You are on page 1of 7

EUROPEAN JOURNAL OF BEHAVIOR ANALYSIS 2001, 2, 35- 98 NUMBER 1 (SUMMER 2001) 35

The (not so) strange death of stimulus equivalence


Dermot Barnes-Holmes
National University of Ireland, Maynooth

Steven C. Hayes
University of Nevada, Reno

Bryan Roche
National University of Ireland, Maynooth

Undertakers play a necessary if somewhat correlations is problematic. Finally, we will argue


unpopular role in modern life. And this is how that a modern behavioral approach is needed to
we see Tonneau’s article — as the undertaker who scale the mountains of human language and cog-
is kindly helping bury our close friend, stimulus nition, and this approach is fundamentally oper-
equivalence. We suspect that few will thank Ton- ant in nature. We will offer Relational Frame
neau for providing this unpleasant service, but it Theory as a map for the course ahead (see Hayes,
needs to be done. Barnes-Holmes, & Roche, 2001).
Funerals are not pleasant occasions, but they
can serve an important social and psychological Stimulus Equivalence as an Exhausted
function in preparing those left behind to move Research Paradigm
on. In this case, moving on will be neither simple
nor easy. Like a widowed spouse considering Tonneau questions the relevance of the data
the future, equivalence researchers may be fright- gathered by equivalence-class researchers to their
ened by what seems to lie ahead. For some, stated purpose – the understanding of symbolic
stimulus equivalence was the behavior-analytic behavior. In large part we agree. Stimulus equiva-
answer to the cognitive challenge. Tonneau has lence is simply too narrow and too limited to
helped show how thin that claim has become. carry the weight some have put on it. It is a re-
The demise of equivalence may appear to search preparation that shows every sign of ex-
threaten the future of behavior analysis – at least haustion. The warning signs are myriad, but two
for those who feel (correctly) that providing an stand out.
adequate account of language and cognition is
essential. Hopefully, however, this very threat will Limited Applied Impact
galvanize the field into action. Human language Behavior analysis is a field that develops be-
and cognition stand before behavior analysis like havioral principles in order to understand human
a vast mountain range. Basic behavioral research- complexity (Skinner, 1938, p. 441), not as an end
ers must not turn back from this challenge if in itself. After 30 years of research, stimulus
their science is to survive. equivalence has produced little impact on applied
In this commentary on Tonneau’s article we behavior analysis. Only very rarely, for example,
will echo some of his concerns pertaining to the do the pages of the Journal of Applied Behavior
study of stimulus equivalence. However, we will Analysis contain an article bearing on the applica-
also explain why we believe that his focus on tion of stimulus equivalence. If stimulus equiva-
what he calls functional equivalence and stimulus lence is an adequate account of human symbolic
behavior, surely much more should be expected.
Address correspondence concerning this article to E-mail:
Dermot.Barnes-Holmes@may.ie or Hayes@SCS.UNR.EDU
35
36 COMMENTARY

Limited Basic Impact lence has oriented the field toward an important
Basic behavior analysis, and the field as a whole, property of human language and cognition, that
needs a theory that will lead to an experimental of the bi-directionality of stimulus relations
analysis of human language and cognition. If such (“symmetry”). Curiously, this is precisely the area
a theory were available, we would expect it to that Tonneau brushes over in his analysis of func-
inspire basic research on intelligence, problem tional equivalence.
solving, emotion, reasoning, language develop- The four types of function transfer that Ton-
ment, and a whole host of similar areas. Stimulus neau describes in his article could be based on
equivalence has failed to do so. Instead of broad- varieties of forward associative conditioning (sen-
ening over time, the focus of equivalence research sory preconditioning, second-order conditioning,
seems to be narrowing to increasingly arcane is- etc.), and he argues quite reasonably that transi-
sues, disconnected from human language and tivity in matching to sample could also be con-
cognition as a general phenomenon. What was sidered in the same light. Then on page 19, in
once an outcome that might orient the field to key parentheses, he suggests that symmetry might also
process issues in the analysis of language, is now yield to a similar analysis, but he refuses to “press
being studied as an end in itself. such arguments any further.” It is this property
Considered as an end in itself, equivalence is of bi-directionality (particularly when it is com-
of dubious value. It has already been studied for bined with transitivity), that most requires an ac-
30 years, and the field could spend another 30 count, because it seems to require backward asso-
years in research primarily characterized by its ciative conditioning. Backward conditioning is a
precise irrelevancy. Behavior analysis does not notoriously weak effect that occurs in a rather
have that time to waste. restricted set of conditions (see Spetch, Wilkie,
As a result of this narrowing and the limits & Pinel, 1981). Hall (1996), a leader in the field
of the phenomenon itself, the study of stimulus of associative learning, has stated that backward
equivalence seems to be adding little that cogni- associations in matching to sample contexts “are
tive psychology had not already considered. The not readily formed” (p. 248).
rich literatures on semantic network theory and Interestingly, Hall (1996) has also argued that
transitive inference, for example, provide con- when backward conditioning effects apparently
ceptual and empirical analyses that can incorpo- occur, using matching to sample, they may actu-
rate the phenomenon of equivalence class for- ally reflect a mediated form of forward condi-
mation and a great deal more as well. tioning. Imagine, for example, that a discrimina-
We are not calling for behavior analysts to tive function transfers from C to B after A-B
embrace cognitive psychology. To the contrary, and A-C pairings. Apparently, a result such as this
it is precisely because it is the mission of behav- would require backward associative condition-
ior analysis to provide its own account of hu- ing (i.e., A may acquire some of the functions of
man complexity that this solution is unacceptable. B via forward conditioning, but C can only ac-
But the field cannot wait forever for real progress. quire the functions of A, and thus B, via back-
If stimulus equivalence is not the vehicle for fun- ward conditioning). Several researchers (Hall,
damental progress - and it seems undeniably clear 1996; L. J. Hayes, 1992; Urcuioli, 1996) have ar-
that it is not - we must move on to approaches gued, however, that C-B transfer, in this case, may
that are more productive. be based on a mediated form of forward con-
ditioning (i.e., mediated generalization) in the ab-
Symmetry and Backward Associative sence of any backward associative processes.
Conditioning According to this account, A-B pairing estab-
lishes a private perceptual function (or represen-
Our foregoing concerns clearly echo some of tation) of B whenever A is presented (i.e., when
Tonneau’s criticisms of the stimulus equivalence the subject sees A, an image or representation of
research program. However, we do not share all B immediately occurs). Thus, whenever A and C
of his views. The concept of stimulus equiva- are paired, the private image of B is also paired
COMMENTARY 37

with C (i.e., when the subject sees A, s/he imme- lus equivalence, but as we will discuss below it
diately thinks of B but then C is presented, and leaves untouched the substantial evidence that
thus the private image of B occurs before C). In bidirectionality characterizes many different types
this case, the discriminative functions of C may of stimulus relations (e.g., Dymond & Barnes,
transfer to B via forward respondent condition- 1995, 1996; Roche & Barnes, 1996, 1997; Roche,
ing. Barnes-Holmes, Smeets, Barnes-Holmes, &
Urcuioli (1996) also argued that in order for McGeedy, 2000; Steele & Hayes, 1991). Some-
this type of mediated generalization to occur, the thing much more flexible than backward condi-
A-B training must occur before the A-C training. tioning seems needed.
If, however, A-B and A-C training occur con- A ready behavioral alternative is available: the
currently, “it is unlikely that [the A] samples could bidirectional stimulus relations observed in
generate the necessary prospective mediators to equivalence, and other types of stimulus relations,
support transfer” (Urcuioli, 1996, p. 65) to the B reflects a generalized operant behavioral process
stimuli. (Hall also implies this argument in his (Barnes-Holmes & Barnes-Holmes, 2000; Hayes,
analysis, but he does not state it explicitly). In many Gifford, & Wilson, 1996). This is the core con-
equivalence studies, however, A-B and A-C train- ception of Relational Frame Theory (RFT; Hayes,
ing trials are presented concurrently (e.g., Barnes- Barnes-Holmes, & Roche, 2001). Associative or
Holmes, Keane, Barnes-Holmes, & Smeets, 2000; stimulus-correlational effects are certainly also
Barnes & Keenan, 1993), and thus by this account involved (Barnes, 1994; Leader, Barnes, & Smeets,
C-B transfer should not have occurred unless one 1996), but they do not play a primary explana-
wishes to invoke the problematic concept of tory role in RFT (Barnes & Roche, 1996). In fact,
backward associative conditioning. from the RFT perspective, the effects of stimu-
Given these difficulties, it is perhaps not hard lus correlations on behavior may often depend
to understand why Tonneau refused to address upon the generalized operant process of rela-
the issue of symmetry in his article. If we are to tional framing itself (see Barnes, Smeets, &
take his arguments seriously, however, he will have Leader, 1996, pp. 167-168; Leader, Barnes, &
to provide an account of those instances of func- Smeets, 1996, pp. 702-704; Leader, Barnes-
tion transfer that appear to require backward as- Holmes, & Smeets, 2000; Leader & Barnes-
sociative conditioning. Without this, his work sim- Holmes, 2001, in press; Smeets, Leader, & Barnes,
ply questions the set-theoretic analysis of stimu- 1997).
lus equivalence, but offers nothing else in its place. Many RFT studies focus on non-equivalence
As it is, Tonneau has pointed to stimulus associa- relations, and it is here that a simple explanation
tions or correlations, rather than equivalence class in terms of stimulus correlations (involving ei-
formation, as the basis for function transfer, but ther forward or backward associations) seems
he has completely failed to address the very issue almost completely untenable. Consider, for ex-
that made the concept of stimulus equivalence ample, a finding reported by Dymond and Barnes
necessary in the first place — the emergence of (1995). In this study, what we call a relational net-
bidirectional stimulus relations. work was established using matching to sample
procedures. The derived stimulus relations most
Relational Frame Theory: Moving Beyond relevant in the current context were as follows:
Function Transfer C1 same as B1; B2 less than B1; C2 more than
B1 (note the stimuli were nonsense syllables and
One “solution” to the problem of thus were not related to each other along any
bidirectionality is simply to accept that backward consistent physical dimensions). A “one-re-
associative conditioning occurs readily and in most sponse” discriminative function was then estab-
conditions with humans, but with great difficulty lished for B1 using direct reinforcement. Subse-
in nonhumans. This is not unlike Sidman’s argu- quently, a “one-response” function emerged for
ment that equivalence is a basic stimulus function C1, a “zero-response” function emerged for B2,
(Sidman, 1994). It “solves” the problem of stimu-
38 COMMENTARY

and a “two-response” function emerged for C2, compound (they were all incorrect stimuli in com-
without explicit training. parison sets), and thus there appears to be no
In effect, the directly reinforced “one-re- basis on which to predict the choice of B2C2.
sponse” function established for B1 was trans- In another part of the Barnes et al. study (data
formed in accordance with the derived same, less- not reported in the published version) the fol-
than, and more-than relations. B1 controlled one- lowing trial-type was presented; B1C1 as sample
response by direct training, and thus C1, which with B2C2 and A1C4 as comparisons. In this case,
participated in a derived same relation with B1, the A1C4 choice contained a stimulus (A1) that
also controlled one-response. B2, however, par- had been directly correlated, during training, with
ticipated in a derived less than relation with B1, both of the elements contained in the sample,
while C2 participated in a derived more than rela- whereas B2C2 contained two stimuli that were
tion with B1. Thus, B2 controlled a zero-response, never correlated with either element. The sub-
and C2 controlled two-responses. jects chose B2C2, as RFT would predict, because
We do not see how Tonneau’s account can an equivalence relation is in a “frame of coordi-
explain such data. During the training and testing nation” with an equivalence relation.
that produced these performances each of the In case after case (we have presented only
stimuli were correlated with each other (indirectly two, but the possible examples are many), a pri-
during training and directly during testing). If both mary focus on stimulus correlations, per se, ap-
B2 and C2 were correlated with B1, why would pears to run into trouble when we consider some
they acquire two different discriminative func- of the data that has been gathered over the years
tions, each of which differ from the trained B1 under the rubric of RFT. We have made a simi-
function? Although difficult to explain using the lar argument against purely class-based accounts
concept of stimulus-correlations, data such as of equivalence (Barnes & Roche, 1996; Hayes &
these are readily explained by RFT in terms of Barnes, 1997). It might be possible to save these
generalized operant response classes. more limited accounts by attempting to confine
Other research findings that appear to stretch them only to certain kinds of studies of derived
a stimulus correlational account are also handled stimulus relations and not others, but this saves
with relative ease by RFT. Consider, for example, the accounts by killing them. It is not a theory of
a study reported by Barnes, Hegarty, and Smeets a narrow research preparation that we need. If
(1997) in which matching to sample procedures RFT can explain equivalence and a host of other
were used to establish four, three-member phenomena, while other accounts cannot even
equivalence classes (A1 B1 C1, A2 B2 C2, A3 B3 explain equivalence data such as those reported
C3, A4 B4 C4). During one of the probe trials, by Barnes, et al. (1997), parsimony alone suggests
the B1 and C1 stimuli were presented together that RFT deserves the more serious consideration.
as a compound sample stimulus, and two other
compound stimuli were presented as compari- Conclusion
sons; B2C2 and B3C4. Subjects reliably chose the
B2C2 comparison over the B3C4 comparison, Equivalence research was rooted in the po-
and the authors described this outcome as relat- tential it provided for the analysis of language.
ing one equivalence relation to a second equiva- The stultification of equivalence research comes
lence relation. This performance is readily pre- from forgetting that mission. Tonneau appears
dicted by RFT (see Stewart, Barnes-Holmes, to agree. However, like others before him
Hayes, & Lipkens, 2001, for an extended discus- (whether pro- or anti-equivalence), Tonneau fails
sion of relations among relations), but it is not to address the fundamental issues that seem to
obvious how the concept of stimulus correla- underlie language and cognition, providing in-
tions, per se, can accommodate these data. The stead an account that does not lead directly to
individual stimuli in the B2C2 and B3C4 com- powerful new approaches to human complex-
pounds were each equally correlated, during train- ity.
ing and testing, with the two stimuli in the B1C1
COMMENTARY 39

We believe that it is time for behavioral theo- clearly on the table. Of course, some may prefer
rists to face directly the challenge posed by other to ignore these fundamental issues and hope in-
forms of derived stimulus relations, the trans- stead that the corpse of stimulus equivalence will
formation of functions through those relations, be reincarnated in a different guise (e.g., as the
relational networks, relations among relations, and study of naming, stimulus control topographies,
similar areas that RFT appears to handle with or stimulus correlations). We believe, however,
relative ease. We have shown that these phenom- that it is time to leave the graveside of stimulus
ena lead directly to new empirical approaches to equivalence, turn to the mission that this research
the analysis of such topics as metaphor, allego- area first provided, and attempt once more to
ries, thinking, problem-solving, self, and similar scale the mountains of human language and cog-
topics (Hayes, Barnes-Holmes, & Roche, 2001). nition.
Furthermore, we have described how they lead
directly to new approaches to psychological de- References
velopment, education, social behavior, psycho-
pathology, psychotherapy, and even spirituality Barnes, D. (1994). Stimulus equivalence and rela-
(Hayes, Barnes-Holmes, & Roche, 2001). Rela- tional frame theory. The Psychological Record, 44,
tional Frame Theory provides a relatively simple, 91-124.
operant account of the phenomenon of stimu- Barnes, D., Hegarty, N., & Smeets, P. M. (1997).
lus equivalence, but in addition it accounts for Relating equivalence relations to equivalence
more complex phenomena as well, and in so relations: A relational framing model of com-
doing gives behavior analysis a powerful new plex human functioning. The Analysis of Ver-
angle on human language and cognition itself. bal Behavior, 14, 57-83.
To state it in simple terms, RFT claims that Barnes, D., & Keenan, M. (1993). A transfer of
relating is an operant. This is an empirical matter, functions through derived arbitrary and non-
and must be so (see Barnes-Holmes, Barnes- arbitrary stimulus relations. Journal of the Ex-
Holmes, Roche, & Smeets, in press a, in press b; perimental Analysis of Behavior, 59, 61-81.
Healy, Barnes-Holmes, & Smeets, 2000). Behav- Barnes, D., & Roche, B. (1996). Relational frame
ior analysis is not a field worth having if it cannot theory and stimulus equivalence are fundamen-
determine, on empirical grounds, what is or is tally different: A reply to Saunders. The Psycho-
not an operant. Thus, the question “is relating an logical Record, 46, 489-508.
operant?” must be answerable in one of three Barnes, D., Smeets, P.M., & Leader, G. (1996).
ways: “no,” “not enough data yet,” or “yes.” If Procedures for generating emergent match-
the answer is “no” then one must say why, in ing performances: Implications for stimulus
clear conceptual or evidentiary terms. If the an- equivalence. In T.R. Zentall and P.M. Smeets
swer is “don’t know yet” one must say what other (Eds.), Stimulus class formation in humans and
evidence is needed. If the answer is “yes,” then animals(pp. 153-171). Holland: Elsevier.
behavior analysis will be forever changed. If re- Barnes-Holmes, D., & Barnes-Holmes, Y. (2000).
lating is an operant, it operates on operant and Explaining complex behavior, Two perspec-
respondent processes themselves (Hayes, Fox, tives on the concept of generalized operant
Gifford, Wilson, Barnes-Holmes, Healy, 2001; see classes. The Psychological Record, 50, 251-265.
also Barnes-Holmes & Barnes-Holmes, 2000; Barnes-Holmes, D., Barnes-Holmes, Y., &
Barnes-Holmes, Barnes-Holmes, & Cullinan, Cullinan, V. (2000). Relational frame theory
2000). and Skinner’s “Verbal Behavior:” A possible
Evidence is mounting in favor of the RFT synthesis. The Behavior Analyst, 23, 69-84.
interpretation of equivalence responding, and Barnes-Holmes, D., Keane, J., Barnes-Holmes, Y.,
relational responding more generally (see Hayes, & Smeets, P. M. (2000). A derived transfer
Barnes-Holmes, & Roche, 2001, for a compre- of emotive functions as a means of estab-
hensive review). The issue may not be closed as lishing differential preferences for soft drinks.
an empirical matter, but we believe that it is now The Psychological Record, 50, 493-512.
40 COMMENTARY

Barnes-Holmes, Y., Barnes-Holmes, D., Roche, (Eds.), Stimulus class formation in humans and ani-
B., & Smeets, P. M. (in press a). Exemplar mals (pp. 279-299). New York: Elsevier.
training and a derived transformation of Healy, O., Barnes-Holmes, D., & Smeets, P. M.
functions in accordance with symmetry. The (2000). Derived relational responding as gen-
Psychological Record. eralized operant behavior. Journal of the Ex-
Barnes-Holmes, Y., Barnes-Holmes, D., Roche, perimental Analysis of Behavior, 74, 207-227.
B., & Smeets, P. M. (in press b). Exemplar Leader, G., Barnes, D., & Smeets, P. M. (1996).
training and a derived transformation of Establishing equivalence relations using a re-
functions in accordance with symmetry: II. The spondent-type training procedure. The Psycho-
Psychological Record. logical Record, 46, 685-706.
Dymond, S., & Barnes, D. (1995). A transfor- Leader, G., Barnes-Holmes, D., & Smeets, P. M.
mation of self-discrimination response func- (2000). Establishing equivalence relations us-
tions in accordance with the arbitrarily appli- ing a respondent-type procedure III. The Psy-
cable relations of sameness, more-than, and chological Record, 50, 63-78.
less-than. Journal of the Experimental Analysis of Leader, G., & Barnes-Holmes, D. (2001). Es-
Behavior, 64, 163-184. tablishing fraction-decimal equivalence using
Dymond, S., & Barnes, D. (1996). A transfor- a respondent-type training procedure. The Psy-
mation of self-discrimination response func- chological Record, 51, 151-166.
tions in accordance with the arbitrarily appli- Leader, G., & Barnes-Holmes, D. (in press).
cable relations of sameness and opposition. Matching-to-sample and respondent-type
The Psychological Record, 46, 271-300. training as methods for producing equivalence
Hall, G. (1996). Learning about associatively relations: Isolating the critical variables. The
activated stimulus representations: Implications Psychological Record.
for acquired equivalence and perceptual learn- Roche, B., & Barnes, D. (1996). Arbitrarily ap-
ing. Animal Learning and Behavior , 24, 233- plicable relational responding and human
255. sexual categorization: A critical test of the
Hayes, L. J. (1992). Equivalence as process. In S. derived difference relation. The Psychological
C. Hayes & L. J. Hayes (Eds.), Understanding Record, 46, 451-475.
verbal relations (pp. 97-108). Reno, NV: Con- Roche, B., & Barnes, D. (1997). A transforma-
text Press. tion of respondently conditioned stimulus
Hayes, S. C., & Barnes, D. (1997). Analyzing de- function in accordance with arbitrarily appli-
rived stimulus relations requires more than the cable relations. Journal of the Experimental
concept of stimulus class. Journal of the Ex- Analysis of Behavior, 67, 275-300.
perimental Analysis of Behavior, 68, 235-270. Roche, B., Barnes-Holmes, D., Smeets, P. M.,
Hayes, S. C., Barnes-Holmes, D., & Roche, B. Barnes-Holmes, Y., & McGeady, S. (2000).
(2001). (Eds.), Relational Frame Theory: A post- Contextual control over the derived transfor-
Skinnerian account of human language and cogni- mation of discriminative and sexual arousal
tion. Plenum Press. functions. The Psychological Record,50, 267-291.
Hayes, S. C., Fox, E., Gifford, E. V., Wilson, K. Sidman, M. (1994). Stimulus equivalence: A research
G., Barnes-Holmes, D., & Healy, O. (2001). story. Boston: Authors Cooperative.
Derived relational responding as learned be- Skinner, B. F. (1938). Behavior of organisms. New
havior. In S. C. Hayes, D. Barnes-Holmes, & York: Appelton-Century-Crofts.
B. Roche (Eds.), Relational Frame Theory: A post- Smeets, P.M., Leader, G., & Barnes, D. (1997).
Skinnerian account of human language and cognition Establishing stimulus classes with adults and
(pp. 21-50). Plenum Press. children using a respondent training proce-
Hayes, S. C., Gifford, E. V., & Wilson, K. G. dure: A follow-up study. The Psychological
(1996). Stimulus classes and stimulus relations: Record, 47, 285-308.
Arbitrarily applicable relational responding as Spetch, M . L., Wilkie, D. M., & Pinel, J. P. J.
an operant. In T. R. Zentall & P. M. Smeets (1981). Backward conditioning: A reevalua-
COMMENTARY 41

tion of the empirical evidence. Psychological Hayes, D. Barnes-Holmes, & B. Roche (Eds.),
Bulletin, 89, 163-175. Relational Frame Theory: A post-Skinnerian account
Steele, D. L., & Hayes, S. C. (1991). Stimulus of human language and cognition (pp. 73-86). Ple-
equivalence and arbitrarily applicable relational num Press.
responding, Journal of the Experimental Analysis Urcuioli, P. J. (1996). Acquired equivalences and
of Behavior, 56, 519-555. mediated generalization in pigeon’s matching-
Stewart, I., Barnes-Holmes, D., Hayes, S. C., & to-sample. In T.R. Zentall and P.M. Smeets
Lipkens, R. (2001). Relations among relations: (Eds.), Stimulus class formation in humans and ani-
Analogies, metaphors, and stories. In S. C. mals (pp. 55-70). Holland: Elsevier.

Emergent matching to sample and equivalence relations


Harrie Boelens
Leiden University

Tonneau has presented a well-written, many different puzzles at the same time might
thoughtful, and erudite critique of the literature be more fruitful than the massive concentration
on stimulus equivalence. I will here react to two on emergent matching to sample that we see
of the issues he has raised: the importance of nowadays.
studying emergent matching to sample, and the The applications of the mathematical “equiva-
proper application of the mathematical “equiva- lence relation” concept to emergent matching to
lence relation” concept. sample often show misunderstandings. Tonneau
The performances seen on tests of stimulus will agree with me. I will here develop his point
equivalence are no obvious conditioning effects. further, in the hope of producing more clarity.
I propose that that is justification enough for Following Tonneau, define xRy to mean that com-
studying them. Emergent matching to sample is parison y is reliably selected in the presence of
a puzzle for conditioning theory. We can hope sample x. We could suggest that R is an equiva-
that working on this puzzle will lead to a revision lence relation (i.e., a relation that is reflexive, sym-
of the theory, and that the revised theory will metric, and transitive). This suggestion has many
make sense of other emergent phenomena, such consequences. I will here consider five that are
as those of generalized imitation, instruction fol- not often appreciated. First, it does not make
lowing, and grammatical speech. For me, it is not sense to call the individual relations that are trained
necessary that the puzzle has a feature (such as or tested equivalence relations. Suppose we have
“symbolic behavior”) that makes it look relevant taught a subject conditional relations A1B1, A2B2,
for something in everyday life. I agree with Ton- B1C1, and B2C2. We can then ask if the condi-
neau that the cognitive journals present many facts tional relation is an equivalence relation (for this
that do not have obvious behavior-analytic inter- subject, at this time). It does not make sense to
pretations, but does it matter? Studying them will ask if A1B1, A2B2, etc. are equivalence relations.
be worthwhile too! I find it a pity only that there Second, the relation R applies to matching to
is not much diversity in this work. Working on sample, not to conditional discrimination in gen-
eral. Suppose A1 is a red light, A2 a green light,
Address correspondence concerning this article to E-mail: B1 a low tone, and B2 a high tone. A rat’s lever
boelens@fsw.leidenuniv.nl

You might also like