Table Of ContentJOURNALOFAPPLIEDBEHAVIORANALYSIS 2009, 42, 171–176 NUMBER1 (SPRING2009)
USING PERCENTILE SCHEDULES TO INCREASE EYE CONTACT IN
CHILDREN WITH FRAGILE X SYNDROME
SCOTT S. HALL, NATALEE P. MAYNES, AND ALLAN L. REISS
STANFORDUNIVERSITYSCHOOLOFMEDICINE
AversiontoeyecontactisacommonbehaviorofindividualsdiagnosedwithFragileXsyndrome
(FXS); however, no studies to date have attempted to increase eye-contact duration in these
individuals. In this study, we employed a percentile reinforcement schedule with and without
overcorrectiontoshapeeye-contactdurationof6boyswithFXS.Resultsshowedthatalthough
aversiontoeyecontactisoftenthoughttobeunamenabletochangeinFXS,itcanbeshapedin
someindividualsusingpercentilescheduleseitheraloneorincombinationwithovercorrection.
DESCRIPTORS: eye contact, Fragile X syndrome, overcorrection, percentile schedules,
shaping
_______________________________________________________________________________
Studies have shown that children who have desiredgoal;Martin&Pear,1992).Traditional
been diagnosed with particular genetic syn- shapingprocedures,however,maybelimitedby
dromes often engage in highly specific forms of inconsistent identification of criterion respons-
aberrant social behavior (Oliver, Demetriades, es. To overcome this problem, Galbicka (1994)
& Hall, 2002; Reiss & Dant, 2003). These has advocated the use of a percentile reinforce-
behavioral phenotypes have been considered to ment schedule, a mathematical rule that
be important examples for investigations of the determines whether an observed response is a
interplay between genes and environment (i.e., criterion response. Percentile schedules may
behavioral neurogenetics research; Kennedy, allow more consistent identification of criteria
Caruso, & Thompson, 2001). for reinforcement during shaping.
ChildrenwithFragileXsyndrome(FXS),the To date, percentile schedules have been
most common known form of inherited employed to shape smoking cessation (Lamb,
developmentaldisability,oftenshowprominent Morral, Kirby, Iguchi, & Galbicka, 2004),
aversion to eye contact (Cohen, Vietze, Sud- computer game playing (Miller & Neuringer,
halter, Jenkins, & Brown, 1991; Hall, DeBer- 2000), and academic task engagement (Athens,
nardis, & Reiss, 2006). For example, in the Vollmer, & Pipkin, 2007), with successful
study by Hall et al., boys with FXS showed this
outcomes in all cases. In the present study, we
aversion 80% of the time during a 5-min social
employed a percentile reinforcement schedule
interaction with an unfamiliar person, despite
to allow more consistent shaping of eye-contact
repeated prompts for eye contact.
duration in children with FXS. Specifically, we
To our knowledge, no studies have specifi-
wanted to examine to what extent duration of
cally attempted to improve eye contact in
eye contact could be increased by employing
children with FXS. One way to train succes-
percentile schedules in time-limited sessions
sively longer durations of eye contact would be
(e.g., 1 to 2 hr) to show that the behavior was
toemployashapingprocedure(e.g.,differential
amenable to behavioral treatment. For those
reinforcement of successive approximations to a
children who were initially noncompliant (i.e.,
did not look at the experimenter when
Address correspondence to Scott S. Hall, Center for
Interdisciplinary Brain Sciences Research, Department of prompted), we implemented an overcorrection
Psychiatry and Behavioral Sciences, Stanford University, procedure similar to that employed by Foxx
401 Quarry Road, Stanford, California 94305 (e-mail:
(1977) to ensure that all children would make
[email protected]).
doi:10.1901/jaba.2009.42-171 contact with the percentile schedule.
171
172 SCOTT S. HALL et al.
METHOD participant eye contact. Successive durations of
eye contact were then used by the computer to
Participants and Setting
rank response durations ordinally and to
Participants were 6 boys, aged 8 to 17 years,
indicate a criterion for reinforcement. The
who had been diagnosed with FXS. The
computer then determined whether a reinforcer
developmental levels of the participants, as
should be delivered. Behavior during overcor-
measured by the Vineland Adaptive Behavior
rection was recorded using a stopwatch.
Scales (Sparrow, Balla, & Cicchetti, 1984),
In six of the 12 sessions, interobserver
ranged from 1 to 5 years. All children were
agreement data for duration of eye contact
recruitedfromthelocalareaandhadundergone
and duration of behavior during overcorrection
previous genetic testing to confirm their
were collected. For each observer, the number
diagnoses of FXS. All participants were able to
ofsecondsofeyecontactorminutesofbehavior
understand simple instructions and could speak
during overcorrection was summed over blocks
in short sentences using at least two words.
of five trials. Total agreement was calculated by
Sessions were conducted in a room contain-
dividing the smaller number by the higher
ing a large desk and several chairs. The
number and converting this ratio to a percent-
participant sat in a chair located against the
age. The mean level of agreement across
backwalloftheroom,andtheexperimentersat
participants was 91% (range, 84% to 96%)
inachair directlyoppositetheparticipant, with
for eye-contact duration and 100% for over-
knees almost touching the participant. An
correction duration.
observer (the second author), who was trained
inobservationalprocedurestogreaterthan80% Experimental Design and Procedure
interobserver agreement, sat 1 m away and to An ABAB design was used to evaluate the
the right of the experimenter. Preferred items effects of percentile schedules on eye-contact
were identified using the protocol described by duration. The experimenter conducted two
Fisher et al. (1992). sessions approximately 1 hr in length on
separate days with each participant. During
Response Measurement and
each session, each participant was exposed to
Interobserver Agreement
two conditions. In the baseline condition, on
Eye contact was defined as the child orienting
each trial the experimenter and participant sat
his head toward the experimenter so that the facing each other while the experimenter held
participant’s eyes looked directly at the exper- an edible item (e.g., Skittles) or play material
imenter’s eyes. Behavior during overcorrection (e.g., Lego) close to one eye and gave the verbal
was defined as the child or experimenter prompt ‘‘Look at me for as long as possible,
moving the child’s head in one of three please.’’ If the participant made eye contact
directions—up, down, or straight ahead—and within 5 s, the observer recorded the duration
maintaining each position for 15 s. of eye contact on the computer. Once eye
The percentile schedule was implemented on contact had been recorded, the computer
a laptop computer. (The program was written simultaneously determined whether the exper-
in Visual Basic, runs under Windows, and is imenter should deliver reinforcement with a
available from the corresponding author.) On random probability of .5. If a reinforcer was
each trial, when the experimenter delivered the scheduled, the computer sounded a beep, and
verbal prompt, participant eye contact was thiswasacuefortheexperimentertodeliverthe
recorded on the computer by the observer. edible item or play material. Both experimenter
The observer pressed a specific key to indicate and observer then said ‘‘Good job, good eye
the onset and again to indicate the offset of contact.’’ If there was no eye contact, no
EYE CONTACT 173
reinforcement was delivered, and the experi- 1977). During overcorrection, the participant
menterwaitedfor10 sbeforeinitiatingthenext was required to move his head in one of three
trial.Thisconditionwasconductedfor20trials directions—up, down, or straight ahead—and
tomaintaineachposturefor15 satatime.The
andallowedthecurrent level of eyecontact and
experimenter prompted the participant at the
the effects of intermittent reinforcement of eye
beginning of each 15-s period by saying either
contact to be assessed, independent of the
‘‘head down,’’ ‘‘head forward,’’ or ‘‘head up.’’
duration of the response.
No prompts for eye contact were delivered. If
In the percentile schedule condition, succes-
the child did not move his head in the required
sively longer durations of eye contact were direction or did not maintain each posture, the
required before the experimenter delivered experimenter stood beside the child and gently
reinforcement. As in the baseline condition, guided the child’s head with his hands. Once
on each trial, the experimenter and participant four postures had been maintained, overcorrec-
tion was completed, and the experimenter then
sat facing each other and the experimenter
waited 10 s before initiating the next trial. In
delivered the verbal and nonverbal prompts
total, the percentile schedule condition was
described above. If the participant made eye
conducted for a maximum of 80 trials or until
contact with the experimenter within 5 s, the
1 hr had passed, whichever came first.
observer recorded the duration of eye contact
on the computer. To determine whether the
RESULTS AND DISCUSSION
response should be reinforced, the computer
program operated a percentile reinforcement Figure 1 shows that for both Bob and Sal,
schedule (see Galbicka, 1994). The percentile eye-contact duration was relatively low during
schedule requires two parameters to be prespec- baseline but increased steadily in duration in
ified: the probability of reinforcing a criterion eachofthepercentileschedulephases.ForBob,
eye contact increased from a mean of 2 s at
response (w), and the number of prior
baseline to a minimum of 4 s in the last five
observations (m) to be included in the calcula-
trials of the second percentile-schedule phase.
tion. The percentile schedule equation, k 5 (m
For Sal, eye contact increased from a mean of
+1)(12w),specifiesthekthrankthatmustbe
2 s in the first baseline condition to 6 s in the
exceeded before the current response becomes a
first percentile-schedule phase and from a mean
criterion response. In this study, we set w 5 .5
of4 sinthesecondbaselinephaseto12 sinthe
and m 5 5. Using the equation above, k 5 3.
second phase in which the percentile schedule
This means that the value of the current
was implemented. However, the data for Jesse
response would be considered a criterion
indicated that only a marginal increase in eye-
response only if it exceeded three of the
contact duration occurred during the first
previous five response values. The criterion
percentile-schedule phase, and that eye contact
response would then have a 50% chance of
did not improve in the second percentile-
being reinforced. The starting criterion for the
schedule phase.
percentile schedule was any duration of eye
For Ben, Jay, and Eric (Figure 2), levels of
contactwiththeexperimenter.Iftheparticipant
eye contact were zero on some trials during the
did not engage in eye contact within 5 s of the percentile schedule condition, which necessitat-
prompts (as occurred for Ben, Jay, and Eric), edtheuseoftheovercorrectionprocedure.All3
the experimenter said, ‘‘You did not look at children received overcorrection during the first
me,’’ in a stern voice and proceeded to 30 trials of the percentile schedule condition,
implement the overcorrection procedure (i.e., but once compliance to the eye-contact prompt
functional movement training) for 60 s (Foxx, was established, this contingency was no longer
174 SCOTT S. HALL et al.
Figure 1. Eye-contact duration (in seconds) for Bob, Jesse, and Sal at baseline and in the percentile schedule
condition. Datapoints representfive-trial means.
required. For Jay, eye contact increased from a that eye-contact duration can be improved in
mean of 0.5 s at baseline to a minimum of 2 s somechildrenwithFXS(Bob,Sal,Jay,andEric),
in the last five trials of each percentile schedule atleastintheshortterm.Futurestudiesshouldbe
phase. For Eric, eye contact increased from a conducted to determine whether parents or
mean of 1 s at baseline to a minimum of 3 s in teachers could implement these training sessions
the last five trials of the second percentile- in school or at home, perhaps leading to longer
schedule phase. For Ben, eye contact did not term improvements in eye contact. Although
appear to improve above baseline in the first Galbicka (1994) suggested ways in which
percentile-schedulephase.However,eyecontact percentile schedules could be implemented in
increased from a mean of 0.5 s at baseline to a naturalistic settings using pencil-and-paper tech-
minimum of 3 s in the last five trials of the niques, it seems likely that efficient implementa-
second phase in which the percentile schedule tion of percentile schedules would, at the very
was implemented. minimum, require a computer program running
Previousinvestigatorshaveconsideredaversion on a handheld computer, which may not be
toeyecontactinFXStobeaprominentfeatureof available to all consumers.
thesyndrome that may bedifficulttoremediate. For 3 of the participants, the presence of the
The data from the current investigation indicate overcorrectionprocedureinthepercentilesched-
EYE CONTACT 175
Figure2. ResultsforBen,Jay,andEricatbaselineandinthepercentilescheduleplusovercorrection.Opencircles
indicate eye-contact duration (in seconds) plotted along the left y axis. Bars indicate duration of the overcorrection
consequence (inminutes) plottedalong the right yaxis. Datapoints represent five-trial means.
ule condition precludes an unbiased assessment FXS is the most common known form of
that the changes observed were due to the inherited developmental disability and is caused
percentile schedules. It is possible that the byamutationtotheFMR1genelocatedontheX
overcorrection procedure would have been effec- chromosome. However, it is unclear how the
tiveindependentofthepercentileschedule;some FMR1mutationplacesanindividualwithFXSat
studieshaveshownthattheeffectofpunishment riskforeye-contactaversion.Itispossiblethatthe
may not reverse when it is withdrawn (e.g., mutation alters the relative reinforcing property
Piazza, Hanley, & Fisher, 1996). The data from ofescapefromsocialstimulationinthesechildren
the current investigation also are limited in that and that aversion to eye contact may allow the
levels of eye contact were increasing in baseline child to escape social stimulation. Once estab-
for several of the participants prior to the lished,itseemslikelythatenvironmentalcontin-
initiation of the percentile schedule, the baseline gencies may further shape and maintain this
condition was implemented substantially less behavior. To date, no functional analysis of
often than the percentile schedule, and the aversion to eye contact in FXS has been
improvements in eye contact were gradual in conducted in which escape as a consequence has
the percentile schedule condition. been manipulated systematically. It is therefore
176 SCOTT S. HALL et al.
possible that such aversion may serve other Lamb,R.J.,Morral,A.R.,Kirby,K.C.,Iguchi,M.Y.,&
Galbicka, G. (2004). Shaping smoking cessation
functions.
using percentile schedules. Drug and Alcohol Depen-
dence,76,247–259.
Martin, G., & Pear, J. (1992). Behavior modification:
REFERENCES
What it is and how to do it (4th ed.). Englewood
Athens, E. S., Vollmer, T. R., & Pipkin, C. C. (2007). Cliffs,NJ:Prentice Hall.
Shaping academic engagement with percentile sched- Miller,N.,&Neuringer,A.(2000).Reinforcingvariabil-
ules.JournalofAppliedBehaviorAnalysis,40,475–488. ity in adolescents with autism. Journal of Applied
Cohen,I.L.,Vietze,P.M.,Sudhalter,V.,Jenkins,E.C., BehaviorAnalysis,33,151–165.
& Brown, W. T. (1991). Effects of age and Oliver,C.,Demetriades,L.,&Hall,S.(2002).Effectsof
communication level on eye contact in fragile X environmental events on smiling and laughing
males and non-fragile X autistic males. American behavior in Angelman syndrome. American Journal
JournalofMedical Genetics,38,498–502. onMentalRetardation, 107,194–200.
Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L. Piazza, C. C., Hanley, G. P., & Fisher, W. W. (1996).
P., Owens, J. C., & Slevin, I. (1992). A comparison Functional analysis and treatment of cigarette
of two approaches for identifying reinforcers for pica. Journal of Applied Behavior Analysis, 29, 437–
personswithsevereandprofounddisabilities.Journal 449.
ofAppliedBehavior Analysis,25,491–498. Reiss, A. L., & Dant, C. C. (2003). The behavioral
Foxx, R. M. (1977). Attention training: The use of neurogeneticsoffragileXsyndrome:Analyzinggene-
overcorrectionavoidancetoincreasetheeyecontactof brain-behavior relationships in child developmental
autistic and retarded children. Journal of Applied psychopathologies. Development and Psychopathology,
BehaviorAnalysis,10, 489–499. 15,927–968.
Galbicka,G.(1994).Shapinginthe21stcentury:Moving Sparrow, S. S., Balla, D. A., & Cicchetti, D. V. (1984).
percentile schedules into applied settings. Journal of The Vineland adaptive behavior scales. Circle Pines,
AppliedBehavior Analysis, 27,739–760. MN:American GuidanceService, Inc.
Hall,S.,DeBernardis,M.,&Reiss,A.(2006).Socialescape
behaviorsinchildrenwithfragileXsyndrome.Journal
ofAutismandDevelopmentalDisorders,36,935–947.
Kennedy, C. H., Caruso, M., & Thompson, T. (2001).
Experimental analyses of gene–brain–behavior rela- Received February 23,2007
tions: Some notes on their application. Journal of Final acceptance March19, 2007
AppliedBehavior Analysis, 34, 539–549. Action Editor,James Carr
