Professional Documents
Culture Documents
ALAN M . GROSS
DANIEL A. WOJNILOWER
Emory University
The present paper examines whether children can learn to effectively manage
their own behavior. Focusing on self-reinforcement strategies, studies in which
children were reported to be successful and unsuccessful in their attempts to alter
their own responding are examined. Special attention is given to the role played
by external treatment, mediator-imposed contingencies on the effectiveness of
self-management programs. Conclusions regarding the necessary conditions for
successful self-control in children are presented.
This research was supported by grant #HD17090 from the National Institutes of Health.
The authors thank Thomas A. Brigham for his helpful comments on an earlier draft of this
manuscript. Requests for reprints should be sent to Alan M. Gross, Department of Psy-
chology, Emory University, Atlanta, GA 30322.
50 1 0005-7894/84/0501-051451.00/0
Copyright1984by Associationfor Advancementof BehaviorTherapy
AUrightsof reproductionin any formreserved.
502 GROSS A N D WOJNILOWER
adults can spend more time teaching other skills (O'Leary & Dubey, 1979).
Additionally, it has been suggested that teaching children to control their
own behavior might result in stronger maintenance effects than those
observed when contingencies have been administered by external agents.
As a behavior change strategy, self-control training with children has
involved a variety of procedures (e.g., self-monitoring, self-evaluation,
self-reinforcement). However, self-reinforcement is generally recognized
as the component most integral to the development of self-control. Self-
reinforcement refers to the process by which individuals, in the relative
absence of controlling influences regulated by others, increase and main-
tain their own behavior both by freely imposing certain contingencies for
the self-administration of reinforcing stimuli and by exhibiting full control
over available reinforcers (Jones, Kazdin, & Nelson, 1977). In an applied
setting this usually translates into teaching children the following set of
responses: monitoring their own behavior, establishing behavior perfor-
mance criteria (response contingency), selecting a reinforcer that normally
is freely available, and taking the reinforcer only following the perfor-
mance of the target behavior.
Numerous studies on the effectiveness of self-reinforcement procedures
with children have been conducted. In a recent review of this area, O'Leary
and Dubey (1979) concluded that these procedures can produce effects
equivalent to, or better than, those achieved when contingencies are ex-
ternally administered. However, further scrutiny of these studies reveals
that the reported examples of children's self-control may not be as free
of externally administered contingencies as this statement implies (Gross
& Drabman, 1982). In fact, externally imposed contingencies may be a
critical variable in successful self-management with children.
The purpose of the present paper is to examine the necessary conditions
for effective self-directed behavior change in children. Focusing on self-
reinforcement strategies, studies in which youngsters successfully and un-
successfully altered their own behavior through self-administered contin-
gencies will be explored in order to determine the relative contribution
of mediator-controlled contingencies in children's self-control.
Efficacy of Self-Reinforcement
There exist today a large number of studies examining the effectiveness
of self-reward as a behavior change strategy with children. Typically,
elementary school children serve as subjects and are afforded the oppor-
tunity to reward themselves for appropriately engaging in both academic
and general classroom behavior.
Several studies have investigated the relative efficacy of self-reinforce-
ment and external reinforcement systems. Glynn (1970) monitored the
daily history test performance of four classes of ninth-grade girls. Every
day, the teacher read aloud the test answers, and the girls noted on the
top of their papers the number of problems they scored correctly. FoP
lowing this phase, one group of youths received token reinforcers at the
rate of one for every four correct answers (total of 5 tokens possible per
test). A second group awarded themselves one to five tokens based on
SELF-CONTROL 50 3
what they thought their test performance merited. The third class of girls
was yoked to the teacher-determined token group and given tokens based
on their performance. The remaining subjects did not receive tokens. It
was reported that students in the teacher- and self-delivered token groups
exhibited the largest improvements in academic performance. Moreover,
there was no significant difference in performance between the self-deliv-
ered and externally delivered token subjects. Bolstad and Johnson (1972),
Felixbrod and O'Leary (1973, 1974), and Frederiksen and Frederiksen
(1975) also have reported self-reinforcement to be equally effective as
externally delivered reinforcers in altering behavior.
Ballard and Glynn (1975) improved story writing in elementary school
children using self-reinforcement. In a multiple baseline design, the num-
ber of sentences, number of action words, and number of describing words
used in the children's daily story-writing exercise were monitored. The
children were then instructed to self-record the occurrence of these target
behaviors. This phase was followed by the sequential introduction of self-
reinforcement across target behaviors. Although self-recording had little
effect on writing behavior, the addition of self-reinforcement resulted in
substantial increases in the number of sentences written as well as in the
usage of action and describing words. Additionally, independent ratings
of the children's essays reflected a large improvement in story-writing
ability.
Self-reinforcement has also been shown to be more effective than self-
imposed response cost in altering the reading behavior of second-graders
(Humphrey, Karoly, & Kirschenbaum, 1978). In a counterbalanced ABAC
design, the children either self-delivered tokens for correct responses or
fined themselves tokens for errors on reading workbook assignments.
Although both self-management procedures produced increases in work
accuracy, the children performed best during the self-reinforcement con-
dition. In contrast, Kaufman and O'Leary (1972) found self-imposed
response cost and self-reinforcement to be equally effective in decreasing
disruptive behavior in adolescents in a psychiatric hospital school.
In addition to the modification of academic responding, self-reinforce-
ment has been used to alter children's disruptive behavior. Bolstad and
Johnson (1972) monitored "on-task" behavior in a second-grade class-
room. Following baseline, the youngsters received training in self-record-
ing, self-evaluation, and self-reinforcement. At various intervals while the
children worked, a tape recorder sounded a tone. The children were told,
upon hearing this tone, to note on a data sheet if they were on task.
Subsequently, they awarded themselves rewards based on the number of
intervals on task. This procedure resulted in a large decrease in disruptive
behavior. Glynn and Thomas (1974) and Glynn, Thomas, and Shee (1973)
have reported similar findings.
The Role of External Treatment Mediators
As can be seen from the previous material, numerous investigators
have suggested that self- and external reinforcement systems in children
seem to be equally effective in exerting positive influence upon academic
504 GROSS AND WOJNILOWER
and disruptive behavior. However, Bass (1972) has argued that self-re-
inforcement necessitates that children both understand and believe they
have complete control over all contingencies and can thereby "cheat" or
engage in noncontingent self-reinforcement behavior without bringing
about any other contingent external consequences of either a reinforcing
or punishing nature. If these conditions are not met, the child is not
exhibiting self-control, but is environmentally controlled. An important
question is whether the majority of self-reinforcement studies with chil-
dren meet these criteria.
Edgar and Clement (1980) had four fourth-grade boys meet with a
teacher in a tutoring class three times weekly for a 1-hour period. During
two week-long baseline phases, students did not receive any programmed
rewards. In the treatment conditions, the order of presenting self- and
teacher-imposed contingency systems was counterbalanced across stu-
dents. Regardless of the condition in effect, a tape recorder emitted beeps
on a variable interval schedule of 2 minutes. These beeps signaled when
rewards were to be either self-delivered by the student or by the teacher
provided that the student was engaged in the appropriate target behavior
at that moment. At the conclusion of each class session, the student
recorded the number of points appearing on the point counter on his desk.
Points were accumulated across sessions and could be traded for back-
up reinforcers at the end of each 2-week treatment phase.
The authors concluded that self-reinforcement was more effective than
teacher-controlled reinforcement in increasing academic behaviors. How-
ever, scrutiny of the procedure employed reveals that environmental fac-
tors may have influenced the effectiveness of self-reinforcement proce-
dures. During baseline, for example, children were paid one dollar for
simply attending all three tutoring sessions in any given week. Further,
during self-reinforcement phases, students were able to engage in self-
reinforcement only when beeps were emitted from a tape recorder. This
constraint, in conjunction with the teacher's ease in observing whether
or not reinforcers were being administered contingently, may have influ-
enced the students to self-reinforce only at appropriate times. Finally,
inaccurate self-administration of token points was prevented by the teach-
er watching each student record the points. Taken together, environmental
variables all but eliminated the children's ability to discontinue their work
and self-reinforce noncontingently. Similar studies have encountered many
of the same problems (Clement et al., 1978; Fantuzzo & Clement, 1981;
Fantuzzo, Harrell, & McLeod, 1979; Glynn, 1970; Lovitt & Curtiss, 1969).
Bolstad and Johnson (1972) examined the relative efficacy of self- and
external contingency systems upon the disruptive behavior of first- and
second-grade children. Students who exhibited high rates of such behavior
were either placed in a control condition or assigned to one of three
experimental groups after prebaseline measures of classroom disruptive-
ness were obtained. Those children in the experimental conditions were
exposed to a reinforcement system in which an adult evaluated the chil-
dren's behavior and then dispensed reinforcers contingent upon that eval-
SELF-CONTROL 505
uation. In the next phase, pupils in two self-regulation groups were taught
to record their own disruptive behavior and self-reinforce based upon
scientific criteria, while the third group remained under adult-controlled
reinforcement.
This procedure resulted in large decreases in disruptive behavior for
children in both the self- and adult-regulated conditions. However, prior
exposure to an external contingency confounds these findings for the self-
reinforcement group. Other sources of environmental influence which
may have contributed to the effectiveness of self-reinforcement included
informing children that they should refrain from self-administering points
unless they emitted less than a specified number of disruptive behaviors,
penalizing students if their self-evaluations did not match observer eval-
uations, and observing pupils as they took prizes from the reward box.
Finally, the presence of observers who had previously dispensed points
throughout the self-regulatory phases may have served as cues for children
to employ particular self-reinforcement standards. Many of these same
confounding environmental variables can be found in other studies of
self-control in children (Drabman, Spitalnik, & O'Leary, 1973; Freder-
iksen & Frederiksen, 1975; Glynn et al., 1973; Kaufman & O'Leary, 1972;
Santogrossi, O'Leary, Romanczyk, & Kaufman, 1973; Wood & Flynn,
1978).
Felixbrod and O'Leary (1973, 1974) compared the effects of self-de-
termined and externally imposed reinforcement systems upon math pro-
ductivity and on-task behavior of lower elementary school pupils. Chil-
dren were assigned to self-, external-, and no-reinforcement conditions.
Children in the self-reinforcement group were allowed to choose their
own performance standard, and this standard was applied to a matched
counterpart in the external reward condition.
The authors reported the two reinforcement groups to be equally effi-
cacious as well as significantly better than the control condition at pro-
ducing increases in productivity. However, analysis of the methodology
employed reveals that children in the self-reinforcement condition were
unable to meet the criteria for self-reinforcement; that is, they were unable
to self-administer reinforcers noncontingently. More specifically, the chil-
dren circled their self-selected performance criteria on a piece of paper
prior to working on the math problems and then had to hand in their
answers for grading before rewards were delivered. Teacher-imposed re-
strictions on noncontingent self-reinforcement are also found in studies
by Ballard and Glynn (1975), Brownell, Colletti, Ersner-Hershfield,
Hershfield, and Wilson, (1977), Dickerson and Creedon (1981), and Hum-
phrey et al. (1978).
Finally, Wall (1982) examined the effects of systematic self-monitoring
and self-reinforcement components upon the test performance of fourth-
grade children. Four different classes participated and each was assigned
to a different condition. After three baseline sessions, self-reinforcement
procedures were introduced into two of the groups. One class had pre-
viously engaged in systematic self-monitoring and continued to do so.
506 GROSS A N D W O J N I L O W E R
et al., 1979) may well have functioned as a cue both to increase children's
motivation to please the experimenter and to suggest that compliance
with experimental procedures was expected.
Demand characteristics cannot be eliminated from self-control re-
search. Nevertheless, they may play a significant role in the outcome of
such studies and therefore must be monitored. Two ways o f assessing the
potential ability of demand characteristics to alter behavior are to utilize
a postexperimental questionnaire or a preinquiry technique (Orne, 1969).
In a postexperimental questionnaire, subjects are interviewed after the
experiment is over. Self-control researchers could use this procedure to
obtain "subjects'" perceptions regarding the extent to which they felt free
o f environmental constraints on their behavior (Wilson & O'Leary, 1980).
For example, children could be asked if they thought they would get in
trouble if they stopped working and took a reward or what they thought
would happen if they took too much reward or simply took rewards
without doing any o f the suggested task.
Alternatively, a preinquiry technique employs one group of subjects to
predict the experimental results without actually going through the ex-
perimental process. Preinquiry subjects serve as a regular control group,
performing on the dependent measures in the same fashion as would
subjects who actually participated in the procedures. These preinquiry
findings are then compared to data obtained from subjects actually ex-
posed to experimental methods. The smaller the difference between these
two sets of findings, the more likely it will be that demand characteristics
influenced subject behavior.
While relatively easy to implement, neither of these procedures has
been employed in self-control research with children. However, recent
investigations have begun to examine the differential effects of demand
characteristics often found in self-control procedures (Jones & Evans,
1980; Jones & Ollendick, 1979; Speidel & Tharp, 1980). In these studies,
external controls are purposely included within the experimental meth-
odology and then are taken into account when interpreting the data. While
these investigations have only attended to the more general types of ex-
ternal demand which have influenced children's self-controlling behavior,
they most certainly are a step in the right direction.
Generalization
Self-management skills have frequently been advocated as a method of
enhancing generalization (Stokes & Baer, 1977). While 15 classes of gen-
eralization have been enumerated (Drabman, Hammer, & Rosenbaum,
1979), time generalization (maintenance) has received the most attention
in self-control research with children.
Drabman et al. (1973) developed a procedure to teach and maintain
accurate self-reinforcement. In a classroom setting, dn-task behavior was
monitored. Following baseline, the teacher evaluated and reinforced (to-
kens) each child for appropriate classroom responding. The youngsters
were then asked to rate their own behavior and to self-reinforce at the
SELF-CONTROL 509
end o f specific time intervals. They were told that if their evaluation
matched that of the teacher, then they would earn bonus tokens. Over
time, the matching of teacher and student ratings was faded by gradually
selecting fewer children each day for matching. When matching was dis-
continued, the children independently rated their responding and rein-
forced themselves with tokens based on these evaluations. The procedures
resulted in accurate self-reinforcement which was maintained over 12
days. Robertson, Simon, Pachman, and Drabman (1979)and Turkowitz,
O'Leary, and Ironsmith (1975) have replicated these findings.
Maintenance of behavior change following exposure to self-reinforce-
ment training has also been observed by Wood and Flynn (1978). Pre-
delinquent youths in a group home were taught a number of room-cleaning
responses. Following baseline and instruction-to-clean phases, a token
system based on experimenter evaluation of behavior was implemented.
Half o f the youngsters were then taught to monitor the occurrence of the
target behavior and self-reinforce contingent on appropriate performance.
The others continued to receive tokens from the experimenters based
upon their performance evaluations. When contingencies on room clean-
ing were removed, youths who had experienced training to self-administer
reinforcers maintained high levels of on-task behavior over a 60-day
period. Youngsters exposed only to externally administered tokens showed
a substantial decline in room-cleaning behavior.
Of the numerous self-control studies with children, few present data on
time generalization. The four studies reported here suggest that self-con-
trol training may facilitate the maintenance of behavior change. However,
follow-up periods were relatively short (12-60 days). Moreover, many of
the environmental influences affecting self-control behavior cited earlier
were also operating in these investigations and very well could have con-
tributed to the long-term effects noted. More important, other investi-
gators have failed to observe maintenance (Felixbrod and O'Leary, 1974;
Santogrossi et al., 1973). Although the evidence seems favorable, the issue
is by no means clear. Further research with longer follow-up periods is
needed.
CONCLUSIONS
Environmental sources of control appear to exist in all demonstrations
of children's self-control. While the number of environmental variables
observed may vary, there do appear to be specific environmental con-
ditions that must exist if children are to manage their own behavior
successfully. A contingency on accuracy of self-reinforcement must be
applied. If children discriminate that there are no aversive contingencies
for noncontingent self-reinforcement, they will rate their behavior as
appropriate and self-deliver reinforcers regardless of whether they are
performing the target behavior. This creates a serious problem because
it results in contingent reinforcement for inappropriate behavior.
The role of environmental variables in children's self-control is not
denied by investigators in this area. They argue (e.g., Thoresen & Ma-
510 GROSS AND WOJNILOWER
honey, 1974) that it is the relative frequency (absence), timing, form, and
magnitude o f such factors that differentiate self- from externally controlled
behavior. As such, the m e t h o d o l o g y used to study self-control in children
has generally involved attempts to eliminate as m a n y externally controlled
contingencies as possible and then to observe whether a child could suc-
cessfully self-administer a reinforcement contingency. It is clear that even
if all obvious, immediate, externally administered reinforcement contin-
gencies could be controlled (and this literature review suggests that this
is not possible), such control would not negate their effect either on be-
havior or on a child's learning history (Skinner, 1953). Moreover, this
view also raises the question o f the n u m b e r o f external contingencies that
can exist while still allowing us to call the child's self-administration o f
a reinforcer a d e m o n s t r a t i o n o f self-control. Unfortunately, this appears
to be a question m o r e suited to a philosophical analysis o f behavior rather
than an experimental analysis o f behavior.
In an early discussion o f self-control, Skinner (1953) m a d e an i m p o r t a n t
distinction between controlling responses and controlled responses. He
suggested that a controlling response manipulated variables in a m a n n e r
that resulted in an alteration o f the probability o f occurrence o f a future
behavior. He called the affected future behavior the controlled response.
W h e n individuals emit a controlling response in order to alter the behavior
o f another person, it is labeled behavior modification. However, when
individuals emit controlling responses in order to modify their own re-
sponding, it is labeled self-control. The apparent difference in these two
situations depends from where the controlling response is seen to origi-
nate.
Identifying a controlling response in a behavior chain is not always a
simple task. More i m p o r t a n t the segment o f the response chain that is
observed can influence the process o f determining the controlling re-
sponse. This review suggests that children who exhibit i m p r o v e m e n t in
academic performance following an intervention in which they are taught
to self-monitor, self-evaluate, and self-reward for the display o f appro-
priate academic behavior are frequently seen as exhibiting self-control.
The children's self-monitoring, self-evaluation, and self-reward behaviors
are considered controlling responses affecting their school work. However,
closer inspection o f this situation reveals that in the absence o f a teacher-
imposed accuracy contingency, the children would award themselves rein-
forcers noncontingently. Moreover, in order for the children to use their
self-control skills effectively, the teacher must m o n i t o r their behavior.
Does the requirement that there be a teacher-mediated contingency sug-
gest that the teacher is displaying the controlling response and the chil-
dren's b e h a v i o r is not self-controlled? After all, it is the children's con-
trolling response (noncontingent self-reward) that results in the occurrence
o f the accuracy contingency.
The example presented illustrates that attempting to identify a con-
trolling response as originating from the responding individual or from
the e n v i r o n m e n t is an arbitrary decision at best. A more fruitful approach
SELF-CONTROL 5 11
REFERENCES
BaUard, K. D., & Olynn, T. (1975). Behavioral self-management in story writing with
elementary school children. Journal of Applied Behavior Analysis, 8, 387-398.
Bandura, A. (1971). Vicarious and self-reinforcement processes. In R. Glaser (Ed.), The
nature of reinforcement (pp. 228-278). New York: Academic.
Bass, B. A. (1972). Reinforcement history as a determinant of self-reinforcement. The
Journal of Psychology, 81, 195-203.
Bolstad, O. D., & Johnson, S. M. (1972). Self-regulation in the modification of disruptive
classroom behavior. Journal of Applied Behavior Analysis, 5, 443-454.
Brownell, K. D., Colletti, G., Ersner-Hershfield, R., Hershfield, S. M., & Wilson, G. T.
(1977). Self-control in school children: Stringency and leniency in self-determined and
externally imposed performance standards. Behavior Therapy, 8, 442-455.
Clement, P. W., Anderson, E., Arnold, J,, Butman, R., Fantuzzo, J., & Mays, R. (1978).
Self-observation and self-reinforcement as sources of self-control in children. Biofeed-
back and Self-Regulation, 3, 247-267.
Dickerson, E. A., & Creedon, C. F. (1981). Self-selection of standards by children: The
relative effectiveness of pupil-selected and teacher-selected standards of performance.
Journal of Applied Behavior Analysis, 14, 425-433.
Drabman, R. S., Hammer, D., & Rosenbaum, M. S. (1979). Assessing generalization in
behavior modification with children: The generalization map. Behavioral Assessment,
1, 203-219.
Drabman, R.S.,Spitalnik, R.,&O'Leary, K.D. (1973). Teaching self-eontrol to disruptive
children. Journal of Abnormal Psychology, 82, l O- 16.
Edgar, R., & Clement, P. (1980). Teacher-controlled and self-controlled reinforcing with
underachieving black children. Child Behavior Therapy, 2(4)~33-56.
Fantuzzo, J. W., & Clement, P. W. (1981). Generalization of the effects of teacher- and
self-administered token reinforcers to nontreated students. Journal of Applied Behavior
Analysis, 14, 435-447.
Fantuzzo, J., Harrell, K., & McLeod, M. (1979). Across-subject generalization of attending
SELF-CONTROL 5 13
RECEIVED:December 2, 1983
FINAL ACCEPTANCE:April 12, 1984