Journal of Experimental Child Psychology 132 (2015) 221–229

Contents lists available at ScienceDirect

Journal of Experimental Child

Psychology
journal homepage: www.elsevier.com/locate/jecp

Brief Report

Children’s developing understanding of what and

how they learn
David M. Sobel ⇑, Susan M. Letourneau
Department of Cognitive, Linguistic, and Psychological Sciences, Brown University, Providence, RI 02912, USA

a r t i c l e i n f o a b s t r a c t

Article history: What do children know about learning? Children between 4 and
Available online 26 February 2015 10 years of age were asked what they thought the word learning
meant and then engaged in a structured interview about what
Keywords: kinds of things they learned and how they learned those things.
Children’s learning
Most of the 4- and 5-year-olds’ responses to these questions indi-
Metacognition
cated a lack of awareness about the nature of learning or how
Theory of mind
Knowledge change
learning occurs. In contrast, the 8- to 10-year-olds showed a strong
Source memory understanding of learning as a process and could often generate
Cognitive development explicit metacognitive responses indicating that they understood
under what circumstances learning would occur. The 6- and
7-year-olds were in a transitional stage between these two levels
of understanding. We discuss the implications of this development
with children’s theory-of-mind development more generally.
Ó 2015 Elsevier Inc. All rights reserved.

Introduction

A fundamental goal of research in cognitive development is to describe how children learn. There
are many descriptions of learning mechanisms that allow children to acquire knowledge (e.g., Carey,
2009; Gopnik & Meltzoff, 1997; Harris & Koenig, 2006; Mandler, 1992; Piaget, 1955; Saffran, Aslin, &
Newport, 1996; Smith & Heise, 1992; Tenenbaum, Kemp, Griffiths, & Goodman, 2011). Debates in cog-
nitive development revolve around what kind of learning mechanism best describes how children
acquire knowledge generally or in specific domains (see Piattelli-Palmarini (1980) for a classic discus-
sion and Johnson (2010) for a more contemporary one).

⇑ Corresponding author. Fax: +1 401 863 2255.

E-mail address: dave_sobel@brown.edu (D.M. Sobel).

http://dx.doi.org/10.1016/j.jecp.2015.01.004
0022-0965/Ó 2015 Elsevier Inc. All rights reserved.
222 D.M. Sobel, S.M. Letourneau / Journal of Experimental Child Psychology 132 (2015) 221–229

Another important, but often overlooked, goal of research in cognitive development is to describe
at what age children understand that learning has occurred. There is little evidence that young chil-
dren accurately reflect on their own learning, nor do researchers appreciate when children begin to
conceptualize learning as a process. It is also unclear whether understanding learning as a process
matters for reflecting on one’s own learning.
Investigations of metacognitive awareness suggest that preschoolers find articulating an under-
standing of learning difficult. Although children begin to talk about learning during the preschool
years (Bartsch, Horvath, & Estes, 2003), it is only after 4 years of age that they appreciate the relations
among various mental states that are necessary for learning to occur—for example, attention to the
task or a desire or intention to learn (Sobel, Li, & Corriveau, 2007). Preschoolers’ developing under-
standing of the mental states that are involved in learning has implications for their ability to monitor
and reflect on their own knowledge. When taught new pieces of knowledge, preschoolers often claim
that they knew it all along (Esbensen, Taylor, & Stoess, 1997; Taylor, Esbensen, & Bennett, 1994). Such
studies indicate that preschoolers are unable to understand mental states in a dynamic way—as active
and changing with thought and reflection (see, e.g., Flavell, Green, & Flavell, 1995; Johnson & Wellman,
1982).
Only a few studies have directly examined preschoolers’ abilities to reflect on their own learning.
Tang & Bartsch (2012; see also Tang, Bartsch, & Nunez, 2007) demonstrated that preschoolers typically
struggled to report when they learned a specific piece of information but could track the source of that
knowledge. When information was learned in laboratory settings, 4- and 5-year-olds displayed some
ability in describing how a particular piece of information was learned but not when it was learned.
One week after being shown or told a piece of information, children could state whether they were
shown or told the information but could not report that this was done a week prior.
Although these results suggest that children can track how they learn information, the controlled
nature of the task makes it difficult to generalize the findings to more naturalistic learning situations.
Children can learn different kinds of information in a multitude of ways, thereby making the forced-
choice response options in the previous study analogous to a recognition memory test; when asked
the test question, children only need to recognize one of the response options, not recall the actual
learning event from their memory. Bemis, Leichtman, and Pillemer (2011) considered this by asking
4- to 9-year-olds a series of factual questions that they were likely to be able to answer. Children were
then asked to describe how they had learned that piece of information. Even the youngest children in
their sample could generate details about how they learned the information, although there was sig-
nificant age-related change (i.e., older children could generate more details) and differences between
genders (i.e., girls generated more instances of source monitoring than boys). Bemis, Leichtman, and
Pillemer (2013) followed up on this finding by first teaching 4- and 5-year-olds new pieces of infor-
mation and then, in a subsequent session, examining whether those children remembered how they
had learned it. They again found that even the 4-year-olds could generate accurate details about how
they learned the information and, therefore, suggested that young children possessed some under-
standing of learning itself.
That said, those authors’ data suggest that this understanding develops. Bemis and colleagues
(2011) showed that only 25% of 4- and 5-year-olds generated an account of where they learned a par-
ticular fact, whereas 7- to 9-year-olds did so only 45% of the time. Moreover, because this study asked
children to recall how they had learned arbitrary facts that they were exposed to at some point in the
past (see also Tang & Bartsch, 2012), differences among ages might reflect children’s developing
source memory capacities (e.g., Lindsay, Johnson, & Kwon, 1991).
The current investigation considers what children know about learning in general and whether
they can reflect on instances of their own learning. Children were asked to describe what they believed
learning was. Then, they were asked to generate their own examples of what and how they had
learned in the past. Our goal was to document whether children regarded learning as a process
(articulating something about their own knowledge changing) or whether they conceived of learning
as the content or knowledge itself. Moreover, we were interested in whether children’s understanding
of learning as a process influenced the way in which they reflected on particular instances of their own
past learning.
 D.M. Sobel, S.M. Letourneau / Journal of Experimental Child Psychology 132 (2015) 221–229 223

Method

Participants

In total, 101 children (41 male and 60 female, mean age = 79.04 months, range = 48–131) were
recruited through a list of birth records and from a local children’s museum. For ease of analysis,
we divided our sample into three age groups: 4- and 5-year-olds (n = 43, 17 male and 26 female, mean
age = 60.72 months, range = 48.0–70.3), 6- and 7-year-olds (n = 39, 16 male and 23 female, mean
age = 81.71 months, range = 72.0–94.5), and 8- to 10-year-olds (n = 19, 8 male and 11 female,
mean age = 115.11 months, range = 97.2–131.6). No formal measures of socioeconomic status or
ethnicity were taken.

Procedure

All children were tested in one-on-one interviews conducted in a quiet room at a children’s muse-
um or in a testing room at a laboratory. Interviews were capped at 5 min so that children would not
become distracted or bored.
At the start of the interview, children were asked an open-ended learning question: ‘‘What do you
think ‘learning’ means?’’ If children did not offer a response or did not understand, the question was
restated: ‘‘What does it mean ‘to learn’?’’ After children answered this open-ended question, the
experimenter prompted children to provide specific examples of learned content by asking, ‘‘Can
you think of something that you have learned?’’ and, as a follow-up, ‘‘How did you learn that?’’ These
two questions were repeated to allow children to give multiple examples of what they had learned
until they had no further examples to give. Finally, children were asked to offer specific examples
of process, ‘‘How else can you learn?’’ with an additional prompt, ‘‘What are some other ways you
can learn?’’ (if children did not respond or did not understand the question). Again, this question
was repeated to allow children to give multiple examples until they had no further examples to give
or until the 5 min allocated for the interview was complete. All children were able to answer all of the
interview questions and were given opportunities to provide more than one example in response to
each question within the time available for the interview.

Coding

All interviews were videotaped and then transcribed for coding. Children’s definitions of learning
were divided into the following categories: (a) no response, including ‘‘I don’t know,’’ ‘‘I’m not sure,’’
and no answer; (b) identity responses, in which children simply used the word ‘‘learn’’ or ‘‘learning’’
to define learning (e.g., ‘‘learning is when you learn’’); (c) content responses, in which children defined
learning as involving a subject or topic that was or could be learned (e.g., ‘‘like reading and math’’);
and (d) process responses, in which children defined learning as involving either a source (e.g., ‘‘when
your teacher tells you something’’) or a strategy (e.g., ‘‘when you practice again and again until you
know it’’) that would result in gaining knowledge.
Responses to the next question—about children’s examples of learned content—were divided into
the following categories: (a) no response, because some children were unable to offer any examples
of what they had learned; (b) subjects, either academic or proto-academic topics or other generalizable
knowledge (e.g., reading, counting); (c) skills, either motor skills or procedural knowledge (e.g., how to
tie shoes); (d) conventions, including social and nonsocial rules (e.g., red lights mean stop); and (e)
facts, non-generalizable knowledge such as single observations or statements of trivia (e.g., ants have
six legs). When children gave multiple examples of their own learning, each example was coded
separately.
Finally, children’s examples of process (their descriptions of how they had learned each content
example and their examples of how else they could learn in general) were divided into the following
categories: (a) no response, because some children were unable to state how they had learned or could
learn; (b) source, citing a person (e.g., ‘‘from my teacher’’) or a place (e.g., ‘‘in school’’) as the source of
224 D.M. Sobel, S.M. Letourneau / Journal of Experimental Child Psychology 132 (2015) 221–229

knowledge; and (c) strategy, involving an active process through which knowledge was gained (e.g.,
practicing, reading it in a book, watching someone else). Unlike the previous two examples, the cate-
gories here were not mutually exclusive (with the exception of no response). For example, children
could state that they learned from their teacher and also by practicing. Children’s responses to the fol-
low-up question, ‘‘How did you learn that?’’ (following each content example) and their responses to
the question, ‘‘How else can you learn?’’ were each coded separately.
An undergraduate research assistant, blind to the purpose of the experiment, coded the entire data-
set. A second undergraduate research assistant (also blind to the purpose of the experiment) coded a
subset of these responses (25% of the dataset). Both coders were blind to the age and gender of the
participants they were scoring. Agreement was 90%, with discrepancies resolved through discussion
with the second author.

Results

Table 1 shows the distribution of children’s responses to the first open-ended learning question,
‘‘What do you think ‘learning’ means?’’ Preliminary analyses showed that for none of our dependent
measures did we find a significant difference between boys and girls in the sample; therefore, gender
was not included in the remainder of the analyses.
The 4- and 5-year-olds were frequently unable to offer any definition of learning (39.53% of this age
group), whereas 41.86% of this age group gave process-based responses. The 6- and 7-year-olds sel-
dom failed to respond (only 10.26% of children in this age group offered no response); instead, their
responses were divided among the other response categories, with 66.67% of their responses catego-
rized as process based. The 8- to 10-year-olds most frequently gave process-based definitions of learn-
ing (94.74% of children in this age group), and none failed to respond to the question. The extent to
which children were coded as giving process-based definitions of learning was significantly related
to their age (in months), q(99) = .49, p < .01. We were concerned that this correlation resulted from
children’s developing language capacities. Although we did not explicitly measure such capacities,
we did measure the number of words children used to define learning. The correlation between age
and the presence of a process-based definition of learning remained significant when controlling for
the length of children’s definitions, q(98) = .28, p < .01. This suggests that the relations between age
and process definitions of learning cannot be explained solely by developmental differences in verbal
ability. We discuss this further below.
Because our sample of 8- to 10-year-olds mostly generated process definitions, we wanted to
ensure that the significant correlation with age was not due to their inclusion. This correlation
remained significant when only the 4- to 7-year-olds were considered, q(80) = .38, p < .01, and when
the 4- and 5-year-olds were compared directly with the 6- and 7-year-olds, we observed more pro-
cess-based definitions of learning in the older group, v2(1, N = 82) = 5.06, p = .02, / = .25. This suggests
that our significant results were not due to the oldest age group’s ceiling performance.
Next, children were asked to give specific examples of content they had learned in the past and
were prompted to describe the processes through which they had learned. Our goal was to determine
whether children who initially gave process definitions of learning would give qualitatively different
examples of what they had learned. These questions also allowed us to examine whether children

Table 1
Numbers of children providing different definitions of learning by age group.

Response category
Age group (years) No response Identity Content Process
4–5 17 (39.53) 1 (2.33) 7 (16.28) 18 (41.86)
6–7 4 (10.26) 6 (15.00) 3 (7.69) 26 (66.67)
8–10 0 (0) 0 (0) 1 (5.26) 18 (94.74)
Total 21 (20.79) 7 (6.93) 11 (10.89) 62 (61.39)

Note. Percentages of responses within age groups are in parentheses.

 D.M. Sobel, S.M. Letourneau / Journal of Experimental Child Psychology 132 (2015) 221–229 225

Table 2
Numbers of children providing examples of content across age group and whether they provided a process-based definition of
learning.

Age (years) n Response type

Subject Skill Convention Fact No response
Process definitions 4–5 18 7 (38.89) 11 (61.11) 3 (16.67) 6 (33.33) 0 (0)
6–7 26 20 (76.92) 13 (50.00) 3 (11.54) 8 (30.77) 0 (0)
8–10 18 16 (88.89) 13 (72.22) 2 (11.11) 5 (27.78) 0 (0)
Total 62 43 (69.35) 37 (59.68) 8 (12.90) 19 (30.65) 0 (0)
Other definitions 4–5 25 12 (48.00) 5 (20.00) 4 (16.00) 3 (12.00) 8 (32.00)
6–7 13 9 (69.23) 0 (0) 0 (0) 3 (23.08) 3 (23.08)
8–10 1 1 (100) 1 (100) 0 (0) 0 (0) 0 (0)
Total 39 22 (56.41) 6 (15.38) 4 (10.26) 6 (15.38) 11 (28.21)

Note. Counts indicate the numbers of children who gave each type of response. Percentages of responses within age groups are
in parentheses. With the exception of ‘‘no response,’’ response types are not mutually exclusive.

would be able to describe learning processes in the context of a specific example (through the follow-
up questions).
Children generated between 0 and 6 content examples within the 5-min interview time (M = 2.33,
SD = 1.36). There was a significant correlation between children’s age in months and the number of
examples they gave, q(99) = .27, p < .01. This appeared to be driven by the fact that several 4- and
5-year-olds offered no examples at all (eight children in this age group vs. three 6- and 7-year-olds
and no 8- to 10-year-olds). Because interviews were capped at 5 min, and children elaborated on
the follow-up questions to different extents, the remainder of the analyses involved the number of
children who offered each type of example (e.g., whether the child ever offered a subject example
or a skill example over the course of the interview), not the total number of examples given.
The number of children offering each type of content example is shown in Table 2. Children’s age in
months related to whether they generated a subject response, q(99) = .39, p < .01, and a skill response,
q(99) = .24, p = .01, but did not relate to any other category. These were the two most popular respon-
se types (offered by 64.36% and 42.57% of children, respectively), even among the youngest age
groups. Direct comparisons between age groups revealed that for subject examples, 6- and 7-year-olds
were more likely than 4- and 5-year-olds to offer this response type, v2(1, N = 82) = 7.67, p < .01,
/ = .31, whereas 6- and 7-year-olds and 8- to 10-year-olds did not differ, v2(1, N = 58) = 1.78,
p = .18. For skill examples, 8- to 10-year-olds were more likely to offer this kind of example than
6- and 7-year-olds, v2(1, N = 58) = 8.36, p < .01, / = .38, whereas the two younger age groups did
not differ, v2(1, N = 82) = 0.134, p = .71.
We also analyzed whether generating a process-based definition of learning correlated with gen-
erating each type of content example. This analysis revealed significant correlations with skill exam-
ples, q(99) = .44, p < .01, and a marginal correlation with fact examples, q(99) = .17, p = .09. These
correlations remained present even when factoring out children’s age in months: skill, q(98) = .38,
p < .01; fact, q(98) = .18, p = .08. There were no significant relations for subject or convention
responses.1
For each topic that children said they had learned, the follow-up question, ‘‘How did you learn
that?’’ prompted them to describe their learning process in the context of that specific example.
Table 3 shows the distribution of these responses, again broken down by age and by whether children
generated a process-based description of learning in response to the first question. As with the
analyses of content examples, analyses of these responses to this question involved the number of
children who offered each type of example (e.g., whether the child ever offered a source or strat-
egy-based response), not the total number of examples given. Children’s age correlated with whether
they generated a strategy-based response or a source response, q(99) = .32 and .30, respectively, both

1
For this and subsequent analyses, there was also a significant negative correlation between generating a process-based
definition of learning and not responding to this question, q(99) = .44, p < .01.
226 D.M. Sobel, S.M. Letourneau / Journal of Experimental Child Psychology 132 (2015) 221–229

Table 3
Numbers of children who described different kinds of processes in response to ‘‘How did you learn that?’’ based on whether they
generated a process definition of learning.

Age (years) n Response type

Strategy Source None
Process definitions 4–5 18 7 (38.89) 13 (72.22) 7 (38.89)
6–7 26 14 (53.85) 22 (84.62) 5 (19.23)
8–10 18 12 (66.67) 14 (77.78) 4 (22.22)
Total 62 33 (53.23) 49 (79.03) 16 (25.81)
Other definitions 4–5 25 7 (28.00) 8 (32.00) 19 (76.00)
6–7 13 2 (15.38) 7 (53.85) 7 (53.85)
8–10 1 1 (100) 0 (0) 1 (100)
Total 39 10 (25.64) 15 (38.46) 27 (69.23)

Note. Counts indicate the number of children who gave each type of response. Percentages of responses within age groups are in
parentheses. Strategy and Source response types are not mutually exclusive.

Table 4
Numbers of children who described different kinds of processes in response to ‘‘How else can you learn?’’ based on whether they
generated a process definition of learning.

Age (years) n Response type

Strategy Source None
Process definitions 4–5 18 7 (38.89) 4 (22.22) 9 (50.00)
6–7 26 10 (38.46) 12 (46.15) 6 (23.08)
8–10 18 16 (88.89) 6 (33.33) 0 (0)
Total 62 33 (53.23) 22 (35.48) 15 (24.19)
Other definitions 4–5 25 8 (32.00) 5 (20.00) 14 (56.00)
6–7 13 4 (30.77) 3 (23.08) 7 (53.85)
8–10 1 1 (100) 0 (0) 0 (0)
Total 39 13 (33.33) 8 (20.51) 21 (53.85)

Note. Counts indicate the number of children who gave each type of response. Percentages of responses within age groups are in
parentheses. Strategy and Source response types are not mutually exclusive.

p values < .01. When age groups were compared directly, 6- and 7-year-olds were more likely than
4- and 5-year-olds to generate source examples, v2(1, N = 82) = 5.60, p = .02, / = .26, whereas the
two older age groups did not differ, v2(1, N = 58) = 0.00, p = .96. In contrast, 8- to 10-year-olds were
more likely than 6- and 7-year-olds to generate strategy examples, v2(1, N = 58) = 3.84, p = .05,
/ = .26, whereas the two younger age groups did not differ, v2(1, N = 82) = 0.63, p = .43.
Process definitions of learning also correlated with children’s descriptions of strategies, q(99) = .27,
p < .01, and sources, q(99) = .41, p < .01. This correlation with source examples held when controlling
for age, q(98) = .32, p < .01; however, correlations with descriptions of strategies were nonsignificant
after controlling for age, q(98) = .14, p = .16.
Finally, responses to the question, ‘‘How else can you learn?’’ are shown in Table 4. As with previ-
ous analyses, analyses of responses to this question involved the number of children who offered each
type of example (e.g., whether the child ever offered a source or strategy-based response) rather than
the total number of examples given. Generation of a strategy-based response was correlated with chil-
dren’s age, q(99) = .34, p < .01, and with the presence of a process definition of learning, q(99) = .19,
p = .05. These correlations were not found for descriptions of sources.
When age groups were compared, 6- and 7-year-olds were marginally more likely than 4- and
5-year-olds to give a source example, v2(1, N = 82) = 3.04, p = .08, / = .19, whereas the two older age
groups did not differ, v2(1, N = 58) = 0.26, p = .61. For strategy examples, 8- to 10-year-olds were more
likely than 6- and 7-year-olds to generate this kind of response, v2(1, N = 58) = 14.74, p < .01,
/ = .50, whereas the two younger age groups did not differ, v2(1, N = 82) = 0.00, p = .92. The relation
between generating a process-based definition of learning and generating a strategy-based response
to this question did not remain significant when age was factored out of the analysis, q(98) = .04, p = .73.
 D.M. Sobel, S.M. Letourneau / Journal of Experimental Child Psychology 132 (2015) 221–229 227

Discussion

The current study investigated the relations among children’s ability to define ‘‘learning’’ in gener-
al, their age, and their descriptions and understanding of particular instances of their own learning.
The ability to define learning as a process develops roughly between 4 and 8 years of age and is related
both to different types of self-generated examples of learning and to changes in children’s descriptions
of the ways in which they had learned in the past.
One potential reason why 4- and 5-year-olds generated less sophisticated definitions of learning or
examples of their own learning was that they failed to understand the nature of the question.
However, we do not think that this is the case. Natural language analyses have shown that children
hear the words ‘‘learn’’ and ‘‘learning’’ in conversations with adults and use these words appropriately
before their fourth birthday (Bartsch et al., 2003; Sobel et al., 2007). Given this, we might ask why
many 4- and 5-year-olds failed to generate a response to the first open-ended ‘‘What is ‘learning’?’’
question in our interview. Simply because children use these words in natural language does not mean
that they have the metacognitive capacity to reflect on the word’s meaning. Our use of the open-ended
interview format attempted to elicit this more metacognitively complex ability, which appears to be
still developing.2
Previous research on children’s understanding of learning has also often asked children to reflect on
the mental states involved in others’ learning, not their own, and has involved the learning of arbitrary
pieces of information as opposed to their own knowledge. This raises the question of how children
apply their emerging understandings of mental states in these scripted situations to their own learn-
ing in everyday life. Bemis and colleagues (2011, 2013) did ask preschoolers about particular instances
of their own learning, but they also involved somewhat arbitrary facts and information. In the current
study, we found that with age, children were increasingly likely to describe learning in general as a
process (involving mental processes such as strategies and sources) and that this understanding
was related to the kinds of descriptions they offered of their own learning. Children rarely generated
isolated facts when asked for instances of their own learning, which might account for why Bemis and
colleagues found positive but nascent capacities as they limited their investigation to this kind of
knowledge. Moreover, by asking children to generate their own examples of what they had learned,
the current study may have made it easier for children to reflect on their learning by allowing them
to focus on topics or examples that were most meaningful to them.
Here, we found that even the youngest children could often cite an example of something they had
learned, and for those children with process-based definitions of learning, the examples that they pro-
vided increasingly involved not only subjects and singular facts but also skills (e.g., motor or procedu-
ral knowledge). This reflects their view of learning as an active process and even a change in their
conception of the type of knowledge that can result from this process. Critically, this relation remained
even when controlling for age.
One other significant difference between our results and those of Bemis and colleagues (2011,
2013) is the gender difference that they observed. In their work, girls appealed more to instances of
what the authors called ‘‘active learning’’—engaging in an activity with the goal of acquiring knowl-
edge. These results might suggest differences between genders, specifically that boys in our study
should have been more likely to describe sources and girls should have been more likely to describe
strategies. However, we did not find any significant differences between boys and girls in the age-re-
lated changes that we saw or their tendency to cite sources/strategies (or in general throughout our
investigation). This might be because self-generated examples made the task moderately easier or
helped children to reflect on their own learning.
Finally, the current study potentially relates to children’s developing conceptions of teaching and
knowledge. Although learning and teaching are highly related (LeBlanc & Bearison, 2004; Strauss &
Shilony, 1994), the current data suggest specific ways in which they might also be separate or only

2
We ran a second interview with a group of 4- and 5-year-olds (n = 18), replicating our procedure, replacing ‘‘learning’’ with
‘‘figure out’’ (i.e., ‘‘What does it mean to figure something out?’’ ‘‘Can you tell me times you have figured something out?’’ etc.).
Although children generated fewer proto-academic examples when asked about specific content, the general pattern of results
echoes the reported data. As such, we do not believe that the difficulty with the word ‘‘learn’’ underlies young children’s results.
228 D.M. Sobel, S.M. Letourneau / Journal of Experimental Child Psychology 132 (2015) 221–229

partially overlapping. For example, children’s developing understanding of belief appears to have a
direct relation with their understanding of teaching, but not necessarily with their understanding of
learning. Ziv & Frye (2004; Ziv, Solomon, & Frye, 2008; see also Strauss, Ziv, & Stein, 2002) suggested
that children who succeeded on false belief measures understood more about when another person
was engaged in a teaching act and would use their knowledge of others’ beliefs to decide whom a tea-
cher would teach. Although we did not measure children’s false belief capacities, given the ages of the
children in our study, most of our sample should pass a standard false belief measure (Wellman, Cross,
& Watson, 2001) and yet children’s understanding of learning changed markedly with age. This sug-
gests that false belief capacities may be independent of children’s developing understanding of learn-
ing as a process or may only partially contribute to their general understanding of learning.
In contrast, children’s understanding of learning and teaching might be more closely related in
other situations such as when children are explicitly instructed to facilitate another child’s learning.
There is evidence of teaching-like behavior in children younger than 7 years (specifically when a more
knowledgeable child is paired with a less knowledgeable one; see Ashley & Tomasello, 1998; Azmitia,
1988). These studies, however, did not explicitly ask children to engage in such behavior (much like
the distinction between children’s use of the word ‘‘learning’’ in their natural conversation and the
current findings that young children might struggle to define learning explicitly). Wood, Wood,
Ainsworth, and O’Malley (1995) asked children to engage in teaching behavior with less knowledge-
able peers and found that 7-year-olds were more likely than 3- and 5-year-olds to interact contingent-
ly on one another, with the more knowledgeable children assessing their peers’ level of understanding
as they attempted to complete a task and offering assistance only when needed. Wood and colleagues
appeal to older children’s understanding the recursive nature of belief (e.g., Perner & Wimmer, 1985)
as the mechanism that allows for this explicit strategy.
We observed similar patterns of development. By 7 years of age, children begin not only to define
learning as a process but also to articulate the way in which they learn information and skills through
enacting such processes. By this age, children can also hold multiple perspectives and understand that
others might interpret events differently from themselves (or that two people might interpret the
same event differently; see, e.g., Carpendale & Chandler, 1996; Eisbach, 2004; Heiphetz, Spelke,
Harris, & Banaji, 2013; Lagattuta & Wellman, 2001). This development might underlie children’s abil-
ity to integrate the roles of various mental states in learning (Sobel et al., 2007) as well as their ability
to reflect on learning as a dynamic process (in the current study). A prediction that comes out of this
work, therefore, would be that children who generate process-based definitions of learning may be
more likely to succeed on these ‘‘interpretive’’ theory-of-mind measures.
An alternative possibility is that individual differences among children, as well as the developmen-
tal differences observed here, are due to children’s developing language capacities. Although we can-
not rule out this possibility, we do think that it is unlikely. The length of children’s definition did not
affect the relation between age and the type of definition children generated. This suggests that the
overall amount of language children generate is unrelated to their ability to define learning. However,
future studies could administer an age-appropriate standardized measure of language competence
(e.g., the Peabody Picture Vocabulary Test) to examine whether this factor influences children’s ability
to demonstrate their understanding of learning.
To conclude, the current study suggests that by 8 years of age, children understand learning as a
process and can reflect on the ways in which they learned in the past. Such capacities appear to be
developing between 4 and 8 years of age, but regardless of age, having a process-based understanding
of learning is associated with children being able to offer more varied examples of what they have
learned (describing skills and not just topics or pieces of information) and an improved ability to
reflect on the ways in which they have learned.

Acknowledgments

This work was supported by the National Science Foundation (1223777). We thank Chris Erb, Eva
Lai, Deanna Macris, and Tiffany Tassin for helpful discussions and thank Jenna Eldridge, Chris Erb, and
Julia Franckh for assistance with data collection and analysis.
 D.M. Sobel, S.M. Letourneau / Journal of Experimental Child Psychology 132 (2015) 221–229 229

References

Ashley, J., & Tomasello, M. (1998). Cooperative problem-solving and teaching in preschoolers. Social Development, 7, 143–163.
Azmitia, M. (1988). Peer interaction and problem solving: When are two heads better than one? Child Development, 59, 87–96.
Bartsch, K., Horvath, K., & Estes, D. (2003). Young children’s talk about learning events. Cognitive Development, 18, 177–193.
Bemis, R. H., Leichtman, M. D., & Pillemer, D. B. (2013). I remember when you taught me that!: Preschool children’s memories of
realistic learning episodes. Infant and Child Development, 22, 603–621.
Bemis, R. H., Leichtman, M. D., & Pillemer, D. B. (2011). ‘‘I remember when I learned that!’’: Developmental and gender
differences in children’s memories of learning episodes. Infant and Child Development, 20, 387–399.
Carey, S. (2009). The origin of concepts. New York: Oxford University Press.
Carpendale, J. I., & Chandler, M. J. (1996). On the distinction between false belief understanding and subscribing to an
interpretive theory of mind. Child Development, 67, 1686–1706.
Eisbach, A. O. D. (2004). Children’s developing awareness of diversity in people’s trains of thought. Child Development, 75,
1694–1707.
Esbensen, B. M., Taylor, M., & Stoess, C. (1997). Children’s behavioral understanding of knowledge acquisition. Cognitive
Development, 12, 53–84.
Flavell, J. H., Green, F. L., & Flavell, E. R. (1995). Young children’s knowledge about thinking. Monographs of the Society for
Research in Child Development, 60(1, Serial No. 243).
Gopnik, A., & Meltzoff, A. N. (1997). Words, thoughts, and theories. Cambridge, MA: MIT Press.
Harris, P. L., & Koenig, M. A. (2006). Trust in testimony: How children learn about science and religion. Child Development, 77,
505–524.
Heiphetz, L., Spelke, E. S., Harris, P. L., & Banaji, M. R. (2013). The development of reasoning about beliefs: Fact, preference, and
ideology. Journal of Experimental Social Psychology, 49, 559–565.
Johnson, C. N., & Wellman, H. M. (1982). Children’s developing conceptions of the mind and brain. Child Development, 53,
222–234.
Johnson, S. P. (2010). Neoconstructivism: The new science of cognitive development. New York: Oxford University Press.
Lagattuta, K. H., & Wellman, H. M. (2001). Thinking about the past: Early knowledge about links between prior experience,
thinking, and emotion. Child Development, 72(1), 82–102.
LeBlanc, G., & Bearison, D. J. (2004). Teaching and learning as a bi-directional activity: Investigating dyadic interactions between
child teachers and child learners. Cognitive Development, 19, 499–515.
Lindsay, D. S., Johnson, M. K., & Kwon, P. (1991). Developmental changes in memory source monitoring. Journal of Experimental
Child Psychology, 52, 297–318.
Mandler, J. M. (1992). How to build a baby II: Conceptual primitives. Psychological Review, 99, 587–604.
Perner, J., & Wimmer, H. (1985). ‘‘John thinks that Mary thinks that . . .’’: Attribution of second-order beliefs by 5- to 10-year-old
children. Journal of Experimental Child Psychology, 39, 437–471.
Piaget, J. (1955). The construction of reality in the child. New York: Basic Books.
Piattelli-Palmarini, M. (1980). Language and learning: The debate between Jean Piaget and Noam Chomsky. Cambridge, MA:
Harvard University Press.
Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science, 274, 1926–1928.
Smith, L. B., & Heise, D. (1992). Perceptual similarity and conceptual structure. In B. Burns (Ed.), Percepts, concepts, and categories
(pp. 233–272). Amsterdam: Elsevier.
Sobel, D. M., Li, J., & Corriveau, K. H. (2007). ‘‘They danced around in my head and I learned them’’: Children’s developing
conceptions of learning. Journal of Cognition and Development, 8, 345–369.
Strauss, S., & Shilony, T. (1994). Teachers’ models of children’s minds and learning. In L. A. Hirshfield & S. A. Gelman (Eds.),
Mapping the mind: Domain specificity in cognition and culture (pp. 455–473). New York: Cambridge University Press.
Strauss, S., Ziv, M., & Stein, A. (2002). Teaching as a natural cognition and its relations to preschoolers’ developing theory of
mind. Cognitive Development, 17, 1473–1487.
Tang, C. M., & Bartsch, K. (2012). Young children’s recognition of how and when knowledge was acquired. Journal of Cognition
and Development, 13, 372–394.
Tang, C. M., Bartsch, K., & Nunez, N. (2007). Young children’s reports of when learning occurred. Journal of Experimental Child
Psychology, 97, 149–164.
Taylor, M., Esbensen, B. M., & Bennett, R. T. (1994). Children’s understanding of knowledge acquisition: The tendency for
children to report that they have always known what they have just learned. Child Development, 65, 1581–1604.
Tenenbaum, J. B., Kemp, C., Griffiths, T. L., & Goodman, N. D. (2011). How to grow a mind: Statistics, structure, and abstraction.
Science, 331, 1279–1285.
Wellman, H. M., Cross, D., & Watson, J. (2001). Meta-analysis of theory-of-mind development: The truth about false belief. Child
Development, 72, 655–684.
Wood, D., Wood, H., Ainsworth, S., & O’Malley, C. (1995). On becoming a tutor: Toward an ontogenetic model. Cognition and
Instruction, 13, 565–581.
Ziv, M., & Frye, D. (2004). Children’s understanding of teaching: The role of knowledge and belief. Cognitive Development, 19,
457–477.
Ziv, M., Solomon, A., & Frye, D. (2008). Young children’s recognition of the intentionality of teaching. Child Development, 79,
1237–1256.