You are on page 1of 17

EUROPEAN JOURNAL OF BEHAVIOR ANALYSIS 2009, 10, 31 - 47 NUMBER 1 (SUMMER 2009)

31

Some thoughts on the relation


between derived relational responding
and verbal behavior
Jay Moore
University of Wisconsin-Milwaukee

The present paper critically examines the bold claims of Relational Frame Theory (RFT) advocates that
RFT is a comprehensive approach to the phenomena referred to in traditional parlance as language and
cognition, and is manifestly preferable in both scope and detail to that found in B. F. Skinner’s book
Verbal Behavior. Although some data do indicate a high positive correlation between derived relational
responding and verbal behavior, in keeping with RFT, other data indicate at best a low correlation.
The reasons for the differences between expected and actual correlations across the several data sets
are not clear. We conclude that despite the value of RFT, the nature and causes of derived relational
responding, as well as the relation between derived relational responding and verbal behavior more
generally, remain an important area of investigation.
Keywords: relational framing theory, verbal behavior, mutually entailed relations, combinatorially
entailed relations, transformation of stimulus function, autoclitic behavior

Hayes, Barnes-Holmes, and Roche (2001, p. B. F. Skinner regarded his book Verbal Be-
viii) stated that after several years in incipient or havior (Skinner, 1957) as his most important
inchoate status, relational frame theory (here- work. In its more charitable moments, of which
after, RFT) was first presented to the behavior there are few, RFT champions itself as “post-
analytic community in any detailed way in 1985 Skinnerian,” applying the empirical principles
(Hayes & Brownstein, 1985). Since then, it of operant behavior, reinforcement theory,
has spawned numerous research articles, book and environment-based accounts of behavior
chapters, heated theoretical debates, as well as to language and the complex instances of hu-
an important book (Hayes, Barnes-Holmes, man behavior referred in traditional accounts
& Roche, 2001) that has itself spawned too as cognition. In its less charitable moments,
many reviews, replies, and counter-replies to go of which there are many, “RFT directly chal-
over here. The present discussion reviews the lenges Skinner’s perspective on verbal events”
nature of RFT, and then examines what we can (Hayes, 1994, p. 20), suggesting that Skinner’s
say about what causes it and how it is related perspective enjoys an “unwarranted dominance”
to verbal behavior, with due recognition of the (Hayes, 1994, p. 20) because it is actually con-
work of others who have written or conducted ceptually limited, empirically incorrect, and
research on these same issues. theoretically flawed. We note with interest that
RFT explicitly, forcefully, and in our opinion
Author note: I thank Per Holth and Erik Arntzen for their distastefully casts itself as an adversary rather
kind invitations to participate in their Seminar on Verbal Behav- than an extension or continuation of Skinner’s
ior, and for their distinguished leadership in the Behavior Analysis
Community. Correspondence concerning this article should be approach. Exactly why RFT advocates feel
addressed to J. Moore, Ph. D., Dept of Psychology, University the need to adopt a strident adversarial stance
of Wisconsin-Milwaukee, Milwaukee, WI 53201, phone: (414)
229-4746, email: jcm@uwm.edu is not abundantly clear. A particular and fre-
31
32 Jay Moore

quent RFT charge is that Skinner’s definition does not therefore himself call him Jones, nor
of verbal behavior is not functional and is too for this reason report that Jones was present, nor
broad, including virtually any socially medi- point to Jones in replay to the question Which
ated behavior as verbal (e.g., Hayes, 1994, p. is Jones? All these stages are developed through
20; Hayes, Barnes-Holmes, & Roche, 2001, p. the use of autoclitics.... A purely verbal defini-
12). We revisit this charge at the conclusion of tion appears to use the same process. Thus An
this paper, when we consider the definition of amphora is a Greek vase with two handles has at
verbal behavior. least three effects upon the listener. As a result
At this point, we can acknowledge that Skin- of having heard this response he may (1) say
ner didn’t directly or extensively engage many of amphora when asked What is a Greek vase with
the topics of interest to RFT advocates. How- two handles called?, (2) say A Greek vase having
ever, passages in Skinner’s writing do indicate two handles when asked What is an amphora?, and
he was genuinely sensitive to these topics. For (3) may point appropriately when asked Which of
example, consider the following passage from these is an amphora? Again, these are not results
Verbal Behavior: which occur spontaneously in the naive speaker
Something less than full-fledged relational but rather as the product of a long history of
autoclitic behavior is involved when partially verbal conditioning. Education is largely con-
conditioned autoclitic “frames” combine with cerned with setting up the behavior necessary to
responses appropriate to a specific situation. permit these changes to occur. (Skinner, 1957,
Having responded to many pairs of objects with pp. 359-360)
behavior such as the hat and the shoe and the gun In this passage, readers may note that the
and the hat, the speaker may make the response two sentences that point out some instances
the boy and the bicycle on a novel occasion. If of verbal behavior come about only as a con-
he has acquired a series of responses such as the sequence of a long process or history of verbal
boy’s gun, the boy’s shoe, and the boy’s hat, we may conditioning. Skinner did not specify the
suppose that the partial frame the boy’s _____ is precise nature of that history, but presumably
available for recombination with other responses. further knowledge of that history is very impor-
The first time the boy acquires a bicycle, the tant for an adequate understanding of verbal
speaker can compose a new unit the boy’s bicycle. behavior (see also Barnes-Holmes, Barnes-
This is not simply the emission of two responses Holmes, & Cullinan, 2000, p. 71).
separately acquired. The process resembles the
multiple causation of Chapter 9. The relational What is Derived Relational Responding?
aspects of the situation strengthen a frame, and
specific features of the situation strengthen Detailed presentations of the fundamentals
responses fitted into it. (Skinner, 1957, p. 336) of RFT are available elsewhere. Accordingly,
In this passage, readers may note that the last we give only a brief summary here. According
sentence actually contains the words “relational” to Hayes, Barnes-Holmes, and Roche (2001),
and “frame.” Relational frame theory is a behavior analytic
Next, consider another passage from Verbal approach to human language and cognition.
Behavior: RFT treats relational responding as a general-
Thus, we hear a man called Jones and see him ized operant, and thus appeals to a history of
respond appropriately to this “vocative.” As a multiple-exemplar training. Specific types of
result, we may also address him as Jones, or later relational responding, termed relational frames,
reply Jones to the question Who was there? or are defined in terms of the three properties of
correctly designate him when asked Which man mutual and combinatorial entailment, and the
was Jones? But this does not all happen in the transformation of functions. Relational frames
naive speaker or listener; it is the end result of a are arbitrarily applicable, but are typically not
long process of verbal conditioning. The young necessarily arbitrarily applied in the natural
child hearing someone called Jones many times language context. (p. 141)
Relational responding 33

Relating means responding conditionally to criminative, reinforcing, punishing). A network


one stimulus or event on the basis of its relation of relations may develop among elements of
to another. For RFT, the important cases are the stimuli so framed. Conditional contextual
when the stimuli are related in some abstract stimuli may further modify the nature of the
way, rather than through directly trained rela- relation or stimulus function as it participates
tions or through a shared physical or formal in a frame or a relational network. For RFT,
property along a continuous dimension like verbal behavior may therefore be understood as
length, width, height, weight, wavelength, or the action of framing events relationally. Verbal
frequency, as in primary stimulus generaliza- stimuli have an effect because they participate in
tion. Arbitrarily applicable suggests the impor- relational frames (e.g., Hayes, Barnes-Holmes,
tance of conventional practices and distinctions & Roche, 2001, pp. 43-44). Stimulus equiva-
in the verbal community as the abstract basis for lence, an earlier topic of investigation, may be
these relations. For example, all chairs do not understood as concerned with specific cases of
look alike. However, we may learn to call object derived relational responding. What is called
X a chair because it has been verbally related to symmetry in the vocabulary of stimulus equiva-
other objects conventionally called chairs, even lence is one kind of mutually entailed relation;
though we haven’t had the experience of suc- there are others. What is called transitivity
cessfully sitting on object X and it doesn’t look in the vocabulary of stimulus equivalence is
very much like other objects conventionally one kind of combinatorially entailed relation;
called chairs. Verbal behavior is bi-directional, there are others. In summary, the general term
in the sense that a verbal stimulus will often “derived relational responding” emphasizes that
function as the equivalent for an object, and responding is not the result of direct experiences
vice versa. In addition, words often refer or that can be traced to a history of reinforce-
function equivalently to each other. When ment involving the stimuli in question, or to
two stimuli become related to each other in the formal similarity between or among the stimuli.
absence of a history of direct reinforcement or Rather, the responding may be understood as
shared formal properties, we speak of mutual a function of more generalized and abstract
entailment. Mutual entailment means that relations between and among classes that come
interactions with the environment have led about through the arbitrary (i.e., nonformal,
speakers to respond on the basis of the particular conventional) practices of a verbal community.
relations that exist between the two objects in
the environment. When three or more stimuli How Does Derived Relational Responding
become related to each other in the absence Come About?
of a history of direct reinforcement or shared
formal properties, we speak of combinatorial As operant behavior, verbal behavior devel-
entailment. Combinatorial entailment means ops according to the contingencies imposed by
that interactions with the verbal community the verbal community. For RFT, those con-
have led speakers to respond to particular rela- tingencies involve multiple exemplar training.
tions that exist among multiple objects in the For example, a child may learn through direct
environment on the basis of the words that are experience to say that John is older than Mary.
used to describe those relations. Examples of The child may then learn again through direct
entailed relations are coordination (e.g., nam- experience to say that Mary is younger than
ing, similarity, identity, sameness), opposition, John. Now suppose the child learns through
distinction, comparison, hierarchical relations, direct experience to say that Paul is older than
temporal relations, spatial relations, condition- Sue. The child can presumably now state the
ality/causality, and deictic relations. We speak mutually entailed relation that Sue is younger
of the transformation of stimulus function when than Paul. Reinforcement from the verbal
the stimulus acquires a new function by virtue community across several examples has been
of its relation to other stimuli or events (e.g., dis- effective. If the child learns through direct
34 Jay Moore

experience that Tom is older than John, and ticipants with developmental disabilities. Lan-
Sarah is younger than Sue, the child can now guage rehabilitation programs are often aimed
state the combinatorially entailed relations that at these populations, in an effort to promote
Sue is younger than Tom, and that Sarah is verbal self-sufficiency. Indeed, the study of
younger than Paul. equivalence relations as an important feature
With regard to transformation of function, of verbal behavior grew out of such programs.
suppose an adult red wine drinker favors the If behavior needs to involve the principles of
bold, hard-hitting flavor of syrah but is unfa- RFT to be genuinely verbal, then it stands to
miliar with shiraz. Upon learning that shiraz reason that training involving derived relational
is an Australian equivalent of the French syrah, responding is indeed a useful component of a
then name brands associated with shiraz, as language rehabilitation program. In addition,
well as an unopened bottle of shiraz itself, may data from these populations may allow an as-
take on a new function in the adult’s life, even sessment of the processes, developmental or
though the adult has not had any direct experi- otherwise, that are held to underlie derived rela-
ence with shiraz. tional responding (e.g., Hayes, Barnes-Holmes,
Verbal behavior doesn’t cause derived & Roche, 2001, chapter 9).
relational responding. Rather, both verbal In one such study, Devany, Hayes, and
behavior and other forms of complex respond- Nelson (1986) explicitly studied the relation
ing based on derived relations are caused by between language use and stimulus equivalence.
contingencies involving multiple exemplars. As they reported in their abstract, children
These other forms of complex responding may were assigned to one of three groups: nor-
include equivalence, performance on RFT tasks mally developing preschoolers, children with
in laboratory experiments, psychopathology, a developmental disability who used speech
verbal regulation, and what is called cognition or signs spontaneously and appropriately,
more generally. The responding is said to be and children with a developmental disability
generalized higher-order operant behavior, as who did not. All children were then taught a
part of an overarching or generalized operant series of four related conditional discrimina-
class. The phenomenon bears at least superfi- tions and subsequently tested to determine if
cial resemblance to the process of generalized classes of equivalent stimuli had formed. All of
imitation (Hayes, Barnes-Holmes, & Roche, the language‑able children (normal as well as
2001, pp. 23-26). those with a developmental disability) formed
We may now turn to data that speak to the equivalence classes. In contrast, none of the
generality of the RFT account of verbal behav- language‑disabled children did so. The authors
ior and other forms of higher-order behavior concluded that even though the exact nature of
designated in the vernacular as “cognition.” We the relation between stimulus equivalence and
do not intend this paper to be a review of all language remains to be clarified, their results
studies, or to constitute a “box-score analysis” of supported the view that stimulus equivalence
how many studies can be interpreted as support- is a phenomenon with direct relevance to the
ing RFT versus how many opposing. Rather, we development of language.
review selected studies in an effort to highlight However, other studies involving par-
what we believe to be important issues in the ticipants with developmental disabilities have
study the relation among RFT, verbal behavior, found somewhat different results. For example,
and other forms of sophisticated responding Carr, Wilkinson, Blackman, and McIlvane
linked with verbal processes. (2000) reported two studies in which individu-
als with severe mental retardation and minimal
Data From Participants with verbal repertoires were tested for equivalence
Developmental Disabilities classes. Readers may recall conjectures to the
effect that performance on equivalence tasks is
We start by examining data from par- linked with language repertoires (e.g., Hayes,
Relational responding 35

Barnes-Holmes, & Roche, 2001 p. 160). In one veloping children. As they described in their
study, three individuals learned several match- abstract, Lipkens, Hayes, and Hayes (1993, p.
ing‑to‑sample performances: matching picture 201) gave a normally developing child a series
comparison stimuli to dictated‑word sample of experiences relevant to derived relational
stimuli (AB), matching those same pictures to responding as the child developed from the
printed letter samples (CB), and also matching age of 16 months to 27 months. In conjunc-
the pictures to nonrepresentative forms (DB). tion with these experiences, they also tested the
On subsequent tests, all individuals immedi- child for the derivation of relations. In four
ately displayed new relations AC, AD, BC, BD, longitudinal studies the child was (a) taught to
CD, and DC, together constituting a positive match names to pictures or pictures to names
demonstration of the formation of equivalence and was tested for derived mutually entailed
relations. A second study also found equiva- relations, (b) tested for retention of trained
lence in one of two individuals with similarly and derived relations after a 2‑week delay and
minimal verbal repertoires. Taken together, for derived mutually entailed relations after a
these studies question whether equivalence 1‑week delay from training, (c) taught to match
classes are demonstrable only in individuals with sounds to pictures and names to pictures and
well‑developed language repertoires. tested for derived mutually entailed relations
In an even more recent study, Wilkinson, and name‑sound and sound‑name combinato-
Rosenquist, and McIlvane (2009) evaluated rially entailed relations, and (d) tested for the
the formation of simple symbolic categories matching of a novel picture to a novel name and
from initial learning to specific dictated word- for subsequent naming of the novel excluded
picture relations through derived relations. The picture. The subject demonstrated mutually
participants were 10 individuals with several entailed relations as early as 17 months. Com-
intellectual or language limitations. Three ex- binatorially entailed relations emerged later. The
perimental categories were constructed, each authors concluded that their findings supported
containing one spoken word (category A), the view that derivation of relations does not
one photograph (category B), and one visual- depend on sophisticated verbal abilities, and
graphic lexigram (category C). Exclusion-based was consistent with expectations based on RFT.
learning procedures were then used to teach Another study is Luciano, Gomez Becerra,
first the three auditory-visual relations (A-B), and Rodriguez-Valverde (2007). As they de-
and then the three visual-visual relations (B- scribed in their abstract, they conducted three
C). Seven participants acquired these rela- studies that attempted to analyze the impact of
tions. These participants then demonstrated multiple‑exemplar training on the emergence of
virtually error-free performances on C-B and visual–visual equivalence relations and recep-
A-C derived relations. With regard to applied tive symmetry with a very young child named
matters, the study helped to define a useful and Gloria. Gloria was described as a healthy,
productive path for systematic instruction in happy, restless, and curious child who showed
symbolic functioning for persons with intel- no evidence of receptive symmetry or naming
lectual and language disabilities. With regard when tested at the age of 15 months 24 days (15
to more theoretical matters, the authors point m 24 d). In the first study, multiple‑exemplar
out that their study adds to the growing body training in immediate and delayed receptive
of findings that question the close relation be- symmetry or listener behavior (from object–
tween language development and the capacity sound to immediate and delayed sound–object
to exhibit emergent stimulus-stimulus relations selection) proceeded for one month with 10
characterized by stimulus equivalence (p. 200). different objects. This training was followed,
at 16 m 25 d, by a second test conducted with
Data From Typically Developing Children six new objects. Gloria showed generalized re-
ceptive symmetry with a 3‑hr delay. However,
We next examine data from typically de- no evidence of naming with new objects was
36 Jay Moore

found. A second study with Gloria began when were examined on the subtests of the WAIS,
she was 17 m. The aim of this study was to the scores on the vocabulary and arithmetic
establish derived visual–visual equivalence rela- subtests of the 31 students who completed the
tions using a matching‑to‑sample format with task were significantly higher than the scores of
two comparisons. Visual–visual equivalence the 44 participants who failed. However, the
responding emerged at 19 m, although Gloria scores on the digit‑symbol encoding subtest of
still did not show evidence of naming. A third the students who successfully completed the
study with Gloria was then carried out when she relational task did not differ significantly from
was between 22 m and 23 m 25 d. This study those who didn’t. Post‑hoc statistical analyses
used a three‑comparison matching‑to‑sample found a low but significant correlation between
procedure to establish visual–visual equivalence. the vocabulary and arithmetic scores and the
Equivalence responding emerged as in the sec- percentage of correct responses emitted in one
ond study, and naming emerged by the end of particular training phase of the relational task.
the third study, after over a month of further However, overall performances of monolingual
training experiences. The authors stated that and bilingual participants did not differ signifi-
these data suggest the relation between extant cantly on either the relational task or the WAIS
verbal skills and derived relational responding subtests. The authors interpreted their findings
differs from that ordinarily found in the main- as supporting the position that a behavioral
stream RFT literature. account of language must be informed by the
literature on derived relational responding.
Data on Monolinguals and Bilinguals, and
Correlations Between Derived Relational Data on Covariance of Elements of the RFT
Responding and Scores on IQ Tests Repertoire

Other relevant data can be found in stud- As reviewed in Hayes, Barnes-Holmes, and
ies with participants who are already verbally Roche (2001, e.g., pp. 31 ff.), some studies have
competent. Of particular interest are compari- shown derived responding and transformation
sons of the performances of monolinguals and of stimulus function with respect to condi-
bilinguals on RFT laboratory tasks, as well as tioned reinforcement, discrimination, elicited
the correlations between those performances conditioned emotional responses, extinction,
and scores on IQ tests (for discussion of the and even self-discrimination. Indeed, derived
relevance of RFT to intellectual development, responding and the transformation of stimulus
particularly “verbal intelligence,” see Hayes, functions within a network are the very heart
Barnes-Holmes, & Roche, 2001, pp. 160 ff.). of RFT. We can now review three other studies
In an important study, O’Hora, Pelaez, examining whether the elements of the RFT
and Barnes-Holmes (2005) first determined repertoire systematically covary in speakers
whether students were monolingual or bilin- who are already verbally competent. In the
gual. They next exposed these participants to a first study, Pilgrim and Galizio (1990) began
complex relational task (i.e., an empirical model by training two, three‑member equivalence
of instructional control), taken from O’ Hora, classes (A1-B1-C1 and A2-B2-C2) with five
Barnes‑Holmes, Roche, & Smeets (2004). college students as participants. One or more
Finally, they examined the participants on the of the reinforcement contingencies controlling
vocabulary, arithmetic, and digit‑symbol encod- baseline conditional discriminations were then
ing subtests of the WAIS‑III. On the basis of a changed. AC relations were either reversed (i.e.,
language examination, the researchers classified choosing C2 was reinforced and C1 punished
26 students as monolingual, and 49 as bilingual. when A1 was the sample; choosing C1 was
When participants were exposed to the complex reinforced and C2 punished when A2 was the
relational task, 31 students successfully com- sample) or arranged randomly (i.e., choosing
pleted it, and 44 did not. When participants C2 and C1 were reinforced and punished
Relational responding 37

equally often in the presence of A1 and A2). equivalence phenomena or relational framing
In a third condition, the original AB and AC processes.
relations were reversed. The results showed In a third study, which is related to Pilgrim
that although baseline conditional discrimina- and Galizio (1990), Pilgrim, Chambers, and
tion performances were under the control of Galizio (1995) used 5- to 7-year old children
reinforcement contingencies, and performances as participants. As they reported in their
on symmetry trials (i.e., mutual entailment) abstract, the children first learned two con-
varied with baseline responding for three of ditional discriminations (i.e., A1-B1, A2-B2,
four participants when contingencies were A1-C1, and A2-C2) in a two‑choice arbitrary
reversed, performances on transitivity probes match‑to‑sample task. After this training, two,
(i.e., combinatorial entailment) remained con- three‑member equivalence classes emerged (A1-
sistent with the initial equivalence class. These B1-C1 and A2-B2-C2). Baseline conditional
inconsistencies between probe and baseline discrimination performances were quickly
performances were striking because baseline controlled by reversals of the AC reinforce-
conditional discriminations are thought to ment contingencies (i.e., choosing comparison
determine equivalence class performance. stimulus C2 was reinforced given sample A1,
Similarly, the contrast between performances and choosing C1 was reinforced given sample
on symmetry and transitivity probes was of A2) when the reversals were introduced in
theoretical interest because equivalence classes restricted baselines. On reflexivity, symmetry,
are defined by congruent patterns of respond- and transitivity/equivalence probes following
ing on probe trials (see also Fields et al., 1992). the reversal, there was some limited indication
In a second study, Pilgrim and Galizio of equivalence‑class reorganization (i.e., A1-
(1995) similarly trained two four‑member B1-C2 and A2-B2-C1) in keeping with the
equivalence classes (A1-B1-C1-D1 and A2- concurrently performed baseline relations for
B2-C2-D2) again with five college students as two of five participants, but the predominant
participants. They then gave the participants pattern across probe trials was one of inconsis-
a series of further training experiences involv- tent conditional control. These findings suggest
ing the relations. These experiences included that, given similar challenges, equivalence‑class
a reversal of the baseline conditional discrimi- performances may be even more easily disrupted
nations (i.e., choosing D2 was reinforced and in young children than in adults.
D1 punished given Sample A1; choosing D1
was reinforced and D2 punished given Sample Data From Procedural Variations
A2), the delayed introduction of CD/DC
transitivity/equivalence probes, DE conditional Some data may now be considered from
discrimination training, a second baseline con- research in which features of the experimental
ditional discrimination reversal (i.e., choosing procedure are varied. The point is that if equiva-
C2 was reinforced given B1, etc.), and a return lence or derived relational responding varies
to original baseline reinforcement contingen- systematically when aspects of the training or
cies. Their results showed that performances on testing procedure are varied, then perhaps those
baseline and symmetry probes were extremely procedural variations reveal something critical
sensitive to baseline modifications. However, about the relation between training histories
performances on transitivity/equivalence probes and the demonstration of equivalence. Speci-
remained predominantly consistent with the fication of those training histories would then
originally established equivalence classes, al- seem directly relevant to the existence, analysis,
though there were exceptions on some E probe and understanding of any underlying processes
relations for two participants. They concluded (e.g., see discussion in Saunders et al., 1993,
that the dissociation between baseline and sym- as to whether equivalence is “underanalyzed”).
metry versus transitivity/equivalence patterns is For example, research has investigated
not easily accounted for by current models of whether equivalence varies when the order in
38 Jay Moore

which baseline training and testing for emergent protocol) that led to the emergence of two,
relations (i.e., symmetry, transitivity, equiva- 4-member equivalence classes (A1-B1-C1-D1
lence) is manipulated. In this regard, three com- and A2-B2-C2-D2). After a mild electrical
mon training-testing protocols are the simple- shock was paired with stimulus B1, partici-
to-complex protocol, the complex-to-simple pants did not exhibit a conditioned emotional
protocol, and the simultaneous protocol. With response to other members of the equivalence
a simple-to-complex protocol, participants are class. These results failed to replicate Dougher
first directly trained in one conditional relation et al. (1994), which had earlier shown that
(e.g., AB), then tested for symmetry (e.g., BA). after a related manipulation, participants did
Participants are then directly trained in a second exhibit a conditioned emotional response to
conditional relation (e.g., BC) and tested for other members of the equivalence class. Then
symmetry (e.g., CB). Finally, participants are Rodriguez-Valverde, Luciano, and Barnes-
tested for transitivity (e.g., AC) and for com- Holmes manipulated several features of the
bined symmetry-transitivity (e.g., CA). With experimental procedure. For instance, direct
a complex-to-simple protocol, participants are respondent conditioning was carried out with
first directly trained in all conditional relations. several members of each class, instead of just
They are then tested for symmetry, transitivity, one member. This time, participants did exhibit
and combined symmetry-transitivity in that or- the transformation of respondent function with
der. With a simultaneous protocol, participants other members of the equivalence class. Again,
are first directly trained in all conditional rela- procedural features needed to be identified in
tions. They are then tested for symmetry, transi- order to properly understand when equivalence
tivity, and combined symmetry-transitivity in a did and did not occur.
random order. Fields and colleagues (e.g., Fields A final important study we may consider
et al., 1997) have found a high percentage of is Holth and Arntzen (2000). This study ex-
participants demonstrated equivalence relations panded on earlier reports showing that laten-
with the simple-to-complex protocol but only cies to respond on equivalence test trials were
a low percentage with a simultaneous protocol. noticeably longer than latencies on conditional
A further procedural variation is whether discrimination trials during training (e.g., Arn-
training involves a linear series design, a many- tzen & Holth, 1997; Spenser & Chase, 1996;
to-one design, or a one-to-many design. With Wulfert & Hayes, 1988). Holth and Arntzen
a linear series design, participants are directly began by directly training conditional discrimi-
trained with AB and BC conditional discrimi- nations using a favorable one-to-many protocol.
nations before testing for equivalence. With a After the conditional discriminations had been
many-to-one design, participants are directly established, Holth and Arntzen then explicitly
trained with AB and CB conditional discrimi- required college-student participants to respond
nations before testing. With a one-to-many with a very short latency on equivalence test
design, participants are directly trained with trials. If participants did not respond in time,
AB and AC conditional discriminations before the trial was terminated and participants were
testing. Arntzen and Holth (1997) found the returned to the training protocol. Holth and
one-to-many design to be the most effective in Arntzen found that on a majority of trials,
demonstrating equivalence. With a linear series participants did not respond before the time
training design, they found the probability of limit elapsed. When participants did respond,
experimenter-defined equivalence class forma- they did not reliably exhibit equivalence, even
tion may not even exceed chance. though they had performed appropriately on
A more complex example of procedural the baseline conditional discriminations.
variations is Rodriguez-Valverde, Luciano, As Holth and Arntzen (2000) point out,
and Barnes-Holmes (2009). In this study, these results question assumptions that stimulus
participants were first trained in a condi- equivalence is a basic process that “emerges” au-
tional discrimination procedure (one-to-many tomatically and immediately as a consequence
Relational responding 39

of a certain training history. Holth and Arntzen disabilities. Nevertheless, the data suggest the
concluded that “Even a perfectly maintained correlation among language, derived relational
baseline with respect to the prerequisite con- responding, and developmental level may not
ditional discriminations does not ensure the be as high as originally thought. Although
emergence of experimenter-predicted emergent Devany, Hayes, and Nelson (1986) do show
performances in accord with equivalence” (p. derived relational responding does correlate ap-
331). That response latencies increased during propriately with a verbal repertoire, other stud-
equivalence testing presumably implies that ies reviewed above show it does not. If derived
participants were doing something different relational responding is as tightly linked with
on test as compared with training trials. That verbal behavior as RFT advocates suggest when
equivalence was not demonstrated when trials they tout the superiority of their approach, then
terminated after a short period of time without it seems to us that question can legitimately be
a response presumably implies that the trial raised as to why the other studies have shown
interrupted the “something” they were doing, that relational responding is not more positively
preventing them from completing it. At issue correlated with verbal behavior.
is what that something is. Holth and Arntzen We move next to data from typically devel-
suggested the possibility of “precurrent” re- oping children. Hayes (1994) among others has
sponding. We discuss this possibility and others interpreted the outcome of Lipkens, Hayes, and
in a later section. Hayes (1993) as consistent with expectations
based on RFT. Similarly, RFT advocates may
Analysis of Studies claim the outcome of Luciano, Gomez Becerra,
and Rodriguez-Valverde (2007) with Gloria is
We may now comment more extensively on consistent with RFT. At issue is whether it is
the implication of the studies reviewed above legitimate to so claim. In principle, it seems to
for the generality and indeed validity of RFT. us that the essential features of RFT (mutual
Where possible, expectations based on our entailment, combinatorial entailment, trans-
straightforward interpretation of RFT will be formation of function, as reflected in verbal
offered, although we recognize that on the basis behavior and contrived RFT laboratory tasks)
of past evidence, such as reviews of the RFT should at some point covary in the repertoires
book, RFT advocates typically suggest conclu- of individuals. If a participant exhibits some
sions about their position offered by others, features of the repertoire, it follows that the
particularly when they are critical, are in error. participant must have had sufficient exposure
We begin with data from populations with to multiple exemplar training to establish the
developmental disabilities. To be sure, com- other appropriate features of RFT. In other
parison across participants with developmental words, a child should be far enough along in the
disabilities in these studies is difficult because list of 43 steps in Hayes, Barnes-Holmes, and
the variance among the existing repertoires of Roche (2001, p. 151) that the vast proportion
participants can be great. For example, the of features should be present–after all, that’s
participants may have had widely varying expe- what being verbally competent means. We
riences with language, both from being spoken point out that Carr, Wilkinson, Blackman, and
to by others and from their own behavior, how- McIlvane (2000) gave equivalence training to
ever limited or idiosyncratic those experiences participants with developmental disabilities.
might have been. As a prelude, we acknowledge Even though these individuals had minimal
that Hayes, Barnes-Holmes, and Roche (2001, verbal repertoires, they exhibited equivalence.
p. 151) sketched a representative sequence of However, for Gloria, important aspects of the
43 milestones for the development of derived RFT repertoire do not all seem to be present.
relational responding, based on a typical train- It took something over a month of further
ing history. Atypical training histories might training to establish naming, even though other
well exist in populations with developmental aspects of equivalence were present much earlier.
40 Jay Moore

Perhaps some additional experiences occurred than IQ tests. Empirical validation of this
during this period. In any case, this outcome post-hoc claim awaits.
suggests a disparity wherein the critical features A further matter concerns between-subject
of derived relational responding do not neces- variability. A casual inspection of research ar-
sarily covary. ticles employing RFT laboratory tasks suggests a
We move next to data comparing mono- very large number of verbally competent partici-
linguals and bilinguals (O’Hora, Pelaez, & pants fail to acquire the target behavior. Often
Barnes-Holmes, 2005) . At issue is whether data from these subjects is simply dismissed in
it is reasonable to expect bilinguals to perform reports in the literature. For example, O’Hora,
at a higher level than monolinguals on RFT Pelaez, and Barnes-Holmes (2005) found 31
laboratory tasks. Our position is that such participants acquired their relational task, but
an expectation is in fact reasonable. Presum- 44 did not. When between-subject variance is
ably, bilingualism involves exactly the kind this high, one wonders about the validity of the
of tasks that RFT deems critical. If so, then premise that RFT laboratory tasks reveal pro-
performances of bilinguals should be generally cesses that lie at the heart of verbal behavior and
superior to those of monolinguals. However, the general intellectual functioning popularly
the data suggest the performances of bilinguals called cognition.
are comparable, rather than superior. We can now move to a final consideration,
Similarly, at issue is whether we can reason- namely, whether the elements of the RFT
ably expect the correlations between perfor- repertoire covary in prescribed ways in partici-
mance on RFT laboratory tasks and scores on pants who are already verbally competent. It
the verbal components of a standard IQ test to follows that if what RFT says is correct about
be highly positive (O’Hora, Pelaez, & Barnes- equivalence, derived relational responding, the
Holmes, 2005). Again, our position is that transformation of function, relational frames,
such an expectation is reasonable. Presumably, and relational networks, then the elements of
the environments of the individuals taking the RFT repertoire should systematically covary
these tests have encouraged the development in verbally competent speakers: mutual entail-
of flexible thinking and other intellectual skills ment, combinatorial entailment, and transfor-
associated with “being smart.” This reper- mation of function. Presumably, speakers who
toire should in turn be highly correlated with are already verbally competent have already
performance on laboratory tasks that require experienced the necessary degree of multiple-
relational framing, as both are concerned with exemplar training to possess the repertoires
the same aspects of a repertoire (see Hayes, required by RFT, otherwise they would not be
Barnes-Holmes, & Roche, 2001, pp. 160 ff.). verbally competent in the first place. Those
Although the correlations between performance repertoires require covariation, which is to say
on laboratory tasks and scores on the vocabulary high positive correlations across elements: To
and arithmetic subtests were consistent with use the language of stimulus equivalence, if
this expectation, correlations with the scores already verbally competent speakers exhibit the
on the digit-symbol test were not. This result is relation called symmetry, then so also should
surprising, as the digit-symbol test seems most they exhibit the relation called transitivity.
closely related to relational framing and derived In other words, it seems to us that from
relational responding. the standpoint of RFT, individuals exhibit in-
A possible RFT rejoinder re IQ tests is that stances of sophisticated intellectual repertoires
they are at best merely imperfect measures of and are verbally competent because one way
verbal and intellectual repertoires established or another they have experienced the necessary
via developmental training. If so, it follows that degree of (e.g., multiple exemplar) training from
the nature of demands posed in RFT lab task verbal community during their lifetimes. This
are actually better measures of individual dif- training has produced frames regarding mu-
ferences in verbal and intellectual performance tual entailment, combinatorial entailment, and
Relational responding 41

transformation of function. To be sure, some studied. However, researchers have to recognize


individuals may have had better training than that questions about whether something unap-
others, resulting in stronger repertoires, but all preciated but nevertheless behavioral is going
must have some minimum level of training that on are legitimate.
necessarily and universally provides the foun- With regard to these matters, Hayes, Barnes-
dation for the development of the repertoires Holmes, and Roche (2001) commented as
in question and for being verbally competent. follows:
Readers will recall that Pilgrim and Galizio It is expected that these units [mutual entail-
(1990), Pilgrim and Galizio (1995), and Pil- ment, combinatorial entailment, and transfor-
grim, Chambers, and Galizio (1995) explicitly mation of stimulus functions] may not always
investigated the extent to which transitivity covary and specific contingencies may break
covaried with symmetry in an equivalence task. them apart or pull them together... For the
Participants in these studies were already verbal- purposes of our analysis, a relational frame is a
ly competent: Participants were college students “unit” for two primary reasons: it seems to be
in the first and second studies, and children in the simplest unit that can describe the key ele-
the third. It seems to us that if participants in ments of speaking with meaning and listening
RFT laboratory tasks are already verbally com- with understanding..... We do not believe that
petent when they participate in the experiment, relational frames are “primitives” or “elemental
then mutual entailment (e.g. symmetry), com- units” however.... The flexibility of operants is
binatorial entailment (e.g., transitivity), and one of their hallmarks. Several studies show that
transformation of function should necessarily relations among an equivalence class are quite
and universally co‑vary for these participants, changeable, and even once formed, relations
unless specific operations (e.g., contextual) are among stimuli in a class may change individu-
in effect to dictate otherwise in designated cases. ally or en masse depending on conditions. For
Again, this research found that when baseline example, having provided training sufficient to
conditions were modified, symmetry changed generate equivalence classes, .... if we change only
but transitivity did not. The authors concluded a small number of baseline discriminations, some
that symmetry and transitivity did not covary of the derived relations will change while others
in the ways expected on the basis of RFT (see will remain intact.... This would be expected
also Fields et al., 1992). from the point of view of RFT. (pp. 34, 41-42)
Research involving procedural variations The passage above strikes us as entirely post-
strongly suggests derived relational responding hoc. It makes RFT unassailable, by saying that
consists of component processes. Otherwise, all possible outcomes are consistent with RFT.
the procedural variations would not influence If behavior changes, then that result is consistent
the demonstration of equivalence or derived with RFT, but if behavior doesn’t change, then
relational responding in the observed ways. In that result is consistent, too. Strange reasoning,
any event, this is not an appeal to mentalism that, in light of the avowed functional contex-
and the mediationism of cognitive psychol- tualist principle of a pragmatic truth criterion
ogy, any more than talking to oneself is an with prediction and influence of events as a
appeal to mentalism and the mediationism of single integrated goal. Ironically, the reasoning
cognitive psychology. Rather, the point is that appeals to mechanistic conceptions of behavior
procedural variations such as those reviewed and to correspondence as a truth criterion, to
here may be understood as tools to investigate which RFT says it stands opposed.
whether something unappreciated but never-
theless behavioral is going on in research on Summary and Conclusions
equivalence or derived relational responding. If
participants do something in response to these RFT advocates argue that their accounts of
variations that influences the demonstration of verbal behavior, equivalence, performance on
equivalence, then it can be known about and derived relational responding tasks in labora-
42 Jay Moore

tory experiments, psychopathology, verbal answer. We agree with Palmer (2004): “[M]
regulation, and what is called cognition more uch more detail is required about the variability
generally, are superior to any other account, of the history and its relation to variability in
particularly any account linked to Skinner’s behavior” (pp. 195-196).
Verbal Behavior. It follows from RFT accounts A further troublesome issue regarding RFT
that any time already verbally competent par- is the vocabulary. One concern is the nature
ticipants are in an experiment that involves an of “deriving” and “relating.” In many of their
equivalence or RFT task, they should neces- expository treatments, RFT advocates treat de-
sarily and universally show a high percentage riving and relating as some sort of prebehavioral
of the various features of derived relational activity: There is first the deriving or relating,
responding, by virtue of already being verbally and then subsequently there is speaking or re-
competent (i.e., by virtue of already having the sponding. Deriving or relating is portrayed as
necessary history of multiple exemplar training something else that is responsible for behavior,
that has led to the development of the relational rather than as an early link in a chain. This sort
frames that have made them verbally compe- of conceptual approach has proved troublesome
tent). As noted in the review, although the in the past, as it is an invitation to mentalism.
various features of derived relational responding A second and related concern is the frequent
do covary for some participants, they do not use of the word cognition (e.g., Hayes, Barnes-
for many others, either between participants Holmes, & Roche, 2001, p. 144). To be sure,
(e.g., high percentage of verbally competent both casual and professional writing are often
participants–44 of 75--who failed to acquire under audience control, reflecting that speakers
the relational task: O’Hora, Pelaez, & Barnes- or writers will use different words with different
Holmes, 2005; routine and deliberate disregard audiences. In casual speech, even radical behav-
in studies of verbally competent participants iorists say they change their minds, when they
who fail to acquire the baseline RFT task) or recognize that literally they do not have minds
within (failure of transitivity to covary with to change, and that their choice of words reflects
symmetry, and other failures of covariation: other considerations. Perhaps use of cognition
Fields et al., 1992; Pilgrim & Galizio, 1990, in the RFT literature reflects nothing more
1995; Pilgrim, Chambers, & Galizio, 1995). than the attempt to “get the message across”
To be sure, RFT advocates can easily iden- with a nontechnical, non-behavior-analytic
tify some data from some participants in some vocabulary. At issue is whether the current
studies that differ from the data reviewed above. RFT vocabulary surrenders too much to the
Moreover, Hayes, Barnes-Holmes, and Roche vernacular, and by so doing, invites mentalism.
(2001, pp. 41 ff.) have already claimed that We think so. It is all to easy to invoke derived
diverse results are to be expected from the point relational responding, relational frames, and so
of view of RFT. However, the validity of this on as “evidence” of cognition, or as the “mea-
claim is not at all clear. That relational framing sure” of cognition. Cognition then becomes
might develop according to some sequence of some nonbehavioral activity in a nonbehavioral
developmental experiences is not at issue. Of dimension that precedes and causes behavior.
course it does. What is at issue is that for speak- Hayes, Barnes-Holmes, and Roche seek to
ers who are already verbally competent, the defuse concerns by arguing that “cognizing is
necessary repertoire of relational framing and relating.... Cognition is not a mental event, it
derived relational responding should already be is a behavioral event” (p. 145). Regrettably,
in place, such that responding on RFT tasks and however, this treatment is muddled, as it el-
verbal behavior should covary in some general evates “cognizing” to some causally effective
and uniform sense. They clearly do not. It fol- antecedent process that is itself not behavioral.
lows, then, that RFT is not the answer it claims Readers may note that Palmer (2004) similarly
to be. It may contribute to an answer for some commented on conceptual difficulties with the
participants, but it is neither the whole nor only vocabulary of RFT.
Relational responding 43

The present recapitulation brings us to the in environment (for example, see discussion in
very definition of verbal (see also Palmer, 2008, Hayes, Barnes-Holmes, & Roche, 2001, p. 27).
and Normand, 2009, for a thorough evaluation Similarly Skinner (1957) talks of abstraction as
of the definition of verbal behavior). Readers “peculiarly a verbal process because a nonver-
will recall that RFT advocates make much of bal environment cannot provide the necessary
the superiority of their definitional approach, restricted contingency” (p. 109). Once again,
and return to this matter time and again as the Skinner’s approach is consistent with what RFT
basis for their position. To our way of thinking, advocates identified 50 years later.
they adopt an essentialist approach when they Further, RFT advocates make much of
say that behavior is verbal if and only if it en- “arbitrarily applicable relational responding.”
tails framing events relationally (see Normand, Presumably, what arbitrarily applicable rela-
2009, for similar discussion). They contrast tional responding signifies is responding based
their definition with Skinner’s (1957) initial on the conventional practices of a verbal com-
statement that talked of “verbal behavior as munity, as distinguished from responding that
behavior reinforced through the mediation of could be interpreted as controlled by histories of
other persons” (p. 2). Less conspicuous in their direct reinforcement involving absolute stimu-
treatments is that Skinner then acknowledged lus properties. Again, Skinner commented in
his provisional definition needed certain refine- numerous cases about the importance of un-
ments, which were provided later: derstanding conventional practices in a verbal
To say that we are interested only in behavior community:
which has an effect upon the behavior of an- The “languages” studied by the linguist are the
other individual does not go far enough, for the reinforcing practices of verbal communities.”
definition embraces all social behavior.... If we (Skinner, 1957, p. 461).
make the provision that the “listener” must be Different verbal communities shape and main-
responding in ways which have been conditioned tain different languages in the same speaker,
precisely in order to reinforce the behavior of who then possesses different repertoires having
the speaker, we narrow our subject to what is similar effects upon different listeners. (Skinner,
traditionally recognized as the verbal field. (pp. 1974, p. 92).
224-225) As a result, this aspect of Skinner’s perspec-
Skinner’s refined definition clearly acknowl- tive is again consistent with what RFT advocates
edges the importance of going beyond social argue about verbal behavior 50 years later.
behavior, and including a listener who has a In a more positive vein, it seems to us that
history that involves reinforcing the behavior equivalence and derived relational responding
of the speaker. As such, it seems consistent may best be understood as stimulus control ef-
with what RFT advocates argue about verbal fects. We agree that multiple exemplar respond-
behavior nearly 50 years later but claim as ing is important in the development of that
their own unique insights (cf. Barnes-Holmes, stimulus control. For example, in a discussion
Barnes-Holmes, & Cullinan, 2001). of abstraction, Skinner (1953) stated that
The passages cited from Skinner (1957) at By reinforcing responses to a circular red spot
the beginning of this article are from a chapter while extinguishing response to circular spots of
on the autoclitic. Autoclitic processes involve other colors, we may give the red spot exclusive
a speaker who modifies the effect of ongoing control over the behavior.... We have, in other
verbal behavior. Autoclitic processes are occa- words, brought the response under the control
sioned by a speaker’s own discriminative verbal of circular red spots but not of the “property of
behavior, rather than some object or property of redness” alone. To achieve the latter, we must
the environment. This concept seems entirely reinforce responses to many objects, all of which
consistent with what RFT advocates identify as are red, but which differ widely in their other
important when they talk of relating or deriv- properties.... There are no “natural” contingen-
ing, rather than properties of stimulus objects cies which reinforce a response in the presence
44 Jay Moore

of a single property without respect to other of events in motion.


properties. The necessary contingency appar- Interestingly, the terminal relations in a
ently requires the mediation of other organisms. demonstration of mediated generalization are
Abstraction, therefore, appears to have become virtually the same as identified in RFT, although
possible only with the development of verbal RFT typically views as mechanistic, and rejects
behavior. (pp. 134-136) out of hand, appeals to such devices as mediat-
Clearly, equivalence and derived relational ing proprioceptive feedback. The possibility of
responding are very important stimulus control mediated generalization is relevant because ex-
effects linked to verbal behavior. planations of the various effects in the literature
If RFT is not the answer it claims to be, what wouldn’t appeal to new principles. An adequate
then is the answer? RFT explicitly disparages understanding of existing principles would suf-
“associative explanations.” At issue is whether fice. The foregoing is not a claim that mediated
it is reasonable to do so. For example, Ton- generalization or any other associative process
neau and colleagues (Tonneau & Gonzalez, is the total explanation of derived relational
2004; Tonneau, Arreola, & Martinez, 2006) responding, but rather only a suggestion that
have suggested that Pavlovian processes may well-known processes could conceivably play a
be relevant to understanding behavior similar role. The possibility could be explored empiri-
to derived relational responding. As Tonneau cally, rather than be dogmatically dismissed.
and Gonzalez (2004) point out, Honey and Another possibility is a chaining process,
Hall (1989) reported the transfer of discrimi- perhaps involving covert verbal behavior in
native functions from one stimulus to another certain cases. Palmer (2004) identified this
with rats as subjects, by virtue of pairing with possibility in an earlier review. This possibility
a common stimulus. Markham and Markham is also consistent with the results of Holth and
(2002) reported an analogous transfer of func- Arntzen (2000), in which time constraints on
tion without reinforcement with humans as responding disrupted the demonstration of
participants, based on Pavlovian relations. That equivalence in test trials. In their study, perhaps
Pavlovian processes can influence operant pro- the “precurrent” behavior that was disrupted
cesses is well documented (e.g, Moore, 1986). by the termination of a trial was an ongoing
One such Pavlovian process could be medi- chaining process responsible for equivalence.
ated generalization. A mediated generalization Again, no new process may be needed to explain
experiment is typically carried in three phases. results of equivalence or RFT laboratory tasks.
In Phase 1, suppose a tone is sounded, and fol- The variability in results across experiments may
lowed shortly thereafter by a light tap on the then be understood as variability in the history
cheek sufficient to cause a subject to blink. In of the participants that contributes to control
Phase 2, suppose the same light tap on the cheek by covert events. Of course, whether chaining is
is followed shortly thereafter by an electric shock actually involved in derived relational respond-
sufficient to cause an autonomic response. In ing is an empirical question. As Skinner (1953)
Phase 3, suppose we sound the tone from Phase commented in one place (we have made minor
1. Research of this sort shows that the subject changes in the wording of the comments to
will show the autonomic response from Phase 2, make the wording consistent with the present
even though there has been no formal, directly context) , “There are great individual differences
experienced relation between tone and shock in the extent to which private [stimulation] is
that produced the Phase 2 autonomic events. used..... [S]uch differences may be traced either
The standard interpretation of these sorts of to differences in the extent to which [control by
experiments is that the tone causes the blink in private stimuli] has been established or to dif-
Phase 1, and proprioceptive feedback from the ferences in the ability to describe the resulting
blink then acquires stimulus control over the self-stimulation or use it as a basis for further
autonomic response by virtue of its relation with behavior” (pp. 273-274).
the shock in Phase 2. The tone sets the cascade The forms of behavior identified as derived
Relational responding 45

relational responding clearly do occur, and they cognition more generally, and then turn around
are clearly linked with verbal behavior. The to claim that the observed lack of covariance
present concern is whether derived relational re- between derived relational responding and
sponding is the only answer to questions about verbal behavior, equivalence, performance on
the nature and causes of verbal behavior. RFT RFT tasks in laboratory experiments, psycho-
advocates repeatedly claim that only responding pathology, verbal regulation, and what is called
that shows derived relational responding and cognition more generally is to be expected.
relational framing is verbal. However, it appears
that such a claim is unnecessarily restrictive. References
Indeed, it may be counterproductive. To our
way of thinking, enough data suggest other Arntzen, E., & Holth, P. (1997). Probability
behavior that meets all the other requirements of stimulus equivalence as a function of
of being verbal can occur without the prescribed training design. The Psychological Record,
features of derived relational responding, and 47, 309-320.
the presence of derived relational responding Arntzen, E., & Holth, P. (2000). Probability
does not guarantee verbal behavior. Thus, of stimulus equivalence as a function of class
mutual entailment, combinatorial entailment, size vs. number of classes. The Psychological
transformation of function, and relational fram- Record, 50, 79-104.
ing don’t covary in necessary and universal ways, Barnes-Holmes, D., Barnes-Holmes, Y., & Cul-
either within or between verbally competent linan, V. (2000). Relational frame theory
participants. Is it appropriate for behavior ana- and Skinner’s Verbal Behavior: A possible
lysts to be concerned with the relation between synthesis. The Behavior Analyst, 23, 69-84.
verbal behavior and love, compassion, and Carr, D., Wilkinson, K. M., Blackman, D., &
community? Absolutely yes. Is it appropriate McIlvane, W. J. (2000). Equivalence classes
for behavior analysts to be concerned with the in individuals with minimal verbal reper-
relation between verbal behavior and thinking, toires. Journal of the Experimental Analysis
reasoning, and problem solving? Absolutely of Behavior, 74, 101‑114.
yes. Is it appropriate for behavior analysts to Devany, J., Hayes, S., & Nelson, R. (1896).
be concerned with the relation between verbal Equivalence class formation in language-able
behavior and therapeutic processes and inter- and language-disabled children. Journal of
ventions? Absolutely yes. Is it appropriate for the Experimental Analysis of Behavior, 46,
behavior analysts to do more research on what 243-257.
are called equivalence and derived relational Dougher, M. J., Augustson, E., Markham, M.
responding? Absolutely yes. Is it appropriate R., Greenway, D. E., & Wulfert, E. (1994).
for behavior analysts to claim that “A new day The transfer of respondent eliciting and ex-
has dawned” (Hayes, Barnes-Holmes, & Roche, tinction functions through stimulus equiva-
2001, p. xii) because the recent flood of research lence classes. Journal of the Experimental
on equivalence, derived relational responding, Analysis of Behavior, 62, 331-351.
therapeutic processes, and interventions could Fields, L., Adams, B. J., Newman, S., &
only have come after Skinner’s (1957) definition Verhave, T. (1992). Interactions among
of verbal behavior was repudiated and a new emergent relations during equivalence class
definition based on derived relational respond- formation. Quarterly Journal of Experimental
ing embraced in a post-Skinnerian world? We Psychology, 45B, 125-138.
may never know. What we do know is that we Fields, L., Reeve, K. F., Rosen, D., Varelas, S.,
can’t have it both ways. We can’t claim that Adams, B. J., Belanich, J. & Hobbie, S. A.
derived relational responding uniquely explains (1997). Using the simultaneous protocol
verbal behavior, equivalence, performance on to study equivalence class formation: The
RFT tasks in laboratory experiments, psycho- facilitating effects of nodal number and
pathology, verbal regulation, and what is called size of previously established equivalence
46 Jay Moore

classes. Journal of the Experimental Analysis ing. The Psychological Record, 54, 437-450.
of Behavior, 67, 367-389 O’Hora, D., Pelaez, M., & Barnes-Holmes, D.
Hayes, S. (1994). Relational frame theory: A (2005). Derived relational responding and
functional approach to verbal events. In S. performance on verbal subtests of the WAIS-
Hayes, L. Hayes, M. Sato, & K. Ono (Eds.), III. The Psychological Record, 55, 155-175.
Behavior Analysis of Language and Cognition Palmer, D. (2004). Data in search of a prin-
(pp. 9-30). Reno, NV: Context Press. ciple: A review of Relational Frame Theory:
Hayes, S., Barnes-Holmes, D., & Roche, B. A Post-Skinnerian Account of Human
(2001). Relational frame theory: A post- Language and Cognition. Journal of the Ex-
Skinnerian account of human language and perimental Analysis of Behavior, 81, 189-204.
cognition. New York: Kluwer/Plenum. Palmer, D. (2008). On Skinner’s definition of
Hayes, S., & Brownstein, A. (1985, May). Ver- verbal behavior. International Journal of Psy-
bal behavior, equivalence classes, and rules: chology & Psychological Therapy, 8, 295-307.
New definitions, data, and directions. Paper Pilgrim, C., & Galizio, M. (1990). Relations
presented at the convention of the Associa- between baseline contingencies and equiva-
tion for Behavior Analysis, Columbus, OH. lence probe performances. Journal of the Ex-
Holth, P., & Arntzen, E. (2000). Reaction perimental Analysis of Behavior, 54, 213‑224.
times and the emergence of class consistent Pilgrim, C., & Galizio, M. (1995). Reversal of
responding: A case for precurrent respond- baseline relations and stimulus equivalence:
ing? The Psychological Record, 50, 305-337. I. Adults. Journal of the Experimental Analy-
Honey, R., & Hall, G. (1989). Acquired sis of Behavior, 63, 225‑238.
equivalence and the distinctiveness of cues. Pilgrim, C., Chambers, L., & Galizio, M.
Journal of Experimental Psychology: Animal (1995). Reversal of baseline relations and
Behavior Processes, 15, 338-346. stimulus equivalence: II. Children. Journal
Lipkens, R., Hayes, S., & Hayes, L. (1993). of the Experimental Analysis of Behavior, 63,
Longitudinal study of derived stimulus rela- 239‑254.
tions in an infant. Journal of Experimental Rodriguez-Valverde, M., Luciano, C., &
Child Psychology, 56, 201-239. Barnes-Holmes, D. (2009). Transfer of
Luciano, C., Gomez Becerra, I., & Rodri- aversive respondent elicitation in accordance
guez-Valverde, M. (2007). The role of with equivalence relations. Journal of the Ex-
multiple‑exemplar training and naming perimental Analysis of Behavior, 92, 85-111.
in establishing derived equivalence in an Saunders, K. J., Saunders, R. R., Williams, D.
infant. Journal of the Experimental Analysis C., & Spradlin, H. E. (1993). An interac-
of Behavior, 87, 349‑365. tion of instructions and training design on
Markham, R., & Markham, M. (2002). On stimulus class formation: Extending the
the role of covarying functions in stimulus analysis of equivalence. The Psychological
class formation and transfer of function. Record, 43, 725-744.
Journal of the Experimental Analysis of Be- Skinner, B. F. (1953). Science and human
havior, 78, 376-386. behavior. New York: Macmillan.
Moore, J. (1986). On the consequences of con- Skinner, B. F. (1957). Verbal behavior. New
ditioning. Psychological Record, 36, 39-61. York: Appleton-Century-Crofts.
Normand, M. (2009). Much ado about noth- Skinner, B. F. (1974). About behaviorism. New
ing? Some comments on B. F. Skinner’s York: Knopf.
definition of verbal behavior. The Behavior Spenser, T. J., & Chase, P. N. (1996). Speed
Analyst, 32, 185-190. analyses of stimulus equivalence. Journal
O’ Hora, D., Barnes‑Holmes, D., Roche, B., & of the Experimental Analysis of Behavior, 65,
Smeets, P. (2004). Derived relational net- 643-659.
works and control by novel instructions: A Tonneau, F., & Gonzalez, C. (2004). Func-
possible model of generative verbal respond- tion transfer in human operant experiments.
Relational responding 47

Journal of the Experimental Analysis of Behav- with severe language impairments and intel-
ior, 81, 239-255. lectual disabilities. The Psychological Record,
Tonneau, F., Arreola, F., & Martinez, A. (2006). 59, 187-206.
Function transformation without reinforce- Wulfert, E., & Hayes, S. C. (1988). The trans-
ment. Journal of the Experimental Analysis fer of conditional ordering response through
of Behavior, 85, 393-405. conditional equivalence classes. Journal of
Wilkinson, K., Rosenquist, C., & McIlvane, W. the Experimental Analysis of Behavior, 50,
(2009). Exclusion learning and emergent 125-144.
symbolic category formation in individuals

You might also like