Professional Documents
Culture Documents
2017-When Things Are Not The Same
2017-When Things Are Not The Same
Responding to stimuli as same and different can be considered a critical component of a variety
of language and academic repertoires. Whereas responding to “sameness” and generalized iden-
tity matching (i.e., coordination) have been studied extensively, there appears to be a significant
gap in behavior analytic research and educational programs with regard to nonmatching rela-
tions or relations of difference. We review research on difference relations from a variety of
domains, including comparative psychology, as well as experimental, and translational behavior
analysis. We examine a range of studies, including research on the perception of difference and
oddity responding, as well as investigations on establishing relational frames of distinction. We
present suggestions for future research and describe potential methods for teaching skills related
to relations of difference.
Key words: frame of distinction, same/different, relational frame theory, autism, difference
One of these things is not like the others, behaviors (Rosales-Ruiz & Baer, 1997). One
One of these things just doesn’t belong. such behavior change occurs when individuals
Can you tell which thing is not like the learn to make generalized same/different judg-
others ments (McIlvane et al., 2011).
By the time I finish my song? —Sesame Identifying stimuli as being different from
Street (Raposo, Stone & Hart, 1970, one another is considered by cognitive, devel-
track A7)
opmental, comparative, and educational psy-
chologists to be fundamental to cognition
Behavior analysts charged with developing
(e.g., Addyman & Mareschal, 2010; Thomp-
individualized intervention programs for clients
son & Oden, 1995). Using abstract language
with developmental disabilities, including
concepts, particularly those of “sameness” or
autism spectrum disorders (ASD), have long
“difference,” to classify the relations among
relied on the concept of behavioral cusps
objects and events has been described as being
(Bosch & Fuqua, 2001; Rosales-Ruiz & Baer,
the “hallmark of human intelligence”
1997). A behavioral cusp has been defined as
(Blaisdell & Cook, 2005), the “very keel and
any behavior change that allows an organism to
background of our thinking” (James, 1981/
come into contact with new contingencies that
1890, p 434). One of the main lines of
have broad and significant effects with respect
research in comparative psychology for decades
to establishing and maintaining other new
has been to determine whether nonhuman sub-
This article was prepared as part of the first author’s
jects are capable of same/different responding,
doctoral thesis at the National University of Ireland, long considered to be evidence of “conceptu-
Galway. ally-mediated” behavior (e.g., Nissen, Blum, &
The authors would like to thank Dr. Caio Miguel for Blum, 1948). Finally, within educational psy-
his helpful comments, edits, and suggestions through
many versions of this manuscript. chology, identifying similarities and differences
Correspondence concerning this article should be between stimuli or concepts has been high-
addressed to Siri Ming, School of Psychology, National lighted as having a particularly powerful impact
University of Ireland, Galway, Ireland.
E-mail: siri@siriming.com on student achievement (Marzano, Pickering, &
doi: 10.1002/jaba.367 Pollock, 2001).
© 2017 Society for the Experimental Analysis of Behavior
1
2 SIRI MING and IAN STEWART
Same/different responding features promi- identical items and picking out objects that do
nently in academic curricula. At the preschool not share a feature, function, or class with a
level, important milestones for 48-month-old named object. However, this is still a relatively
children include showing interest in how peo- limited sampling of difference responding.
ple are the same and different, and describing More traditional programs often include some
simple relations between objects such as point- lessons on difference (e.g., Center for Autism
ing to pictures of two different animals and say-
and Related Disorders, 2013; Leaf & McEa-
ing that they are “different” (e.g., see California
Department of Education, 2008). At 60 chin, 1999; Maurice, Green, & Luce, 1996),
months, preschoolers are expected to describe which vary in comprehensiveness. The most
differences between items based on informa- detailed includes a number of lessons
tional text, such as by “communicat[ing] sequenced from matching same/different
important differences of jet airplanes and pro- through identifying items as being the same or
peller planes after being read a story about air- different and identifying how two presented
planes and airports” (California Department of items are the same or different (i.e., in terms of
Education, 2008, p. 68). Marzano, Pickering, & their attributes or categories; Center for Autism
Pollock (2001) noted that identifying similari- and Related Disorders, 2013). However, even
ties and differences is the foundation of numer- these programs provide a relatively limited task
ous effective instructional activities at all
analysis of training procedures, given the poten-
academic levels and across subjects involving
comparing, classifying, and creating metaphors tial range from very simple to very complex
and analogies. Identification of similarities and same/different skills.
differences, including comparing/contrasting, The emphasis on relations of sameness
are also clearly recognized academic standards within the behavior analytic basic and applied
across multiple grades and subjects within the literature (e.g., Dube, McIlvane & Green,
widely adopted Common Core curriculum 1992; Leaf & McEachin, 1999; Lovaas, 2003;
(National Governors Association Center for Serna, Dube & McIlvane, 1997) may suggest
Best Practices & Council of Chief State School that when teaching “same,” we are inherently
Officers, 2010). also teaching “different,” as the two concepts
In light of the emphasis on both sameness cannot exist except in contrast to one another
and difference within the broad educational lit- (i.e., if something is not same, it is necessarily
erature, it is somewhat surprising that there is different, and vice versa). However, there are a
relatively little explicit emphasis on teaching number of reasons why focusing on relations of
differences (as opposed to sameness) in behavior difference is important. First, although it might
analytic intervention programs for children seem logical to assume that if a person responds
with autism and other developmental disabil- to same, then they also respond to different,
ities. Programs based on an analysis of verbal empirical evidence suggests that this is not the
behavior (e.g., Partington, 2006; Sundberg, case. As we will discuss further, research with
2008) place little to no emphasis on identifying individuals with developmental disabilities has
differences, considering it to be too complex shown that individuals who respond accurately
for early language intervention programs during identity matching tasks may not neces-
(e.g., autoclitics, see Sundberg, 2004). A more sarily respond accurately during oddity match-
recent curriculum (Promoting the Emergence ing (e.g., selecting a comparison different from
of Advanced Knowledge [PEAK]; Dixon, a sample), and vice versa (e.g., Mackay, Soraci,
2014a, 2014b, 2015) includes some potentially Carlin, Dennis & Strawbridge, 2002; Soraci
relevant lessons such as selecting an item that et al., 1987). A similar argument can be made
does not belong from an array of otherwise in response to the suggestion that teaching
A REVIEW OF DIFFERENCE RELATIONS 3
condition responding; for example, delivery of Rehfeldt, 2011; Sidman, 1971, 2009). For
Thai food might result in criticism and an argu- example, if a child is told that the sound lemur
ment, whereas delivery of something other than (A) is the name of the animal in the picture
Thai food might be met with approval. (B) and also that the written name for lemur is
Relational Frame Theory (RFT; Dymond & the textual stimulus lemur (C), she may then
Roche, 2013; Hayes, Barnes-Holmes, & respond as if the three stimuli A, B and C were
Roche, 2001) provides a useful foundation for substitutable or equivalent, despite the fact that
analyzing research into the emergence and they are not physically similar. This pattern
development of difference relations, and thus might be cued by words such as is, called, or
we will use it as a framework in our review. name of, as well as by more subtle cues such as
One key distinction is that between nonarbi- the use of a matching-to-sample (MTS) proce-
trary and arbitrary relations. dure (which has been employed to study this
pattern in the laboratory). In other words, these
features of the context are what determine the
Arbitrary versus Nonarbitrary Relations relational response, not the topography of the
There are two fundamentally different types stimuli being related. Arbitrary relational
of relational responding: nonarbitrary and arbi- responding in accordance with difference
trary (or arbitrarily applicable). Nonarbitrary involves responding to two stimuli under the
relational responding is based on physical char- control of a cue for a pattern of distinction. For
acteristics of stimuli. An example of nonarbi- example, when discussing with one child how
trary sameness responding would be selecting a other children in her class compare and con-
triangle as the comparison in the presence of a trast in terms of their likes and dislikes, she or
triangle as the sample (i.e., identity matching). he might learn that Johnny likes the same food
Another example would involve selecting an as Susie, but different food from Alan. In such
item that is physically different from a sample a case, despite there being no immediately
(i.e., oddity matching). The type of response physically obvious same or different dimensions
(selecting same or different) required in a given of the individuals being related, the child may
circumstance may be determined by the con- identify that Susie and Alan like different food
textual cue, such as the words same or different from each other (Dymond & Roche, 2013). In
(as in, “Select same.”). However, despite the this case, again, cues such as the words same
fact that contextual control can determine and different, which are used to compare the
which physical relation is appropriate, nonarbi- preferences of Johnny and Susie and of Johnny
trary relational responding is always based to at and Alan, respectively serve as the determinant
least some extent on the physical characteristics for the response.
(e.g., color, shape, etc.) of the stimuli being In the next sections, we will consider both
related. nonarbitrary and arbitrary difference relations
In the case of arbitrary (or arbitrarily applica- as investigated from a number of different theo-
ble) relational responding, by contrast, physical retical perspectives. We will describe a sample
features of the stimuli being related are no of the available literature, with the intent to
longer a defining feature. It is the contextual provide a logically and conceptually systematic
control that determines the relational response. organization to stimulate applied research.
Stimulus equivalence provides an example of Towards this end, a variety of search terms
this for a relation of arbitrary sameness were used (e.g., identity and nonidentity
(e.g., McLay, Sutherland, Church, & Tyler- matching, oddity, difference responding, dis-
Merrick, 2013; O’Donnell & Saunders, 2003; tinction, same-different responding), which
A REVIEW OF DIFFERENCE RELATIONS 5
were assembled organically2 over the course of and early types of nonarbitrary relational
the project rather than beginning with a prede- responding—identity and oddity—have also
fined set of search terms and methods. been well studied with humans (e.g., Mackay
A variety of terms for specific response forms et al., 2002; Serna et al., 1997; Soraci, Carlin &
relevant to difference relations were identified, Bray, 1992; Soraci et al., 1987). As noted
depending upon theoretical orientation. For above, nonarbitrary relational responding in
consistency, we have chosen to use a specific accordance with a relation of sameness or dis-
set of terms, as follows (also see Miguel et al., tinction can include matching pairs of stimuli
2015): We will use the term relational matching (e.g., matching two cows to two dogs rather
to describe pair matching on the basis of the than to a cat and horse; relational matching);
relation exemplified by the pairs involved selecting an item that is the same as, or differ-
(e.g., matching AB [different] with XY [differ- ent from a sample (relational listener discrimi-
ent] rather than ZZ [same]). We will term the nations); or producing/selecting the name of a
selection of comparison stimuli in a MTS for- relation in response to items either physically
mat, under the control of a specific relational different or the same as each other (relational
cue (such as, “Find different.”) as relational lis- tacting). Furthermore, the experimental litera-
tener discriminations. Finally, we will term pro- ture would suggest that prior to acquiring any
duction or selection of the name of a relation of these repertoires, organisms must first dem-
(e.g., “different”) in response to two stimuli in onstrate responses relevant to perception of dif-
a pair (e.g., two physically dissimilar stimuli) as ference relations (such as orienting towards
relational tacting, whether the response is topo- stimuli that are different from those previously
graphical (such as saying “different”) or seen) and learn to discriminate different stimuli
selection-based (such as selecting the textual based on a history of training rather than based
stimulus different). on contextual cues (i.e., oddity matching).
As noted, we will structure this review The literature related to responding to non-
around core elements of relational responding arbitrary relations of difference includes: a) per-
as identified by RFT. We will thus begin with ception of sameness/difference and nonarbitrary
early nonarbitrary relational responding and difference responding that is not controlled by
proceed to arbitrary (arbitrarily applicable) rela- a specific relational cue (oddity matching or
tional responding. As we do so, we will provide nonmatching to sample); b) nonarbitrary rela-
recommendations for future applied research, tional matching; c) nonarbitrary contextually
as well as clinical practice (when there is suffi- controlled relational listener discriminations;
cient basis in the literature to do so). and d) nonarbitrary relational tacting.
of difference versus sameness among a variety (e.g., Soraci et al., 1987). Even though several
of nonhuman species, as well as among human factors may influence performance at early
infants. For example, research has found that developmental levels of functioning, some
7- to 8-month-old human infants, but not authors (Carlin et al., 2003; Soraci et al., 1992)
younger, and infant chimpanzees as young as suggest that difficulty in establishing successful
9 months, differentially respond to novel sti- oddity matching may be due to the complexity
muli (i.e., those different from previously seen of the task requirements. For example, it may
stimuli), as well as to matching versus non- be more difficult for individuals at early devel-
matching pairs of stimuli measured by how opmental or chronological ages to perform cor-
long they look at the stimulus, and by anticipa- rectly in an oddity task involving a small array
tory eye movements (Addyman & Mareschal, of stimuli in which the odd stimulus is similar
2010; Oden, Thompson, & Premack, 1990; in some dimension or feature to the other sti-
Tyrrell, Stauffer, & Snowman, 1991). Such muli. In addition, performance on some oddity
results indicate that responding to relations of tasks requires attending both to the oddity rela-
sameness and difference starts to be shaped tion itself (i.e., responding to the difference
early on, and that systematic training might between stimuli) and to a specific dimension
benefit young individuals. along which the stimuli differ (e.g., form, color,
One very early type of discrimination that is or size). Children may have a predisposition to
based on a relation of difference is oddity attend to one dimension more than another,
matching—selecting the item that is different with younger children demonstrating a dimen-
from the others in an array. Responding on the sional preference for color, and older children
basis of oddity relations has been trained in a for form (e.g., Suchman & Trabasso, 1966).
variety of nonhuman species, including pigeons Thus, increasing the salience of the relevant
(e.g., Zentall, Edwards, Moore, & Hogan, relations between stimuli (e.g., making the dif-
1981), corvids (e.g., Smirnova, Lazareva, & ferent stimulus more distinct, such as one ele-
Zorina, 2000), rats (e.g., April, Bruce, & Gali- phant among many mice) and/or the relevant
zio, 2011), bees (Avarguès-Weber, Dyer, dimensions of stimuli could be expected to
Combe, & Giurfa, 2012), sea lions (e.g., Hille, facilitate oddity (and identity) matching.
Dehnhardt, & Mauck, 2006), and various pri- As mentioned, one general method of
mates (e.g., Davis, Leary, Stevens, & Thomp- increasing the salience of interstimulus relations
son, 1967). In these studies, subjects are involves organizational manipulations of the
trained to respond to a MTS task in which array, including (a) increasing the number of
selecting the comparison stimulus that is differ- nonodd stimuli and (b) increasing stimulus
ent from the sample is reinforced, and then contiguity (see Figure 1). For example, Zentall,
generalization to novel stimuli is tested. Hogan, Edwards, and Hearst (1980) found that
Although early research on oddity with increasing the number of nonodd stimuli from
humans seemed to indicate that responding to two to eight not only facilitated acquisition of
oddity tasks was very difficult for individuals oddity learning in pigeons, but also facilitated
under 4 years of age (e.g., Ellis & Sloan, 1959; transfer of their performance to an array of only
House, 1964), other early researchers consid- two nonodd and one odd stimuli. Soraci
ered oddity to be a prerequisite for identity et al. (1987) found similar results with young
matching (e.g., Fellows, 1968). More recent children identified as at risk for developmental
studies, moreover, have been successful in disabilities. Increasing stimulus contiguity
training oddity with young children and indivi- (i.e., eliminating space between stimuli) has
duals with developmental disabilities also been found to facilitate oddity learning
A REVIEW OF DIFFERENCE RELATIONS 7
with either small (e.g., three nonodd stimuli) sound—and one rattle—with its sound) facili-
or large (e.g., eight nonodd stimuli) arrays tated oddity performance. In this study, pre-
(Soraci et al., 1992). school children who had previously failed
Soraci et al. (1992) also described two meth- oddity tasks with visual stimuli alone were able
ods of increasing the salience of the relevant to pass using the paired stimuli, and that skill
dimensions (e.g., color vs. form) of the match- transferred to oddity tasks using novel visual
ing stimuli. The first is to assess individual stimuli alone. Using stimuli in this way might
dimensional preferences by using stimuli that be a useful remediation for students who show
vary in both color and form and determining difficulty with oddity responding.
which dimension is attended to in a matching By increasing the salience of the relevant
task (e.g., given a red airplane as a sample, and variables, children with and without develop-
a red car or a blue airplane as the comparisons, mental delays have been taught to perform
which comparison does the child choose?). A accurately during identity and oddity tests.
second method is to specifically train attending Mackay et al. (2002) worked with children
to the relevant dimension, by reinforcing iden- who had failed a test of two-choice identity
tity matching on the basis of the relevant matching (the cohort included individuals with
dimension (e.g., one could reinforce matching developmental disabilities, as well as typically
on the basis of color rather than form, such as developing 3- to 4-year-old children). Partici-
by matching a red car and a red airplane) pants were trained in an identity MTS task that
before testing oddity on that dimension. first used a comparison array of eight incorrect
Finally, Soraci, Deckner, Baumeister, and comparisons (all identical to each other but dif-
Bryant (1991) found that pairing a tone with ferent from the sample), and one correct com-
visual stimuli during oddity tests (such as using parison (identical to the sample, but different
an array consisting of several bells—with that from the other comparisons) to guide attention
Figure 1. Types of arrays used for oddity tasks, progressing from large arrays of stimuli in close proximity to small
arrays in less proximity, using MET with different stimulus sets on each trial.
8 SIRI MING and IAN STEWART
to the correct comparison. Initially, the chil- relation as a sample (see Figure 2 for examples).
dren could make a correct selection on the basis For instance, such tasks might require match-
of oddity alone, within the comparison array. ing a pair of squares (same) to a pair of trian-
Over time, the size of the comparison array was gles (same) rather than a pair that contains
gradually reduced, until the array consisted of both a circle and a star (different), or matching
only one correct and one incorrect comparison a pair containing a cow and a pig (different) to
stimulus. Using this procedure, children with a pair containing a horse and a duck (different)
and without developmental delays passed the rather than a pair of cats (same). This type of
identity matching test which they had previ- responding might be seen as a simple
ously failed. (or precursor) form of analogy, which in turn is
Previous work on identity matching may also seen as an important skill in itself as well as a
serve to guide research on oddity matching measure of intellectual performance
(e.g., Green, 2001; McIlvane, 2009). For (e.g., Ortony, 1979; Sternberg 1977).
instance, Serna et al. (1997) described critical There are several decades of research into
variables to consider when teaching individuals this type of relation with primates
to respond to tasks requiring identity matching. (e.g., Thompson & Oden, 1995, 2000), as well
These include whether the MTS task is non- as more recent research with corvids (Smirnova,
conditional (i.e., the S+ and S- stimuli do not Zorina, Obozova, & Wasserman, 2015). For
change across trials) or conditional (i.e., the S+ example, Thompson, Oden, and Boysen
stimulus on one trial could change to be an S- (1997) demonstrated that chimpanzees with a
stimulus on another trial). Their recommended history of training to select a specific token (tri-
teaching sequence is to proceed from simple angle) when shown a “same pair”, and a differ-
discrimination to nonconditional to conditional ent token (diagonal) when shown a “different
identity matching; preverbal toddlers (16-21 pair” could match a same pair to another same
months) have also been successfully trained in pair, and a different pair to another different
identity matching using this sequence pair. Meanwhile, a chimpanzee without similar
(de Alcântara Gil, de Oliveira, & McIlvane, training did not do so. Thompson and Oden
2011). Even though this sequence has not yet (1995, 2000) concluded that organisms must
been studied with oddity matching, it might be be trained to emit a relational tact in order to
a reasonable one to further examine. perform a relational match. However, more
Although there is insufficient evidence to recent research indicates that some primates
determine whether oddity and identity training and pigeons can be trained to perform nonarbi-
should proceed simultaneously or in a particu- trary relational matching without such a his-
lar sequence, research by Mackay et al. (2002) tory, given sufficient training exemplars and
suggests that identity and oddity are indeed sufficiently individualized stimulus presenta-
separable at the level of behavioral repertoires tions (e.g., small or large array size; Cook,
and thus an individual might learn one before Kelly & Katz, 2003; Flemming, Beran,
learning the other. As such, future research Thompson, Kleider, & Washburn, 2008;
might examine whether these repertoires should Truppa, Piano Mortari, Garofoli, Privitera, &
be trained sequentially or simultaneously. Visalberghi, 2011; Vonk, 2003).
To date, there appears to be little or no
research on teaching nonarbitrary relational
Nonarbitrary Relational Matching matching to humans that parallels the type of
Relational matching involves selecting an relational matching of same and different rela-
item pair that exemplifies the same type of tions seen in the nonhuman literature. Most
A REVIEW OF DIFFERENCE RELATIONS 9
Figure 2. Examples of nonarbitrary relational matching tasks. I: Relational matching same—distinct stimuli. II:
Relational matching same—more difficult discrimination. III: Relational matching different.
10 SIRI MING and IAN STEWART
research with humans has focused on analogy matching responses (e.g., given a sample of a pic-
based on arbitrary relations. This seems to be a ture of a banana and the vocal or textual cue
significant gap, as it would appear to be a “same,” selecting another picture of a banana
potentially important element of a generalized rather than an orange, and vice versa given the
repertoire of responding to differences, as well vocal or textual cue “different”). This is a similar
as a potentially important precursor to analogy. task to the previously described oddity matching
A number of different issues might be exam- tasks in that it involves the selection of a stimu-
ined. One question concerns the stage in the lus that is different from another stimulus; how-
typical developmental sequence at which non- ever, in the case of nonarbitrary listener
arbitrary relational matching begins to be seen discriminations, the response is under the
in children. For example, does relational specific—and changing—control of the contex-
matching typically precede or follow relational tual cue of either same or different. In oddity
tacting and/or listener behavior, and might this matching, the task does not change between
order be changed based on targeted training? A selection of same and different—only the differ-
second question concerns the best methods by ent item must be selected from the array. Nonar-
which to establish and strengthen this reper- bitrary relational tacts require identifying the
toire. For instance, how might training of relation between items, either through a topo-
related repertoires such as relational tacting graphical response (e.g., emitting a vocal or sign-
and/or listener training compare, or interact based response of “same” or “different” when
with multiple exemplar training (MET) of the shown two bananas vs. a banana and an orange)
repertoire itself as a means of establishing or or selection of the appropriate relational name
strengthening this capacity? Might training (e.g., selecting a text or symbol card indicating
involving the matching of a range of different same vs. different when shown a pair of identical
varieties of relations be better than training vs. nonidentical stimuli).
involving a more limited set? How effective Relational tacts and listener responses involve
might the use of stimulus arrays similar to responding to identical versus nonidentical
those used in establishing oddity selection items as the same or different, as well as
(e.g., a comparison array of many same pairs responding on the basis of how two items
and one different pair, with a different pair as might be the same or different from one
sample, or vice versa) be? A further issue con- another. Examples abound not only in pre-
cerns the effect of training nonarbitrary rela- school worksheets, but also in other arenas,
tional matching on the emergence of arbitrary such as training musical listening skills by
relational matching (the form of behavior to requiring learners to listen to two tones and
which the label “analogical reasoning” is most identify if they are the same or different
frequently applied) and/or other arbitrary rela- (e.g., EarTrainingandImprov.com, 2012). The
tional (i.e., relational framing) abilities as well classic Sesame Street game of “One of these
as on intellectual performance more generally. things is not like the others” (Raposo, Stone &
Hart, 1970, track A7) is a good example of
these kinds of tasks at a slightly more advanced
Nonarbitrary Relational Listener level—that is, having to identify an item as
Discriminations and Relational Tacts being different from the others on the basis of
Nonarbitrary relational listener discrimina- features other than simple identity/oddity (see
tions require selecting a comparison under the Figure 3 for examples). Examples at the early
control of relevant contextual cues—stimuli elementary level include Common Core Math
discriminative for either same or different standards that require students to “analyze and
A REVIEW OF DIFFERENCE RELATIONS 11
compare two- and three-dimensional shapes, in successfully trained to make one response in
different sizes and orientations, using informal the presence of two same novel items, and
language to describe their similarities, differ- another response in the presence of two differ-
ences, parts (e.g., number of sides and vertices/ ent novel items. For example, Katz and Wright
”corners”) and other attributes (e.g., having (2006) taught pigeons to peck an upper picture
sides of equal length),” and Common Core Lit- as the sample, which brought up two compari-
eracy standards that require students to “iden- sons: another picture and a white rectangle.
tify basic similarities in and differences between The pigeons were taught to peck the compari-
two texts on the same topic (e.g., in illustra- son picture if it was the same as the sample,
tions, descriptions, or procedures)” (National and to peck the white rectangle if the two pic-
Governors Association Center for Best Prac- tures were different.
tices & Council of Chief State School Offi- Pepperberg (1987, 1988) trained an African
cers, 2010). Grey parrot (Alex) in several more complex
Nonarbitrary relational listener discrimina- tasks requiring second order same/different
tions have been studied and demonstrated with relational tacting. As Alex had been previously
a number of primate species, both in a MTS trained to identify abstract categories of color,
format in which either an identical or noniden- shape, and material, these categories were used
tical comparison is selected (e.g., Burdyn & for training him to tact what properties of two
Thomas, 1984; Robinson, 1955; Thomas & items were either the same or different, or if no
Kerr, 1976), and in which an identical or noni- properties were same or different. After multi-
dentical comparison pair is selected, depending ple exemplar training with various pairs of
on a cue (e.g., Flemming, Beran, & Washburn, items with the questions, “What’s same?” or,
2007). Depending on the cue displayed “What’s different?” Alex correctly responded to
(e.g., colors or patterns), selecting one or the pairs of novel objects by vocally tacting same/
other of the comparisons is reinforced. Such different relations with respect to color, shape,
tasks might require responding on the basis of material, or stated that none of the properties
a single dimension (e.g., shape) throughout, or were the same or different.
might require a more complex level of respond- Relevant to relational tacting of same/differ-
ing such as selecting among multidimensional ent, Serna et al. (1997) described research on
stimuli for an item that is the same or different developing nonvocal procedures for humans
with respect to a specific dimension; for exam- that are very similar to some of the procedures
ple, same color or different size (e.g., Vonk, used with nonhuman subjects as described
2003; see Figure 4 for examples). above. A “blank comparison” procedure allows
There is also substantive data from nonhu- nonvocal individuals to identify (i.e., tact)
man research that has investigated nonarbitrary whether or not a comparison is the same as the
same/different relational tacting. Using discrim- sample. With these procedures, an individual
ination training and MET, a variety of species can be taught to press a matching comparison
have been taught to identify pairs of items as if it is the same as (or similar to) the sample,
being the same or different (e.g., Blaisdell & and to press a blank comparison if not.
Cook, 2005; Flemming et al., 2007; Katz & Training nonarbitrary relational listener and
Wright, 2006; Mercado, Killebrew, Pack, tacting skills with humans could reasonably be
Mácha, & Herman, 2000; Pepperberg, 1987, guided by procedures used in the nonhuman
1988). Though required response topographies literature in addition to procedures commonly
have varied (e.g., token selection, key press, used in language intervention programs for
paddle press), in all instances, the subjects were other types of tacts and listener behaviors
12 SIRI MING and IAN STEWART
(e.g., Sundberg & Partington, 2010). However, this task being present in several curricula for
we have not found any studies examining children with autism (e.g., Center for Autism
teaching nonarbitrary relational same/different and Related Disorders, 2013; Leaf & McEa-
listener discriminations with humans, despite chin, 1999; Maurice et al., 1996; Partington,
A REVIEW OF DIFFERENCE RELATIONS 13
2006). Obvious directions for future research, investigated as a foundation for later arbitrary
therefore, might involve assessment and where yes/no responding. This would involve identi-
necessary, training of these skills in both typi- fying whether a pair of items is or is not same
cally developing and developmentally delayed or different with nonarbitrary stimuli
children. This work might involve assessing (e.g., responding to the question, “Are these
and training listener and tacting repertoires the same?” or, “Are these different?” with phys-
using protocols similar to those previously ically identical vs. different items), or finding
employed in nonhuman work, and perhaps an item that is not the same or not different
comparing them in terms of their efficacy. It is (e.g., see Hayes, Stewart, & McElwee, in press).
also important to investigate at what point arbi- Students who have difficulty with this task
trary (rather than nonarbitrary) responding to might benefit from the procedures described by
relations of difference should be introduced. Serna et al. (1997) and similar procedures used
Research has found that training nonarbitrary with nonhumans—that is, procedures in which
responding can provide an important founda- the learner is taught to select a matching com-
tion for arbitrary responding (e.g., Barnes- parison or to select a blank card (or make some
Holmes, Barnes-Holmes, & Smeets, 2004; other response) if the comparison is different
Berens & Hayes, 2007). Similarly, yes/no from the sample. It would be informative to
responding with nonarbitrary stimuli in rela- determine if nonarbitrary yes/no responding is
tions of same and different might also be facilitative or necessary for arbitrary yes/no
14 SIRI MING and IAN STEWART
relational responding. At the arbitrary level, an example, imagine that we tell a child that we
example of such responding would use picture- have three foreign coins (A, B and C) and that
name combinations (e.g., “Is this a cat?” or, A is the same as B and B is different from C. If
“Find the one that is not a cat.”), in which the we next ask her whether A and C are same or
relation is based on social convention (the word different, she may be able to tell that they are
is the name of the picture) rather than physical different, even though she hasn’t seen any of
attributes. Parenthetically, we would note that the coins. She can do this based on a learning
many programs for children with autism teach history involving the contextual cues same and
such yes/no tacting and listener discriminations different, rather than on physical properties.
without necessarily testing for generalized Responding to stimuli in arbitrary same/dif-
responding with yes/no at the earlier, nonarbi- ferent relations such as this facilitates rapid
trary level of physical identity/difference learning and generativity to an extent that nei-
(e.g., Center for Autism and Related Disorders, ther nonarbitrary relational responding nor
2013; Leaf & McEachin, 1999; Partington, direct operant training can (e.g., Stewart &
2006). Research on this issue might be Roche, 2013). Of particular importance, such
extremely helpful for developing intervention skill likely provides the foundation for further
curricula. key repertoires including classifying objects and
In summary, nonarbitrary difference rela- events together and distinguishing them across
tional responding has been demonstrated with multiple different verbal dimensions (see, for
a variety of species across several types of rela- example, Hayes, Gifford, Townsend &
tional responses, and there are well- Barnes-Holmes, 2001; Gil, Luciano, Ruiz &
documented experimental procedures in the lit- Valdivia-Salas, 2012; Slattery & Stewart, 2014).
erature. However, there appears to be a signifi- Furthermore, recent empirical evidence of sub-
cant gap with respect to using these procedures stantial improvements in reading comprehension
with children who have skill deficits in this based on MET of arbitrary same and different
area, and there is no clear evidence for a partic- relations would appear to cohere with this con-
ular sequence or hierarchy of teaching relational clusion (Newsome, Berens, Ghezzi, Aninao, &
tacting, listener discriminations, or matching. Newsome, 2014).
Relevant research into assessment, training, and Arbitrary relational responding of difference/
interaction of these skills is needed, including, distinction might also involve listener
for example, exploring which ones might most responses, relational tacting, or relational
efficiently facilitate the others, and thus be tar- matching. The common factor in all cases is
geted first, as well as what effect this might that responses would be primarily controlled by
have on arbitrary relational responding within contextual cues (e.g., different) and involve arbi-
frames of difference and beyond. trarily designated relations of difference
(e.g., food or music preferences, the coins men-
tioned above, or that cat is the same as gato but
ARBITRARY RELATIONAL RESPONDING different from chien). Moreover, such respond-
As discussed earlier, relational responding ing can involve the emergence of a new
can be either nonarbitrary or arbitrary. In arbi- response on the basis of taught relations (i.e., a
trary relational responding (also called relational derived response, as when stimulus equivalence
framing; Hayes, Barnes-Holmes & Roche, is demonstrated for a relation of sameness, or as
2001), stimuli are related (e.g., as the same as when the child in the example above could tell
or different from each other) based on contex- that coins A and C were different; see also
tual cues rather than physical properties. For Ming, Moran, & Stewart, 2014). In the
A REVIEW OF DIFFERENCE RELATIONS 15
sections to follow, we will consider behavior a young infant. The first version, which was
analytic research relevant to this type of first recorded at 16 months of age, involved
responding, beginning with possible precursors selecting a novel versus known animal picture
to arbitrary difference relational responding and given a novel name (i.e., the straightforward
then arbitrary difference relational responding version of exclusion described in the previous
itself, including arbitrary relational tacts, lis- paragraph). The other, which was first recorded
tener discriminations, and relational matching. at 23 months of age, was a version in which
We will start with precursors to arbitrary differ- responding was based on a combination of
ence relations. both exclusion and derived relational respond-
ing. More specifically, in this case the child had
to not only select the novel animal, but also
Precursors to Responding to Arbitrary Relations tact that animal without further training. This
of Difference. latter performance, which might be referred to
There are two repertoires that can be argued to as derived exclusion, might seem particularly
be precursors to, or at least skills supportive of, relevant as a precursor to demonstrating derived
arbitrary difference relations: exclusion and yes/no relations of difference.
responding. The phenomenon of exclusion is seen Although the exact importance of exclusion
when, given a history of well-established MTS per- in both its basic and derived forms to differ-
formance with a set of arbitrarily related stimuli, a ence responding remains to be investigated,
novel unfamiliar comparison stimulus is selected in exclusion is an important skill for the estab-
the presence of a novel unfamiliar sample stimulus lishment of emergent matching (Wilkinson
without additional training. As such, exclusion pro- et al., 1998). Thus, one important area for
cedures can lead to matching of arbitrary stimuli in future research is to develop and test proce-
the absence of any explicit reinforcement (emer- dures for establishing basic and derived exclu-
gent matching; see Wilkinson, Dube, & McIlvane, sion with individuals who do not yet readily
1998, for a review of this research). Exclusion demonstrate them. In this regard, MET with
seems relevant to arbitrary relations of difference novel sets of stimuli may be an avenue to
because it appears to involve responding influenced explore (that is, by training the exclusion
by perceived incongruity. In fact, some theorists response with one set of stimuli and then test-
have suggested that performance on arbitrary MTS ing with a novel set of stimuli, and repeating
tasks may involve not simply selecting as necessary). Another important direction for
(i.e., responding towards) the correct comparison future research would be to determine any
(S+) but also excluding or rejecting potential importance of demonstrating both
(i.e., responding away from) the incorrect basic and derived exclusion, from the perspec-
comparison(s) (S-; e.g. Carrigan & Sidman, 1992; tive of whether either of these abilities is cor-
Dixon & Dixon, 1978; Stromer & Osborne, related with, or might support the learning of,
1982). In any event, testing for, and if necessary nonarbitrary and arbitrary difference relations.
training exclusion might be important for ensuring For example, research might investigate
a well-established arbitrary MTS repertoire prior to whether children taught to perform nonarbi-
assessing and training arbitrary difference relations. trary and/or arbitrary relational responding in
Exclusion and derived relational responding accordance with difference (such as relational
(DRR) may also be examined in combination. listener discriminations or tacts of difference)
For example, Lipkens, Hayes, and Hayes can also readily show derived exclusion, or
(1993) showed two versions of exclusion when whether teaching derived exclusion speeds up
recording instances of DRR in the behavior of the learning of such repertoires.
16 SIRI MING and IAN STEWART
Another potential precursor to arbitrary rela- and Egel (1984) examined teaching procedures
tions of difference is responding to yes or no for teaching yes/no mand and tact responses.
questions about arbitrary sameness relations. Children with developmental delays (ages 4-6
Relevant to a yes/no repertoire, “go/no-go” pro- years old, with assessed age-equivalencies on
cedures (e.g., Debert, Amelia Matos, & McIl- standardized IQ tests between 2 and 3 years
vane, 2007; Modenesi & Debert, 2015) have old) were taught yes/no mands through
been used to train responding (e.g., clicking a embedded instructional procedures (mand
mouse) only when a pair of stimuli have been training). However, this skill did not generalize
arbitrarily established as “going together” either to yes/no tacts; the latter were only acquired
in training or through equivalence relations when the teaching procedures included specifi-
(and not responding if they have not)—this cally programming alternating trials of tacts and
can be seen as analogous to identifying the pair mands. Shillingsburg, Kelley, Roane, Kisamore,
of items as being the same or different from and Brown (2009) examined procedures for
each other. What would generally be termed teaching children with autism yes/no mands,
descriptive autoclitic tacts of assertion and tacts, and intraverbals, with a focus on deter-
negation (Skinner, 1957) involve similar mining the functional independence of these
responses. In this case, negation would imply responses. They found that responses of yes/no
not-same or different. For example, when could be taught to children as mands
answering the question, “Is this a cat?” when (e.g., “Do you want a chip?”), tacts (e.g., “Is
presented with a dog, saying “no” would this a cup?”), and intraverbals (e.g., “Does a
require the student to identify the item pre- dog say moo?”) and generalized within but not
sented as being not the same as, or different across operants. We would note that additional
from, the word cat. As an intraverbal response, information about participants’ other relational
answering the question, “Does a cat say woof?” responding skills at both nonarbitrary and arbi-
requires the student to identify that cat and trary levels (for example, relational listener dis-
woof don’t belong together or that they relate criminations and tacts of nonarbitrary same
to different things. In these cases, the relation and different relations; arbitrary sameness
of the stimuli cat, dog, or woof to each other is [equivalence]) might be helpful in determining
arbitrary. Similarly, selecting an item on the factors that could potentially facilitate such
basis of its (arbitrarily designated) difference generalization (see also Stewart, McElwee &
from a sample could also include identifying an Ming, 2013).
item that is not the same as a sample—such as A critical direction for future research is to
in common teaching procedures for negation identify any relation between arbitrary yes/no
(e.g., Leaf & McEachin, 1999), in which a stu- responding and both nonarbitrary and arbitrary
dent is asked, for example, to find the stimulus difference responding. For example, arbitrary
that is not a cat from an array that includes a yes/no responding with respect to sameness
cat and a dog. (e.g., saying “yes” or “no” in response to the
There is little applied research on teaching question, “Are these cups?”) might facilitate arbi-
individuals such arbitrary conditional discrimi- trary difference responding (e.g., selecting a
nations with respect to yes/no responding to quarter rather than two nickels when asked to
relations of sameness when they do not already find an amount different from a dime).
do so. Two studies have attempted to teach Responding to yes/no questions with respect to
participants to say “yes” or “no” as tact or intra- nonarbitrary sameness/difference (e.g., answer-
verbal responses, using discrete trial teaching ing “no” to the question, “Are these different?”
procedures. In the first of these, Neef, Walters when shown a pair of identical items) might
A REVIEW OF DIFFERENCE RELATIONS 17
facilitate responding to yes/no questions about of relational framing in accordance with dis-
arbitrary relations (e.g., answering “no” when tinction. There is very little research on how to
shown five dimes and asked, “Are these different establish a repertoire of derived difference rela-
from two quarters?”), or to other forms of arbi- tional responding. Most research into derived
trary difference responding. Future research relations with very young children and children
might also examine the use of yes/no procedures with developmental delays has focused on
as a means of training both nonarbitrary and derived sameness (i.e., stimulus equivalence;
arbitrary relational responding, as discussed pre- Luciano, Gomez Becerra, & Rodriguez Val-
viously with respect to nonarbitrary relational verde, 2007). Nonetheless, a few studies have
tacts and listener discriminations. examined the emergence of untrained (derived)
difference relations. O’Connor, Barnes-Holmes,
and Barnes-Holmes (2011) investigated rela-
Deriving Arbitrary Relational Listener tional responding in typically developing chil-
Discriminations And Relational Tacts: dren and children with autism. Using a two-
Relational Framing comparison MTS task, contextual control was
As we have explained previously, arbitrary established for emitting either the symmetrical
relational responding is based on contextual response (i.e., train A1-B1, test for B1-A1;
control rather than physical properties of sti- e.g., Sidman, 2009) or the alternative asymmet-
muli. As such, if appropriate contextual cues rical response (i.e., train A1-B1, test for B2-A1;
are used with respect to arbitrary stimuli and a see also Boelens & van den Broek, 2000). The
relational pattern is established (e.g., A = B and cues used in this protocol were blue or red cir-
C 6¼ B, as in our earlier example involving the cles, which might be interpreted as the same
child and the coins), then additional responses and not the same, respectively. In this series of
may emerge without further training. These experiments, children were first taught to read
untrained responses can be seen as derived rela- nonsense words (e.g., saying “vug” [A1] when
tional listener discriminations or derived rela- presented with the text card VUG [B1; A1-B1]
tional tacts, dependent on the topography of or saying “lup” when presented with LUP [A2-
the response, but all can be viewed as part of B2]). They were then trained to select the
the generalized operant of relational framing. related (same) comparison when presented with
For example, if a child is taught to tact coin A the sample and a blue circle, and to select the
when shown coin B as being the same and to unrelated (not same) comparison when pre-
tact coin C when shown coin B as being differ- sented with the sample and a red circle. Thus,
ent and can then tact coin A when shown coin upon hearing “vug” in the presence of the red
C as being different, she will have demon- circle, the participants were trained to select the
strated a derived relational tact. If, on the other LUP text (B1-A2). The red circle might be
hand, she has been taught to select various described as cuing selection of “not the same”
coins given a sample and the contextual cue of or “different from” stimuli. Once contextual
same or different, and can then select coins control was established over trained same and
based on combining taught relations not same relations, the typically developing
(e.g., selecting coin A when shown C and told children demonstrated generalized and derived
to “find different”), then she would be demon- contextually controlled relational responses to
strating a derived relational listener response. novel stimulus sets. Not all the children with
More importantly, both of these response topo- autism did so initially, but for these children,
graphies involve deriving a new arbitrary rela- MET resulted in contextual control over the
tional response and can be considered examples derived responses. This could be described as
18 SIRI MING and IAN STEWART
et al. argued that arbitrary relational responding establishing responding in accordance with dis-
is critical to reading comprehension, and that tinction would be to use procedures for estab-
responding in terms of both distinction and lishing yes/no responding that have been
coordination is foundational for skills such as developed with the Relational Evaluation Pro-
comparing and contrasting, making predictions, cedure (REP; Barnes-Holmes, Hayes,
and integrating concepts found in text. In this Dymond & O’Hora, 2001; Cullinan, Barnes, &
study, explicit training was provided in relating Smeets, 1998; Cullinan, Barnes-Holmes &
stimuli on the basis of similarity or difference. Smeets, 2000, 2001) and the Implicit Rela-
Following training in identifying the categories, tional Assessment Procedure (IRAP; see Barnes-
features and functions of a variety of common Holmes, Barnes-Holmes, Stewart, & Boles,
stimuli, participants were next trained to flu- 2010). As the names suggest, the aim of these
ency in several different relational tasks invol- procedures is to test and train participants to
ving distinction and coordination. For example, evaluate stimulus relations (both nonarbitrary
participants might be asked to identify how a and arbitrary) using relational tacting proce-
bus and a dog are the same or different. Partici- dures. For example, the IRAP has participants
pants were also asked questions about activities, respond to relations between natural language
such as, “How is playing in the park like stimuli through time-pressured selection of nat-
[or different from] swimming in the pool?” ural language cues (i.e., the actual words yes,
This protocol improved both responding on no, same, different, etc.) when presented with
similar relational tasks with novel stimuli and pairs of stimuli and a relational cue (see
performance on standardized measures of read- Figure 5 for examples).
ing comprehension. In applying this procedure to establishing a
As noted, there is little research on establish- repertoire of arbitrary difference responding, for
ing a repertoire of responding to arbitrary rela- example, one might present a picture, a text
tions of difference, and thus much further stimulus, and a same or different stimulus, and
investigation is needed. Luciano et al. (2009) require a yes/no response—for instance, if a
made a number of general suggestions for teach- picture of a cat is paired with the text dog and
ing the earliest frames of relational responding, the word different (or the student is asked, “Are
including coordination and distinction, such as they different?”), then the required response
using MET in the specific pattern of relational would be “yes.” Training could also be directed
responding, mixing different relations to at establishing and strengthening second order
improve flexibility, using specific and consistent contextual control over arbitrary relations
cues for the relations being trained, and begin- (e.g., same or different according to specified
ning with nonarbitrary relations before transi- features, functions, or categories, and identify-
tioning to arbitrary relations. With respect to ing how stimuli are the same or different;
distinction, these suggestions require further e.g., Newsome et al., 2014). For example, a
empirical investigation, although the results of game might be devised with key vocabulary
O’Connor et al. (2011) with respect to symmet- words (such as from passages in a remedial
rical/asymmetrical responding, and Ming (2015) reading text), and the teacher might ask the
with respect to transformation of function in student to sort them in particular ways or iden-
combined relations of coordination and distinc- tify how they are related. Given a selection of
tion would certainly suggest that MET in the animals, questions might include finding some-
relevant relational patterns may be successful. thing that lives in the same/different place as
Another worthwhile avenue to investigate (than) another, or has the same/different num-
with respect to possible procedures for ber of legs, and so on.
20 SIRI MING and IAN STEWART
SAME DIFFERENT
Cat . . Cow
YES NO YES NO
SAME DIFFERENT
Pig Horse .
YES NO YES NO
Figure 5. IRAP-type relational responding tasks. Tabletop versions could use text cards for the names, and the
teacher could ask the relevant question, for example, “Are they different?”.
Beginning with familiar stimuli, and estab- the combination of difference relations. For
lishing new, arbitrary relations under the con- example, if the lion likes different food from
textual control of cues for same/different, and the zebra and the zebra likes different food
testing for transformation of function might from the frog, then the student could be asked
also provide a foundation for the more complex whether the lion and the frog like the same or
relational responding required at the level of different. The correct answer is that we do not
know. Another way of proceeding might of
deriving responses in frames of distinction and
course be to look for generalization of either
provide another avenue for future research. At combined same/different or different/different
first, this is likely best accomplished in combi- relations to a novel context; for example, one
nation with coordinate relations, because iden- could teach that a cow and a horse live in the
tification of a relation as unspecified—that is, if same place, but a cow and an axolotl live in dif-
A is different from B, and C is different ferent places, and then ask if a horse and an axo-
from B, we cannot know what the relation is lotl live in the same place or in different places.
between A and C—is relatively difficult even
for typically developing adults (see, for exam- Relational Matching
ple, Vitale, Barnes-Holmes, Y., Barnes-Holmes, We will conclude by examining relational
D., & Campbell, 2008). For instance, the food matching, or analogy, involving arbitrary rela-
preference protocol described earlier involved tions of difference. A number of behavior ana-
testing correct derivation of an untrained food lytic studies have modeled analogy as the
preference based on the combination of same derivation of relations between derived rela-
and difference relations. One more advanced tions, including both equivalence relations and
test within this protocol might be to look for nonequivalence relations. For example, Barnes,
A REVIEW OF DIFFERENCE RELATIONS 21
Hegarty, and Smeets (1997) first trained and responding with individuals who do not yet
tested participants (typically developing adults) demonstrate such skills. These protocols pro-
for the formation of four three-member equiva- ceed from training baseline conditional discri-
lence relations (which can be designated as A1- minations (e.g., A1-B1, A2-B2, A1-C1, A2-
B1-C1, A2-B2-C2, A3-B3-C3, A4-B4-C4), C2) to training and testing relations between
and then showed that, without further training, both matched (e.g., A1B1-A2B2) and
they matched compound stimuli composed of unmatched (e.g., A1B2-A2B1) pairs of stimuli,
stimuli in an equivalence relation to each other and then testing for derived relational respond-
(e.g., B1C1 to B3C3; referred to as ing with respect to relations between both
equivalence–equivalence responding) and that matched and nonmatched pairs (e.g., B1C1-
they also matched compound stimuli composed B2C2, C1B2-C2B1). In line with these recom-
of stimuli not in an equivalence relation to each mendations, matching pairs of stimuli that have
other (e.g., B1C2 to B3C4; referred to as been trained in arbitrary relations of either
nonequivalence–nonequivalence responding). same or different might be an important step
The latter phenomenon of matching nonequi- in teaching frames of distinction as well as
valence relations to nonequivalence responding establishing analogy. For example, if a student
can be considered an example of matching arbi- is taught to identify sets of animals as liking
trary relations of distinction and a number of particular foods (e.g., the bear, lion, and tiger
other studies since have replicated this phe- like popcorn, and they like different food from
nomenon (e.g., Carpentier, Smeets & Barnes- the giraffe, zebra, and elephant, which like
Holmes, 2002, 2003; Stewart, Barnes-Holmes, cake) a task might be to match on the basis of
Roche, & Smeets, 2001, 2002; Stewart, whether particular pairs of animals represent
Barnes-Holmes, & Roche, 2004). the same relation. For example, a pair com-
Recently, Miguel et al. (2015) combined the posed of the lion and giraffe (different) should
use of relational tacting with analogy proce- be matched to a pair involving the bear and
dures. In these studies, college students were zebra (different) rather than a pair involving
trained to tact abstract stimuli belonging to the the lion and tiger (same). This type of activity
same class with a common name and then to could be conducted first with taught relations
tact the relations between pairs of stimuli as as the elements making up the pairs, and subse-
same or different. Following this training, parti- quently with derived relations in the pairs.
cipants tacted novel pairs as same or different Again, research is necessary to determine an
and accurately matched pairs that were same or appropriate sequence of such training in rela-
different (e.g., A1C1 to A2C2 or A1C2 to tion to the other skills described in this
A2C1). Participants who failed tests of rela- section as well as to determine the most effec-
tional tacting also failed the matching (analogy) tive set of procedures for establishing this reper-
tests. toire when it appears to be absent.
Finally, most of the work carried out so far
has involved typically developing individuals.
However, this research provides an important CONCLUSIONS
foundation for work with populations with As noted earlier, comparing/contrasting and
developmental delays. Stewart, Barnes-Holmes, identifying similarities and differences are preva-
and Weil (2009) describe protocols for asses- lent in curricular standards, and teaching activ-
sing and training analogy on the basis of equiv- ities that focus on these skills have been singled
alence relations that might be investigated from out as having particularly important impacts on
the perspective of establishing analogical academic performance (Marzano, et al., 2001).
22 SIRI MING and IAN STEWART
In addition, recent research supports the sequencing of these skills and the most effective
importance of same/different relational teaching procedures for establishing each
responding in various contexts. Performance of them.
on relational tasks that include derived In summary, in this paper we attempted to
responding to same and different relations has organize existing literature on same/different
also been correlated with the vocabulary and relational responding from comparative psy-
arithmetic subtests of the Wechsler Adult chology, experimental behavior analysis, and
Intelligence Scale–Third Edition (O’Hora, applied behavior analysis both to stimulate
Pelaez, & Barnes-Holmes, 2005) and with the applied research and to encourage practitioners
verbal subtest of the Kaufmann Brief Intelli- to consider relations of difference more system-
gence Test (O’Toole & Barnes-Holmes, atically in their programming. Practitioners
2009), and activities that capitalize on deriving might begin to incorporate this work by using
arbitrary relations of sameness and difference existing research protocols (such as for oddity
have been shown to improve reading compre- and relational matching) and considering differ-
hension (Newsome et al., 2014). O’Toole and ence responding as a continuum from nonarbi-
Barnes-Holmes further suggest that fluent trary to arbitrary relational responses as they
responding to congruent and incongruent rela- develop teaching plans. There are clearly many
tions under contextual control may be a key gaps to be filled, however, as there is very little
indicator of relational flexibility, which has applied research with humans who do not yet
itself long been regarded as a critical compo- demonstrate either arbitrary or nonarbitrary
nent of cognitive ability (e.g., Premack, 2004). responding in accordance with difference.
Even though there are no studies examining Additionally, there is little in the way of curri-
how best to establish frames of distinction cula advising how to teach difference respond-
when individuals cannot demonstrate such ing to children with autism. As Rehfeldt
untaught relational responses, this skill is clearly (2011) has also noted, it is critical for the field
important to language development as well as to move beyond relations of sameness and
many academic skills and cognitive develop- apply the derived relational responding para-
ment more generally (Marzano et al., 2001; digm to other relations if we are to develop
McIlvane et al., 2011; O’Toole & Barnes- practical teaching procedures for the many
Homes, 2009). We argue that by conceptualiz- complex language and cognitive skills that are
ing same/different responding as a continuum necessary for academic success. We hope that
of responding from nonarbitrary to arbitrary the summaries and suggestions in this review
distinction relational responding, a path will stimulate such research with regard to rela-
becomes clear for research (and future curricu- tions of difference and that a more conceptually
lum development) on a likely hierarchy of com- systematic and empirically validated set of cur-
ponent/composite skills as well as teaching ricular recommendations might develop in tan-
procedures. Considerable research is warranted dem with such work.
-0
to investigate the influence and optimal
REFERENCES
Addyman, C., & Mareschal, D. (2010). The perceptual de Alcântara Gil, M. S., de Oliveira, T. P., &
origins of the abstract same/different concept in McIlvane, W. J. (2011). Conditional discriminations
human infants. Animal Cognition, 13, 817-833. by preverbal children in an identity matching-to-
doi:10.1007/s10071-010-0330-0 sample task. The Psychological Record, 61(3), 327-
A REVIEW OF DIFFERENCE RELATIONS 23
to-sample. American Journal on Mental Retardation, Murphy, C., & Barnes-Holmes, D. (2010). Establishing
107, 445-454. Retrieved from http://aaiddjournals. complex derived manding in children with and with-
org/loi/ajmr.1 out a diagnosis of autism. The Psychological Record,
Marzano, R. J., Pickering, D., & Pollock, J. E. (2001). 60, 489-504. Retrieved from http://www.springer.
Classroom instruction that works: Research-based strate- com/psychology/psychology+general/journal/40732
gies for increasing student achievement. Alexandria, VA: National Governors Association Center for Best Prac-
Association for Supervision and Curriculum tices, & Council of Chief State School Officers
Development. (2010). Common core standards. Washington, D.C.:
Maurice, C., Green, G., & Luce, S. (1996). Behavioral National Governors Association Center for Best Prac-
intervention for young children with autism: A manual tices, Council of Chief State School Officers.
for parents and professionals. Austin, TX: PRO-ED. Retrieved from http://www.corestandards.org/
McIlvane, W. J. (2009). Translational behavior analysis: Neef, N. A., Walters, J., & Egel, A. L. (1984). Establish-
From laboratory science in stimulus control to inter- ing generative yes/no responses in developmentally
vention with persons with neurodevelopmental dis- disabled children. Journal of Applied Behavior Analysis,
abilities. The Behavior Analyst, 32, 273-280. 17, 453-460. doi:10.1901/jaba.1984.17-453.
Retrieved from http://www.ncbi.nlm.nih.gov/pmc/ Newsome, K.B., Berens, K.N., Ghezzi, P.M.,
journals/557/ Aninao, T., & Newsome, W.D. (2014). Training
McIlvane, W. J., Dube, W. V., Serna, R. W., Lionello- relational language to improve reading comprehen-
DeNolf, K. M., Barros, R. D. S., & Galvao, O. F. sion. European Journal of Behaviour Analysis, 15, 165-
(2011). Some current dimensions of translational 197. Retrieved from http://www.tandfonline.com/loi/
behavior analysis: From laboratory research to inter- rejo#.V6SvomgrLb0
vention for persons with autism spectrum disorders. Nissen, H. W., Blum, J. S., & Blum, R. A. (1948). Anal-
In J. A. Mulick & E. A. Mayville (Eds.), Behavioral ysis of matching behavior in chimpanzee. Journal of
foundations of effective autism treatment (pp. 155- Comparative and Physiological Psychology, 41, 62-74.
181). Cornwall-on-Hudson, NY: Sloan Publishing. doi:10.1037/h0061040
McLay, L. K., Sutherland, D., Church, J., & Tyler- O’Connor, J., Barnes-Holmes, Y., & Barnes-Holmes, D.
Merrick, G. (2013). The formation of equivalence (2011). Establishing contextual control over symme-
classes in individuals with autism spectrum disorder: try and asymmetry performances in typically develop-
A review of the literature. Research in Autism Spec- ing children and children with autism. The
trum Disorders, 7, 418 - 431. doi:10.1016/j. Psychological Record, 61, 287-312. Retrieved from
rasd.2012.11.002 http://www.springer.com/psychology/psychology
Mercado, E., Killebrew, D. A., Pack, A. A., +general/journal/40732
Mácha, I. V., & Herman, L. M. (2000). Generaliza- Oden, D. L., Thompson, R. K. R., & Premack, D.
tion of ‘same–different ‘classification abilities in bot- (1990). Infant chimpanzees spontaneously perceive
tlenosed dolphins. Behavioural Processes, 50, 79-94. both concrete and abstract same/different relations.
doi:10.1016/S0376-6357(00)00100-5 Child Development, 61, 621-631. doi:10.1111/
Miguel, C. F., Frampton, S. E., Lantaya, C. A., j.1467-8624.1990.tb02807.x
LaFrance, D. L., Quah, K., Meyer, C. S., O’Donnell, J., & Saunders, K. J. (2003). Equivalence
Fernand, J. K. (2015). The effects of tact training on relations in individuals with language limitations and
the development of analogical reasoning. Journal of mental retardation. Journal of the Experimental Analy-
the Experimental Analysis of Behavior, 104, 96-118. sis of Behavior, 80, 131-157. doi:10.1901/
doi:10.1002/jeab.167 jeab.2003.80-131
Ming, S. (2015). Assessing and training early emergent O’Hora, D., Pelaez, M., & Barnes-Holmes, D. (2005).
derived relational responding in children with autism Derived relational responding and performance on
(Doctoral thesis, National University of Ireland, Gal- verbal subtests of the WAIS-III. The Psychological Rec-
way, Ireland). Retrieved from http://hdl.handle.net/ ord, 55, 155-175. Retrieved from http://www.
10379/5549 springer.com/psychology/psychology+general/journal/
Ming, S., Moran, L., & Stewart, I. (2014). Derived rela- 40732
tional responding and generative language: Applica- Ortony, A. E. (1979). Metaphor and thought. New York,
tions and future directions for teaching individuals NY: Cambridge University Press.
with ASD. European Journal of Behavior Analysis, 15, O’Toole, C., & Barnes-Holmes, D. (2009). Three chron-
199-224. Retrieved from http://www.tandfonline. ometric indices of relational responding as predictors
com/loi/rejo#.V6SynGgrLb0 of performance on a brief intelligence test: The
Modenesi, R. D., & Debert, P. (2015). Contextual con- importance of relational flexibility. The Psychological
trol using a go/no-go procedure with compound Record, 59, 119-132. Retrieved from http://www.
abstract stimuli. Journal of the Experimental Analysis of springer.com/psychology/psychology+general/journal/
Behavior, 103(3), 542-552. doi:10.1002/jeab.154 40732
26 SIRI MING and IAN STEWART
Partington, J. W. (2006). The assessment of basic language Experimental Analysis of Behavior, 73, 163-176.
and learning skills - revised (the ABLLS-R). Pleasant doi:10.1901/jeab.2000.73-163
Hill, CA: Behavior Analysts, Inc. Smirnova, A., Zorina, Z., Obozova, T., & Wasserman, E.
Pepperberg, I. M. (1987). Acquisition of the same/differ- (2015). Crows spontaneously exhibit analogical rea-
ent concept by an African grey parrot (Psittacus soning. Current Biology: CB, 25(2), 256-60.
erithacus): Learning with respect to categories of doi:10.1016/j.cub.2014.11.063
color, shape, and material. Animal Learning & Behav- Soraci, S. A., Carlin, M. T., & Bray, N. W. (1992). Stim-
ior, 15, 423-432. doi:10.3758/BF03205051 ulus organization and relational learning. Interna-
Pepperberg, I. M. (1988). Comprehension of "absence" tional Review of Research in Mental Retardation, 18,
by an African grey parrot: Learning with respect to 29-53. doi:10.1016/S0074-7750(08)60115-9
questions of same/different. Journal of the Experimen- Soraci, S. A., Deckner, C. W., Baumeister, A. A., &
tal Analysis of Behavior, 50, 553-564. doi:10.1901/ Bryant, J. T. (1991). Generalized oddity performance
jeab.1988.50-553 in preschool children: A bimodal training procedure.
Premack, D. (2004). Is language the key to human intelli- Journal of Experimental Child Psychology, 51, 280-
gence? Science, 303, 318-320. doi:10.1126/ 295. doi:10.1016/0022-0965(91)90037-S
science.1093993 Soraci, S. A., Deckner, W., Haenlein, M.,
Raposo, J., Stone, J., & Hart, B. (1970). One of these Baumeister, A. A., Murata-Soraci, K., &
things [Recorded by Sesame Street]. On The Sesame Blanton, R. L. (1987). Oddity performance in pre-
Street book and record (Record). New York, NY: school children at risk for mental retardation: Trans-
Columbia Records. fer and maintenance. Research in Developmental
Rehfeldt, R. A. (2011). Toward a technology of derived Disabilities, 8, 137-151. doi:10.1016/0891-4222(87)
stimulus relations: An analysis of articles published in 90044-8
the Journal of Applied Behavior Analysis, 1992-2009. Sternberg, R. J. (1977). Intelligence, information processing,
Journal of Applied Behavior Analysis, 44(1), 109-119. and analogical reasoning: The componential analysis of
doi:10.1901/jaba.2011.44-109 human abilities. Oxford, England: Lawrence
Robinson, J. S. (1955). The sameness–difference discrimi- Erlbaum.
nation problem in chimpanzee. Journal of Compara- Stewart, I., Barnes-Holmes, D., & Roche, B. (2004). A
tive and Physiological Psychology, 48(3), 195-197. functional analytic model of analogy using the rela-
doi:10.1037/h0042463 tional evaluation procedure. The Psychological Record,
Rosales-Ruiz, J., & Baer, D. M. (1997). Behavioral cusps: 54, 531-552. Retrieved from http://www.springer.
A developmental and pragmatic concept for behavior com/psychology/psychology+general/journal/40732
analysis. Journal of Applied Behavior Analysis, 30, Stewart, I., Barnes-Holmes, D., Roche, B., & Smeets, P.
533-544. doi:10.1901/jaba.1997.30-533 M. (2001). Generating derived relational networks
Serna, R. W., Dube, W. V., & McIlvane, W. J. (1997). via the abstraction of common physical properties: A
Assessing same/different judgments in individuals possible model of analogical reasoning. The Psycholog-
with severe intellectual disabilities: A status report. ical Record, 51, 381-408. http://www.springer.com/
Research in Developmental Disabilities, 18, 343-368. psychology/psychology+general/journal/40732
doi:10.1016/s0891-4222(97)00015-2 Stewart, I., Barnes-Holmes, D., Roche, B., & Smeets, P.
Shillingsburg, M. A., Kelley, M. E., Roane, H. S., M. (2002). A functional analytic model of analogy: A
Kisamore, A., & Brown, M. R. (2009). Evaluation relational frame analysis. Journal of the Experimental
and training of yes-no responding across verbal oper- Analysis of Behavior, 78, 375-396. doi:10.1901/
ants. Journal of Applied Behavior Analysis, 42, 209- jeab.2002.78-375
223. doi:10.1901/jaba.2009.42-209 Stewart, I., Barnes-Holmes, D. & Weil, T. (2009). Train-
Sidman, M. (1971). Reading and auditory-visual equiva- ing analogical reasoning as relational responding. In
lences. Journal of Speech, Language, and Hearing R.A. Rehfeldt, & Y. Barnes-Holmes (Eds.), Derived
Research, 14, 5-13. doi:10.1044/jshr.1401.05 relational responding: Applications for learners with
Sidman, M. (2009). Equivalence relations and behavior: autism and other developmental disabilities. Oakland,
An introductory tutorial. The Analysis of Verbal CA: New Harbinger copublished with Context Press.
Behavior, 25, 5-17. Retrieved from http://www.ncbi. Stewart, I., & McElwee, J. (2009). Relational responding
nlm.nih.gov/pmc/journals/609/ and conditional discrimination procedures: An appar-
Skinner, B. F. (1957). Verbal behavior. New York: Apple- ent inconsistency and clarification. The Behavior
ton Century Crofts. Analyst, 32(2), 309-317. Retrieved from http://www.
Slattery, B., & Stewart, I. (2014). Hierarchical classifica- ncbi.nlm.nih.gov/pmc/articles/PMC2778812/
tion as relational framing. Journal of the Experimental Stewart, I., McElwee, J., & Ming, M. (2013). Language
Analysis of Behavior, 101, 61-75. doi:10.1002/jeab.63 generativity, response generalization and derived rela-
Smirnova, A. A., Lazareva, O. F., & Zorina, Z. A. tional responding. The Analysis of Verbal Behavior,
(2000). Use of number by crows: Investigation by 29, 137-155. Retrieved from http://www.ncbi.nlm.
matching and oddity learning. Journal of the nih.gov/pmc/journals/609/
A REVIEW OF DIFFERENCE RELATIONS 27
Stewart, I., & Roche, B. (2013). Relational frame theory: Truppa, V., Piano Mortari, E., Garofoli, D.,
An overview. In D. Dymond & B. Roche (Eds.), Privitera, S., & Visalberghi, E. (2011). Same/different
Advances in Relational Frame Theory: research and concept learning by capuchin monkeys in matching-
application. Oakland, CA: Context Press. to-sample tasks. PLoS ONE, 6(8), e23809.
Stromer, R., & Osborne, G. (1982). Control of adolescents’ doi:10.1371/journal.pone.0023809
arbitrary matching-to-sample by positive and negative Tyrrell, D. J., Stauffer, L. B., & Snowman, L. G. (1991).
stimulus relations. Journal of the Experimental Analysis of Perception of abstract identity/difference relationships
Behavior, 37, 329-348. doi:10.1901/jeab.1982.37-329 by infants. Infant Behavior and Development, 14,
Suchman, R. G., & Trabasso, T. (1966). Color and form 125-129. doi:10.1016/0163-6383(91)90059-2
preference in young children. Journal of Experimental
Child Psychology, 3, 177-187. doi:10.1016/0022- Vitale, A., Barnes-Holmes, Y., Barnes-Holmes, D., &
0965(66)90092-0 Campbell, C. (2008). Facilitating responding in
Sundberg, M. L. (2004, May). What the autoclitic is and accordance with the relational frame of comparison:
what the autoclitic is not. Presented at the Annual Systematic empirical analyses. The Psychological Rec-
Conference of the Association for Behavior Analysis: ord, 58(3), 365-390. Retrieved from http://opensiuc.
International. Boston, MA. lib.siu.edu/tpr/vol58/iss3/3/
Sundberg, M. L. (2008). VB-MAPP: Verbal behavior mile- Vonk, J. (2003). Gorilla (Gorilla gorilla gorilla) and orang-
stones assessment and placement program—protocol. utan (Pongo abelii) understanding of first- and
Concord, CA: AVB Press. second-order relations. Animal Cognition, 6, 77-86.
Sundberg, M. L., & Partington, J. W. (2010). Teaching doi:10.1007/s10071-003-0159-x
language to children with autism or other developmental Wilkinson, K. M., Dube, W. V., & McIlvane, W. J.
disabilities. Walnut Creek, CA: AVB Press. (1998). Fast mapping and exclusion (emergent
Thomas, R. K., & Kerr, R. S. (1976). Conceptual condi- matching) in developmental language, behavior analy-
tional discrimination in Saimiri sciureus. Animal sis, and animal cognition research. The Psychological
Learning & Behavior, 4, 333-336. doi:10.3758/ Record, 48, 407-422. Retrieved from http://www.
BF03214060 springer.com/psychology/psychology+general/journal/
Thompson, R. K. R., & Oden, D. L. (2000). Categorical 40732
perception and conceptual judgments by nonhuman Zentall, T. R., Edwards, C. A., Moore, B. S., &
primates: The paleological monkey and the analogical Hogan, D. E. (1981). Identity: The basis for both
ape. Cognitive Science, 24, 363-396. doi:10.1207/ matching and oddity learning in pigeons. Journal of
s15516709cog2403_2 Experimental Psychology: Animal Behavior Processes, 7,
Thompson, R. K., & Oden, D. L. (1995). A profound 70-86. doi:10.1037/0097-7403.7.1.70
disparity revisited: Perception and judgment of
abstract identity relations by chimpanzees, human Zentall, T. R., Hogan, D. E., Edwards, C. A., &
infants, and monkeys. Behavioural Processes, 35, 149- Hearst, E. (1980). Oddity learning in the pigeon as a
161. doi:10.1016/0376-6357(95)00048-8 function of the number of incorrect alternatives. Jour-
Thompson, R. K., Oden, D. L., & Boysen, S. T. (1997). nal of Experimental Psychology: Animal Behavior Pro-
Language-naive chimpanzees (Pan troglodytes) judge cesses, 6, 278-299. doi:10.1037/0097-7403.6.3.278
relations between relations in a conceptual matching-
to-sample task. Journal of Experimental Psychology: Received July 17, 2014
Animal Behavior Processes, 23(1), 31-43. doi:10.1037/ Final acceptance June 9, 2016
0097-7403.23.1.31 Action Editor, Caio Miguel