The Emergence of Forgiveness in Young Children

Abstract

Humans depend greatly on our cooperative relationships. Thus, when our relationships are damaged by transgressions, they need to be repaired. Such repair requires that the transgressor show remorse and the victim forgive. Previous research demonstrates that as transgressors, young children show remorse and attempt to repair the harm they caused. However, it remains unclear when children, as victims, forgive remorseful transgressors. In Study 1, 5‐, but not 4‐year‐olds, (n = 20 each) were more forgiving of a remorseful transgressor (who did not explicitly apologize) than an unremorseful transgressor. In Study 2, 4‐year‐olds (n = 20) were more forgiving of an apologetic than unapologetic transgressor. Thus, from early in ontogeny, humans are motivated to repair damaged relationships and thus uphold cooperation.

Full text

The Emergence of Forgiveness in Young Children
Janine Oostenbroek and Amrisha Vaish
The University of Virginia
Humans depend greatly on our cooperative relationships. Thus, when our relationships are damaged by
transgressions, they need to be repaired. Such repair requires that the transgressor show remorse and the vic-
tim forgive. Previous research demonstrates that as transgressors, young children show remorse and attempt
to repair the harm they caused. However, it remains unclear when children, as victims, forgive remorseful
transgressors. In Study 1, 5-, but not 4-year-olds, (n=20 each) were more forgiving of a remorseful transgres-
sor (who did not explicitly apologize) than an unremorseful transgressor. In Study 2, 4-year-olds (n=20) were
more forgiving of an apologetic than unapologetic transgressor. Thus, from early in ontogeny, humans are
motivated to repair damaged relationships and thus uphold cooperation.
Humans are extremely social beings, and we rely
heavily on our cooperative relationships for our
survival and to succeed in achieving our individual
and communal goals (Tomasello, Melis, Tennie,
Wyman, & Herrmann, 2012). Thus, when transgres-
sions occur and damage our relationships, it is vital
that we repair those relationships in order to con-
tinue benefiting from group living and cooperation.
So how do we repair our ruptured cooperative
relationships? One key to reparation is the expres-
sion of guilt and remorse by the transgressor.
Adults, and even young children, experience guilt
after a transgression and attempt to repair harm
they have caused and restore the damaged relation-
ship (Kochanska, Gross, Lin, & Nichols, 2002;
Vaish, Carpenter, & Tomasello, 2016). However,
guilt and remorse only represent the transgressor’s
half of the repair process. The other half is forgive-
ness by the victim (McCullough, 2008; Worthington,
2010). Forgiveness reestablishes a victim’s positive
feelings toward transgressors, fosters reconciliation,
and allows transgressors to reenter mutually benefi-
cial relationships, thus helping to maintain coopera-
tion (McCullough, 2008). Yet we know strikingly
little about the ontogenetic emergence of forgive-
ness. The current research was designed to fill this
gap in our knowledge.
Perhaps the most prominent elicitor of forgive-
ness is transgressors’remorse (Ohbuchi, Kameda, &
Agarie, 1989; Petrucci, 2002). Research with adults
shows that transgressors’remorse signals to the vic-
tim that the transgressor himself is also suffering,
which evokes sympathy, concern, and forgiveness
and thus reduces the likelihood of punishment
(Keltner & Anderson, 2000; Leary, Landel, & Patton,
1996). Expressing remorse also serves as a promise
of more acceptable behavior in the future (Castel-
franchi & Poggi, 1990; Keltner, Young, & Buswell,
1997) and signals the transgressor’s intention of
refraining from harming the victim again (McCul-
lough, Worthington, & Rachal, 1997). Remorse is
thus extremely effective in eliciting forgiveness and
therefore promoting relationship repair in adults.
The literature remains unclear, however, as to
when children respond positively to transgressors’
remorse. Previous studies have shown that after
hearing stories in which one individual transgresses
against another, 4- to 8-year-old children blame and
punish transgressors less, and forgive and like them
more, if they apologized than if they did not apolo-
gize (e.g., Darby & Schlenker, 1982, 1989; Smith,
Chen, & Harris, 2010). Furthermore, children of this
age judge situations in which a transgressor apolo-
gized as better and more just than ones in which
the transgressor was unapologetic (Irwin & Moore,
1971; Wellman, Larkey, & Somerville, 1979), and
they attribute improved feelings to a victim who
received an apology (Smith et al., 2010).
We thank all the families for their participation; Shannon
Savell, Mimi Nguyen, and Shannon McGinnis for assistance with
testing; and Ebony Logan for coding. This research was
supported by a grant from the John Templeton Foundation to
Amrisha Vaish (Grant #55437).
Correspondence concerning this article should be addressed to
Amrisha Vaish, Department of Psychology, University of Vir-
ginia, PO Box 400400, Charlottesville, VA 22904. Electronic mail
may be sent to vaish@virginia.edu.
©2018 Society for Research in Child Development
All rights reserved. 0009-3920/2018/xxxx-xxxx
DOI: 10.1111/cdev.13069
Child Development, xxxx 2018, Volume 00, Number 0, Pages 1–18

In nearly all prior work, however, children were
bystanders to the transgressions and were thus
evaluating transgressors’responses from an obser-
ver’s perspective. Although such third-party evalua-
tions are undoubtedly important, they do not
constitute true forgiveness. In part, this is because
forgiveness is by definition felt and granted by a
hurt party rather than a bystander. That is, victims
are unique in their ability to forgive and thus to
allow relationships to be repaired (see Zitek, Jor-
dan, Monin, & Leach, 2010). Additionally, when
children are bystanders and are not personally
affected by the transgressions, their feelings of sad-
ness or anger (which they must overcome in order
to forgive) are likely not elicited nearly as strongly
as when they are the victims of transgressions. In
other words, as bystanders, they can likely evaluate
and consider transgressors’responses more coolly
(“cold cognition,”see Metcalfe & Mischel, 1999),
whereas as victims, their strong emotions might
make it more challenging for them to do so (“hot
cognition”). This could lead to a more protracted
development of forgiveness in first- than third-party
situations. Indeed, there is evidence of such a dis-
connect from other domains of moral development.
For example, whereas young children endorse
norms of sharing and expect others to share
equally, they themselves do not share equally until
7–8 years of age (Smith, Blake, & Harris, 2013).
It is therefore critical to examine children’s for-
giveness when they themselves are within the
threatened relationship. Only two studies to date
have used this approach. In one study, 4- to 7-year-
old children suffered minor disappointment when
they did not receive the stickers that they expected
to receive from another child. Children then
received either a note of apology from the other
child or a note without an apology. Children
reported feeling better and rated the other child as
being more remorseful and nicer if the other child
had sent an apologetic note versus a nonapologetic
note (Smith & Harris, 2012). This study provided
the first evidence that as early as 4 years of age,
children forgive apologetic transgressors. In another
study, Drell and Jaswal (2016) used a live paradigm
to investigate children’s responses to a transgres-
sor’s remorse. Six- and 7-year-olds painstakingly
built a tower that was then knocked over by a
transgressor. The transgressor either provided an
apology, offered restitution, or did nothing. Chil-
dren shared more stickers with a transgressor who
apologized or offered restitution than with one who
did nothing. Taken together, these two studies sug-
gest that preschool and early school-age children
may well be capable of forgiving remorseful trans-
gressors.
However, from an early age, children are fre-
quently told by their caregivers and teachers to
apologize (see Smith, Noh, Rizzo, & Harris, 2017),
even when they might not feel sorry. As a result,
children’s evaluations of transgressors who say they
are sorry or who are described as having apolo-
gized might be based on hearing key words or
phrases (e.g., “I’m sorry”), which they have learned
are the normative responses after one has commit-
ted a transgression (Vaish, Carpenter, & Tomasello,
2011). Thus, to assess the development of forgive-
ness in response to transgressors’remorse rather
than apologies specifically, it is imperative to study
children’s responses to remorse in the absence of
explicit apologies.
Accordingly, one recent study investigated chil-
dren’s responses to a transgressor who showed
remorse without explicitly apologizing (Vaish et al.,
2011). Using a forced-choice paradigm, this study
revealed that 5-year-old children positively evalu-
ated the remorseful transgressor more than the
unremorseful transgressor, preferred the remorseful
transgressor, and distributed more resources to the
remorseful transgressor. Four-year-olds, on the
other hand, did not show any of these effects. In
Vaish et al.’s (2011) Study 2, when the transgres-
sor’s remorse was accompanied by explicit apolo-
gies (“I’m sorry. I apologize.”), 4-year-olds also
positively evaluated, preferred, and distributed
more resources to the remorseful transgressor.
Therefore, by 5 years of age, children seem to truly
appreciate the implications of—and thus respond
positively to—a transgressor’s expression of
remorse; a year earlier, children show similar
appreciation for more conventional cues of remorse
such as explicit apologies.
Although Vaish et al.’s (2011) findings shed light
on children’s responses to transgressors’remorse in
the absence of apologies, their studies, like most
prior work, also placed children in the role of
bystander rather than victim. Their results thus can-
not be assumed to generalize to situations that
arguably matter most to restoring relationships—
when children are themselves the victims of trans-
gressions—and thus cannot be assumed to reflect
children’s propensity for forgiveness.
The present studies were designed to examine
the developmental emergence of forgiveness. We
asked: When do children, as the victims of trans-
gressions, begin to forgive remorseful transgressors?
In Study 1, we assessed children’s responses to
remorse without explicit apologies. Specifically,
2 Oostenbroek and Vaish

4- and 5-year-olds experienced a minor harm caused
by two transgressors. One transgressor showed
remorse (without explicitly apologizing) and the
other was unremorseful. Children were asked about
their evaluations of and preferences for the two
transgressors, and they distributed resources between
the transgressors. Furthermore, we were interested to
see if children form a generalized positive evaluation
of the remorseful transgressor (i.e., think that she is a
generally positive, prosocial individual) and a gener-
alized negative evaluation of the unremorseful trans-
gressor (i.e., think that she is a generally negative,
unhelpful individual).
On the basis of Vaish et al. (2011), we expected
that by 5 years, but not by 4 years, children would
be more forgiving of the remorseful transgressor,
operationalized as preferring her, evaluating her
more positively, making more positive generaliza-
tions about her, and giving her more resources than
the unremorseful transgressor (McCullough, 2008).
However, given that children were themselves the
victims and would thus be far more affected by the
transgression than as observers, it was possible that
the 5-year-olds in our study would not respond
positively to the remorseful transgressor as robustly
as the 5-year-olds in Vaish et al.’s third-party sce-
narios did. We also asked children to provide justi-
fications for their evaluations, as these may be
useful in revealing which aspects of the situations
children use to make their evaluations. In Study 2,
we examined 4-year-old children’s responses to
transgressors who explicitly apologized versus did
not apologize. Based on Vaish et al. (2011) and
Smith and Harris (2012), we predicted that 4-year-
olds would be more forgiving of a transgressor
who explicitly apologized than one who did not.
Study 1
Method
Participants
Twenty 4-year-olds (M=52;4, i.e., 52 months
and 4 days, SD =2;29; range =48;3–57;19; 10 girls)
and 20 5-year-olds (M=65;2; SD =3;8;
range =60;9 days to 69;13; 10 girls) were included
in the study. Four additional 4-year-old children
were tested but excluded due to unwillingness to
participate (n=3) or not understanding sufficient
English (n=1). Participants were recruited from a
medium-sized mid-Atlantic university town,
through the institution’s database of families inter-
ested in participating in child development
research. Data were collected between October 2016
and February 2017. Of the families that provided
information about race (n=38) and education
(n=40), 92.1% of the parents were Caucasian, and
95% of the parents were at least college educated.
Materials and Setting
Materials included paper and crayons, three
small cardboard boxes, and three cloth flowers.
Children were tested individually in a quiet room
within the laboratory, equipped with a rectangular
table, four chairs, and two video cameras on tri-
pods. Three adults conducted the study: one as
moderator and the other two as actresses.
Procedure
After initial warm-up games with the child, the
moderator (M) led the child and the two actresses
(Anna and Kelly) into the testing room and invited
them to sit around the table. The child sat at one
end of the table and the two actresses sat on either
side of the child, across from one another. On the
table in front of each of them was a sheet of paper.
In front of the child’s seat were three boxes—one
each for Anna and Kelly and one for the child. The
boxes for Anna and Kelly contained an individual
photograph of themselves (the photographs fea-
tured each actress with a neutral facial expression
looking directly at the camera), whereas the box for
the child had a cartoon picture of either a boy or
girl (matched to the participant’s gender) inside the
box. Each box contained three crayons. M dis-
tributed the boxes and then asked each of them
what they would like to draw. If the child asked
what he or she should draw, M said, “You can
draw whatever you like.”M then explained that
whoever drew the best picture would win a special
prize.
Then, M said she needed to do some work and
sat out of sight of the child, leaving the child, Anna,
and Kelly to draw their pictures. After they com-
pleted their pictures, Anna and Kelly admired each
other’s pictures and showed their pictures to the
child. Both actresses then exclaimed that they
would like to look at the child’s picture, simultane-
ously picked up the child’s picture, and each held
one end of the picture while admiring it.
The accident then occurred when Anna and
Kelly tried to return the picture to the child, as in
that moment, each actress simultaneously pulled
the picture in her own direction, thus “accidentally”
tearing the picture (note that the picture did not
Forgiveness in Young Children 3

completely tear in half). Both actresses paused
briefly while still holding the child’s picture and
looking neutrally at it, and then simultaneously
placed the torn picture in front of the child.
The actresses’subsequent reactions were based
closely on Vaish et al. (2011). Specifically, one of the
actresses now looked remorseful and concerned, and
remorsefully said, “Oh, I’ve torn your picture. I
didn’t want that to happen. It’s my fault.”While
speaking, she alternated her gaze between the child
and the torn picture. Then, the other actress
expressed no remorse but instead looked neutral and
said in a neutral tone of voice, “Hmm, I’ve torn your
picture. Hmph [shrugging shoulders], I don’t care,”
while alternating her gaze between the child and the
torn picture. Note that the unremorseful transgressor
was not aggressive or negative in any way; rather,
she was simply neutral about the transgression.
Next, M, seemingly without noticing the pictures
on the table, requested Anna and Kelly to “tidy up
their things”outside the room. The actresses left the
room. M then “noticed”the torn picture and
expressed mild sympathy, saying, “Oh, your pic-
ture.”She then asked the first comprehension ques-
tion: “What happened?”This question was designed
to ensure that children had comprehended that (a)
their own picture was torn, and (b) the two actresses
tore it. Thus, if children provided a complete answer
such as, “Anna and Kelly tore my picture,”M moved
on to the remainder of the comprehension questions
(described next). However, if children provided an
incomplete answer (such as, “My picture is torn”), M
asked for the missing piece of information (e.g.,
“And who tore it?”).
Once children had demonstrated comprehension,
M then moved Anna and Kelly’s boxes in front of
the child (so that she and the child could point to
the photo of each transgressor during the subse-
quent questions and answers). M then asked two
further comprehension questions, specifically, if the
child could recall what Anna and Kelly each said
after the picture was torn. If the child answered
correctly (“She said she didn’t want it to happen”
or “She said it was her fault”or something similar
for the remorseful transgressor; “She said she didn’t
care”or something similar for the unremorseful
transgressor), M said, “Yes, that’s right. You’ve
understood correctly,”and repeated what each
transgressor had said. If, however, the child
answered incorrectly (e.g., “She said it was her
fault”in the unremorseful case), M said, “Hmm,
I’m not sure about that,”and provided the child
with the correct information. M then provided a
reminder of what each actress said. This was done
to ensure that all children received the same, correct
information before being asked the test questions.
We chose to explicitly provide children with the
correct information because our primary aim was
not to test whether children comprehend the trans-
gression and the transgressors’responses on their
own but rather to examine how children respond
when they have this information. Children were
then asked the following test questions (adapted
from Vaish et al., 2011):
1. “Whom are you more upset with, Anna or
Kelly?”(pointing to each actress in turn)
1a. “Why are you more upset with her?”
2. “Whom do you like more, Anna or Kelly?”
(pointing to each)
2a. “Why do you like her more?”
3. “If you fell over, who do you think would
help you, Anna or Kelly?”(pointing to each)
3a. “Why do you think she would help you?”
4. “If you drew another picture, who do you
think would tear it again, Anna or Kelly?”
(pointing to each)
4a. “Why do you think she would tear it
again?”
The order of these question pairs was counterbal-
anced across children. After testing 16 children
(seven 5-year-olds and nine 4-year-olds), we real-
ized that Question 4 taps into children’s expecta-
tions about a future transgression that is the same
as the transgression that occurred in the experi-
ment, but that it would also be interesting to assess
whether children’s expectations generalized to a dif-
ferent future transgression. We thus added a fifth
test question pair:
5. “If you were playing on the swings, who do
you think would push you off, Anna or
Kelly?”(pointing to each)
5a. “Why do you think she would push you
off?”
This fifth question pair was inserted at the very end
of the procedure. Thus, the procedure was identical
for the entire sample and after the first 16 participants,
the remaining children also received this additional
question pair at the end of the entire procedure.
Questions 1–5 were forced-choice questions,
designed to assess whether, when presented with the
choice, children’s evaluations take into account the
transgressors’remorse. In answering the forced-choice
4 Oostenbroek and Vaish

questions, children were expected to name or point to
one of the transgressors. If a child responded
with “Both”or “Neither,”M encouraged the child to
choose one. If a child did not respond, M repeated
the question, but if the child still did not respond, M
moved on to the next question. Questions 1a–5a were
open questions, designed to elicit justifications for
children’s responses to the forced-choice questions. M
therefore did not prompt or probe children when
they were answering the open questions, and did not
provide any feedback on the content of their
responses.
After the first four test questions, M gave the
child three cloth flowers and said, “Here are three
flowers. Later, Anna and Kelly will look inside their
boxes. You can give the flowers out to Anna and
Kelly how ever you like. I just have some work to
do over here for a minute,”(M looked away and
shuffled some papers so the child did not feel
observed). If the child was hesitant or asked for
guidance, M said, “You can give the flowers out
how ever you like.”Children were expected to dis-
tribute all three of the flowers between the two
transgressors. Once children distributed the flowers,
M asked them a final justification question: “Why
did you give [Anna or Kelly] more flowers?”
Again, the child was free to respond, and M did
not prompt or probe the child further. Finally, M
asked the fifth question pair (beginning after the
first 16 children). This concluded the study. In
order to end the procedure on a positive note, both
actresses returned to the room and apologized for
tearing the child’s picture. All three experimenters
then helped the child draw a new picture. Children
were then taken to their parents and were given a
gift for their participation. The experimental proce-
dure took approximately 15 min.
Counterbalancing
Children in each age group were randomly
assigned to one of 20 presentation orders, which
counterbalanced the following: which side of the
child each actress sat, which actress expressed
remorse or no remorse, whether remorse or no
remorse was expressed first, the order of the test
question pairs (with the exception of the fifth ques-
tion pair), as well as the order that the actresses’
names appeared in the test questions.
Coding and Reliability
The primary coder (the first author) used tran-
scriptions to code whether children responded
correctly to the comprehension questions and the
forced-choice test questions (Questions 1–5).
Responses (verbal or pointing) were scored 1 if they
were consistent with the hypotheses that children
should (a) be more upset with the unremorseful
transgressor, (b) prefer the remorseful transgressor,
(c) think the remorseful transgressor would help
them if they fell over, (d) think the unremorseful
transgressor would tear another picture of the
child’s again and, (e) think the unremorseful trans-
gressor would push them off the swing; responses
not consistent with these hypotheses were scored 0.
A second coder (unaware of the hypotheses) coded
100% of the sample. Reliability for both the compre-
hension checks and forced-choice test questions was
perfect, both js=1.00.
The primary coder used video to code whether
children gave 0, 1, 2, or 3 flowers to the remorseful
transgressor. The second coder also coded this vari-
able for the whole sample and agreement between
coders was 100%.
In addition, children’s justifications (in response
to Questions 1a–5a as well as following the distri-
bution of flowers) were coded from the transcrip-
tions. Each justification was assigned a score of 1 or
0 (see Table 1 for details of the coding scheme). Jus-
tifications were assigned a score of 1 if they indi-
cated relevant and sophisticated reasoning about
the transgressors and their responses, including jus-
tifications that referred to feelings of remorse or
apologies, or involved moral evaluations. Refer-
ences to a transgressor’s feelings of remorse (or the
lack thereof) could be of two types: One type
—“Remorse (repeated)”—involved repeating
phrases that the transgressors had used in the inter-
action (e.g., “Because she said it was her fault that
she tore my picture”), whereas the other type
—“Remorse (re-described)”—involved using
phrases other than those used by the transgressors
in the interaction (e.g., “Because she said sorry for
tearing my picture”). Although “Remorse (re-
peated)”was obviously related to our question of
whether children forgive remorseful transgressors,
it was unclear whether children were engaging in
higher level reasoning or not when children
repeated phrases used by the transgressors in the
interaction. That is, did children truly reason about
the transgressors’responses in a sophisticated way
but chose to repeat what the transgressors had said,
or did they not understand the transgressors’
responses in a sophisticated way and therefore sim-
ply repeated what the transgressors had said? To
account for both possibilities, we analyzed chil-
dren’s justifications in two ways: In one, “Remorse
Forgiveness in Young Children 5

(repeated)”was assigned a score of 1, and in the
other, it was assigned a score of 0 (cf. Vaish et al.,
2011).
All other justifications were assigned a 0, includ-
ing those that were not diagnostic. For instance, jus-
tifications based on one of the transgressors tearing
the child’s picture received a score of 0 because in
fact both transgressors tore the child’s picture. A
second coder (unaware of the hypotheses) coded
justifications of a random 25% of children. Reliabil-
ity was perfect, j=1.
Results
We begin with the results of the comprehension
checks in order to provide information about how
well children understood the interaction and how
accurately they recalled each transgressor’s response
after their picture was torn. We then present results
from the test questions and the distribution of
resources task. Preliminary analyses revealed that
for both the 4- and 5-year-old groups, there were no
significant effects of gender. We therefore collapsed
across this variable for all analyses.
Comprehension Checks
Comprehension check 1. What happened to your
picture? Next we report the distribution of chil-
dren’s responses by age group.
Age comparisons. A chi-square analysis
revealed that the older children were somewhat
more likely to respond correctly than the younger
children, p=.057.
4-Year-olds. When asked what had happened
to their picture, eight of twenty 4-year-olds (40%)
correctly responded that the two actresses had torn
it. Of the remaining 12, one child (5%) responded
that only one actress had torn their picture (the
remorseful transgressor); four children (20%) pro-
vided incomplete answers and when asked the fol-
low-up question (Who tore it?), three children said
the remorseful transgressor had torn it and one
said the unremorseful transgressor had torn it; and
seven children (35%) responded that they did not
know what happened to their picture. For the 12
children who provided incomplete or incorrect
answers, M clarified that in fact both actresses
were holding the child’s picture and both had
torn it.
5-Year-olds. When asked what had happened
to their picture, fourteen of twenty 5-year-olds
(70%) correctly responded that the two actresses
had torn it. The remaining six children (30%)
responded that they did not know what happened
to their picture, and M then provided them with
the correct information.
Comprehension check 2. Can you remember what
[Anna or Kelly] said after your picture was torn? Next
we report the distribution of children’s responses
by age group.
Age comparisons. Significantly more 5- than
4-year-olds correctly recalled what the remorseful
transgressor said after the picture was torn, (Fish-
er’s exact test [due to small Ns in some cells],
p=.048). However, there was no significant differ-
ence between the age groups for children’s recall of
what the unremorseful transgressor said (p=.280).
Table 1
Coding Scheme for Justifications: Study 1
Score Category Content
1 Remorse (re-described) Transgressor did (or did not) show remorse (child uses words other than those used
by the actresses in the experiment or by M); for example, “Because she tore my
picture and said she was sorry”
Moral character, evaluation, or norm Transgressor is a good (or bad) person, transgressor’s response to the transgression
was good (or bad), or transgressor broke (or did not break) a moral norm;
for example, “Because she said the right thing”or “Because she’s a nicer person”
Remorse (repeated) [analyses were
conducted with this category scored
as “1”and as “0”]
Transgressor did (or did not) show remorse (child uses words that had been used by
the actresses in the experiment or by M); for example, “Because she said,
‘It was all my fault’” or “Because she said she didn’t care”
0 Own preference Child’s own preference for the transgressor; for example, “Because I like her more”
Action Transgressor tore the picture
Object The picture is torn or ripped and can no longer be repaired
Other, irrelevant, or uncodable Response could not be put into any of the above categories (e.g., “Just because”),
was irrelevant (e.g., “Because my head would hurt”) or could not be coded
(e.g., because the child’s speech could not be understood)
6 Oostenbroek and Vaish

4-Year-olds. When asked what the remorseful
transgressor said, four children (20%) provided no
verbal response. Of the 16 children who provided a
response, 11 (55%) answered correctly (binomial
probability, using a test proportion of .50, p=.210)
and 5 (25%) answered incorrectly. When asked
what the unremorseful transgressor said, four chil-
dren (20%) provided no verbal response. Of the 16
children who provided a response, 8 children (40%)
answered correctly (binomial probability, p=1.00),
and eight children (40%) answered incorrectly.
5-Year-olds. When asked what the remorseful
transgressor said, seven children (35%) provided no
verbal response. The remaining 13 children (65%)
answered correctly (binomial probability,
p<.0005). When asked what the unremorseful
transgressor said, four children (20%) provided no
verbal response. Of the 16 children who provided a
response, 11 (55%) answered correctly (binomial
probability, p=.210) and five children (25%)
answered incorrectly.
Test Questions
Forced-choice questions. Preliminary analyses
revealed no significant effects of either 4- or 5-year-
olds’performance on the comprehension checks on
their performance on the test questions. This vari-
able was thus not included in further analyses.
Age comparisons. Chi-square analyses
revealed that on four of the five forced-choice test
questions, significantly more 5- than 4-year-olds
drew the hypothesized inferences (all ps<.022, φ
range =.36–.51; see Figure 1). The one question that
did not reveal a significant age difference was the
question about which transgressor would tear
another picture that the child drew (p=.239). Addi-
tionally, the proportion of the five test questions
answered in the hypothesized way was signifi-
cantly higher among the 5-year-olds (M=0.86;
SD =.28) than 4-year-olds (M=0.49; SD =.45),
t(38) =3.07, p=.004, Cohen’sd=0.98. We thus
analyzed each age group separately.
4-Year-olds. The 4-year-olds’responses indi-
cated that they did not draw any of the hypothesized
inferences. That is, 4-year-olds did not systematically
choose either the remorseful or unremorseful trans-
gressor when judging whom they were more upset
with, whom they liked more, who they thought
would help them if they fell over, or who would be
more likely to tear another picture that they drew or
push them off a swing. The proportion of children
who responded in the hypothesized way on each
question ranged from .45 to .60; binomial probabili-
ties, all ps>.503. Altogether, only nine of 20 children
(45%) responded as predicted to more than half of
the forced-choice questions (p=.824).
5-Year-olds. In striking contrast to the 4-year-
olds, the majority of the 5-year-olds did draw the
appropriate and hypothesized inferences for each
question: They were more upset with the unre-
morseful transgressor (binomial probability,
p<.0005), liked the remorseful transgressor more
(p<.0005), thought the remorseful transgressor
would be more likely to help them (p=.012), and
thought the unremorseful transgressor would be
Figure 1. Proportion of 4- and 5-year-old children in Study 1 who answered each test question and distributed resources in the hypoth-
esized ways. The dashed line indicates chance level.
Note. *p<.05. **p<.01. ***p<.0005.
Forgiveness in Young Children 7

more likely to tear another picture that the child
drew (p=.031) and to push them off a swing
(p=.003). Furthermore, 18 of 20 children (90%)
responded as predicted to more than half of the
forced-choice questions (p<.0005).
Justifications. Children’s justifications were
compared across the age groups. Justifications were
only included in analyses if children had answered
the prior forced-choice test question in the pre-
dicted way (following Vaish et al., 2011; but see
Appendix S1 for analyses including justifications
from all children). This resulted in thirteen 4-year-
olds and nineteen 5-year-olds being included, as
seven 4-year-olds and one 5-year-old did not
answer any of the forced-choice questions in the
hypothesized ways.
Additionally, three 4-year-olds and three 5-year-
olds did respond to one or more forced-choice
questions in the hypothesized way but did not pro-
vide any verbal responses to the respective justifica-
tion questions, so their data were excluded from
analyses (but note that including these children in
analyses and assigning their responses a score of 0
did not change the pattern of any of the results).
The analyses of justifications were thus conducted
with ten 4-year-olds and sixteen 5-year-olds.
When “Remorse (repeated)”was assigned a
score of 1, only five of the ten 4-year-olds provided
a Level-1 justification, whereas fourteen of the six-
teen 5-year-olds provided at least one Level-1 justi-
fication across all justification questions, indicating
a sophisticated level of understanding and reason-
ing about the transgressors and their responses
among the older children. Comparing across age
groups revealed that somewhat more 5-year-olds
than 4-year-olds provided at least one higher level
justification (Fisher’s exact test, p=.069). When
“Remorse (repeated)”was assigned a score of 0,
less than half of the 4-year-olds (3 of 10) still pro-
vided at least one Level-1 justification, whereas
more than half of the 5-year-olds (10 of 16) did so.
However, this difference between age groups was
no longer significant, p=.107. (Table 2 provides
the numbers obtained using both coding methods.)
Distribution of Resources
Distribution
Age comparisons. A chi-square analysis indi-
cated that a greater proportion of 5- than 4-year-
olds distributed more flowers to the remorseful
transgressor, v
2
(1, N=40) =8.64, p=.003, φ= .46.
The mean number of flowers given to the remorse-
ful transgressor was also higher among 5-year-olds
(M=1.9, SD =.45) than 4-year-olds (M=1.35,
SD =.75), t(38) =2.83, p=.007, Cohen’sd=0.89.
We thus analyzed each group separately.
4-Year-olds. Only eight of twenty 4-year-olds
(40%) gave more flowers (two or three of three) to
the remorseful transgressor (binomial probability,
p=.503). Twelve 4-year-olds (60%) gave more
flowers to the unremorseful transgressor. The mean
number of flowers given to the remorseful trans-
gressor (M=1.35, SD =.75) was not significantly
different from the mean number of flowers given to
the unremorseful transgressor (M=1.65, SD =.75),
t(19) =0.90, p=.379.
5-Year-olds. In contrast, seventeen of twenty
5-year-olds (85%) gave more flowers to the
remorseful transgressor (binomial probability,
p=.003). Three 5-year-olds (15%) gave more flow-
ers to the unremorseful transgressor. The mean
number of flowers given to the remorseful trans-
gressor (M=1.90, SD =.45) was significantly dif-
ferent from the mean number of flowers given to
the unremorseful transgressor (M=1.10, SD =.45),
t(19) =4.00, p=.001, Cohen’sd=1.78.
Justifications
Children’s justifications were compared across
the age groups. Justifications were only included in
analyses if children had distributed the resources as
hypothesized, that is, distributed more flowers to
the remorseful transgressor (but see Appendix S1
for analyses including justifications from all chil-
dren). Furthermore, assigning “Remorse (repeated)”
a score of 1 versus 0 did not change the pattern of
results, so we only report analyses in which it was
assigned a score of 1 (but see Table 2 for numbers
obtained with both coding methods). Of the eight
4-year-olds who distributed the resources as pre-
dicted, two provided no verbal justifications. Four
of the remaining six children in the 4-year-old
group provided Level-1 justifications. Of the seven-
teen 5-year-olds who distributed the resources as
predicted, one provided no verbal justification. Ten
of the remaining 16 children provided Level-1 justi-
fications. There was no significant difference across
age groups in the proportion of children who pro-
vided Level-1 justifications on the distribution of
resources task, Fisher’s exact test, p=1.000.
Discussion
As predicted, 5-year-olds were much more for-
giving of a transgressor who showed remorse after
causing them harm than a transgressor who
8 Oostenbroek and Vaish

showed no remorse. This was evident in their posi-
tive evaluations and expectations of the remorseful
transgressor, their negative evaluations and expec-
tations of the unremorseful transgressor, and their
distribution of more resources to the remorseful
transgressor. Four-year-olds, on the other hand, did
not show these differences in their evaluations,
expectations, or resource distribution. Thus,
between 4 and 5 years, children become capable of
forgiving those who have harmed them if the trans-
gressors display remorse.
Importantly, as the remorseful transgressor in
Study 1 did not apologize, children in our study
could not have relied on superficial cues of apolo-
gies to draw their inferences. Therefore, 5-year-olds
seem to truly understand the meaning of expres-
sions of remorse rather than simply relying on
whether the convention of offering an apology was
followed or not. Children’s justifications also pro-
vided evidence for this, as several 5-year-olds stated
that the remorseful transgressor had “said she was
sorry,”even though she had not offered an explicit
apology. This suggests that 5-year-olds have begun
to perceive the true meaning behind apologizing,
namely, to convey remorse and to demonstrate a
desire to make amends.
The results of Study 1 suggest that 4-year-olds
are no more forgiving of remorseful than unre-
morseful transgressors. Yet, it is unknown whether
the 4-year-olds’at-chance performance was attribu-
table to difficulties with the task or due to the
absence of explicit apologies. We know from Vaish
et al.’s (2011) Study 2 findings that when a third-
party transgressor’s remorse was accompanied by
explicit apologies, 4-year-olds did show the same
positive responses toward the apologetic transgres-
sor as the 5-year-olds did toward the remorseful
transgressor. Therefore, in Study 2, we asked
whether 4-year-olds would forgive an apologetic
transgressor and thus show an improved perfor-
mance compared to the 4-year-olds in Study 1 and
a more similar performance to the 5-year-olds in
Study 1. If so, then the 4-year-olds’at-chance per-
formance in Study 1 is unlikely to have been due to
procedural demands. The method of Study 2 was
identical to Study 1 except that the remorseful
transgressor now explicitly apologized. Based on
Vaish et al. (2011) and Smith and Harris (2012), we
predicted that 4-year-olds would forgive (that is,
prefer, evaluate more positively, make more posi-
tive generalizations about, and give more resources
to) a transgressor who explicitly apologized than an
unremorseful transgressor.
Study 2
Method
Participants
Twenty 4-year-olds (M=53;9; SD =2;63;
range =49;2–57;70; 10 girls) were included in the
study (no children were excluded). Data were col-
lected between February and April 2017. All fami-
lies provided information about race and
education: 75% of the parents were Caucasian,
and 90% of the parents were at least college edu-
cated. Children were recruited and tested as in
Study 1.
Design, Materials, and Setting
The design and materials of Study 2 were identical
to those in Study 1 with the only difference being
Table 2
Children’s Justifications in Study 1 and Study 2
Variable
5-Year-olds
Study 1: Remorse
Score
4-Year-olds
Study 1: Remorse
Score
4-Year-olds
Study 2: Apology
Score
1 010 1 0
“Remorse (repeated)”in Study 1 & “Apology”in Study 2 coded as 1, %
Test questions 70 (14) 10 (2) 25 (5) 25 (5) 65 (13) 25 (5)
Distribution of resources 50 (10) 30 (6) 20 (4) 10 (2) 35 (7) 25 (5)
“Remorse (repeated)”in Study 1 & “Apology”in Study 2 coded as 0, %
Test questions 50 (10) 30 (6) 15 (3) 35 (7) 30 (6) 60 (12)
Distribution of resources 40 (8) 40 (8) 15 (3) 15 (3) 5 (1) 55 (11)
Note. The values represent the percentage of children (raw numbers in parentheses) who provided at least one higher level justification
(score of 1) versus none (score of 0).
Forgiveness in Young Children 9

that after tearing the child’s picture, the remorseful
transgressor’s response was changed to: “Oh, I’ve
torn your picture. I’m sorry. I apologize.”Counter-
balanced variables were exactly as in Study 1.
Procedure
The procedure of Study 2 was identical to that of
Study 1 with only the following changes: (a) when
children were asked the second comprehension
probe about whether they could recall what the
remorseful transgressor said, if the child answered
correctly (“She said she was sorry”or “She apolo-
gized”or something similar), M said, “Yes, that’s
right. You’ve understood correctly,”and repeated
what the remorseful transgressor said. If, however,
the child answered incorrectly, M said, “Hmm, I’m
not sure about that,”and provided the child with
the correct information, and (b) the reminder that
M provided prior to asking the test questions was
changed to reflect the remorseful transgressor’s
explicit apology.
Coding and Reliability
Children’s responses were coded as in Study 1.
For a random 25% of the sample, children’s
responses were coded by a second coder (unaware
of the hypotheses). Reliability was perfect for all
measures: j=1 for the comprehension probes as
well as the forced-choice test questions, and 100%
agreement for the distribution of the three flowers.
Scores of 1 or 0 were again assigned to children’s
justifications. Justifications that indicated relevant
and sophisticated reasoning about the transgressors
and their responses, including justifications that
referred to feelings of remorse or apologies or
involved moral evaluations, were assigned a score of
1. However, unlike in Study 1, justifications involv-
ing apologies in Study 2 could be a product of
repeating phrases that had been used by the actresses
in the experiment or by M (e.g., “Because she said
sorry”or “Because she apologized”) rather than from
a true understanding of the transgressors’feelings or
motivations for apologizing. That is, children might
in fact have understood the transgressors’responses
in a sophisticated way but chose to repeat apology-
related phrases used by the actresses or by M, or they
may not have understood the transgressors’
responses in a sophisticated way and therefore sim-
ply repeated the phrases they heard. To account for
both possibilities, we conducted two sets of analyses
of children’s justifications: In one, “Apology”was
assigned a score of 1, and in the other, it was
assigned a score of 0. All other justifications were
assigned a score of 0 (see Table 3 for details of the
coding scheme). A second coder (unaware of the
hypotheses) coded justifications of a random 25% of
children. Reliability was perfect, j=1.
Results
Results will be reported in the same order as for
Study 1. Preliminary analyses revealed that for both
age groups, there were no significant effects of gen-
der. We therefore collapsed across this variable for
all analyses. For each measure of interest (i.e., com-
prehension checks, test questions and distribution
of resources task), we first report on the perfor-
mance of the 4-year-olds in Study 2 and then com-
pare their performance to that of the 4- and 5-year-
olds in Study 1. These cross-study comparisons
were made to evaluate whether the use of an expli-
cit apology improved the 4-year-olds’performance
in Study 2 to make it more similar to that of the 5-
year-olds in Study 1.
Comprehension Checks
Comprehension check 1. What happened to your
picture?
4-Year-olds, Study 2. Due to experimenter
error, one child was not asked this question. When
asked what had happened to their picture, eleven
of nineteen 4-year-olds (58%) correctly responded
that the two actresses had torn it; one child (5.5%)
responded that only one actress had torn it (the
remorseful transgressor); five children (26%)
responded that they did not know what happened
to their picture; and two children (10.5%) did not
provide any verbal responses. For the eight children
who did not answer correctly or provided no verbal
responses, M clarified that in fact both actresses
were holding the child’s picture and both had
torn it.
Age comparisons. Across studies, the number
of children who answered the first comprehension
check correctly in Study 2 was not significantly dif-
ferent from the number of 4-year-olds (p=.264) or
5-year-olds (p=.431) who answered it correctly in
Study 1.
Comprehension check 2. Can you remember what
[Anna or Kelly] said after your picture was torn?
4-Year-olds, Study 2. When asked what the
remorseful transgressor said, one child (5%) pro-
vided no verbal response. Of the 19 children who
provided a response, 11 (55%) answered correctly
10 Oostenbroek and Vaish

(binomial probability, p=.648) and eight (40%)
answered incorrectly. When asked what the unre-
morseful transgressor said, four children (20%) pro-
vided no verbal response. Of the 16 children who
provided a response, 9 (45%) answered correctly
(binomial probability, p=.804) and seven children
(35%) answered incorrectly.
Age comparisons with 4-year-olds in Study
1. Children’s responses across studies were com-
pared using chi-square tests. When comparing 4-
year-olds’comprehension in Study 1 versus Study
2, there was no significant difference between the
two groups for children’s recall of what the
remorseful transgressor (p=.508) or unremorseful
transgressor (p=.723) said.
Age comparisons with 5-year-olds in Study
1. Significantly more 5-year-olds in Study 1 than
4-year-olds in Study 2 correctly recalled what the
remorseful transgressor said after the picture was
torn (Fisher’s exact test, p=.010). However, there
was no significant difference between the 5-year-
olds in Study 1 and 4-year-olds in Study 2 for chil-
dren’s recall of what the unremorseful transgressor
said, (p=.465).
Test Questions
Forced-choice questions. Preliminary analyses
revealed no significant effects of children’s perfor-
mance on the comprehension checks on children’s
performance on the test questions. This variable
was thus not included in further analyses.
4-Year-olds, Study 2. Children’s responses to
the five test questions indicated that they drew
nearly all of the appropriate, hypothesized infer-
ences. The proportion of children who responded in
the hypothesized way on Questions 1 through 4
ranged from .75 to .85; binomial probabilities, all
ps<.041 (see Figure 2). Fourteen of 20 children
(70%) also responded in the hypothesized way for
Question 5 (“If you were playing on the swings,
who do you think would push you off?”), but this
proportion was not significantly above chance (bi-
nomial probability, p=.115). In spite of this, a sig-
nificant majority of children (17 of 20) responded in
the hypothesized way to more than half of the
forced-choice questions (p=.003).
Age comparisons with 4-year-olds in Study
1. To compare children’s performance across
studies and ages, we conducted chi-square analyses.
Significantly more 4-year-olds in Study 2 than those
in Study 1 responded in the hypothesized way to
two of the five forced-choice test questions (“Whom
are you more upset with?”,p=.008, φ= .42, and
“If you fell over, who do you think would help
you?”,p=.022, φ= .36; see Figure 2). Furthermore,
the proportion of questions answered in the
hypothesized way was significantly higher among
the 4-year-olds in Study 2 (M=0.78, SD =.32) than
the 4-year-olds in Study 1 (M=0.49, SD =.45),
t(38) =2.30, p=.027, Cohen’sd=0.74.
Age comparisons with 5-year-olds in Study
1. Results revealed that the 4-year-olds in Study 2
performed similarly to the 5-year-olds in Study 1.
Specifically, on all forced-choice questions, there
were no significant differences in the numbers of
4-year-olds in Study 2 versus 5-year-olds in Study 1
who responded in the hypothesized ways (all
Table 3
Coding Scheme for Justifications: Study 2
Score Category Content
1 Remorse Transgressor did (or did not) show remorse (child uses words other than
those used by the actresses in the experiment or by M); for example,
“Because she tore my picture and said it was her fault”
Moral character, evaluation, or norm Transgressor is a good (or bad) person, transgressor’s response to the
transgression was good (or bad), or transgressor broke (or did not break)
a moral norm; for example, “Because she said the right thing”or
“Because she’s a nicer person”
Apology [analyses were conducted
with this category scored as “1”and as “0”]
Transgressor did (or did not) apologize; for example,
“Because she said sorry”or “Because she apologized”
0 Own preference Child’s own preference for the transgressor
Action Transgressor tore the picture
Object The picture is torn or ripped and can no longer be repaired
Other, irrelevant, or uncodable Response could not be put into any of the above categories, was irrelevant,
or could not be coded
Forgiveness in Young Children 11

ps>.202). There was also no significant difference
between these two groups in the proportion of ques-
tions answered in the hypothesized way (p=.421).
Together, these analyses show that 4-year-olds’per-
formance on the forced-choice test questions
improved substantially in Study 2 when the remorse-
ful transgressor provided an explicit apology.
Justifications. As in Study 1, justifications were
included in analyses only if children had answered
the prior forced-choice test question in the pre-
dicted way (but see Appendix S1 for analyses
including justifications from all children). This
resulted in 18 children being included in the analy-
ses, as two children did not answer any of the
forced-choice test questions as predicted.
Assigning “Apology”a score of 1 versus 0 did
not change the pattern of results. We will thus only
report analyses in which it was assigned a score of
1 (but see Table 2 for numbers obtained with both
coding methods). The majority of 4-year-olds (13 of
18) provided at least one Level-1 justification across
the justification questions. This proportion was not
significantly different from the proportion of 4-
year-olds in Study 1 (Fisher’s exact test, p=.412) or
the proportion of 5-year-olds in Study 1 (Fisher’s
exact test, p=.405), who provided at least one such
justification when “Remorse (repeated)”was scored
1.
Distribution of Resources
Distribution
4-Year-olds, Study 2. Of the twenty 4-year-
olds in Study 2, one refused to distribute any
flowers. Of the remaining 19 children, 12 children
(63%) gave more flowers (two or three of three) to
the remorseful transgressor (binomial probability,
p=.359). Seven children (37%) gave more flowers
to the unremorseful transgressor. The mean number
of flowers given to the remorseful transgressor
(M=1.68, SD =.58) was not significantly different
from the mean number of flowers given to the
unremorseful transgressor (M=1.32, SD =.58), t
(19) =1.379, p=.185.
Age comparisons with the 4-year-olds Study
1. A chi-square analysis revealed no significant
difference in the proportion of 4-year-olds in Study
2 who gave more flowers to the remorseful trans-
gressor compared to the proportion of 4-year-olds
in Study 1 who did so (p=.148). The number of
flowers that 4-year-olds in Study 2 gave to the
remorseful transgressor (M=1.68, SD =.58) also
did not significantly differ from the number of
flowers that 4-year-olds in Study 1 gave to the
remorseful transgressor (M=1.35, SD =.75;
p=.128).
Age comparisons with the 5-year-olds in Study
1. A chi-square analysis revealed no significant
difference in the proportion of 4-year-olds in Study
2 who gave more flowers to the remorseful trans-
gressor compared to the proportion of 5-year-olds
in Study 1 who did so (p=.155). The number of
flowers that 4-year-olds in Study 2 gave to the
remorseful transgressor (M=1.68, SD =.58) also
did not significantly differ from the number of
flowers that 5-year-olds in Study 1 gave to the
remorseful transgressor (M=1.90, SD =.45;
p=.201).
Figure 2. Proportion of 4-year-old children in Study 1 and 4-year-old children in Study 2 who answered each test question and dis-
tributed resources in the hypothesized ways. The dashed line indicates chance level.
Note. *p<.05. **p<.01.
12 Oostenbroek and Vaish

Justifications
Children’s justifications were included in analy-
ses only if children had distributed the resources as
hypothesized, that is, distributed more flowers to
the remorseful transgressor (but see Appendix S1
for analyses including justifications from all chil-
dren). When “Apology”was assigned a score of 1,
over half of the 4-year-olds in Study 2 (7 of 12)
were scored as providing a Level-1 justification.
This proportion was not significantly different from
the proportion of 4-year-olds in Study 1 (Fisher’s
exact test, p=1.000) or the proportion of 5-year-
olds in Study 1 (Fisher’s exact test, p=1.000) who
provided a Level-1 justification when “Remorse (re-
peated)”was scored 1.
When “Apology”was assigned a score of 0, only
one of 12 children was scored as providing a Level-
1 justification. This proportion was not significantly
different from the proportion of 4-year-olds in
Study 1 (Fisher’s exact test, p=.083) but was signif-
icantly different from the proportion of 5-year-olds
in Study 1 (Fisher’s exact test, p=.039) who
provided a Level-1 justification when “Remorse
(repeated)”was assigned a score of 0.
Discussion
In Study 2, when a transgressor’s remorse was
accompanied by an explicit apology, even 4-year-
olds were willing to forgive her more than an unre-
morseful transgressor. They liked the remorseful
transgressor more, were more upset with the unre-
morseful transgressor, judged the remorseful trans-
gressor as more likely to help them, and judged the
unremorseful transgressor as more likely to tear
another picture that they drew. Across these ques-
tions, the 4-year-olds in Study 2 performed very
similarly to the 5-year-olds in Study 1, indicating
that for this younger age group, transgressors’
explicit apologies are important for eliciting
forgiveness.
Interestingly, unlike the 5-year-olds in Study 1,
4-year-olds in Study 2 did not report that they
expected the unremorseful transgressor to be more
likely to push them off a swing (even though they
did expect the unremorseful transgressor to be
more likely to tear another picture that they drew).
This might suggest that 4-year-olds limit their nega-
tive expectations about nonapologetic transgressors
to the specific transgressions that the transgressors
caused and do not form broader negative evalua-
tions that extend to novel situations (see Rholes &
Rubel, 1984). That would also help explain the lack
of difference in 4-year-olds’distribution of
resources to the apologetic versus nonapologetic
transgressor, as these children might simply not
have generalized their evaluations from the specific
transgression to the new domain of distributing
resources. Note, however, that the distribution of
resources task as well as the final test question
about the swing were presented at the very end of
the procedure, by which point the 4-year-olds
might have been fatigued or no longer able to
remember which transgressor had apologized. As
we will discuss further in the General Discussion,
our use of a live, in-person paradigm (rather than
videos or pictures) prevented us from providing
children with additional memory aids that could
have helped children’s performance. Still, all in all,
the results of Study 2 show that 4-year-olds do
forgive their transgressors if those transgressors
explicitly apologize.
General Discussion
When our cooperative relationships are damaged
by transgressions, they need to be repaired. One
part of this repair involves the transgressor display-
ing remorse, which conveys that the transgressor
regrets his actions and promises more acceptable
behavior in the future. However, remorse displays
alone are not sufficient; to successfully repair the
breached relationship, the victim must also respond
to the remorse and forgive the transgressor. In two
studies, we assessed the ontogeny of this latter pro-
cess by asking when children begin to forgive
remorseful transgressors. As predicted, results from
Study 1 revealed that 5-year-olds were more forgiv-
ing of a remorseful transgressor, even in the
absence of an apology. This was evident in their
positive evaluation of, preference for, and distribu-
tion of more resources to the remorseful transgres-
sor. Four-year-olds, on the other hand, did not
show any of these effects. In Study 2, however,
4-year-olds were more forgiving of a transgressor
who explicitly apologized than an unapologetic
transgressor. Our findings thus indicate that during
the preschool years, children acquire the ability to
forgive those who show remorse after causing them
harm. This forgiveness reestablishes children’s posi-
tive feelings toward their transgressors and fosters
reconciliation, thus helping to repair their relation-
ships and uphold cooperation (McCullough, 2008).
Our results contribute importantly to prior work
examining children’s responses to remorseful trans-
gressors. Unlike previous research that has typically
Forgiveness in Young Children 13

placed children in the role of bystanders, where
children can reason in a relatively “cool”cognitive
state (e.g., Darby & Schlenker, 1982, 1989; Smith
et al., 2010; Vaish et al., 2011), our findings demon-
strate that even as victims, when children may
experience stronger emotions such as sadness and
anger, 4- to 5-year-olds show sensitivity to remorse
and are more forgiving of remorseful transgressors.
Moreover, in the few prior studies in which chil-
dren were the victims, they responded to explicit
apologies, which are conventional cues of remorse
that children are trained from early on to use and
expect (Drell & Jaswal, 2016; Smith & Harris, 2012;
Smith et al., 2017); our findings demonstrate that
by at least 5 years of age, children forgive trans-
gressors who display remorse even in the absence
of explicit apologies (see also Vaish et al., 2011).
Transgressors’remorse is thus a powerful elicitor of
forgiveness, even in young children. Taken
together, we now have a clearer understanding of
when forgiveness first begins to emerge in develop-
ment: By as early as 4 years, children are willing to
forgive a remorseful transgressor, but only if con-
ventional cues of remorse such as apologies are pre-
sent. By 5 years, children are capable of forgiving a
remorseful transgressor even in the absence of
explicit apologies.
In Study 2, we found that 4-year-olds forgave a
transgressor who explicitly apologized more than
one who did not apologize, and that this forgive-
ness was very similar to the forgiveness shown by
5-year-olds to transgressors’remorse without apolo-
gies (Study 1). However, the improvement in
4-year-olds’performance from Study 1 (remorse
without apologies) to Study 2 (remorse with apolo-
gies) was not quite as striking as the improvement
in performance by 4-year-olds in Vaish et al. (2011).
Specifically, Vaish et al. found that 4-year-olds per-
formed consistently better on all four of their test
questions and the distribution of resources task
when the transgressor’s remorse involved an expli-
cit apology (Study 2) than no explicit apology
(Study 1). On the other hand, we found that the
4-year-olds in Study 2 did not perform significantly
better than the 4-year-olds in Study 1 on three of
the five test questions or on the distribution of
resources task (though the proportion of questions
they answered correctly in Study 2 was signifi-
cantly higher).
There are a few possible reasons for this discrep-
ancy between Vaish et al.’s (2011) and our results.
The most interesting possibility is that when chil-
dren are the victims themselves (as in our studies),
they may not respond as positively to transgressors’
remorse as when they are bystanders (as in Vaish
et al.). Thus, at least at age 4, the hot cognition that
children engage in as the victims of a transgression
might indeed make forgiveness somewhat more
challenging than the cooler cognition that children
engage in when they witness third-party transgres-
sions. To address this possibility head-on, future
work should directly compare children’s responses
when they are the victims versus bystanders in a
transgression.
Other, more methodological explanations are also
possible for the discrepancy. For instance, our live,
in-person paradigm may have proved more chal-
lenging for the younger children than other para-
digms (e.g., videos, vignettes, or hypothetical
scenarios). Specifically, our paradigm did not permit
us to provide children with additional cues and
memory aids, such as displaying images of the
transgressors’remorseful or unremorseful responses
throughout the question and distribution tasks, as
was the case in Vaish et al. (2011). Additionally,
whereas children in Vaish et al. witnessed two sets
of third-party transgressions and answered two sets
of the same questions, children in the current stud-
ies only experienced one transgression and
answered one set of questions (as we did not want
children to experience being harmed more than
once). In our live paradigm, therefore, children had
fewer opportunities to experience each transgres-
sor’s response (though the moderator did provide a
reminder of each response prior to asking the test
questions), and fewer opportunities to show their
forgiveness. These methodological differences might
partially explain why 4-year-olds in our Study 2 did
not show the same robust improvement in perfor-
mance as the 4-year-olds in Vaish et al.’s Study 2.
Nevertheless, note that 4-year-olds in our Study
2 did perform similar to the 5-year-olds in our
Study 1. Moreover, some 4-year-olds in both Stud-
ies 1 and 2 provided justifications to suggest that
their forced-choice responses were driven by for-
giveness-related aspects of the interactions. For
example, some children reasoned that the remorse-
ful transgressor “cares about me”and “said nicer
words to me,”while the unremorseful transgressor
“wasn’t being nice”and “was meaner.”There is
thus good reason to think that an understanding of
what transgressors’apologies (or lack thereof) rep-
resent is beginning to emerge around 4 years of age
(see also Smith & Harris, 2012).
It is also important to consider more generally
what might account for the age effects present in
our study. First, research on children’s understand-
ing of traits has shown that preschool-aged children
14 Oostenbroek and Vaish

have difficulties grasping that traits remain stable
and are predictive of behavior over time and across
different contexts (Rholes & Rubel, 1984; Ruble &
Dweck, 1995). Furthermore, preschool-aged children
need more behavioral examples to make trait attri-
butions, especially negative ones (Boseovski & Lee,
2006; Heyman & Dweck, 1998; Ronfard & Lane,
2017). Thus, perhaps the 4-year-olds in our study
were not as proficient as the 5-year-olds at predict-
ing the transgressors’future behaviors (e.g., who
would be likely to help or hurt them in the future)
based on the behaviors they had just witnessed,
and they may also have had difficulty generalizing
information about the transgressors to the new con-
text of distributing resources. Second, 5-year-olds’
emotional understanding is likely more advanced
than that of 4-year-olds (Harris, 1989; Pons, Law-
son, Harris, & de Rosnay, 2003). For instance, 5-
year-olds might have a more sophisticated capacity
to read others’emotions such that even in the
absence of scripted phrases such as “sorry,”they
can nonetheless understand how a remorseful per-
son feels and can thus sympathize with and forgive
that person. Relatedly, these older children might
also be better able to infer an apology from dis-
plays of remorse, and that inference might then eli-
cit forgiveness. Future work is needed to
investigate these possibilities further.
Our study is not without limitations. One concern
is the nature of our unremorseful transgressor’s
response. Specifically, in both studies, the unre-
morseful transgressor said, “Oh, I’ve torn your pic-
ture. Hmph [shrugging shoulders], I don’t care.”
Saying “I don’t care”and shrugging the shoulders
could be seen as somewhat negative rather than
entirely neutral, and it could thus be argued that
children were not only responding to the presence
or absence of remorse but also to the unremorseful
transgressor’s callousness. We found, however, that
it is challenging to compose a suitable neutral
response that is sufficiently distinct from an apolo-
getic or remorseful response. For example, if the
unremorseful transgressor had said, “That’s too
bad”or “Oh, now it’s torn,”these phrases could be
seen as expressions of sympathy and concern for the
victim, both of which are in fact part of expressing
remorse. We thus chose to use the direct translations
of the (German) phrases used by Vaish et al. (2011)
so as to make our results comparable with theirs.
Importantly, some children’s justifications did indi-
cate that they were using the remorse and not the
callousness of the transgressor to guide their choices
(e.g., “I’m more upset with her because she didn’t
say sorry”). Nonetheless, future work would benefit
from using more unambiguously neutral responses
(e.g., having the unremorseful transgressor simply
stay silent) in order to draw stronger conclusions
about children’s responses.
Another possible concern is how well children
comprehended the transgression. In response to the
comprehension checks in Study 1, only 40% of
4-year-olds stated that the two transgressors had
torn their picture whereas 70% of 5-year-olds did so.
This raises the possibility that the majority of 4-year-
olds did not comprehend the transgression and that
perhaps their performance on the test questions
reflects this lack of comprehension. Recall, however,
that children who answered the comprehension
questions incorrectly then received the correct infor-
mation from the moderator, and that all children
received a reminder with the correct information
prior to the test questions. Furthermore, if we only
consider those children who did answer the compre-
hension checks correctly, the pattern of results
remains very similar: The 4-year-olds still performed
at chance on all test questions whereas the 5-year-
olds drew the hypothesized inferences on four of the
five test questions and distributed the resources as
predicted (binomial probabilities, all ps<.013).
Additionally, as the 4-year-olds in Study 2 did per-
form above chance on nearly all of the test questions,
it is unlikely that the poor test performance of the 4-
year-olds in Study 1 was due to comprehension
problems, since the transgression was identical
across studies. Finally, in Vaish et al.’s (2011) Study
1, most 4-year-olds (approximately 80%) answered
the comprehension questions correctly, yet their per-
formance on the test questions was still at chance.
Considered together, it seems unlikely that poor
comprehension of the transgression explains chil-
dren’s test performance.
Nevertheless, we note that there are limitations
in how we assessed children’s comprehension. If
children answered incorrectly, the moderator imme-
diately gave children the correct information about
the event. It may have been helpful to ask children
a second time what had occurred, so that they
could generate the information about the transgres-
sion themselves. However, our primary aim was
not to test whether children comprehend the trans-
gression and the transgressors’responses on their
own, but rather to examine how children respond
when they have the necessary information. We
acknowledge, though, that children often need to
deduce this information on their own in real-world
scenarios and are not explicitly given information
about a transgression, so the format of these ques-
tions reduces ecologically validity.
Forgiveness in Young Children 15

A further issue concerns our use of forced-choice
test questions. The format of these questions made
certain assumptions about children’s responses
(e.g., the question “Whom are you more upset
with?”presumed that they were upset) and forced
participants to choose one of the transgressors
when in fact children may have wanted to choose
both or neither. Indeed, research has shown that
children are less discriminating on tasks where they
are presented with one individual rather than two
contrasting individuals (Boseovski & Lee, 2006;
Ronfard & Lane, 2017). While we acknowledge that
there are limitations in using forced-choice ques-
tions, we chose to use them because as a first step
in investigating children’s forgiveness, we believed
it was important to demonstrate that children can
discriminate between remorseful and unremorseful
transgressors and that they do show a systematic
preference at least within this more constrained for-
mat. We also wanted our findings to be comparable
to those of Vaish et al. (2011) and thus chose to use
their format of questions. Furthermore, we did
include justification questions, which were open-
ended and permitted children to provide reasons
for their choices. It is also important to note that
there may be potential limitations in only using
open-ended questions. For example, when children
are shy or otherwise unwilling to respond verbally,
open-ended questions may make it harder to elicit
responses whereas forced-choice questions allow
children to provide simple verbal or gestural
responses (e.g., pointing at one individual).
Nonetheless, it will be important for future research
to move away from forced-choice to single trans-
gressor scenarios and assess how children respond
when the direct contrast between remorseful and
unremorseful transgressors is absent. Based on
prior work (e.g., Drell & Jaswal, 2016; Smith & Har-
ris, 2012), we would expect our findings to hold
with such paradigms as well.
Another possible limitation of our procedure is
that we chose to use adult actresses instead of child
peers as the transgressors, which may have resulted
in a somewhat imbalanced power dynamic. Our
decision was based on practical considerations and
the desire to increase experimenter control, given
the challenge of having young children serve as
actors, particularly with a complex and demanding
procedure such as ours. It was also not practical to
match the gender of the actors and moderator to
that of the participant; as we did not find any gen-
der differences on any measure in either study,
though, it is unlikely that the gender of the experi-
menters had an effect. It would be interesting for
future research, however, to examine whether chil-
dren’s responses differ when the transgressors are
peers and gender-matched to participants.
It is also important to consider whether we truly
measured children’s forgiveness. Unlike in adult
research, where participants recall or engage in
(seemingly) serious transgressions (e.g., betrayals of
trust), it is not feasible to ask young children to
introspect or, for ethical reasons, to subject them to
serious transgressions. Therefore, transgressions
used in experiments with children are necessarily of
a more trivial nature. As a result, the forgiveness
we have examined may be of a more limited form
than the forgiveness often examined in adults. Nev-
ertheless, children’s responses in our studies indi-
cate that they felt more positively toward the
remorseful transgressor, evaluated her more favor-
ably, and for the 5-year-olds, were more likely to
cooperate with her (by giving her more resources).
These are precisely the outcomes that forgiveness is
argued to promote (McCullough, 2008), and indeed,
from the point of view of repairing and maintaining
cooperation, it is these outcomes that matter most.
We thus believe that children’s responses in the
current studies are indicative of early forgiveness.
The results of this study highlight a number of
interesting future research directions. Adult
research has shown that forgiveness is associated
with numerous positive outcomes for both victims
and offenders. For instance, greater levels of for-
giveness are related to lower stress levels and
greater positive emotions for victims, and enhanced
psychological and physiological well-being for both
victims and offenders (Berry & Worthington, 2001;
Karremans, Van Lange, Ouwerkerk, & Kluwer,
2003; Orcutt, 2006; Witvliet, Ludwig, & Vander
Laan, 2001). It would be interesting for future
research to consider if children also experience posi-
tive outcomes of forgiveness, both as victims and
transgressors. That is, do children feel happier after
they forgive and are forgiven themselves? Notably
though, forgiveness may not always be the most
appropriate response to transgressions, such as in
cases of transgressions that cause the victim great
harm (e.g., Lamb, 2006) or when transgressors do
not believe they did anything wrong (Adams, Zou,
Inesi, & Pillutla, 2015). An important avenue for
future research is whether children always forgive
remorseful transgressors or whether there are con-
texts in which they too withhold forgiveness.
Exploring whether children’s forgiveness is cali-
brated in this way will greatly enrich our under-
standing of the development, functions, and
nuances of forgiveness.
16 Oostenbroek and Vaish

In sum, humans rely on the cooperation of others
for our survival and to achieve individual and com-
munal goals (Tomasello et al., 2012). When coopera-
tive relationships break down, therefore, it is critical
to repair them. Prior work had demonstrated that
when children transgress against others, they experi-
ence remorse and are motivated to repair the harm
caused (Kochanska et al., 2002; Vaish et al., 2016).
The present studies demonstrate that when they are
the victims, children respond to transgressors’
remorse with forgiveness and so allow their broken
relationships to be repaired. From early in ontogeny,
then, humans are motivated to repair damaged rela-
tionships and thus uphold cooperation.
References
Adams, G. S., Zou, X., Inesi, M. E., & Pillutla, M. M.
(2015). Forgiveness is not always divine: When express-
ing forgiveness makes others avoid you. Organizational
Behavior and Human Decision Processes,126, 130–141.
https://doi.org/10.1016/j.obhdp.2014.10.003
Berry, J. W., & Worthington, E. L. (2001). Forgivingness,
relationship quality, stress while imagining relationship
events, and physical and mental health. Journal of Coun-
seling Psychology,48, 447–455. https://doi.org/10.1037/
0022-0167.48.4.447
Boseovski, J. J., & Lee, K. (2006). Children’s use of fre-
quency information for trait categorization and behav-
ioral prediction. Developmental Psychology,42, 500–513.
https://doi.org/10.1037/0012-1649.42.3.500
Castelfranchi, C., & Poggi, I. (1990). Blushing as discourse:
Was Darwin wrong? In W. R. Crozier (Ed.), Shyness and
embarrassment: Perspectives from social psychology (pp.
230–254). Cambridge, UK: Cambridge University Press.
https://doi.org/10.1017/CBO9780511571183
Darby, B. W., & Schlenker, B. R. (1982). Children’s reac-
tions to apologies. Journal of Personality and Social Psy-
chology,43, 742–753. https://doi.org/10.1037/0022-
3514.43.4.742
Darby, B. W., & Schlenker, B. R. (1989). Children’s reac-
tions to transgressions: Effects of the actor’s apology,
reputation and remorse. British Journal of Social Psychol-
ogy,28, 353–364. https://doi.org/10.1111/j.2044-8309.
1989.tb00879.x
Drell, M. B., & Jaswal, V. K. (2016). Making amends: Chil-
dren’s expectations about and responses to apologies.
Social Development,25, 742–758. https://doi.org/10.
1111/sode.12168
Harris, P. L. (1989). Children and emotion: The development
of psychological understanding. Cambridge, MA, US: Basil
Blackwell.
Heyman, G. D., & Dweck, C. S. (1998). Children’s think-
ing about traits: Implications for judgments of the self
and others. Child Development,69, 391–403. https://doi.
org/10.2307/1132173
Irwin, D. M., & Moore, S. G. (1971). The young child’s
understanding of social justice. Developmental Psychol-
ogy,5, 406–410. https://doi.org/10.1037/h0031614
Karremans, J. C., Van Lange, P. A. M., Ouwerkerk, J. W.,
& Kluwer, E. S. (2003). When forgiving enhances psy-
chological well-being: The role of interpersonal commit-
ment. Journal of Personality and Social Psychology,84,
1011–1026. https://doi.org/10.1037/0022-3514.84.5.1011
Keltner, D., & Anderson, C. (2000). Saving face for Dar-
win: The functions and uses of embarrassment. Current
Directions in Psychological Science,9, 187–192. https://
doi.org/10.1111/1467-8721.00091
Keltner, D., Young, R. C., & Buswell, B. N. (1997).
Appeasement in human emotion, social practice, and
personality. Aggressive Behavior,23,359–374. https://doi.
org/10.1002/(SICI)1098-2337(1997)23:5<359::AID-AB5>3.
0.CO;2-D
Kochanska, G., Gross, J. N., Lin, M.-H., & Nichols, K. E.
(2002). Guilt in young children: Development, determi-
nants, and relations with a broader system of stan-
dards. Child Development,73, 461–482. https://doi.org/
10.1111/1467-8624.00418
Lamb, S. (2006). Forgiveness, women, and responsibility
to the group. Journal of Human Rights,5,45–60.
https://doi.org/10.1080/14754830500485874
Leary, M. R., Landel, J. L., & Patton, K. M. (1996). The moti-
vated expression of embarrassment following a self-
presentational predicament. Journal of Personality,64,619–
636. https://doi.org/10.1111/j.1467-6494.1996.tb00524.x
McCullough, M. E. (2008). Beyond revenge: The evolution of
the forgiveness instinct. San Francisco, CA: Jossey-Bass.
McCullough, M. E., Worthington, E. L., & Rachal, K. C.
(1997). Interpersonal forgiving in close relationships.
Journal of Personality and Social Psychology,73, 321–336.
https://doi.org/10.1037/0022-3514.73.2.321
Metcalfe, J., & Mischel, W. (1999). A hot/cool-system
analysis of delay of gratification: Dynamics of will-
power. Psychological Review,106,3–19. https://doi.org/
10.1037/0033-295X.106.1.3
Ohbuchi, K., Kameda, M., & Agarie, N. (1989). Apology
as aggression control: Its role in mediating appraisal of
and response to harm. Journal of Personality and Social
Psychology,56, 219–227. https://doi.org/10.1037/0022-
3514.56.2.219
Orcutt, H. K. (2006). The prospective relationship of inter-
personal forgiveness and psychological distress symp-
toms among college women. Journal of Counseling
Psychology,53, 350–361. http://dx.doi.org/10.1037/
0022-0167.53.3.350
Petrucci, C. J. (2002). Apology in the criminal justice set-
ting: Evidence for including apology as an additional
component in the legal system. Behavioral Sciences and
the Law,20, 337–362. https://doi.org/10.1002/bsl.495
Pons, F., Lawson, J., Harris, P. L., & de Rosnay, M.
(2003). Individual differences in children’s emotion
understanding: Effects of age and language. Scandina-
vian Journal of Psychology,44, 347–353. https://doi.org/
10.1111/1467-9450.00354
Forgiveness in Young Children 17

Rholes, W. S., & Rubel, D. N. (1984). Children’s under-
standing of dispositional characteristics of others. Child
Development,55, 550–560. https://doi.org/10.2307/
1129966
Ronfard, S., & Lane, J. D. (2018). Preschoolers continually
adjust their epistemic trust based on an informant’s
ongoing accuracy. Child Development,89, 414–429.
https://doi.org/10.1111/cdev.12720
Ruble, D. N., & Dweck, C. S. (1995). Self-conceptions, per-
son conceptions, and their development. In N. Eisen-
berg (Ed.), Review of personality and social psychology.
Vol. 15: Social development (pp. 109–139). Thousand
Oaks, CA: Sage.
Smith, C. E., Blake, P. R., & Harris, P. L. (2013). I should
but I won’t: Why young children endorse norms of fair
sharing but do not follow them. PLoS ONE,8, e59510.
https://doi.org/10.1371/journal.pone.0059510
Smith, C. E., Chen, D., & Harris, P. L. (2010). When the
happy victimizer says sorry: Children’s understanding
of apology and emotion. British Journal of Developmental
Psychology,28, 727–746. https://doi.org/10.1348/
026151009X475343
Smith, C. E., & Harris, P. L. (2012). He didn’t want me to
feel sad: Children’s reactions to disappointment and
apology. Social Development,21, 215–228. https://doi.
org/10.1111/j.1467-9507.2011.00606.x
Smith, C. E., Noh, J. Y., Rizzo, M. T., & Harris, P. L.
(2017). When and why parents prompt their children to
apologize: The roles of transgression type and parent-
ing style. Journal of Family Studies,23,38–61. https://d
oi.org/10.1080/13229400.2016.1176588
Tomasello, M., Melis, A. P., Tennie, C., Wyman, E., &
Herrmann, E. (2012). Two key steps in the evolution of
human cooperation: The interdependence hypothesis.
Current Anthropology,53, 673–692. https://doi.org/10.
1086/668207
Vaish, A., Carpenter, M., & Tomasello, M. (2011). Young
children’s responses to guilt displays. Developmental Psy-
chology,47,1248–1262. https://doi.org/10.1037/a0024462
Vaish, A., Carpenter, M., & Tomasello, M. (2016). The
early emergence of guilt-motivated prosocial behavior.
Child Development,87, 1772–1782. https://doi.org/10.
1111/cdev.12628
Wellman, H. M., Larkey, C., & Somerville, S. C. (1979).
The early development of moral criteria. Child Develop-
ment,50, 869–873. https://doi.org/10.2307/1128956
Witvliet, C. v. O., Ludwig, T. E., & Vander Laan, K. L.
(2001). Granting forgiveness or harboring grudges:
Implications for emotion, physiology, and health. Psy-
chological Science,12, 117–123. https://doi.org/10.1111/
1467-9280.00320
Worthington, E. L. (2010). The new science of forgiveness.
In D. Keltner, J. Marsh, & J. A. Smith (Eds.), The
compassionate instinct: The science of human goodness (pp.
62–71). New York, NY: W. W. Norton & Company.
Zitek, E. M., Jordan, A. H., Monin, B., & Leach, F. R.
(2010). Victim entitlement to behave selfishly. Journal of
Personality and Social Psychology,98, 245–255. https://d
oi.org/10.1037/a0017168
Supporting Information
Additional supporting information may be found in
the online version of this article at the publisher’s
website:
Appendix S1. Full justification analyses
18 Oostenbroek and Vaish