ERIC EJ877516: Acquisition of Intraverbal Behavior: Teaching Children with Autism to Mand for Answers to Questions

2010·0.8 MB·English

by  ERIC

JOURNALOFAPPLIEDBEHAVIORANALYSIS 2010, 43, 1–17 NUMBER1 (SPRING2010)
ACQUISITION OF INTRAVERBAL BEHAVIOR: TEACHING CHILDREN
WITH AUTISM TO MAND FOR ANSWERS TO QUESTIONS
EINAR T. INGVARSSON
UNIVERSITYOFNORTHTEXAS
CHILDSTUDYCENTER,FORTWORTH
AND
TATIA HOLLOBAUGH
YOUNGSTOWNSTATEUNIVERSITY
Fourboyswithautismweretaughtviaechoicpromptingandconstantpromptdelaytomandfor
answers to questions by saying ‘‘I don’t know please tell me’’ (IDKPTM). This intervention
resulted in acquisition of the IDKPTM response for all 4 participants and in acquisition of
correct answers to most of the previously unknown questions for 2 participants. For 1
participant,tangiblereinforcementresultedinincreasedfrequencyofcorrectanswers,anddirect
promptingofcorrectanswerswaseventuallyconductedforthefinalparticipant.TheIDKPTM
responsegeneralizedtountargetedunknownquestionswith3participants.Resultsofpersonand
setting generalization probes varied, but some generalization eventually occurred for all
participants following additional training or interspersal ofprobetrials withtraining trials.
Key words: autism, generalization, intraverbals, manding for information, question
answering, verbalbehavior
_______________________________________________________________________________
Autism is a developmental disability that approaches (e.g., Sundberg & Partington,
affects an increasing number of families world- 1998) emphasize language and communication
wide. Recent estimates indicate that 1 of every interventions based on B. F. Skinner’s Verbal
1508-year-oldchildrenintheUnitedStateshas Behavior (1957). In this approach, verbal
an autism spectrum disorder (Centers for behavior is defined as any behavior whose
Disease Control and Prevention, n.d.). One of reinforcement is mediated by other people. A
the defining characteristics of autism spectrum distinction is made between verbal operants
disorders is marked delay in or absence of based on characteristic features of stimulus
functional language or other communication control, motivational operations, and reinforce-
(Filipek et al., 1999). Any comprehensive ment.
intervention program for children with autism Four verbal operants—the tact, the mand,
should therefore emphasize the goal of increas- the echoic, and the intraverbal—are relevant to
ing language and communication skills. Some thecurrentdiscussion.Thetactisunderspecific
early intensive behavioral intervention (EIBI) stimulus control and is maintained by a
generalized social reinforcer (e.g., a child may
We thank Bernadette Treece, Jacob Gailey, Denise
say ‘‘cookie’’ in the presence of a cookie, and
Kurelko, Renee Reagan, and Michael Clayton for their
assistance with the conduct of the study and Anthony the response is reinforced with praise). The
Cammilleri for useful comments on an earlier version of mand is evoked by a specific motivating
this manuscript. In addition, we are grateful to the
operation and reinforced with a characteristic
teachers,staff,andstudentsattheRichCenterforAutism
consequencerelatedtothemotivatingoperation
inYoungstown, Ohio, fortheir cooperation andsupport.
Address correspondence to Einar T. Ingvarsson, (e.g., a child may say ‘‘cookie’’ when hungry,
Department of Behavior Analysis, 1155 Union Circle, andtheresponseisreinforcedwithaccesstothe
Box 310919, University of North Texas, Denton, Texas
cookie). The echoic is a verbal operant that is
76203(e-mail: [email protected]).
doi:10.1901/jaba.2010.43-1 underthestimuluscontrolofaprecedingverbal
1
2 EINAR T. INGVARSSON and TATIA HOLLOBAUGH
stimulus, maintained by generalized reinforce- child acquired intraverbal responding. Gold-
ment, and has a point-to-point topographical smith, LeBlanc, and Sautter (2007) taught 3
correspondence to the preceding stimulus (e.g., young boys with autism to answer questions
a child says ‘‘cookie’’ when an adult says related to categories (e.g., ‘‘What are some
‘‘cookie,’’ and the child receives praise as a things you wear?’’) using tact prompts (e.g.,
result).Theintraverbal,whichisthemainfocus pictures of clothing) and prompt delay.
of the current experiment, is a verbal operant Although it is clearly desirable to teach
thatisunderthestimuluscontrolofapreceding correct answers directly (as in Finkel &
verbal stimulus, without point-to-point corre- Williams, 2001; Goldsmith et al., 2007), an
spondence, and is maintained by generalized alternativestrategyistoteachageneralresponse
reinforcement (e.g., a child says ‘‘cookie’’ when that may lead to acquisition of intraverbals. In
asked ‘‘What did you have for snack?’’ and the other words, the children may be taught to
child receives praise). mand for information (Sundberg & Michael,
Target behaviors in intraverbal training 2001). Some support for this notion comes
include conversational turns, categorization, from research by Sundberg, Loeb, Hale, and
and fill-in-the-blank tasks; the current study Eigenheer (2002) and Endicott and Higbee
focused on question answering. Question (2007). These researchers taught children with
answering is commonly included in EIBI autism to ask the questions ‘‘Where is it?’’ and
curricula, either targeted directly as a separate ‘‘Who has it?’’ with regard to items initially
program(Sundberg&Partington,1998;Taylor presentednoncontingentlyandthenhiddenout
& Jasper, 2001; Taylor & McDonough, 1996) ofviewinspecifiedlocationsoronpeopleinthe
orindirectlythroughprogramsthattargetother immediate environment. These investigators
general skills (e.g., concepts such as yes–no, found that the children manded equally with
emotions, functions of body parts, and general high-preference and low-preference items, sug-
knowledge; McEachin & Leaf, 1999). Previous gesting that the information regarding the
research has indicated that transfer-of-function location of the item may have taken on
procedures are effective in establishing intra- reinforcing quality that may not have been
verbal behavior (Braam & Poling, 1983; completely dependent on the value of the
Partington & Bailey, 1993). Transfer-of- hidden item. This lends some support to the
function procedures include the delivery of notion that questions such as ‘‘Where is it?’’
prompts that reliably evoke the desired re- may be maintained by getting access to
sponse topography. The prompts are then ‘‘information’’ and may thus be conceptualized
faded (e.g., through delayed prompting), and as mands for information. In addition, Wil-
stimulus control is transferred to the desired liams, Perez-Gonzalez, and Vogt (2003) taught
antecedent. 2 children with autism to ask ‘‘What’s in the
Ahandfulofstudieshaveevaluatedtheuseof box?,’’ ‘‘Can I see it?,’’ and finally ‘‘Can I have
such procedures to teach question answering it?’’ regarding preferred items placed out of
and other intraverbal behavior to children with sight in a box. In the latter two cases the
autism. Finkel and Williams (2001) found that questions (i.e., mands) were reinforced by the
textual prompts (i.e., printed text) were more sight of the object and access to the object.
efficient than echoic prompts in establishing However, the question ‘‘What’s in the box?’’
questionansweringinayoungboywithautism. may be conceptualized as a mand for informa-
Thetargetanswersweremultiwordphrases,and tion because the reinforcer consisted of a verbal
the prompts were faded by reducing the length statement describing which preferred item was
of the prompts one word at a time until the hidden in the box.
INTRAVERBAL BEHAVIOR 3
Previous research has evaluated similar pro- difficult rather than easy demand conditions
cedures in order to establish verbal operants (Weeks & Gaylord-Ross, 1981). A mand for
other than intraverbals. For example, Taylor information may morph a difficult and unfa-
and Harris (1995) presented children with miliar demand situation into an easier demand
autism with items they were able to tact, along situation,therebyreducingstereotypyandother
with some items they were not able to tact. A undesirable behavior. Second, the IDKPTM
prompt-delay procedure established the phrase response may enable children to mand for a
‘‘What is that?’’ as a mand for information, more intrusive prompting level while avoiding
which generalized across environments and errors. This may reduce the overall numbers of
resulted in the acquisition of novel tacts. A errors that may otherwise occur as less intrusive
second example was provided by Esbenshade prompts are introduced during most-to-least
and Rosales-Ruiz (2001), who found evidence prompt fading, progressive prompt delay, or
of tact acquisition in a 5-year-old boy with other errorless teaching procedures. Third, the
autism after he was taught to ask ‘‘What is IDKPTM response may enable children to
that?’’ in the presence of unknown items across benefit from a broad range of programmed and
avarietyoftasks.However,weareawareofonly naturally occurring learning opportunities if
onestudythathasevaluatedprocedurestoteach shown to generalize across settings, people, and
a mand that led to the acquisition of new stimuli.
intraverbals. Ingvarsson, Tiger, Hanley, and To date, the majority of research on verbal
Stephenson (2007) first taught the participants behavior has focused on tacts and mands rather
to say ‘‘I don’t know’’ (IDK) in response to than intraverbals (Dymond, O’Hora, Whelan,
unknown questions. Desirable generalization & O’Donovan, 2006; Sautter & LeBlanc,
across teachers and unknown questions was 2006). Although a handful of studies on
found, but undesirable generalization to previ- intraverbal behavior have been published in
ously known questions also occurred (i.e., the the last 2 to 3 years (e.g., Perez-Gonzalez,
children started responding to previously Garcia-Asenjo, Williams, & Carnerero, 2007;
known questions by saying IDK). Second, the Petursdottir, Carr, Lechago, & Almason, 2008;
children were taught to say ‘‘I don’t know, Petursdottir, O´lafsdo´ttir, & Arado´ttir, 2008),
please tell me’’ (IDKPTM) in response to moreresearchisneeded.Hence,abroadgoalof
unknown questions. Every time the children the current study was to strengthen the
engaged in the IDKPTM response, they were empirical basis for intraverbal training for
provided with the correct answer to the children with autism. A more specific goal was
question. IDKPTM generalized across teachers to replicate and extend the Ingvarsson et al.
and questions, but correct answers increased to (2007) study systematically. The current study
acceptable levels only after toy access was made differed from the previous one in the following
contingent on their occurrence. manner. First, the participants were children
Teaching a mand for information, such as with autism spectrum disorders rather than
the IDKPTM response, may be valuable for at children with language delays. Second, it is
least three reasons. First, research has suggested possible that in the previous study, a history of
that levels of stereotypic behavior (e.g., echola- saying IDK to unknown questions reduced the
lia) in children with autism are greatest when acquisition of correct answers after IDKPTM
unfamiliar tasks (e.g., unknown questions) are training occurred. Therefore, IDK was not
presented (Charlop, 1986; Turner, 1999). taught prior to teaching IDKPTM in the
Other studies have shown that demand-related current study. Third, we added generalization
problem behavior is more likely to occur under measures that consisted of asking the partici-
4 EINAR T. INGVARSSON and TATIA HOLLOBAUGH
pants’ regular teachers to carry out generaliza- Setting
tion probes in their classrooms. Fourth, in the Sessions were conducted in small classrooms
previous study, teaching was implemented (henceforth referred to as training rooms)
in the context of toy play, two trials were designed for small group or individualized
presented per minute, and all question sets teaching, located at the participants’ school.
were interspersed randomly in each session The training rooms contained child-sized
(i.e., known and unknown, targeted and furniture, art materials, and toys. During each
untargeted questions were interspersed). This session,theexperimenterandtheparticipantsat
arrangement mimicked everyday classroom inchairsfacingeachother,andtheobserverssat
interaction and probably supported the gener- 2 to 3 m to the side. We conducted classroom
alization of IDK and IDKPTM to untargeted generalizationprobesintheparticipants’regular
sets. However, it may also have reduced the classrooms with other children present. These
speed of acquisition of IDK, IDKPTM, and classrooms varied in size and layout, but were
correct answers. Therefore, in the current large enough to accommodate eight to 10
study, we conducted brief sessions with rapidly students and two or three teachers. During the
presentedtrialsanddidnotinterspersequestion classroom generalization probe sessions, the
sets. classroom teacher asked questions while seated
next to the child at a child-sized desk, and the
observers stood or sat 2 to 3 m to the side.
METHOD
Participants Measurement
Theparticipantswere4boyswithadiagnosis Observers scored the participants’ verbal
of autism—Chris, Neil, Matt, and Jim—who responses using event recording. For each trial
were 10, 7, 6, and 4 years old, respectively. (i.e., the presentation of a single question), the
Chris, Matt, and Jim were Caucasian, and Neil observers circled codes on a data sheet indicat-
was African American. All the boys attended a ing whether the participants gave the correct
university-based school for children with au- answer to the question or whether they said
tism, where they received full-day educational IDKPTM. The observers also scored whether
services 5 days per week. The participants were these responses were prompted (i.e., preceded
selected based on the recommendations of the by an echoic prompt) or independent. The
speech-language therapist who worked at the observers scored responses as prompted if they
school and who was familiar with the verbal were initiated within 5 s after the prompt
abilities of all the students. Specifically, she was presentation and independent if they were
asked to nominate children who had difficulty initiated within 5 s of the question and before
answering common questions. Based on infor- the prompt.
mationfromthespeech-languagepathologist,as
well as informal observations conducted by the Interobserver Agreement
first author, Chris, Jim, and Neil had fairly A second observer simultaneously but inde-
well-established echoic, manding, and tacting pendentlycollecteddataduring55%ofsessions
skills. Matt also had relatively strong echoic for Chris, 58% for Neil, 45% for Matt, and
skills, but more limited tacting and manding 51% for Jim. We scored a trial (i.e., the
repertoires compared with the others. All the presentation of a single question) as an
participants had relatively undeveloped intra- agreement if both observers circled the same
verbal skills compared with same-age peers, code or as a disagreement if any scoring for a
with Matt displaying the greatest deficiency in given trial differed. For each session, the
that skill area. number of trials scored in agreement was
INTRAVERBAL BEHAVIOR 5
divided by the total number of trials and subsequent experimental phases. The partici-
converted into a percentage. Mean agreement pants’ teachers agreed to refrain from including
was99%(range,87%to100%)forChris,98% the selected questions in educational activities
(range, 80% to 100%) for Neil, 99% (range, during other parts of the school day.
89%to100%)forMatt,and99%(range,90% Baseline. In baseline, the experimenter tar-
to 100%) for Jim. geted questions from Known Set 1, Unknown
Set 1, and Unknown Set 2. Thus, a total of 15
Procedure
questions were targeted in each session. This
Pretest and question selection. The pretests
number remained constant for experimental
were similar to those described by Ingvarsson et
sessions (but not generalization probes, see
al. (2007) and included similar questions. The
below) throughout the experiment. The ques-
questions targeted personal information (e.g.,
tions were always asked in the same order;
‘‘Wheredoyoulive?’’),generalknowledge(e.g.,
KnownSet1first,followedbyUnknownSet1,
‘‘Where do you buy groceries?’’), and academic
and finally Unknown Set 2. The order of
skills (e.g., ‘‘How much is a dime?’’). A total of
questions within each set also remained the
56 questions were included in the pretest. We
same. The experimenter asked the 15 questions
divided the questions into four sets, three of
in rapid succession, allowing 5 s for an answer
which contained 15 questions and one con-
to occur. The responses ‘‘I don’t know, please
tained 11 questions. Each set was targeted three
tell me’’ and ‘‘I don’t know’’ would have been
times.Thus,weconductedatotalof12pretests
praised, but no such responses occurred (‘‘I
and asked each question three times. The
don’t know’’ never occurred throughout the
pretests were conducted over a span of 3 to 7
experiment). Correct answers were followed by
days. We delivered no prompts during the
descriptivepraise(e.g.,‘‘That’sright,acowsays
pretest. The experimenter praised correct an-
moo’’). The experimenter delivered descriptive
swers and ignored incorrect answers.
praise throughout the experiment whenever
Basedonthepretestresults,weclassifiedeach
correct answers occurred to any question. If
question as unknown if it was always answered
the participant gave an incorrect or no answer,
incorrectly (or no answer occurred) and known
the experimenter asked the next question but
if it was always answered correctly. We then
created four unique sets of unknown questions provided no other consequence.
and two unique sets of known questions based IDKPTM training. This phase was identical
onthisclassification,withfiveuniquequestions to baseline, except that the questions from
in each set (due to experimenter error, Matt’s Unknown Set 1 were targeted for IDKPTM
Unknown Set 3 included four questions). In training.Theexperimenterusedechoicprompt-
subsequent experimental sessions, the first ing and constant prompt delay (Wolery et al.,
author targeted Known Set 1 and Unknown 1992) to teach the participants to engage in the
Sets 1 and 2 in experimental sessions in the IDKPTM response in the following manner:
training room. One of the research assistants Initially, the experimenter prompted the
targeted Known Set 2 and Unknown Set 3 in IDKPTM response by providing an immediate
generalization probes in the training room, and verbal prompt after asking a question from
each participant’s classroom teacher targeted Unknown Set 1. (e.g., ‘‘How much is a dime?
Unknown Set 4 and Known Set 1 in classroom Say ‘I don’t know, please tell me’’’). After the
generalization probes. Questions that were participant’s IDKPTM response, the experi-
sometimes answered correctly and sometimes menter modeled the correct answer (e.g., ‘‘A
incorrectly during the pretest were not given dime is 10 cents’’). If the child did not repeat
any classification and were not used in the correct answer (‘‘10 cents’’), the experi-
6 EINAR T. INGVARSSON and TATIA HOLLOBAUGH
menter provided a prompt (e.g., ‘‘say ‘10 incorrect answers were ignored). We chose this
cents’’’). When the child stated the correct intervention because his classroom teachers
answer (with or without a prompt), the indicated that food was used successfully in
experimenter provided descriptive praise (e.g., the past to increase participation in the
‘‘That’s right, a dime is 10 cents’’). classroom, and the use of these reinforcers was
When the participant echoed the IDKPTM acceptable to the teachers and his family. His
prompt on two consecutive trials, we intro- preference hierarchy was determined by obtain-
duced a 5-s delay between the question and ingalistof10potentiallypreferrededibleitems
the prompt. We reintroduced an immediate fromhisteachersandthenconductingapaired-
prompt if the participant did not engage in choice preference assessment based on the
either the IDKPTM response or the correct methods of Fisher et al. (1992). The experi-
answer for two consecutive trials. When the menter then delivered the three most preferred
IDKPTM response occurred unprompted (i.e., items in alternation according to the prevailing
during the 5-s delay), only the latter part of the schedule.
teaching sequence was implemented (i.e., the In the initial stages of the edible reinforce-
experimenter provided the correct answer and ment phase, the experimenter delivered edible
prompted the participant to say it if necessary). items contingent on correct answers on a fixed-
Wealsoimplementedonly thelatterpart ofthe ratio (FR) 1 schedule (note that correct answers
teaching sequence if the IDKPTM response were reinforced only if they were not preceded
occurred following questions from Unknown by IDKPTM). Schedule thinning began after
Set 2 and Known Set 1 (i.e., if the IDKPTM random question order had been implemented
response generalized to untargeted questions). according to the criteria described below. The
Otherwise, the contingencies for Known Set 1 schedule of edible delivery was thinned when
and Unknown Set 2 were identical to baseline Neil answered at least 60% of the questions in
(i.e., no prompting was conducted, and incor- each unknown set (Unknown Sets 1 and 2)
rect answers were ignored). An exception correctly across two consecutive sessions. The
occurred for Matt, with whom we eventually schedule thinning proceeded as follows: FR 1,
conductedIDKPTMtrainingforUnknownSet FR 2, FR 3, FR 4, FR 5, NCR. The NCR
2, because generalization across questions did schedule involved delivering the mean number
not occur following initial IDKPTM training of edible items that had been delivered in the
with Unknown Set 1. preceding FR 5 sessions; however, they were
Additionalinterventions.WithNeilandMatt, made available as each session started, indepen-
IDKPTM training did not lead to acceptable dent of any particular behavior other than
acquisition of correct answers. We therefore attendingthesession.Wechosetodeliveredible
implemented additional interventions, but the items independent of correct answers following
procedures differed for these 2 participants as a schedule thinning to strengthen the case that
function of their response patterns. With Neil, the answers functioned as intraverbals and to
correct answers occurred at intermediate levels reduce the likelihood that the answers were
during the IDKPTM training phase, but the maintained primarily by access to edible items
target level of at least 80% correct across two rather than praise and approval.
consecutive sessions was not reached. We We could not implement additional rein-
therefore added edible items contingent on forcement for correct answers with Matt,
correct answers while other contingencies were because no independent correct answers oc-
keptintactfromthepreviousphase(i.e.,correct curred during the IDKPTM training phase.
answers and IDKPTM were praised, and Therefore, the experimenter conducted one
INTRAVERBAL BEHAVIOR 7
session (Session 38) in which correct answers to conducted. Research assistants conducted
the questions in Unknown Set 1 were directly probes in the training room, and each partic-
prompted (using 5-s constant prompt delay) ipant’s classroom teacher conducted probes in
while IDKPTM training was suspended. Be- their regular classroom. A unique set of
cause independent (unprompted) correct an- unknown questions was used for each type of
swers started to occur immediately, we sus- probe: Unknown Set 3 in the training room
pended direct prompting of correct answers probes and Unknown Set 4 in the classroom
following Session 38, in order to evaluate probes. These sets were never targeted in any
whether the correct answers would continue to other condition. A unique set of known
occur under simple differential reinforcement. questions (Known Set 2) also was targeted in
During Sessions 39 to 47, correct answers were the training room probes, but the classroom
praised, and IDKPTM resulted in provision of probes included the same set of known
the correct answer. No prompting was con- questions that had been targeted in the
ducted during these sessions. We reinstated the experimental sessions (Known Set 1). We chose
IDKPTM training contingencies starting with this approach because only a limited number of
Session 48 after we noticed a reduction in known questions were identified for some
correct answers. participants. The questions were always asked
Random question order. As described above, in the same order, with known questions first,
the experimenter always asked the questions in followed by unknown questions.
thesameorderduringbaselineandinitialstages One or both of the authors supervised all
of IDKPTM training. However, after acquisi- probe sessions and instructed the research
tion of correct answers had becomeevident, the assistants and classroom teachers how to
experimenter asked the questions in random proceed. The instructions specified that ques-
order to ensure that question answering was tions should be asked in sequence, incorrect
under the stimulus control of the relevant responses ignored, correct answers praised, and
questions and not other aspects of the experi- IDKPTMshouldresultintheparticipantbeing
mental context. This was accomplished by the given the correct answer and prompted to
experimenteraskingthequestionsfromallthree repeat the correct answer if necessary.
sets (Known Set 1, Unknown Set 1, and Preintervention probes consisted of two
Unknown Set 2) in random order from session training room probes and one classroom probe,
to session, such that all 15 questions were conducted in that order. The experimenter
intermixed randomly. conducted the first probe following Session 1
Generalization probes. The purpose of the forChris,Session2forNeil,Session3forMatt,
generalization probes was to evaluate whether and Session 1 for Jim. The experimenter
IDKPTM would generalize to questions (Un- conducted the remaining two probes following
known Sets 3 and 4) not included in the Session2forChris,Session5forNeil,Session9
IDKPTM training sessions, across persons for Matt, and Session 12 for Jim. The experi-
(research assistant and classroom teacher), and menter conducted postintervention probes
setting (classroom). The procedures during the when acquisition of IDKPTM had been seen
generalization probes were identical to those with Unknown Sets 1 and 2 and when relative
used during baseline in that the experimenter stability in the levels of both IDKPTM and
praised correct answers, ignored incorrect correct answers had been observed. The post-
answers, provided the correct answer if intervention probes were conducted following
IDKTPM occurred, but delivered no prompts. Session 24 for Chris, Session 32 for Neil,
Two types of generalization probes were Session 35 for Matt, and Session 19 for Jim.
8 EINAR T. INGVARSSON and TATIA HOLLOBAUGH
IfgeneralizationoftheIDKPTMresponseto generalization of IDKPTM to known questions
UnknownSets3and4didnotoccurduringthe was not evident, in that correct answers to
probe sessions, we implemented additional Known Set 1 remained stable throughout.
interventions. For Chris and Neil, a research Moreover, the data show that Chris gradually
assistant conducted a single session of direct learned the correct answers to the questions in
training with Unknown Set 3. The procedures Unknown Sets 1 and 2 during this phase. After
during this training were identical to the an initial increase in IDKPTM, a gradually
IDKPTM training described above. For Neil, decreasing trend of IDKPTM was seen, in that
the classroom teacher also conducted direct correctanswersincreasedforbothquestionsets.
training with Unknown Set 1 (previously Toward the end of his participation, he was
targeted during initial IDKPTM training) after answering at least 80% of questions from both
generalization to the classroom failed to occur sets correctly across repeated sessions. His
for the second time. For Matt, the research accuracy remained high during the last five
assistant conducted a single interspersal session, sessions, whichwereimplementedwithrandom
in which Unknown Set 3 trials were inter- question order.
spersed randomly with trials containing ques- Neil’s results are shown in the third and
tions from Unknown Sets 1 and 2 (with which fourth panels of Figure 1. Like Chris, Neil
Matt had already shown acquisition of answered known questions, but not unknown
IDKPTM). questions, correctly in baseline. There were no
instances of IDKPTM in baseline. With the
Experimental Design
initiation of IDKPTM training, acquisition of
The direct and indirect effects of the
IDKPTM was quickly evident across both the
IDKPTM training (and additional interven-
target (Unknown Set 1) and the generalization
tions) were evaluated in a nonconcurrent
(Unknown Set 2) sets. Undesirable generaliza-
multiple baseline design across participants.
tion of IDKPTM to the known questions was
Generalization of the IDKPTM response across
minimal (there were only three instances in the
questions was evaluated by including nontar-
IDKPTM training phase). An increase in
geted (Unknown Set 2) questions in the
correct answers was seen across both unknown
experimental sessions. In addition, generaliza-
sets after three sessions of IDKPTM training.
tion across questions, people, and settings was
However, the number of correct answers
evaluated via pre- and postintervention gener-
remained variable throughout this phase and
alization probes.
did not reach the stable high levels that had
been seen with Chris. Therefore, starting with
RESULTS
Session 38, preferred edible items were deliv-
The results for the experimental sessions are ered contingent on correct answers while other
showninFigure 1.Onlyunpromptedresponses procedures remained identical to the IDKPTM
areshown.ResultsforChrisarepresentedinthe phase. With this intervention, Neil consistently
top two panels. Chris answered the known, but answered three to five questions from each
not the unknown, questions correctly in unknown set correctly for the remainder of the
baseline. There were no instances of IDKPTM experimental sessions. Undesirable generaliza-
in baseline. When we implemented IDKPTM tion of IDKPTM to the known questions
training, evidence of acquisition of the occurred infrequently in this phase, with a total
IDKPTM response was seen quickly, both with of five instances. This performance remained
the target (Unknown Set 1) and the generaliza- stable while the schedule of edible delivery was
tion (Unknown Set 2) sets. Undesirable thinned from FR 1 to FR 5 and during
INTRAVERBAL BEHAVIOR 9
Figure1. ThenumberofcorrectanswersandIDKPTMresponsesforChris,Neil,Matt,andJim.Thedottedphase
linesindicate whenrandomized question order began witheach participant.
10 EINAR T. INGVARSSON and TATIA HOLLOBAUGH
noncontingent edible delivery. His accuracy training, Jim showed quick acquisition of that
remained high after the order of question was response, which generalized to Unknown Set 2.
changed to random during the last 15 sessions. There was only one instance of undesirable
Matt’sresultsareshowninthefifthandsixth generalization of IDKPTM to the known
panels of Figure 1. In baseline, he answered all questions.AcquisitionofIDKPTMwasquickly
of the known questions and none of the un- followedbyanincreaseincorrectanswersacross
known questions correctly. There were no both unknown sets and a corresponding
instances of IDKPTM in baseline. When decrease in IDKPTM. The last four sessions
IDKPTM training was implemented, includedrandomquestionorder;Jim’saccuracy
IDKPTM increased for Unknown Set 1 but remained high.
did not generalize to Unknown Set 2 (with the To evaluate the extent to which individual
exception of one response in Session 10). No unknown questions were acquired by each
undesirable generalization of IDKPTM to participant, we counted the number of ques-
known questions occurred. No evidence of tions answered correctly from Known Set 1,
acquisition of correct answers was observed in Unknown Set 1, and Unknown Set 2 during
this phase. Because of the lack of generalization the last two sessions of baseline, as well as the
of the IDKPTM response across unknown last two sessions of the last intervention phase
questions, IDKPTM training was next carried for each participant. To be counted as correct,
out with Unknown Set 2, resulting in imme- an individual question had to be answered
diate acquisition of IDKPTM with that set of correctly in two consecutive sessions. All 4
questions. However, correct answers did not participants answered all five known questions
emerge. Because no correct answers occurred, correctly towards the end of both baseline and
we could not implement additional reinforce- intervention.Noneoftheparticipantsanswered
ment of correct answers as we had with Neil. anyunknownquestionscorrectlyinthelasttwo
Therefore, we carried out a single session of sessions of baseline, except for Jim, who
direct training of correct answers (Session 38) answered one question from Unknown Set 1
while IDKPTM training was suspended. In the correctly.All4participantsansweredeitherfour
subsequent sessions, correct answers increased or five (out of five) questions from the two
across both unknown sets. However, a decrease unknown sets correctlyin their last twosessions
was seen later without an accompanying of intervention, indicating that acquisition of
increase in IDKPTM. Therefore, IDKPTM correct answers had taken place for most of the
training was reinstated, resulting in a brief previously unknown questions.
increase in IDKPTM and an eventual recovery The results of the generalization probes are
of high levels of correct answers. The last 10 shown in Figure 2. Results are shown only for
sessions were carried out using random session Unknown Set 3 (training room probe) and
order.Accuracywashighduringthemajorityof UnknownSet4 (classroomprobe). KnownSets
these sessions. 1 and 2 also were included in these probe
Jim’s results are shown in the bottom two sessions; however, all 4 participants answered
panels of Figure 1. Jim answered all known the known questions 100% correctly in each
questions correctly in baseline, and there were generalization probe session. Therefore, to
no instances of IDKPTM. Initially, Jim simplify the data presentation, the results for
answered all unknown questions incorrectly, the known questions are not presented.
but during the last five sessions of baseline, he Results for Chris are shown in the top panel
answered one question in Unknown Set 1 of Figure 2. No instances of IDKPTM were
correctly. When we implemented IDKPTM evident in the preintervention probes. Follow-