ASSESSING THE EFFICACY OF PICTORIAL PREFERENCEASSESSMENTS FOR CHILDREN WITH DEVELOPMENTAL

DISABILITIES

MEGAN R. HEINICKE

CALIFORNIA STATE UNIVERSITY, SACRAMENTO

JAMES E. CARR

BEHAVIOR ANALYST CERTIFICATION BOARD

SACHA T. PENCE

AUBURN UNIVERSITY

DANIKA R. ZIAS

CALIFORNIA STATE UNIVERSITY, SACRAMENTO

AMBER L. VALENTINO

TRUMPET BEHAVIORAL HEALTH

AND

JOHN M. FALLIGANT

AUBURN UNIVERSITY

Past research has demonstrated that pictorial preference assessments can predict subsequent rein-forcement effects for individuals with developmental disabilities only when access to the selectedstimulus is provided contingent on a pictorial selection. The purpose of the present investigationwas to assess more comprehensively the feasibility of the pictorial format with children withdevelopmental disabilities. In Experiment 1, prerequisite skill assessments were conducted, andthe role of a contingent reinforcer was assessed by comparing the results from the pictorialassessment without contingent access to a reinforcer assessment. If contingent access was foundto be necessary, the effects of schedule thinning were evaluated to determine whether a pictorialformat could be made more practical in Experiment 2. The pictorial format without contingentaccess was successful with only some participants. However, schedule thinning was found to bean effective method to establish conditioned reinforcement properties for pictorial stimuli to cre-ate a more practical assessment for a subset of participants.Key words: developmental disabilities, pictorial modality, preference assessment, prerequisite

skills, reinforcer assessment, schedule thinning

This article is based on a dissertation submitted by thefirst author, under the supervision of the second and thirdauthors, to the Psychology Department at Auburn Uni-versity in partial fulfillment for the requirements of a PhDdegree. We thank the research assistants who served asdata collectors: Rayven Arzola, Leslie Borromeo, ZamiraGarcia, Emilee Harbin, Anna Lee, Kim Magat, LynnNorwood, Cameron Mosely, and Kaitlyn Terry. We alsothank Christopher Newland and Martha Escobar for their

helpful comments as part of the dissertation committee.This research was supported in part by the Provost’sResearch Incentive Fund from California State University,Sacramento.

Address correspondence to Megan R. Heinicke,Department of Psychology, California State University,Sacramento, 6000 J Street, Sacramento, California 95819(e-mail: megan.heinicke@csus.edu).

doi: 10.1002/jaba.342

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2016, 49, 1–21 NUMBER 4 (WINTER)

1

Behavior analysts are often faced with the chal-lenge of identifying reinforcers for individualswith developmental disabilities. Reinforcers suchas preferred edible items, tangible items, andactivities are often used to increase the effective-ness of skill-acquisition (e.g., language training)and behavior-reduction (e.g., environmentalenrichment) programming for this population(Volkert, Lerman, Trosclair, Addison, & Kodak,2008; Watkins & Rapp, 2014). A commonmethod for identifying reinforcers is a stimuluspreference assessment (SPA), in which an indi-vidual selects a stimulus from an array of optionsoften nominated by caregivers (e.g., DeLeon &Iwata, 1996; Fisher et al., 1992; Roane, Vollmer,Ringdahl, & Marcus, 1998).Graff and Karsten (2012) conducted a survey

to determine how often behavior analysts useSPA methods with their clients with develop-mental disabilities. Nearly 90% of behavioranalysts reported using at least one method.However, over 80% of those same respondentsalso reported a lack of time as a major barrierto conducting SPAs regularly, and the majorityof behavior analysts reported conducting SPAsless than once per month. The time barrier isconcerning because there is evidence to suggestthat preferences are idiosyncratic and maychange over time; therefore, regular SPAsshould be conducted to capture changing pre-ferences and help bolster behavioral program-ming (Mason, McGee, Farmer-Dougan, &Risley, 1989; Zhou, Iwata, Goff, & Shore,2001). One potential solution for decreasingthe duration of SPAs is to conduct the assess-ment using an alternative format or modality.Alternative-modality SPAs involve presenting

pictures or verbal descriptions (e.g., “Wouldyou rather have X or Y?”) of putative reinforcersrather than presenting actual tangible items ona tabletop. Alternative modalities offer the ben-efit of a shorter SPA because clients are notgiven access to a stimulus following a selection.That is, after a client makes a stimulus selec-tion, the next pictorial or verbal choice is

presented. Alternative formats may also savebehavior analysts time in preparing SPAs, espe-cially in the case of the verbal modality inwhich one would only need access to a datasheet and a pencil (i.e., no pictorial or tangiblematerials would be required). In addition, pic-torial and verbal SPAs offer practitioners a for-mat to assess larger items (e.g., bean bag chair),activities (e.g., playing basketball), and pro-tracted events (e.g., going to a local park) thatare not often feasible to assess during a stand-ard, tangible SPA.Previous studies that have compared pictorial

and verbal SPAs to the tangible format suggestthat alternative-modality SPAs are successfulwith only some individuals with developmentaldisabilities. For example, Parsons, Harper, Jen-sen, and Reid (1997) conducted a comparisonof the pictorial and tangible formats for sevenolder adults with profound intellectual disabil-ities and found that the pictorial format wassuccessful in identifying reinforcing leisureitems for only two participants. These authorsdid note that participants had limited verbalcommunication abilities, and there were no for-mal assessments of participants’ skill levelsbefore the comparison was conducted.Researchers then began to investigate the roleof hypothesized prerequisite skills such as IQ(e.g., Cohen-Almeida, Graff, & Ahearn, 2000;Wilder, Ellsworth, White, & Schock, 2003),matching abilities (e.g., Clevenger & Graff,2005; Cohen-Almeida et al., 2000; Graff &Gibson, 2003), verbal repertoires (e.g., Higbee,Carr, & Harrison, 1999; Kuhn, DeLeon, Ter-longe, & Goysovich, 2006), and discriminationabilities (e.g., Conyers et al., 2002; de Vrieset al., 2005). Overall, multiple prerequisiteskills have been assessed across numerous inves-tigations; however, it is difficult to draw con-clusions on which skills are pertinent becausethey have not been assessed systematically.There is some evidence that matching skillsmay be necessary for the pictorial format to beeffective (e.g., Clevenger & Graff, 2005), but

MEGAN R. HEINICKE et al.2

there may be some skills that have not yet beenempirically investigated (e.g., pictorial man-ding; Higbee et al., 1999).The majority of the alternative-modality SPA

comparison investigations have included briefaccess to a stimulus following its selectionrather than immediately presenting the nextpictorial or verbal choice. Researchers likelyprovide contingent access to equate their com-parison conditions (e.g., tangible vs. pictorial);however, in doing so, they are evaluating a lesspractical form of the assessment. That is, pro-viding contingent access increases the durationof the assessment and no longer allows behavioranalysts the benefit of assessing other classes ofstimuli related to quality of life (e.g., protractedevents, residential and employment options). Itseems that the mixed results in this line ofalternative-modality comparison studies may bedue in part to the use of contingent reinforceraccess in some studies and not in others. Morerecently, researchers have directly compared theefficacy of pictorial and verbal SPAs with andwithout contingent access (e.g., Groskreutz &Graff, 2009; Hanley, Iwata, & Lindberg, 1999;Tessing, Napolitano, McAdam, DiCesare, &Axelrod, 2006), and these studies generally sug-gest that alternative-modality SPAs have partialpredictive validity only when they are usedwithout contingent stimulus access.Recently, step-by-step flowcharts have been

published to guide practitioners who serve indi-viduals with developmental disabilities in usingSPA methods (Karsten, Carr, & Lepper, 2011;Virués-Ortega et al., 2014). In one of thesemodels, Karsten et al. (2011) suggested thatalternative modalities may be useful when largetangible items, activities, and protracted eventsare evaluated, but these authors caution the useof alternative-modality SPAs without contin-gent access. Virués-Ortega et al. (2014) guidedpractitioners to use a paired-stimulus methodwith pictorial stimuli if clients are able tomatch pictorial and tangible stimuli reliablyand to make choices from an array of pictorial

stimuli. In their survey, Graff and Karsten(2012) found that behavior analysts report theuse of alternative-modality SPAs even thoughthey have only partial predictive validity whenused without contingent access for individualswith developmental disabilities.The purpose of the present investigation was

to assess the feasibility of pictorial SPAs withchildren with developmental disabilities. Thepictorial modality was examined because it hasbeen the most heavily researched alternativepresentation format, yet no firm conclusionscan be drawn on its efficacy with clinical popu-lations without contingent access to the rein-forcer. First, the role of contingent access tothe stimulus was assessed by comparing theresults of a pictorial SPA without contingentaccess to the results of a reinforcer assessment(RA). Matching and pictorial mand assessmentswere conducted before pictorial SPA and RAsessions to evaluate the role of hypothesizedprerequisite skills. Second, if contingent accessto the stimulus was found to be a necessarycomponent of a pictorial SPA, the effects ofschedule thinning were evaluated to determinewhether a pictorial SPA could be made morepractical for those participants.

EXPERIMENT 1

METHOD

ParticipantsParticipants included three girls and five boys

(2 to 11 years old; see Table 1 for a detailedsummary of participant characteristics). Keron,Eric, Lisa, Greg, and Mitch received school-based behavioral services during the course ofthe study, and Sean received services in a center-based behavioral program. Annah and Connorreceived services in a structured after-schoolsocialization program and had received earlyintensive behavioral intervention in the past,during which they were exposed to SPAs. How-ever, neither participant’s parents were able toreport which methods they had experienced.

3PICTORIAL PREFERENCE ASSESSMENTS

Preference assessments were not being con-ducted at Annah’s and Connor’s after-schoolprogram during the course of the study. Lisahad not been exposed to any formal SPA meth-ods, but she was given the choice between twoor three preferred stimuli multiple times per dayto be used in her programming. Keron was theonly participant who had been exposed toalternative-modality SPAs. All participants wereable to follow one-step instructions, fully partic-ipate in the assessments, and had minimal prob-lem behavior.

Setting and Session DurationAll sessions were conducted in a small treat-

ment room on a university campus or in asmall room or partitioned area of the partici-pant’s school. During all sessions, the experi-menter sat across from the participant at atable. All materials necessary for conductingthe sessions were placed near the experimenterand out of reach of the student. One or twotrained independent observers also were pres-ent during a subset of sessions for interobser-ver agreement and procedural integrity datacollection.

Sessions ranged from 1 min (an RA controlsession) to 17 min 45 s (a picture-to-objectprerequisite assessment session). Experimentalsessions were conducted 2 to 4 days per week,and participants’ total time commitment ran-ged from 2 to 4 weeks.

MaterialsOne caregiver and one staff member (who

had worked closely with the participant) wereasked to list and rank each participant’s favoritetoys and foods using the Reinforcer Assessmentfor Individuals with Severe Disabilities(RAISD) structured interview (Fisher, Piazza,Bowman, & Amari, 1996). The top eight edi-ble items ranked by the caregivers and staffwere used in all subsequent phases of the study.Caregivers also were asked to nominate onefood item that their child would be unlikely toconsume, which was used in RA control condi-tions throughout the study. The edible stimuliidentified in the RAISD, laminated colorphotographs (10.2 cm by 15.2 cm) of the edi-ble stimuli, a timer, a data sheet, and a penwere used in prerequisite assessments and SPAsessions. Additional materials for the RA

Table 1Participant Demographic Information

Age Diagnosis GARS Communication modality and skillsHistory

with SPAsHistory with alternative-

modality SPAs

Keron 5 years11 months

Noonansyndrome

74 Vocal using 4 to 5+ word sentences PairedMSWO

PictorialVerbal

Annah 9 years8 months

Autism,ADHD

83 Vocal using 4 to 5+ word sentences Yes No

Connor 11 years Autism 76 Vocal using 4 to 5+ word sentences Yes NoEric 4 years

9 monthsAutism 89 Vocal using 4 to 5+ word sentences MSWO No

Lisa 4 years8 months

Autism 98 Vocal using 1 to 2 word utterances;some icon exchange

Yes No

Sean 2 years8 months

Developmentaldelay

113 Vocal using 1 to 2 word utterances No No

Greg 5 years Autism 96 Vocal using 1 to 2 word utterances;some icon exchange

No No

Mitch 4 years3 months

Autism 113 Vocal using 2 to 4 word sentences PairedFree operant

No

Note. ADHD = attention deficit hyperactivity disorder; GARS = Gilliam Autism Rating Scale (Gilliam, 2006);MSWO = multiple stimulus without replacement; SPA = stimulus preference assessment.

MEGAN R. HEINICKE et al.4

sessions included paper clips, four containers,three pieces of different-colored paper(i.e., blue, red, yellow; 10.2 cm by 15.2 cm),clear adhesive tape, and preferred toys identi-fied via the RAISD.

Design and ProcedureFor all participants (with the exception of

Connor), RAs were evaluated using an alternat-ing treatments design (Barlow & Hayes, 1979)with the high-preference (HP) and low-preference (LP) items identified via a pictorialSPA. For Connor, RAs were evaluated using a

concurrent-schedules design (Piazza, Fisher,Hagopian, Bowman, & Toole, 1996). Theorder of preexperimental and experimental con-ditions is depicted in Figure 1.Prerequisite assessments. Before experimental

sessions, participants were exposed to the fol-lowing three prerequisite skill assessments: apicture-to-object (P-O) matching assessment,an object-to-picture (O-P) matching assess-ment, and a pictorial mand assessment. In theP-O assessment, the experimenter placed aphoto of one of the food items in front of theparticipant and instructed him or her to matchthat photo (e.g., “match”) to the correct food

Correspondence?

Yes No

Preexperimental Procedures

RAISD for Stimulus Pool

Prerequisite Assessments

Experi m

ent 2

Exp

erim

ent

1

Pictorial SPA Without Contingent Access

Reinforcer Assessment

Participation

Terminated

Pictorial SPA With Access

Reinforcer Assessment

Participation

Terminated

Correspondence?

Yes No

Schedule Thinning

3 VR 3 Pictorial SPAs

(With Clear and Consistent Gradients)

Brief RA

Repeat at Next Schedule Value

(VR 3 VR 5 EXT)

Repeat SPAs and Brief RA

Correspondence?

Yes No

Repeat at Next Schedule Value

(VR 3 VR 5 EXT)

Figure 1. Sequence of preexperimental and experimental conditions for Experiments 1 and 2.

5PICTORIAL PREFERENCE ASSESSMENTS

item in an array of three foods. Similarly, inthe O-P assessment, the experimenter placed afood item in front of the participant andinstructed him or her to match the food itemto the correct picture in an array of three pic-tures. Each of the eight photos or food itemswas presented twice in a random order, for atotal of 16 P-O and 16 O-P matching trials,and the stimuli in the array varied across trials.Data were collected on the participant’sresponse (correct, incorrect, or no response).During the P-O and O-P assessments, the

experimenter provided 30 s to 1 min of accessto a preferred toy within 3 s of a correctresponse. If the participant responded incor-rectly or did not respond after 5 s, the experi-menter provided a nonspecific statementredirecting the child to the next trial(e.g., “Let’s try the next one”). The percentageof correct trials was calculated for the P-O andO-P assessments by dividing the number ofcorrect trials by the total number of trials andconverting the ratio to a percentage.In an effort to contrive an establishing opera-

tion for food items used during pictorial mandassessments, sessions were conducted immedi-ately before lunch or dinner, and caregivers andstaff members were asked to restrict access tothe food items before the session on that day.In the pictorial mand assessment, the experi-menter held up one HP food item for approxi-mately 3 s in the participant’s line of visionand then placed the stimulus out of view(e.g., in an opaque bag, on the experimenter’slap under the table) to ensure the features of apure mand were present. The experimenterplaced the corresponding picture of that fooditem in front of the participant and put out herhand if the participant picked up the pictureand began to move the picture toward theexperimenter. If the participant handed the pic-ture to the experimenter, he or she receivedaccess to the item for 30 s. After 30 s, theexperimenter provided a nonspecific statementand presented the next trial. If the participant

did not respond within 10 s, the experimenterremoved the picture, provided a nonspecificstatement, and presented the next trial. If theparticipant engaged in a vocal mand, the exper-imenter did not provide any differential conse-quences or access to the requested stimulus.The experimenter collected data on vocalmands (data available from the first author),which rarely occurred. Each food item wasassessed twice in a random order, and data werecollected on the participant’s response (correct,incorrect [e.g., touched the picture but did nothand it to the experimenter], or no response)and vocal mands. The percentage of correcttrials was calculated for the mand assessmentby dividing the number of correct trials by thetotal number of trials and converting the ratioto a percentage.Pictorial SPA without contingent access. Before

the first SPA session, participants were given30 s of access to each of the eight edible stimuliin a random order. Thereafter, they wereexposed to a pictorial SPA using a paired-stimulus method (Fisher et al., 1992). In alltrials, the experimenter presented a pair of pic-ture stimuli in front of the participant (approxi-mately 25 cm apart) and asked the participantto “pick one.” After selection of one stimulus,the experimenter presented the next pair of pic-ture stimuli. If the participant did not respondduring a trial, the experimenter waited 10 sand then provided a verbal prompt (i.e., “pickone”) followed by an opportunity for the par-ticipant to make a selection. If a selection wasmade or 5 s elapsed without a response afterthe verbal prompt, the next trial was presented.Each stimulus was paired with every otherstimulus once in a random order, and the orderof left–right positions of the stimuli was coun-terbalanced for a total of 28 trials. Selectionpercentages were calculated, and these datawere graphed to represent the hierarchical rank-ings of the stimuli.Reinforcer assessment. The purpose of this

assessment was to evaluate whether the items

MEGAN R. HEINICKE et al.6

from the pictorial SPA without contingentaccess functioned as reinforcers (i.e., if the pic-torial SPA without access was a valid assess-ment). The RA format and progressive-ratio(PR) schedules were determined based on parti-cipants’ patterns of responding and taskrequirements. We used a single-operant RAwith a paper clip transfer task with an additivePR 3 schedule with Keron, Annah, Eric, Greg,and Mitch. An alternative task was selected forLisa and Sean due to the persistence ofresponding on the paper clip transfer task inthe absence of social consequences, suggestingpotential automatic reinforcement properties ofthe task. Finally, we used an alternative format(concurrent-operants RA) with Connor due toproblem behavior in the control condition ofhis initial single-operant RA.Single-operant PR RA. Items identified as HP

and LP (i.e., the items with the highest andlowest selection percentages) and a control itemwere delivered contingent on responding dur-ing an arbitrary task in a single-operant PRarrangement for all participants except Connor(Roane, Lerman, & Vorndran, 2001). Theexperimenter presented the arbitrary task to theparticipant at the beginning of each session.This included one open container of paper clipsapproximately 25 cm in front of the participantand one colored container approximately25 cm behind the container of paper clips. Theexperimenter modeled how to perform the taskby placing one paper clip from the open con-tainer into the colored container and used aleast-to-most prompting procedure during thetask. During the prompted trial, the experi-menter provided 30 s of access to the itemassociated with the colored container. The con-trol condition was included to ensure that thearbitrary task was not automatically reinforcing,and the item used in the control condition wasa nonpreferred item nominated by the partici-pants’ caregivers via the RAISD. After theprompted trial, the experimenter provided aninstruction to the participant to begin such as,

“If you would like [the item associated with thecondition], you will have to move these[pointed to the paper clips in the open con-tainer] from here to here [pointed to the emptycontainer]. You can stop anytime.”Sessions lasted 10 min, and the timing of the

session began immediately after the experimen-ter’s instruction. The order of the conditions(i.e., HP, LP, control) was quasirandomizedacross sessions. However, the color of the con-tainer paired with the HP, LP, or control itemremained consistent across sessions to aid in thediscrimination of contingencies. During all ses-sions, the HP, LP, and control items werebehind the colored container. When the firstcriterion was met (i.e., placing three paper clipsinto the colored container), the edible item wasdelivered for 30 s. Using an additive PR3 schedule, the ratio increased by threeresponses each time responding met theresponse criterion (e.g., 3, 6, 9).An alternative, arbitrary task was used for Lisa

and Sean due to high break points in the controlcondition. The alternative task included placinga piece of laminated colored paper (10.2 cm by15.2 cm) on a wall and requiring the participantto touch the paper with his or her hand. ForLisa, the paper was placed approximately 5 cmabove where her fingertips touched the wall,requiring her to jump slightly off the ground tomake contact with the paper. For Sean, thepaper was placed at a height where he couldtouch the paper with his fingertips while hestood flat on the ground because of a deficit inhis jumping skills. All other procedures wereconsistent with the paper clip transfer task,except that an additive PR 2 schedule was used.The experimenter sat across from the partici-

pant during all RA sessions, delivered the itemdesignated for the condition contingent on theparticipant meeting the response criterion,blocked the task materials during the 30-s rein-forcement interval, and then re-presented thetask materials. The session timer was notstopped during reinforcer access. Break points

7PICTORIAL PREFERENCE ASSESSMENTS

were recorded for each stimulus evaluated inthe RA. Sessions were terminated either when10 min elapsed or 1 min elapsed with noresponding (i.e., a break point).Concurrent-operants RA. Items (HP, LP, con-

trol) were delivered contingent on respondingduring the same arbitrary task described abovein a concurrent-operants arrangement for Con-nor. The procedures used in the concurrent-operants RA were identical to the single-operantRA, with the following exceptions. First, theexperimenter presented three colored containersbehind the open container of paper clipsapproximately 25 cm apart from one another.Second, when the participant placed a paper clipin one of the colored containers, the experi-menter delivered the item associated with thatcontainer on a continuous schedule (i.e., no PRschedule was used). Finally, sessions lasted5 min, and we recorded the rate of respondingfor each stimulus evaluated in the RA.

Interobserver AgreementWe compared data from a second trained

observer, collected either during experimentalsessions or from videotape, to the data collectedby the experimenter to calculate trial-by-trialinterobserver agreement. For prerequisite assess-ments, an agreement was defined as both obser-vers recording a correct, incorrect, or noresponse for a trial. For SPA sessions, an agree-ment was defined as both observers recordingthe same stimulus selection for a trial. For RAsessions, an agreement was defined as bothobservers recording the same break point persession in the single-operant PR arrangementor recording the same number of responses ineach container in the concurrent-operantsarrangement. All agreement scores were aver-aged and reported per participant per sessiontype. Agreement was calculated for 97% of ses-sions (range, 81% to 100%), and the averagescore for all participants across session typeswas 99.5% (range, 98% to 100%).

Procedural IntegrityA trained observer collected data to deter-

mine whether the procedures were implemen-ted correctly. Procedural integrity scores werecalculated as the percentage of correct responsesmade by the experimenter using a checklist ofexperimenter behavior specific to each type ofexperimental session. In addition, interobserveragreement was assessed for the proceduralintegrity data. Data were compared using thetrial-by-trial agreement formula. An agreementwas defined as both independent observersscoring an experimenter’s responses during atrial as either correct or incorrect. All integrityscores were averaged and reported per partici-pant per session type. Integrity scores were cal-culated for 97% of sessions (range, 81% to100%), and the average integrity score for allparticipants across session types was 99.7%(range, 99% to 100%). Mean procedural integ-rity interobserver agreement scores fell withinthe acceptable range for the discipline and areavailable from the first author, along with thechecklists of experimenter behavior mentionedabove.

RESULTS AND DISCUSSION

Figure 2 shows high levels of correspondencebetween the SPAs without contingent accessand subsequent RAs for Keron, Annah, andConnor. Keron’s pictorial SPA without contin-gent access (top left) produced a clear gradient,with selection percentages ranging from 0% to100%. When Keron’s HP, LP, and controlitems were assessed in a subsequent single-operant PR RA (bottom left), break points wereconsistently higher for her HP item (M = 11)than for the LP (M = 4.5) and control items(M = 3.5), confirming the predictions of thepictorial SPA without contingent access.Annah’s pictorial SPA without access

(Figure 2, top middle) produced a clear gradi-ent with selection percentages of 86% for hertwo HP items. Two LP items were identified

MEGAN R. HEINICKE et al.8

Figure

2.Results

ofKeron

’s,Ann

ah’s,

and

Conner’s

pictorialpaired-stim

ulus

preference

assessments

with

outcontingent

access

(top)and

single-operant

progressive-ratio

(Keron

andAnn

ah)andconcurrent-operants(Connor)reinforcer

assessments(bottom).

9PICTORIAL PREFERENCE ASSESSMENTS

with selection percentages of 14%. Thus, theexperimenters randomly chose one HP and LPitem to be used in the subsequent RA, whichare depicted as the first and last stimuli in allSPA bar graphs. The control (celery), HP, andLP items were evaluated in a subsequent

single-operant PR RA. Break points werehigher for Annah’s HP item (M = 13.3) thanfor the LP (M = 5.6) and control (M = 4.3)items (Figure 2, bottom).Connor’s pictorial SPA without contingent

access (Figure 2, top right) produced a gradient

Figure 3. Results of pictorial paired-stimulus preference assessments without and with contingent access for Eric,Sean, and Greg (bar graphs). Results of progressive-ratio reinforcer assessments following pictorial paired-stimulus prefer-ence assessments without access (EXT) and with access (FR 1) and during schedule thinning for Eric, Sean, and Greg(line graphs).

MEGAN R. HEINICKE et al.10

with selection percentages ranging from 14%to 71%. During his subsequent concurrent-operants RA (bottom right), rate of respondingwas much higher for his HP item (M = 2.4responses per minute) than for the LP(M = 0.1 responses per minute) and controlitems (grape tomatoes; M = 0.03 responses perminute).Figures 3 and 4 show results for the remain-

ing five participants who had low correspond-ence between the results of their SPAs withoutcontingent access (displayed as black bars in theleft panel) and subsequent RAs (displayed inthe first “No Access [EXT]” phase of the linegraphs in the right panel). Eric’s pictorial SPAwithout contingent access (Figure 3, top left)produced a moderate gradient, with selection

percentages ranging from 29% to 86%. DuringEric’s single-operant PR RA (Figure 3, topright), break points were higher for his LP item(M = 12) than for the HP (M = 9) and controlitems (M = 1.5), suggesting that the pictorialSPA without contingent access did not producevalid results. The results of Sean’s pictorial SPAwithout access (Figure 3, middle left) were alsonot validated by his subsequent RA (middleright). Sean had higher break points for his LPitem (M = 3.5) than for his HP (M = 1) andcontrol (M = 1.5) items. Greg’s pictorial SPAwithout contingent access (Figure 3, bottomleft) produced a moderate gradient, with selec-tion percentages ranging from 29% to 86%.During his single-operant PR RA (Figure 3,bottom right), breakpoints were higher for his

Figure 4. Results of pictorial paired-stimulus preference assessments without and with contingent access for Lisaand Mitch (bar graphs). Results of progressive-ratio reinforcer assessments following pictorial paired-stimulus preferenceassessments without access (EXT) and with access (FR 1) and during schedule thinning for Lisa and Mitch (linegraphs).

11PICTORIAL PREFERENCE ASSESSMENTS

LP item (M = 3) than for the HP (M = 1.8)and control (M = 0) items, suggesting that thepredictions of the pictorial SPA without contin-gent access were not valid.Lisa’s pictorial SPA without contingent access

(Figure 4, top left) produced a slight gradient,with selection percentages ranging from 43% to71%. A clear separation between break pointswas not found for her HP (M = 5) and LP(M = 5) items during her initial RA (top right).However, these break points were higher thanin the control condition (M = 1), suggestingthat the HP and LP items likely had similarreinforcing properties. Mitch’s pictorial SPAwithout contingent access (Figure 4, bottom)produced a slight gradient, with selection per-centages ranging from 29% to 57%. There wasno clear separation among the HP (M = 4.7),LP (M = 2.1), and control (M = 1.3) data pathsduring Mitch’s subsequent RA (bottom right),suggesting that the predictions of the pictorialSPA without contingent access were not valid.The results of pictorial SPAs without access

were validated by subsequent RAs for Keron,Annah, and Connor. In addition, Keron andConnor scored at 100% for all three prerequi-site assessments (see Table 2 for summaries ofall participants’ prerequisite assessments scores).

Annah scored at or near 100% during themand and O-P matching assessments, but shescored only 6% on the P-O matching assess-ment. During her P-O assessment, Annah con-sistently engaged in pictorial mands by handingthe sample picture to the experimenter ratherthan matching the picture to the correct objectin the array. Furthermore, both her caregiverand former case manager reported that she hadmastered P-O matching. For these reasons, it islikely that the low score on her P-O assessmentwas due to faulty stimulus control during theassessment rather than a true deficit in her P-Omatching repertoire.Some participants with low correspondence

between their SPAs without access and RAshad different results on their prerequisite assess-ments (see Table 2), suggesting that certainprerequisite skills (i.e., pictorial mands, O-P/P-O matching) may be correlated with the suc-cess of pictorial SPAs without access. Lisascored 100% on her mand assessment, 0% onher P-O matching assessment, and slightlyabove chance levels (i.e., 44%) on her O-Pmatching assessment. Similarly, Greg scored100% on his mand assessment and slightlyabove chance levels on his P-O and O-Pmatching assessments (i.e., 44% and 56%,

Table 2Summary of Results for Experiments 1 and 2

% of correct responding inprerequisite skills assessments Correspondence between reinforcer assessments and SPA types

P-Omatching

O-Pmatching

Pictorialmand

Withoutcontingentaccess SPA

With (FR 1)contingentaccess SPA

VR3 contingentaccess SPAs

VR5 contingentaccess SPAs

ExtinctionSPAs

Keron 100 100 100 YesAnnah 6 81 100 YesConnor 100 100 100 YesEric 100 100 100 No Yes Yesa Yes YesLisa 0 44 100 No Yes Yes Yes YesSean 87.5 87.5 0 No Yes Yesb

Greg 44 56 100 No Yes Yes Yes YesMitch 93.8 93.8 100 No Yes Yes Yes Yes

Note. P-O = picture to object; O-P = object to picture; SPA = stimulus preference assessment.a Correspondence was observed for Eric only following two sets of VR 3 SPAs.b Correspondence was observed for Sean only following a set of denser (VR 2) SPAs.

MEGAN R. HEINICKE et al.12

respectively). Sean scored 0% on his mandassessment and 87.5% for both of his matchingassessments.The alternative-modality SPA literature indi-

cates that pictorial SPAs without contingentaccess are not successful with some individualswith developmental disabilities (e.g., Hanleyet al., 1999; Higbee et al., 1999), and theresults for five of our participants were consist-ent with these previous findings. Nevertheless,the addition of contingent access to the rein-forcer in alternative-modality SPAs creates a lesspractical form of the assessment. AlthoughGroskreutz and Graff (2009) suggested thatresearchers should examine intermittent sche-dules of contingent access (e.g., deliveringaccess to stimuli on an interval or ratio sched-ule) to maintain both meaningful selectionsand the benefits of an alternative modality forsuch individuals, to date, no empirical investi-gations of these procedural manipulations havebeen conducted. Therefore, the purpose ofExperiment 2 was to investigate the role of con-tingent access to the reinforcer for participantswho did not have valid results from a pictorialSPA without access when compared to a subse-quent RA. For this subset of participants, theeffects of schedule thinning were evaluated todetermine if a pictorial SPA could be mademore practical. That is, access to the stimulidepicted on the pictures was delivered on aschedule that was thinned to extinction(i.e., no contingent access).

EXPERIMENT 2

METHOD

Participants, Setting, Materials, and SessionDurationEric, Lisa, Sean, Greg, and Mitch partici-

pated in Experiment 2. Session arrangementsand materials were consistent with Experiment1. Sessions ranged from 1 min (RA control) to16 min 22 s (SPA with contingent access). Ses-sions were conducted 2 to 4 days per week,

and each participant’s total time commitmentranged from 4 to 8 weeks.

Design and ProcedureReinforcer assessments were evaluated using

an alternating treatments design (Barlow &Hayes, 1979). To demonstrate experimentalcontrol over the effects of schedule thinning ondiscriminated responding in RA sessions, weused a nonconcurrent multiple baseline designacross participants. The order of preexperimen-tal and experimental conditions is depicted inFigure 1.Pictorial SPA with contingent access and RA.

The trials for this assessment were conductedin the same manner as described for the picto-rial SPA without contingent access, except thatthe participant was given the opportunity toconsume a small amount of the edible stimulusafter each selection. A single-operant PR RAwas conducted after the pictorial SPA with con-tingent access using the HP, LP, and controlitems. These sessions were conducted in thesame manner as described in Experiment 1.Schedule thinning. These schedule manipula-

tions were conducted with participants whoshowed high levels of correspondence betweenthe results of the pictorial SPA with contingentaccess and the RA to evaluate whether condi-tioned reinforcement properties could be estab-lished for the pictorial stimuli to create a morepractical pictorial SPA (i.e., to eliminate theneed for contingent access). During schedulethinning, access to the stimuli depicted on thepictures was delivered on a schedule that wasthinned to extinction (i.e., no contingentaccess). The results of the previous RAs usingthe HP and LP items from the SPA withoutcontingent access served as the baseline data foreach participant.First, three pictorial SPAs were conducted,

with contingent access provided on a variable-ratio (VR) 3 schedule. If a clear gradient wasachieved and the gradients were consistent

13PICTORIAL PREFERENCE ASSESSMENTS

across the three SPAs, three RA sessions (i.e., abrief RA) that evaluated the HP and LP itemsfrom the VR 3 schedule SPAs and the controlitem were conducted in a quasirandom order.A clear gradient was defined as each SPA pro-ducing selection percentages ranging from atleast 14% to 86%. A consistent gradient wasdefined as all three SPAs identifying the samestimulus as the HP item (i.e., having the high-est or second highest selection percentage) andthe LP item (i.e., having the lowest or secondlowest selection percentage). If the participant’sbreak points were higher for the HP item thanfor the LP and control in the brief RA (i.e., theRA validated the prior SPA’s prediction), threepictorial SPAs, using a VR 5 schedule of rein-forcement, were conducted.Schedule thinning progressed (i.e., VR

3, VR 5, extinction) until no access was pro-vided in a 28-trial SPA session. If clear gradi-ents or consistent results across the SPAs werenot obtained, the three SPAs were repeatedusing the same schedule value. If preferencegradients remained unclear or results continuedto be inconsistent, participation was either ter-minated or the experimenters implementedschedule thinning at a denser schedule value.In addition, if the RA did not validate the SPAfor any of the schedule values, the experimenterconducted an additional three SPAs at the sameschedule value with subsequent RA sessions.No more than one schedule-thinning SPA

occurred each day, and all SPAs at each sched-ule value were conducted within the sameweek. The brief RA after schedule-thinningSPAs was conducted within the same week asthe SPAs, and each RA session within the briefRA was separated by a 5- to 10-min play break.

Interobserver Agreement and ProceduralIntegrityData on interobserver agreement and proce-

dural integrity were collected in the same man-ner as described above in Experiment 1. All

interobserver agreement and procedural integ-rity scores were averaged and reported per par-ticipant per session type. Interobserveragreement was calculated for 100% of sessions,and for all participants across session types itwas 100%. Procedural integrity was calculatedfor 99.8% of sessions (range, 97% to 100%),and the average score for all participants acrosssession types was 100%.

RESULTS AND DISCUSSION

The duration of pictorial SPA sessions with-out contingent access were shorter (M = 6 min52 s) than those that included contingentaccess (M = 13 min 59 s) for all participants.The results of all participants’ SPAs with con-tingent (FR 1) access are displayed as whitebars in Figures 3 and 4 (left), and subsequentRA data are displayed in the second “WithAccess (FR 1)” phase of the line graphs inFigures 3 and 4 (right). Greg’s pictorial SPAwith contingent access (Figure 3, bottom left)produced a clear gradient, with selection per-centages ranging from 0% to 100%. Breakpoints were higher for Greg’s HP item (M = 6)during his second RA than for the LP (M = 2)and control (M = 0) items (bottom right), con-firming the predictions of the pictorial SPAwith contingent access. We then conducted aschedule-thinning manipulation. Due to spaceconstraints, we describe and display allschedule-thinning data (including bar graphs ofpictorial SPAs at each schedule value of themanipulation) for Greg as a representative par-ticipant and display the PR RA data after picto-rial SPAs without access (extinction), withaccess (FR 1), and during schedule thinning forthe remaining participants.Greg had consistent results across his first

three VR 3 pictorial SPAs (Figure 5, top left),and each pictorial SPA produced a clear gradi-ent. The first pictorial VR 3 SPA identified theHP item with a selection percentage of 100%,and the second and third pictorial SPA

MEGAN R. HEINICKE et al.14

identified this same item as having the secondhighest selection percentage (i.e., 86%). Allthree VR 3 pictorial SPAs identified the sameLP item (selected on 0% of opportunities). Greghad a higher break point for his HP item thanfor his LP and control items in his subsequentbrief RA (see “Schedule Thinning” phase inFigure 3, bottom right), validating the results ofhis VR 3 pictorial SPAs. When the schedule ofreinforcement was thinned to VR 5 and then toextinction (Figure 5, bottom left and top right),results were consistent; the same two items wereidentified as HP and LP. In addition, Greg hadhigher break points for his HP item than for hisLP and control items when they were assessed

in brief RAs that followed his VR 5 and extinc-tion manipulations (Figure 3, bottom right).The SPA without contingent access(i.e., extinction) and brief RA results remainedconsistent when they were tested 2 weeks afterhis last schedule-thinning sessions (Figure 5,bottom right, and Figure 3, bottom). Theseresults suggest that systematic thinning of thereinforcement schedule across pictorial SPAswas a successful procedure to establish condi-tioned reinforcement properties of the pictorialstimuli for Greg.Eric’s pictorial SPA with contingent access

produced a clear gradient, with selection per-centages ranging from 0% to 86% (Figure 3,

Figure 5. Results of Greg’s schedule-thinning manipulation. Each bar graph depicts selection percentages from pic-torial preference assessments conducted on VR 3 (top left), VR 5 (bottom left), and EXT (top right) schedules. The bargraph in the bottom right panel depicts selection percentages conducted without contingent access (EXT) at 2-weekmaintenance sessions.

15PICTORIAL PREFERENCE ASSESSMENTS

top left), and the break points were higher forhis HP item (M = 4) during his RA (Figure 3,top right) after the pictorial SPAs with contin-gent access than for the LP (M = 1) and con-trol (M = 2) items. Eric had highly consistentresults across his first three VR 3 pictorialSPAs, and each pictorial SPA produced a cleargradient. However, he had the same high breakpoint for his HP and control items as he didfor his LP item when these items were assessedin a subsequent brief RA. Thus, the VR 3 picto-rial SPAs were repeated. Results were consistentduring his second set of three VR 3 pictorialSPAs, and a new HP item was identified. Erichad a higher break point for his HP item thanfor his LP and control items in his subsequentbrief RA, validating the results of his second setof VR 3 pictorial SPAs. When the reinforce-ment schedule was thinned to VR 5 and thento extinction, results were consistent, and hehad higher break points for his HP item thanfor his LP and control items during thebrief RAs.Sean’s RA and schedule-thinning data are

depicted in Figure 3 (left and right middle,respectively). The assessment with contingentaccess produced a clearer gradient and identi-fied new HP and LP items. His RA showed aclear separation between data paths, withhigher break points in the HP condition(M = 3.7) than in the LP (M = 1.3) and con-trol (M = 0) conditions. These results sug-gested that the pictorial modality was successfulwhen contingent access was implemented.Unlike the other participants, Sean did not

have consistent results across his VR 3 pictorialSPAs after schedule thinning. Despite the threeadditional pictorial SPAs with the same itemsand the same schedule value, Sean continued tohave inconsistent results. He did have success-ful VR 3 pictorial SPA results after exposure toa denser (i.e., VR 2) schedule; however, wewere unable to replicate these findings when weused a VR 5 schedule. Due to these inconsis-tencies, Sean’s participation was terminated.

These data suggest that the VR 5 schedulevalue was too lean to produce valid SPAoutcomes.Lisa’s SPA with contingent access produced

a clearer gradient, with selection percentagesranging from 14% to 100%, and separationbetween break points for her HP item(M = 5.3) and her LP (M = 0.7) and control(M = 0.7) items in her RA (Figure 4, topright). Lisa had consistent results across herthree pictorial SPAs conducted with VR 3 andVR 5 contingent access and with no access(i.e., extinction) during schedule thinning.When these items were assessed along with hercontrol item in subsequent brief RAs, she had ahigher break point for the HP item than forthe LP and control items. These results suggestthat schedule thinning was a successful proce-dure for Lisa.Mitch’s SPA with contingent access pro-

duced a clear hierarchy, with selection percen-tages ranging from 14% to 100%, and thisSPA identified new HP and LP items. In a sub-sequent RA (Figure 4, bottom right), breakpoints were higher for the HP item (M = 8)than for the LP (M = 0) and control (M = 0)items, suggesting that contingent access wasnecessary for a pictorial SPA to be valid. Dur-ing schedule-thinning manipulations, Mitchhad consistent results across his three pictorialSPAs conducted with VR 3 contingent access.However, the results from his first set of VR5 pictorial SPAs did not identify a consistentLP item. His results were consistent during hissecond set of VR 5 pictorial SPAs and whenthe pictorial SPAs were conducted with noaccess (i.e., extinction). During subsequentbrief RAs, he had consistently higher breakpoints for the HP items, suggesting that sched-ule thinning was successful in creating a validpictorial SPA for Mitch.A nonconcurrent multiple baseline design

across participants was used to demonstrateexperimental control over the effects of sched-ule thinning on discriminated responding in

MEGAN R. HEINICKE et al.16

RA sessions (Figures 3 and 4). In baseline, lowlevels of correspondence were observed betweenthe results of the pictorial SPAs without accessand the subsequent RAs for all participants.That is, there was no clear separation of theHP data paths from the LP and control datapaths, suggesting that the results of the pictorialSPAs without access were not valid. However,pictorial SPAs became valid when contingentaccess was provided after stimulus selections onan FR 1 schedule for all participants. Whenschedule thinning was introduced, high corre-spondence continued to be observed for Eric,Greg, Lisa, and Mitch, and this high corre-spondence remained stable as the schedule ofreinforcement was thinned to a VR 5 scheduleand then to extinction (i.e., no access). Rein-forcer assessments were not conducted after theinitial VR 3 pictorial SPAs for Sean because ofinconsistent results across his three assessments.However, the same high correspondence thatwas observed for the four other participantsduring schedule thinning was also observedwhen a denser schedule value was implemented(Figure 3, middle). Sean continued to havehigh correspondence when the schedule wasthinned to a VR 3 value; however, RAs werenot conducted after the VR 5 SPAs because ofinconsistent results across his pictorial SPAs.The results of Experiment 2 demonstrate

that schedule thinning was an effective methodto establish conditioned reinforcement proper-ties for pictorial stimuli for participants whodid not have correspondence between pictorialSPAs without access and subsequent RAs.Beginning schedule thinning using a VR3 schedule of contingent access was effectivefor Lisa, Greg, and Mitch. Eric required a sec-ond VR 3 manipulation due to high respond-ing in his first RA control condition. Mitchrequired a second VR 5 manipulation due toinconsistencies in identifying an LP item, andSean required a denser schedule of reinforce-ment at the beginning of his schedule-thinningmanipulation. Sean’s need for a denser schedule

may have been due to deficits observed in hispictorial mand repertoire. Although he had agrowing vocal mand repertoire during thecourse of the study, he did not have a historywith using pictorial mands. Overall, this experi-ment suggests that schedule thinning may be auseful procedure to make pictorial SPAs morepractical for participants who require contin-gent access to make meaningful pictorialselections.

GENERAL DISCUSSION

The research on alternative-modality SPAsindicates that alternative SPA formats have onlypartial predictive validity when they are admi-nistered without contingent access(e.g., Hanley et al., 1999; Higbee et al., 1999).The results of Experiment 1 demonstrate thatcertain prerequisite skills, such as pictorialmands and O-P/P-O matching, might be corre-lated with the success of the pictorial modalitywhen contingent access is not provided. Twoof the eight participants had high scores on theprerequisite skills and consistent outcomes forthe pictorial SPA without contingent accessand RA. In addition, three participants had lowscores on the prerequisite skills, and the resultsof their pictorial SPAs without contingentaccess and RAs were inconsistent. Thus, theresults for five of eight participants show corre-lations between their prerequisite skills and thevalidity of the pictorial SPA without contingentaccess.Annah was an exception in Experiment

1, with a low score of 6% on her P-O match-ing assessment, although the RA confirmed theaccuracy of her SPA without contingent access.However, there is anecdotal evidence to showthat these low scores were due to faulty stimu-lus control rather than a true P-O matchingdeficit. Also, despite high prerequisite assess-ment results, Eric’s and Mitch’s pictorial SPAswithout access were not successful in identify-ing reinforcers. One explanation for Eric’s and

17PICTORIAL PREFERENCE ASSESSMENTS

Mitch’s differing results could be that we useda brief initial assessment to determine if partici-pants had basic exchange-based communicativeskills, similar to what would be observed inearly mand training. However, we did not eval-uate discriminated responding to ensure thatthe participant selected the picture of the HPitem when multiple icons were presented in anarray. It may be that a discrimination betweennonpreferred and preferred icons(e.g., Boelter & Hagopian, 2011) would bemore predictive of the effectiveness of pictorialSPAs. Future researchers might consider gather-ing more information on participants’ priorexposure to picture exchange communicationsystems and directly evaluating different skillsrelated to pictorial mands to help inform neces-sary prerequisite skills.It could also be the case that conducting the

pictorial SPA under extinction (i.e., withoutcontingent access) was not sufficient to identifyreinforcers for Eric and Mitch. That is, theseparticipants may have made meaningful selec-tions during the initial trials of the SPA with-out access, but indiscriminate responding mayhave occurred over successive trials of no access.Researchers interested in the topic of contin-gent access in alternative-modality SPAs mightconsider attending to this concern via a within-session analysis of selection percentages duringthe initial and final trials of an SPA withoutaccess compared to a pictorial SPA with access.Such an analysis may help to determine if inva-lid results of SPAs without access are related toprerequisite skill deficits or possible extinction-induced variability in responding.Another explanation for the results for the

SPA without access between the participants inExperiments 1 and 2 is that the learning his-tories may have differed between these two setsof participants. For example, participants whohad valid results for the SPA without access inExperiment 1 were older and may have had alonger history of engagement in selectionresponses under lean reinforcement schedules.

However, through schedule thinning, theresults of the SPAs became valid for the partici-pants in Experiment 2. Lisa and Greg showedthe most success with schedule thinning, onlyrequiring one set of SPAs at each schedulevalue (i.e., VR 3, VR 5) before they mademeaningful choices in their pictorial SPAs con-ducted without access (i.e., extinction). Mitchrequired a second set of pictorial SPAs at theVR 5 schedule value because a clear LP itemwas not consistently identified during his firstset of VR 5 pictorial SPAs. Eric required a sec-ond set of pictorial SPAs at the VR 3 schedulevalue due to a high control break point in hisfirst brief RA. The similar HP and controlbreak points during his first schedule-thinningRA may have been due to automatic reinforce-ment properties associated with the task. How-ever, this explanation seems unlikely becauseresponding during subsequent RA control con-ditions was low. Anecdotally, he vocallymanded for “blue” (i.e., the HP condition)during the first control condition, and hemanded for “chocolate” (i.e., the Hershey’sKisses) during the first HP condition whengummy candies were being delivered. He alsooccasionally consumed the control item duringthe RA sessions but pushed the LP item out ofhis reach and never consumed it. This anecdo-tal evidence suggests that the LP item may havebeen less preferred than the nonpreferred itemnominated via the RAISD. Therefore, it seemslikely that the results of Eric’s first schedule-thinning RA were accurate in invalidating theresults of his first set of VR 3 SPAs.Sean did not have initial success with sched-

ule thinning at the VR 3 schedule value; how-ever, he did have consistent results when weimplemented the pictorial SPA using a denser(i.e., VR 2) schedule. In addition, he did nothave consistent results when we implemented aleaner schedule (i.e., VR 5). One possibleexplanation for these inconsistent VR 3 andVR 5 pictorial SPA results is that his prefer-ences changed over the course of the repeated

MEGAN R. HEINICKE et al.18

assessments. This possibility is unlikely becausehe did have consistent results when the amountof access was increased, and he did have highbreak points during his subsequent RA. It ismore likely that that the inconsistent results forthe VR 3 pictorial SPA were due to the noveltyof using pictures to request items. That is,Sean’s therapists verbally reported that he didnot use pictorial mands, and his 0% score onhis pictorial mand prerequisite assessment sup-ports this report.Overall, schedule thinning was successful for

establishing conditioned reinforcement proper-ties of the pictorial stimuli assessed in the SPAs.One limitation of these schedule-thinning pro-cedures was that we conducted a brief RA aftereach schedule value rather than an extendedRA. Another limitation of our schedule-thinning procedures was that we did not collectfollow-up data to assess the maintenance ofeffects for the participants other than Greg.Researchers interested in extending the line ofresearch on schedule thinning for alternative-modality SPAs should consider conducting anadditional SPA without access at week-longintervals after the end of schedule thinningboth with the original stimulus set and with anovel stimulus set to assess maintenance andgeneralization.Some additional limitations of the investiga-

tion should be mentioned. First, we assessededible reinforcers in this study. Edible reinfor-cers can be quite powerful in motivating behav-ior, but practicing behavior analysts also need toassess the reinforcing effects of other classes ofstimuli, such as toys and activities, to use inbehavioral programming. Alternative-modalitySPAs may also be useful in identifying otherimportant preferences related to quality of life(e.g., Whitehouse, Vollmer, & Colbert, 2014).Future studies should consider evaluating theuse of alternative-modality SPAs for otherclasses of stimuli because most stimuli in otherclasses are difficult or impossible to present dur-ing a traditional tangible SPA (e.g., riding a

bike, going to the movies, vocationalpreferences).Second, we chose to evaluate schedule thin-

ning for participants who scored poorly ontheir prerequisite skill assessments and did nothave valid results for the SPA without access.Although Lohrmann-O’Rourke and Browder(1998) made the recommendation to teach pre-requisite skills that may be necessary to performsuccessfully in SPAs with nontangible stimuli,to date only one investigation has focused onthis potential solution (Browder, Cooper, &Lim, 1998). Future research could evaluate therole of hypothesized prerequisite skills by con-ducting remediation training to examine theeffects of prerequisite skill acquisition on thesuccess of a pictorial presentation format. Ourfindings in Experiment 2, especially for Ericand Mitch, call into question whether therepertoires assessed in the current investigationare meaningful prerequisites for pictorial SPAswithout contingent access to the reinforcer. Itmay be the case that the prerequisite skillsneeded for alternative-modality SPAs willremain speculative until a more rigorous assess-ment (i.e., remediation training) is conductedto assess failures of prerequisite assessments.Third, we chose to focus solely on the pic-

torial modality in the current investigation.That is, we did not include comparisons ofthe other alternative SPA formats (i.e., verbal,textual, video). The pictorial modality wasselected because it has been the most heavilyresearched alternative presentation format.Researchers are encouraged to replicate theprocedures of this study with other alternativemodalities. It may be useful to focus on theverbal modality because it has the quickestadministration time.In addition, some limitations of our RA pro-

cedures should be mentioned. First, we used anarbitrary task in our RAs rather than evaluatingthe reinforcing effects of the HP and LP stimuliin the participants’ ongoing behavioral pro-gramming. Second, we identified the

19PICTORIAL PREFERENCE ASSESSMENTS

nonpreferred item used in the RA control con-ditions via informant assessment rather thanusing a direct observation measure. Eventhough the RA data for each participant sup-port the function of the control stimulus asnonpreferred (i.e., break points during the con-trol conditions were consistently low), research-ers interested in continuing this line of researchusing a similar RA arrangement should consideridentifying nonpreferred items via a single-stimulus preference assessment. Third, wedecreased the duration of participation consid-erably with the concurrent-operants arrange-ment for Connor’s RA compared to that of thethe other participants, suggesting that thisarrangement could be advantageous in futurestudies. However, a single-operant PR arrange-ment does allow therapists to obtain informa-tion about absolute reinforcement effects(Roscoe, Iwata, & Kahng, 1999). Future stud-ies should directly compare the results of thesedifferent RA arrangements.The results of this study have implications

for behavior analysts who work with individualswith developmental disabilities. We found thatpictorial SPAs were successful for less than halfof the participants who completed the protocolwhen these assessments were conducted with-out contingent access. We also found that pro-viding contingent access to the participantswho did not have success with the initial picto-rial SPA created an effective SPA. However,this solution is not practitioner-friendly becausecontingent access after item selection eradicatesthe main benefits of alternative modalities.That is, alternative modalities are most likelydictated by the need to assess preferences forthe multitude of reinforcers that cannot beassessed in a tangible tabletop arrangement andthe need to decrease preparation and adminis-tration time of SPAs. However, schedule thin-ning may be a viable solution for practitionersto maintain both meaningful selections and thebenefits of a pictorial format for clients whorequire contingent access.

REFERENCES

Barlow, D. H., & Hayes, S. C. (1979). Alternating treat-ments design: One strategy for comparing the effectsof two treatments in a single subject. Journal ofApplied Behavior Analysis, 12, 199–210. doi:10.1901/jaba.1979.12-199

Boelter, E. W., & Hagopian, L. P. (2011). Effects of pref-erence on verification of discriminated mands. Journalof Applied Behavior Analysis, 44, 931–935. doi:10.1901/jaba.2011.44-931

Browder, D. M., Cooper, K. J., & Lim, L. (1998).Teaching adults with severe disabilities to expresstheir choice of settings for leisure activities. Educationand Training in Mental Retardation and Developmen-tal Disabilities, 33, 228–238.

Clevenger, T. M., & Graff, R. B. (2005). Assessingobject-to-picture and picture-to-object matching asprerequisite skills for pictorial preference assessments.Journal of Applied Behavior Analysis, 38, 543–547.doi: 10.1901/jaba.2005.161-04

Cohen-Almeida, D., Graff, R. B., & Ahearn, W. H.(2000). A comparison of verbal and tangible stimuluspreference assessments. Journal of Applied BehaviorAnalysis, 33, 329–334. doi: 10.1901/jaba.2000.33-329

Conyers, C., Doole, A., Vause, T., Harapiak, S.,Yu, D. C. T., & Martin, G. L. (2002). Predictingthe relative efficacy of three presentation methods forassessing preferences of persons with developmentaldisabilities. Journal of Applied Behavior Analysis, 35,49–58. doi: 10.1901/jaba.2002.35-49

de Vries, C., Yu, C. T., Sakko, G., Wirth, K. M.,Walters, K. L., Marion, C., & Martin, G. L. (2005).Predicting the relative efficacy of verbal, pictorial, andtangible stimuli for assessing preferences of leisureactivities. American Journal on Mental Retardation,110, 145–154. doi: 10.1352/0895-8017(2005)110<145:PTREOV>2.0.CO;2

DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of amultiple-stimulus presentation format for assessingreinforcer preferences. Journal of Applied BehaviorAnalysis, 29, 519–533. doi: 10.1901/jaba.1996.29-519

Fisher, W. W., Piazza, C. C., Bowman, L. G., &Amari, A. (1996). Integrating caregiver report with asystematic choice assessment to enhance reinforceridentification. American Journal on MentalRetardation, 101, 15–25.

Fisher, W. W., Piazza, C. C., Bowman, L. G.,Hagopian, L. P., Owens, J. C., & Slevin, I. (1992).A comparison of two approaches for identifying rein-forcers for persons with severe and profound disabil-ities. Journal of Applied Behavior Analysis, 25,491–498. doi: 10.1901/jaba.1992.25-491

Gilliam, J. E. (2006). Gilliam Autism Rating Scale (2nded.). Austin, TX: ProEd.

MEGAN R. HEINICKE et al.20

Graff, R. B., & Gibson, L. (2003). Using pictures toassess reinforcers in individuals with developmentaldisabilities. Behavior Modification, 27, 470–483. doi:10.1177/0145445503255602

Graff, R. B., & Karsten, A. M. (2012). Assessing prefer-ences of individuals with developmental disabilities: Asurvey of current practices. Behavior Analysis inPractice, 5, 37–48.

Groskreutz, M. P., & Graff, R. B. (2009). Evaluating pic-torial preference assessment: The effect of differentialoutcomes on preference assessment results. Researchin Autism Spectrum Disorders, 3, 113–128. doi:10.1016/j.rasd.2008.04.007

Hanley, G. P., Iwata. B. A., & Lindberg, J. S. (1999).Analysis of activity preferences as a function of differ-ential consequences. Journal of Applied BehaviorAnalysis, 32, 419–435. doi: 10.1901/jaba.1999.32-419

Higbee, T. S., Carr, J. E., & Harrison, C. D. (1999).The effects of pictorial versus tangible stimuli instimulus-preference assessments. Research in Develop-mental Disabilities, 20, 63–72. doi: 10.1016/S0891-4222(98)00032-8

Karsten, A. M., Carr, J. E., & Lepper, T. L. (2011).Description of a practitioner model for identifyingpreferred stimuli with individuals with autism spec-trum disorders. Behavior Modification, 35, 347–369.doi: 10.1177/0145445511405184

Kuhn, D. E., DeLeon, I. G., Terlonge, C., &Goysovich, R. (2006). Comparison of verbal prefer-ence assessments in the presence and absence of theactual stimuli. Research in Developmental Disabilities,27, 645–656. doi: 10.1016/j.ridd.2005.08.001

Lohrmann-O’Rourke, S., & Browder, D. M. (1998).Empirically based methods to assess the preferences ofindividuals with severe disabilities. American Journalon Mental Retardation, 103, 146–161. doi: 10.1352/0895-8017(1998)103<0146:ebmtat>2.0.co;2

Mason, S. A., McGee, G. G., Farmer-Dougan, V., &Risley, T. R. (1989). A practical strategy for ongoingreinforcer assessment. Journal of Applied BehaviorAnalysis, 22, 171–179. doi: 10.1901/jaba.1989.22-171

Parsons, M. B., Harper, V. N., Jensen, J. M., &Reid, D. H. (1997). Assisting older adults with severedisabilities in expressing leisure preferences: A proto-col for determining choice-making skills. Research inDevelopmental Disabilities, 18, 113–126. doi:10.1016/S0891-4222(96)00044-3

Piazza, C. C., Fisher, W. W., Hagopian, L. P.,Bowman, L. G., & Toole, L. (1996). Using a choiceassessment to predict reinforcer effectiveness. Journalof Applied Behavior Analysis, 29, 1–9. doi: 10.1901/jaba.1996.29-1

Roane, H. S., Lerman, D. C., & Vorndran, C. M.(2001). Assessing reinforcers under progressive sched-ule requirements. Journal of Applied Behavior Analysis,34, 145–167. doi: 10.1901/jaba.2001.34-145

Roane, H. S., Vollmer, T. R., Ringdahl, J. E., &Marcus, B. A. (1998). Evaluation of a brief stimuluspreference assessment. Journal of Applied BehaviorAnalysis, 31, 605–620. doi: 10.1901/jaba.1998.31-605

Roscoe, E. M., Iwata, B. A., & Kahng, S. (1999). Relativeversus absolute reinforcement effects: Implications forpreference assessments. Journal of Applied BehaviorAnalysis, 32, 479–493. doi: 10.1901/jaba.1999.32-479

Tessing, J. L., Napolitano, D. A., McAdam, D. B.,DiCesare, A., & Axelrod, S. (2006). The effects ofproviding access to stimuli following choice makingduring vocal preference assessments. Journal ofApplied Behavior Analysis, 39, 501–506. doi:10.1901/jaba.2006.56-05

Virués-Ortega, J., Pritchard, K., Grant, R. L., North, S.,Hurtado-Parrado, C., Lee, M. S. H., …Yu, C. T.(2014). Clinical decision making and preferenceassessment for individuals with intellectual and devel-opmental disabilities. American Journal on Intellectualand Developmental Disabilities, 119, 151–170. doi:10.1352/1944-7558-119.2.151

Volkert, V. M., Lerman, D. C., Trosclair, N.,Addison, L., & Kodak, T. (2008). An exploratoryanalysis of task-interspersal procedures while teachingobject labels to children with autism. Journal ofApplied Behavior Analysis, 41, 335–350. doi:10.1901/jaba.2008.41-335

Watkins, N., & Rapp, J. T. (2014). Environmentalenrichment and response cost: Immediate and subse-quent effects on stereotypy. Journal of Applied Behav-ior Analysis, 47, 186–191. doi: 10.1002/jaba.97

Whitehouse, C. M., Vollmer, T. R., & Colbert, B.(2014). Evaluating the use of computerized stimuluspreference assessments in foster care. Journal ofApplied Behavior Analysis, 47, 470–484. doi:10.1002/jaba.148

Wilder, D. A., Ellsworth, C., White, H., & Schock, K.(2003). A comparison of stimulus preference assess-ment methods in adults with schizophrenia. Behav-ioral Interventions, 18, 151–160. doi: 10.1002/bin.132

Zhou, L., Iwata, B. A., Goff, G. A., & Shore, B. A.(2001). Longitudinal analysis of leisure-item prefer-ences. Journal of Applied Behavior Analysis, 34,179–184. doi: 10.1901/jaba.2001.34-179

Received February 3, 2015Final acceptance February 10, 2016Action Editor, Tiffany Kodak

21PICTORIAL PREFERENCE ASSESSMENTS