Theory of Mind Development in Chinese Children: A Meta-Analysis of
False-Belief Understanding Across Cultures and Languages
David Liu, Henry M. Wellman, and Twila Tardif
University of Michigan
Mark A. Sabbagh
Queen’s University at Kingston
Theory of mind is claimed to develop universally among humans across cultures with vastly different
folk psychologies. However, in the attempt to test and confirm a claim of universality, individual studies
have been limited by small sample sizes, sample specificities, and an overwhelming focus on AngloEuropean
children. The current meta-analysis of children’s false-belief performance provides the most
comprehensive examination to date of theory-of-mind development in a population of non-Western
children speaking non-Indo-European languages (i.e., Mandarin and Cantonese). The meta-analysis
consisted of 196 Chinese conditions (127 from mainland China and 69 from Hong Kong), representing
responses from more than 3,000 children, compared with 155 similar North American conditions (83
conditions from the United States and 72 conditions from Canada). The findings show parallel developmental
trajectories of false-belief understanding for children in China and North America coupled with
significant differences in the timing of development across communities—children’s false-belief performance
varied across different locales by as much as 2 or more years. These data support the
importance of both universal trajectories and specific experiential factors in the development of theory
of mind.
Keywords: theory of mind, false belief, meta-analysis, Chinese, culture
Everyday attribution of mental states—a “theory of mind”—
distinguishes human social cognition from that of other animals,
including nonhuman primates. Human theory of mind is not only
phylogenetically distinctive, it is arguably ontogenetically universal.
Except in cases of neurodevelopmental disorders, such as
autism (Baron-Cohen, 1995), or in cases of severely restricted
linguistic inputs, such as deaf children of nonsigning parents
(Peterson & Siegal, 2000), the claim is that everyone develops a
theory of mind (Scholl & Leslie, 2001; Wellman, 1998). Confirming
and characterizing universality in theory of mind, however,
requires the collection of cross-cultural data, which poses several
challenges with regard to appropriate measures and samples.
The most widely used measures of theory-of-mind development
are false-belief tasks. One version involves an unseen change of
location (Wimmer & Perner, 1983): A child sees Maxi put his
chocolate in a kitchen cupboard and then leave. While Maxi is
outside, his mother moves the chocolate from the cupboard to a
drawer. Maxi returns, and the child is asked, “Where does Maxi
think the chocolate is?” or “Where will Maxi look for his chocolate?”
Children develop from answering these sorts of questions
according to reality to answering according to false belief. This
developmental pattern is consistent across numerous task modifications,
thereby establishing that false-belief tasks measure a critical
conceptual development (Wellman, Cross, & Watson, 2001).
Whether false-belief understanding shows a similar trajectory
and/or timetable across disparate cultures has not been thoroughly
and convincingly addressed to date because the bulk of the studies
have been with children in communities dominated by Western
European cultures, speaking Indo-European languages. Recently, a
large number of studies (both published and unpublished) have
investigated the emergence of false-belief understanding in Chinese
children. Indeed, false-belief research in China now constitutes
the largest sample of non-Western data. This large body of
work provides a unique opportunity to use powerful meta-analytic
procedures to evaluate whether the character of development in
false-belief understanding is similar across divergent cultural and
linguistic communities.
Although many researchers have argued that Western cultures’
conception of mind is not always wholly shared by other cultures
(Lillard, 1998; Shweder & Bourne, 1984), only a few studies have
directly compared false-belief performance between Western and
non-Western communities (e.g., Callaghan et al., 2005; Vinden,
1999; Wellman et al., 2001). These studies came to inconsistent
conclusions about developmental trajectory and timing. Callaghan
et al. (2005) tested a modest number of children (from 13 to 31
children per age for each locale) on a single task across five
David Liu, Henry M. Wellman, and Twila Tardif, Department of Psychology
and Center for Human Growth and Development, University of
Michigan; Mark A. Sabbagh, Department of Psychology, Queen’s University
at Kingston, Kingston, Ontario, Canada.
This research was supported by a National Science Foundation Graduate
Fellowship to David Liu; Grant HD-22149 from the National Institute of
Child Health and Human Development to Henry M. Wellman; Grant
CUHK4116/99H from the Research Grants Council of the Hong Kong
Special Administrative Region to Twila Tardif; and an award from the
Social Sciences and Humanities Research Council of Canada to Mark A.
Sabbagh.
We are extremely grateful to all the researchers who shared their
unpublished data with us.
Correspondence concerning this article should be addressed to David
Liu, who is now at the Department of Psychology, University of California,
San Diego, 9500 Gilman Drive, #0109, La Jolla, CA 92093-0109. E-mail:
davidliu@ucsd.edu
Developmental Psychology Copyright 2008 by the American Psychological Association
2008, Vol. 44, No. 2, 523–531 0012-1649/08/$12.00 DOI: 10.1037/0012-1649.44.2.523
523
countries (Canada, India, Peru, Samoa, and Thailand). They found
a universal trajectory from below- to above-chance performance
and, although they did not conduct any statistical analyses directly
comparing data between locales, concluded that there is a tightly
synchronous timetable for the onset of false-belief competence.
Wellman et al. (2001) conducted a meta-analysis comparing falsebelief
performance across age, various tasks, and countries. They
argued that the trajectory from below- to above-chance performance
is universal but (in contrast to Callaghan et al.) also argued
that the timing for this development varies across countries. These
prior findings are intriguing, but significant questions remain. In
particular, all prior analyses involved small samples from outside
Western European cultures, which did not allow for precise estimation
of developmental timetables across cultures. The current
meta-analysis encompasses a large number of non-Western children
to better compare the acquisition timetables of false-belief
understanding.
Chinese children provide an excellent non-Western comparison
because folk psychologies, societal expectations, and parental
practices differ significantly between Chinese and Western cultures
(Nisbett, Peng, Choi, & Norenzayan, 2001; Wang, 2004), all
of which could potentially affect children’s developing understanding
of mental states. Moreover, there are cultural differences
that are directly relevant to factors that are known to influence
theory-of-mind development. For one, exposure to mental state
verbs, such as think, know, and want, influences theory-of-mind
development among Western children (Bartsch & Wellman, 1995;
Dunn, Brown, Slomkowski, Tesla, & Youngblade, 1991; Ruffman,
Slade, & Crowe, 2002). However, verb meanings and acquisition
of verbs differ for Chinese versus English (Tardif, 2006). Whereas
English terms for talking about belief, such as think and believe,
are neutral with respect to the belief’s truthfulness (Moore, Bryant,
& Furrow, 1989), Chinese languages contain belief terms that
specifically mark someone’s belief as false. For Mandarinspeaking
adults, the term yi3wei2 denotes false belief (and the term
dang1 can also have connotations of false belief), whereas xiang3
is neutral (Lee, Olson, & Torrance, 1999). Similarly, in Cantonese,
the term ji5wai4 denotes false belief, whereas soeng2 and nam5
are neutral (Tardif, Wellman, & Cheung, 2004). Second, research
has shown that within Western samples, executive functioning
correlates with performance on false-belief tasks (Carlson &
Moses, 2001). Impulse control is especially valued by Chinese
parents and appears early in Chinese children (X. Chen et al.,
1998), and Chinese children on the mainland (Sabbagh, Xu, Carlson,
Moses, & Lee, 2006) and in Hong Kong (Tardif, So, &
Kaciroti, 2007) outperform American children in executive func-
tioning.
For our meta-analysis, we assembled published and unpublished
studies encompassing almost 200 conditions (typically 16–24 children
per condition) from mainland China (Mandarin speaking) and
Hong Kong (Cantonese speaking). For comparison, we included
available studies from North America (the United States and
Canada), which constitutes the largest sample of Western data
available. Note that we are not specifically predicting earlier or
different development for one group or another. More simply, we
argue for the value of comprehensive data from non-Western
children and outline factors that make Chinese children an important
comparison case.
Method
Unit of Analysis
Conditions within studies, rather than individual participants or
entire studies, make up the unit of analysis in a meta-analysis
(Glass, McGraw, & Smith, 1981). Our dependent variable was the
proportion of false-belief questions answered correctly in each
condition. A meta-analysis of such data (readily available in falsebelief
studies) circumvents the use of effect sizes as the dependent
variable, circumventing problems with the appropriateness, assumptions,
and robustness of various techniques of pooling different
indirect measures of effect size.
Studies
We first searched relevant databases, journals, and abstracts of
recent conferences for studies on theory of mind, false belief,
belief reasoning, understanding mental states, and folk or naı¨ve
psychology in China. Because a potential weakness of any metaanalysis
is publication bias against certain, usually nonsignificant,
results, we contacted Chinese researchers with as yet unpublished
data. We ended our search for relevant studies in April 2004.
Of the studies found, we included only those with false-belief
conditions with typically developing Chinese children (not autistic
or other delayed samples). We did not include conditions with
Chinese children outside of China (e.g., in Taiwan or in Europe).
For the North American data, we used the conditions from Wellman
et al.’s (2001) meta-analysis that were from the United States
and Canada. We selected only conditions that matched the Chinese
data in task characteristics. For example, none of the Chinese
conditions involved deception, so we did not include any North
American conditions that involved deception. As shown in Table
1, we included 196 separate Chinese conditions (127 from mainland
China and 69 from Hong Kong) and 155 separate North
American conditions (83 from the United States and 72 from
Canada).
Coding
Each condition was coded for the proportion of false-belief
judgments answered correctly and for various features making up
the independent variables. Some variables used by Wellman et al.
(2001) were not included because they were unavailable in the
Chinese data (e.g., deception); some levels of a few variables were
truncated because there were no or few conditions with those
levels. The resulting coding scheme was
1. Age: Mean age (in months) of children in a condition.
2. Locale of participants: Mainland China, Hong Kong, the
United States, or Canada.
3. Type of false-belief task: Change of location, unexpected
contents, or deceptive identity (use of appearance–reality
stimuli in a false-belief task).
4. Nature of the protagonist: A puppet or doll, a pictured
character, or a real person (present or absent). Protagonists
as real persons were also coded as either the self or
another person.
524 LIU, WELLMAN, TARDIF, AND SABBAGH
5. Nature of the target object: A real or representative object
or a pictured object.
6. Salience of the protagonist’s mental state: The mental
state had to be inferred from the protagonist’s simple
absence during the key transformation, or the mental
state was demonstrated initially on the children themselves
(e.g., children initially discovered that the candy
box contained pencils).
7. Temporal marker: The false-belief question explicitly
marked the time frame (e.g., “When Maxi returns for his
chocolate, where will he first look for his chocolate?”), or
did not.
8. Type of question: The false-belief question asked children
to judge what the protagonist will think or know
(excluding conditions using think-falsely verbs) or to
judge how the protagonist will behave (e.g., “Where will
Maxi look?”).
9. Chinese verb form: The false-belief question used a
think-falsely verb or a more neutral verb.
Results
Here we present the results from analyzing the full set of
assembled data, which includes both published and unpublished
data. However, to preview, all of the effects found to be significant
for the full dataset were also found to be significant when only the
published data were analyzed (93 of the 127 conditions from
mainland China, 45 of the 69 conditions from Hong Kong, 76 of
the 83 conditions from the United States, and 62 of the 72 conditions
from Canada came from studies that have been published in
peer-reviewed journals). Likewise, all of the effects found to be
nonsignificant were replicated in both analyses.
Chinese Studies
We first considered the Chinese studies to characterize the
regularities in that data necessary for careful comparisons across
locales and languages. We began with the 152 Chinese conditions
that asked children to judge someone else’s false belief. As shown
in Figure 1A, Chinese children’s false-belief performance improves
dramatically with age. Because all the conditions used tasks
with two possible responses, chance equals 50%. Figure 1B shows
the same data with the dependent variable, proportion correct,
transformed via a logit transformation to allow for examination of
the data via linear regression. Chinese children’s answers develop
from being incorrect (27% correct around the age of 39 months) to
being increasingly correct.
The Chinese data came from numerous studies but were often
conducted by a handful of collaborative lab groups. Lab groups
that contributed 20 or more conditions are identified by the name
of the primary collaborator in Table 1. We included lab group as
a blocking factor (covariate) entered first in our regression analyses
to minimize the influence of data from any single lab. The lab
group factor was significant with the 152 Chinese conditions that
asked children to judge someone else’s false belief, F(3, 148) ϭ
18.24, p Ͻ .001. However, all of the effects found to be significant
with lab group as a blocking factor were also found to be significant
without the blocking factor, except for temporal marker.
The first row of Table 2 summarizes the effect of age alone on
performance (blocked by lab group); age accounted for 36% of the
variance in children’s performance. The far right column reports
the measure of effect size, computed as an odds ratio. The odds of
being correct increase 4.32 times for every 1-year increase in age.
We next examined the effects of the independent variables on
this age trend. First consider possible patterns of results: (a) Levels
of an additional variable could have no effect on developmental
trajectory or timing; (b) an additional variable could be significant
Table 1
Listing of the Studies and Conditions in the Meta-Analysis
Location and study Year
Total conditions
included in
meta-analysis Lab group
Mainland China
M. J. Chen & Lin 1994 4 Other
Deng 2001 32 Deng
Goetz 2003 8 Other
Lee et al. 1999 36 Lee
Sabbagh et al. 2006 9 Other
Sang et al. 2004 12 Deng
Tardif et al. 2000 12 Tardif
Tardif et al. 2001 12 Tardif
Volling et al. 1999 2 Other
Hong Kong
Tardif et al. 2001 12 Tardif
Tardif et al. 2004 21 Tardif
Tardif et al. 2007 24 Tardif
Tardif & Ng 2001 12 Tardif
Total conditions from China 196
United States
Bartsch & Wellman 1989 1
Carlson et al. 1998 2
Dalke 1995 14
Davis 1997 3
Frye et al. 1995 12
Hickling et al. 1997 2
Lillard & Flavell 1992 3
Moses 1993 2
Robinson & Mitchell 1995 8
Sheffield et al. 1993 2
Slaughter & Gopnik 1996 4
Sullivan & Winner 1991 12
Sullivan & Winner 1993 10
Taylor & Carlson 1997 4
Winner & Sullivan 1993 2
Zaitchik 1990 2
Canada
Astington et al. 1989 8
Carpendale & Chandler 1996 13
Chandler & Hala 1994 2
Gopnik & Astington 1988 42
Hala et al. 1991 3
Moore et al. 1990 2
Ruffman et al. 1993 2
Total conditions from North
America
155
Note. The Chinese data came from numerous studies but were often
conducted by a handful of collaborative lab groups. Lab groups that
contributed 20 or more conditions are identified by the name of the primary
collaborator.
525THEORY OF MIND IN CHINESE CHILDREN
only as a main effect without an interaction with age—the timing
differs, but the trajectories remain equivalent; (c) the variable
could interact significantly with age—levels of that variable
change not simply the timing but also the trajectory of development.
To assess these possibilities, we used linear regression to
check for all two-way interactions with age. For variables that did
not interact with age, the interaction term was dropped from the
regression and main effects were tested.
As summarized in Table 2, none of the variables we tested
interacted significantly with age. In addition, six of the eight task
variables had no significant influence on performance. For example,
performance is the same for any given age whether the
protagonist is presented as a real person, a puppet or doll, or a
pictured storybook character, and performance is the same across
the three types of false-belief tasks: change-of-location tasks,
unexpected-contents tasks, and deceptive-identity tasks. To examine
possible self versus other differences, we compared conditions
that asked children about their own beliefs to conditions that asked
them to judge others’ beliefs. All self conditions in China used
unexpected-contents or deceptive-identity tasks. Therefore, we
included only conditions with those tasks in comparing judgments
for self (44 conditions) versus other (64 conditions). As reported in
Table 2, Chinese children’s false-belief judgments of themselves
and of others follow identical developmental trajectories. For all
these nonsignificant variables, the Chinese data replicate those
from Western results (Wellman et al., 2001).
Two task variables, temporal marker and Chinese verb form,
were significant as main effects, as listed in Table 2. Use of an
explicit temporal marker hindered children’s performance. Because
this is counter to the effect typically found when such
modifications are provided in English (Wellman et al., 2001), it
serves as a reminder that different languages work differently; a
change that helps clarify linguistic intent in one language may add
difficulty in another language. Nevertheless, as discussed above,
temporal marker was the only significant effect that was not
significant without lab group as a blocking factor, and as we show
in the combined regression model analyses below, the temporal
marker effect is not robust.
More important, Chinese verb form yielded a main effect.
Chinese verb form captures whether the false-belief question used
a think-falsely verb or a more neutral verb. This finding indicates
that the use of think-falsely verbs in the experimental protocol
enhances performance for children of all ages. As demonstrated
with the absence of an interaction, children proceed from below- to
above-chance performance with increasing age, even with thinkfalsely
verbs in the false-belief tasks.
Mainland China Versus Hong Kong
Chinese conditions in our dataset encompass two different Chinese
languages—Cantonese and Mandarin—and they come from
different geographical locales—mainland China (primarily from
Beijing but also from Shanghai, Hangzhou, and Wenzhou) and
Hong Kong. As shown in Table 3, there was a significant main
effect of mainland China versus Hong Kong but no interaction of
locale with age. Thus, the trajectory of development is the same in
both locales, with timing earlier for mainland Chinese children.
Because only one lab group (Tardif’s) provided conditions from
Hong Kong, we also compared mainland China with Hong Kong
using only conditions from this particular lab group. A main effect
of mainland China versus Hong Kong, F(1, 66) ϭ 5.91, p Ͻ .05,
remains even with this more controlled analysis.
We constructed a combined regression model with the significant
variables (age, Chinese verb form, temporal marker, and
mainland China versus Hong Kong) to best predict false-belief
performance and to address the possibility that a variable’s significant
effect may disappear if other significant variables are controlled
for as well. In this first multivariate combined model (R2
ϭ
.679), temporal marker was no longer significant. Therefore, in the
final combined model, we excluded temporal marker; together,
.0
.1
.2
.3
.4
.5
.6
.7
.8
.9
1.0
30 40 50 60 70 80 90
Age (Months)
ProportionCorrect
-5
-4
-3
-2
-1
0
1
2
3
4
5
30 40 50 60 70 80 90
Age (Months)
Correct(Logit)
A
B
Figure 1. Scatterplots of Chinese conditions with increasing age showing
best-fit line: (A) raw scatterplot of mean proportion correct data with log
fit; (B) scatterplot of logit transformed data with linear fit.
526 LIU, WELLMAN, TARDIF, AND SABBAGH
age, Chinese verb form, and mainland China versus Hong Kong
accounted for 67.5% of the variance in Chinese children’s falsebelief
performance.
Cross-Locale Comparisons
The regularities found in the Chinese data allow us to directly
and comprehensively compare performance across cultures and
languages. Given a main effect of locale within the Chinese
conditions, we compared Chinese and North American conditions
in terms of four locales: mainland China, Hong Kong, the United
States, and Canada. Just as with the Chinese data, the vast majority
of North American conditions (61 from the US and 45 from
Canada) asked children to judge someone else’s false belief. As
listed in Table 3, there was a significant main effect of the four
locales and no interaction with age. Looking at the effect sizes and
Figure 2, it is clear that the developmental trajectories are all very
similar, because locale does not interact with age. However, there
was a significant pattern of timing differences across the four
locales. Specifically, the timing of development for children in
mainland China and children in the United States is very similar
(an odds ratio effect size close to 1.0 translates into even odds of
mainland Chinese and U.S. children being correct at any age).
However, Canadian children develop earlier than mainland Chinese
and U.S. children, whereas Hong Kong Chinese children
develop much later. The differences in timing are clearly shown in
Figure 2. Whereas Canadian children start performing above
chance around 38 months of age, Hong Kong children do so
around 64 months—a difference of more than 2 years.
We demonstrated earlier that Chinese children perform better on
conditions with think-falsely verbs; would locales still show differences
in timing if only Chinese think-falsely conditions were
compared with the North American conditions? To address this,
we compared Chinese conditions with think-falsely verbs (58 from
mainland China and 30 from Hong Kong) with conditions from the
United States and Canada. As listed in Table 3, even when only
Chinese think-falsely conditions—the “best” Chinese data—were
included, the main effect of locale remained significant, and there
was no significant interaction with age. Furthermore, the effect
sizes remained very similar.
Discussion
Our meta-analysis provides the most comprehensive test to date
of theory-of-mind development in non-Western cultures with children
speaking non-Indo-European languages. The findings add
substantially to claims about the universal, early development of
theory of mind. Specifically, we demonstrate parallel developmental
trajectories but substantially different timetables across numerous
locales. Because we included numerous unpublished as well as
published studies, our findings are unlikely to represent publication
bias tilted toward a certain pattern of results (i.e., expected
Table 2
Summary of Results From Meta-Analysis With Chinese Data
Variable Main effect Interaction with age Effect sizea
Age: F(1, 147) ϭ 145.55, p Ͻ .001 4.32 for 1 year
Nonsignificant
Nature of the protagonist F(2, 145) ϭ 1.07, p Ͼ .30 F(2, 143) ϭ 1.35, p Ͼ .20
Nature of the target object F(1,146) ϭ .50, p Ͼ .40 F(1, 145) ϭ 2.75, p Ͼ .10
Salience F(1, 146) ϭ .59, p Ͼ .40 F(1, 145) ϭ .02, p Ͼ .50
Type of question F(1, 58) ϭ .80, p Ͼ .30 F(1, 57) ϭ .42, p Ͼ .50
Type of task F(2, 145) ϭ 1.26, p Ͼ .20 F(2, 143) ϭ 1.51, p Ͼ .20
Self vs. other F(1, 102) ϭ .95, p Ͼ .30 F(1, 101) ϭ .28, p Ͼ .50
Main effects
Temporal marker F(1, 146) ϭ 5.85, p Ͻ .05 F(1, 145) ϭ .26, p Ͼ .50 1.76
Chinese verb form F(1, 146) ϭ 13.20, p Ͻ .001 F(1, 145) ϭ 1.68, p Ͼ .10 1.97
a
Effect sizes are presented only for significant variables; effect sizes were computed as odds ratios.
Table 3
Summary of Results From Meta-Analysis With Chinese and North American Data
Variable Main effect Interaction with age Effect sizea
Hong Kong vs. mainland China F(1, 146) ϭ 5.26, p Ͻ .05 F(1, 145) ϭ 1.50, p Ͼ .20 2.02
Locales across North America and China F(3, 253) ϭ 37.64, p Ͻ .001 F(2, 251) ϭ .58, p Ͼ .50
U.S. vs. mainland China 1.27
Hong Kong vs. U.S. 5.38
Hong Kong vs. Canada 13.21
Locales across North America and China
(Chinese conditions with think-falsely verbs)
F(3, 189) ϭ 22.49, p Ͻ .001 F(2, 187) ϭ .18, p Ͼ .50
U.S. vs. mainland China 1.55
Hong Kong vs. U.S. 3.52
Hong Kong vs. Canada 9.68
a
Effect sizes are presented only for significant variables; effect sizes were computed as odds ratios.
527THEORY OF MIND IN CHINESE CHILDREN
similarity to prior research in Western European cultures). In
addition, because the same pattern of findings was also found
when only the data from published studies were analyzed, our
findings are not skewed by the inclusion of unpublished data that
have not undergone peer review.
First consider developmental trajectory. Chinese children develop
from below- to above-chance performance during early
childhood, and this parallels the trajectory of North American
children (because there was not a significant interaction effect
between locale and age in our results). Certainly, the developmental
trajectory of Chinese versus North American children’s falsebelief
understanding could have been nonparallel (because of
differences in folk psychologies, differences in executive function
development, the presence of think-falsely verbs in Chinese languages,
and so on, as outlined in the Introduction).
None of the task variables we tested interacted significantly with
age; variables showed either nonsignificance or simple main effects.
With the nonsignificant findings, our data demonstrate that
Chinese children’s false-belief judgments are robust across many
potentially relevant variables. Some have questioned whether the
false-belief task always has validity in a non-Western context (e.g.,
Lillard, 1998). Here we show that not only is the trajectory the
same in China as in Western cultures, but also it is important that
similar task manipulations have the same non-effects (Wellman et
al., 2001). This strengthens the case that false-belief tasks are
measuring the same construct across cultures. Note also that the
presence of think-falsely verbs in Chinese languages (and in many
Chinese false-belief tasks) does not change the overall developmental
trajectory for false-belief understanding; even with questions
that use think-falsely verbs, children still develop from
below- to above-chance performance.
The lack of an interaction for Chinese verb form with age
undermines early competence accounts. Early competence accounts
of theory of mind (Chandler, Fritz, & Hala, 1989; Scholl &
Leslie, 2001) have suggested that changes in false-belief performance
during the preschool years reflect children’s increasing
ability to deal with verbal ambiguities of typical false-belief tasks
or other domain-general cognitive demands of the tasks, such as
executive functioning, rather than development in mental-state
understanding. As such, if verbal ambiguities or domain-general
cognitive demands were removed from false-belief tasks, falsebelief
performance would not improve with age during the preschool
years, because 2- or 3-year-olds would already have the
requisite mental-state understanding. Thus, the expected pattern in
a meta-analysis for factors with different levels of verbal ambiguities
would be an interaction with age on false-belief performance
(Wellman et al., 2001). That is, for example, conditions that use
the think-falsely verbs (which significantly lower verbal ambiguities)
should show no (or considerably less) change with age in
contrast to typical conditions. However, our results show that
Chinese verb form did not interact with age. Furthermore, even in
especially facilitative linguistic environments (i.e., Chinese children
growing up with think-falsely verbs), understanding false
belief poses genuine conceptual difficulties for young children and
performance changes with age. It also appears that Chinese children
have earlier competence at executive function tasks than
North American children (Sabbagh et al., 2006; Tardif et al.,
2007), although this does not translate into superior false-belief
performance. Therefore, our pattern of findings undermines a
variety of early competence accounts that contend that executive
functioning limitations or linguistically demanding tasks have
masked otherwise apparent early false-belief competence (see Sabbagh
et al., 2006).
Next, consider developmental timing. Several variables—most
clearly verb form and locale or language—visibly and significantly
influenced the timing of false-belief understanding. Using thinkfalsely
verbs in false-belief tasks improves the performance of
Chinese children relative to using more neutral verbs. Two individual
studies with Mandarin- (Lee et al., 1999) and Cantonesespeaking
(Tardif et al., 2004) children also found an effect of using
think-falsely verbs in the question. However, those studies did not
compare Chinese children’s performance to that of children in
Western societies, as we were able to do in our analyses. As is
clear in Figure 2, the presence of think-falsely verbs in Chinese
languages does not enhance Chinese children’s overall false-belief
understanding relative to North American children.
It is most intriguing that the timing of development varies
significantly across different locales, as shown in Figure 2. Such a
large difference in timing between specific locales counters Callaghan
et al.’s (2005) suggestion that the timing of false-belief
understanding might be tightly synchronous across very different
cultures. Instead, we find different developmental timetables exhibiting
more than 2 years’ difference across communities.
Although our results demonstrate significant differences in developmental
timing across locales, the specific differences in timing
(including the difference between mainland China and Hong
Kong) admit no straightforward interpretations. Although there are
numerous differences between the four populations that could lead
to differences in false-belief performance, these differences do not
easily account for our findings. For example, mainland Chinese
children are almost certainly singletons, whereas the Hong Kong
children are more likely to have siblings. Given research showing
-5
-4
-3
-2
-1
0
1
2
3
4
5
30 40 50 60 70 80 90 100
Age (Months)
Correct(Logit)
United States
Hong Kong
Mainland China
Canada
Figure 2. Developmental trajectories for the four communities—
mainland China, Hong Kong, the United States, and Canada—of logit
transformed data with increasing age.
528 LIU, WELLMAN, TARDIF, AND SABBAGH
that having older siblings is associated with enhanced false-belief
understanding (Perner, Ruffman, & Leekam, 1994), one might
expect Hong Kong’s timetable to be closer to North American
trajectories than to mainland China’s timetable, but we found the
opposite pattern of results. Likewise, because Hong Kong is more
“Westernized” than mainland China (Bond & Cheung, 1983), one
might expect a pattern of results opposite from the observed
results. Also, children in Hong Kong are more likely to be bilingual
(Tardif, Fletcher, Marchman, & Liang, 2006), a factor associated
with enhanced false-belief understanding (Goetz, 2003), and
yet their developmental timetable is the furthest behind.
It is also important to consider socioeconomic status (SES). The
studies in our dataset contain little SES data; almost all the studies
provide no such data. More importantly, we consider SES to be too
broad a factor and unlikely to provide a simple explanation for the
pattern of observed differences. It is unclear what it would mean to
compare such a broad concept as SES in the different cultural
contexts of the North American and the Chinese populations. For
example, does having college-educated parents or being middle
class mean the same thing in the cultural contexts of North America,
mainland China, and Hong Kong? When we compare mainland
China and Hong Kong, very little is known about the effect of
SES on cognitive and language development. However, the little
data that exists would suggest that SES has very little influence on
comparisons of cognitive and language development between
mainland China and Hong Kong. For example, in a norming study
of the MacArthur-Bates Communicative Development Inventories
with normative samples of 3,270 children from both Beijing and
Hong Kong, the average educational levels of both parents (mothers
and fathers) and both sets of grandparents (maternal and
paternal) were all significantly lower for the Hong Kong sample
than for the Beijing sample (Tardif et al., 2006). Nevertheless, the
effect of Hong Kong versus Beijing on vocabulary score yielded
an R2
of only .03, and the effect of SES yielded an even smaller R2
of .003. That is, there were only small differences between Hong
Kong and Beijing in children’s verbal ability, and SES played very
little role in that difference. These findings suggest that the developmental
timing differences observed between Hong Kong and
mainland China (and thus between Hong Kong and North America)
for false-belief understanding were not due to (or only very
minimally influenced by) SES or, for that matter, verbal ability.
Still, without direct data on SES or other possible differences
between the populations between the locales, we cannot completely
rule out the possibility that these variables play a larger
role.
Generally, although one can speculate and point to numerous
comparisons across the populations that could account for differences,
it is noteworthy that no single sociocultural or linguistic
factor (especially broad factors such as SES, East–West culture, or
bilingualism) can provide a straightforward account of the pattern
of differences observed in this meta-analysis. Instead, it seems
more likely that a coalescence of factors as yet unknown will prove
responsible. This counts as an important conclusion from the
current research; it directs future research to examine specific
variables (e.g., particular parental practices or linguistic features)
to address the question of how specific adult folk psychologies,
enculturation, and linguistic factors jointly shape the development
of theory of mind.
Conclusion
The current meta-analysis supports three main points. First, the
results indicate that theory of mind universally develops. Specifically,
Chinese and North American children all develop from
below- to above-chance performance, and this trajectory does not
disappear with the inclusion of any of the task variables we
examined. Second, although it reflects parallel developmental trajectories,
theory of mind understanding appears on substantially
different timetables across numerous cultures and languages. Summing
across these two points, the developmental course of theory
of mind necessarily includes influences of both universal trajectories
and specific experiential factors. Third, the observed timetable
differences are unlikely to be explained by straightforward, simplistic
accounts, but are the product of (and could be used to
reveal) the multiple sociocultural and linguistic factors that jointly
shape theory of mind development.
References
References marked with an asterisk indicate studies included in the
meta-analysis.
*Astington, J. W., Gopnik, A., & O’Neill, D. (1989). Young children’s
understanding of unfulfilled desire and false belief. Unpublished manuscript,
Ontario Institute for Studies in Education, Toronto, Ontario,
Canada.
Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory of
mind. Cambridge, MA: MIT Press.
*Bartsch, K., & Wellman, H. M. (1989). Young children’s attribution of
action to beliefs and desires. Child Development, 60, 946–964.
Bartsch, K., & Wellman, H. M. (1995). Children talk about the mind. New
York: Oxford University Press.
Bond, M. H., & Cheung, T. S. (1983). College students’ spontaneous
self-concept: The effect of culture among respondents in Hong Kong,
Japan, and the United States. Journal of Cross-Cultural Psychology, 14,
153–171.
Callaghan, T., Rochat, P., Lillard, A., Claux, M. L., Odden, H., Itakura, S.,
et al. (2005). Synchrony in the onset of mental-state reasoning: Evidence
from five cultures. Psychological Science, 16, 378–384.
Carlson, S. M., & Moses, L. J. (2001). Individual differences in inhibitory
control and children’s theory of mind. Child Development, 72, 1032–
1053.
*Carlson, S. M., Moses, L. J., & Hix, H. R. (1998). The role of inhibitory
processes in young children’s difficulties with deception and false belief.
Child Development, 69, 672–691.
*Carpendale, J. L., & Chandler, M. J. (1996). On the distinction between
false belief understanding and subscribing to an interpretive theory of
mind. Child Development, 67, 1686–1706.
Chandler, M., Fritz, A. S., & Hala, S. (1989). Small scale deceit: Deception
as a marker of 2-, 3-, and 4-year-olds’ early theories of mind. Child
Development, 60, 1263–1277.
*Chandler, M., & Hala, S. (1994). The role of personal involvement in the
assessment of early false belief skills. In C. Lewis & P. Mitchell (Eds.),
Children’s early understanding of mind (pp. 403–425). Hillsdale, NJ:
Erlbaum.
*Chen, M. J., & Lin, Z. X. (1994). Chinese preschoolers’ difficulty with
theory-of-mind tests. Bulletin of the Hong Kong Psychological Society,
32, 34–46.
Chen, X., Hastings, P. D., Rubin, K. H., Chen, H., Cen, G., & Stewart, S. L.
(1998). Child-rearing attitudes and behavioral inhibition in Chinese and
Canadian toddlers: A cross-cultural study. Developmental Psychology,
34, 677–686.
*Dalke, D. E. (1995). Explaining young children’s difficulty on the false
529THEORY OF MIND IN CHINESE CHILDREN
belief task: Representational deficits of context-sensitive knowledge?
British Journal of Developmental Psychology, 13, 209–222.
*Davis, D. L. (1997, April). Children’s understanding of the role of
knowledge and thinking in pretense. Paper presented at the biennial
meeting of the Society for Research in Child Development, Washington,
DC.
*Deng, C. (2001). A study on development and representational mechanism
of young children’s theory of mind. Unpublished doctoral dissertation,
East China Normal University, Shanghai, China.
Dunn, J., Brown, J., Slomkowski, C., Tesla, C., & Youngblade, L. (1991).
Young children’s understanding of other people’s feelings and beliefs:
Individual differences and their antecedents. Child Development, 62,
1352–1366.
*Frye, D., Zelazo, P. D., & Palfai, T. (1995). Theory of mind and rulebased
reasoning. Cognitive Development, 10, 483–527.
Glass, G. V., McGraw, B., & Smith, M. L. (1981). Meta-analysis in social
research. Beverly Hills, CA: Sage.
*Goetz, P. J. (2003). The effects of bilingualism on theory of mind
development. Bilingualism: Language and Cognition, 6, 1–15.
*Gopnik, A., & Astington, J. W. (1988). Children’s understanding of
representational change and its relation to the understanding of false
belief and the appearance–reality distinction. Child Development, 59,
26–37.
*Hala, S., Chandler, M., & Fritz, A. S. (1991). Fledgling theories of mind:
Deception as a marker of 3-year-olds’ understanding of false belief.
Child Development, 62, 83–97.
*Hickling, A. K., Wellman, H. M., & Gottfried, G. (1997). Preschoolers’
understanding of others’ mental attitudes toward pretend happenings.
British Journal of Developmental Psychology, 15, 339–354.
*Lee, K., Olson, D. R., & Torrance, N. (1999). Chinese children’s understanding
of false beliefs: The role of language. Journal of Child Language,
26, 1–21.
Lillard, A. (1998). Ethnopsychologies: Cultural variations in theories of
mind. Psychological Bulletin, 123, 3–32.
*Lillard, A. S., & Flavell, J. H. (1992). Young children’s understanding of
different mental states. Developmental Psychology, 28, 626–634.
Moore, C., Bryant, D., & Furrow, D. (1989). Mental terms and the
development of certainty. Child Development, 60, 167–171.
*Moore, C., Pure, K., & Furrow, P. (1990). Children’s understanding of the
modal expression of certainty and uncertainty and its relation to the
development of a representational theory of mind. Child Development,
61, 722–730.
*Moses, L. J. (1993). Young children’s understanding of belief constraints
on intention. Cognitive Development, 8, 1–25.
Nisbett, R. E., Peng, K., Choi, I., & Norenzayan, A. (2001). Culture and
systems of thought: Holistic versus analytic cognition. Psychological
Review, 108, 291–310.
Perner, J., Ruffman, T., & Leekam, S. R. (1994). Theory of mind is
contagious: You catch it from your sibs. Child Development, 65, 1228–
1238.
Peterson, C. C., & Siegal, M. (2000). Insights into theory of mind from
deafness and autism. Mind & Language, 15, 123–145.
*Robinson, E. J., & Mitchell, P. (1995). Masking children’s early understanding
of the representational mind: Backwards explanation versus
prediction. Child Development, 66, 1022–1039.
*Ruffman, T., Olson, D. R., Ash, T., & Keenan, T. (1993). The ABCs of
deception: Do young children understand deception in the same way as
adults? Developmental Psychology, 29, 74–87.
Ruffman, T., Slade, L., & Crowe, E. (2002). The relation between children’s
and mothers’ mental state language and theory-of-mind understanding.
Child Development, 73, 734–751.
*Sabbagh, M. A., Xu, F., Carlson, S. M., Moses, L. J., & Lee, K. (2006).
The development of executive functioning and theory-of-mind: A comparison
of Chinese and U.S. preschoolers. Psychological Science, 17,
74–81.
*Sang, B., Ma, L., & Deng, C. (2004). An experimental study between the
using of mental terms and the development of theory of mind. Psychological
Science (China), 27, 584–589.
Scholl, B. J., & Leslie, A. M. (2001). Minds, modules, and meta-analysis.
Child Development, 72, 696–701.
*Sheffield, E. G., Sosa, B. B., & Hudson, J. A. (1993, March). Understanding
metarepresentation: 2- and 3-year-olds’ comprehension of
false belief. Paper presented at the biennial meeting of the Society for
Research in Child Development, New Orleans, LA.
Shweder, R. A., & Bourne, E. J. (1984). Does the concept of the person
vary cross-culturally? In R. A. Shweder & R. A. LeVine (Eds.), Culture
theory: Essays on mind, self, and emotion (pp. 158–199). Cambridge,
England: Cambridge University Press.
*Slaughter, V., & Gopnik, A. (1996). Conceptual coherence in the child’s
theory of mind: Training children to understanding belief. Child Development,
67, 2967–2988.
*Sullivan, K., & Winner, E. (1991). When 3-year-olds understand ignorance,
false belief and representational change. British Journal of Developmental
Psychology, 9, 159–171.
*Sullivan, K., & Winner, E. (1993). Three-year-olds’ understanding of
mental states: The influence of trickery. Journal of Experimental Child
Psychology, 56, 135–148.
Tardif, T. (2006). But are they really verbs? Chinese words for action. In
K. Hirsh-Pasek & R. M. Golinkoff (Eds.), Action meets word: How
children learn verbs (pp. 477–498). New York: Oxford University
Press.
Tardif, T., Fletcher, P., Marchman, V., & Liang, W. (2006, June).
Comprehension and production of nouns and verbs: Data from the
CDI norming studies in English, Mandarin, and Cantonese. Paper
presented at the International Conference on Infant Studies, Kyoto,
Japan.
*Tardif, T., Fung, K., Wellman, H. M., Liu, D., & Fang, F. (2001, April).
Preschoolers’ understanding of knowing how, knowing that, and false
belief. Poster session presented at the biennial meeting of the Society for
Research in Child Development, Minneapolis, MN.
*Tardif, T., & Ng, M. C. (2001). Emotional understanding and theory of
mind development. Unpublished manuscript, Chinese University of
Hong Kong, China.
*Tardif, T., So, C., & Kaciroti, N. (2007). The development of false belief
and its relation to syntactic complements in Cantonese-speaking preschoolers.
Developmental Psychology, 43, 318–340.
*Tardif, T., & Wellman, H. M. (2000). Acquisition of mental state language
in Mandarin- and Cantonese-speaking children. Developmental
Psychology, 36, 25–43.
*Tardif, T., Wellman, H. M., & Cheung, K. M. (2004). False belief
understanding in Cantonese-speaking children. Journal of Child Language,
31, 779–800.
*Tardif, T., Yang, X. D., & So, C. (2000). False belief and complement
understanding in Mandarin. Unpublished manuscript, Chinese University
of Hong Kong, China.
*Taylor, M., & Carlson, S. M. (1997). The relation between individual
differences in fantasy and theory of mind. Child Development, 68,
436–455.
*Tomasello, M. (1999). The cultural origins of human cognition. Cambridge,
MA: Harvard University Press.
Vinden, P. G. (1999). Children’s understanding of mind and emotion: A
multi-cultural study. Cognition and Emotion, 13, 19–48.
*Volling, B. L., Xiang, Z., & Niu, W. (2001). Children’s false belief
understanding in Chinese and U.S. preschoolers. Unpublished manuscript,
University of Michigan, Ann Arbor.
530 LIU, WELLMAN, TARDIF, AND SABBAGH
Wang, Q. (2004). The emergence of cultural self-construct: Autobiographical
memory and self-description in American and Chinese children.
Developmental Psychology, 40, 3–15.
Wellman, H. M. (1998). Culture, variation and levels of analysis in our folk
psychologies. Psychological Bulletin, 123, 33–36.
Wellman, H. M., Cross, D., & Watson, J. (2001). Meta-analysis of theoryof-mind
development: The truth about false belief. Child Development,
72, 655–684.
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and
constraining function of wrong beliefs in young children’s understanding
of deception. Cognition, 13, 103–128.
*Winner, E., & Sullivan, K. (1993). Deception as a zone of proximal
development for false belief understanding. Unpublished manuscript,
Boston College.
*Zaitchik, D. (1990). When representations conflict with reality: The
preschooler’s problem with false beliefs and “false” photographs. Cognition,
35, 41–68.
Received January 16, 2006
Revision received July 6, 2007
Accepted August 21, 2007 Ⅲ
531THEORY OF MIND IN CHINESE CHILDREN