RESISTANCE TO DISRUPTION IN A CLASSROOM SETTING

DIANA E. PARRY-CRUWYS, CARRIE M. NEAL, AND WILLIAM H. AHEARN

NEW ENGLAND CENTER FOR CHILDREN AND NORTHEASTERN UNIVERSITY

AND

EMILY E. WHEELER, RASEEKA PREMCHANDER, MELISSA B. LOEB, AND

WILLIAM V. DUBE

UNIVERSITY OF MASSACHUSETTS MEDICAL SCHOOL AND SHRIVER CENTER

Substantial experimental evidence indicates that behavior reinforced on a denser schedule is moreresistant to disruption than is behavior reinforced on a thinner schedule. The present experimentstudied resistance to disruption in a natural educational environment. Responding duringfamiliar activities was reinforced on a multiple variable-interval (VI) 7-s VI 30-s schedule for 6participants with developmental disabilities. Resistance to disruption was measured by presentinga distracting item. Response rates in the disruption components were compared to within-sessionresponse rates in prior baseline components. Results were consistent with the predictions ofbehavioral momentum theory for 5 of 6 participants.

Key words: behavioral momentum, relative reinforcement density, behavioral persistence,multiple schedules, developmental disabilities

_______________________________________________________________________________

Reinforcement schedule thinning is a com-monly used tactic for maintaining acquiredskills. However, basic research on behavioralmomentum suggests that this strategy may becounterproductive, particularly if schedule thin-ning occurs in a distracting classroom environ-ment (Nevin, 1974, 1992; Nevin & Grace,

2000). This study examined the strength oftask-related behavior under differing schedulesof reinforcement in the presence of commonlyoccurring distracting stimuli using the behav-ioral momentum metaphor. The behavioralmomentum metaphor borrows its terminologyfrom classical physics. In physics, the momen-tum of an object is the product of its velocityand mass. If an external force is applied to theobject, the greater the mass, the greater theresistance to change or disruption. In behavioralmomentum, rate of responding is analogous tovelocity and is largely determined by theschedule of reinforcement (the response–rein-forcer relation). The characteristic rate ormagnitude of obtained reinforcement in thesituation (the stimulus–reinforcer relation) de-termines the behavioral analogue of mass(Nevin, 1974).

Studies with nonhuman subjects have mea-sured relative resistance to change using amultiple-schedule arrangement with two com-ponents with different rates of reinforcement.Relative resistance to change was determined byintroducing disrupters such as extinction or

Correspondence concerning this article should beaddressed to W. V. Dube, University of MassachusettsMedical School Shriver Center, 333 South Street, Shrews-bury, Massachusetts 01545 (e-mail: william.dube@umassmed.edu) or W. H. Ahearn, New England Centerfor Children, 33 Turnpike Road, Southborough, Massa-chusetts 01772 (e-mail: bahearn@necc.org).

doi: 10.1901/jaba.2011.44-363

Research and manuscript preparation were supported byGrant HD033802 from the National Institute of ChildHealth and Human Development and by the NewEngland Center for Children. The contents of this paperare solely the responsibility of the authors and do notnecessarily represent the official views of NICHD. Someof the data were presented at the Association for BehaviorAnalysis International annual convention in May, 2007.We thank Lauren Abraham, Jason Krienke, and SeanTobyne for assistance with data collection; KarenDeGregory and David Pasterchik of Abilities Softwarefor computer programming; and the staff and students ofthe New England Center for Children for their cooper-ation.

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2011, 44, 363–367 NUMBER 2 (SUMMER 2011)

363

prefeeding and measuring disruption in re-sponding as a proportion of baseline responserate (e.g., Nevin, Mandell, & Atak, 1983).Responding in the more densely reinforcedcomponent was generally more resistant todisruption (see Nevin & Grace, 2000, for areview).

The majority of studies with human subjectswith developmental disabilities have examinedbehavioral momentum effects using computer-based tasks in laboratory settings (e.g., Dube,McIlvane, Mazzitelli, & McNamara, 2003). Incontrast, Mace et al. (1990) examined behav-ioral momentum with two adults with intellec-tual disabilities in the participants’ group home.The responses were sorting plastic dinnerware,and the disrupter was a video. Results showedthat behavior exposed to a higher rate ofreinforcement (i.e., a richer schedule of rein-forcement) was more resistant to disruption.

The present study replicated and extendedthe findings of Mace et al. (1990). Subjects werechildren with developmental disabilities, andthe settings were their special education class-rooms. The behavior measured was respondingto regularly scheduled academic tasks orstructured leisure activities. The reinforcerswere those ordinarily used in the classroom toteach or maintain such behavior (e.g., tokens),and the disrupters were potentially distracting

objects or events that were also part of thesubjects’ typical educational environment.

METHOD

Participants

Six boys enrolled in special educationprograms participated in the study. All partic-ipants were selected based on their ability tocomplete academic or simple leisure tasks andparental consent for their participation. Table 1shows participants’ ages, diagnoses (from schoolrecords), and Peabody Picture VocabularyTest–Revised mental age-equivalent scores.Ben, Cody, Paul, Jon, and Ryan attended aday school for children with autism anddevelopmental disabilities, and Noah attendeda behavioral residential program.

Settings and Materials

All sessions took place at the desk or tablewhere the student usually worked. Ben andCody sat at a table in their classroom, and theacademic task materials were spelling and mathworksheets copied on blue or white paper. Paul,Jon, Ryan, and Noah regularly received indi-vidual instruction in small areas (approximately1.5 m by 1.5 m) separated from the commonarea of the classroom with room dividers onthree sides. A camcorder mounted above the

Table 1

Experimental Details

Participant, age, diagnosis,PPVT MAEa VI 7-s task VI 30-s task Reinforcer Disrupter

Number ofbaseline sessions

Ben, 13, ADHD andOCDb, 12.3

spelling math tokens laptop computer with racinggame, computermagazine

18

Cody, 12, autism, 8.9 math spelling tokens toy activity set 12Paul, 11, autism, ,1.75 beads puzzles candy sing-along video 31Jon, 5, PDDc, ,1.75 beads blocks onion rings, chips cartoons DVD 17Ryan, 4, autism, 5.6 blocks puzzles pretzels, raisins cartoons DVD 20Noah, 8, autism, ,1.75 blocks puzzles pretzels push-button musical books 16

a Peabody Picture Vocabulary Test–Revised mental age-equivalent score; score of ,1.75 years indicates failure toachieve the basal score.

b Attention deficit hyperactivity disorder and obsessive compulsive disorder.c Pervasive developmental disorder.

364 DIANA E. PARRY-CRUWYS et al.

student’s desk recorded each session. Leisureactivity materials included 8- and 24-piecepuzzles, 1-in. wooden beads with string, and1- to 2-in. interlocking blocks. Disrupterstimuli were chosen individually for eachparticipant based on teacher reports of prefer-ence among items that could be manipulatedwhile one hand was kept free for taskengagement (see Table 1).

The experimenter used a handheld computerto run implement a multiple variable-interval(VI) VI schedule. The computer program cuedthe experimenter through an earpiece as towhen to begin each component of the multipleschedule, when a reinforcer should be deliveredfollowing the next response, and when to endeach component of the session.

Reinforcers were the same as those used forparticipants’ regular instruction. Ben and Codyreceived tokens that were exchangeable for abreak with a toy or computer game after thesession. Paul, Jon, Ryan, and Noah receivedpreferred food items; these items were deter-mined by recent preference assessments con-ducted either by the student’s teacher or by theexperimenter. Reinforcers for each participantare listed in Table 1.

Response Measurement and Reliability

The dependent variable was the rate of task-related responding in each component of themultiple schedule. For Ben and Cody, aresponse was defined as making a written markon a page. The number of responses percomponent was determined after sessions bycounting letters, numerals, lines, punctuation,and any marks still visible on the worksheetafter erasing. Incorrect answers were alsocounted as responses. For Paul, Jon, Ryan,and Noah, reinforcement was delivered forspecific responses that indicated task-appropri-ate engagement with the activity. Puzzlebuilding included picking up a puzzle piecefrom the desk, touching it to the board, movinga piece already on the board, or placing a pieceanywhere on the desk or puzzle board. Block

building included picking up a block, attachinga block, or detaching a block. Bead stringingincluded picking up a bead, touching the tip ofthe string to a bead, or pulling the stringthrough a bead. For these four participants, thevideotaped sessions were downloaded to acomputer and coded for number of responsesper component using the Video Frame Codersoftware program.

Interobserver agreement was calculated for100% of test sessions by comparing theresponse counts of two independent scorers.Interobserver agreement was determined bydividing the number of agreements by agree-ments plus disagreements and multiplying by100%. Mean interobserver agreement for allparticipants was 96% (range for individualsessions, 80% to 100%).

Procedure

During all sessions, the experimenter verballyprompted Ben and Cody with the phrase ‘‘It’stime to work’’ if no response occurred for 5 s.Paul, Ryan, and Noah were prompted torespond if no response occurred for 10 s, firstverbally (‘‘It’s time to work on —’’) and thenwith a light physical prompt repeated ifnecessary every 5 s. Prompting rarely occurredfor these participants. For Jon, who pausedmore often than the other participants, if noresponse occurred for 5 s, the experimenter usedfull manual guidance to prompt him torespond, and the prompt was repeated at 5-sintervals if he did not respond independently.His prompted responses were never followed byreinforcers during the sessions and were notincluded in response totals.

Baseline sessions. Sessions included alternatingcomponents with two tasks, with a VI 7-sschedule for one task and a VI 30-s schedule forthe other task (Table 1). The task–schedulepairings remained the same throughout allsessions for each participant. The VI 7-sschedule included the following values: 2, 3,4, 5, 7, 9, 10, 11, and 12 s; the VI 30-s scheduleincluded 10, 15, 20, 25, 30, 35, 40, 45, and

RESISTANCE TO DISRUPTION 365

50 s. Each schedule value was programmed inrandom order, with the restriction that everyvalue occurred before any was repeated. Thetwo tasks or activities alternated within sessions,and the order in which they appeared alternatedacross sessions. Each component was 90 s long,followed by a 20-s intercomponent interval.Ben’s and Cody’s sessions consisted of eightcomponents, and the remaining participants’sessions consisted of six components. Thebaseline phase continued for a minimum of12 sessions and until response rates for bothactivities were judged to be stable by visualinspection. The initial baseline phase averaged19 sessions across participants (Table 1).

Test sessions. Following the initial series ofbaseline sessions, a multielement design thatalternated baseline and test sessions was imple-mented for each participant (similar to Dube etal., 2003). The structure of test sessions was thesame as for baseline sessions, with the exceptionthat during test sessions a distracting item (i.e.,disrupter stimulus, see Table 1) was introducedduring the fifth and sixth components. Asecond experimenter sat next to Ben and Codyand engaged with the item so that theparticipant could easily watch. For the otherparticipants, the distracting item was place onthe table. Responding to the tasks was rein-forced on the designated schedule duringdistracter components just as it was duringbaseline components. The order of componentswithin test sessions alternated across successivetest sessions. Ben, the first participant in theexperiment, received six test sessions. All otherparticipants received a minimum of six testsessions, and testing continued until respondingwas consistently higher for one task across atleast five of six consecutive sessions or until amaximum of 10 test sessions were conducted.

RESULTS AND DISCUSSION

The data of interest are the rate of taskresponding during test components, expressedas a proportion of average response rates in the

prior baseline components within the samesession. A test–baseline response ratio for eachtask was calculated by dividing the response rateduring the test component by the response ratein the first two baseline components for thesame task in the same test session. The meanproportional response rate for each participant,from the final six test sessions, is shown inFigure 1. Individual session data, which areconsistent with the overall means, are availablefrom the first author.

For five of six participants, resistance todistraction was greater for the task associatedwith the richer schedule of reinforcement,although the effect was small for Cody. Forthe sixth participant (Noah), resistance todistraction was slightly greater for the taskassociated with the thinner schedule. Theseresults, along with those of Mace et al. (1990),indicate that the predictions of behavioral

Figure 1. Mean response rate as a proportion of baselineresponse rate for the last six test sessions for all participants.

366 DIANA E. PARRY-CRUWYS et al.

momentum theory may be applicable to clinicalsettings. Notably, the settings, instructional orleisure materials, responses, reinforcers, anddistracters all occurred in the participants’everyday environments. Thus, ongoing relativereinforcer rates may affect behavioral persistencein the absence of the relatively high level ofcontrol found in the laboratory. Althoughschedules of reinforcement are typically thinnedto maintain responding, these results suggestthat behavior will be more likely to bemaintained in the presence of common dis-tracters if the behavior is reinforced on a denserschedule. Additional research is needed toreconcile the benefits and limitations of sched-ule thinning.

One distinction between the current studyand Mace et al. (1990) concerns responsetopography. The task in Mace et al. (sortingplastic dinnerware) was the same in both richand lean components. In the present study, theacademic tasks differed in subject content forBen and Cody, and the leisure activities differedin response requirements for the remainingparticipants. (Note that response rates werenever compared across activities; only within-task proportional measures of response disrup-tion were compared.) Although the participantswere familiar with and had previously masteredall of the tasks, it is possible that the two taskswere not of equivalent response effort for someparticipants. If so, the results may reflect an

interaction between effort and reinforcer rate.This possibility is a topic for future research,along with other variables that influenceresistance to change in natural settings (e.g.,Ahearn, Clark, Gardenier, Chung, & Dube,2003).

REFERENCES

Ahearn, W. H., Clark, K. M., Gardenier, N. C., Chung,B. I., & Dube, W. V. (2003). Persistence ofstereotypic behavior: Examining the effects of externalreinforcers. Journal of Applied Behavior Analysis, 36,439–448.

Dube, W. V., McIlvane, W. J., Mazzitelli, K., &McNamara, B. (2003). Reinforcer rate effects andbehavioral momentum in individuals with develop-mental disabilities. American Journal on MentalRetardation, 108, 134–143.

Mace, F. C., Lalli, J. S., Shea, M. C., Lalli, E. P., West, B.J., Roberts, M., et al. (1990). The momentum ofhuman behavior in a natural setting. Journal of theExperimental Analysis of Behavior, 54, 163–172.

Nevin, J. A. (1974). Response strength in multipleschedules. Journal of the Experimental Analysis ofBehavior, 21, 389–408.

Nevin, J. A. (1992). An integrative model for the study ofbehavioral momentum. Journal of the ExperimentalAnalysis of Behavior, 57, 301–316.

Nevin, J. A., & Grace, R. C. (2000). Behavioralmomentum and the law of effect. Behavioural andBrain Sciences, 23, 73–130.

Nevin, J. A., Mandell, C., & Atak, J. R. (1983). Theanalysis of behavioral momentum. Journal of theExperimental Analysis of Behavior, 39, 49–59.

Received January 23, 2008Final acceptance July 28, 2009Action Editor, Iser DeLeon

RESISTANCE TO DISRUPTION 367