ERIC EJ941692: Examination of the Influence of Contingency on Changes in Reinforcer Value

2011·0.51 MB·English

by  ERIC

JOURNALOFAPPLIEDBEHAVIORANALYSIS 2011, 44, 543–558 NUMBER3 (FALL2011)
EXAMINATION OF THE INFLUENCE OF CONTINGENCY ON
CHANGES IN REINFORCER VALUE
ISER G. DELEON
KENNEDYKRIEGERINSTITUTE
ANDJOHNSHOPKINSUNIVERSITYSCHOOLOFMEDICINE
MEAGAN K. GREGORY
KENNEDYKRIEGERINSTITUTE
MICHELLE A. FRANK-CRAWFORD
KENNEDYKRIEGERINSTITUTE
ANDUNIVERSITYOFMARYLAND,BALTIMORECOUNTY
MELISSA J. ALLMAN
KENNEDYKRIEGERINSTITUTE
ANDJOHNSHOPKINSUNIVERSITYSCHOOLOFMEDICINE
ARTHUR E. WILKE
CENTERFORAUTISMANDRELATEDDISORDERS,INC.
ABBEY B. CARREAU-WEBSTER
KENNEDYKRIEGERINSTITUTE
ANDUNIVERSITYOFMARYLAND,BALTIMORECOUNTY
AND
MANDY M. TRIGGS
KENNEDYKRIEGERINSTITUTE
This study examined how the amount of effort required to produce a reinforcer influenced
subsequent preference for, and strength of, that reinforcer in 7 individuals with intellectual
disabilities. Preference assessments identified four moderately preferred stimuli for each
participant, and progressive-ratio (PR) analyses indexed reinforcer strength. Stimuli were then
assigned to one of four conditions for 4 weeks: fixed-ratio (FR) 1 schedule, escalating FR
schedule,yokednoncontingent(NCR)delivery,andrestrictedaccess.Preferenceassessmentsand
PR schedules were then repeated to examine changes in selection percentages and PR break
points. Selection percentages decreased for all NCR stimuli but increased for most of the
restricted stimuli. There were no systematic changes in selection percentages for either of the
contingent stimuli. Break points increased, on average, for all conditions, but the increase was
highestfortherestrictedstimuliandlowestfortheNCRstimuli.Theseresultsarediscussedin
relationtorecent basic research addressing the influence ofefforton stimulus value.
Research and manuscript preparation were supported by Grant R03 MH072845 from the National Institute of
MentalHealth(NIMH)andGrantR01HD049753fromtheEuniceK.ShriverNationalInstituteofChildHealthand
HumanDevelopment(NICHD).Itscontentsaresolelytheresponsibilityoftheauthorsanddonotnecessarilyrepresent
theofficial views ofNIMH or NICHD.
Address correspondence to Iser G. DeLeon, Neurobehavioral Unit, Kennedy Krieger Institute, 707 N. Broadway,
Baltimore,Maryland 21205(e-mail: [email protected]).
doi:10.1901/jaba.2011.44-543
543
544 ISER G. DELEON et al.
Key words: noncontingent reinforcement, preference assessments, progressive-ratio analysis,
stimulus value
_______________________________________________________________________________
One important observation that has emerged 2007, p. 275). One sort of relatively aversive
from the literature on stimulus preference is event is the amount of effort required to
that preferences can change over time and produce the stimuli. Thus, stimuli reliably
experience (Hanley, Iwata, & Roscoe, 2006). preceded by greater effort become preferred
Items or events identified as more preferred on over stimuli preceded by less effort (e.g.,
oneoccasionmaybelesspreferredonaseparate Clement et al., 2000; Friedrich & Zentall,
occasion (Lohrmann-O’Rourke & Browder, 2004; Johnson & Gallagher, 2011; Kacelnik &
1998; Mason, McGee, Farmer-Dougan, & Marsh, 2002; Klein, Bhatt, & Zentall, 2005).
Risley, 1989; Zhou, Iwata, Goff, & Shore, Although some experimenters have been unable
2001). These fluctuations in stimulus prefer- to replicate this effect (e.g., Arantes & Grace,
ence sometimes translate into fluctuations in 2008; Vasconcelos, Urcuioli, & Lionello-De-
reinforcer effectiveness (e.g., DeLeon et al., nolf, 2007), most studies that have examined
2001). within-trial contrast effects have observed
Although preferences are dynamic, few positive instances (Zentall & Singer, 2007b).
studies have directly examined the variables If increases in the effort historically associated
that influence the stability of preferences or with earning a stimulus are positively related
durability of reinforcers. The exceptions have with the subsequent value of that stimulus
often examined the local effects of satiation and relative to others, it follows that changes in
deprivation on preference assessment outcomes stimulus value over time may vary systemati-
(Gottschalk, Libby, & Graff, 2000; Hanley, cally as a function of whether the stimuli were
Tiger, Ingvarsson, & Cammilleri, 2009; Mc- delivered in a contingent (higher effort) or
Adametal.,2005).Thesestudieshavegenerally noncontingent (lower effort) fashion.
revealed that stimulus deprivation can increase Research on changes in the value of a
preference rankings and satiation can decrease stimulus given past effort generally has exam-
preference rankings. For example, Hanley et al. ined the effects of these manipulations on
(2006) observed that repeated noncontingent preferences for discriminative stimuli associated
delivery of a stimulus resulted in lower indices withsymmetricalreinforcers(e.g.,S+associated
of preference for that stimulus. However, given with greater effort to obtain grain vs. S+
that one typically provides reinforcers on a associated with lesser effort to obtain grain).
contingent basis, it may also be important to From an applied standpoint, a more pertinent
examinechanges instimulusvalueasafunction question might be if these effects transfer to the
of contingent reinforcer delivery. reinforcers themselves when they are qualita-
Recent research suggests that events histori- tively distinct (e.g., one reinforcer associated
cally associated with producing a stimulus can with greater effort vs. a different reinforcer
influence subsequent changes in the value of associated with lesser effort). Birch, Zimmer-
that stimulus. For example, the effect some- man, and Hind (1980) observed something
times termed within-trial contrast (Zentall & akin to this effect. They first asked preschool
Singer, 2007a), or the work ethic effect (Clem- children to rate snack foods (peanuts, carrots,
ent, Feltus, Kaiser, & Zentall, 2000), predicts crackers, etc.). Four moderately preferred foods
that ‘‘reinforcers that follow relatively aversive were then subjected to various manipulations
events become preferred over those that follow for several weeks, after which thechildren again
less aversive events’’ (Singer, Berry, & Zentall, rated the foods. In one manipulation, the
THE INFLUENCE OF CONTINGENCY 545
children received the designated snack item as a ing. The last ratio requirement met, termed the
reward contingent on the display of varying breakpoint,providesanindexoftheamountof
behaviors (e.g., responding to verbal requests, workthereinforcerwillsupport.Comparisonof
cooperative play) twice a day for 21 days. In mean break points provides a gauge of relative
another condition, the child received the potency across stimuli. Recent applied studies
designated item in a nonsocial context; the (e.g., DeLeon, Frank, Gregory, & Allman,
items were simply placed in the children’s 2009; Francisco, Borrero, & Sy, 2008; Glover,
lockerstwiceadayandthechildrenwerefreeto Roane, Kadey, & Grow, 2008) suggest that PR
consume them when they visited their lockers. schedules perform fairly well in validating
After 21 days, the children again rated the differences in preference levels by gauging
items, and the experimenters found an overall independent stimulus value.
increase in the ratings of the snack items AnaddedadvantageofusingPRbreakpoints
delivered as rewards, leading them to conclude is that reinforcer potencycan be gauged for one
that ‘‘presenting a food as reward enhances stimulus independent of the potency of other
preferencesforthatfood’’(p.860).Bycontrast, stimuli. Recent examinations of within-trial
ratings of the items delivered independent of a contrast effects have been restricted to measur-
programmed contingency did not increase. In a ing relative preferences among two or more
manner consistent with the more recent stimuli, arranged in concurrent schedules.
nonhuman animal research, items associated Although these studies have shown that prior
withanearningrequirementincreasedinrating, manipulations of effort can influence relative
and those associated with no effort did not. responseallocationtoonestimulusoveranother
Although this study suggested that contin- concurrently available stimulus, differences in
gency might be positively related to subsequent relative response allocation under concurrent
reinforcer value, the only dependent measure schedules do not always indicate a difference in
was children’s ratings of the items. Verbal self- the absolute strength of a reinforcer in
reports of relative preferences, however, may supporting operant behavior on independent
notalwayscorrespondwithbehavioralmeasures (nonconcurrent) schedules (Roscoe, Iwata, &
of the relative strength of the stimuli in Kahng, 1999). It is therefore important to
supporting operant behavior (e.g., Northup, determine whether changes in relative prefer-
2000; Northup, George, Jones, Broussard, & ence translate into meaningful changes in the
Vollmer, 1996). Therefore, in addition to some independent strength of the stimuli as reinforc-
index of changes in relative preference, as ers. DeLeon, Williams, Gregory, and Hagopian
offered by established stimulus preference (2005) reported pilot data for two individuals
assessments, it is important to conduct a more with developmental disabilities from such an
direct test of changes in reinforcer strength analysis. The experimenters delivered moder-
before and after arranging manipulations of ately preferred stimuli on an FR 1 schedule
effort.Onemeasurethathasreceivedincreasing during daily training sessions and noncontin-
attention in the applied behavioral literature is gently on a schedule yoked to the FR 1
the break point on a progressive-ratio (PR) schedule. Thus, the individuals received equiv-
schedule (Hodos, 1961). Under a PR schedule, alent exposure to the two stimuli, but they
reinforcers are contingent on the completion of worked for one and not the other. PR analyses
response requirements that increase within conducted before and after these arrangements
sessions. The schedule continues to increase revealed increases in break points for the
until reaching some termination criterion, contingent stimuli but decreases for the non-
usually some period of time without respond- contingent stimuli, again suggesting a positive
546 ISER G. DELEON et al.
relation between prior effort and changes in ney’ssessionstookplaceatatablelocatedinthe
stimulus value. center of a room (7.7 m by 7.7 m) adjacent to
The present study sought to expand this line the main living area. Sessions for Todd and
of investigation on the relation between con- Cathytookplaceataworkstation(deskandtwo
tingency and subsequent stimulus value. The chairs) located in the classroom (6.8 m by
influence of contingent and noncontingent 6.8 m) where they completed their daily
delivery on changes over time was measured academics. Partitions divided the workstations.
by (a) changes in preference assessment selec-
Response Definitions, Data Collection, and
tion percentages and (b) changes in reinforcer
Interobserver Agreement
strength via PR schedules. To further examine
the effects of the amount of effort, the current Trained graduate students and research
study evaluated two levels of contingent effort: assistants served as data collectors for all
contingentdeliveryonaconstantFR1schedule assessments. Observers used laptop computers
and contingent delivery on an escalating FR to collect data for preference assessments and
schedule in which the schedule requirements reinforcer-value manipulation sessions. During
increased across successive weeks. Finally, the the preference assessments, thetherapist record-
same measures were applied to stimuli entirely ed responses in the presence of stimuli
restricted (i.e., no exposure) between preference includingselection,avoidance,andnoresponse.
assessments and PR analyses. This latter Selection included touching, reaching toward,
condition served as a control to assess what oraskingforthetoyorfooditem.Anavoidance
might happen in the absence of any effort response was scored if the participant actively
manipulation. pushed or threw the stimulus away, moved
away from the stimulus within 3 s of
presentation, or engaged in negative vocaliza-
METHOD
tionsuchascryingorsaying‘‘no.’’Noresponse
Participants and Settings was scored if the participant exhibited no
Seven individuals with developmental dis- reaction to the stimuli within 5 s of presenta-
abilities admitted to an inpatient unit for the tion.
assessment and treatment of behavior disorders During reinforcer-value manipulation ses-
participated in the study (see Table 1 for sions, observers scored task completion each
participant description and stimuli assigned to time the participant completed the task follow-
each condition). Four of the participants ing a verbal or gestural prompt. The task for
(Cathy, Courtney, Mary, and Thomas) had Mary and Thomas was placing a block in a
been previously involved in a study examining bucket (scored when the block passed the lip of
the correspondence between preference assess- the bucket and the participant let it go).
ment outcomes and PR schedule analyses Jonathan’s task was to sort paper clips by size.
(DeLeon et al., 2009), and thus were familiar Thetherapistplacedtwobaskets onthetablein
with the PR schedule analyses. frontofhim,onewithasmallclipattachedtoit
Sessions were conducted in different areas of and one with a large clip attached. A correct
the inpatient hospital. Sessions for Mary and response involved placing a clip in the basket
Thomastook placeinasession room(2.4m by with the same-sized clip attached. The task for
3 m) equipped with a one-way observation Todd, Cathy, April, and Courtney was to place
window,twochairs,andatable.April’ssessions apeginaPeg-Board.Observersscoredacorrect
took place at one of two tables located on the response when the participant placed the peg
main living area (9.4 m by 9.4 m) of the completely in any one of 25 holes in the board
inpatient hospital unit. Jonathan’s and Court- such that it stood upright when the participant
THE INFLUENCE OF CONTINGENCY 547
Table 1
ParticipantDescription andStimuli Assigned toEachCondition
Name Age Gender Diagnosis Condition Stimulus
Mary 9 female autism,moderatementalretardation,mood FR1 musicradio
disorder(NOS) NCR band
esclFR phone
FRrest ballpopper
Jonathan 13 male autism,severementalretardation,pervasive FR1 popcorn
developmentaldisorder(PDD) NCR chips
esclFR crackers
rest M&Ms
Todd 11 male autism,disruptivebehaviordisorder(NOS), FR1 sphere
stereotypicmovementdisorderwithself-injury NCR ballpopper
esclFR caterpillar
rest drum
Cathy 11 female autism,moderatementalretardation,stereotypic FR1 radio
movementdisorderwithself-injury NCR shakingdog
esclFR ballpopper
rest band
April 10 female mentalretardation(NOS),PDD,stereotypic FR1 radio
movementdisorderwithself-injury NCR playdough
esclFR ballpopper
rest Elmobook
Courtney 20 female criduchat,severementalretardation FR1 playdough
NCR caterpillar
esclFR shakingdog
rest spintoy
Thomas 16 male mentalretardation(NOS),PDD,stereotypic FR1 squishyball
movementdisorderwithself-injury NCR Spongebob
esclFR wormball
rest car
Note. NOS 5 not otherwise specified; FR 1 5 fixed-ratio 1; NCR 5 noncontingent reinforcement; escl FR 5
escalating fixedratio; rest 5restricted.
let it go. For all participants, stimulus delivery erence assessments consisted of both observers
consisted of placing the stimulus directly in recording the same selection, avoidance, or no
front of the participant. Stimulus interaction response during each trial. Mean percentage
included touching, turning on, or otherwise agreement across participants for the paired-
interacting with the item. For Jonathan only, choice preference assessments was 97% (range
consumptionconsistedofmovingthefooditem across sessions, 81% to 100%). During the PR
past the plane of his lips. analyses, an agreement consisted of both
During the PR analyses, trained observers observers placing a check or no check next to
usedpaperandpenciltorecordeachcompleted each FR schedule value. Mean percentage
FR schedule value. The data sheet listed agreement across participants for the PR
individual FR values ranging from FR 1 to analyses was 99% (range, 92% to 100%).
FR25.Nextto eachFR valuewasa boxforthe Interobserver agreement data for the reinforcer
data collector to check after the participant manipulation sessions were calculated for
completed the respective schedule value. compliance with prompts, delivery of the
Asecondindependentobservercollecteddata stimulus, and item interaction using the exact
during an average of 87%, 58%, and 29% of agreement within intervals method (Mudford,
paired-choice preference assessments, PR anal- Martin, Hui, & Taylor, 2009). Sessions were
yses, and reinforcer-value manipulation ses- dividedintoconsecutive10-sintervals.Intervals
sions, respectively. Agreement during the pref- in which the same number was scored by both
548 ISER G. DELEON et al.
observerswereassignedavalueof1.Intervalsin preference assessment results) were selected
which one observer scored 0 target responses from the preference assessment for inclusion
andtheotherscoredanythingotherthan0were in the experimental manipulations for each
assigned a value of 0. For intervals in which participant, with two exceptions. For Mary, the
different numbers were scored, the smaller ninth-ranked stimulus (the ball popper) re-
number of responses scored was divided by placed the eighth-ranked stimulus (the rattle)
the larger number. These quotients were because she engaged in self-injury with the
summed, divided by the total number of rattle. Thomas’s seventh-ranked stimulus (the
intervals in the session, multiplied by 100%,
musical band) was broken after the initial three
and averaged across sessions. Mean percentage
preference assessments were completed. There-
agreement across participants for reinforcer-
fore, the fourth-ranked stimulus (the worm
value manipulation sessions was 95% (range,
ball) was used during the PR analysis and
57% to 100%), 89% (range, 42% to 100%),
reinforcer value manipulation sessions. The
and 86% (range, 53% to 100%) for compli-
musical band continued to be included in the
ance, stimulus delivery, and item interaction,
preference assessments. Across all participants,
respectively.
the difference in selection percentage for the
stimulichosenforinclusionintheremainderof
Procedure
the study was no greater than 28%. All stimuli
Paired-choice preference assessment. The
were entirely restricted from the participants’
paired-choice preference assessments (Fisher et
environment outside the experimental condi-
al., 1992) included 12 leisure or food items
tions. This seemed important because uncon-
identified by caregivers or hospital staff. Each
trolled access to the stimuli could affect
item was paired once with every other item in a
subsequent preference rankings.
quasirandom order for a total of 66 trials.
Progressive-ratio analyses. Three PR sessions
During each trial, the therapist presented two
items to the participant. If he or she selected were conducted, in a randomized order, with
eitheritem,thetherapistprovided30-saccessto each of the four selected stimuli for each
theleisureitem or delivereda smallpiece ofthe participant(i.e.,12PRsessionsperparticipant).
food to consume. The therapist blocked all A task that the participant could readily
attempts to select both items simultaneously. If complete was chosen from prior educational
the participant made no response toward both plans. Before each session, the therapist
items,thetherapistphysicallyguidedhimorher prompted the participant to complete the task
to sample each item for 5 s and then initiated a threetimesusingsuccessiveverbal,gestural,and
second trial. The assessment was repeated twice physical prompts. After completion of each
more, and the selection percentages were task, regardless of the level of prompting, the
averaged across the three assessments. Selection therapist delivered the appropriate stimulus for
percentage was determined by calculating the 30 s. When the session began, the therapist
percentage of assessment trials in which stimuli placed the available stimulus within the view of
were selected when available. The items were the participant and issued a single verbal
then arranged in descending order according to prompt to engage in the task. No additional
their selection percentages (i.e., the item prompting was delivered. The stimulus being
selectedwiththegreatest percentagewasranked assessed during agiven sessionwas deliveredfor
1, the item with the second highest selection 30 s on a PR schedule beginning with a single
percentage was ranked 2, etc.). response (i.e., the stimulus was delivered for a
The middle four stimuli (i.e., those ranked single correct completion of the task). During
fifth,sixth,seventh,andeighthinthecombined each trial thereafter, the schedule requirement
THE INFLUENCE OF CONTINGENCY 549
was increased by one in an arithmetic progres- 1 min. For Jonathan only, asmall piece of food
sion (e.g., two responses, three responses, etc.). was delivered. At the end of the reinforcement
The therapist removed task materials after the interval,theFR1stimuluswasremovedandthe
deliveryofeachreinforcerandreturnedthemas next prompt to complete the task was issued.
soon as the 30-s reinforcement period had The therapist issued the next prompt for
elapsed.Sessionscontinueduntiltheparticipant Jonathan after he consumed the food. This
ceased to respond for 1 min. The highest continued until the participant had earned the
schedulerequirementcompletedbeforemeeting stimulus 10 times.
the termination criterion constituted the break The escalating FR condition was identical to
point for the session. the FR 1 condition, with the exception that the
Reinforcer-value manipulations. Following the FR schedule value increased after every fifth
completion of the PR analyses, the four stimuli session.Three-steppromptingwasagainusedto
were randomly assigned to one of four ensure completion of each task presented. The
experimental conditions: FR 1, escalating FR, therapist delivered the escalating FR stimulus
noncontingent reinforcement (NCR), or re- after the predetermined schedule requirements
stricted (see Table 1 for assignment of stimuli had been met, regardless of the level of
to conditions). Sessions were conducted each prompting required. During the 1st week, the
day, 5 days per week for 4 weeks. Each day, participantearnedtheitemonanFR1schedule
three experimental conditions were conducted identical to that described for the FR 1
inthefollowingorder:FR1,escalatingFR,and condition. The reinforcement schedule was
NCR. For all conditions, sessions were termi- increased to FR 2 during the 2nd week, FR 5
nated after the stimulus assigned to that during the 3rd week, and FR 10 during the 4th
condition was delivered 10 times. Thus, session week. The response requirements used in the
length varied while the total number of Escalating FR condition were adopted from
reinforcers delivered per session for each prior studies in which these values produced
condition remained constant. meaningful outcomes in related analyses (e.g.,
During the FR 1 condition, the participant DeLeon, Iwata, Goh, & Worsdell, 1997). All
and therapist sat next to each other at a table. sessions, regardless of the reinforcement sched-
The therapist presented the task using a three- ule in place, were terminated after the partic-
step prompting procedure that consisted of ipant earned the escalating FR stimulus 10
sequential verbal, gestural, and physical times.
prompts. The task used for each participant During the NCR condition, no academic
was identical to the tasks used during the PR task was present. The therapist delivered the
analyses. Three-step prompting was used for stimulusassignedtothisconditionfor1minon
both effort manipulations (i.e., FR 1 and a fixed-time (FT) schedule that was yoked to
escalatingFRconditions).Thetherapistinitially the schedule of delivery for the stimulus in the
delivered a verbal prompt to complete the task. precedingFR1condition.TheFTschedulewas
If the participant did not comply with the determined by dividing the duration of the FR
verbal prompt, a gestural prompt was delivered 1 session by 10 (i.e., the number of times the
(i.e., the correct response was modeled for the participant earned the reinforcer in the FR 1
participants). If the participant still did not condition). For example, if the participant
comply, he or she was physically guided to earned the FR 1 stimulus 10 times and the
complete the task. Each time the participant session lasted 12 min, the NCR session would
completed the task, regardless of the level of be 12 min in duration and the stimulus would
prompting, the FR 1 stimulus was delivered for be delivered every 72 s, with the first reinforcer
550 ISER G. DELEON et al.
being delivered at the start of session. The the four stimuli identified for each participant
therapist removed the stimulus when the and the break-point values for each of the four
reinforcement interval elapsed and withheld stimuli during the PR schedule analyses.
the stimulus until the next scheduled interval. Changes in selection percentage were deter-
NCR sessions ended when the stimulus had mined by comparing selection percentages for
been delivered 10 times. each of the stimuli during the initial preference
The stimulus assigned to the restricted assessments and preference assessments that
condition was placed in a locked closet such followed the reinforcer value manipulations.
that the participant would not have contact Changes in break-point values were determined
withorseethestimulus.Therestricteditemwas by comparing the means of the last completed
used only during subsequent preference assess- FR schedule in thethree initial PRanalyses and
ments and PR analyses. final three PR analyses with each stimulus.
Prior research on events that may affect
stimulus preference have suggested that (a) RESULTS
noncontingent stimulus delivery can result in
Figure 1 displays change scores (postmanip-
a decrease in preference indices (e.g., Hanley et
ulation average minus premanipulation aver-
al., 2006), (b) stimuli that require greater effort
age) in the preference assessment selection
toobtainmaybe preferredrelative tothosethat
percentages for each type of stimulus for each
require less effort (e.g., Friedrich & Zentall,
participant. Selection percentages for the NCR
2004),and(c)stimulusdeprivationcanincrease
stimuli decreased for all seven participants. The
preference (e.g., McAdam et al., 2005). We
restricted stimuli moved upwards in selection
therefore hypothesized that preferences (indi-
percentage for five of the seven participants and
cated by selection percentage) and effectiveness
moved down for two participants. As noted
(measured by PR break points) for the stimulus
previously, the restricted stimuli served as a
assigned to the NCR condition would decrease
form of control to indicate what might happen
when compared to preferences for the stimuli
across evaluations in the absence of stimulus
assigned to the three other experimental
access between preference assessments. The FR
conditions. Furthermore, preference for the
Escalating FR stimulus was expected to increase 1 stimuli showed a mixed pattern, increasing in
to a greater extent than that for the FR 1 selectionpercentageforfouroftheparticipants,
stimulus. Finally, we hypothesized that depri- but decreasing for the remaining three partic-
vation resulting from restricted access would ipants. The escalating FR stimuli also showed a
result in an increase in preference for the mixed pattern, but decreases in selection
restricted stimulus relative to that for the percentages were more common than increases.
noncontingent stimulus. Figure 2depictsthepercentagechangescores
inthePRbreakpointsforeachtypeofstimulus
Design and Data Analysis for each of the participants. The change scores
At the end of Week 4, three preference were calculated by subtracting the mean
assessments were again conducted with all premanipulation break point from the mean
participants. In addition, the three PR sessions postmanipulation break point, dividing the
were repeated with the stimuli assigned to each difference by the mean premanipulation break
condition (i.e., 12 PR sessions per participant). point, and multiplying by 100%. Percentage
A before-and-after comparison was used to change, rather than raw change, was used
assess changes in the value of a given stimulus. because an absolute change in the mean
The dependent variables included the selection break-point value would have very different
percentages across preference assessments for interpretations depending on the base value
THE INFLUENCE OF CONTINGENCY 551
Figure 1. Differences in selection percentages between the preference assessments conducted before and after the
experimentalmanipulations for each participant.
(e.g., an increase of two mean responses is a increases in mean break points were observed
substantial increase for a stimulus with a forsixofthesevenFR1stimuli,withadecrease
premanipulation mean break point of one in one stimulus. Finally, mean break points
response but not for a stimulus with a mean increasedforfourofsevenescalatingFRstimuli
premanipulation break point of 20). and decreased for three stimuli. Although break
A great deal of variability was observed points increased from the first to the second
during the second PR analysis, with PR break administration in all conditions, increases
points increasing for 17 of the 28 stimuli, occurred less often in the NCR condition
decreasing for nine of the 28 stimuli, and not (two of seven cases) than in the other three
changingfortwostimuli.Themeanbreakpoint conditions. These results provide further sup-
decreased for four of the NCR stimuli, port for the notion that the noncontingent
increased in two cases, and remained the same delivery of a stimulus can result in a reduction
inonecase.Themeanbreakpointincreasedfor in the value of that stimulus.
five of seven restricted stimuli, decreased for Figure 3 (top) depicts the mean change in
one, and stayed the same for one. Percentage selection percentage for each type of stimulus
552 ISER G. DELEON et al.
Figure2. Percentagechangescoresfromthemeanpremanipulationbreakpointtothepostmanipulationbreakpoint
for eachtype of stimulus foreach participant.
collapsed across participants. Error bars depict Figure 3 (bottom) depicts the mean percent-
the standard deviation. The data for each bar agechangeinbreak-pointvaluesforeachtypeof
were calculated by subtracting the premanipula- stimulus across all participants. Overall, the
tion selection percentage from the postmanipu- mean break point increased for each type of
lation selection percentage and averaging the stimulus. Mean break points for Restricted
differences across all participants. Consistent stimuli made the largest percentage gain (M 5
with the individual data, the mean difference 355%,SD5548),followedindescendingorder
increasedfortheRestrictedstimuli(M59.1,SD by the FR 1 stimuli (M 5 258.3%, SD 5
5 18.6) and, to a lesser extent, for the FR 1 446.7), the escalating FR stimuli (M 5 84.6%,
stimuli (M 5 7.2, SD 5 32.6). The mean SD 5 152.3), and finally, the NCR stimuli (M
differencedecreasedfortheEscalatingFRstimuli 5 5.3%, SD 5 83.9). Although there was an
(M526.8,SD524.4),butthelargestchange averageincreaseinPRbreakpointsfortheNCR
wasadecreaseintheselectionpercentageforthe stimuli, it was accounted for by the relatively
NCR stimuli (M 5 217.2, SD 5 16). large increases observed in the only two NCR