1 23
Review of Philosophy and
Psychology
ISSN 1878-5158
Rev.Phil.Psych.
DOI 10.1007/s13164-011-0067-
y
On the Long Road to Mentalism in
Children’s Spontaneous False-Belief
Understanding: Are We There Yet?
Jason Low & Bo Wang
1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Science+Business Media B.V.. This e-offprint
is for personal use only and shall not be selfarchived
in electronic repositories. If you
wish to self-archive your work, please use the
accepted author’s version for posting to your
own website or your institution’s repository.
You may further deposit the accepted author’s
version on a funder’s repository at a funder’s
request, provided it is not made publicly
available until 12 months after publication.
On the Long Road to Mentalism in Children’s
Spontaneous False-Belief Understanding:
Are We There Yet?
Jason Low & Bo Wang
# Springer Science+Business Media B.V. 2011
Abstract We review recent anticipatory looking and violation-of-expectancy studies
suggesting that infants and young preschoolers have spontaneous (implicit)
understanding of mind despite their known problems until later in life on elicited
(explicit) tests of false-belief reasoning. Straightforwardly differentiating spontaneous
and elicited expressions of complex mental state understanding in relation to an implicitexplicit
knowledge framework may be challenging; early action predictions may be
based on behavior rules that are complementary to the mentalistic attributions under
consideration. We discuss that the way forward for diagnosing early mentalism is to
analyze whether young candidate mind-readers’ visual orienting cohere across different
belief-formation by belief-use combinations. Adopting this formal cognitive analysis,
we conclude that whilst some studies come tantalizingly close to sign-posting mentalism
in infants and young children’s spontaneous responses, the bulk of evidence for early
mentalism grades into behaviorism.
Human beings can interpret others’ actions, motivations and thoughts in terms of
underlying mental states—this everyday reasoning is termed as ‘theory of mind’
(ToM). Thus, when watching the film ‘The Sixth Sense’, we understand the plot
twist wherein the child psychologist character wants to treat the boy for hallucinating
seeing dead people but—not realizing that ghosts do not always know that they are
dead—falsely believes that he himself is alive. Decades of research, using elicited
response tasks that ask direct verbal questions, concluded that children only start to
understand false-belief from around their 4th birthday onwards (Wellman et al.
2001). Intriguingly, recent research findings suggest that false-belief understanding
may be present much earlier in life—when children are tested on spontaneous
response tasks such as anticipatory looking (AL—Southgate et al. 2007) or
violation-of-expectancy (VOE—Onishi and Baillargeon 2005).
Rev.Phil.Psych.
DOI 10.1007/s13164-011-0067-y
J. Low (*) :B. Wang
School of Psychology, Victoria University of Wellington, PO Box 600, Wellington, New Zealand 6140
e-mail: Jason.Low@vuw.ac.nz
Author's personal copy
In this review, we discuss whether extant evidence on children’s spontaneous
looking responses in false-belief tasks warrants ascription of early mentalism. We
explain that extant research has not yet managed to arbitrate between behavioral and
mental state accounts that explain the same input–output function, and thus suggest
that it is not yet clear whether mentalism can even be signposted in the spontaneous
looking responses displayed by infants and young preschoolers. We then go on to
discuss possible approaches that may do the job in determining when and how early
in life children may be mentalists.
1 Dissociation Between Spontaneous and Elicited False-Belief Responding
Despite the quantitative strengths of Wellman et al.’s (2001) seminal meta-analysis of
research on children’s ToM development, only children’s correct versus incorrect
judgments to direct false-belief questions were treated as the essential dependent
variable. Data on children’s non-judgmental eye gazing to two candidate locations in
the false-belief task were entirely excluded (Moses 2001). Indeed, studies
emphasizing young preschoolers possessing a kernel of non-judgmental knowledge
can have substantial theoretical stake in illuminating how false-belief understanding
develops (e.g., Clements and Perner 1994; Ruffman et al. 2001). Clements and
Perner were the first to introduce a spontaneous response task that incorporated an
AL modality to test preschoolers’ false-belief sensitivity. The task was based on the
classical unexpected transfer false-belief scenario. However, before asking children a
direct false-belief question, participants heard the story narrator wonder aloud: “I
wonder where [the agent] will look for the object”. Clements and Perner used the “I
wonder” prompt to trigger children’s spontaneous visual anticipation of where an
agent with a false-belief about the location of an object will search for the object.
The results showed that 3-year-olds, but not younger, spontaneously looked towards
the correct location where the agent thought the object was, even though they made
wrong explicit verbal judgments of the agent’s future searching behavior. Clements
and Perner suggested that accuracy in 3-year-olds’ AL might reflect an implicit
understanding of false-belief.
Subsequent research replicated and extended Clements and Perner’s (1994)
findings. Ruffman et al. (2001) confirmed that 3-year-olds who gave incorrect verbal
false-belief answers did not even want to bet a modest number of counters on the
location as indicated by their correct eye gaze. Thus, children do not appear to be
conscious of the knowledge conveyed in their AL responses. Low (2010) attempted
to give functional meaning to the implicit-explicit knowledge framework by
speaking to the different patterns of correlations involving spontaneous and elicited
false-belief responding. Studying individual differences amongst 3- to 4-year-olds,
Low found that AL was not correlated with language ability or executive functioning
performance but complex grammatical understanding and conflict control ability
were associated with direct false-belief answering. Furthermore, whilst there was a
robust dissociation between AL and verbal answering, individual differences in
children’s looking time towards the correct belief-based location uniquely correlated
with individual differences in explicit verbal false-belief answering. Hence, there is
some degree of plausibility in the representational redescription of false-belief
J. Low, B. Wang
Author's personal copy
understanding such that implicit awareness is transformed for participation in direct
judgments and, further, the development of fully explicit understanding is partly
advanced by executive functioning and language abilities.
Whilst accurate AL amongst 3-year-olds were securely replicated across Clements
and Perner (1994), Low (2010) and Ruffman et al. (2001), these studies also shared
the same limitation of visual orienting being triggered by a verbalized “I wonder”
prompt. It is possible that early false-belief understanding does not undergo implicitexplicit
representational format changes per se. For example, young children may be
unable to express their latent understanding due to task specific factors (cf. Csibra
and Southgate 2006; Southgate et al. 2007). It is possible that very young
preschoolers might interpret “where” in the verbal prompt as referring to the
location of the hidden object, rather than the location of the actor’s subsequent
actions (however, see Clements 1995, for counter arguments).
2 Spontaneous Non-Verbal False-Belief Performance
Southgate et al. (2007) developed an elegant non-verbal AL paradigm to test
whether 25-month-olds were able to show signs of false-belief understanding. In
their task, children were first familiarized with two events in which a puppet hid a
toy in one of two boxes whilst a female agent looked on. Then the agent opened
one of two windows (which was above the boxes) to retrieve the toy when both
windows were illuminated. The illumination of the windows was paired with a
simultaneous bell sound. In doing so, Southgate et al. ingeniously created the light
and sound pairing as a non-verbal prompt to elicit AL toward one of the windows
later in the test trial. In the test trial, the agent saw the puppet hide the toy in the
left or the right box. A phone then rang behind the agent, leading her to turn
toward the sound. Whilst the agent was facing away, the puppet retrieved the toy
and left with it. Another important modification compared to previous
spontaneous-response tasks was that the target toy was always removed from the
scene in the test trial, avoiding the possibility that knowledge of the object’s
location may mask children’s sensitivity to false-belief. By making the task nonverbal
and removing the target object from the scene, Southgate et al. uncovered
that 2-year-olds correctly looked in anticipation of where an agent with a falsebelief
would search for an object—indicating that toddlers might be able to
attribute false-beliefs to others.
Even earlier sensitivity to false-belief has been claimed by another group of
studies using the VOE paradigm in which looking fixations to unexpected as
compared to expected outcomes are contrasted. In Onishi and Baillargeon’s (2005)
ground-breaking study, they found that 15-month-olds looked much longer when an
agent’s behavior was inconsistent, as opposed to consistent, with her false-belief,
indicating that infants might implicitly attribute to the agent a false-belief about the
location of an object. Onishi and Baillargeon’s findings have been replicated
(Träuble et al. 2010) and even extended to 13-month-olds (Surian et al. 2007).
Recently, using the VOE paradigm, Kovács et al. (2010) reported that infants as
young as 7-months could compute an agent’s belief and maintain it even in the
absence of the agent.
Mentalism in Children’s Spontaneous False-Belief Understanding
Author's personal copy
The above results, however, are intensely debated. On one hand, the findings
appear to be commensurate with the possibility that false-belief understanding may
be present very early in life, and may have an innate basis (Leslie 2005). On the
other hand, Perner and Ruffman (2005) offered a number of alternative explanations.
For instance, with respect to Onishi and Baillargeon’s data, they suggested that
looking times may be due to infants grappling with new agent-object-location
association links in the test event that appear different from the ones formed during
familiarization. It is also not clear whether infants attributed false-belief or ignorance
to the agent (Southgate et al. 2007).
Senju et al. (2011) ruled out the possibility that infants’ success is based only on
associations in the stimuli. In their AL study, 18-month-olds first experienced either
a visually identical opaque or trick blindfold (looked opaque but could be seen
through), and subsequently watched a video sequence (similar to Southgate et al.
2007) in which an actor wore the respective blindfold while an object was displaced
in front of her. Infants who experienced the opaque blindfold showed AL that the
actor’s search action would accord with her having a false-belief about the object’s
location, but infants who experienced the trick blindfold did not. It is important to
emphasize that infants could not observe either themselves or others wearing the
blindfold and, further, the same video sequences was presented in both conditions.
The results, then, are unlikely due to infants acquiring and grappling with possible
associations between blindfold-wearing and observable behaviors. It is an important
discovery that 18-month-olds are at least able to represent what an agent did and did
not see at a given point in time, and are able to use such information for anticipating
an agent’s future action.
Ignorance-based interpretations are readily contained by Southgate et al.’s (2007)
AL study. Since the target item was removed completely from the scene in Southgate
et al.’s design, an ‘ignorance leads to error’ interpretation would have no basis in
predicting preschoolers’ correct visual orienting: with the object removed, both
locations are incorrect. And yet, 2-year-olds looked first at the empty belief-based
location when anticipating where the agent would search for the toy. Recent VOE
data also qualifies against ignorance interpretations of looking fixations. Scott and
Baillargeon (2009) tested 18-month-olds in an unexpected identity belief-inducing
event and reported that infants’ looking patterns were different depending upon
whether the agent had a mistaken belief or was ignorant. Their results indicated that
infants in the 2nd year of life distinguish that mistaken agents should act in accord
with their false-belief whilst ignorant agents should act randomly.
3 Efforts in Meeting Perner and Ruffman’s (2005) Challenge: Ruling
out Behavior-Rules
Whilst new generation AL and VOE evidence speak against association and
ignorance explanations, behavior rule interpretations may be the final frontier to
confidently ascribing mentalism as underwriting infants and young preschoolers’
spontaneous false-belief responding. For example, extent AL research has only
managed to test young preschoolers’ spontaneous false-belief understanding via a
single-shot unexpected transfer belief-inducing scenario (e.g., Clements and Perner
J. Low, B. Wang
Author's personal copy
1994; Low 2010; Ruffman et al. 2001; Senju et al. 2011; Southgate et al. 2007).
According to Perner and Ruffman (2005), young children’s AL can be inevitably
open to a rule-based explanation (e.g., children attribute that the agent will look for
the object where he or she last saw it). Restricted documentations of accurate
responses being coherent across diverse belief-inducing scenarios have implications
for whether we can straightforwardly classify responses stripped to the level of eye
gaze as reflecting implicit mental-state understanding (as opposed to implicitly
implemented in procedural rules). For Perner and Ruffman, “the [mentalistic]
conclusions from the standard false-belief task are warranted only because
understanding of false-belief around 4 years can be demonstrated in a variety of
belief-inducing situations” (p. 216).
To meet Perner and Ruffman’s (2005) challenge, different VOE studies have
tested and confirmed infants’ reasoning in different belief-inducing scenarios, such
as false-beliefs about location, contents, identity, and properties. Infants show
accurate looking across these contexts (e.g., Scott and Baillargeon 2009; Scott et
al. 2010; Song and Baillargeon 2008; Song et al. 2008; Surian et al. 2007). VOE
research has been cutting-edge in documenting the breadth of infants’ looking
fixations as being potentially fitting with an early implicit mind-reading
hypothesis. That said, the combined evidence of success amongst infants in
diverse belief-inducing contexts is largely based on data between age groups and
between children. Researchers are only beginning to uncover whether children of a
given age who show accurate looking responses in one false-belief inducing
scenario will necessarily show accurate looking responses in other false-belief
inducing scenarios (He et al. 2011). Comprehensive data bearing on coherence in
spontaneous responses across diverse false-belief inducing tasks in the very same
age cohort (weaker evidence) and the very same infants/children (stronger
evidence) are still lacking.
Furthermore, behavior rule analyses of individual VOE studies are possible.
Consider, for example, Song and Baillargeon’s (2008) study. In their study, 14.5month-olds
watched events where an agent faced a stuffed skunk and a doll with
blue pigtails; the agent consistently reached for the doll indicating that she preferred
it over the skunk. In the false-belief trial, whilst the agent was absent, the
experimenter hid the doll in a plain box and hid the skunk in a hair box (the hair box
had a tuft of blue hair protruding from under its lid). The results indicated that
infants who saw the agent grasp the plain box looked reliably longer than those who
saw the hair box outcome event. For Song and Baillargeon, the findings suggested
that infants mentalistically reasoned that the agent mistakenly perceived the tuft of
hair as part of the doll and hence falsely believed that the doll was hidden inside the
hair box and so were surprised when she reached for the hair box. However, it is
equally possible for infants to base their looking responses upon the general
behavior rule that a person will reach for the box where attributes of the object to be
retrieved are visible. Song and Baillargeon also found that when the agent had a
preference for the stuffed skunk over the doll, infants looked reliably longer when
the agent grasped the hair box than those who saw the plain box test event. Even in
this situation, a complementary general behavior rule could also explain the pattern
of looking fixations: a person will avoid reaching to a location where attributes of
the object not to be retrieved are partly visible.
Mentalism in Children’s Spontaneous False-Belief Understanding
Author's personal copy
This challenge in arbitrating between mind-reading and behavior-reading explanations
does not mean that VOE research has failed to control for any behavior rule
interpretations. Scott et al. (2010), for example, reasoned that if 18-month-olds
possessed a robust implicit understanding of false-belief and are not constrained by
behavior search rules for predicting agents’ actions, they should be able to show
appropriate visual expectations on a non-search task. Their reasoning parallels research
indicating that 4-year-olds who verbally pass search based false-belief tasks (e.g.,
unexpected transfer) also verbally pass non-search false-belief tasks (e.g., appearancereality)
(e.g., Gopnik and Astington 1988). In Scott et al.’s study, as an agent watched,
an experimenter demonstrated that her object had a non-obvious property (it made a
rattling sound when shaken). Next, the experimenter asked the agent to perform the
action by choosing between two test objects: one was identical to the experimenter’s
object and the other differed in color and pattern. However, infants (but not the agent)
were shown in a prior trial that the different test object rattled when shaken, but the
identical test object did not. Thus, the agent will have a false-belief about similar
objects having similar non-obvious properties. Infants looked longer when the agent
(upon being asked by the experimenter, “Can you do it?”) reached for the nonidentical
test object compared to the identical test object. Infants appear to be able to
attribute others’ false-beliefs about non-obvious properties. Scott et al. anticipated that
such findings could still be open to infants following a particular behavior rule that
agents will select an object that looks most like a prior object when reproducing an
interesting sound. Thus, they conducted a clever follow-up experiment whereby the
experimenter first demonstrated to the agent the properties of her object and the two
test objects by shaking each object in turn. The agent had knowledge that the different
test object rattled whilst the identical test object did not make any sound. Then, whilst
the agent was absent, the experimenter removed the lids from the two test objects and
poured the marbles from the different test object into the identical test object. In this
reversed false-belief scenario, infants flexibly accommodated to the reversal of the
agent’s awareness—they looked reliably longer when the agent reached for the
identical test object compared to the different test object.
Träuble et al. (2010) also introduced variations in agents’ information access in
order to disentangle between rule-based and mind-reading accounts of infants’
looking fixations. In their study, a box was placed on either arms of a balance beam.
The agent could operate the balance beam by lifting one side manually leading the
ball to roll from one box to another without making any noise. Infants assigned to
the false-belief condition watched as the agent observed the ball roll from one
location to the other. Then, whilst the agent’s back was turned, infants saw the ball
change its location. For infants allocated to the manual-control condition, the only
difference was that whilst the agent’s back was turned, the agent herself used her
hand to lift one side of the balance beam causing the ball to change its location.
Infants in the false-belief condition looked longer when the agent reached inside the
ball box compared to the empty box (replicating Onishi and Baillargeon 2005).
Infants in the manual-control condition, however, looked longer when the agent
reached inside the empty box compared to the ball box. Träuble et al. suggest that
differences in reactions between the two conditions indicate that infants’ looking
responses are not pinned to the behavior rule that agents will look where they last
saw. If this rule was dominating, then infants should look longer when the agent
J. Low, B. Wang
Author's personal copy
reached for the ball box in the manual control condition, regardless of whether she
might know the object’s whereabouts during the test event even if she did not see the
location reversal. Whilst Scott et al. (2010) and Träuble et al.’s (2010) attempts to
contain non-mentalistic interpretations of children’s looking responses are forward
thinking, we will go on to discuss how each of these studies has not yet managed to
control for behavior-rule explanations that are complementary to the mind-reading
explanations under consideration.
4 Plausibility Arguments are Relevant but Not Decisive
To date, no experiment has yet been designed whose positive results can only be
explained and predicted by a mental-state account but not by a complementary
behavior-rule account. To differentiate between the two accounts, researchers have
raised plausibility arguments, especially about parsimony (Perner 2010, 2011).
According to the principle of parsimony, the account that makes the fewest
assumptions will be preferred. However, as Perner (2011) has pointed out, depending
on how one “counts” assumptions, each account may claim to have parsimony on
their side.
For researchers who favor a rich interpretation, a mental-state account is argued to
involve just “one rule” that governs infants’ expectations about agents’ behavior: people
act based on the information available to them (He et al. 2011; Scott et al. 2010). Whilst
it is possible to spin out in specific situations which specific false-belief infants might
attribute (e.g., about the location, identity, properties, or contents of an object), the
bottom line is that, if children can determine that the agent holds false information
about an aspect of a scene, they will expect the agent to act on it. For example, He et
al. sought to uncover whether VOE fixations amongst children in same age group
(2.5-year-olds) were consistent between the unexpected transfer and unexpected
contents belief-inducing scenarios. The results of their study showed that—across the
two belief-formation scenarios—toddlers consistently looked longer at unexpected
outcomes wherein agents’ search actions conflicted with their false-beliefs about
desired items being in a particular location or were incongruent with their false-beliefs
about which packages contained items they wanted. A recent study by Scott and
Baillargeon (2009) also suggested that infants may be capable of attributing two falsebeliefs
simultaneously: the agent’s false-belief about the identity of the penguin toy
under the transparent cover being a single piece leads her to hold the additional falsebelief
that the 2-piece penguin toy must be under the opaque cover. If we considered
that the mental-state account only involved one rule that people act on information
available to them, then no new rule would be needed for reasoning in new false-belief
inducing scenarios: the rule is always the same and makes the same predictions. In
contrast, when more and more belief-inducing situations are examined, a behavior-rule
account must come up with more and more local heuristics to explain positive results.
On this analysis, there appears to be theoretical parsimony in favoring a ToM
explanation over a behavior-rule explanation of infants and young preschoolers’
spontaneous false-belief responding.
The analysis changes dramatically, however, depending on how else researchers
define and “count” behavior and ToM rules. Perner (2010) has argued that when one
Mentalism in Children’s Spontaneous False-Belief Understanding
Author's personal copy
explicates the computational abstractions infants and young preschoolers must make
in order to anticipate how agents’ minds interface with their environment and their
actions, a mentalism account also requires a number of rules (see also Whiten 1996).
Importantly, whilst Perner views the input–output functions to be the same in both
accounts, their internal structures differ. A behavior rule captures a direct and
specific link between a situation and an action (S→A). If we combine different
situations with different actions, then the number of the behavior-rules needed for
correct predictions in all cases will be the number of different situations multiplied
by the number of different actions. On the other hand, in a mentalism account, a
mental state (e.g., false-belief) is considered as an intervening variable (S→M→A)
(Whiten). If we combine different situations with different actions, then the number
of mental-state rules for correct predictions in all cases will be the number of
different situations plus the number of different actions. Put formally, increases in
rule complexity for determining coherent false-belief attribution are determined by
computational procedures needed for combining belief-formation (f) by belief-use
(u) situations wherein the number of rules required for claiming ToM-rules over
behavior-rules for action prediction are (f+u) and (f×u), respectively. Applied to He
et al.’s (2011) study, toddlers’ attributions were diagnosed across 2 belief-formation
scenarios (unexpected transfer and unexpected contents) by 1 belief-use scenario
(predicting where agents would look for a target object). AToM-rules account would
require a total of 3 (2+1) mentalistic abstractions to predict coherent looking
responses. The first mentalistic abstraction is that if the agent did not see the object
transferred from location 1 to location 2, then the agent will have a false-belief that
the object is in location 1. The second abstraction is that if the agent did not see the
contents of containers at locations 1 and 2 being swapped, then the agent will have a
false-belief that the container at location 1 will contain the desired item. In order to
correctly predict or anticipate the agent’s desires to look for the target object across
both belief-inducing situations, the final abstraction in the web is that if the agent has
a false-belief, then the agent will look for the target item at location 1. By
comparison, a behavior-rules account would only require a total of 2 (2×1)
abstractions to accommodate coherent looking responses. The first behavior-rule is
that agents will look for an item in a location where they last saw it. The second
behavior-rule is that agents will look for an item in containers that usually contain it.
Of course, a hidden assumption here is that the unit cost for a given rule is the same,
regardless of whether that rule is behavioral or mentalistic. Overall, based on this
alternative treatment, a behavior-rules account turns out to be more economic of
toddlers’ representational resources compared to a ToM-rules account.
Other VOE studies that have attempted to contain behavior rule interpretations are
amenable such an alternative cognitive analysis. With respect to Träuble et al.’s
(2010) study, for example, the complexity of a ToM-rules account for accommodating
where agents will look for unexpectedly and manually transferred objects could take
the following form. First, in the false-belief condition, if the agent did not see the ball
roll from locations 1 to 2, she would have a false-belief of its whereabouts in location
1. Second, in the manual-change condition, if the agent did not see but intentionally
caused the ball to roll from locations 1 to 2, she would have a true-belief of its
whereabouts in location 2. Third, if the agent has a false- or true-belief, she will look
in locations 1 or 2, respectively. Nonetheless, compared to the total of 3 ToM-rules
J. Low, B. Wang
Author's personal copy
required to deploy correct looking across Träuble et al.’s tasks, a behavior-rules
analysis is also possible, and more economic of infants’ representational resources.
The first behavior-rule is that agents will look for objects where they last saw them roll
to. The second rule is that agents will look for objects where they last felt them roll to.
Similarly, Scott et al.’s (2010) data indicating toddlers’ coherent VOE
responses across two belief-formation scenarios (standard and reversed falsebelief
property induction) for anticipating what agents will do to reproduce
interesting sounds (1 belief-use situation) is also qualified by behavior-rules when
subjected to such a formal cognitive analysis. AToM-rules account would require 3
(2+1) abstractions to cover all correct predictions for what agents’ actions might
be. First, if agents are ignorant of a dissimilar test item sharing non-obvious
properties with the pivot item, they will have a false-belief of the hidden property
of the perceptually similar test item. Second, if agents are ignorant of the insides of
the similar and dissimilar test items being reversed, they will have a false-belief of
the hidden property of the dissimilar test item. Combining across scenarios, the
third ToM-rule stipulates that if agents have a false-belief of the similar or
dissimilar items they will pick up those respective items to reproduce an interesting
sound. However, only two (2×1) complementary behavior-rules are needed to
account for coherence in looking times: agents will reach for matching items to
reproduce an interesting sound; and agents will reach for items that they last saw/
heard made an interesting sound to reproduce an interesting sound. We believe that
He et al. (2011), Scott et al. and Träuble et al. (2010) are on the right track; but they
only combined multiple belief-inducing situations with a single use of belief (e.g.,
predicting where an agent will look for an object or predicting what an agent will
do to produce an effect). In such instances, a ToM-rules account would have an
explanation deficit whilst a behavior-rules account would have a natural
explanation by being more economic of representational resources (f+1 rules>f×1
rules, respectively).
Based on the extant findings, it is premature to conclude that theoretical
parsimony favors an early mentalism explanation per se. Advocates of either a
behavior-rule account or a ToM account can claim to have theoretical parsimony on
their side, depending on how one treats stimulus-action and mental-state rules. We
also cannot straightforwardly evaluate between the relative parsimony of behavior-rule
compared to mentalistic models when children’s information processing systems are not
the outcomes of a rational design but a messy evolutionary process (Apperly and
Butterfill 2009; Perner 2010, 2011). Plausibility arguments in and of themselves are
not conclusive; they are only relevant considerations for guiding research. We need
more decisive empirical data to differentiate between mentalistic and behavior-rule
accounts.
5 Potential Approaches for Differentiating Mentalism and Behavior-rules
Depending on whether researchers focus on spontaneous looking responses or
elicited verbal responses, existence of the ability to attribute false-belief mental states
ranges from 7-months to 4-years of age (Kovács et al. 2010; Wellman et al. 2001).
Presently, it is difficult to rule out behavior-rule explanations of children’s early
Mentalism in Children’s Spontaneous False-Belief Understanding
Author's personal copy
spontaneous looking responses. In fact, an argument in favor of behavior-rules is
also possible for 4-year-olds. Although 4-year-olds start to give correct verbal
predictions on standard false-belief tasks, their verbal justifications still smack of
behavior-rule following (e.g., “because that’s where he put the chocolate”; “because
he couldn’t see mother putting it there”) (Perner 2011). Wimmer and Mayringer
(1998) reported that mentalistic justifications only became prominent from 6-years
of age (e.g., “because he thinks the ice cream man stays there”). Nonetheless, in the
current literature, it is generally agreed that children from their 4th birthday do show
ToM understanding. Scientists’ agreement over such a conclusion stems from
decades of research showing that 4-year-olds are able to flexibly master different
belief-formation situations (e.g., unexpected transfer, unexpected contents, misinformation,
appearance-reality) and, further, are able to deploy their attributions for
different uses or demands (e.g., judging where agents will look for, where they will
think, what they will say, where they will point to) (Wellman et al.). The fact that
there is no great variation occurring at this onset age for solving different tasks with
different demands is important in terms of constraining a behavior-rule account. As
Apperly (2011) pointed out, “it would be truly surprising if children [at this age] just
happened to learn simultaneously a whole set of behavioral rules suitable for
predicting agents’ actions on the basis of false-beliefs” (p. 42). Evidence of 4-yearolds’
verbal predictions cohering across a variety of tasks combining different beliefformation
scenarios with belief-use situations around the same time partly provides
strong evidence that children at this age are reasoning about mental content.
By the same token, if we wish to conclude that children younger than 4-years
possess an understanding of ToM at some level, we also need to provide evidence of
infants and young preschoolers passing tests of coherence. Specifically, we will need
to provide evidence that infants and young preschoolers’ spontaneous looking
responses cohere across different false-belief inducing situations combined with
different response demands at the same time. Let us canvas one such scenario.
Consider the idea of testing whether 3-year-olds’ spontaneous false-belief responses
are robust across interactions between three different belief-inducing scenarios (e.g.,
unexpected contents, unexpected transfer, and misinformation) and three different
desire-inducing situations (e.g., anticipating through the “I wonder” prompt where
agents would look for an item, say where an item is, or point to where an item is)
(see Fig. 1).
If a sample of 3-year-olds showed robust and correct AL or VOE responding
across all of these combinations at the same time, a mentalistic account would have a
natural explanation by needing only to posit a total of 6 (3+3) ToM-rules (see Fig. 1,
bottom panel). In contrast, a behavior-rules account would have an explanatory
deficit since a total of 9 (3 x 3) such rules would be needed to cover the range of
evidence (see Fig. 1, top panel). If researchers were to discover that young
preschoolers’ pass in their spontaneous looking on all of the combinations at the
same time, it would suggest that the cognitive system can compute how different
epistemic and connative states interact and generalize to an invariant and intervening
mental state (false-belief) which, in turn, supports correct action prediction in all
cases. If such evidence were obtained, we could say that young children possess
implicit mentalistic understanding because it is unlikely that children acquire all of
the 9 behavior-rules at the same time.
J. Low, B. Wang
Author's personal copy
It should be clear, then, studying visual orienting or visual fixation across multiple
belief-formation scenarios combined with multiple belief-use scenarios is at least one
critical step for diagnosing when young candidate mind-readers show implicit
recognition of states of mind (see Fig 2).
Whilst researchers have investigated spontaneous responding across different
belief-formation scenarios, they have only required children to put their belief
attribution to a single use (e.g., anticipating where agents will search/look for a
desired object) (e.g., He et al. 2011). Further, whilst infants succeed at VOE, helping,
and referential communication tasks wherein each of the 3 tasks required belief
B1: Will look where last saw X is
B2: Will say where last saw X is
B3: Will point to where last saw X is
B4: Will look where last told X is
B5: Will say where last told X is
B6
B7: Will look in container that usually has X
B8
: Will point to where last told X is
: Will say in container that usually has X
B9: Will point to container that usually has X
T1: If saw X in L1, then will believe X in L1
T2: If misinformed X in L1, then will believe X in L1
T3: If saw X container in L1, then will believe X in L1
T4: If possess false-belief, then will look for X in L1
T5: If possess false-belief, then will say X in L1
T6: If possess false-belief, then will point to X in L1
T5
f1: Unexpected
Transfer u1: Look
u2: Say
u3: Point
u1: Look
u2: Say
u3: Point
f2: Misinformation
f3: Unexpected
Contents
Belief
For Behavior-Rules account: (3) formation × (3) use = 9 behavior-rules
For Theory of Mind rules account: (3) formation + (3) use = 6 ToM-rules
f2:
Misinformation
f3: Unexpected
Contents
B3
B1
B2
B6
B4
B5
B7
B8
B9
f1: Unexpected
Transfer
T2
T1
T3
T4
T6
Fig. 1 Differences in complexity of behaviour-rules (B) and ToM-rules (T) in a 3 (belief-formation) by 3
(belief-use) combination
Mentalism in Children’s Spontaneous False-Belief Understanding
Author's personal copy
attribution to be put to a different use, all of the tasks involved only 1 beliefformation
scenario—unexpected transfer (Buttelmann et al. 2009; Onishi and
Baillargeon 2005; Southgate et al. 2010). Even in these cases, if we adopt the
specific cognitive analysis introduced here, behavior-rules will always turn out to be
more economical of neural resources underpinning the coherence of looking
responses (see Fig. 2a). The empirical turning-point favoring ToM-rules over
behavior-rules, however, is revealed if and when infants and young preschoolers
show accurate spontaneous looking responses simultaneously across at least 2
different belief-inducing situations combined with at least 3 different desire-inducing
situations, or vice versa. At these break points, a ToM-rules account would only need
to posit a total of 5 abstractive links (2+3 or 3+2) to explain breadth of spontaneous
false-belief responding, as compared to the total of 6 links needed under a behaviorrules
account (2×3 or 3×2) (see Fig. 2b, c). Of course, a ToM-rules account would
have significant edge over a behavior-rules account by being substantially
economical of representational resources if individual infants and young preschoolers
turn out to be even capable of spontaneously classing large batches of
belief-formation scenarios as leading to the same mental state and, further, putting
this classification to an armamentarium of uses (e.g., where belief-formation is ≥2
and belief-use is ≥4) (see Fig. 2d).
0
2
4
6
8
10
TotalRuleComplexity
TotalRuleComplexity
TotalRuleComplexity
TotalRuleComplexity
Belief-Formation (f)Belief-Formation (f)
Belief-Formation (f)Belief-Formation (f)
0
2
4
6
a: where Belief-Use (u) = 1
ToM-
Rules
(f+u)
ToM-
Rules
(f+u)
BeH-
Rules
(fu)
0
2
4
6
8
10
12
14
c: where Belief-Use (u) = 3
b: where Belief-Use (u) = 2
d: where Belief-Use (u) = 3
BeH-
Rules
(fu) 0
3
6
9
12
15
18
1 2 3 41 2 3 4
1 2 3 4 1 2 3 4
Fig. 2 Changes in total rule complexity for predicting coherent responses across different belief-formation
(f) by belief-use (u) combinations as stipulated in a ToM-rules account (f+u) compared to a behavior-rules
account (f×u). In (a) where looking responses are based on f≥1 and u=1, a behavior-rules (BeH) account
is consistently economical of representational resources. In (b and c) a ToM-rules account is more
economical of representational resources, as compared to a BeH-rules account, if u=2 and f≥3, or u=3
and f≥2, respectively. In (d) the complexity of computations needed for covering breadth of correct
response changes dramatically in favor of a representationally economical ToM-rules account when large
batches of belief-use situations (u≥4) are combined with multiple belief-formation scenarios (f≥2)
J. Low, B. Wang
Author's personal copy
At the time of writing, we are aware that Wang et al. (2011a, b) adopted the
specific cognitive analysis outlined here and used a within-subjects design to test
whether Chinese preschoolers (3- and 4-year-olds) showed accurate spontaneous
false-belief understanding across two belief-formation scenarios (misinformation and
unexpected contents) combined with three belief-use situations (anticipating where
an agent will look, say or point to where a desired item was located). Wang et al.
used a neutral Mandarin translation of the verbal “I wonder” prompt to trigger
children’s AL. As far as we know, they are also the first to document spontaneous
AL-based false-belief understanding amongst children growing up in a non-Western
culture learning to speak a non-Indo-European language. Wang et al. found that,
despite extant indications of Chinese children possessing a surfeit of executive
functioning abilities (see Sabbagh et al. 2006) that could enhance their expression of
ToM understanding, 3-year-old Chinese preschoolers consistently showed correct
first looks across the full range of belief-formation by belief-use combinations just as
they consistently erred on elicited judgments. The consistency of 4-year-olds’
accurate AL responses dovetailed with their correct verbal answering. It is promising
that Wang et al.’s Chinese data confirmed accurate AL simultaneously spreading
across a 2 (belief-formation) by 3 (belief-use) combination—thus suggesting
mentalism (5 ToM-rules), as compared to behaviorism (6 stimulus-action rules), as
being more economical of young preschoolers’ representational resources. It remains
unknown as to whether 2-year-olds (in Mainland China or elsewhere) would pass
such stringent tests of coherence. Extrapolating from current directions in the work
of Baillargeon and colleagues (e.g., He et al. 2011; Scott et al. 2010) and Wang et al.,
we are hopeful that it is only a matter of time when VOE and AL researchers boldly
go on to discover that the very same infants and toddlers are also capable of making
accurate spontaneous responses across multiple belief-formation by multiple beliefuse
combinations. The field can make significant advances in exploring these issues
if we adopt an explicit cognitive analysis that arbitrates in a principled manner when
early spontaneous responses to false-belief tasks are mentalistic and when they are
not. As argued here, we cannot just yet confidently ascribe mentalism over smart
behavior rule following to the full range of evidence on infants, toddlers and young
preschoolers’ spontaneous responses in false-belief tasks. We can get there if we start
to gauge children’s spontaneous responses against the empirical yardstick of
minimally showing understanding across different belief-situation by desirepurpose
combinations required to rule out application of behavior-rules.
We have canvassed the promising solution of designing experiments to test the
coherence of infants and toddlers’ spontaneous responses so as to differentiate which
account is more economical of neural resources, thereby being the more efficient
account. But as acknowledged earlier, evolution and development do not always
yield the most rational or economical solutions to problems. Whilst tests of
coherence for differentiating which account requires the fewest rules (and thus the
most parsimonious) are relevant, they are not decisive. Other kinds of converging
empirical evidence will also be needed to indicate that behavior-rules may be
insufficient to explain all instances of efficient mentalistic reasoning. Other
approaches include documenting: whether dual task interference has an impact on
explicit verbal responses but not AL responses (Newton and de Villiers 2007);
whether there are positive transfer effects from trained to untrained false-belief tasks
Mentalism in Children’s Spontaneous False-Belief Understanding
Author's personal copy
(for transfer tests to be potent, novel belief-formation by belief-use scenarios will be
needed so that researchers can gain some degree of control over the acquisition of
children’s knowledge) (Perner 2010); and whether there are real signature upper
limits to the flexibility by which infants and young preschoolers can apply and
generalize their early tacit ToM understanding (Apperly and Butterfill 2009). We turn
to spotlight the issue of limits in spontaneous knowledge because it is analogous to
our discussion of the issue of coherence in spontaneous understanding across
combinations of belief-formation by belief-use scenarios.
If one is right to interpret that there exists two types of false-belief reasoning
systems—an early developing implicit representational system that has a modular
basis and a slowly emerging explicit representational system that is partly
underpinned by important growths in executive functioning and language ability
(Low 2010; Low and Simpson 2011)—then modular understandings must exhibit
efficiency at the expense of flexibility (Apperly and Butterfill 2009). Apperly and
Butterfill suggested that the economy/efficiency inherent in children’s implicit
understanding of mental states may be achieved by representing only links between
agents, objects and properties and avoiding complexities entailed in processing links
between agents and propositions. An implicit representational system would then
allow for the processing of links between visual access, knowledge or false-belief
and consequent behavior (e.g., Baillargeon et al. 2010; Kovács et al. 2010; Senju et
al. 2011). At the same time, however, implicit mindreading should not be expected to
support processing of Level-2 types of mental representation problems which require
recognition that an object may present different visual appearances to two people if
they view it from different positions (Apperly 2011). Indeed, Samson et al. (2010)
found that in a Level-1 visual perspective taking task that required tracking what an
agent can or cannot see, adults exhibited egocentric interference when judging how
many dots an agent could see (participants were slower at judging the agent’s
perspective when they themselves saw more dots). Participants also showed
altercentric interference when judging how the number of dots they themselves
could see (adults were slower when the agent saw fewer dots than when the agent
saw the same number). In this case, altercentric interference is an important finding
because it shows that adults are implicitly and automatically computing the agent’s
perspective even when there was no need to do so. Nonetheless, there appears to be
limits to the mileage in implicit automatic mental-state processing. In a Level-2 type
of task that required representation of the particular way in which an agent thinks
about something (e.g., a digit is read as “9” from the participant’s perspective but
read as “6” from the agent’s point of view), however, Surtees et al. (2010, cited in
Apperly 2011) found no evidence of altercentric interference: adults only showed
egocentric interference when judging the agent’s perspective. The presence of
altercentric interference in the Level-1 but not Level-2 tasks suggests that whilst we
do implicitly and automatically process an agent’s perspective, there are limits to the
deployment of implicit mentalistic understanding. The study of limits in implicit
mentalistic understanding is still in its early stage. It will be important to suitably
replicate the null results for automatic Level-2 perspective processing (and in more
than one paradigm) with infants and young preschoolers. As such, it will be critical
to discover whether the implicit mental-state reasoning system available to infants
and young preschoolers will efficiently cohere across a range of spontaneous falseJ.
Low, B. Wang
Author's personal copy
belief tasks but will distinctively not support the spontaneous processing of
propositional content for a range of Level-2 ToM tasks. To echo Apperly and
Butterfill, the discovery of signature limits in children’s spontaneous responses to
ToM tasks is another powerful technique that allows us to tell when an implicit
mentalistic reasoning system is and is not being deployed.
6 Going Beyond Spontaneous Visual Responses
Almost all current evidence on children’s implicit insight about false-belief rests on
eye movements or looking time (Baillargeon et al. 2010; Low 2010). If this were the
only way in which children’s implicit understanding was manifest, then we would
not expect this early conceptual insight to have much impact on their everyday
behavior. To assess what pre-linguistic children understand of others and their
worlds, some researchers have begun using interactive tasks, including helping (e.g.,
Buttelmann et al. 2009) and referential-communication paradigms (e.g., Southgate et
al. 2010). An important characteristic of such interactive paradigms is that infants
and children are not required to predict what an agent will do by looking to the
correct location, but they are required to initiate a more active behavioral response.
For example, in the false-belief condition of Buttelmann et al., an experimenter
showed infants (16- and 18-month-olds) two lidded boxes and demonstrated how to
lock and unlock them, and thereupon the boxes were left unlocked. Next, a male
agent entered the room, hid a toy in one of the boxes, and then left. While the agent
was absent, the experimenter moved the toy to the other box and locked both boxes.
When the agent returned, he tried to open the box where he had hidden the toy,
without success, and then sat centered behind the boxes. The rationale was that if
participants understood the agent’s false-belief and wanted to help, they should infer
that he wanted the toy he thought was in there. In this case, children should not go
and help him open the first box but, instead, go to the other box and extract the
desired item for him. Buttelmann et al. found that when prompted to help the agent,
the majority of 18-month-olds approached the other box containing the toy which
the agent did not approach.
Southgate et al. (2010) provided convergent evidence using a referentialcommunication
task to test whether 17-month-olds were able to understand a
communicator’s false-belief and, further, use this understanding in a pragmatic
context. The infants watched as a female agent hid two different toys in two lidded
boxes and then left. While she was gone, an experimenter switched the toys. The
agent then returned, pointed to one of the boxes, and announced that the toy inside it
was a “sefo”. Southgate et al. reported that when prompted to get the sefo, most
infants approached the other box (i.e., the one that the agent did not point to). There
is narrow age range tested across these interactive paradigms (17- and 18-months) to
promisingly suggest that infants might be able to coherently deploy their false-belief
understanding across both helping and communication contexts.
That said, it will still be important to test whether the very same infants (i.e., using a
within-subjects design) are able to use their false-belief understanding actively across
both task contexts. Indeed, in Buttelmann et al.’s (2009) study, 16-month-olds
performed randomly in the true-belief condition as compared to their 18-month-old
Mentalism in Children’s Spontaneous False-Belief Understanding
Author's personal copy
counterparts. Since 16-months-olds failed to attribute the agent’s mental state in
the true-belief condition, their accurate responses in the false-belief condition
may not be fully indicative of understanding false-belief per se. This
complication raises the possibility that 16-month-olds might not necessarily
transfer an apparent false-belief understanding from an inactive looking context
(as in the VOE scenarios) to an active helping context or vice versa.
Demonstrating that infants’ spontaneous responses cohere across inactive and
active false-belief inducing situations at the same time will be partly important
for ascribing mentalism per se. Buttelmann et al. suggested that 16-month-olds
might possess a genuine understanding of others’ mental state, but had
difficulty inhibiting going for the target toy in the true-belief condition. Scott
et al. (2010) have also raised this possibility, suggesting that interactive situations
might require more inhibitory control compared to spontaneous visual responding.
Scott et al. argued that in the spontaneous visual response paradigms such as VOE
and AL, children are passive observers, who can focus solely on the agent’s
perspective on the scene, with their own perspectives being less salient. On the
other hand, interactive tasks may impose higher levels of inhibitory control
because children need to actively hold in mind both their own and the agent’s
perspectives. Future studies investigating whether children’s inhibitory skills
predict their performance across non-interactive and interactive response tasks
can provide insight into the processing load underlying these tasks. Nonetheless,
helping and referent retrieval goals can be the basis of generating different beliefuse
situations in spontaneous false-belief tasks so that researchers are not focused
on measuring only children’s eye gazing responses. To that aim, documenting
whether looking responses can simultaneously spread to accurate helping
responses across diverse belief-inducing and desire-inducing tasks will help
establish when young children grasp false-belief and how such understanding is
implemented.
7 Conclusion
Overall, whilst extant work on early false-belief understanding is suggestive, the
evidence is not yet compelling. There is currently insufficient evidence of young
children’s visual fixations or anticipations cohering across different situation-purpose
combinations so as to rule out learning of behavior-rules. Although this kind of
evidence is hard to come by, researchers are now starting to analyze whether
spontaneous understanding—in children of a given age cohort and amongst the very
same children—coheres across different false-belief inducing situations combined
with different response demands. An explicit cognitive analysis of how early
understanding is structured and spread can partly help us take a step in the right
direction on the long road to answering how early in development children
mentalistically and implicitly grasp false-belief.
Acknowledgments We are grateful to Josef Perner for helpful theoretical encouragements. We also
thank Victoria Southgate and two anonymous reviewers for their generous comments.
J. Low, B. Wang
Author's personal copy
References
Apperly, I. 2011. Mindreaders: The cognitive basis of “theory of mind”. Hove: Psychology.
Apperly, I.A., and S.A. Butterfill. 2009. Do humans have two systems to track beliefs and belief-like
states? Psychological Review 116: 953–970.
Baillargeon, R., S.M. Scott, and Z. He. 2010. False-belief understanding in infants. Trends in Cognitive
Science 14: 110–118.
Buttelmann, D., M. Carpenter, and M. Tomasello. 2009. Eighteen-month-olds show false beliefs
understanding in an active helping paradigm. Cognition 112: 337–342.
Clements, W.A. 1995. Implicit theories of mind. Unpublished doctoral dissertation, University of Sussex.
Clements, W.A., and J. Perner. 1994. Implicit understanding of belief. Cognitive Development 9: 377–395.
Csibra, G., and V. Southgate. 2006. Evidence for infants’ understanding of false beliefs should not be
dismissed. Trends in Cognitive Science 10: 4–5.
Gopnik, A., and J.W. Astington. 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.
He, Z., M. Bolz, and R. Baillargeon. 2011. False-belief understanding in 2.5-year-olds: Evidence from
change-of-location and unexpected-contents violation-of-expectation tasks. Developmental Science
14: 292–305.
Kovács, A.M., E. Téglás, and A.D. Endress. 2010. The social sense: Susceptibility to others’ beliefs in
human infants and adults. Science 330: 1830–1834.
Leslie, A.M. 2005. Developmental parallels in understanding minds and bodies. Trends in Cognitive
Sciences 9: 459–462.
Low, J. 2010. Preschoolers’ implicit and explicit false-belief understanding: Relations with complex
syntactical mastery. Child Development 81: 579–615.
Low, J., and S. Simpson. 2011. Effects of labeling on preschoolers’ explicit false-belief performance:
Outcomes of cognitive flexibility or inhibitory control? Child Development.
Moses, L.J. 2001. Executive accounts of theory of mind development. Child Development 72: 688–690.
Newton, A.M., and J.G. de Villiers. 2007. Thinking while talking: Adults fail nonverbal false belief
reasoning. Psychological Science 18: 574–579.
Onishi, K.H., and R. Baillargeon. 2005. Do 15-month-old infants understand false beliefs? Science 308:
214–216.
Perner, J. 2010. Who took the cog out of cognitive science? Mentalism in an era of anti-cognitivism. In
Perception, attention, and action: International perspectives on psychological science (Volume 1), ed.
P.A. Frensch & R. Schwarzer, 241–261. Psychology Press.
Perner, J. 2011. Theory of mind—an unintelligent design: From behaviour to teleology and perspective. In
Handbook of theory of mind, ed. A.M. Leslie and T.C. German. NJ: Erlbaum.
Perner, J., and T. Ruffman. 2005. Infants’ insight into the mind: How deep? Science 308: 214–216.
Ruffman, T., W. Garnham, A. Import, and D. Connolly. 2001. Does eye gaze indicate implicit knowledge
of false belief? Charting transitions in knowledge. Journal of Experimental Child Psychology 80:
201–224.
Sabbagh, M.A., F. Xu, S.M. Carlson, L.J. Moses, and K. Lee. 2006. The development of executive
functioning and theory of mind. Psychological Science 17: 74–81.
Samson, D., I.A. Apperly, J.J. Braithwaite, B.J. Andrews, and S.E. Bodley Scott. 2010. Seeing it their
way: Evidence for rapid and involuntary computation of what other people see. Journal of
Experimental Psychology. Human Perception and Performance 36: 1255–1266.
Scott, R.M., and R. Baillargeon. 2009. Which penguin is this? Attributing false beliefs about identity at
18 months. Child Development 80: 1172–1196.
Scott, R.M., R. Baillargeon, H. Song, and A.M. Leslie. 2010. Attributing false beliefs about non-obvious
properties at 18 months. Cognitive Psychology 61: 366–395.
Senju, A., V. Southgate, C. Snape, M. Leonard, and G. Csibra. 2011. Do 18-month-olds really attribute
mental states to others? A critical test. Psychological Science.
Song, H., and R. Baillargeon. 2008. Infants’ reasoning about others’ false perceptions. Developmental
Psychology 44: 1789–1795.
Song, H., K.H. Onishi, R. Baillargeon, and C. Fisher. 2008. Can an agent’s false belief be corrected
through an appropriate communication? Psychological reasoning in 18-month-old infants. Cognition
109: 295–315.
Southgate, V., A. Senju, and G. Csibra. 2007. Action anticipation through attribution of false belief by 2year-olds.
Psychological Science 18: 587–592.
Mentalism in Children’s Spontaneous False-Belief Understanding
Author's personal copy
Southgate, V., C. Chevallier, and G. Csibra. 2010. Seventeen-month-olds appeal to false beliefs to interpret
others’ referential communication. Developmental Science 13: 907–912.
Surian, L., S. Caldi, and D. Sperber. 2007. Attribution of beliefs by 13-month-old infants. Psychological
Science 18: 580–586.
Träuble, B., V. Marinović, and S. Pauen. 2010. Early theory of mind competencies: Do infants understand
others’ belief? Infancy 15: 434–444.
Wang, B., J. Low, J. Zhang, and Q. Qinghua. 2011a. Chinese preschoolers’ implicit and explicit falsebelief
understanding. British Journal of Developmental Psychology: Special Issue on Implicit-Explicit
False-Belief Understanding.
Wang, B., J. Low, J. Zhang, and Q. Qinghua. 2011b. Chinese 3-year-olds show coherent false-belief
anticipatory looking across different belief-formation and belief-use task combinations. Manuscript in
preparation.
Wellman, H.M., D. Cross, and J. Watson. 2001. Meta-analysis of theory-of-mind development: The truth
about false belief. Child Development 72: 655–684.
Whiten, A. 1996. When does smart behavior-reading become mind-reading? In Theories of theories of
mind, ed. P. Carruthers and P.K. Smith, 277–292. Cambridge: Cambridge University Press.
Wimmer, H., and H. Mayringer. 1998. False belief understanding in young children: Explanations do not
develop before predictions. International Journal of Behavioral Development 22: 403–422.
J. Low, B. Wang
Author's personal copy