You are on page 1of 6

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2017, 9999, n/a–n/a NUMBER 9999 ()

GO/NO-GO PROCEDURE WITH COMPOUND STIMULI WITH


CHILDREN WITH AUTISM
RAFAEL AUGUSTO SILVA AND PAULA DEBERT
UNIVERSIDADE DE SÃO PAULO INSTITUTO NACIONAL DE CIÊNCIA E TECNOLOGIA SOBRE COMPORTAMENTO,
COGNIÇÃO E ENSINO (INCT-ECCE)

The go/no-go with compound stimuli is an alternative to matching-to-sample to produce condi-


tional and emergent relations in adults. The aim of this study was to evaluate the effectiveness
of this procedure with two children diagnosed with autism. We trained and tested participants
to respond to conditional relations among arbitrary stimuli using the go/no-go procedure. Both
learned all the trained conditional relations without developing response bias or responding to
no-go trials. Participants demonstrated performance consistent with symmetry, but not
equivalence.
Key words: autism, compound stimuli, conditional discrimination, equivalence, go/no-go

Sidman (1971) showed that after training Galloway, 1967; Kangas & Branch, 2008).
some conditional relations (e.g., AB and BC) Position bias may develop because selections
using matching-to-sample (MTS), conditional based on a specific location in the array (left,
relations that were not directly trained (e.g., AC middle, or right) produce intermittent rein-
and CA) emerged when tested in the absence of forcement (Kangas & Branch, 2008). To rem-
programmed consequences. This allowed for edy this problem, extra instructional time
the development of a teaching technology to and/or effort modifying procedures may be
establish socially valid relations in a fast and effi- needed (Bourret et al., 2012; Galloway, 1967;
cient manner (e.g., Rehfeldt, 2011). However, Kangas & Branch, 2008). Thus, the develop-
the establishment of conditional relations via ment of an alternative to MTS that would
MTS with children diagnosed with autism may avoid position bias seems warranted.
depend upon modified teaching and testing An efficient alternative to MTS, which has
conditions (Eikeseth & Smith, 1992; McLay, been investigated in basic research, is the go/no-
Sutherland, Church, & Tyler-Merrick, 2013). go procedure with compound stimuli
One of the possible difficulties associated (e.g., Debert, Matos, & Mcllvane, 2007). One
with MTS is the development of position bias of the advantages of this procedure is that com-
(e.g., Bourret, Iwata, Harper, & North, 2012; pound stimuli (two stimuli presented side-by-
side in a trial) are successively presented, and
the selection response is made in the unique
This study is based on a thesis submitted to the Uni- location provided, so no position bias can be
versity of São Paulo in partial fulfillment of the require-
ments for an M.A. in Experimental Psychology. This developed. Additionally, the go/no-go proce-
research was part of the scientific program of Instituto dure relies on a series of simple discriminations,
Nacional de Ciência e Tecnologia sobre Comportamento, making it more appropriate for use with learners
Cognição e Ensino, with support from the Brazilian
National Research Council (CNPq, Grant # 465686/
lacking prerequisite skills for conditional dis-
2014–1) and the São Paulo Research Foundation (Grant crimination. Considering the importance of
# 2014/50909–8). basic research to inform applied research and
Address correspondence to: Rafael Augusto Silva. Depar- practice, the present study is the first attempt to
tamento de Psicologia Experimental, Universidade de São
Paulo, Av. Prof. Mello Moraes, 1721, São Paulo - SP, Brazil,
assess the use of the go/no-go procedure with
CEP: 05508-900. E-mail: rafael.augustosi@gmail.com compound stimuli with children diagnosed with
doi: 10.1002/jaba.421 autism.
© 2017 Society for the Experimental Analysis of Behavior
1
2 RAFAEL AUGUSTO SILVA and PAULA DEBERT

METHOD compound (two side-by-side stimuli belonging


Participants, Setting and Materials to the same class) was presented, and not press-
ing the space bar (no-go) when a nonrelated
Participants were two 10-year old boys (Bob
compound (two side-by-side stimuli belonging
and Vinnie) diagnosed with autism. They fol-
to different classes) was presented. Incorrect
lowed two-step instructions and performed a
responses consisted of pressing the space bar
variety of self-care routines without prompts.
when a nonrelated compound was presented,
They could also repeat common words (general-
and not pressing when a related compound was
ized echoic repertoire), tact and mand for more
presented. The computer software recorded
than 200 items, answer questions such as “What
participant’s responses. Percentages of correct
is your name?” (intraverbals), and demonstrate
responses per session were calculated by adding
some autoclitic responses (e.g., “this one”).
the total number of correct responses divided
Although Vinnie engaged in three- to five-word
by 24 trials.
sentences, and manded and tacted independ-
We used a nonconcurrent multiple baseline
ently, his intraverbals were mostly prompted.
design across participants. Participants com-
Both participants demonstrated limited sight-
pleted the following experimental conditions:
word vocabulary. Participants’ listener behavior
equivalence relations pretest, pretraining, base-
was evaluated using the Portuguese translation
line relations training, baseline relations test,
of the Peabody Picture Vocabulary Test
symmetry relations test and equivalence rela-
(Capovilla et al., 1997), and their scores yielded
tions posttest.
an age equivalence of 6 years.
We conducted two to four sessions daily,
5 days per week, in a separate room at the parti-
Procedure
cipants’ school. Each session consisted of three
Equivalence relations pre- and posttests. Tests
blocks of eight trials or six blocks of four trials
were conducted before and after training to
depending on the experimental condition (see
evaluate whether participants would respond
below). During sessions, a participant sat at a
discriminatively to transitivity (AC) and equiva-
table with a computer equipped with the soft-
lence (CA) relations when no consequences
ware Compound® (Debert et al., 2007) that
were provided. Participants used the space bar
controlled stimuli presentation and recorded
to respond to each compound stimulus. Com-
participants’ responses. The experimenter sat on
pound stimuli were successively and randomly
the left side of the participant. Each compound
presented for 6 s followed by an intertrial inter-
stimulus consisted of two images presented side-
val of 3 s. Related compounds were A1C1,
by-side (adapted from Devany, Hayes, & Nel-
A2C2, C1A1 and C2A2 and nonrelated com-
son, 1986). The experimenter conducted a brief
pounds were A1C2, A2C1, C1A2 and C2A1
multiple-stimulus preference assessment
(see Figure 1). Each session consisted of three
(DeLeon, & Iwata, 1996) prior to the first ses-
blocks with eight trials each.
sion of each day. After a session, completion of
Prior to the beginning of the session, the
a token board (24 tokens) resulted in access to a
experimenter provided the following instruc-
preferred item for 2 min.
tion: “Let’s play a game. Press the key for cor-
rect figures, and do not press it for incorrect
Response Measurement and Experimental figures. Remember that sometimes you have to
Design press and sometimes you don’t. At the end of
Correct responses consisted of pressing the the game, you will find out how many you got
keyboard space bar (go) when a related correct.” The experimenter delivered praise
GO/NO-GO PROCEDURE 3

Experimental
Related (Go) Nonrelated (No-go)
Phase
A1B1 A2B2 A1B2 A2B1

Baseline
Relations
B1C1 B2C2 B1C2 B2C1
Training

B1A1 B2A2 B1A2 B2A1

Symmetry
Relations
C1B1 C2B2 C1B2 C2B1
Test

A1C1 A2C2 A1C2 A2C1


Equivalence
Relations
Pre- and C1A1 C2A2 C1A2 C2A1
Post-test

Figure 1. Compound stimuli presented in each of the experimental phases.

contingent upon attending (i.e., sitting quietly, Correct responses were followed by the presen-
looking at the experimenter and stimuli) tation of a smiley face on the screen, the experi-
approximately every 30 s. Participants had to menter’s voice (“very good”) issued by the
score at chance levels (approximately 50% cor- software, praise by the experimenter, and a
rect) during pretests to progress to the next token. Incorrect responses were followed by a
phase. The posttest was repeated for three ses- black screen for 2 s.
sions to assess for delayed emergence. To prevent responding to all stimuli, the
Pretraining. We conducted pretraining to experimenter covered the space bar with his
familiarize participants with the task. The right hand as soon as a nonrelated compound
experimenter provided the following instruc- was presented. Initially, the experimenter
tion: “Press the key when two characters from restricted the response during the full 4 s of the
the same cartoon are presented, and do not compound stimulus presentation. The criterion
press when they are from different cartoons.” for decreasing the duration of this restriction
Each of four familiar compound stimuli was (prompt) was one session (24 trials) with a
successively presented for 4 s. Related com- minimum of 92% correct responses. The
pounds were two characters from the same car- restriction was gradually faded (4 s, 3 s, 2 s,
toon and nonrelated compounds were two 1 s, and no prompt). When correct responding
characters from different cartoons. Each session fell below 92% for two consecutive sessions,
consisted of six blocks of four trials each. the restriction was increased by 1 s. The
4 RAFAEL AUGUSTO SILVA and PAULA DEBERT

mastery criterion was 92% accuracy for two TI by dividing the number of correctly imple-
consecutive sessions. The participants required mented trials by the total number of trials and
six (Bob) and 13 (Vinnie) sessions, respectively, converting the result to a percentage. Mean
to learn the task during pretraining. integrity was 99.6% (range, 82%-100%) across
Baseline relations training. Participants participants. We did not collect interobserver
received the following instruction by the experi- agreement given that data collection was
menter: “This is the same activity, but the fig- automated.
ures will be different. You may begin.”
Procedures were the same as in pretraining
except that the compound stimuli were unfa- RESULTS AND DISCUSSION
miliar shapes (see Figure 1). Each session con- Figure 2 shows the percentage of independ-
sisted of three blocks of eight trials each. The ent correct responses across conditions for Bob
training sequence was AB, BC, and mixed AB and Vinnie. During the equivalence relations
and BC relations. The criterion to move to pretest, participants pressed the space bar in
each of these training relations and to the next almost all trials (about 50% accuracy). Bob and
phase was two consecutive sessions with 92% Vinnie required 36 (864 trials) and
of correct responses with no prompt. 18 (432 trials) sessions, respectively, to reach
Baseline relations test. This condition was criterion during training. Vinnie and Bob
identical to baseline relations training described passed the baseline relations test and demon-
above, with the exception that the experimenter strated the emergence of symmetry within one
withheld programmed consequences and (24 trials) and three (72 trials) sessions, respec-
prompts. This served to verify that participants tively. During equivalence relations posttests,
could emit independent correct responses to both participants performed at 50% within
tests in the absence of programmed conse- three sessions, responding in almost all trials.
quences. Immediately after each session, the Although none of the participants passed
experimenter provided access to one of the pre- equivalence relations posttests, our results
ferred items for 2 min to maintain the partici- showed that the go/no-go procedure with com-
pant’s motivation. The passing criterion to pound stimuli was successful in establishing
proceed to the next phase was 92% correct in baseline conditional relations (AB and BC) in
one session. children with autism without the development
Symmetry tests. We tested for accurate of position bias (Galloway, 1967). Most nota-
responding in the presence of the BA and CB bly, the prompting procedure used during
compounds. Otherwise, this condition was training was effective in preventing participants
identical to the equivalence relations test as from responding across all stimulus
described above. compounds.
Although failing to accurately perform dur-
ing equivalence tests is not an unusual outcome
Treatment Integrity with this population (McLay et al., 2013), it is
Treatment integrity (TI) was assessed during possible that failures were related to lack of
33% of the sessions by a second observer. experience with multiple exemplar instruction
Trials were recorded as correct when the experi- for this type of responding (Rosales, Rehfeldt, &
menter used the prompt procedure for the pre- Lovett, 2011), poor verbal skills (Lee, Miguel,
scribed duration and praised correct responses, Darcey, & Jennings, 2015), or simply a prod-
did not praise incorrect responses, and provided uct of prolonged testing conditions in the
no feedback during testing trials. We calculated absence of reinforcement. The inclusion of
GO/NO-GO PROCEDURE 5

Figure 2. Percentage of independent correct responses (go and no-go) across conditions for Bob (upper panel) and
Vinnie (lower panel).

symmetry tests served to verify whether partici- the location of the elements of compounds are
pants had the prerequisite relation for equiva- simply reversed (a form of stimulus generaliza-
lence (Sidman, 1994). However, exposure to tion), but not in equivalence tests. Future stud-
unreinforced symmetry trials could have nega- ies could evaluate different ways to avoid the
tively affected subsequent equivalence class per- development of this type of control, including
formances. Future studies should include teaching simple discrimination of stimulus
multiple exemplar instruction, as well as assess components (e.g., Miguel et al., 2015).
whether reinforced testing conditions would This was the first attempt to use the go/no-
yield better outcomes (e.g., LeBlanc, Miguel, go procedure with compound stimuli as an
Cummings, Goldsmith, & Carr, 2003). alternative to establish conditional relations
Another explanation for the failures observed with children diagnosed with autism. Our
during equivalence tests was that participants results showed that is possible to establish base-
may have learned to respond to compounds as line (AB and BC) conditional discriminations
single stimuli as opposed to attending to each using a simple discrimination procedure invol-
of their elements, as well as their relation. This ving go/no-go responses without developing
type of stimulus control would lead to positive position bias. Future studies should attempt to
outcomes in symmetry tests, as observed when produce equivalence-class performances, as well
6 RAFAEL AUGUSTO SILVA and PAULA DEBERT

as compare the effectiveness of this procedure Kangas, B. D., & Branch, M. N. (2008). Empirical vali-
against MTS. We hope to inspire researchers to dation of a procedure to correct position and stimulus
biases in matching-to-sample. Journal of the Experi-
investigate new and potentially more efficient mental Analysis of Behavior, 90, 103-112. https://doi.
technologies to teach conditional discrimina- org/10.1901/jeab.2008.90-103.
tion and symbolic behavior to children with LeBlanc, L. A., Miguel, C. F., Cummings, A. R.,
Goldsmith, T. R., & Carr, J. E. (2003). The effects
autism. of three stimulus-equivalence testing conditions on
emergent U.S. geography relations of children diag-
nosed with autism. Behavioral Interventions, 18, 270-
REFERENCES 289. https://doi.org/10.1002/bin.144.
Lee, G. P., Miguel, C. F., Darcey, E. K., &
Bourret, J. C., Iwata, B. A., Harper, J. M., & Jennings, A. M. (2015). A further evaluation of the
North, S. T. (2012). Elimination of position-biased effects of listener training on the emergence of
responding in individuals with autism and intellectual speaker behavior and categorization in children with
disabilities. Journal of Applied Behavior Analysis, 45, autism. Research in Autism Spectrum Disorders, 30,
241-250. http://doi.org/10.1901/jaba.2012.45-241. 286-293. https://doi.org/10.1016/j.rasd.2015.
Capovilla, F. C., Nunes, L. R., Nunes, D. R., Araújo, I., 04.007.
Nogueira, D., Bernat, A. B., & Capovilla, A. G. McLay, K. L., Sutherland, D., Church, J., & Tyler-
(1997). Desenvolvimento do vocabulário receptivo- Merrick, G. (2013). The formation of equivalence
auditivo da pré-escola à oitava série: Tabelas de nor- classes in individuals with autism spectrum disorder:
matização fluminense baseadas e aplicação coletiva A review of the literature. Research in Autism Spec-
em sala de aula da tradução brasileira do Peabody Pic- trum Disorders, 7, 418-431. https://doi.org/10.1016/
ture Vocabulary Test. Ciência Cognitiva: Teoria, Pes- j.rasd.2012.11.002.
quisa e Aplicação, 1, 371-430. Miguel, C. F., Frampton, S. E., Lantaya, C. A.,
Debert, P., Matos, M. A., & McIlvane, W. (2007). Con- Quah, K., Meyer, C. S., LaFrance, D., …
ditional relations with compound abstract stimuli Fernand, J. K. (2015). The effects of tact training on
using a go/no-go procedure. Journal of the Experimen- the development of analogical reasoning. Journal of
tal Analysis of Behavior, 87, 89-96. https://doi.org/10. the Experimental Analysis of Behavior, 104, 96-118.
1901/jeab.2007.46-05. https://doi.org/10.1002/jeab.167.
DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a Rehfeldt, R. A. (2011). Toward a technology of derived
multiple-stimulus presentation format for assessing stimulus relations: An analysis of articles published in
reinforcer preferences. Journal of Applied Behavior the Journal of Applied Behavior Analysis, 1992–2009.
Analysis, 29(4), 519-533. https://doi.org/10.1901/ Journal of Applied Behavior Analysis, 44, 109-119.
jaba.1996.29-519. http://doi.org/10.1901/jaba.2011.44-109.
Devany, J. M., Hayes, S. C., & Nelson, R. O. (1986). Rosales, R., Rehfeldt, R. A., & Lovett, S. (2011). Effects
Equivalence class formation in language-able and of multiple exemplar training on the emergence of
language-disabled children. Journal of the Experimen- derived relations in preschool children learning a sec-
tal Analysis of Behavior, 46, 243-257. https://doi.org/ ond language. The Analysis of Verbal Behavior, 27,
10.1901/jeab.1986.46-243. 61-74.
Eikeseth, S., & Smith, T. (1992). The development of Sidman, M. (1971). Reading and auditory-visual equiva-
functional and equivalence classes in high-functioning lences. Journal of Speech and Hearing Research, 14,
autistic children: The role of naming. Journal of the 5-13.
Experimental Analysis of Behavior, 58, 123-133. Sidman, M. (1994). Equivalence relations and behavior: A
https://doi.org/10.1901/jeab.1992.58-123. research story. Boston, MA: Authors Cooperative.
Galloway, C. (1967). Modification of a response bias
through differential amount of reinforcement. Journal Received November 10, 2015
of the Experimental Analysis of Behavior, 10, 375-382. Final acceptance February 14, 2017
https://doi.org/10.1901/jeab.1967.10-375. Action Editor, Caio Miguel

You might also like