Developing Helping Behavior in Young Children Through Multiple Exemplar Training

Abstract

Research has demonstrated that engaging in helping behavior is highly valued by children and adults and has diverse benefits for the recipient, helper, and larger group. Not surprisingly, raising children who exhibit pro-social behavior such as helping others is a central concern for parents and societies. However, the learning process that leads to the emergence of helping remains understudied. The current study examined the establishment of generalized helping behavior in young, typically developing children, in a context in which helping competed with ongoing toy play. Additionally, we examined the emergence of vocalizations about behavior that suggest the adoption of a socially conventional rule that helping is a good thing to do. Generalized helping was initially established through multiple exemplar training, with some participants also receiving rule instruction and behavioral feedback. Generalized helping emerged across all participants, and 2 of 3 participants made vocalizations demonstrating a behavioral rule that helping is good.

Full text

Developing Helping Behavior in Young Children Through Multiple
Exemplar Training
George H. Noell, Jeanne M. Donaldson, Kristin A. Gansle, Rachel L. Bradley,
Aijah K. Goodwin, Emma Larson, Philip R. Richard, James J. Upright,
Catherine R. Lark, Katherine L. Moore, and Ashley E. Bordelon
Louisiana State University
Research has demonstrated that engaging in helping behavior is highly valued by
children and adults and has diverse benefits for the recipient, helper, and larger group.
Not surprisingly, raising children who exhibit pro-social behavior such as helping
others is a central concern for parents and societies. However, the learning process that
leads to the emergence of helping remains understudied. The current study examined
the establishment of generalized helping behavior in young, typically developing
children, in a context in which helping competed with ongoing toy play. Additionally,
we examined the emergence of vocalizations about behavior that suggest the adoption
of a socially conventional rule that helping is a good thing to do. Generalized helping
was initially established through multiple exemplar training, with some participants
also receiving rule instruction and behavioral feedback. Generalized helping emerged
across all participants, and 2 of 3 participants made vocalizations demonstrating a
behavioral rule that helping is good.
Keywords: generalization, helping, multiple exemplar training, rule governed behavior
Raising children who behave in ethical and
prosocial ways is a paramount concern for most
parents (Smith, Noh, Rizzo, & Harris, 2017).
Although there are religious and cultural varia-
tions in parents’ specific expectations, demon-
strating concern for others (e.g., helping, altru-
ism, kindness, generosity) is commonly highly
valued. Young adults also describe the develop-
ment of greater consideration for others as an
important developmental outcome (Arnett, 2003).
Helping is one of the important behaviors that
demonstrate concern for others. Parents and teach-
ers typically value helping behavior in children
(Gralinski & Kopp, 1993;Pettygrove, Hammond,
Karahuta, Waugh, & Brownell, 2013). Addition-
ally, children and adolescents evaluate helping
others as virtuous (Kahn, 1992).
Helping behavior has been found to benefit
not only recipients, but also helpers and the
broader community in a variety of ways (see
Eisenberg, Fabes, & Spinrad, 2006, for a re-
view). Implementation of a curriculum to pro-
mote pro-social behavior in schools that in-
cluded helping behavior has been found to
result in less conflict, greater agreeableness, and
less physical aggression within the group ex-
posed to the curriculum (Caprara, Luengo Ka-
nacri, Zuffianò, Gerbino, & Pastorelli, 2015). It
is important to recognize that Caprara et al.
studied the impact of a comprehensive curricu-
lum on adolescents who engaged in aggressive
behavior, and as a result, it is not possible to
isolate the importance of the helping compo-
nent. Plötner, Over, Carpenter, and Tomasello
(2015) suggested that helping peers may be a
means of establishing collaboration between
them. This would be valuable given their find-
ings that children were more likely to help,
trust, and like their collaborators. At an individ-
This article was published Online First January 17, 2019.
George H. Noell, Jeanne M. Donaldson, Kristin A.
Gansle, Rachel L. Bradley, Aijah K. Goodwin, Emma Lar-
son, Philip R. Richard, James J. Upright, Catherine R. Lark,
Katherine L. Moore, and Ashley E. Bordelon, Department
of Psychology, Louisiana State University.
Correspondence concerning this article should be ad-
dressed to George H. Noell, Department of Psychology,
Louisiana State University, Baton Rouge, LA 70803.
E-mail: gnoell@lsu.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Behavioral Development
© 2019 American Psychological Association 2019, Vol. 24, No. 1, 6–17
1942-0722/19/$12.00 http://dx.doi.org/10.1037/bdb0000083
6

ual level, children who help other students ac-
ademically (tutoring) have been found to make
academic, social, and self-concept gains (Dineen,
Clark, & Risley, 1977;Franca, Kerr, Reitz, &
Lambert, 1990). Engaging in acts that benefit
others has been found to be associated with
diverse benefits for the child emitting helping
behavior, including improved social relation-
ships and positive psychological functioning
(Caprara et al., 2015;Eisenberg et al., 2006;
Williamson, Donohue, & Tully, 2013).
Although helping is a pervasive developmen-
tal expectation for children, relatively little re-
search has examined what developmental and
learning experiences lead to its emergence. In-
terestingly, the pervasive evidence of helping
by children at the group level in laboratory
studies and the increasing prevalence of helping
as children age has led some developmental re-
searchers to suggest that helping may be a be-
havior that emerges as a result of preparedness,
rather than learning (Pettygrove et al., 2013;
Warneken & Tomasello, 2013). Research has
suggested that children’s helping is influenced
by their environment. For example, children are
more likely to exhibit helping behaviors after
watching a model help (Pettygrove et al., 2013;
Williamson et al., 2013).
The studies described above are limited in
their ability to inform a scientific understanding
of helping at the individual level for several
reasons. First, they consistently focused on
group means in attempts to estimate population
parameters. As a result, the data may represent
the behavior of groups of children, while pro-
viding little insight into how individual children
behave (Branch, 1999). Second, the emphasis
on helping as a prepared behavior (e.g., Petty-
grove et al., 2013) that simply emerges through
the passage of time places analysis of helping
outside of a scientific context and does not
consider the substantial variation in helping be-
havior within and across children. Finally, the
emphasis on examination of laboratory condi-
tions that are distinctly unlike children’s every-
day lives that occasion (or fail to occasion)
helping behavior in young children results in a
literature that has limited value in understand-
ing how to teach children to help as part of
development.
In contrast with the broader developmental
literature, the behavior-analytic literature exam-
ining prosocial behaviors in young children has
examined behavior at the level of the individual
as well as those learning experiences and con-
tingencies that occasion prosocial behaviors.
These have included prosocial behaviors such
as self-control, sharing, honesty, and respecting
others’ rights (Beaulieu & Hanley, 2014;Dunkel-
Jackson, Dixon, & Szekely, 2016;Marzullo-
Kerth, Reeve, Reeve, & Townsend, 2011). Be-
havior-analytic approaches to pro-social
behavior have focused on the creation of a
learning history for individuals that fosters the
emission of specific prosocial behaviors such as
sharing (e.g., Marzullo-Kerth et al., 2011).
Surprisingly, given its social significance, the
behavior-analytic study of helping has been
limited. Research has demonstrated that a treat-
ment package that included progressive prompt-
ing, fading, and social gratitude was sufficient
to teach three adolescent boys with Autism
Spectrum Disorder (ASD) to emit helping re-
sponses following three different statements in-
dicating an adult needed assistance (Harris,
Handleman, & Alessandri, 1990). In addition,
some generalization to other settings, individu-
als, and statements signaling the need for help
were observed. Subsequent research examining
teaching helping behavior to children with ASD
used a multicomponent treatment package ar-
ranged within a multiple exemplar training (MET)
program (Reeve, Reeve, Townsend, & Poulson,
2007). MET is a teaching strategy that is de-
signed to promote generalization by training
across sufficient exemplars to establish a di-
verse set of responses. Reeve et al. trained four
verbal discriminative stimuli (S
D
s) that signaled
the need for help while assessing for general-
ization to four untrained categories. Training
included video models, verbal prompting, mo-
tor prompting, praise, and tokens that could be
exchanged for rewards. All four participants
learned to emit trained helping behaviors and
exhibited high levels of generalization to un-
trained exemplars.
Research extending the application of MET
to the developmental milestones of typically
developing children is quite limited, but gener-
ally encouraging. Extensions of MET to chal-
lenges that can emerge during normative devel-
opment have shown initial promise both for
addressing practical challenges and for testing
theoretical propositions. For example, the same
study demonstrated that MET is an effective
way to teach children relational responding
7HELPING BEHAVIOR
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(e.g., is one amount more or less than another)
and provided evidence that arbitrary compara-
tive responses can emerge as operant responses
(Berens & Hayes, 2007). Similarly, Gilic and
Greer (2011) demonstrated that MET was an
effective way to establish bidirectional higher-
order language operants as well as providing
some insight into how those abilities may
emerge. MET also has been demonstrated to be
effective in establishing perspective-taking
skills and derived tact relation skills in a second
language (Rosales, Rehfeldt, & Lovett, 2011;
Weil, Hayes, & Capurro, 2011). Although each
of these studies might be described as a produc-
tive initial inquiry into an area of investigation,
they are sufficiently heterogeneous in their tar-
gets, participants, and methods that few themes
emerge across studies. The area in which they
are consistent is in demonstrating that MET has
been an effective procedure for evoking the
emergence of untaught relations, whether that is
generalization across stimuli, responses, or
combined generalization.
The present study sought to extend existing
research by analyzing the emergence of gener-
alized helping behavior in young children of
typical development. This target was selected
based on the broadly endorsed societal value
that helping others is valuable and conforming
to the socially endorsed rule that “you should
help because it is good.” Additionally, we were
interested in several specific issues as they re-
late to MET targeting helping for young chil-
dren. First, is MET an efficient approach for
teaching typically developing children social
skills they do not presently emit? Second, is it
possible to obtain response and stimulus gener-
alization with few exemplars? We used fewer
exemplars than prior MET studies that taught
prosocial behaviors such as rule following, em-
pathy, and sharing to children with ASD (Mar-
zullo-Kerth et al., 2011;Reeve et al., 2007;
Sivaraman, 2017;Wymer, Tarbox, Beavers, &
Tullis, 2016). Third, would MET be effective in
establishing a social skill in a context in which
children were already engaged with an activity
they had chosen (toy play)? Prior research with
typically developing children has demonstrated
the effectiveness of MET in contexts in which
reinforcement for competing behaviors was
purposefully minimized (Berens & Hayes,
2007;Gilic & Greer, 2011;Rosales et al., 2011;
Weil et al., 2011). By contrast, our interest was
in the emergence of helping behavior in a con-
text in which the consequences of helping were
socially conventional (gratitude), and in compe-
tition with other consequences the child was
likely to find reinforcing in the natural environ-
ment (i.e., continuing access to the toys the
child selected).
Method
Participants and Setting
Three typically developing 3-year-old boys,
Jack, Macon, and Andy, participated in the
study. All participants exhibited age-appropri-
ate language and social and motor skills based
on informal observations at entry into the study,
and all were able to follow two-step directions,
provided there was not a distraction present.
Participants were enrolled in a university-
affiliated early childhood education laboratory
preschool center and were nominated by the
center director to participate in the study. Pa-
rental consent and child assent were obtained
prior to child participation in any research ac-
tivities. Sessions were conducted in large pre-
school classrooms containing child-sized furni-
ture at times when the classrooms were not in
use.
Materials
Children were provided with an array of toys
from which to choose at the beginning of each
session. The array included interlocking blocks,
plastic toys, crayons and coloring books, wooden
inset puzzles, Play-Doh, and iPad mini 4s. The
iPads were loaded with an array of games that
were appropriate for 3-year-old children. The
iPads were also used to present the videos (de-
scribed below) to the participants. In addition to
the toys available to the children, several props
were needed to arrange the helping situations
(described below). The props included playing
cards, picture flash cards, first-aid supplies, sil-
verware, office supplies, cups, paper towels,
small bins with lids, papers, and clipboards.
Four video recordings were prepared for the
study. Each video depicted a different adult
sitting at a table working with materials (e.g.,
papers, a clipboard, cards, etc.). A young boy
(age 5– 8 years) was seated approximately 1 m
away from the adult and was either playing with
8 NOELL ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

a toy or reading a book. Every 30 s during the
video, the adult made a verbalization that was
drawn from either the helping S
D
s or the control
vocalizations (described below). A total of 10
vocalizations were made during the 5-min
video, with vocalizations equally divided be-
tween S
D
s and control vocalizations. Each of
the five S
D
s were drawn from different exem-
plars (see the next section).
Response Definitions, Data Collection
Procedures, and Interobserver Agreement
The primary target behavior was the child
initiating a relevant helping behavior within 15
s of the occurrence of a verbal S
D
. Helping
behavior was defined as a motor response that
would help resolve the problem specified by the
S
D
(e.g., picking up for spills or getting the
first-aid kit for injuries). Six groups of S
D
s were
used in the study. These groups of S
D
s, helping
S
D
exemplars, were statements indicating that
the experimenter either had injured her/himself,
had lost something, had mixed up materials,
could not open something, had spilled some-
thing, or had a mess to clean up. If children
asked for permission to help, they were told
“yes” (see Table 1). If children asked how they
could help, the experimenter described a simple
motor action the child could carry out to help
(e.g., “please bring me the first-aid kit”). These
interactions did not influence the criteria for
scoring help as having occurred.
In addition to recording the occurrence of
helping behaviors for discriminative vocaliza-
tions by the experimenter, observers also re-
corded the occurrence of offers to help within
10 s and attempts to help within 15 s of control
vocalizations (S
⌬
s). Control vocalizations were
scripted to fall into six categories that included
statements about how the experimenter felt,
comments on the weather, statements about
how well the task the experimenter was engaged
in was proceeding, observations about the task
the experimenter was engaged in, statements
about future expectations, and a statement indi-
cating something was misplaced. No child of-
fered to help or emitted a helping behavior
following a control verbalization. As a result,
those data are not discussed further (data avail-
able upon request).
Participants’ vocalizations while watching
the videos described above were also tran-
scribed. The transcriptions were reviewed and
scored for the presence of two types of state-
ments. First, any statement that indicated that
the child in the video should engage in a helping
behavior following a helping S
D
emitted by the
adult was recorded. Second, a statement that
described how the participant child would en-
gage in helping behavior in the situation de-
picted in the video was recorded. This latter
category was coded because young children of-
ten speak in self-referent ways when describing
how others should behave.
Observers recorded the occurrence of helping
behavior following each verbal S
D
. Graduate
students in psychology independently collected
interobserver agreement (IOA) data during 49%
of sessions across phases. IOA was calculated
using exact agreement for the occurrence of
helping following each S
D
, with the number of
agreements divided by number of agreements
plus disagreements multiplied by 100. IOA was
100% for Jack and Macon and was 96.7% for
Andy.
All transcriptions of child statements while
viewing the video were reviewed by two re-
searchers. Each statement was reviewed and
scored for the presence of a helping verbaliza-
tion by both researchers. IOA was calculated as
number of agreements divided by number of
agreements plus disagreements multiplied by
100. IOA was 100% for transcriptions.
Experimental Design and Procedures
Experimental data were evaluated in a con-
current multiple baseline across participants de-
sign. Experimental phases included baseline,
general prompt, specific prompt, MET, rule in-
struction plus feedback, and maintenance. Prior
to the beginning of the experiment and at the
conclusion, each child watched the 5-min video
described above. Prior to starting the video, the
Table 1
Categories of Helping Discriminative Stimuli
Category Example
Something lost Shoot, where is my highlighter?
Injury Ouch, I cut myself. (paper cut)
Mixed-up materials Darn, these got mixed up. (cards)
Something spilled Uh oh, I spilled my paper clips.
Top is stuck Shoot, I can’t open this jar.
Someone left a mess Oh no, what a mess.
9HELPING BEHAVIOR
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

experimenter said, “We are going to watch a
short video that has a boy and a grownup in it.
If you think the boy in the video should do
something different than what he is doing,
please tell me. Do you have any questions?”
The assessment of responding to the video stim-
ulus was included in the study as a means of
assessing the degree to which participants may
have developed verbal rules regarding the im-
portance of helping as a result of the experimen-
tal procedures. Our interest in the possible
emergence of verbal rules was motivated by the
hypothesized importance of rules and rule gov-
erned behavior in varied prosocial behaviors
potentially including helping (Peláez, 2001).
We chose observation of video rather than ask-
ing questions about helping because it had
lower social desirability demand characteristics
and required recognizing both a relevant occa-
sion and an appropriate action.
Baseline. The child was brought to an empty
classroom by the experimenter and seated at a
table. The experimenter explained that the child
needed to wait for a short time before they could
start their activity together. The child was told
that he would be allowed to play with toys
during his wait. The child was shown the array
of toys (described above) and allowed to pick
four toys to keep out to play with. The experi-
menter then moved to a nearby table and began
working on the material props described under
Materials. Once the child and experimenter were
seated and engaged with their respective materi-
als, the experimenter made his or her first state-
ment drawn from the list of helping S
D
s and S
⌬
s
(i.e., S-delta, control vocalizations), which be-
gan the session.
Each session consisted of six experimenter
vocalizations, with three S
D
s and three S
⌬
s pre-
sented. The order of presentation was random-
ized within sessions. The presentation of S
D
s
was randomized across sessions such that each
S
D
exemplar occurred once within two consec-
utive blocks of sessions. Experimenter state-
ments were spaced 1 min apart, and experiment-
ers used a vibrating timer to cue when to make
the next randomized verbalization. If the child
continued playing with his toys, the experi-
menter ignored the child and continued to in-
teract with her or his materials. If the child
offered to help, asked how they could help, or
asked what was the matter, the experimental
plan had been to respond in a socially conven-
tional fashion: accept the help, explain how they
could help, or explain the problem. Addition-
ally, if the child helped, the experimenter would
have thanked him. However, because no child
emitted any of these behaviors during baseline,
these procedures were not needed. At the end of
the baseline session, children engaged in an
animal categorization activity with the experi-
menter.
General prompt. To test the possibility
that a general priming statement would be suf-
ficient to evoke helping behavior, sessions were
conducted with a general prompt provided prior
to the session. The following statement was
added to the presession instructions provided to
participants: “I think helping people is good.”
General prompt sessions were conducted in the
same manner as baseline sessions except for the
inclusion of the prompt.
Specific prompt. To test the possibility that a
prompt to help in specific situations would
evoke helping behavior, we conducted Specific
Prompt sessions. The randomization of the pre-
sentation of S
D
s was modified to ensure that the
experimenter prompt corresponded to the first
helping S
D
that occurred during the session.
Two helping exemplars, something lost and
something spilled, were selected to be the first
S
D
to occur in each session. These two S
D
s were
randomly assigned to alternate session pairs as
the first S
D
to occur without replacement. As a
result, one of the two would be the first helping
S
D
in Session 1 and the other in Session 2. The
randomization continued across two session
blocks. The remaining four helping exemplars
were randomized across sessions so that all four
occurred once across two session blocks and
that three S
D
s were randomly assigned to each
session. When the experimenter was giving the
presession instructions to the child, a statement
was added to the end of the instructions stating
that helping is good, and then the experimenter
provided the first S
D
to occur in that schedule as
a specific example. For example, for the spilled
items’ S
D
, the statement was “Helping people is
good, like when helping pick up when some-
thing gets spilled.” Specific prompting sessions
were conducted in the same manner as baseline
session, except as described above.
Multiple exemplar training. MET was con-
ducted to examine whether teaching children to
help following two S
D
exemplars would result
in generalization of helping to untrained S
D
10 NOELL ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

exemplars. The randomization of S
D
exemplars
across sessions followed the procedures de-
scribed under Specific Prompt. MET was ap-
plied to the first S
D
to occur in each session (the
something lost and the something spilled exem-
plars).
Training consisted of a least-to-most teaching
procedure with three levels of prompts. First, if
the child did not either begin moving toward the
experimenter or ask how they could help imme-
diately, the experimenter said, “I need help
please.” If the child did not respond, the second
prompt was provided: “Please come here and
ask how you can help.” The third level of
prompt was to move to the child’s play area,
remove the toy the child was playing with, and
say, “Please help me,” followed by a description
of the specific motor action necessary to help
(e.g., “Please help pick up the cards”). The third
prompt was only needed with Jack for one ses-
sion. Contingent upon helping for both trained
and untrained S
D
exemplars, the child was
thanked and praised. The specific content of the
praise was varied to help prevent satiation (e.g.,
“You are such a good helper,” “What a nice boy
you are,” “What a big boy you are to help,”
etc.).
Rule instruction plus feedback. We were
interested in the extent to which our procedures
would produce high sustained levels of re-
sponding. For Jack and Macon, helping in-
creased during MET but did not occur on all
response opportunities. These participants were
provided rule instruction plus feedback (RI ⫹
FB), which was introduced to determine whether
explicitly talking about helping others and the
positive consequence of helping others might
evoke rule-governed behavior (RGB, Skinner,
1971). Our interest in the possible utility of
rules and RGB is the importance of language in
a considerable range of human learning and the
extensive support that RGB appears to have for
generalization (Fryling, Johnston, & Hayes,
2011;Malott, 1989).
RI ⫹FB followed the same procedures de-
scribed under MET, but added a brief preses-
sion conversation in which the experimenter
told the child that she or he thought helping was
good and provided an age-appropriate rationale
such as “Helping is a way to show others we
care.” The experimenter then provided a second
rationale for why either parents or teachers like
helpers, providing a different age-appropriate ra-
tionale such as “Helping is a way to be a good
team member.” The experimenter then asked
the child if he thought helping was good. Both
participants responded “yes.” The experimenter
then asked the child why helping was important.
If the child provided a rationale that was irrel-
evant to helping, which occurred in one session,
the child was prompted to provide a reason
related to helping. The children provided ratio-
nales as to why helping was good, and they
were praised for their responses.
At the end of the session, prior to transition-
ing to the categories activity, the experimenter
told the participant how many times he had
helped during the session out of the three op-
portunities. If the child helped once or twice, the
experimenter told the child that he or she
thought the child could do better. If the child
helped all three times, the experimenter praised
the child incorporating the rationale the child
had provided for helping. For example, for a
child who gave making his mother proud as
a rationale, the feedback would be, “You did a
great job helping all three times today. Your
mom would be proud of you.”
Maintenance. During the maintenance phase,
procedures were identical to baseline.
Treatment Integrity
The treatment integrity of the experimental
procedures was assessed using procedural
checklists devised for each condition. An ob-
server recorded the completion of each step in
the checklist during 40% of sessions. In order to
qualify as completing a step, the experimenter
had to complete the relevant step on all oppor-
tunities during the session. For example, if the
child helped three times during the session
the experimenter would be required to thank the
child on all three occasions to receive credit for
that step. Treatment integrity was 100% across
all assessments.
Results
The percentage of helping opportunities in
which participants emitted helping within 15 s
of an S
D
for all three participants are presented
in Figure 1. None of the participants exhibited
helping behavior during the baseline or general
prompt phases. Jack did not emit a helping
response during his specific prompt session. In
11HELPING BEHAVIOR
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

his initial MET session, he did not begin help-
ing quickly enough to meet the criterion. Sub-
sequently, he helped on all trials across four
sessions. Across the last four sessions of MET,
he missed 25% of the helping opportunities. In
one of those instances, he did approach the
experimenter and emitted a helping behavior
(sorting cards), but it was not the relevant re-
Figure 1. Percentage of helping behavior opportunities across sessions for Jack, Macon, and
Andy. BL denotes baseline, GP denotes general priming, SP denotes specific priming, MET
denotes multiple exemplar training, RI denotes rule instruction plus feedback, and Maint
denotes maintenance.
12 NOELL ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

sponse (picking up the spilled cards). Anecdot-
ally, we observed that periodically he appeared
reluctant to break away from the toy he was
playing with to help. RI ⫹FB was initiated to
test whether explicitly talking with him about
reasons helping is important and having him
state those consequences would result in RGB
stabilizing his behavior at earlier maximal lev-
els since his responding had become less stable.
His helping during RI ⫹FB was similar to that
at the end of MET. During the maintenance
sessions, Jack exhibited helping on all possible
occasions.
Macon exhibited two instances of helping
during specific prompting (22% of opportuni-
ties). Interestingly, in neither case was the help-
ing S
D
he responded to the one that had been
prompted. During MET, he exhibited variable
responding, with helping responses ranging
from 33% to 100% of opportunities per session.
During the MET phase, Macon began intermit-
tently emitting socially supportive statements
that were not part of our response definition of
helping. For example, in an instance in which
the experimenter said she had cut herself, he
responded, “That’s okay,” but did not offer as-
sistance. Like Jack, it appeared to the research
team that at times Macon was reluctant to dis-
engage from his toys to help. As an example, on
one trial he delayed disengaging from his toy
for too long to meet our helping definition, but
he did get up, approach the experimenter, and
offer help. RI ⫹FB was implemented with
Macon due to the variability in his responding
during MET to test the possibility that direct
RI ⫹FB would yield more consistent and max-
imal levels of responding. Macon exhibited
helping responses on 66% to 100% of opportu-
nities during RI ⫹FB. It was not possible to
obtain maintenance data for Macon due to
scheduling constraints.
Andy helped once during the specific prompt
condition. During MET, Andy’s helping behav-
ior was variable; however, he helped in all
opportunities during the last two sessions. It
was not possible to extend Andy’s analysis to
test whether his responding would remain stable
at maximal levels due to limitations on how
much longer it would be possible to work with
the participants. As a result, a maintenance
phase was implemented. During maintenance,
Andy helped on 83% of opportunities across 2
sessions.
The percentage of helping opportunities on
which participants emitted helping within 15 s
of an S
D
for all three participants across cate-
gories of helping by phase are presented in
Figure 2. All three participants responded to a
high percentage of opportunities for the two
trained exemplars, lost and spilled, once these
S
D
s were trained in MET during that phase and
subsequent phases. It is also worth noting that
when responding was below 100% for these
categories, the nonresponding within 15 s oc-
curred on either the first or first and second
instance that response was trained. Jack’s re-
sponding was similar across all generalization
categories. He responded at moderately high
rates during training (60% to 80%) and re-
sponded on all occasions at maintenance. Ma-
con’s and Andy’s responding demonstrated
more variability across categories. Both Macon
and Andy quickly mastered the top is stuck
category and responded at levels similar to Jack
for someone left a mess. Macon and Jack also
responded at substantially lower levels to the
injury and mixed-up materials in one or more
conditions across MET, RI ⫹FB, or mainte-
nance.
Prior to the experiment, all three participants
viewed the video tapes described in the Mate-
rials section presenting five S
D
s and five control
vocalizations across a 5-min video. Although
the participants attended to the videos and made
a number of statements about the video, none of
their statements was relevant to helping the
adult following an S
D
. Posttraining, only Macon
made a statement relevant to helping. While
observing the video, Macon observed that the
boy in the video was not helping, but that he
(Macon) helped people find things.
Discussion
All participants acquired a generalized help-
ing repertoire relevant to the conditions exam-
ined in this study during MET. The addition of
RI ⫹FB did not yield clear additional gains in
helping behavior for Jack or Macon over MET.
The participants for whom maintenance data
were available (Jack and Andy) continued to
help during a return to a naturalistic baseline.
Two of our participants made spontaneous vo-
calizations about helping being good during
MET sessions, and one described helping as
important when viewing the post-experimental
13HELPING BEHAVIOR
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Figure 2. Percentage of helping behavior opportunities for which participants emitted
helping behavior by helping category across phases. Lost and spill were the trained exemplars.
Baseline and general prompt in which no helping occurred are omitted.
14 NOELL ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

video recording (Macon). None of our partici-
pants exhibited helping behaviors during base-
line while playing with a toy in a classroom at
their school or made a response to the videos
describing helping behavior. Specific prompting
was followed by very low levels of helping
(12.5% of trials across participants). Similar to
prior research (Reeve et al., 2007), exposure to
MET was sufficient to occasion helping behav-
ior for both trained and untrained exemplars.
Our study differed from Reeve and colleagues
in that we trained half the number of exemplars
and training proceeded more quickly (6 to 9
sessions of MET vs. 11 to 43 sessions). This
difference may be attributable to differences
between our participants, who were 3-year-olds
of typical development, versus 5- and 6-year-
old children with ASD in Reeve et al.
Our study differed from prior research em-
ploying MET to establish pro-social behavior in
young children in the context of training. Prior
research in this area could be described as being
conducted in a discrete-trial context with con-
tingencies for any response that was not the
target of training minimized (Marzullo-Kerth et
al., 2011;Reeve et al., 2007;Wymer et al.,
2016). Although Reeve and colleagues did col-
lect substantive generalization data in the natu-
ral environment, these generalization trials were
presented as trials without a competing source
of reinforcement programmed. Additionally,
our training procedures were substantially less
intense and applied across fewer exemplars than
prior studies. For example, Reeve and col-
leagues provided repeated presentations of S
D
s,
video models, verbal prompts, and physical
prompts on each trial until the participant emit-
ted the target response. These studies demon-
strate that MET is sufficient to develop gener-
alized responding in a carefully controlled
context. In contrast, our study demonstrated that
it was possible to use MET to obtain general-
ized helping behavior in a natural context using
relatively few teaching trials and under condi-
tions of competing reinforcement for alternative
socially appropriate and sanctioned behavior. It
is worth recalling that at the beginning of the
wait period the children had been instructed to
play with the toys.
Some interesting patterns also emerged in the
differences in responding across the generaliza-
tion helping S
D
exemplars. The someone left a
mess and a top is stuck exemplars occasioned
relatively high rates of responding for all our
participants. It is possible that these are situa-
tions that participants were more familiar with
because they had encountered them pre-experi-
mentally. In contrast, the injury and mixed-up
materials exemplars resulted in lower levels of
responding for two of our participants. It is
possible that our 3-year-old participants had
fewer pre-experimental experiences with these
exemplars. We may have obtained more consis-
tent results had we used extended training on
these two exemplars for those participants who
struggle with them.
Exposure to the MET instruction also re-
sulted in two participants stating a rule that
helping is good. On various trials after exposure
to MET, the participants made statements that
could be described as socially conventional ex-
pressions of concern or support for the experi-
menter, such as the statements “I’ll help you”
and “How did you do that?” Also, two partici-
pants spontaneously made statements explicitly
stating the rule that helping is good (e.g., “Oh,
it’s good to help people”). Only one of the
participants articulated the rule post-experimen-
tally when observing the video. This may be
because the other two participants developed
generalized helping without developing a rule
that helping is good. Alternatively, it may be
that the task demands for the video assessment
were sufficiently novel and our participants suf-
ficiently young (3 years old) that they were not
able to generalize from what they learned in
session to the video. This result appears more
likely for Andy, who spontaneously vocalized
the value of helping during his own session, but
not when watching the video.
In placing our study in context, it is important
to recognize its limitations. We did not collect
data on toy engagement for our participants,
although anecdotally it was essentially continu-
ous. Additionally, we did not test whether help-
ing would have been more likely in a wait state
without toys. Also, our MET phase was rela-
tively short, and we trained only two exemplars.
We might have achieved more consistent re-
sponding by training a larger array of exemplars
and extending training. Our informal observa-
tions of how the sessions progressed would
suggest an alternative hypothesis. It may be the
case that reinforcement in the form of the grat-
itude that a child receives for helping an adult
would not successfully compete with the rein-
15HELPING BEHAVIOR
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

forcement available from ongoing toy play dur-
ing 100% of opportunities. In some cases, fin-
ishing a game, putting the next piece in the
puzzle, or completing the art project may sim-
ply be more reinforcing than the consequences
associated with stopping to help. The boys who
participated in our study stopped playing to help
at high levels by the end of the study. It is also
interesting to note that during trials in which
they did not help, they sometimes continued to
play while offering verbal encouragement or
simply waited beyond our time limit to begin
helping. Finally, our maintenance phases were
introduced immediately following completion
of the preceding condition. Thus, future re-
search could strengthen our methods by exam-
ining generalization across different contexts,
persons in need of help, and helping situations
to see if a broader response class emerged. For
young children, it is not clear how broadly we
should expect generalization to occur.
Our study replicated prior research demon-
strating the utility of MET in establishing a
pro-social behavior in young children (Mar-
zullo-Kerth et al., 2011;Reeve et al., 2007;
Wymer et al., 2016). Our study extended prior
research by demonstrating the emergence of
generalized helping in a context where the so-
cially conventional contingency for helping
successfully competed with access to ongoing
toy play. Additionally, our study is the first
study of which we are aware that found the
emergence of unprompted vocalizations of so-
cial behavioral rules in young children follow-
ing MET to establish a pro-social behavior.
Future research readily could build upon this
study by examining a number of issues. For
example, is there a format of rule instruction
alone that would result in rule-governed helping
behavior in young children? Would a format
that folded in more exemplars systematically to
assure that all exemplars are consistently ac-
quired prove efficacious? Will procedures such
as ours lead to the emergence of helping across
contexts? What learning history supports pro-
social behavior and rule conformity when the
consequences of behavior are delayed, ambigu-
ous, or uncertain?
References
Arnett, J. J. (2003). Conceptions of the transition to
adulthood among emerging adults in American
ethnic groups. In J. J. Arnett & N. L. Galambos
(Eds.), Exploring cultural conceptions of the tran-
sition to adulthood (pp. 63–76). San Francisco,
CA: Jossey-Bass. http://dx.doi.org/10.1002/cd.75
Beaulieu, L., & Hanley, G. P. (2014). Effects of a
classwide teacher-implemented program to pro-
mote preschooler compliance. Journal of Applied
Behavior Analysis, 47, 594–599. http://dx.doi.org/
10.1002/jaba.138
Berens, N. M., & Hayes, S. C. (2007). Arbitrarily
applicable comparative relations: Experimental ev-
idence for a relational operant. Journal of Applied
Behavior Analysis, 40, 45–71. http://dx.doi.org/10
.1901/jaba.2007.7-06
Branch, M. N. (1999). Statistical inference in behav-
ior analysis: Some things significance testing does
and does not do. The Behavior Analyst, 22, 87–92.
http://dx.doi.org/10.1007/BF03391984
Caprara, G. V., Luengo Kanacri, B. P., Zuffianò, A.,
Gerbino, M., & Pastorelli, C. (2015). Why and how
to promote adolescents’ prosocial behaviors: Direct,
mediated and moderated effects of the CEPIDEA
school-based program. Journal of Youth and Adoles-
cence, 44, 2211–2229. http://dx.doi.org/10.1007/
s10964-015-0293-1
Dineen, J. P., Clark, H. B., & Risley, T. R. (1977).
Peer tutoring among elementary students: Educa-
tional benefits to the tutor. Journal of Applied
Behavior Analysis, 10, 231–238. http://dx.doi.org/
10.1901/jaba.1977.10-231
Dunkel-Jackson, S. M., Dixon, M. R., & Szekely, S.
(2016). Self-control as generalized operant behav-
ior by adults with autism spectrum disorder. Jour-
nal of Applied Behavior Analysis, 49, 705–710.
http://dx.doi.org/10.1002/jaba.315
Eisenberg, N., Fabes, R. A., & Spinrad, T. L. (2006).
Prosocial behavior. In W. Damon, R. M. Lerner,
and N. Eisenberg (Eds.), Handbook of child psy-
chology: Vol. 3.Social, emotional, and personality
development (6th ed., pp. 646–718). New York,
NY: Wiley.
Franca, V. M., Kerr, M. M., Reitz, A. L., & Lambert,
D. (1990). Peer tutoring among behaviorally dis-
ordered students: Academic and social benefits to
tutor and tutee. Education and Treatment of Chil-
dren, 13, 109–128.
Fryling, M. J., Johnston, C., & Hayes, L. J. (2011).
Understanding observational learning: An interbe-
havioral approach. Analysis of Verbal Behavior, 27,
191–203. http://dx.doi.org/10.1007/BF03393102
Gilic, L., & Greer, R. D. (2011). Establishing naming
in typically developing two-year-old children as a
function of multiple exemplar speaker and listener
experiences. Analysis of Verbal Behavior, 27,
157–177. http://dx.doi.org/10.1007/BF03393099
Gralinski, J. H., & Kopp, C. B. (1993). Everyday
rules for behavior: Mothers’ requests to young
16 NOELL ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

children. Developmental Psychology, 29, 573–584.
http://dx.doi.org/10.1037/0012-1649.29.3.573
Harris, S. L., Handleman, J. S., & Alessandri, M.
(1990). Teaching youths with autism to offer as-
sistance. Journal of Applied Behavior Analysis, 23,
297–305. http://dx.doi.org/10.1901/jaba.1990.23-
297
Kahn, P. H., Jr. (1992). Children’s obligatory and dis-
cretionary moral judgments. Child Development, 63,
416– 430. http://dx.doi.org/10.2307/1131489
Malott, R. W. (1989). The achievement of evasive
goals: Control by rules describing contingencies that
are not direct acting. In S. C. Hayes (Ed.), Rule-
governed behavior: Cognition, contingencies, and
instructional control (pp. 269–322). New York,
NY: Plenum Press. http://dx.doi.org/10.1007/978-
1-4757-0447-1_8
Marzullo-Kerth, D., Reeve, S. A., Reeve, K. F., &
Townsend, D. B. (2011). Using multiple-exemplar
training to teach a generalized repertoire of sharing to
children with autism. Journal of Applied Behavior
Analysis, 44, 279–294. http://dx.doi.org/10.1901/jaba
.2011.44-279
Peláez, M. (2001). Morality as a system of rule-
governed behavior and empathy. Behavioral De-
velopment Bulletin, 10, 8–14. http://dx.doi.org/10
.1037/h0100475
Pettygrove, D. M., Hammond, S. I., Karahuta, E. L.,
Waugh, W. E., & Brownell, C. A. (2013). From
cleaning up to helping out: Parental socialization
and children’s early prosocial behavior. Infant Be-
havior and Development, 36, 843– 846. http://dx
.doi.org/10.1016/j.infbeh.2013.09.005
Plötner, M., Over, H., Carpenter, M., & Tomasello,
M. (2015). The effects of collaboration and mini-
mal-group membership on children’s prosocial be-
havior, liking, affiliation, and trust. Journal of Ex-
perimental Child Psychology, 139, 161–173.
http://dx.doi.org/10.1016/j.jecp.2015.05.008
Reeve, S. A., Reeve, K. F., Townsend, D. B., & Poul-
son, C. L. (2007). Establishing a generalized reper-
toire of helping behavior in children with autism.
Journal of Applied Behavior Analysis, 40, 123–136.
http://dx.doi.org/10.1901/jaba.2007.11-05
Rosales, R., Rehfeldt, R. A., & Lovett, S. (2011). Ef-
fects of multiple exemplar training on the emergence
of derived relations in preschool children learning a
second language. Analysis of Verbal Behavior, 27,
61–74. http://dx.doi.org/10.1007/BF03393092
Sivaraman, M. (2017). Using multiple exemplar
training to teach empathy skills to children with
autism. Behavior Analysis in Practice, 10, 337–
346. http://dx.doi.org/10.1007/s40617-017-0183-y
Skinner, B. F. (1971). Beyond freedom and dignity.
New York, NY: Knopf/Random House.
Smith, C. E., Noh, J. Y., Rizzo, M. T., & Harris, P. L.
(2017). When and why parents prompt their chil-
dren to apologize: The roles of transgression type
and parenting style. Journal of Family Studies, 23,
38– 61. http://dx.doi.org/10.1080/13229400.2016
.1176588
Warneken, F., & Tomasello, M. (2013). Parental pres-
ence and encouragement do not influence helping in
young children. Infancy, 18, 345–368. http://dx.doi
.org/10.1111/j.1532-7078.2012.00120.x
Weil, T. M., Hayes, S. C., & Capurro, P. (2011).
Establishing a deictic relational repertoire in young
children. The Psychological Record, 61, 371–390.
http://dx.doi.org/10.1007/BF03395767
Williamson, R. A., Donohue, M. R., & Tully, E. C.
(2013). Learning how to help others: Two-year-
olds’ social learning of a prosocial act. Journal of
Experimental Child Psychology, 114, 543–550.
http://dx.doi.org/10.1016/j.jecp.2012.11.004
Wymer, S. C., Tarbox, J., Beavers, G. A., & Tullis,
C. A. (2016). Teaching children with autism to
follow rules specifying a behavior and conse-
quence. Analysis of Verbal Behavior, 32, 265–274.
http://dx.doi.org/10.1007/s40616-016-0059-1
Received January 18, 2018
Revision received November 2, 2018
Accepted November 2, 2018 䡲
17HELPING BEHAVIOR
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.