Predicate Matching in NLP

Predicate Matching in NLP: A Review of Research on the Preferred Representational System

Christopher F. Sharpley

Faculty of Education, Monash University, Clayton, Victoria, Australia

The increasing publicity of Neurolinguistic Programming (NLP) has not been accompanied by marked research support. As a first review of the 15 studies performed so far that have investigated the use of the Preferred Representational System (PRS) in NLP, this article describes each of these studies, compiling a summary of data collected. Aspects of design, methodology, population, and dependent measures are evaluated, with comments on the outcomes obtained. Results of this review suggest that there is little supportive evidence for the use of the PRS in NLP in these 15 studies, with much data to the contrary. Questions of accountability are raised, with suggestions for future research.

Beginning with their first publications (Bandler & Grinder, 1975; Grinder & Bandler, 1976), the proponents of Neurolinguistic Programming (NLP) have seen a notable increase of interest in their theory of human communication. Harman and O’Neill (1981) commented that “hardly a month goes by that we do not receive two or more notices of some kind of NLP seminar” (p. 449). Although applications of NLP have been principally oriented toward counseling (e.g., Eliasoph, 1981; Stevens, 1978), the fields of personnel training (Maron, 1979) and marketing (Brownlee, 1981) have also been suggested as benefiting from NLP. Yet, in spite of this increase in interest (and presumed application), no review has been published of research evidence supporting NLP’s claim as an effective intervention procedure for use by counselors and others seeking to facilitate human communication. The present article examines 15 such reports, performed to evaluate one of the basic tenets of NLP, and discusses outcomes from the viewpoint of variables of design, including methodology, subjects, and dependent measures. Findings are evaluated, with particular reference to the counselor in the field.

Neurolinguistic Programming (NLP)

and the Preferred Representational System (PRS)

Based on their observations of “therapeutic wizards” (Bandler & Grinder, 1979, p. 3) such as Satir, Erickson, and Peris, Bandler and Grinder proposed that persons process reality through five sensory and representational systems: visual, auditory, kinesthetic, gustatory, and olfactory (Dilts, Grinder, Bandler, Bandler, & Delozier, 1980), although this theory is applicable to right-handed persons only. In our culture it is the first three of these systems that are most used, with individuals seldom using only one system for all interactions with physical reality and showing variability in the prevalence they have for each system. Although persons can distort their experiences of or reactions to reality, Bandler and Grinder (1975) suggested that counselors can define the “deep structure” of clients’ thinking processes by reference to the “surface structure” of their verbal and nonverbal responses to the elements of reality that are discussed within a counseling interview. This “surface structure” can be classified into (most commonly) visual, auditory, or kinesthetic representational systems by careful observation of client eye movements or verbalizations. The system that an individual uses most of the time is termed the Preferred Representational System (PRS), and Bandler and Grinder suggested that matching this PRS is the key to effective counseling. The presence of PRS, procedures for identifying it, and the effectiveness of matching or otherwise utilizing the PRS of a selected individual is the focus of the studies reviewed here, with results either supporting or failing to support this basic tenet of NLP. Although it may be that the “therapeutic wizards” referred to by Bandler and Grinder do in fact match the PRS of their clients, it does not follow that such matching is a necessary or sufficient condition for effective communication, particularly in counseling situations.

The Studies Reviewed

A detailed examination of the literature by computer search and cross-referencing revealed a total of 15 studies investigating NLP, all focused on the issue of the PRS of individuals and matching of it to achieve more effective communication for a variety of purposes. Each of the studies is described in this section, with a summary for comparison drawn up in Table 1. The studies are presented in the following development: (a) examinations of the presence of PRS, (b) the congruence of PRS as detected via differing measures, (c) the use of matching PRS for noncounseling intervention, (d) the effectiveness of matching PRS in counseling interviews.

Presence of the PRS

Prior to discussing findings related to the effectiveness of PRS matching with clients, the issues of verifying the existence of the PRS and then evaluating the reliability of methods suggested to identify the PRS arise. Only one study has attempted to verify the presence of PRS in individuals over time. Birholtz (1981) examined the preferences that 27 college students showed for words that reflected one sensory modality over others. The students were asked to describe positive and negative experiences in their past, present, and projected future. Questions used to elicit these responses were phrased in a neutral modality and were audio-taped. Results indicated that there was one preferred (p < .01) mode for all subjects and this was kinesthetic, although there was no significant correlation between this and subjects’ self-report of their PRS. These results were replicated 1 week after the initial interview, suggesting that a preference for kinesthetic predicates did exist and was stable over 1 week. Whether this preference is the result of neurological “style” or societal influence is open to question. However, it does comply with the NLP theory and as such offers some support for the contention that persons possess PRS and these systems are stable over 1 week.

Identifying the PRS

One of the earlier pieces of research performed on NLP was Owens’s (1977/1978) investigation of congruity of PRS as identified by eye movements, verbalizations, and self-report. Owens posed nine stimulus cues that required subjects to respond according to their PRS. Counts of eye movements and verbalizations were performed by independent raters on 128 undergraduate psychology students, and all subjects were later asked to report on their own PRS by answering a short questionnaire. Results of all ratings were classified as either visual, auditory, or kinesthetic on each of the three identification procedures and data were analyzed for agreement between these procedures. No significant correlations were noted (see Gumm, Walker, & Day, 1982, p. 329, for a note on these results), thus failing to verify NLP’s theory of preferred representational systems as dominant across modalities! Although not in direct contradiction to Bir- holtz’s (1981) suggestions that the PRS does actually exist, Owens’s results seriously question the process by which these PRS can be identified, and therefore the chances of implementing any process of matching PRS to facilitate communication.

A similar study was performed by Gumm et al. (1982) in which they also failed to verify NLP predictions of agreement between three methods of determining PRS. In a study of congruity between eye movements, verbalizations, and self:report with 50 female college students confirmed as right-handed, subjects were asked to speak for 1 min in response to each of five questions and answer




  Subject Age   Clinical  
Focus of Study F M (years) Setting features Dependent variable
Presence of stable PRS          
Birholtz (1981) — 27— NR (college University language Nil (a) Response to audiotape
    students) laboratory   (b)  Repeated, 1 week later

(c)  Self-report

Congruity of PRS assessed on          
EMs, verbalizations, selfreport          
Owens (1977) —128— NR (college Psychology Nil PRS agreement across EMs,
    students) laboratory   verbals, self-report
Guram, Walker, & Day 60 0 NR (college NR Nil (right-handed PRS agreement across EMs,
(1982)   students)   only)® verbals, self-report
Congruity of sensory-specific          
information and EMs Beale (1980) — 40— NR (college University language Nil EMs with visual, auditory,
    students) laboratory   kinesthetic content
Thomasen, Arbuckle, & — 40— 18-25 (college Psychology Nil (right-handed EMs with visual, auditory,
Cady (1980)   students) laboratory only)® kinesthetic content
Hernandez (1981) 44 20 18-51 (college Psychology Nil (right-handed EMs with visual, auditory,
    students) laboratory only)® kinesthetic content
PRS as hypnotic inductor          
Yapko (1981) 17 13 NR (college Hypnosis class Nil Frontalis EMG
PRS for recall          
Shaw (1977) —108— NR (college Psychology Nil Number of facts recalled from
    students) laboratory   story
Counselor predicate matching          
vs. mismatching for counselor effectiveness          
Paxton (1981) — 48— 26-35 Clinic Intake patients BLRI


Falzett (1981) 24 0 NR (college Psychology Nil (right-handed Trustworthiness scale of CRF
      students) laboratory only)®  
Ellickson (1981) ‘ 36 36 NR (college Psychology Nil (right-handed (a) BLRI (Empathy scale)
      students) laboratory only)® (b)  Ease of Communication Inventory

(c)  Multiple Affect Adjective

            Check List
Frieden (1981) 2 0 21,22 Psychology Seeking counseling (a) BLRI (client)
        laboratory   (b) Rating of verbal and
            nonverbal client behaviors (observers)
            (c) TCS (client)
Dowd & Pety (1982) 50 34 20-35 (college Human Relations Nil (a) CRF
      students) class   (b) CEI
            (c) Willingness to see a
Dowd & Hingst (1983) _ 54— NR (college Human Relations Nil (a) CRF
      students) class   (b) CEI

a NLP only relevant to right-handed persons (Handler & Grinder, 1979). * p < .001. (Notes continued at end of table.)

How PBS determined Supportive Outcomes


Uncertain Comments
Verbalization (a)  PBS significantly > chance

(b)  Stable over 1 week*

(c) PBS not matched by selfreport   Verified presence of PBS as stable over 1 week; subject not aware of own PBS.
EMs, verbalizations, self* report

EMs, verbalizations, self* report

  No significant agreement between any two

No significant agreement between any two

  Failed to verify NLP methods of determining PBS as congruent

Failed to verify as congruent NLP methods of determining PBS


EMa   No significant correlation of   Failed to verify NLP sensory-
    EMs with content of   system categories
EMs   No significant correlation of   Failed to verify NLP sensory-
    EMs with content of   system categories
EMs Visual PRS congruent Kinesthetic PBS not Auditory PRS partially Partial verification of NLP
    congruent congruent sensory-system categories
Verbalization PRS more effective than     Verified use of PRS for
  other RS for relaxation     relaxation instructions
Verbalization   No significant effect due to   Failed to verify PRS as
    PRS   assisting recall


Verbalization before   No significant difference Matching and Failed to directly support PRS

EMs before interview

Matching > mismatching on PRS* between matching vs. mismatching PRS mismatching significant superior to nonmatching matching -*• counselor effectiveness

Verified NLP that PRS matching -*■ counselor effectiveness (variable of trustworthiness in CRF)

EMs during interview Significant effect (ease) due to PRS matching for males No significant effect due to PRS matching for females   Failed to support PRS matching -*■ counselor effectiveness as independent of sex


Counselor matched/ mismatched

Client statements during interview

(b)  Eye contact increased

(c)  Improvements on TCS

(a)  BLRIns

(b)  Counselor-Client head distance increased

Partial support for PRS matching counselor effectiveness (eye contact, target symptoms only) ‘
Counselor matched/,   (a) CRF ns Failed to support PRS
mismatched     matching -*• Counselor
    (b) CEI ns . effectiveness;
      subjects not clients
Client statements   (c) No significant change in Failed to support PRS
during interview   willingness for PRS matching -* counselor
    matching effectiveness
Counselor matched/   (a) CEF ns  
Client statements during   (b) CEI ns  

(table continued)

  Subject   Clinical


Focus of Study Age .

F M (years)

Setting Dependent variable
Hammer (1983) 63 0 Ы — 19.8 (college students) Psychology


Nil (a)       BLRI (Empathy scale) for clients

(b)  IES for counselors

Note. All studies took place in the United States. F = females; M = males; PRS = Preferred Representational

Counselor Rating Form; CEI = Counseling Evaluation Inventory; IES = Interviewer Experience Scale; EMG


a 24-item self-report questionnaire. Eye movements were recorded on videotape during and between each of these tasks, and later rated by two independent and naive observers. Agreement between any two of these methods of categorizing PRS was not significant at the .05 level. This study therefore verifies Owens’s (1977/1978) study in finding no evidence of agreement between any of the three NLP methods of determining the PRS, and similarly seriously questions the likelihood of implementing PRS- matching processes.

Three studies that examined the relationship between eye movements and PRS were performed by Beale (1980/1981), Thomason, Arbuckle, and Cady (1980), and Hernandez (1981). In the first of these, Beale (1980/1981) examined the congruence of sensory-specific information with eye movements for 40 undergraduates. Videotaping of subjects’ eye movements in response to 24 stimulus items was used to identify each PRS as visual, auditory, or kinesthetic. Data showed that eye movements were in an upward direction regardless of stimulus changes in sensory content, thus failing to verify this NLP basic tenet.

Thomason, Arbuckle, and Cady (1980) investigated the congruity of sensory-specific information with recorded eye movements of 40 right-handed psychology undergraduates. Subjects’ responses to 30 questions (10 each with visual, auditory, and kinesthetic content) were coded by three naive observers, and the level of congruity between eye movements and question content calculated. Results showed that most eye- movement responses were visual, thus failing to verify the NLP prediction as to number of responses.

Finally, Hernandez (1981) tested for congruity between subjects’ eye-movement responses to 24 statements previously coded as visual, auditory, kinesthetic, or nonspecific and the content of the statements, finding partial verification of NLP theory. Subjects were 64 undergraduates (44 women). Visual statements showed significant correlations with visual-category eye movements; only half of the auditory statements resulted in auditory eye movements; none of the kinesthetic statements were correlated with kinesthetic eye movements in subjects, thus agreeing with Beale (1980/1981) and Thomason et al. (1980) in failing to provide conclusive support for NLP on this issue of congruity between eye movements and PRS. The usefulness of eye movements to identify the PRS (assuming it exists) is seriously in doubt. Lack of agreement between the three measures used in these studies (i.e., eye movements, verbalizations, self-report) does not in itself refute the existence of the PRS, but suggests that only one (if any) measure can be accurate. No evidence to date strongly supports any one of these three measures.

Matching PRS for Noncounseling Interventions

Yapko (1981) compared the relative effectiveness of hypnotic induction via verbal instructions in the subject’s preferred modality with similar instructions in a nonpreferred modality. Three representational modalities were used (visual, auditory, and kinesthetic) on audio recordings standardized for content and delivery that were intended “to influence subjects to physically relax” (p. 171). Prior to induction for each of the 30 university students (17 females) who were all volunteers with no noted clinical features, the preferred modality of each subject was determined by verbalizations made to open-ended questions. Frontalis electromyograph (EMG) readings were


How PRS determined Supportive Outcomes


Uncertain Comments
Counselor matched/ mismatched

Client statements during interview

(a) Significant effect (empathy) due tp PRS matching (b) No significant effect for PRS matching on IES   Partial support for PRS matching for client- empathy response

System; EMs = eye movements; NR = not reported; BLRI = Barrett-Lennard Relationship Inventory; CRF = * electromyograph; TCS = Target Complaint Scale; NLP = Neurolinguistic Programming.


taken every 30 s for 5 min baseline and for each of the three audiotapes of relaxation instructions (i.e., visual, auditory, kinesthetic). Data were analyzed to detect differences between EMG levels and baseline and preferred versus nonpreferred PRS for subjects and modalities, and indicated that subjects’ EMG levels were lowest when relaxation instructions were delivered in their preferred modality as identified by verbalizations (F = 3.55, df = 29, p < .05).

The second of these two studies of the effectiveness of PRS was performed by Shaw (1977/1978), who examined the hypothesis of preferred modality (or PRS) by asking 108 undergraduate students to recall facts from a videotaped story. The story was presented in three modalities to match the three PRS of subjects that had been previously determined by verbalization (visual, auditory, kinesthetic), and recall was hypothesized to be superior when the story was presented in the subject’s PRS. Although not a complete evaluation of all three PRSs (none of the subjects had a visual PRS), the hypothesis was rejected when auditory and kinesthetic subjects did not respond differentially to the three modalities. For the purposes of recall, the NLP theoretical tenet of PRS matching as identified via verbalizations was not verified in this study.

Matching PRS in Counseling

Paxton (1980/1981) investigated the effects of counselor verbal style on client perceptions of the counseling relationship with 48 intake clients at a Family and Children’s Service Clinic. All subjects were volunteers who agreed to participate in an evaluation of the agency’s services and were reported as mostly women between 26 and 35 years of age. The PRS of each of these subjects was determined by verbalization prior to intervention and subjects were randomly assigned

to a PRS-matching, -mismatching, or -nonmatching treatment. After the counseling interview, subjects rated the counseling relationship on the Barrett-Lennard Relationship Inventory. Results indicated that both matching and mismatching of PRS treatments were significantly superior to nonmatching PRS (p < .01), but there was no significant difference between matching and mismatching treatments on the variable of client perception of the counseling relationship. The author concluded that the frequent use of sensory modality predicates enhances counseling, but there is no support for matching the specific PRS of clients in counseling.       v

In a study examining the NLP suggestion that trust in a relationship will be enhanced if the counselor matches the PRS of the client, Falzett (1981) had counselors match or mismatch predicates with 24 right-handed female volunteer college students whose PRS had been determined by eye-movement responses to questions prior to predicate matching. Subjects rated the counselor on the Trustworthiness scale of the Counselor Rating Form, with results indicating a significantly higher level of perceived trustworthiness when counselors matched predicates with clients (p < 1001). These data should be treated with caution because the measure used to identify the PRS was eye movements—previously shown to be unreliable for this task. Extraneous variables may have been causal in this study.

Both Paxton (1980/1981) and Falzett (1981) categorized clients’ PRS prior to the predicate-matching process and had counselors respond to this PRS for the length of the interview. The remaining five studies reviewed in this section did not categorize clients’ PRS in this manner but had counselors respond to the specific predicates used by clients during the interview (except El- lickson, 1980/1981, who had counselors respond to clients’ eye movements during the interview). As such, this represents a departure from the “identify for matching a priori” procedures used in the previously reviewed literature toward the use of specific cues for matching during the interview. Concerning the use of verbal cues, this may not be a test for NLP so much as an evaluation of the effectiveness of empathic verbal responding.

Thirty-six men and 36 women volunteer undergraduates were randomly assigned to a PRS-matching or PRS-mismatching interview by Ellickson (1980/1981) in a study designed to evaluate the effects of PRS matching on subjects’ responses on the Empathy scale of the Barrett-Lennard Relationship Inventory, an author-made Ease of Communication Inventory and the Today form of the Multiple Affect Adjective Check List. The PRS of subjects were determined during the counseling interview by client eye movements in response to questions and all subjects were reported as right-handed. Although females showed no significant effects due to matching/mismatching of PRS as determined by eye movements, males were more (p = .012) at ease in the matching than in the mismatching condition, suggesting sex differences in the usefulness of this NLP procedure in counseling, although the use of eye movements to identify the PRS of subjects casts doubt upon whether the PRS-matching process here could have been accurate and therefore whether it was predicate matching, which was the causal variable in male responses.

In order to examine the effects of predicate matching versus mismatching on client perceptions of counselor trust, the level of communicative behaviors and therapy outcome, Frieden (1981) applied predicate matching to an actual counseling intervention with two female college students who were seeking help for “personal problems” (p. iv). Averaged data showed that, both eye contacts and head distance increased; no significant changes were evident in client perception of counselor trustworthiness via the Barrett-Lennard Relationship Inventory; and both clients showed a decrease in symptoms on a Target Complaints Scale. Although the improvement in client perception of counselor trustworthiness and decrease in target symptoms are supportive of the predicate-matching process, the data regarding the increase in eye contact and head distance appear to contradict each other. Increases in eye contact generally indicate increased rapport, whereas head distance usually decreases with rapport. Frieden suggested that these data provide “no unequivocal support” for NLP (p. 85) due to marked individual differences in subject responses that were not revealed in averaged data. Of note is the use of ongoing predicate-matching procedures rather than prior detection of the PRS. Counselors were not told which PRS clients had, but were asked to “track” client verbalizations within the interview, responding in the same modality as used by the client. This process represents a significant step toward isolating the effectiveness of predicate matching from the issue of accurate identification of the PRS prior to interview. This study is notable also for the use of the Target Complaints Scale designed to detect changes in symptoms and is the only study to date that has addressed the issue of change in client behavior, although again by use of paper- and-pencil test rather than actual observations.

When evaluating the effectiveness of predicate matching versus mismatching on (a) social influence of counselor, (b) perceived client satisfaction, and (c) willingness of subjects to see a counselor who was portrayed in a series of audiotapes, Dowd and Pety (1982) found no significant effects due to matching of counselor predicates with client predicates. Subjects were 84 undergraduate Human Relations students (50 female) who volunteered to take part in a study “dealing with communication in counseling” (p. 208). Subjects were asked to think of a personal problem they were experiencing and then rate their willingness to see a counselor about that problem on a 1-10 scale. They then listened to an audiotape that depicted a female student who was experiencing friendship-making difficulties and who was working with either a male or female counselor who either matched or mismatched predicates with the client on 24 of 44 interchanges. Subjects were then asked to rerate their willingness to


see the counselor they had heard and to evaluate him or her on the Counselor Rating Form (Barak & LaCrosse, 1975) and assess total counseling interaction and client satisfaction on the Counseling Evaluation Inventory (Linden, Stone, & Shertzer, 1965). Analysis of covariance on postinterview willingness to see the counselor showed the male counselor to be more highly rated (F = 8.36, df = 1,74, p < .005), but no significant differences were noted on the Counselor Rating Form, the Counseling Evaluation Inventory, or general willingness to see a counselor. Although this study (like Frie- den, 1981) also used matching of actual client predicates by counselor predicates during the interview rather than to a PRS as predetermined by eye movements, self-report, or verbalizations, subjects in this study were not part of a counseling interview themselves (they listened to a recording of an interview) and as such this study examines a different situation to those from the other six studies in this section wherein subjects were clients. Because of this, these data may be of questionable value in resolving the issue of the effectiveness of predicate matching, but in any case are not supportive of NLP on this point.

In a study designed to evaluate the effects of predicate matching on clients who were also subjects, Dowd and Hingst (1983) used 54 volunteer undergraduates as “client- subjects” (p. 208), arguing that this represented an investigation of the effects of predicate matching, mismatching, and nonmatching on “clients’ perceptions of therapists’ social influence in an actual therapeutic interview” (p. 208). Therapists were six postgraduate counseling students who were trained in predicate matching, and who were instructed to match, mismatch, or norimatch client predicates during the interview. These therapists were monitored via bug-in-the-ear devices during interviews to ensure that they maintained their correct treatments. Clients completed the Counselor Rating Form and the Counseling Evaluation Inventory immediately after the interview, with results indicating no significant differences due to treatment, although there was a trend that favored the match group over the mismatch group. Of interest was the finding that of all three treatments, the no-match group scored highest on all dependent variables. Dowd and Hingst suggested that the spontaneity allowed to therapists in this treatment may account for this finding, further indicating that extraneous variables may have accounted for “treatment effects” in this study. As an adjunct to the previous study by Dowd and Pety (1982), these results may be considered as more valid due to the collection of data from persons who were actually experiencing predicate matching rather than listening to others do so, The nonsupportive outcomes of this study cast doubt on the efficacy of predicate matching, either first- or secondhand.

Hammer (1983) also reasoned that the PRS of clients could more accurately be evaluated during the counseling interview rather than prior to it and therefore trained therapists to respond to the ongoing predicates of clients rather than to only one PRS. Sixty-three female undergraduates were volunteers in this study, with three female postgraduate counseling student therapists either matching or mismatching client predicates according to which of the six treatments was used (3 therapists X matching/mismatching). Client perception of counselor empathy was measured by the Empathy Scale of the Barrett-Lennard Relationship Inventory, and counselors completed the Interviewer Experience Scale (constructed by Hammer) to determine the presence of confounding effects due to intersubject differences. There was no significant difference between treatments on the Interviewer Experience Scale, but clients in the matching predicates condition scored their counselors higher on perceived empathy than those in the mismatching condition (p < .05). Client means for predicates used showed auditory to be most common (22.78), followed by kinesthetic (19.40) and visual (9.71), although Hammer noted that “the difference in frequency between the PRS and the next most frequently used type was less than or equal to 5 predicates” (p. 177) in 50% of subjects, suggesting that the prediction of there being one PRS may not be accurate, and further supporting the use of ongoing predicate matching over prior determination of clients’ PRS.

These 15 studies are summarized in Table

1,  and reference to variables depicted there indicates that although there are several specific findings that provide support for NLP, the majority are either nonsupportive (17/29) or uncertain (3/29), with only nine of these findings (i.e., less than one third) in support of NLP on this issue of the PRS and its use.


The implementation of predicate matching as suggested by NLP raises the issue of identification of the PRS of clients. By reference to Table 1, it is apparent that three procedures are suggested, one under two conditions. First, many researchers used an a priori evaluation via eye movements, verbalizations, or self-report. The findings from Owens (1977/1978) and Gumm et al. (1982) that there was no significant agreement between any two of these procedures suggest that at least two are unreliable and question the value of any of the three. They are not interchangeable, with verbalizations emerging as the only procedure that shows any reliability (Birholtz, 1981). Perhaps least accurate is eye movements (strongly advocated by NLP), with data from Beale (1980/1981), Hernandez (1981), and Thomason et al. (1980) failing to verify this particular process as better than chance in identifying the modality of utterances or the PRS of subjects. From this, the findings of studies that used eye movements to establish the PRS prior to matching predicates must be questioned. It is of note that two of the studies that support predicate matching in counseling (Ellickson, 1980/1981; Falzett, 1981) both used eye movements to determine client’s PRS (Falzett prior to interview, Ellickson during interview), whereas the remaining five studies that used verbalizations as cues to predicate matching collected data that primarily failed to support NLP on the issue of predicate matching.

Second, verbalizations were also used both prior to interview and during interview. When used prior to interview to determine clients’ PRS (Paxton, 1980/1981), results were not supportive of predicate matching. When counselors tracked clients’ predicates during interview, varying the modality of their responses to the modality of client statements (Dowd & Hingst, 1983; Dowd & Pety, 1982; Frieden, 1981; Hammer, 1983), only partial support was found on empathy (Hammer, 1983), eye contacts, and some target symptoms (Frieden, 1981), with no significant ease of interviewing reported by counselors, nor significant effects on the Counselor Rating Form or the Counseling Evaluation Inventory (Dowd & Hingst, 1983; Dowd & Pety, 1982). Hammer (1983, p. 178) made the point that “consistently and effectively matching any behavior, regardless of the specific cue” may be the causal variable here. Empathy has been referred to by many authors as one of the primary ingredients of effective counseling (e.g., Ivey & Simek-Downing, 1980; Rogers, 1957) and if NLP is suggesting that counselors who demonstrate high levels of reflection and empathy will be more effective than those who do not, then little new is being said. If NLP seeks to promote empathic responses from counselors, then scales designed to measure empathy ought to, and do, show this (e.g., Hammer, 1983). Although this is a worthwhile procedure for counselors, it does not justify NLP as a separate theoretical position (nor as the “magic” its proponents quote).

Table 1 shows that for the seven studies that attempted to evaluate predicate matching in counseling, the only type of measure used was pencil-and-paper questionnaire, with only one study reporting data from an instrument designed to detect changes in client behaviors (Frieden, 1981), data that showed “no unequivocal support for NLP” (p. 85). This limited range of dependent measures used in the studies reported to date leaves the therapeutic benefit of NLP to clients as yet undemonstrated. Data from the Counselor Rating Form, the Barrett-Lennard Relationship Inventory, and the Counseling Evaluation Inventory, when used with subjects who are nonclinical and within a laboratory situation do not argue for the effectiveness of predicate matching as therapy, even if these data were supportive. Because they are principally nonsupportive of this process it may be concluded that, at best, predicate matching has yet to be shown to effectively enhance either the counseling interview or client outcomes.

Although proponents of NLP maintain that predicate matching is useful only with right-handed individuals, few of the studies reviewed have stipulated that this was a precondition for subject inclusion. Of the seven studies that evaluated the effectiveness of predicate matching within counseling situations, only Ellickson (1980/1981) and Falzett (1981) reported subjects as righthanded. (Unfortunately, both of these studies used eye movements to determine PRS and therefore results must be treated with caution.) Further, of these seven studies, three used only females as subjects, two were mixed, and two did pot report sex of subjects explicitly, The issues of right- handedness and sex of subject require further investigation, particularly because Hammer (1983) found significant support for predicate matching with 88 females, whereas Ellickson (1980/1981) found no significant effects for females, but did find significant effects for males.

Although the overall conclusion reached here is that the NLP process of predicate matching has yet to be conclusively demonstrated as either effective or idiosyncratic to NLP, several minor points needed to be made regarding the research methodologies used to date.

  1. Although it is not a criticism of these studies, the fact that all of them have been conducted in the United States without any cross-cultural comparisons even within that setting, leaves the generalizability of NLP as an open issue, even if all research data were completely supportive of this theory.
  2. Similarly, young (mostly aged 18-25) students do not represent either the general or the clinical population, and data obtained from such persons, even if “volunteers,” should be treated with caution. Of the 834 subjects reported in Table 1, only 50 (5.9%) were in any way “clinical,” although “seeking counseling for personal problems” (Frieden, 1981, p. iv) on the part of two women in their early twenties (21,22) offers few implications to the general counselor who usually will be confronted with a wider variety of ages and complaints. (Frieden’s two clients “did not consider their problem to be a crisis requiring immediate attention,” p. 34.) Paxton (1980/1981), while utilizing a typical clinical sample, did not find significant support for

predicate matching.

  1. The use of laboratories in Psychology Departments as settings for data collection is not unusual in research of this nature, but raises the issue of the generalizability of findings once again.
  2. Finally, even though 445 subjects were not reported as to sex, the remaining 389 were predominantly women (286 or 73.5%), and all were presumed right-handed although not always reported as such. One may hesitate to generalize from a sample of right-handed female college students to the general or clinical population.


Perhaps the PRS does exist. Birholtz (1981) provided some evidence based on verbalizations collected over 1 week to demonstrate that college students do prefer one set of sensory predicates. However, the identification of this PRS (if it is a PRS and not merely current language style) by either eye movements or self-report is not supported by the research data. The cuing effect of client verbalizations is valuable, not to identify PRS but to alert counselors to phrase their responses in such a way as to maximize empathy within the interview. The existence or stability of the PRS is irrelevant to predicate matching as a counseling process, and parsimony argues for the process rather than the as yet unverified theory. This point was best demonstrated by Paxton (1980/1981), when the use of predicates was shown to be more effective than nonuse, regardless of matching or mismatching. Use of eye movements or self-report is not justified either as an indicator of the (questioned) presence of the PRS or as a cue to predicate matching by counselors.

At present, there is no consistent support for the use of the predicate-matching process of NLP in either contrived counseling situations or actual clinical realities. Of most importance, there are no data reported to date to show that NLP can help clients change. A series of controlled studies using reliable indicators of change in clients’ behavior (rather than their perceptions of counselors, which may not be correlated with problem dissolution by clients) is called for.

No psychotherapeutic procedure can claim credibility until it has shown its effectiveness with persons who present themselves for counseling. Although the users of NLP may believe it to be valuable and enjoy applying it, serious questions need to be decided by controlled studies with typical clients. Issues of accountability are of vital importance in the mental health field and the use of NLP because it is comfortable to counselors alone is hardly justifiable. Bandler and Grinder (1979, p. 18) stated that “We’re not offering you something that’s true, just things that are useful… we’re only interested in what works.” The present review suggests that this claim has yet to be verified.


Bandler, R., & Grinder, J. (1975). The structure of magic I. Palo Alto, CA: Science and Behavior Books.

Bandler, R.,& Grinder, J. (1979). Frogs into princes.

Moab, UT: Real People Press.

Barak, A., & LaCrosse, M. B. (1975). Multidimensional perceptions of counselor behavior. Journal of Counseling Psychology, 22,471-476. Beale, R. (1981). The testing of a model for the representation of consciousness (Doctoral dissertation, Fielding Institute, 1980). Dissertation Abstracts International, 41, 3565B. (University Microfilms No. 8106799)

Birholtz, L. (1981). Neurolinguistic Programming: Testing some basic assumptions (Doctoral dissertation, Fielding Institute, 1981). Dissertation Abstracts International, 42, 3565B. (University Microfilms No. 8118324)

Brownlee, M. (1981). Communicating research: NLP—a highway of diamonds with (almost) no-one on it? Journal of Australian Marketing, Advertising and Communication, 3, 42-50.

Dilts, R., Grinder, J., Bandler, R., Bandler, L., & Delo- zier, J. (1980). Neurolinguistic Programming (Vol. 1). Cupertino, CA: Meta Publications.

Dowd, T., & Hingst, A. (1983). Matching therapists’ predicates: An in vivo test of effectiveness. Perceptual and Motor Skills, 57, 207-210.

Dowd, T., & Pety, J. (1982). Effect of counselor predicate matching on perceived social influence and client satisfaction. Journal of Counseling Psychology, 29, 206-209.

Eliasoph, E. (1981). Practical magic—a translation of basic Neurolinguistic Programming into clinical psychotherapy. Journal of Group Psychotherapy, Psychodrama and Sociometry, 34,148-151. Ellickson, J. (1981). The effect of interviewers responding differentially to subjects’ representational systems as indicated by eye movement (Doctoral dissertation, Michigan State University, 1980). Dissertation Abstracts International, 41, 2754B. (University Microfilms No. 81-01102)

Falzett, W. C. (1981). Matched versus unmatched primary representational systems and their relationship to perceived trustworthiness in a counseling analogue. Journal of Counseling Psychology, 28, 305-308.

Frieden, F. (1981). Speaking the client’s language: The effects of Neurolinguistic Programming (predicate matching) on verbal and nonverbal behaviors in psychotherapy: A single case design (Doctoral dissertation, Virginia Commonwealth University, 1981). Dissertation Abstracts International, 42, 1171В. (University Microfilms No. 8118960)

Grinder, J., & Bandler, R. (1976). The structure of magic 11. Palo Alto, CA: Science and Behavior Books.

Gumm, W., Walker, M., & Day, H. (1982). Neurolinguistics Programming: Method or myth? Journal of Counseling Psychology, 29, 327-330.

Hammer, A. (1983). Matching perceptual predicates: Effect on perceived empathy in a counseling analogue. Journal of Counseling Psychology, 30, 172-179.

Harman, R.L.,& O’Neill, C. (1981). Neurolinguistic Programming for counselors. Personnel and Guidance Journal, 59, 449-453.

Hernandez, V. (1981). A study of eye movement patterns in the Neuro-Linguistic Programming model (Doctoral dissertation, Ball State University, 1981). Dissertation Abstracts International, 42, 1587B. (University Microfilms No. 8120505)

Ivey, A., & Simek-Downing, L. (1980). Counseling and psychotherapy: Skills, theories, and practice. Englewood Cliffs, NJ: Prentice-Hall.

Linden, J., Stone, S., & Shertzer, B. (1965). Development and evaluation of an inventory for rating counseling. Personnel and Guidance Journal, 44, 267-276.

Maron, D. (1979). Neurolinguistic Programming—the answer to change. Training and Development Journal, 33,68-71.

Owens, L. (1978). An investigation of eye movements and representational systems (Doctoral dissertation, Ball State University, 1977). Dissertation Abstracts International, 38, 4992B. (University Microfilms No. 7803828)

Paxton, L. (1981). Representational systems and client perception of the counseling relationship (Doctoral dissertation, Indiana University, 1980). Dissertation Abstracts International, 41, 3888A. (University Microfilms No. 8105941)

Rogers, C. (1957). The necessary and sufficient conditions of therapeutic personality change. Journal of Consulting Psychology, 21, 95-103.

Shaw, D. (1978). Recall as effected by the interaction of presentation representational system and primary representational system (Doctoral dissertation, Ball State University, 1977). Dissertation Abstracts International, 38, 5931A. (University Microfilms Order No. 7803830)

Stevens, B. (1978). Gestalt and neurolinguistic programming. Gestalt Journal, 1,89-91.

Thomason, T. C., Arbuckle, T., & Cady, D. (1980). Test of the eye movement hypothesis of neurolinguistic programming. Perceptual and Motor Skills, 51, 230.

Yapko, M. (1981). Neurolinguistic Programming, hypnosis, and interpersonal influence (Doctoral dissertation, United States International University, 1981). Dissertation Abstracts International, 41, 3204B. (University Microfilms No. 8103393)

Received October 24,1983 ■