JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2015, 104, 223–240 NUMBER 3 (NOVEMBER)

INTRAVERBAL NAMING AND EQUIVALENCE CLASS FORMATION IN CHILDREN

CHARLOTTE L. CARP AND ANNA INGEBORG PETURSDOTTIR
TEXAS CHRISTIAN UNIVERSITY

Six typically developing children between 5 and 7 years of age underwent match-to-sample training to
establish three-member equivalence classes after first acquiring a unique name for each stimulus. Horne
and Lowe’s (1996) naming hypothesis predicts that under those circumstances, match-to-sample training
contingencies may establish intraverbal relations between the unique names, which in turn guide correct
responses on a subsequent test for stimulus equivalence. Following training of baseline relations (AB and
AC), participants received an equivalence test followed by an intraverbal test. Performance on the two tests
co-varied, such that three participants passed both tests, and three participants failed repeated
administrations of both tests, including a modified version of the equivalence test designed to promote
intraverbal responding. The participants who failed the equivalence test, however, did so primarily due to
poor performance in transitivity trials, but performed accurately in symmetry trials. After training of a
third relation (BC), all three participants performed accurately in a symmetry test for the remaining
untrained relations (BA, CA, and CB); two of them in the absence of relevant intraverbal repertoires.
Key words: stimulus equivalence, naming, intraverbal, matching to sample, children

An equivalence class, as defined by Sidman formation was mediated by verbal naming of

and Tailby (1982), refers to a class of stimuli stimuli in the class (e.g., Dugdale & Lowe, 1990),
related by the properties of reflexivity, symme- which ran contrary to the predominant viewpoint
try, and transitivity. Stimuli within an equiva- among equivalence researchers (e.g., Sidman &
lence class are treated as interchangeable for Tailby, 1982; Sidman, Willson-Morris, & Kirk,
one another within a particular context, in the 1986). Since then, equivalence class formation
absence of any physical similarity among them has been demonstrated with minimally verbal
and in the absence of a reinforcement history humans (Carr, Wilkinson, Blackman, & McIlvane,
for treating them as such. In humans, stimulus 2000; Luciano, G
omez Becerra, & Rodriguez
equivalence is often demonstrated using arbi- Valverde, 2007) and evidence is mounting that
trary match-to-sample procedures. The training some of its defining features may be observed in
of two or more overlapping sample-comparison other species (for reviews, see Lionello-DeNolf,
relations (e.g., matching comparison B to 2009; Zentall, Wasserman, & Urcuioli, 2014).
sample A and comparison C to sample B) is Thus, the notion that human language is
said to result in equivalence class formation if necessary for equivalence class formation may
the participant’s subsequent test performance no longer be tenable.
is consistent with transitivity (matching C to A On the other hand, naming does not appear
and A to C) and symmetry (matching A to B to be irrelevant to equivalence class formation.
and B to C). Children who initially fail tests of stimulus
Early research on equivalence classes engen- equivalence may pass them following training
dered the hypothesis that equivalence class in possible verbal mediation strategies (e.g.,
Eikeseth and Smith, 1992; Lowe & Beasty,
Charlotte L. Carp, Department of Psychology, Texas
1987). This finding, along with various others
Christian University; Anna Ingeborg Petursdottir, Depart- (e.g., Holth & Arntzen, 1998; Mandell &
ment of Psychology, Texas Christian University Sheen, 1994; Randell & Remington, 1999,
Charlotte L. Carp is now at McNeese State University. 2006) suggest that although perhaps not a
This manuscript is based on a dissertation submitted necessary condition for generating equivalence
by the first author, under the direction of the second
author, to Texas Christian University in partial fulfillment classes, overt or covert naming may facilitate
of the Ph.D. degree in Psychology. We thank Lindsey their formation. The inaccessibility of subvocal
Carnes, Tayla Cox, and Tori VerPloeg for assisting with verbal behavior complicates empirical investi-
data collection. gation into the nature of this functional role.
Address correspondence to Anna Ingeborg Petursdottir,
Department of Psychology, TCU Box 298920, Fort Worth,
However, to the extent that the relevant verbal
Texas 76129. E-mail: a.petursdottir@tcu.edu. behavior can be established at, or later brought
doi: 10.1002/jeab.183 to the overt level, such investigation might serve

223
224 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

to enhance prediction and control of equiva- the squiggly line”). Thus, in a transitivity trial,
lence class formation and related phenomena performance is mediated by a chain of intra-
in humans. verbal responses that may occur either in the
Horne and Lowe (1996) proposed two order established by the training contingencies,
mechanisms by which naming may facilitate or in the reverse order (“the squiggly line goes
performance on tests of stimulus equivalence: with the funny triangle, and the funny triangle
common naming and intraverbal naming. A crucial goes with the Roman nose” or “the Roman nose
assumption is that when humans undergo goes with the funny triangle, and the funny
relational training (e.g., baseline training in a triangle goes with the squiggly line”).
stimulus equivalence study), they tend to name Horne and Lowe’s (1996) naming hypothesis
the stimuli to be related, even if the stimuli are has yielded a substantial body of literature
previously unfamiliar and no contingencies showing that the application of a common
are placed on naming. In the common naming name to several stimuli results in visual
scenario, the participant applies the same stimulus class formation in the absence of
name to all stimuli belonging to the same class. any direct training to relate visual stimuli to
The source of the name may be, for example, one another (Horne, Hughes, & Lowe, 2006;
generalization from a familiar stimulus that Horne, Lowe, & Randle, 2004; Kobari-Wright
shares a perceptual feature with a nodal & Miguel, 2014; Lowe, Horne, Harris, &
stimulus in the class. In intraverbal naming, Randle, 2002; Lowe, Horne, & Hughes, 2005;
the participant generates a unique name for Miguel et al., 2015; Miguel & Kobari-Wright,
each stimulus. When this happens (e.g., the 2013; Miguel, Petursdottir, Carr, & Michael,
participant generates the names “squiggly 2008; Sprinkle & Miguel, 2012). Similarly, the
line,” “funny triangle,” and “Roman nose” for establishment of intraverbal relations between
stimuli A1, B1, and C1), the unique names the unique names of visual stimuli has been
associated with different stimuli may become found to produce novel conditional discrim-
linked intraverbally in accordance with the inations among the stimuli in the absence of
training contingencies (e.g., the participant other training (Petursdottir, Carp, Peterson,
may emit the responses “the squiggly line goes & Lepper, 2015; Santos, Ma, & Miguel, 2015).
with the funny triangle” and “the funny Together with other findings on the effects of
triangle goes with the Roman nose” as he or experimenter-supplied names or intraverbal
she performs the match-to-sample task, and relations (Eikeseth & Smith, 1992; Lowe &
these intraverbal relations are reinforced along Beasty, 1987), these findings suggest that
with correct responses on the primary relational common and intraverbal naming are, indeed,
task). Horne and Lowe, then, hypothesized capable of generating classes of stimuli that are
that successful equivalence test performance treated as equivalent. The data do not imply
depended on the participant’s emission of with certainty, however, that these naming
either common names (in the first scenario) processes operate spontaneously in the context
or intraverbally linked unique names (in the of the usual training and testing procedures in
second scenario) during testing. In the case of human stimulus equivalence research. In fact,
common names, the participant may begin some details of the data have been inconsistent
each trial by naming the sample stimulus, and with specific predictions of the naming hy-
respond to that name as a listener by selecting pothesis. For example, Petursdottir et al.
the comparison stimulus given the same (2015) found that for preschool-age children,
name. In intraverbal naming, initial naming intraverbal training produced novel visual
of the sample evokes one or more intraverbal conditional discriminations without producing
responses that guide the selection of a correct bidirectional intraverbal relations.
comparison. An important assumption is that Postexperimental naming tests and recordings
when the training contingencies establish a of spontaneous vocalizations during equivalence
particular intraverbal relation between two training and testing have often failed to reveal
unique names (e.g., “the squiggly line goes evidence that participants have assigned either
with the funny triangle”), the relation is common or unique names to the experimental
usually bidirectional (Horne & Lowe, 1996, stimuli (e.g., Green, 1990; Sidman & Tailby, 1982;
pp. 218–221), such that the order of names can Sidman et al., 1986; Smeets & Barnes-Holmes,
be reversed (e.g., “the funny triangle goes with 2005). These findings might be interpreted to
 INTRAVERBAL NAMING 225

suggest that although it is possible to establish differences should be expected between partic-
equivalence relations through naming, it is not ipants who demonstrate acquisition of the
how they are necessarily or usually established in relevant intraverbal relations following match-
research using typical match-to-sample proce- to-sample training and equivalence testing,
dures. However, postexperimental naming tests and participants who have not acquired intra-
have been criticized for potentially failing to verbal relations between the unique names.
capture the names that participants actually At one extreme, if verbal behavior is necessary
use during the experiment (Dugdale & Lowe, for successful equivalence test performance,
1990), and recordings of spontaneous vocal- all participants who pass the equivalence test
izations suffer from the obvious limitation that should also pass the intraverbal test, unless there
the relevant verbal behavior could be occurring is evidence that they are using other verbal
covertly. mediation strategies. Thus, we asked if partic-
In the present study, we employed a novel ipants who passed an equivalence test would in
approach to capturing potential verbal media- all cases show emergent intraverbal relations
tion repertoires in postexperimental tests fol- between the names they had been taught
lowing stimulus equivalence training and for the stimuli in each equivalence class. We
testing, focusing on the intraverbal naming elected to study children between the ages of
mechanism described by Horne and Lowe 5 and 7 years because children in this age range
(1996). The participants were children who may or may not pass tests for intraverbal
underwent match-to-sample training to establish relations that have not been instructed directly
three-member equivalence classes consisting of (e.g., Carp & Petursdottir, 2012; P
erez-Gonz
alez,
visual stimuli. Before training began, we sup- Herszlikowics, & Williams, 2008) and their
plied the participants with a unique name for emergent intraverbal relations are not necessar-
each stimulus through vocal tact (Skinner, 1957) ily bidirectional (e.g., Petursdottir, Olafsdottir,
training, but unlike Lowe and Beasty (1987) & Aradottir, 2008), but they are often reported
and Petursdottir et al. (2015), we did not to be successful on tests for three-member
teach any intraverbal relations between the equivalence classes or larger (e.g., Lazar, Davis-
unique names. Instead, we tested intraverbal Lang, & Sanchez, 1984; Sidman et al., 1986).
relations following equivalence testing to exam-
ine whether they had emerged without explicit Method
instruction. We reasoned that if match-to-sample
training contingencies can produce either Participants
common naming or intraverbal naming (Horne Six children between 5 and 7 years of age
& Lowe, 1996), supplying a unique name for (see Table 1) completed the study with their
each stimulus might predispose the participants parents’ permission, following approval by the
to engage in intraverbal naming involving Texas Christian University Institutional Review
those particular unique names, as opposed to Board. The participants were recruited from
common naming or intraverbal naming involv- day care or after-school care programs at two
ing participant-generated names. If this behav- church-based child care centers. All partici-
ior occurs and plays a functional role in pants spoke English as their first language.
equivalence test performance, performance Sessions (20–30 min each) were conducted

Table 1
Participant Information

Participant Sex Chronological Age PPVT-4a Age Equivalent Score

Niels M 7 years, 0 months 6 years, 3 months

Rita F 6 years, 9 months 9 years, 2 months
Emilie F 6 years, 4 months 6 years, 9 months
Charles M 5 years, 6 months 6 years, 1 month
Marie F 5 years, 5 months 6 years, 5 months
Leonardo M 6 years, 3 months 7 years, 4 months
a
Peabody Picture Vocabulary Test, Fourth Edition
226 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

at each participant’s child care center up to selected a comparison stimulus with a mouse
four times per week in a room that was click.
otherwise not in use during session time, with During other training and testing procedures
the experimenter and the participant seated (category pretraining, tact training, intraverbal
side by side at a table. The total number of testing, and postexperimental tests), auditory
sessions per participant ranged from 18 (Rita) stimuli, when used, were presented vocally by
to 49 (Marie). During most training and testing the experimenter and visual stimuli, when used,
conditions, participants could earn tokens that were presented in tabletop format. The visual
were placed on a token board. When the token stimuli were printed on white 20.3 cm by
board was full, the participant received 5 min of 27.9 cm sheets of paper encased in transparent
access to a box that contained a variety of toys sheet protectors and contained in a three-ring
and games. binder. Stimulus sheets were presented to the
participant by turning pages in the binder,
which was placed on the table in front of the
Stimuli and Apparatus participant in landscape orientation. Depend-
Visual stimuli consisted of three colored ing on what was being trained or tested, each
outline maps of states (A1 [Virginia], A2 stimulus sheet contained either one stimulus
[Alabama], and A3 [California]), three color presented in the center of the sheet or all nine
photographs of the corresponding state birds visual stimuli printed in three horizontal rows
(B1 [cardinal], B2 [yellowhammer], and B3 centered on the sheet.
[quail]), and three color photographs of the
corresponding state flowers (C1 [dogwood],
C2 [camellia], and C3 [poppy]). Each stimulus Procedure
was contained within a black border measuring Overview. The sequence of events in the
5.5 cm by 5.7 cm. We use the alphanumerical experiment is summarized in Table 2. Partic-
designations A1’ through C3’ to refer to the ipants were first exposed to category-name
names associated with the visual stimuli when pretraining to ensure control over selections of
dictated by the experimenter or spoken by the visual stimuli by the spoken category names
participant. “state”, “bird”, and “flower”, which would later
Six additional stimuli that were previously be presented in intraverbal tests. Next, partic-
familiar to the participants (i.e., pictures of a ipants received tact training, in which we
bowl, pants, a shirt, shoes, a spoon, and socks) established discriminative control over vocal
and their corresponding conventional names responses A10 through C3’ by visual stimuli A1
were used to familiarize participants with the through C3. Tact maintenance trials (not
conditional discrimination training procedures. shown in Table 2) were included in subsequent
During conditional discrimination training training and testing phases. Following tact
and equivalence testing (or components training, we conducted intraverbal pretests that
thereof), stimulus presentation was controlled probed relations between spoken auditory
by the software application Mestre Libras1 stimuli and vocal responses (A’B’, A’C’, B’A’,
(Goyos, Elias, & Ribeiro, 2005). The application B’C’, C’A’, and C’B’ relations) in the presence
ran on a laptop computer equipped with a of instructions to name a state, a bird, or a
Windows XP operating system and a 29.0 cm by flower (Table 3). Next, participants received
19.8 cm monitor. Visual stimuli were presented AB/AC conditional discrimination training. Once
on the screen against a blue background. Each the training criterion was met, the participant
trial began with the presentation of a visual entered the first test phase. First, we conducted
sample stimulus in the horizontal center of the an equivalence test that included baseline (AB
top third of the screen. Once the participant and AC), symmetry (BA and CA), and transitiv-
clicked on the sample stimulus using the ity (BC and CB) trials. If the participant
computer mouse, three comparison stimuli performed to criterion in at least the baseline
were presented equidistant from one another trials (indicating no need for additional AB and
across the lowest third of the screen, with the AC training), the equivalence test was
middle stimulus centered below the sample followed by an intraverbal posttest identical to
stimulus. The sample and comparison stimuli the pretest. If a participant’s performance did
remained on the screen until the participant not meet a predetermined criterion in either
 INTRAVERBAL NAMING 227

Table 2
Sequence of Events in the Experiment

Procedural Steps Exit Criterion Participants Included

1. Category-name pretraining 100% accuracy in three consecutive All

3-trial blocks
2. Tact training (AA’, BB’, CC’) 100% accuracy in three consecutive All
9-trial blocks on VR 3 schedule
3. Intraverbal pretest (A’B’, A’C’, B’A’, Completion of one to three 18-trial All
B’C’, C’A’, C’B’) w/ tact maintenance blocks (variable across participants to
trialsa rule out testing effects)
4. AB/AC conditional discrimination 100% accuracy in three consecutive All
a
training and tact maintenance 9-trial blocks on VR 3 schedule
5. Equivalence test (AB, AC, BA, CA, BC, Passing performance on one test pair All
CB) þ intraverbal post-test (A’B’, A’C’, (18-trial equivalence testb and 18-trial
B’A’, B’C’, C’A’, C’B’) w/ tact intraverbal test), or three consecutive
maintenance trialsa test pairs with no upward trend
6. Equivalence test (AB, AC, BA, CA, BC, Same as Step 5 Participants who did not pass
CB) with sample tact requirement þ equivalence test in Step 5
intraverbal post-test w/ tact (Emilie, Charles, Marie,
a
maintenance trials Leonardo)
7. BC conditional discrimination training 100% accuracy in three consecutive Participants who did not pass
and tact maintenancea 9-trial blocks on VR 3 schedule equivalence test in Step 6
(Charles, Marie, Leonardo)
8. Symmetry test (AB, AC, BA, CA, BC, Passing performance on one test pair Participants who underwent
CB) þ intraverbal post-test w/ tact (18-trial symmetry test and 18-trial Step 7 (Charles, Marie,
maintenance trialsa intraverbal test), or three consecutive Leonardo)
test pairs with no upward trend
9. Postexperimental listener test Completion of one 9-trial block All
10. Additional postexperimental tests (tact- Completion of 18 tact-intraverbal, Participants who never passed
intraverbal, alternate-tact test, and 9 alternate-tact, and 18 category- the intraverbal test (Charles
category-name intraverbal tests) name trials and Marie)
a
Response decrement in tact maintenance trials resulted in return to and re-completion of Step 2 before continuing the
current step.
b
Below-criterion responding in baseline trials (AB and AC) resulted in readministration of the test without an intervening
intraverbal test.

the equivalence or the intraverbal test, the tests relations (AB, AC, and BC) and emergent
were repeated until the participant either met symmetry (BA, CA, and CB).
criterion on both tests or failed three tests with At the conclusion of testing, we conducted
no upward trend in performance. several postexperimental tests. First, all partic-
If a participant did not pass the equivalence ipants were given a listener test to verify that tact
test during the first testing phase, we added a training had also produced appropriate listener
modified testing phase in which the participant behavior, or in Horne and Lowe’s (1996) terms,
was required to tact the sample stimulus in each produced name relations. Second, participants
equivalence test trial (i.e., given a possible who had not passed the intraverbal test were
prompt to engage in intraverbal mediation). As given a series of tests designed to probe the
before, each equivalence test in which criterion possibility that relevant intraverbal relations
was met for the baseline relations was followed were in their repertoires, but did not occur
by an intraverbal test. If a participant did not during intraverbal tests.
pass the equivalence test in spite of repeated Category-name pretraining. In each trial, the
and modified testing, the participant received experimenter presented a stimulus sheet con-
BC conditional discrimination training, fol- taining one A stimulus, one B stimulus, and one
lowed by additional repetitions of the equiva- C stimulus, and asked the participant to point to
lence and intraverbal tests. We refer to the the state, the bird, or the flower. Correct
equivalence test as a symmetry test in this responses were followed by praise and the
phase, as it now addressed only trained baseline delivery of a token. If an incorrect response
228 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

Table 3
Antecedent Stimuli and Correct Responses in Intraverbal Test Trials

Trial Type Verbal Stimulus Correct Response

Baseline-Like
A1’B1’ Virginia goes with which bird? Cardinal
A2’B2’ Alabama goes with which bird? Yellowhammer
A3’B3’ California goes with which bird? Quail
A1’C1’ Virginia goes with which flower? Dogwood
A2’C2’ Alabama goes with which flower? Camellia
A3’C3’ California goes with which flower? Poppy
Symmetry-Like
B1’A1’ Cardinal goes with which state? Virginia
B2’A2’ Yellowhammer goes with which state? Alabama
B3’A3’ Quail goes with which state? California
C1’A1’ Dogwood goes with which state? Virginia
C2’A2’ Camellia goes with which state? Alabama
C3’A3’ Poppy goes with which state? California
Transitivity-Like
B1’C1’ Cardinal goes with which flower? Dogwood
B2’C2’ Yellowhammer goes with which flower? Camellia
B3’C3’ Quail goes with which flower? Poppy
C1’B1’ Dogwood goes with which bird? Cardinal
C2’B2’ Camellia goes with which bird? Yellowhammer
C3’B3’ Poppy goes with which bird? Quail

occurred, the experimenter pointed to the (i.e., repetition of the trial) in which correct
correct comparison and said, for example, responses produced praise only. Training
“This is a state,” and proceeded to the in Phase 1 continued until the participant
next trial. Training continued until the partici- responded with 100% accuracy in a single six-
pant responded with 100% accuracy in three trial block. Phases 2 and 3 were identical to
consecutive three-trial blocks (no participant Phase 1 except that Phase 2 included B and
required more than four trial blocks to C stimuli, and Phase 3 included A and C stimuli.
complete training). In Phase 4, procedures remained similar (with
Tact training (AA’, BB’ and CC’). Tact the prompt delay at 5 s from the beginning), but
training was completed in five phases. In all nine stimuli (A, B, and C) were included, and
Phase 1, trials were presented in blocks of six training continued until 100% accuracy was
that each contained one presentation of each of achieved in three consecutive nine-trial blocks.
the A and B stimuli, with presentation order Finally, Phase 5 was identical to Phase 4 except
varied across blocks. In each trial, the experi- that the density of praise and token delivery
menter presented a stimulus sheet containing a for correct responses was reduced to a variable
single visual stimulus and asked “What is this?” ratio (VR) 3 schedule, and incorrect responses
In the first two trial blocks, the experimenter were followed only by a prompt, not by an error
provided an immediate vocal prompt and then correction trial. The mastery criterion was
repeated the trial without a prompt, delivering identical to Phase 4. In Phases 4 and 5, if
praise for correct responses. In the next two accuracy dropped below 67% correct (two of
trial blocks, participants were allowed 2 s to three trials) for a particular type of stimulus
respond independently before a prompt was (A, B, or C) in two consecutive trial blocks,
provided, and from the fifth trial block on, the the participant returned to a previous phase. A
delay to the prompt remained constant at 5 s. A participant in Phase 4 returned to Phase 1, and
correct response before the end of the prompt a participant in Phase 5 returned to Phase 4.
delay resulted in the delivery of praise and a The attainment of the Phase 5 mastery criterion
token, whereas an incorrect response (includ- marked the completion of tact training.
ing no response before the prompt delay was Following tact training, reinforced tact main-
up) produced the vocal prompt, followed by the tenance trials were (a) interspersed between test
implementation of an error correction trial trials during intraverbal tests, and (b) presented
 INTRAVERBAL NAMING 229

in a single trial block prior to the beginning of sample (e.g., pants, shoes, and a bowl).
each AB/AC conditional discrimination train- Procedures were similar to AB/AC training,
ing session. During AB/AC training, if a partici- and the criterion for completion was 100%
pant did not respond with 100% accuracy in correct responding in a single six-trial block, in
this single trial block, the participant returned which each of the six familiar stimuli served as
to Phase 5 of tact training and did not return to the sample once.
AB/AC training until tact training was com- Next, AB/AC conditional discrimination
pleted again. training was conducted in four phases. Phase 1
Intraverbal test (A’B’, A’C’, B’A’, B’C’, targeted the three AB relations. Each trial
C’A’, C’B’). Intraverbal tests were conducted began with the presentation of a sample (A)
prior to AB/AC conditional discrimination stimulus, and the three B stimuli appeared as
training and following equivalence tests after comparisons as soon as the participant clicked
the completion of AB/AC training. Each test on the sample. The relative location (left,
contained 18 intraverbal trials presented in a center, right) of positive and negative compar-
randomized order that probed each of the isons was counterbalanced across trials. Selec-
18 relations shown in Table 3. Randomly tion of a positive comparison produced a 2-s
interspersed between intraverbal trials were animation on the computer screen that showed
nine reinforced tact maintenance trials (one a coin falling into a piggy bank; in addition, the
per each of nine visual stimuli). experimenter provided praise and delivered a
An intraverbal trial began with the experi- token to the participant. An intertrial interval
menter’s vocal presentation of the instruction (ITI) of 2 s then preceded the next trial. During
“[name] goes with which [state/bird/flower]?” the ITI, the screen was blue. Selection of a
Up to 10 s were allowed for the participant’s negative comparison produced a black screen
vocal response. To minimize inadvertent cue- for 2 s, followed by the 2-s ITI. Phase 1 training
ing, the experimenter sat next to and slightly continued until the participant responded with
behind the participant while delivering the 100% accuracy in three consecutive three-trial
instruction and waiting for a response. Follow- blocks. Phase 2 was identical to Phase 1, except
ing a correct or an incorrect response, the that AC relations were taught instead of AB
experimenter proceeded to the next trial relations. Phase 3 contained mixed AB and AC
without presenting any other consequences. trials, and continued until the participant
The criterion for the emergence of an intra- responded with 100% accuracy in three conse-
verbal repertoire was correct responses in 16 cutive six-trial blocks. Finally, Phase 4 was
of 18 intraverbal trials, or 89% accuracy. identical to Phase 3 except that consequences
None of the tested intraverbals were directly for correct responses were delivered on a VR 3
trained in the study; however, the six A’B’ and schedule. When a participant responded cor-
A’C’ intraverbals corresponded to the baseline rectly in a trial that was not scheduled for
relations established with visual stimuli in reinforcement, the correct response produced
AB/AC training, and will hereafter be referred the 2-s ITI. In Phase 3, if performance fell below
to as baseline-like intraverbals. B’A’ and C’A’ 83% accuracy in three consecutive trial blocks,
relations will similarly be referred to as the participant returned to Phase 1. In Phase 4,
symmetry-like, and the B’C’ and C’B’ relations if performance fell under 100% accuracy in
as transitivity-like. three consecutive trial blocks, the participant
AB/AC conditional discrimination training. returned to Phase 3.
In this phase, the laptop computer that Equivalence test. An equivalence test con-
controlled stimulus presentation was placed sisted of 18 trials that included six trials targeting
on the table directly in front of the participant. the trained baseline relations (AB and AC), six
Before AB/AC conditional discrimination symmetry trials (BA and CA), and six transitivity
training began, participants received brief trials (BC and CB). Within each test, each of the
pretraining that served to familiarize them 18 possible trial types (e.g., A1B1) was presented
with the computer program. In each pretrain- once, and the order of presentation was
ing trial, a familiar visual stimulus (e.g., a randomized. Trials were conducted in a similar
spoon) was presented as the sample, and three manner as in AB/AC training. In all trial
familiar stimuli were presented as comparisons, types (i.e., baseline, symmetry, and transitivity),
one of which was thematically related to the all responses produced the ITI, regardless of
230 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

whether they were correct or incorrect. The as the participant displayed passing perfor-
passing criterion for the equivalence test was mance in baseline trials (this was always
correct responding in at least five of six baseline the case).
trials, five of six symmetry trials, and five of Postexperimental tests. Following the com-
six transitivity trials. Each administration of the pletion of all equivalence and intraverbal
equivalence test was typically followed immedi- testing, all participants were exposed to a
ately by an intraverbal test. However, if a listener test. Further, participants who never
participant responded correctly in fewer than passed the intraverbal test were exposed to a
five baseline trials on the equivalence test tact-intraverbal test, an alternate tact test, and a
(suggesting a possible need for more AB/AC category-name intraverbal test.
training), the test was repeated without an Listener test (A’A, B’B, and C’C). The listener
intervening intraverbal test. If performance in test consisted of nine trials, one per each of the
baseline trials had not improved to criterion, the nine visual stimuli (A1 through C3). At the
participant would have returned to Phase 4 of beginning of each trial, the experimenter
AB/AC training. presented a stimulus sheet containing all nine
Equivalence test with sample tact require- stimuli and delivered the instruction “Point
ment. Participants who did not pass repeated to ____” (e.g., “Point to Virginia”). When the
equivalence tests were exposed to a modified participant pointed to a stimulus or 5 s were up,
version of the test that required the participant the trial ended and the next trial was presented;
to tact the sample stimulus in each trial. In no other consequences followed correct or
each trial, as the sample stimulus was presented incorrect responses. The passing criterion was
on the computer screen, the experimenter correct responding in at least eight trials, or
asked “What is this?” If the participant emitted 89% accuracy.
an incorrect tact or did not respond within 5 s, Tact-intraverbal test. This test examined the
the experimenter vocally prompted a correct possibility that relevant intraverbal relations
response. Following the participant’s correct or were in the participants’ repertoires, but did
prompted response, the participant could click not occur on the intraverbal test because they
the sample stimulus to produce the comparison were partially under the control of the visual
stimuli; no other consequences were provided. sample stimuli presented in AB/AC training
All other procedures remained identical to the and equivalence test trials. The test consisted
original equivalence test. of nine trials, one for each visual stimulus (A1
BC conditional discrimination training and through C3). Each trial began with the
symmetry test. Participants who repeatedly experimenter’s presentation of a visual stimu-
failed the original and the modified equiva- lus on a trial sheet, along with the question “If
lence test received BC training. Training was you see this, what do you pick?” Two correct
completed in two phases, both of which responses were possible in each trial; for
targeted all three BC relations. In the first example, if the visual stimulus was A1, the
phase, all correct responses were reinforced possible correct responses were B1’ and C1’.
(animation screen, praise, and token), whereas Following the first vocal response from the
in the second phase, correct responses were participant, the experimenter asked “What
reinforced on a VR 3 schedule. The criterion for else do you pick?” and waited 10 s for another
completion of each phase was 100% accuracy in response. Thus, the participant could make
three consecutive three-trial blocks. Procedures up to 18 correct responses, and 16 of 18
were otherwise identical to AB/AC training. BC (89% accuracy) was considered passing
training was followed by a symmetry test that was performance.
identical to the equivalence test (without the Alternate-tact test. This test examined the
sample tact requirement); however, because possibility that during conditional discrimina-
the participants had now completed AB, tion training, participant-generated names for
AC, and BC training, the test addressed only the stimuli were substituted for the names
baseline and symmetry (BA, CA, and CB) supplied to the participant in tact training,
relations. The passing criterion was correct resulting in either common name relations
responding in eight of nine baseline trials and or intraverbal relations that differed from
eight of nine symmetry trials. Each symmetry those tested in the intraverbal test. The test
test was followed by an intraverbal test, as long consisted of nine trials, one for each visual
 INTRAVERBAL NAMING 231

stimulus (A1 through C3). At the beginning of testing conditions, responses were recorded
each trial, the experimenter presented a visual manually on data sheets by the experimenter.
stimulus and asked “What is this?” Following a When the target response was vocal, the
vocal response from the child, the experi- experimenter scored the first state, bird, or
menter asked “Do you call it anything else?” flower name that the participant vocalized
and wrote down the child’s response. No as correct if it matched the experimentally
consequences were delivered for correct or defined target response. The response was
incorrect responses. scored as incorrect if the participant vocalized
Category-name intraverbal tests. This two-part any state, bird, or flower name that differed
test examined the possibility that poor perfor- from the target response, or if the participant
mance on the intraverbal test occurred due to did not vocalize the name of a state, bird, or a
deficient intraverbal repertoires with respect flower within 5 s during training or 10 s during
to the category names “state,” “bird,” and testing. An exception occurred for responses to
“flower” that were presented in intraverbal test the second question in each trial of the
trials. Although pretraining had documented alternate-tact test, in which the experimenter
that the participants could respond as listeners simply wrote down the participant’s response,
to these category names, it seems possible and no responses were considered correct or
that the stimuli nevertheless did not gain incorrect. When the target response consisted
intraverbal control over the vocal responses of selecting a visual stimulus on a stimulus sheet,
acquired as tacts in the experiment. The first the first stimulus that the participant touched
part of the test examined whether the was scored as correct or incorrect, and the trial
participants would produce the responses was scored as incorrect if no stimulus was
“state”, “bird” and “flower” vocally when selected within 10 s.
presented with individual state, bird, and A second observer independently recorded
flower names from the experiment. Nine trials data on vocal responding during at least 30% of
were presented, one for each name (A1’ each participant’s tact training sessions and at
through C3’). In each trial, the experimenter least 90% of each participant’s intraverbal test
asked “What is ____?” (e.g., “What is Vir- sessions, either live or from video. For each trial,
ginia?). A correct response consisted of “state” an agreement was scored if both the experi-
for A names, “bird” for B names, and “flower” menter and the second observer scored the trial
for C names. No consequences were provided as correct, or both scored it as incorrect.
for correct or incorrect responses. The second Interobserver agreement (IOA) was calculated
part of the test consisted of three trials that by dividing the number of agreements by the
addressed control over responses A1’ through number of trials and multiplying by 100. Mean
C3’ by “state,” “bird,” and “flower.” At the IOA across sessions for each participant ranged
beginning of each trial, the experimenter from 96% to 100% for tact training, and from
asked “Name a ___” (e.g., “Name a state.”) A 94% to 99% for intraverbal tests (details
response from the participant was followed by available from authors).
“Name another ___”, which was then repeated
a second time. Passing performance on each Results
test was considered eight of nine correct
responses, or 89% accuracy. Training Data
The participants completed tact training in a
total of 144 (Charles) to 300 (Marie) trials
Response Measurement and Interobserver (M ¼ 205 trials; the minimum number of trials
Agreement required to complete training was 72). Follow-
During conditional discrimination training ing intraverbal pretests, AB/AC conditional
and equivalence testing, the software applica- discrimination training was completed in 126
tion automatically recorded the participants’ (Leonardo) to 462 (Emilie) trials (M ¼ 259
correct and incorrect selection responses, as trials; the minimum number of trials required
well as data on response latency, defined as the to complete AB/AC training was 54). During
time in ms between a participant’s click on the AB/AC training, four participants (Rita,
sample stimulus and the participant’s click on Charles, Marie, and Leonardo) made brief
a comparison stimulus. In other training and returns to tact training due to failures to
232 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

maintain the tact relations; the number of Following tact training, Niels was exposed to
additional tact trials ranged from 36 (Charles) one intraverbal pretest, and Rita and Emilie
to 99 (Marie). to three. All participants showed a tendency to
emit A’ responses when asked about a state, B’
responses when asked about a bird, and C’
Equivalence and Intraverbal Tests: Niels, Rita, responses when asked about a flower. However,
and Emilie the accuracy of pretest responding was in all
The left three panels of Figure 1 show the cases around or below chance level and did
equivalence (white and gray bars) and intra- not improve with repeated administrations of
verbal test (open and filled circles) perfor- the test.
mance of the three participants who ultimately Following AB/AC training, all three partic-
passed the equivalence test: Niels, Rita, and ipants consistently showed passing perfor-
Emilie. White bars and open circles represent mance in baseline trials on the equivalence
failing performance, whereas gray bars and test, so it was never necessary to return to
filled circles represent passing performance. AB/AC training. As shown in Figure 1, Niels

Intrav. Post-AB/AC Intrav. Post-AB/AC Post-BC

Pretest Pretest
100 100 Sample tact

50 50

NIELS CHARLES
0 0
1 2 3 4 5 1 2 3 4 5 6 7 8 9 10 11 12 13

Intrav. Pretest Post-AB/AC Intrav. Post-AB/AC Post-BC

Pretest
Percent Correct

100 100 Sample tact

80
60
50
40
20
RITA MARIE
0 0
1 2 3 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Intrav. Post-AB/AC Intrav. Post-AB/AC Post-BC

Pretest Pretest Sample
Sample tact
100 100 tact

50 50

EMILIE LEONARDO
0 0
1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8

Test
Passed Equivalence Test Passed Symmetry Test Failed Equivalence or Symmetry Test
Passed Intraverbal Test Failed Intraverbal Test

Fig. 1. The figure shows all participants’ performances on equivalence and intraverbal tests. The left three panels show
the performance of participants who passed the equivalence test. The right three panels show the performance of
participants who never passed the equivalence test but passed the symmetry test following BC training. White bars denote
failed equivalence or symmetry tests, grey bars denote passing performance on equivalence tests, and diagonally shaded
bars denote passing performance on symmetry tests. Filled circles denote passing performance on intraverbal tests and
white circles denote failing performance.
 INTRAVERBAL NAMING 233

and Rita passed their first equivalence tests. Rita bars represent passing performance on equiva-
also passed the accompanying intraverbal test, lence tests and diagonally shaded bars represent
but Niels’s intraverbal performance did not passing performance on symmetry tests (i.e.,
meet passing criterion, although accuracy was equivalence tests conducted following BC train-
high compared to baseline. With repeated ing). White bars represent failing performance
administrations of the equivalence and intra- on an equivalence or symmetry test.
verbal tests, Niels’s intraverbal test performance Following tact training, Charles was exposed
improved and eventually met criterion. Emilie to one intraverbal pretest, and Marie and
initially failed repeated administrations of both Leonardo to two. Intraverbal pretest perfor-
the equivalence and the intraverbal tests. When mances did not exceed chance levels and did
the equivalence test was modified to include a not improve with repeated administrations of
sample tact requirement, Emilie passed two of the test. Following AB/AC training, Marie
three administrations of the test and was close consistently showed passing performance in
to meeting the criterion on the third. She also baseline trials on the equivalence test. Three
passed the intraverbal test that accompanied equivalence tests were repeated for Charles and
the second and the third administrations of the one equivalence test was repeated for Leonardo
modified equivalence test. because they did not show passing performance
Horne and Lowe’s (1996) analysis of intra- in baseline trials the first time. In all cases, they
verbal naming may not imply that all of the showed passing performance in baseline trials
intraverbal relations included in the intraverbal when the test was repeated, so a return to
test must necessarily be present in order for a AB/AC training was not implemented. These
participant to pass the equivalence test. Specifi- four tests with below-criterion performance in
cally, successful performance in transitivity baseline trials were not accompanied by intra-
trials could be mediated by chains of baseline- verbal tests and are therefore omitted from the
like and symmetry-like intraverbals (here, data shown in Figure 1.
A’B’, A’C’, B’A’, and C’A’ relations) that could As Figure 1 shows, Charles, Marie, and
conceivably fail to occur in corresponding Leonardo consistently failed equivalence tests
transitivity-like intraverbal trials. As a result, following AB/AC training, despite showing
we further analyzed responding in failed intra- passing performance in baseline trials, and
verbal tests that accompanied passed equiva- also failed the accompanying intraverbal tests.
lence tests; that is, Niels’s first three Thus, all participants received a modified
posttraining tests¸ and Emilie’s first modified equivalence test with a requirement to tact
test. In each of Niels’s first three tests, he the sample. The modified test was administered
responded correctly in at least five of six several times to Charles and Marie, but only
baseline-like and five of six symmetry-like trials. once to Leonardo due to time constraints (i.e.,
Thus, below-criterion intraverbal performance impending end of the school year). No
that accompanied highly accurate equivalence participant’s performance on either type of
test performance was exclusively due to lower test improved when the sample-tact require-
accuracy in transitivity-like intraverbal trials. In ment was introduced, with the exception that
Emilie’s first modified test, she responded Marie passed the second modified equivalence
correctly in four of six baseline-like and five test. Because she went on to fail four subsequent
of six symmetry-like trials. Thus, her perfor- tests, it is possible that her one-time passing
mance also appeared mostly related to lower performance occurred by chance.
accuracy in transitivity-like trials. A further analysis of Charles’s, Marie’s, and
Leonardo’s equivalence test performances
revealed that all three of them met the passing
Equivalence and Intraverbal Tests: Charles, criterion for symmetry in the majority of their
Marie, and Leonardo administrations of the test (Fig. 2). Specifically,
The right panels of Figure 1 show equivalence Charles met the symmetry criterion in six of
and intraverbal test data for the three participants nine tests, Marie in eight of nine tests, and
who never passed the equivalence test: Charles, Leonardo in three of four tests. By contrast, all
Marie, and Leonardo. As before, white and filled participants’ accuracy in transitivity trials was
circles represent failing and passing perfor- around chance level; Charles and Leonardo
mance, respectively, on intraverbal tests. Gray never met the transitivity criterion, and Marie
234 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

Sample-Naming
Requirement
6
Symmetry
5 Transitivity

1
Charles
0
21 32 43 54 6
5 76 87 98 109
6

5
Correct Responses

1
Marie
0
13 42 53 64 7
5 68 97 10
8 11
9

1
Leonardo
0
31 4
2 5
3 6
4 5 6 7 8 9
Testing Sessions

Fig. 2. The figure shows performance in symmetry and transitivity trials in failed equivalence tests prior to BC training
for Charles, Marie, and Leonardo. Light gray bars represent the symmetry relation and black bars represent the transitivity
relation. Light gray horizontal dashed lines represent criterion performance for each relation on the equivalence test.

met it in two of nine equivalence tests. As the intraverbal tests did not improve from the
Figure 2 shows, the introduction of the sample- previous phase. Marie’s performance on the
tact requirement was not followed by increased symmetry test was variable, but met the passing
accuracy in either symmetry or transitivity trials. criterion in four of eight administrations, and
Following BC training, Leonardo immedi- in three of the remaining administrations
ately passed the symmetry test, and in the she was close to meeting criterion with only
second administration he also passed the two symmetry errors instead of the one error
accompanying intraverbal test (see Fig. 1). permitted. Marie’s intraverbal test performance
Charles also passed the symmetry test in two did not meet the passing criterion and did not
of three administrations, but performance on improve from the previous phase.
 INTRAVERBAL NAMING 235

Table 4
Charles and Marie’s percentage accuracy in symmetry tests and accompanying intraverbal tests by
trial type.

Charles Marie
Trial Type Symmetry Test Intraverbal Test Symmetry Test Intraverbal Test

A1B1 / A’1B’1 100 67 100 25

A2B2 / A’2B’2 100 33 100 75
A3B3 / A’3B’3 100 67 88 38
B1A1 / B’1A’1 100 67 88 50
B2A2 / B’2A’2 67 33 100 75
B3A3 / B’3A’3 100 100 88 63
A1C1 / A’1C’1 100 33 88 38
A2C2 / A’2C’2 100 33 100 50
A3C3 / A’3C’3 67 33 100 25
C1A1 / C’1A’1 100 33 75 38
C2A2 / C’2A’2 100 67 100 88
C3A3 / C’3A’d 67 0 100 25
B1C1 / B’1C’1 100 0 88 38
B2C2 / B’2C’2 67 67 100 75
B3C3 / B’3C’3 67 0 100 38
C1B1 / C’1B’1 67 0 75 25
C2B2 / C’2B’2 100 0 63 50
C3B3 / C’3B’3 100 0 100 25
Note. Data are aggregated from tests 11 through 13 for Charles, and tests 12 through 19 for Marie.

Because Charles and Marie passed symmetry with low to intermediate accuracy in all types of
tests without ever passing the intraverbal tests, intraverbal trials; intermediate accuracy was
Table 4 shows details of these participants’ primarily related to stimulus class 2. Thus,
symmetry test performance and performance although Figure 1 may indicate some degree of
on the accompanying intraverbal tests. Charles covariation between accuracy in the symmetry
responded either with 100% accuracy (three of test and the intraverbal test, the latter test did
three trials) or 67% accuracy (two of three not indicate reliable correct responding in a
trials) in each type of trial on the symmetry test. sufficient number of trial types for verbal
On the intraverbal test, performance ranged mediation to be plausible. Covariation may
from no correct responses in most B’C’ and have been related to uncontrolled factors, such
C’B’ trials to 100% accuracy (three of three) in as fatigue or inattention that could have
B’3A’3 trials. Intermediate levels of accuracy decreased accuracy in both tests given that
(two of three trials) were observed in A’1B’1, each pair of tests was conducted back-to-back on
A’3B’3, B’1A’1, C’2A’2, and B’2C’2 trials. the same day.
Because at least intermediate accuracy was
observed in A’B’ and B’A’ intraverbal trials Response Latency Analyses
for two of the three stimulus classes, it The upper panel of Figure 3 shows Niels’s,
is conceivable that performance in BA symme- Rita’s, and Emilie’s mean response latencies in
try trials was verbally mediated, with correct baseline, symmetry, and transitivity trials across
stimuli selected by exclusion in trials involving all administrations of the equivalence test, along
the third class (B2A2 trials). However, it would with an overall mean across the three partic-
be difficult to make the same argument for CA ipants. A repeated-measures ANOVA revealed a
and CB symmetry trials, based on generally low significant difference in mean response laten-
accuracy in A’C’, C’A’, B’C’ and C’B’ trials. cies between trial types, F(2, 4) ¼ 33.39, p < .05.
Marie’s accuracy on the symmetry test ranged A Least Significant Difference (LSD) pairwise
from 63% (C2B2 trials) to 100%, with highly comparison revealed that mean response
accurate responding (correct responding in at latencies in transitivity trials (M ¼ 10,800 ms)
least seven of eight trials, or 88%) in all but were significantly longer (p < .05) than mean
three trial types. By contrast, Marie responded response latencies in baseline (M ¼ 5,900 ms)
236 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

14
Niels
12
* Rita

10 Emilie

2
Response Latency (s)

0
Trained Symmetry Transitivity

14
Charles
12 Marie
10 Leonardo

0
Trained Symmetry

Relations

Fig. 3. The upper panel shows mean response latencies by trial type on the equivalence test for participants who passed
the test. The lower panel shows mean response latencies by trial type on the symmetry test for participants who never passed
the equivalence test but passed the symmetry test following BC training.

and symmetry (M ¼ 5,720 ms) trials, whereas Charles and Marie were exposed to three
there was no significant difference (p ¼ 0.54) in additional tests due to their failures to pass the
mean response latencies between baseline and intraverbal test. First, on the tact-intraverbal test,
symmetry trials. The lower panel of Figure 3 Charles’s accuracy was 72% and Marie’s accuracy
shows Charles’s, Marie’s, and Leonardo’s mean was 50%. Thus, neither participant passed the
response latencies in baseline and symmetry test, although Charles’s performance was
trials across all administrations of the symmetry more accurate than in intraverbal tests in the
test, along with an overall mean across the three absence of visual stimuli. Second, on the alter-
participants. A paired-samples t-test revealed no nate tact test, Charles responded “flower” in all
significant difference between their response flower trials, “bird” and “red bird” in two of the
latencies in baseline (M ¼ 6,320 ms) and sym- bird trials with no intelligible response in the
metry (M ¼ 6,540 ms) trials. third, and “state,” “purple state,” and “cookies”
in state trials. Marie responded “no” in all
trials. Thus, there was no indication that
Postexperimental Tests either participant was applying class-consistent
All six participants passed the listener test at names to the stimuli, and no indication that
100% accuracy, suggesting that in Horne and Marie was applying alternative unique names.
Lowe’s (1996) terms, tact training had estab- Charles’s responses perhaps indicated that he
lished name relations. had substituted alternative names for the bird
 INTRAVERBAL NAMING 237

and state names he had been taught. However, participants’ repertoires. For both Niels and
none of the alternate names he gave in this test Emilie, relatively poorer performance on the
had ever appeared as incorrect responses in intraverbal test than on the equivalence test
intraverbal test trials; all of his incorrect was primarily due to incorrect responses in
responses consisted of incorrect state, bird, transitivity-like trials, whereas performance in
and flower names or “I don’t know.” Third, baseline-like and symmetry-like trials was mostly
both Charles and Marie passed both parts of the accurate.
category-name intraverbal test with 89-100% Second, participants who passed the equiva-
accuracy, suggesting that “state,” “bird,” and lence test had significantly longer response
“flower” exerted intraverbal control over the latencies in transitivity than symmetry trials.
emission of names of states, birds, and flowers, This is a common finding in the stimulus
respectively, and vice versa. equivalence literature (Arntzen, Braaten, Lian,
& Eilifsen, 2011; Arntzen & Lian, 2010; Arntzen
Discussion & Nikolaisen, 2011; Bentall, Dickins, & Fox,
1993; Holth & Arntzen, 1998; Spencer & Chase,
No participant passed the equivalence test 1996) that may or may not reflect the covert
without also passing the intraverbal test. Intra- occurrence of verbal or other potential mediat-
verbal test performance covaried with equiva- ing behavior. Although we cannot be certain, it
lence test performance, such that participants is consistent with the notion that longer chains
who passed the equivalence test also performed of trained intraverbal responses are required
with high accuracy on the intraverbal test, and for successful performance in transitivity than
participants who failed the equivalence test in symmetry trials. Thus, nothing about the
performed poorly on the intraverbal test. This performance of the participants who passed the
finding is consistent with the predictions of the equivalence test appears to directly contradict
naming hypothesis (Horne & Lowe, 1996). Horne and Lowe’s (1996) analysis of mediation
However, it does not necessarily follow that by intraverbal naming, although the alternative
successful performance required intraverbal possibility is not ruled out.
behavior in test trials, or that intraverbal By contrast, the performance of the three
responding played a functional role in equiva- participants who failed the equivalence test may
lence test performance. Alternatively, it is be less consistent with intraverbal mediation.
possible that for participants who formed Charlie, Marie, and Leonardo performed poorly
equivalence classes, the names of the stimuli in intraverbal tests that followed failing equiva-
in each class were incorporated into the lence test performance. However, failing perfor-
equivalence class through the combination of mance in the equivalence tests was mostly due to
tact training and AB/AC training, resulting in poor transitivity performance. All three partic-
the emergence of intraverbal relations collater- ipants performed with high accuracy in symme-
ally with the emergence of novel conditional try trials, suggesting that symmetry emerged in
discriminations. Two aspects of the perfor- the absence of an intraverbal repertoire that
mance of participants who passed the equiva- could potentially guide correct responses. Fur-
lence test are relevant to consider. ther, after receiving baseline training on BC
First, for Niels and Emilie, criterion perfor- relations, all three participants ultimately passed
mance in the intraverbal test appeared to lag symmetry tests involving the three relations that
behind criterion performance in the equiva- had not been trained (CA, BA, and BC), whereas
lence test. At first glance, this finding might Charles and Marie (notably, the youngest two
seem to contradict a functional role of the participants in the study) continued to fail
intraverbal repertoire. However, it should be the intraverbal test. Unlike Niels and Emilie’s
kept in mind that relative to the intraverbal initial successes on the equivalence test,
tests, performance on the equivalence tests may Charles and Marie’s accurate symmetry test
have been inflated due to explicitly presented performance could not be attributed to acquisi-
response options increasing the likelihood of tion of a crucial subset of intraverbal relations,
correct guessing. In addition, it might be because analyses of intraverbal test performance
argued that Horne and Lowe’s (1996) intra- revealed only intermediate performance in
verbal naming mechanism requires only base- baseline-like and symmetry-like trials. Perfor-
line-like and symmetry-like intraverbals in the mance did not improve to mastery levels in the
238 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

postexperimental tact-intraverbal test, suggest- Leonardo’s failures to demonstrate equivalence

ing that the presence of the visual stimuli that class formation provided an interesting oppor-
served as samples in the equivalence test did not tunity to observe the emergence of symmetry in
increase the likelihood of intraverbal respond- the absence of transitivity, and in the apparent
ing. In addition, postexperimental tests did not absence of any verbal relations that might play a
suggest that the participants’ failure to pass the role in symmetry test performance.
intraverbal tests could be traced to lack of Another potential limitation is that the
control by supplemental verbal stimuli (state, equivalence test was never administered prior
bird, flower) presented in intraverbal test trials, to training. Because the stimulus relations
nor that the participants had assigned alternative trained and tested in this study were nonarbi-
unique or common names to the stimuli. Finally, trary, it is conceivable that the participants
Charlie, Marie, and Leonardo had approxi- could have acquired them prior to the study or
mately equal response latencies in symmetry outside of the study after their participation
and baseline trials on the equivalence test, as did began. However, we find this implausible based
the three participants who passed the equiva- on the data reported here. First, intraverbal
lence test. By the nature of the inaccessibility of relations were pretested and did not appear to
covert verbal behavior, it may not be possible to be present in any participant’s repertoire.
rule out that during symmetry trials, Charles and Intuitively, it seems unlikely that in the natural
Marie were nevertheless engaging in accurate environment, children would learn to relate
intraverbal responding involving either the outline maps of states other than their home
experimenter-supplied or other names. How- state with images of corresponding state birds
ever, because none of our multitude of measures and state flowers without explicit teaching of
appear to support this being the case, it seems any intraverbal relations in the form of “__ is the
reasonable to conclude that these participants’ state bird of __,” for example. Second, all
symmetry performance was not aided by intra- participants required a fairly large number of
verbal responding. trials to complete AB/AC training.
That only three of six participants passed Overall, the data suggest that when partic-
the equivalence test was unexpected, given that ipants have a unique name for each stimulus in
typically developing children of similar ages a putative equivalence class, accurate perfor-
or younger have readily demonstrated three- mance in subsequent transitivity trials tends to
member equivalence classes in numerous other be accompanied by emergent intraverbal rela-
studies (e.g., Barnes, Browne, Smeets, & Roche, tions between the unique names. Regardless of
1995; Goyos, 2000; Lazar et al., 1984; Sidman why this is the case, this finding may have
et al., 1986; Smeets & Barnes-Holmes, 2003). practical implications; for example, for the
However, comparable failures among children development of educational software applica-
of similar ages or older are not unprecedented tions for children. By contrast, accurate perfor-
in the literature (e.g., Arntzen & Lian, 2010; mance in symmetry test trials may occur in the
Smeets & Barnes-Holmes, 2005). In the present absence of relevant intraverbal relations. As
study, it is possible that equivalence test out- previously noted, the response latency data for
comes were affected by the use of a one-to-many baseline, symmetry, and transitivity trials may
training structure to establish baseline rela- also lend some support to the notion that
tions. Although some discrepancies exist in the participants talked themselves through transi-
literature, at least some studies suggest that one- tivity, but not symmetry trials. Thus, one
to-many training, in which the baseline nodal possible interpretation is that for children of
stimuli serve as samples in all training trials, the ages represented in the study, responding in
yields equivalence-consistent performance less accordance with transitivity depends on verbal
reliably than many-to-one training in which the behavior emitted during training and testing,
nodal stimuli serve as comparisons (e.g., Fields, whereas symmetry requires no such mediation.
Hobbie-Reeve, Adams, & Reeve, 1999; Hove, If symmetry was not verbally mediated, then
2003; Saunders, Drake, and Spradlin, 1999). what accounts for its emergence? Sidman (2000)
Alternatively, some other procedural limitation proposed that equivalence relations consist of
may have contributed to the variable equiva- ordered pairs of events produced as a direct
lence test results across participants. Regardless outcome of the reinforcement contingency
of why they occurred, Charles’s, Marie’s, and that establishes baseline relations. As such, a
 INTRAVERBAL NAMING 239

mediating mechanism is not required to account Behavior analysis in theory and practice: Contributions and
for them. From this perspective, however, it is controversies (pp. 115–138). Hove: Lawrence Erlbaum
not clear why baseline training should produce Associates.
Eikeseth, S., & Smith, T. (1992). The development of
emergent symmetry but not transitivity for functional and equivalence classes in high-functioning
some participants. Alternatively, it is possible autistic children: The role of naming. Journal of the
that symmetry in the present study was not Experimental Analysis of Behavior, 58, 123–133.
emergent, but rather, directly reinforced. Base- Fields, L., Hobbie-Reeve, S. A., Adams, B. J., & Reeve, K. F.
line relations were established through a simul- (1999). Effects of training directionality and class size
on equivalence class formation by adults. The Psychologi-
taneous MTS procedure, such that in each trial, cal Record, 49, 703–724.
the sample remained visible on the screen Goyos, C. (2000). Equivalence class formation via common
while the participant selected a comparison. If, reinforcers among preschool children. The Psychological
during the trial, the participant looked back Record, 50, 629–654.
Goyos, C., Elias, N. C., & Ribeiro, D. M. (2005).
and forth between the sample and the selected Desenvolvimento de um programa informatizado
comparison, it is possible that the establishment para ensino de LIBRAS. In II Congresso Brasileiro de
of the baseline relation also entailed the EducaSc a~o Especial, S~
a o Carlos. II Congresso Brasileiro de
establishment of a prerequisite for selecting EducaSc a~o Especial, Vol. 1, p. 363.
the correct comparison in a symmetry trial. Green, G. (1990). Differences in development of visual–
visual and auditory–visual equivalence relations. Ameri-
Another possibility is that the participants had, can Journal on Mental Retardation, 95, 260–270.
through a history of prior exposure, acquired Hayes, S. C., Barnes-Holmes, D., & Roche, B. (Eds.) (2001).
symmetry responding as a generalized operant Relational frame theory: A post-Skinnerian account of human
in the context of matching objects (cf. Hayes, language and cognition. New York: Plenum.
Barnes-Holmes, & Roche, 2001), but did not Holth, P., & Arntzen, E. (1998). Stimulus familiarity and the
delayed emergence of stimulus equivalence or consis-
have such a history with respect to transitive tent nonequivalence. The Psychological Record, 48,
inference. Variables that influence the develop- 81–110.
ment of symmetry and transitivity in young Horne, P. J., Hughes, J. C., & Lowe, C. F. (2006). Naming
children may warrant further research. and categorization in young children: IV. Listener
behavior training and transfer of function. Journal of the
Experimental Analysis of Behavior, 85, 247–273.
Horne, P. J., & Lowe, C. F. (1996). On the origins of naming
References and other symbolic behavior. Journal of the Experimental
Arntzen, E., Braaten, L. F., Lian, T., & Eilifsen, C. (2011). Analysis of Behavior, 65, 185–241.
Response-to-sample requirements in conditional dis- Horne, P. J., Lowe, C. F., & Randle, V. R. L. (2004). Naming
crimination procedures. European Journal of Behavior and categorization in young children: II. Listener
Analysis, 12, 505–522. behavior training. Journal of the Experimental Analysis of
Arntzen, E., & Lian, T. (2010). Trained and derived Behavior, 81, 267–288.
relations with pictures versus abstract stimuli as nodes. Hove, O. (2003). Differential probability of equivalence
The Psychological Record, 60, 659–678. class formation following a one-to-many versus a many-
Arntzen, E., & Nikolaisen, S. L. (2011). Establishing to-one training structure. The Psychological Record, 53,
equivalence classes in children using familiar and 617–634.
abstract stimuli and many-to-one and one-to-many Kobari-Wright, V. V., & Miguel, C. F. (2014). The effects of
training structure. European Journal of Behavior Analysis, listener training on the emergence of categorization
12, 105–120. and speaker behavior in children with autism. Journal of
Barnes, D., Browne, M., Smeets, P. M., & Roche, B. (1995). Applied Behavior Analysis, 47, 431–436.
A transfer of functions and a conditional transfer of Lazar, R. M., Davis-Lang, D., & Sanchez, L. (1984). The
functions through equivalence relations in three- to formation of visual stimulus equivalences in children.
six-year-old children. The Psychological Record, 45, 405–430. Journal of the Experimental Analysis of Behavior, 41,
Bentall, R. P., Dickins, D. W., & Fox, S. R. A. (1993). Naming 251–266.
and equivalence: Response latencies for emergent Lionello-DeNolf, K. M. (2009). The search for symmetry:
relations. Quarterly Journal of Experimental Psychology, 25 years in review. Learning & Behavior, 37, 188–203.
46B, 187–214. Lowe, C. F., & Beasty, A. (1987). Language and the
Carp, C. L., & Petursdottir, A. I. (2012). Effects of two emergence of equivalence relations: A developmental
training conditions on the emergence of novel intra- study. Bulletin of the British Psychological Society, 40, A42.
verbals: An extension of P
erez-Gonz
alez et al. (2008). Lowe, C. F., Horne, P. J., Harris, F. D. A., & Randle, V. R. L.
The Psychological Record, 62, 187–206. (2002). Naming and categorization in young children:
Carr, D., Wilkinson, K. M., Blackman, D., & McIlvane, W. J. Vocal tact training. Journal of the Experimental Analysis of
(2000). Equivalence classes in individuals with minimal Behavior, 78, 527–549.
verbal repertoires. Journal of the Experimental Analysis of Lowe, C. F., Horne, P. J., & Hughes, J. C. (2005). Naming
Behavior, 74, 101–114. and categorization in young children: III. Vocal tact
Dugdale, N., & Lowe, C. F. (1990). Naming and stimulus training and transfer of function. Journal of the
equivalence. In D. E. Blackman & H. Lejeune (Eds.), Experimental Analysis of Behavior, 83, 47–65.
240 CHARLOTTE L. CARP and ANNA INGEBORG PETURSDOTTIR

Luciano, C., G
omez Becerra, I., & Rodriguez Valverde, M. Santos, P. M., Ma., M. L., & Miguel, C. F. (2015). The role of
(2007). The role of multiple-exemplar training and intraverbal naming in the establishment of baseline
naming in establishing derived equivalence in an and symmetrical conditional relations. The Analysis of
infant. Journal of the Experimental Analysis of Behavior, Verbal Behavior, 31, 162–182.
87, 349–365. Saunders, R., Drake, K. M., & Spradlin, J. E. (1999).
Mandell, C., & Sheen, V. (1994). Equivalence class Equivalence class establishment, expansion, and modi-
formation as a function of the pronounceability of fication in preschool children. Journal of the Experimental
the sample stimulus. Behavioural Processes, 32, 29–46. Analysis of Behavior, 71, 195–214.
Miguel, C. F., Frampton, S. E., Lantaya, C. A., LaFrance, Sidman, M. (2000). Equivalence relations and the rein-
D. L., Quah, K., Meyer, C. S., . . . Fernand, J. K. (2015). forcement contingency. Journal of the Experimental
The effects of tact training on the development of Analysis of Behavior, 74, 127–146.
analogical reasoning. Journal of the Experimental Analysis Sidman, M., & Tailby, W. (1982). Conditional discrimina-
of Behavior, 104, 96–118. tion vs. matching to sample: An expansion of the testing
Miguel, C. F., & Kobari-Wright, V. V. (2013). The effects of paradigm. Journal of the Experimental Analysis of Behavior,
tact training on the emergence of categorization and 37, 5–22.
listener behavior in children with autism. Journal of Sidman, M., Willson-Morris, M., & Kirk, B. (1986).
Applied Behavior Analysis, 46, 669–673. Matching-to-sample procedures and the development
Miguel, C. F., Petursdottir, A. I., Carr, J. E., & Michael, J. of equivalence relations: The role of naming. Analysis
(2008). The role of naming in stimulus categorization and Intervention in Developmental Disabilities, 6, 1–19.
by preschool children. Journal of the Experimental Analysis Skinner, B. F. (1957). Verbal behavior. New York: Appleton-
of Behavior, 89, 383–405. Century Crofts.
P
erez-Gonz
alez, L. A., Herszlikowicz, K., & Williams, G. Smeets, P. M., & Barnes-Holmes, D. (2003). Children’s
(2008). Stimulus relations analysis and the emergence emergent preferences for soft drinks: Stimulus equiva-
of novel intraverbals. The Psychological Record, 58, lence and transfer. Journal of Economic Psychology, 24,
95–129. 603–618.
Petursdottir, A. I., Carp, C. L., Peterson, S. P., & Lepper, Smeets, P. M., & Barnes-Holmes, D. (2005). Auditory–visual
T. L. (2015). Emergence of visual–visual conditional and visual–visual equivalence relations in children. The
discriminations following intraverbal training. Jour- Psychological Record, 55, 483–503.
nal of the Experimental Analysis of Behavior, 103, Spencer, T. J., & Chase, P. N. (1996). Speed analysis of
332–348. stimulus equivalence. Journal of the Experimental Analysis
Petursdottir, A. I., Olafsdottir, A. R., & Aradottir, B. (2008). of Behavior, 65, 643–659.
The effects of tact training and listener training on the Sprinkle, E. C., & Miguel, C. F. (2012). The effects of
emergence of bidirectional intraverbal relations. Jour- listener and speaker training on emergent relations in
nal of Applied Behavior Analysis, 41, 411–415. children with autism. The Analysis of Verbal Behavior, 28,
Randell, T., & Remington, B. (1999). Equivalence relations 111–117.
between visual stimuli: The functional role of naming. Zentall, T. R., Wasserman, E. A., & Urcuioli, P. J. (2014).
Journal of the Experimental Analysis of Behavior, 71, Associative concept learning in animals. Journal of the
395–415. Experimental Analysis of Behavior, 101, 130–151.
Randell, T., & Remington, B. (2006). Equivalence relations,
contextual control, and naming. Journal of the Experi- Received: September 25, 2015
mental Analysis of Behavior, 86, 337–354. Final Acceptance: November 19, 2015