for each individual.

Having determined the nature of the complaint, NLP next explores potential client outcomes: What do they want to accomplish? What do they want to change? This process provides the desired state that will be used to test the effectiveness of the treatment (the Test portions of the TOTE model). Next, the therapist focuses on the structure of the problem: How do you do that? How do you “become” depressed (what is the structure)? When do you know it is time for this to happen (what is the trigger?)? Where does it happen most often (what is the context?)? The diagnostic process continues with the hypothesis that there is some structured set of perceptions and behaviors that is the direct source of the present-time complaint. That structure is presumed to be a set of sensory representations (internal sights, sounds, feelings, tastes and smells) that by a combination of their order and distortion, give rise to the problem behavior or response. This tentative structural dynamic is held provisionally as further levels of structure are examined. Each of these steps may become the focus at any time as the therapist works to ensure that she has created an accurate model.

Having determined the simple sequence of sensory-motor data that creates the response, NLP now looks for the specific submodality distinctions that give meaning and importance to the internal images, sounds, feelings tastes and smells and that give the behavior its special compelling quality. This level allows the clinician to make hypotheses about the internal dynamics of the pattern (the client’s present state) and begin to formulate a possible restructuring of those dynamics that would succeed in reshaping the behavior.

The final hypothesis is a plan of action for restructuring the perceptions, actions and the submodality structures that are the direct precursors to and/or responsible for maintenance of the problem. It too, however, is held provisionally and must be tested against the client’s map. If the client confirms the structure as valid, if they can use it to start an episode (which the therapist will be careful to end before it overwhelms the patient) the pattern receives a certain level of validation.  The final test appears as the therapist and client systematically modify a set of submodalities, or sequences in the structure of the behavior, that seem to lie at its heart.  This is the “operate” phase of the TOTE model.  In our case study, it represents the therapeutic intervention.  During this process, the new pattern is associated with past and future circumstances which may have evoked the behavior or would threaten to do so in the future.  This is frequently done using a fantasizing process such as:  “Imagine that when this happened back then, if you had structured it the way we just rehearsed, what would have been different?  If the client finds that his response would have changed in the past if this restructuring had been available, the model is partially confirmed.  The next step would be “Imagine the next time the ‘right time and place’ occur, and you structure it the way we just rehearsed, what happens?”  If the client, using the new structure cannot imagine a circumstance that would provoke the response in the present or the future, no matter how hard he tries, the hypothesized remedy is presumed to have been successful until tested again at follow-up. Should the intervention have failed between the last formulation and follow-up, new hypotheses and possible remedies are generated.

Insofar as it is susceptible to cognitive-behavioral intervention, one example of a DSM IV diagnosis of Severe Depression might be described in NLP practice (as determined by the process just described) as “an obsessive internal voice in the same tone and timbre as the client’s long-deceased, overbearing mother.” This diagnostic description provides the therapist with the client’s own internal representations which generate and maintain the depression, and presents specific submodality points that can be altered to alleviate it. Using this example, the therapist might typically invite the client to try several things such as: Move the source of the sound to a place (farther away, from another room, over the telephone, etc.) where it has less impact. Change the tone and timbre of the voice so that it sounds very different (like Mickey Mouse).  Make the voice distant and barely discernable.  Each of these, singly or in combination, could significantly reduce or even eliminate the client’s experience of depression and, with some imaginal practice, could produce a long-term remedy.

GT and NLP

At this point, the parallels between GT and NLP should be obvious.  Both begin with a target of investigation, establish rapport and begin a conversation with their informant or model about how they do the behavior of interest. In both cases there is an attempt to find certain basic categories of information:

1. What are the limits of the behavior or response (When does it start; when does it end? In what contexts does it arise?)

2. What are the essential elements of the phenomenon? In NLP this would be sequences of sensory experiences (motor responses are treated as a subtype of kinesthetic perception), their distortion by submodalities and whether there are recursive elements.  In GT this is the process of categorization: definition of the elements.

3. At every stage in both GT and NLP, any new information is tested against any current model or hypothesis of the phenomenon.

4. In both NLP and GT, as coherent models arise, outside information may be brought to bear in order to clarify the process from previous research. In NLP this includes the use of the meta-model, the structural assumptions about sensory modalities and submodalities as well as other basic elements and tools we have reviewed that are in the NLP tool kit.

5. Axiological analysis involves re-questioning the data and the informant as to the importance of various elements and assigning them sequence and value.  It also includes the assessment of bias or predisposition on the part of the investigator or modeler.

6. After multiple courses of testing and retesting the various levels of model-making against the data, there may come a stage of data saturation in which all possible elements have been defined.

   7. In both cases, the next step is either to perform the same investigation using other subjects, to see whether the same analysis arises from the data, or to

       test the model in other subjects to determine whether it can reproduce the target behavior. 

In this comparison it appears that whereas the early developers of NLP claimed to have no theory or that their practices could not be evaluated theoretically, they may have been unaware of the parallel development of GT during the same era. Viewing each behavioral analysis—the development of an NLP strategy — as a theory in the sense that GT uses the word, NLP becomes a substantive theory of the structure of subjective experience. If each model that is developed using NLP modeling techniques as described is creates a theory of individual perception and action, then the ability to replicate that behavior in others, or to end the negative symptoms of some pathological conditions, constitute validations of those theories.

Basis for Evaluative Research

Following the lead of GT, we strongly recommend NLP be subjected to large-scale evaluations of its techniques (generalizable theories and its basic postulates). For any individual NLP-based theory — the structure of a limiting behavior for one individual, the general utility of the theories (interventions) created by NLP, the research would utilize subjective scales such as the Likert Scale in order to measure relative satisfaction with the intervention. These evaluations would then become subject to statistical evaluation by standard means. Several such studies have already been done and can be reviewed in the work of Konefal and colleagues (1992, 1998), and more recently in the work of Stipancic, Renner, Schütz, and Dond R., (2010), and Genser-Medlitsch & Schütz (1997, 2004).

In addition, we suggest further testing of NLP-based techniques would utilize Comparative Effectiveness Research models. Here, the results from databases tracking the treatment course of a diverse population of patients is compared for its effectiveness as an intervention, and a cost-benefit analysis may be made. These kinds of studies are cost-effective and can be structured so as to provide high quality information about which treatment is better suited to the problem at hand. Typically, the approach monitors the responses of patients receiving different treatments for a similar problem or disorder. Data for comparison is extracted from a database of patients who have received the two treatments, and insofar as possible, the anonymous records are matched for medical and psychiatric history, socioeconomic status, and other relevant variables. The groups are then compared to determine which treatment provided the best and most cost-effective results. Once a database of NLP treatments has been amassed, this approach could produce significant information about the clinical effectiveness of the approach against more standard treatments (Begley, 2011; Morton, & Ellenberger, 2012; Neely, Sharon, Graboyes, Paniello, Nussenbaum, Grindler, & Dassopoulos, 2013). 

Further evaluations of NLP-based patterns may be applied in those cases where NLP has tested individually developed patterns (the Fast Phobia Cure — V/K-D — RTM procedure, the NLP Spelling Strategy, and the Rapport procedure) with larger groups to develop techniques of general application. These techniques, as theories on another level (using the NLP modeling processes previously described to identify the general structure of a specific phenomenon and how to remedy it or reproduce it), are susceptible to evaluation using standard quantitative measures. In these cases, significant clinical trials and or statistical evaluations have already been performed. The Fast Phobia Cure, also known as the RTM or Rewind Technique provided significant relief to 85 percent of a small sample of survivors of the Rwandan Genocide (Utuzza, Joseph & Muss, 2011). The same technique is currently the subject of a thirty subject RCT wait-group control study funded by the state of New York (www.Researchandrecognition.org). The Rapport pattern has been extensively supported in multiple studies (Gray, Wake, Andreas, & Bolsted, 2012). 

At this stage, however, a caveat must issue and that is this: once the NLP/GT process of basic theory generation is completed and a technique of general (or specific) applicability arises, any quantitative test performed on that pattern becomes a test of the pattern or technique, not NLP. In such cases, NLP as a species of GT has generated a theory or technique that stands apart from the process that created it and is tested as an application in its own right.

Axioms

Since its inception, NLP has been subject to empirical investigations. Because, however, there was no clear theoretical model of NLP which could guide the research, researchers found themselves testing patterns (such short-lived patterns previously discussed, like the PRS, eye accessing cues as related to the PRS, models like rapport and techniques such as the fast phobia cure). Because they were consistently assuming that these were the essential elements of NLP, the bulk of the researchers came back with an almost uniformly damning analysis (for reviews see Heap, 1998; Sharpley, 1984, 1987; Witkowski, 2010).

We have argued that NLP can be understood as a species of Grounded Theory. As noted by Fassinger (2005), Grounded Theory takes on many aspects, some related to Positivism, some to Constructivist and others more or less structured. Each variety may include a varying number of tenets that are more or less axiomatic. These may frame the entire inquiry or may be included into the modeling process in the phase of analytical questioning. NLP holds (however loosely) to certain basic principles that inform its search for patterns in creating models. In order to provide an accurate frame for the evaluation of NLP according to its own precepts, we offer the following list:

• All of its principles are held as patterns of behavior that are by their nature stochastic, they have a strong tendency to approach a certain value but that value is a statistical artifact (Bandler & Grinder, 1975; Bostic St. Clair & Grinder, 2001; Dilts & DeLozier, 2001).

• Most perceptions and behaviors can be reduced to the data of sensory experience (VAKOG). In this case movement is understood as a variety of kinesthetic perception (Bandler & Grinder, 1975; Andreas & Andreas, 1980, Bandler 1985; Lewis & Pucelik, 1990; O’Connor & Seymour, 1990).

• The structures of behavior are often recursive and self-referential and can be described in terms of Hall’s Metastates or the TOTE heuristic (Dilts, Grinder, et al., 1980; Hall, 2001)

• Sensory data are distorted by submodality distinctions that code for salience and affective valence (Andreas & Andreas, 1980; Bandler 1985; Gray, Wake, Andreas, & Bolsted, 2012; Lewis & Pucelik, 1990; O’Connor & Seymour, 1990).

• Structure and syntax are more important to the structure of perception and behavior than content (Andreas & Andreas, 1980; Bandler & Grinder, 1975, 1979; Dilts, Grinder et al., 1990).

• Internal representations of sensory data do not correspond to the external objects that they represent in an exact one to one relationship (the map is not the territory) (Andreas & Andreas; 1980, Bandler & Grinder, 1975, 1979; Lewis & Pucelik, 1990; O’Connor & Seymour, 1990).

• The Meta model of language is a primary tool for uncovering lost sensory experience from linguistic constructions (Bandler & Grinder, 1975; Dilts, Grinder, Bandler & Delozier, 1980).

• The presuppositions of NLP form a frame for communication and modeling which may be held as personal beliefs or as a working perspective (Andreas, 2006; Dilts & DeLozier, 2001, IASH & DeLozier, 2010; Lewis & Pucelik, 1990; O’Connor & Seymour, 1990).

These are some of the basic concepts that underlie the perspective of NLP. Although held loosely as noted, there is a large consensus regarding these foundational ideas. Any evaluation of the field of NLP must include some analysis of these axiomatic principles. The presuppositions and submodality distinctions include multiple propositions and assertions, most of which have already been supported by parallel research (Gray, 2013; Gray, Wake, Andreas & Bolsted, 2012). The root concepts of sensory modalities and submodalities have been present in psychology since Titchner’s Lectures published in 1909.