Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The scope and limitations of young children’s belief understanding
(USC Thesis Other)
The scope and limitations of young children’s belief understanding
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
THE SCOPE AND LIMITATIONS OF YOUNG CHILDREN’S BELIEF UNDERSTANDING
by
Qianhui Ni
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
[PSYCHOLOGY]
May 2024
Copyright 2024 Qianhui Ni
BELIEF UNDERSTANDING
Acknowledgments
I would like to thank my advisor, Dr. Henrike Moll, whose guidance, patience, and
expertise were instrumental in the completion of this dissertation. This PhD journey has been
enriched and shaped by your mentorship, making the challenging path of research a rewarding
voyage of discovery. Thank you for being more than an advisor—a mentor, a guide, and a role
model in my academic and personal growth.
I am immensely grateful to my committee members—Dr. Leor Hackel, Dr. David
Schwartz, Dr. Mark Schroeder, Dr. Adena Schachner, and former member Dr. Frank Manis—for
their solid support, invaluable feedback, rigorous scrutiny, and constructive criticism.
My sincere thanks go to the research assistants and graduate student fellows in the Minds
in Development Lab at USC for their invaluable help with my research. I am also grateful to the
families who signed up for my studies, whose participation was essential to this dissertation.
To my family and friends, and especially to the friends from my 2019 cohort, the
developmental area, the Social Learning & Choice Lab, the Annenberg Game Lab, and
Activision: I appreciate the encouragement, understanding, and belief you have shown in me.
Your love and support have shaped my journey in ways I could never fully express.
A special thank you to my beloved guinea pig, Huyang: you are the source of light and
joy in my life. Your tiny squeaks and soft, warm cuddles have provided immeasurable comfort
and emotional support through every page written and every challenge faced.
Lastly, I would like to acknowledge myself for making this journey incredibly fulfilling,
marked by countless moments of happiness and deep satisfaction. I am genuinely excited to see
where my future will lead, armed with the knowledge and experiences gained from this
significant phase of my life.
ii
BELIEF UNDERSTANDING
Table of Contents
Acknowledgments........................................................................................................................... ii
List of Tables.................................................................................................................................. iv
List of Figures..................................................................................................................................v
Abstract...........................................................................................................................................vi
Chapter 1 Introduction..................................................................................................................... 1
1.1 What is a Theory of Mind?.................................................................................................. 1
1.2. Developing Belief Understanding.......................................................................................1
1.3 Three Waves of Research on Belief Understanding in Young Children..............................3
1.4 This Dissertation’s Contribution to the Debate and Research on Theory of Mind............11
1.5 The Affective Dimension of a Theory of Mind................................................................. 12
1.6 Dissertation Overview....................................................................................................... 14
Chapter 2 Experience Tracking and Belief Understanding........................................................... 18
2.1 Introduction........................................................................................................................18
2.2 Current Study.....................................................................................................................20
2.3 Experiment 1......................................................................................................................21
2.4 Experiment 2......................................................................................................................32
2.5 General Discussion............................................................................................................ 40
Chapter 3 Experiential Records: What Are They Made of?..........................................................43
3.1 Introduction........................................................................................................................43
3.2 Current Study.....................................................................................................................44
3.3 Experiment 1......................................................................................................................45
3.4 Experiment 2......................................................................................................................51
3.5 General Discussion............................................................................................................ 57
Chapter 4 Backward Belief Reasoning: From Action to Belief and Experience...........................60
4.1 Introduction........................................................................................................................60
4.2 Current Study.....................................................................................................................63
4.3 Experiment 1......................................................................................................................65
4.4 Experiment 2......................................................................................................................75
4.5 General Discussion............................................................................................................ 85
Chapter 5 Concluding Discussion..................................................................................................90
References....................................................................................................................................101
Appendix......................................................................................................................................111
iii
BELIEF UNDERSTANDING
List of Tables
Table 1: Number of Children and Their Scores in Experiment 1 Tasks from Study 3 .................72
Table 2: Number of Children and Their Scores in Experiment 2 Tasks from Study 3 .................83
iv
BELIEF UNDERSTANDING
List of Figures
Figure 1: Materials Used in the Puppet Shows Broken Down by Story from Study 1.................23
Figure 2: Procedural Steps in Experiment 1 from Study 1 .......................................................... 26
Figure 3: The Number of Expressions Grouped by Condition and Group from Study 1............. 28
Figure 4: The Timing of Expressions Grouped by Condition and Group from Study 1 ..............30
Figure 5: Procedural Steps in Experiment 2 from Study 1 .......................................................... 35
Figure 6: Correct Response Rate to Test Question Broken Down by Condition and Group from
Study 1 .......................................................................................................................................... 37
Figure 7: Procedural Steps of Puppet Show Story in Experiment 1 from Study 2 ...................... 47
Figure 8: Number of Expressions Broken Down by Condition from Study 2 .............................49
Figure 9: Procedural Steps of a Puppet Show Story in Experiment 2 from Study 2 ................... 54
Figure 10: Mean Number of Expressions by Condition and Trial Type in Experiment 2 from
Study 2 .......................................................................................................................................... 56
Figure 11: Four Stories in the Witness Task in Experiment 1 from Study 3 ................................66
Figure 12: Procedural Steps of the Witness Task in Experiment 1 from Study 3 ........................68
Figure 13: The Classic Change-of-Location Tasks in Experiment 1 from Study 3 .....................69
Figure 14: Age Differences in the Witness Task in Experiment 1 from Study 3 .........................71
Figure 15: Predicting Witness Task Scores using Classic Change-of-Location Tasks Scores
Grouped by Age Group from Study 3 ...........................................................................................73
Figure 16: Four Stories in the Witness Task in Experiment 2 from Study 3 ............................... 77
Figure 17: The Classic Unexpected-Content Tasks in Experiment 2 from Study 3 .................... 78
Figure 18: Procedural Steps of the Witness Task in Experiment 2 from Study 3 ........................80
Figure 19: Performance of the 3- and 5-Year-Olds in the Witness Task from Study 3 ................82
v
BELIEF UNDERSTANDING
Abstract
This dissertation explored the scope and limitations of young children’s understanding of
others’ mental states. In three consecutive studies, I investigated 3- to 5-year-olds’ understanding
of others’ action-relevant beliefs or expectations, using a combination of classic
(question-answer format) and novel (facial expressions) measures. The first study (N = 120)
found that 3-year-olds can anticipate an agent’s false expectations about a container with
unexpected contents only when they can track the agent’s prior experience with objects. This was
revealed through their expressions. In verbal judgments, children showed no understanding of
the agent’s false expectations. The second study (N = 90) further examined toddlers’ need to
track others’ perceptual experiences. The third study (N = 120) introduced a novel approach to
understanding the relationship between beliefs and actions by asking children to reason
backward from observed actions to beliefs. It revealed a strong relation between backward and
forward reasoning about beliefs, as well as a developmental progression between 3 and 5. Taken
together, the findings show that 3-year-olds possess limited theory of mind skills: They fail to
understand that an object’s appearance can induce false beliefs or expectations, and the only kind
of belief-like states—expectations—that children this young can grasp are those grounded in
direct prior experience with the object. This dissertation supports a middle position on the
development of theory of mind, according to which a theory of mind is neither innately given nor
late-emerging.
vi
BELIEF UNDERSTANDING
Chapter 1 Introduction
1.1 What is a Theory of Mind?
There is wide agreement today that humans have a remarkably deep and “recursive”
understanding of the mind. We can, for example, entertain thoughts such as “Maxi thinks that I
think he is generous and kind.” Humans are, we might say, natural-born “mindreaders” (see
Apperly, 2010; Tomasello, 1999). Even young children, as we shall see, are impressively
competent at discerning others’ mental states, such as their intentions, beliefs, or desires. The
capacity to attribute mental states to other persons or to oneself, and the ability to make sense of
or predict actions based on these mental states has been referred to as a “theory of mind”.
As anthropologist Sarah Hrdy (2009) and others have pointed out, a theory of mind is an
essential capacity for a kind of creature that is as interdependent and relational as humans are
(see also Brüne & Brüne-Cohrs, 2006), and it is a skill that is important even for the very young,
who are reliant on others’ care, affection, and attention. Humans are what has been called
“cooperative breeders”: we divide the burden of childrearing between several adults, including
non-kin members of one’s community. This, as Hrdy (2009) argues, requires toddlers to be
attentive to their many caregivers’ states of mind. There is thus reason to think that throughout
the human lifespan, the daily social interactions on which our livelihoods depend often
necessitate awareness of what goes on in the minds of our interaction partners. Being able to
discern others’ beliefs, wants, and intentions subserves our capacity to communicate and
cooperate (Moll, Ni, & Stekeler, 2020).
1.2 Developing Belief Understanding
The studies I conducted for this dissertation address a crucial component of human theory
of mind: the capacity to discern what beliefs people hold. Beliefs are important mental states to
1
BELIEF UNDERSTANDING
understand because it is our beliefs about reality, rather than reality itself, that guides and
motivates our actions. We form our beliefs based on personal experience, testimonial reports,
teaching, or inference (Egan, 2008). Having a belief means taking something to be the case or
true; and it is what we take to be the case, not what is the case, that determines our actions. For
example, I might head to the store in the late evening because I believe it to be open, although
the store is, in fact, closed. Understanding my false belief would allow others to explain and
make sense of my seemingly irrational behavior.
Being able to detect or ascribe beliefs is considered the “litmus test” for having a theory
of mind because beliefs are among the most difficult mental states to comprehend (Wellman &
Woolley, 1990). This is because beliefs are usually less observable than many other mental
states. For example, what a person wants or is aiming to achieve (their intention) can often be
directly perceived in their behavior: in the way they approach an object or strive toward some
goal. Beliefs, on the other hand, need not directly manifest in action. In line with this
observation, it has been shown to be harder for young children to arrive at the understanding that
others may misrepresent states of affairs than, for example, to detect others’ desires or intentions.
Even infants and toddlers have been shown to understand, at some fundamental level, others’
wants and intentions (Gergely et al., 1995; Repacholi & Gopnik, 1997; Woodward, 1998, 2009).
Belief tests, however, have been notoriously challenging for children until the age of 4 or 5
(Wellman et al., 2001).
Another reason for the relatively late onset of belief understanding is beliefs’
“mind-to-world” direction of fit (Searle, 1983). Desires, hopes, and intentions have a
“world-to-mind” direction of fit: the world shall be as I am representing it to be. Seeing a person
express their liking for a certain object or perform goal-directed activities allows us to infer, or
2
BELIEF UNDERSTANDING
even directly perceive, what they want or are trying to achieve. Not so for beliefs. The “job” of
beliefs is to represent the world as it is, which entails the possibility of error. For example, I
might believe that the store I am heading to is open, not realizing that it has closed for the
holidays. False beliefs have implications for our decisions, actions, and feelings, and they often
entail other false assumptions and unrealizable plans. I will be surprised and disappointed,
perhaps even angry, when parked in front of the store, realizing that I cannot shop. I might even
have to break the promise I made to a friend to purchase something for her. Only someone who
recognizes false beliefs will be able to make sense of these reactions and implications that follow
from false beliefs.
Since the introduction of the concept of a theory of mind, an enormous amount of
research has been conducted to address its development in humans, with a special focus on belief
understanding. In a recent overview article (Moll et al., 2022), we distinguished between three
waves of research, which differ in their methodologies as well as the theoretical assumptions
about when and how belief understanding emerges. In the next subsection, I will review each of
the three waves of research and their corresponding theoretical commitments.
1.3 Three Waves of Research on Belief Understanding in Young Children
1.3.1 The First Wave: Classic False Belief Tasks and the Late-Onset View
The first wave of research created and relied on what is known today as the classic or
standard false belief tasks: the change-of-location task (Wimmer & Perner, 1983) and the
unexpected-content task (Perner, Leekam, & Wimmer, 1987). Both tasks are “direct” tests of
belief understanding because children are straightforwardly asked to identify an agent’s belief
about a specific state of affairs. In these tasks, an agent holds a false assumption about where
(change of location) or what (unexpected content) an object is. What differs between the tasks is
3
BELIEF UNDERSTANDING
the way in which the false belief is induced. In the change-of-location task, the child is shown
that an agent, e.g., Maxi, places an object, e.g., a chocolate bar, in location A (e.g., a green
cupboard). The child then sees the agent leave and she (the child) witnesses how another
character enters the scene and moves the object from location A to location B (e.g., the blue
cupboard). The relocation of the object has the effect that the agent now has a false, outdated,
belief about where her object is. The test question is posed at the moment when the agent returns:
“Where is Maxi going to look for his chocolate?”
In the unexpected-content task, an agent’s belief is not rendered false by any kind of
change or transformation occurring within the experiment. Instead, the agent (who is often the
child herself rather than a third person) is set up to hold a false belief by presenting her with a
conventional container, such as an M&M’s box that contains an unexpected object, e.g., hair
clips. The child is first asked what she believes the box contains (“M&M’s”) and is then shown
the counter-expected content. The test question is simply: “When you first saw the box, what did
you think was inside?” Another critical test question asks children about a third person’s
assumption regarding the content in the box, who did not witness the reveal. The question posed
is, “What will [name of the third person] think is in there?”
Studies consistently found that children first pass both of these tasks between ages 4 and
5 (Baron-Cohen et al., 1985; Wellman et al., 2001; Wimmer & Perner, 1983). Younger children
reliably give the wrong answer, claiming the agent will look for the object where it actually is
(change of location) or correctly identify, rather than misidentify the object (unexpected content).
A great deal of other theory of mind tasks with similar question-and-answer techniques likewise
indicate a critical turning point at age 4 to 5 (Flavell, et al., 1992; Gopnik & Astington, 1988;
Trautner et al., 2003; also see a meta-analysis in Wellman et al., 2001). For example, children
4
BELIEF UNDERSTANDING
come to understand that people can be deceived by misleading appearances, which is a specific
cause of false beliefs (see Flavell, 1989; Moll & Tomasello, 2012).
This first wave of empirical research culminated in the theory that belief understanding is
the product of a conceptual change occurring at age 4 to 5 years (Wellman & Woolley, 1990). On
this view, now known as the “late-onset view”, age 4 to 5 marks a threshold because this is when
children first come to comprehend how an agent’s represents an object or situation, and that this
mode of representation can conflict or clash with reality (Flavell, 1988; Gopnik, 1993; Wimmer
& Weichbold, 1994). Children now acquire a representational understanding of the mind, being
able to recognize, attribute, and reason about others’ belief states and subjective experiences.
1.3.2 The Second Wave: Looking Tasks and the Early-Onset View
The second wave of research was highly critical of the first, arguing that the above-cited
research program grossly underestimates younger children’s understanding of the mind and their
appreciation of beliefs. Part of the critique focused on the high linguistic demands placed on
children (Brown et al., 1996; Moore et al., 1990; Papafragou et al., 2007), and on the active
elicitation of responses (Scott et al., 2012) in the standard, direct tests. There was thus a push for
undemanding measures that involve no language use or response prompts. Baillargeon and
colleagues (2010; Onishi & Baillargeon, 2005; Scott & Baillargeon, 2009; Song & Baillargeon,
2008) used the so-called “violation of expectation” method to show that even infants represent
others’ beliefs. In their adaptation of the change-of-location task, infants observe scenes in which
the agent, upon returning to the scene, either acts in a way that is congruent (looking for an
object where she last saw it) or incongruent (looking for an object where she did not see it last)
with her beliefs. Longer looks are said to indicate that the agent’s action violated the child’s
expectations, leading to increased attention (Woodward, 1998). These and similar studies found
5
BELIEF UNDERSTANDING
that infants by age 15-month-old look longer at belief-incongruent than at belief-congruent
actions, suggesting that they registered the agent’s belief (Onishi & Baillargeon, 2005; Scott &
Baillargeon, 2009; Song & Baillargeon, 2008; Song et al., 2008; Surian et al., 2007; Träuble et
al., 2010). Helping and pointing studies also seem to indicate that infants in the second year of
life are keeping track of others’ beliefs (Buttelmann et al., 2009; Knudsen & Liszkowski, 2012a,
2012b). In an adaptation of a task by Clements and Perner (1994), Southgate et al. (2007) found
that 2-year-olds correctly anticipated where the agent in the change-of-location task would look
for her object as evidenced by looking toward the empty location (see also, Senju et al., 2011;
Surian & Geraci, 2012).
Most authors of these infancy studies defend an early-onset view, arguing that a theory of
mind is present from birth. Following Leslie (1992), Baillargeon and colleagues (2005, 2010)
suggest that a theory of mind is an innate processing system designed to interpret behaviors on
the basis of mental states. This system, which is said to occupy a distinct area in the brain
(involving the temporoparietal junction, see Saxe, 2006, for details) is “online” from the
beginning of life. As Onishi and Baillargeon (2005, pp. 256-257) write, “infants already possess
a representational theory of mind: They realize that others act on the basis of their beliefs and
that these beliefs are representations that may or may not ‘mirror reality’.”
One prominent problem with the early-onset view is that the empirical data on which it
rests are difficult to reproduce. Several of the looking time and anticipatory looking tests have
failed to replicate (Burnside et al., 2018; Dörrenberg et al., 2018; Kampis et al., 2021; Kulkarni
et al., 2018; Poulin-Dubois et al., 2018; Schuwerk et al., 2018), and, even if the data could be
reproduced, the interpretation of longer looks as reflecting violated expectations is questionable
6
BELIEF UNDERSTANDING
(Aslin, 2007). Therefore, the early-onset view and the data on which it is based stand on shakier
ground than does the late-onset view, whose results have been reliably reproduced for decades.
1.3.3 The Third Wave: Interactive Tasks and the “Middle Position”
Currently, there is a third wave of research that supports a middle position between the
early- and the late-onset views. This new position recognizes that while infants and toddlers
show some sensitivity to others’ belief states, they lack a full, explicit appreciation of beliefs and
other representational states until the age of 4 or 5. The position thus sits between the early-onset
and the late-onset view, arguing that toddlers are limited in their understanding of beliefs in
significant and predictable ways. In this dissertation, I propose that the first limit is children’s
dependency on tracking the agent’s prior experience (Ni et al., 2023), and the second postulated
limit is their struggle to understand different beliefs regarding a single, deceptive object (Low et
al., 2014; Low & Watts, 2013; Ni et al., 2023). Both of these limits will be introduced and
discussed later in more detail in this section.
The third wave of research uses, in addition to looking and verbal measures, a range of
innovative active and interactive methods (including motor tasks and expression tasks) in which
the child is in direct contact with the agent holding a false belief. The eclectic use of various
methods stems from the aim to determine more specifically what toddlers do and what they do
not yet understand about others’ mental states. Garnham and Perner (2001) had 3-year-olds place
a mat to provide a soft landing for an agent who was about to take one of two slides to fetch her
relocated object. Children promptly placed the mat where the agent thought the object was,
although it was no longer there. The same children, however, when asked where the agent would
come down the slide, were confidently giving the wrong answers–thus showing that they lacked
explicit knowledge of the agent’s false belief. Other studies also confirmed a dissociation
7
BELIEF UNDERSTANDING
between an early, implicit grasp, as measured by actions, and an explicit appreciation of the
agent’s false belief, as measured by children’s statements. Moll et al. (2016, 2017) used an
expression-based measure of early belief sensitivity with 2.5- and 3-year-olds. They found that
toddlers showed facial expressions of tension, such as lip biting and furrowed eyebrow, when the
misled agent approached unexpected reality–suggesting that the toddlers noticed the clash
between what the agent was expecting and what she would find. Again, however, children
reliably failed classic false belief tasks, with no association between their expressions and their
performance on the explicit measures.
To further determine what children younger than 4 to 5 years old understand and do not
understand about others’ representational states, Low and colleagues had toddlers witness how
agents interact with objects with dual identities, such as objects that appear to be one color from
one side but a different color from the other (Low & Watts, 2013), or ambiguous figures, such as
a figure that can be viewed as a duck or a rabbit (Low et al., 2014). Toddlers proved to be
sensitive (as indicated by differences in their looking time) to where an agent expected to find a
given object, but they failed to understand under which mode of presentation or aspectual shape
the agent would encounter the object. They lacked appreciation for the specific way in which the
agent viewed or construed a given object (e.g., as a rabbit or duck).
Furthermore, in another study using the expression-based measure, Ni et al. (2023)
directly addressed the two postulated limits of 3-year-olds’ theory of mind. In their study,
3-year-olds watched puppet shows where an agent’s original food item (e.g., a chocolate bar) was
replaced by an object that looked identical but was actually inedible (a chocolate-shaped eraser).
One group of children witnessed the agent’s prior interaction with the original food item, while
the other group only witnessed the agent focusing on and approaching the fake chocolate
8
BELIEF UNDERSTANDING
intending to fetch it. The results showed that only the group that tracked the agent’s prior
experience with the actual chocolate expressed suspense when the agent approached the inedible
look-alike. This suggests that they relied on tracking the agent’s prior experience to discern what
the agent expected to find. Moreover, in the second experiment of the study, even when tracking
experience, children’s responses did not differ whether the original food item (e.g., a chocolate
bar) was replaced with a look-alike or something completely different and non-resembling (e.g.,
an orange block). This suggests that 3-year-olds struggled to understand how deceptive
appearances can induce false assumptions or beliefs about the state of the world.
The resulting pattern of findings–with toddlers solving certain tasks but not others–has
led us and others to defend the middle position, according to which belief understanding
develops in steps over the first few years of life (Apperly & Butterfill, 2009; Moll, Ni, &
Stekeler, 2022; Perner & Roessler, 2010). One variant of the middle position is the two‐systems
account, which postulates two mindreading systems in humans: one is innate, efficient but
inflexible (System 1) and the other acquired, inefficient but flexible (Apperly & Butterfill, 2009;
Butterfill & Apperly, 2013). This account provides a nuanced, conciliatory, answer to the debate:
Children younger than 4 possess only System 1 because System 2 has yet to develop. Any
discriminatory responses by toddlers in looking time, anticipatory looking, and other implicit
tasks (e.g., those measuring manual actions or facial expressions) reflect the tracking of
encounters and expectations by System 1. Direct or standard false belief tasks, however, require
the meta-representational power (e.g., belief ascriptions) of System 2. These more demanding
tasks are first solved at 4 or 5 years old because this is when System 2, helped by advances in
language and executive functions, takes shape (Apperly et al., 2011; Qureshi et al., 2010).
9
BELIEF UNDERSTANDING
Another theory addressed the question from the perspective of teleology (Perner &
Roessler, 2010). This teleology-in-perspective account states that toddlers’ understanding of the
mind is at best shallow. When explaining others’ intentional actions (e.g., go to the store),
children under 4 give only objective reasons, provided by actual states of affairs (e.g., because
the store is open), but not subjective reasons, provided by subjective thoughts (e.g., because the
person believes the store is open). To explain why toddlers are nonetheless sensitive to belief
manipulations in indirect tests, the authors invoke the notion of “experiential record” (Perner et
al., 2007). Perner and Roessler (2012, p. 522) write that “infants are keen to register and record
what others did and did not perceptually track; that is, they keep an experiential record.” When
seeing an agent return to where she previously experienced an object, that record gets activated.
In the authors’ words, “the agent’s reappearance is likely to evoke these records and make the
infants think of what the agent had experienced.” In anticipatory looking studies, the experiential
record makes the child look to where the agent expects the object: “What the agent had
experienced (the object in its original location) corresponds to the content of the agent’s false
belief. Thinking of this content allows children to anticipate a possible action based on this
content.” (Perner & Roessler, 2012). Analogously, in looking-time experiments (e.g., He et al.,
2011; Kovács et al., 2010), infants look longer at false-belief than true-belief scenes because the
experiential record lets them vicariously experience the agent’s surprise, resulting in extended
gaze.
Moll et al. (2022) proposed that the development of belief understanding occurs in two
steps. The first step is practical and interactive, based on the ability to take perspectives. It
emerges in interactive situations where young children react directly to the actions of a person
acting on false assumptions. The second step is theoretical: at the age of 4 to 5 years, children
10
BELIEF UNDERSTANDING
develop the ability to reflect upon and form third-personal judgments about the perspectives of
others, enabling them to process conflicting beliefs. This two-step development of belief
understanding might help explain the signs of sensitivity to false belief observed in studies using
looking behavior measures, as well as in studies employing action-based measures where
children interact with the agent in a practical manner.
Given the necessity for evidence regarding what young children can and cannot do in
understanding others’ mental states, this dissertation aims to delineate the scope and limits of the
theory of mind in children just before the critical age threshold of 4 to 5 years. In doing so, we
seek to make a meaningful contribution to the continuing debate in the field.
1.4 This Dissertation’s Contribution to the Debate and Research on Theory of Mind
This dissertation project endorses the middle position and aims at better identifying the
scope and limits of toddlers’ nascent understanding of others’ belief states. As recently
articulated in a theoretical paper (Moll et al., 2022), we argue that toddlers have an early, implicit
grasp of the mind that we have called “practical” belief understanding. This understanding is
practical in the sense that it manifests exclusively in direct interaction with other people or in the
moment when false beliefs are acted upon by other people. Toddlers’ early understanding of
beliefs is thus “practical” as it is limited to practical or pragmatic interactions with others.
The first limitation in young children’s theory of mind understanding is their reliance on
tracking an agent’s experiences with an object or event, in anticipating the agent’s subsequent
actions or encounters (Perner & Roessler, 2012). The importance of tracking experience in
theory of mind understanding was briefly touched upon in previous studies (Rubio-Fernández,
2019; Rubio-Fernández & Geurts, 2016), where the availability of the agent’s perspective helped
3-year-olds correctly deduce an agent’s expectation. Ni et al. (2023) directly examined
11
BELIEF UNDERSTANDING
3-year-olds’ reliance on tracking experience by showing that they expressed suspense about the
agent’s unexpected encounter only after tracking the agent’s prior experiences. Building upon
these findings, this dissertation will take a step further to explore the detailed aspects of tracking
experience in the mindreading process.
Secondly, young children may lack a clear awareness of how others represent objects,
often failing to detect when others misidentify or misrepresent them. As introduced above,
children at age 3 do not yet fully understand that another person can hold a different, false belief
about the same object simultaneously, particularly when the object’s deceptive appearance or
tricky identity is the cause of the false belief (Low et al., 2014; Low & Watts, 2013; Moll &
Tomasello, 2012; Ni et al., 2023). Furthermore, they struggle to comprehend that another person
might, for instance, falsely assume that the morning star and the evening star are two distinct
stars (Perner et al., 2011; Sprung et al., 2007). Understanding that a single object associates with
two different or even conflicting beliefs simultaneously (as in the appearance-reality distinction
task) is more complex for children, compared to scenarios involving two objects, each associated
with a different belief (as in the change-of-location task).
1.5 The Affective Dimension of a Theory of Mind
This dissertation project is among the first to acknowledge and systematically explore the
affective dimension of belief understanding. False beliefs have often been treated as
affect-neutral, but erring or finding out that one is wrong can have profound affective or
emotional consequences (Scott, 2017). Consider Maxi’s reaction when realizing that his
chocolate bar is no longer in the green cupboard. He is going to be surprised and disappointed:
“Wait, where did my chocolate go?” Only a couple of studies have explored children’s grasp of
the nexus between beliefs and emotions (Moses & Flavell, 1990; Wellman & Banerjee, 1991).
12
BELIEF UNDERSTANDING
This small body of research showed elementary schoolers understand that surprise is a reaction
to having one’s beliefs violated, and that it is often one’s mental state, not reality itself, that
causes us to have certain emotions.
Children’s emotional expressions provide a valuable but underused window into their
understanding of the mind, especially for ages when children cannot yet articulate their
understanding. Expressions can thus help identify the extent of mindreading prior to the
threshold age of 4 to 5 that the late-onset view postulates. Humans express, both voluntarily and
involuntarily, a wealth of emotions and affective nuances through their faces. Comparative
research has revealed that humans are much more expressive than other animals (Kret et al.,
2020, see also Moll & Ni, in press). The human face has evolved features, including a
verticalized facial area, elongated eyes with white scleras, and red lips that enhance the face’s
expressiveness. In face-to-face interactions, humans facially express our psychological states to
one another–and even infants have a remarkable expressive repertoire (Parrott & Gleitman,
1989). In contrast with language, facial expressions are a more direct, natural, and non-symbolic
mode of “speaking one’s mind” (Bar-On, 2013, 2017).
In prior work, we capitalized both on the affective impact of false beliefs and on humans’
expressiveness by studying children’s facial expressions as they observe others act on account of
false beliefs (Moll et al. 2016, 2017; Ni et al., 2023). In these studies, we recorded and analyzed
children’s suspenseful expressions while witnessing others approach reality with false
expectations. Suspense is a feeling of tension that arises from foreseeing someone naively
approaching a situation that will turn out problematic for that person (Zillmann, 1981). In Moll et
al.’s (2016) study, children were shown, for example, how Cookie Monster was approaching a
box of cookies. In one condition (false belief condition), Cookie Monster had failed to see how
13
BELIEF UNDERSTANDING
another agent had removed most cookies from the box and thus returned to the box with false
expectations. In another condition (true belief condition), Cookie Monster had seen the removal,
and so he was approaching the box knowing exactly what he would find. The finding was that
children showed more tense expressions (e.g., lip bite, smirk, furrowed eyebrow): 2.5- and
3-year-olds expressed suspense as Cookie Monster moved towards the box, unaware that
someone had removed cookies from his box. These expressions indicated that children
anticipated Cookie Monster’s “rude awakening” when making contact with unexpected reality.
Building on the work of Moll et al. (2016, 2017), Ni et al. (2023) used the suspense
paradigm to study children’s understanding of the possibility that appearance and reality can
diverge and that people are led to misidentify objects due to these objects’ (deceptive)
appearances. Again, in line with the middle position, the authors found that 3-year-olds were
unable to discern how appearances induce false assumptions or beliefs. The study confirmed the
sensitivity of expression-based measures and included the number of expressions, in addition to
the presence of expressions, as an indicator of the tension children may sense.
In conclusion, expressions provide a powerful window for researchers to delve into
children’s understanding of the mind, as they offer an unspoken yet eloquent narrative of internal
cognitive processes. The studies conducted by Moll et al. (2016, 2017) and Ni et al. (2023)
represent only the beginning of a promising research avenue. There is a substantial need for
further investigations to harness the rich and nuanced information that children’s expressions can
provide about their cognitive and affective worlds.
1.6 Dissertation Overview
This dissertation reports three studies, each comprising two experiments, that aim to
explore the scope and limitations of young children’s theory of mind. The studies focus on the
14
BELIEF UNDERSTANDING
two key postulated limits in 3-year-olds’ belief understanding: 1) their need to track others’
object interactions leading up to a false-belief involving situations and 2) their inability to
understand how, absent such prior interactions, an agent will view or construe an object based on
its appearance. Study 1 (in Chapter 2), using an adapted version of the unexpected-content task
(Gopnik & Astington, 1988; Hogrefe et al., 1986), examined the role of tracking an agent’s
experiences—what they have and have not witnessed—in children’s ascriptions of false
expectations about the contents of a container to the agent. In Experiment 1 (N = 60), the
expressions of 3-year-olds were recorded as they watched a puppet show where an agent
approached a container whose original food content was secretly replaced with inedible items.
The container was either transparent, thus non-deceptive, or deceptive (e.g., a conventional
cookie box). The experiment also varied whether the children were able to track the agent’s prior
experience with the original content and the container. To contrast the expression-based measure
with the traditional verbal question-based measure, in Experiment 2 (N = 60), using the same
design, we verbally asked 3-year-olds what the agent believed was in the container. We predicted
that tracking experience would help children anticipate the agent’s false expectation (measured
through expressions) but not articulate the agent’s false belief (measured through verbal
questions).
Study 2 (in Chapter 3) further explored what experiential records are made of, i.e., what
kind of experience toddlers need to track to be able to ascribe beliefs or belief-like states
(“registrations”) to an agent. The hypothesis is that young children’s theory of mind skills are
rooted in their capacity to directly track others’ perceptual experiences with objects, as proposed
by Apperly & Butterfill (2009) and Perner & Roessler (2010). In two experiments (N = 120), we
measured 3-year-olds’ expressions as they watched an agent approaching a box whose content
15
BELIEF UNDERSTANDING
had, unbeknownst to the agent, been replaced. To test the hypothesis, the study varied the kinds
of interaction between the agent and the original object initially: children either directly
witnessed the agent perceptually interacting with the object, witnessed the agent receiving
non-experiential information, or had to infer the agent’s perceptual experience with the object.
We predicted that it is essential for 3-year-olds to directly track the agent’s perceptual
experiences with objects in order to anticipate the agent’s to-be-violated expectation.
Study 3 (in Chapter 4) focused on when and how children understand the relationship
between experience, beliefs, and actions. It introduced a novel “backward” reasoning approach,
in which participants witness the agent’s action and infer their previous experience. In two
experiments, 3- and 5-year-olds (N = 120) were shown stories modeled after the
change-of-location task (Experiment 1) and the unexpected-content task (Experiment 2). In
Experiment 1, an agent placed an object in Location A, which was later moved to Location B. It
was initially kept ambiguous whether or not the agent witnessed the relocation. Children then
observed the agent approaching either Location A or Location B and were asked whether the
agent was asleep or awake when the object was relocated. Similarly, in Experiment 2, it was kept
ambiguous whether the agent was asleep or awake during the demonstration of non-food
contents (e.g., forks) in a food container (e.g., cookie box). The agent then expressed her appetite
for, e.g., cookies, and either happily reached into the container or sadly walked away. Children
also completed the corresponding forward false belief tasks. We predicted that 5-year-olds would
be better at backward belief reasoning than 3-year-olds. We also predicted that children’s
performance in the backward reasoning tasks would be associated with their performance in the
classic forward belief reasoning tasks, which would suggest that children, between the ages of 3
and 5, develop a general understanding of how experience, belief, and action are interrelated.
16
BELIEF UNDERSTANDING
Taken together, this dissertation project focuses on a number of important but
understudied aspects of theory of mind and its development. We deliberately focused the
investigation on children ages 3 to 5 to examine the potential limitations exhibited by children in
this critical age range, both just below and just above the pivotal age of 4 years. This project
contributes to the current debate on theory of mind by revealing toddlers’ limits, exploring
backward reasoning skills, and blending prior methods to learn more about their reasoning in
theory of mind contexts.
17
BELIEF UNDERSTANDING
Chapter 2 Experience Tracking and Belief Understanding
Study 1 examined the two key limits in toddlers’ understanding of beliefs: (1) their
reliance on tracking experiences to reason about expectations, and (2) their lack of understanding
of deceptiveness and how it may induce false beliefs. Experiment 1 recorded children's
expressions as a measure of their sensitivity to an agent’s false expectations concerning the
content of a container. Experiment 2 assessed whether 3-year-olds could articulate the agent’s
belief in the same scenario. We predicted that children could anticipate the agent’s unexpected
encounter only when they had tracked the agent’s prior experience with a container and its
original content. This sensitivity was predicted to be detected through expressions, but toddlers
were not yet able to verbally articulate their understanding of the agent’s false belief.
2.1 Introduction
The change-of-location task and the unexpected-content task are often treated as
equivalent in measuring theory of mind because children younger than 4 usually fail both
(Bernard, et al., 2015; Grosse Wiesmann et al., 2017, 2018). However, interestingly, older
children did not always perform equally well in these two tasks (Grosse Wiesmann et al., 2017;
Holmes et al., 1996). For example, 4-year-olds were significantly above chance in the
change-of-location task and at chance level in the unexpected-content task (Grosse Wiesmann et
al., 2017). This suggests that the unexpected-content task might be more challenging for children
than the change-of-location task.
In this study, we therefore focused on the unexpected-content task and its specific
challenges. We propose two such challenges. First, we identify that a unique prerequisite for
passing the unexpected-content task is for children to understand that people–themselves
included–form assumptions based on the type of container they are encountering in this task.
18
BELIEF UNDERSTANDING
Children need to grasp how the container’s conventionality shapes agents’ expectations. Second,
toddlers generally rely on tracking others’ experiences with objects over time to arrive at an
understanding of their perspectives. This kind of tracking, however, is absent in the classic
unexpected-content task, where the agent forms a belief instantaneously based on the looks of
the container.
Previous studies have demonstrated that children under age 4 do not understand deceptive
appearances and how they impact belief formation (Flavell, 1986; Flavell et al., 1983; Low &
Watts, 2013; Moll & Tomasello, 2012; Ni et al., 2023). Specifically, toddlers fail to predict
another person’s belief about an object that appears to be one thing but is actually another
(Flavell et al., 1983), or a dual-faced object that looks to be one thing from one side but looks to
be something else from the other (Low & Watts, 2013). This consideration prompts the
hypothesis that in the classic unexpected-content task, children grapple with the deceptive nature
of conventional containers, which may be one of the reasons why they fail to understand that a
person would expect Smarties when there are candles inside a Smarties box. In this study, we
therefore compared scenarios in which the container is deceptive (because it suggests some
content other than what it contains) with ones in which the container is not deceptive
(transparent). We eliminate the deceptiveness by using transparent containers, allowing direct
visibility of the actual contents.
We further propose that for children to correctly reason about an agent’s assumption
about what is in the container, they rely on tracking the agent’s experiences with the container
earlier. A notable challenge in the classic unexpected-content task arises from the absence of this
tracking experience, which was mentioned by previous studies (Lewis, et al., 1994; Ni et al.,
2023; Rubio-Fernández, 2019). In this task, the agent has no prior engagement with the container
19
BELIEF UNDERSTANDING
or its content–she is simply taking a look at the container. Therefore, tracking the agent’s
experience or encounter may also help young children’s performance in understanding the
unexpected nature of the actual content caused by the deceptive look of the container.
It is worth noting that simply tracking the agent’s experience would not allow young
children to pass the classic unexpected-content task, i.e., give a verbal answer about what the
agent thinks is in the container. This is because young children are argued to possess a minimal
theory of mind (Butterfill & Apperly, 2013), and they do not yet understand deceptive
appearances and their effect on agents’ object representations (Ni et al., 2023). However,
previous studies have shown that children who are able to track the direct experience of the agent
with the object showed signs of their early implicit understanding, which can be revealed by
non-verbal measures, such as the expression-based measures (Moll et al., 2016, 2017; Ni et al.,
2023). This means that the effect of tracking experience may only manifest in implicit measures,
not in explicit measures using verbal questions.
2.2 Current Study
Taken together, the current study aimed to investigate how 3-year-old children relied on
tracking an agent’s prior experiences to anticipate the agent’s unexpected encounter in a puppet
show version of the unexpected-content task. We hypothesized that while 3-year-olds might not
explicitly understand the role of deceptiveness in creating false expectations for the agent, their
anticipation of the agent’s unexpected encounter could be detected through measuring
expressions. These expressions capture their spontaneous affective responses without
necessitating verbal questioning.
To test children’s reliance on tracking the agent’s prior experience with the container, we
created two groups: the Tracking Group and the No Tracking Group. Children in the Tracking
20
BELIEF UNDERSTANDING
Group witnessed the agent initially interacting directly with the food content in the container. In
contrast, children in the No Tracking Group did not have the opportunity to track the agent’s
prior experience; they started by being demonstrated the actual non-food content in the container.
Given that the container’s deceptiveness was the cause of the agent’s false expectation, we
established two conditions within each group: the Deceptive Condition, featuring a conventional
food container (e.g., a cookie box), and the Transparent Condition, where deceitfulness was
removed by using a transparent, see-through container.
Experiment 1 utilized expression measurement (Moll et al., 2016, 2017; Ni et al. 2023),
where 3-year-olds watched puppet shows, and their expressions were recorded as they observed
the agent approaching the container with either a false or true expectation. Experiment 2, in the
same design, used verbal questioning similar to the classic unexpected-content task. The study
received ethical approval from the Institutional Review Board (IRB) at the authors’ affiliated
university, ensuring compliance with all relevant ethical guidelines.
2.3 Experiment 1
Experiment 1 aimed to investigate how 3-year-old children anticipate unexpected
encounters in a puppet show using the expression measurement. As the children watched the
puppet show, their facial and bodily expressions were closely monitored, especially as they
observed the agent approaching a container that either held the expected or unexpected items.
This recorded their sensed tension as the agent neared the unexpected outcome. We predicted
that children in the Tracking Group would show more expressions in the Deceptive Condition
than in the Transparent Condition. However, we anticipated no such difference in the No
Tracking Group, based on our hypothesis that at age 3, children rely on tracking the agent’s
21
BELIEF UNDERSTANDING
experience to form their anticipatory reactions. The experiment was pre-registered on the Open
Science Framework (https://osf.io/hyb5j/?view_only=8fa36ef31c854328a37e0bb173491acf).
2.3.1 Methods
Participants
Children were recruited from social media, local preschools, and neighborhoods in the
Greater Los Angeles area. The sample size, N = 60, was determined through a priori power
analysis (F test in G*Power version 3.1), with alpha set at .05, a power of .80, and an effect size
of .30. One additional child was tested but excluded due to uncooperativeness. All participants
were native English speakers. The final sample included 60 (30 female) 3-year-olds (M = 38;26
months, range = 33;28–43;08 months). Ten percent were Latinx participants, and the racial
composition was 68% White, 13% Asian, 5% Black, 3% multiracial, 3% other (7% did not
disclose their racial category). Family household incomes varied widely, ranging from less than
$20,000 to over $120,000.
Materials
Videos of pre-recorded puppet shows were presented to children online. Each of the four
puppet show stories was presented in a wooden puppet theater (71x74x14cm), using a pair of
hand puppets (25-35 cm in height). As shown in Figure 1, each story contained a conventional
food container (e.g., a cookie box), several food (e.g., cookies) and non-food items (e.g., blocks).
A transparent cube-shaped container (8 cm in diameter) was used as the non-deceptive container.
22
BELIEF UNDERSTANDING
Figure 1
Materials Used in the Puppet Shows Broken Down by Story
Note. In the Cookies Story, the conventional container was a cookie box (20cm x 12cm x 7cm),
the food items were three cookies, and the non-food items used for replacement were three
colorful wooden blocks. In the Chips Story, the conventional container was a red Pringles potato
chips box (13cm high, 6.5 cm in diameter), the food items were four potato chips, and the
non-food items were four metal keys. In the Popcorn Story, the conventional container was a
yellow popcorn box (18cm high, 11.5 cm in diameter), the food items were a handful of popcorn,
and the non-food replacements were a handful of colorful puzzle pieces. In the Ice-Cream Story,
the conventional container was an ice-cream bar box (18cm x 15cm x 5cm), the food items were
three colorful ice-cream bars, and the non-food replacements were three black plastic forks.
Design
A mixed design was used, with group as between-subjects variable, and condition as
within-subjects variable. Each participant was randomly assigned to either the Tracking Group or
the No Tracking Group (“Tracking” means the participant was able to witness and track the
agent’s prior experience). Each participant watched a total of four stories (i.e., four trials), two of
23
BELIEF UNDERSTANDING
which were in the Deceptive Condition, and the other two were in the Transparent Condition.
Condition order and story order were counterbalanced.
Procedures
Parental consent was given via a secure electronic signature platform. Parents were then
invited to join a scheduled video call on Zoom. The session was run by a female experimenter
(E). Participants joined the video call using computers with a minimum screen size of 11 inches.
Parents were instructed to ensure their child was positioned at an arm’s distance from the screen.
To establish rapport, E initiated a brief warm-up conversation with the child. E then started the
screen share, and the parent was instructed to hide the video panel, ensuring that only the shared
screen was visible. E then started to play the puppet show videos following the pre-assigned
order.
Each story started with a bell ringing and curtain opening. As shown in Figure 2, in the
Tracking Group, Puppet A entered the stage from the left and greeted the child. Puppet A then
grabbed the container from the right end of the stage. In the Deceptive Condition, the container
was a conventional opaque food box (e.g., a cookie box), while in the Transparent Condition, the
container was a transparent box. Next, Puppet A took out the food items (e.g., cookies) from the
container, and held them in their hands to show to the participant saying, “These look good!”
Puppet A then put the food items back into the container and put the container back on the right
end of the stage. Puppet A excused themselves and left by saying “OK, I have got to go now. I
am going to come back and get them later. Bye!” Later, Puppet B appeared with some non-food
items, e.g., wooden blocks (“Look at what I brought!”), replaced the food items (“Let me put
these in here instead.”) and exited. After a 3-second interval, Puppet A returned, which was the
start of the test phase. Puppet A looked at the container, exclaiming “Ooh! Alright! I will go get
24
BELIEF UNDERSTANDING
them now!” and walked slowly towards the container. When Puppet A touched the container, the
bell rang, and the curtain shut, which marked the end of the test phase (lasting 20–23 s,
depending on the story).
In the No Tracking group, children watched similar puppet shows with the difference that
Puppet A did not have prior experience with the container. Instead, after the bell rang and the
curtains were open, Puppet B showed up, fetched the container, and took everything out of the
container. Puppet B then showed the participant that there were no food items in the container,
only the non-food items. After Puppet B left, Puppet A entered the stage, looked at the container,
expressed the intent to get the container, and then walked slowly towards the container.
After watching all four puppet show stories, which lasted around 15 minutes in total, E
concluded the session and debriefed with the parent. Each family who participated in the
experiment received a $15 gift card to compensate for their time.
25
BELIEF UNDERSTANDING
Figure 2
Procedural Steps in Experiment 1
Note. The upper panel illustrates the procedural steps for the Tracking Group, while the lower
panel depicts those for the No Tracking Group. Each line represents one condition, with the story
progressing temporally from left to right.
Coding
An independent coder, unaware of the group or experimental condition, coded children’s
expressions in the test phase of each trial. Using a modified version of Moll et al.’s (2016)
coding scheme, the coder judged how many expressions each child showed in the test phase (see
Appendix) and noted the time of each expression. A baseline of expressive behavior was
obtained from the initial 10s of the first puppet story that the child received, and expressions in
the test phase were only coded if they were absent during that baseline. This was to ensure that
the expressions observed in the test phase were a response to the tension the participant sensed,
26
BELIEF UNDERSTANDING
rather than the participant’s habitual expressions. Consistent with prior studies (Moll et al., 2016,
2017; Ni et al., 2023), common expressions included facial expressions like brow-raising,
lip-biting, smirking, and lip tightening, as well as bodily movements such as hand-raising and
shoulder-shrugging. A second coder, also unaware of condition and group, independently coded
the expressions of 25% of the sample. Inter-rater reliability was high, Kappa = .73, and any
disagreements were resolved through discussion.
2.3.2 Results
All analyses were performed using R (version 4.0.2). There were four trials in which the
child participant did not finish watching the story; these were excluded from the data analysis.
We ran generalized linear mixed-effects (GLME) models and found no significant effects of
gender, age in days, racial/ethnicity, story, or story order (ps > .08) on the number of expressions.
To examine the difference across conditions and groups, we ran a GLME model with
condition (contrast coded: 1 for the Deceptive Condition, -1 for the Transparent Condition) and
group (contrast coded: 1 for the Tracking Group, -1 for the No Tracking Group) as independent
variables and the number of expressions as dependent variable. This model adjusted for the
interdependence within individual participants. As shown in Figure 3, the main effect of
condition was significant, b = .23, se = .08, p < .01, as was the main effect of group, b = .39, se =
.09, p < .001. These indicated that overall children showed more suspenseful expressions when
they saw the agent getting close to the conventional opaque container than the transparent
container, as well as when they were able to track the agent’s prior experience. Furthermore, the
interaction between condition and group was also significant, b = .19, se = .08, p < .05. A post
hoc analysis revealed that only in the Tracking Group was there a difference between the
Deceptive and Transparent Conditions, b = .96, se = .12, p < .001. In the No Tracking Group, the
27
BELIEF UNDERSTANDING
expressions did not differ between conditions, b = .20, se = .12, p = .19. This suggested that
children could anticipate the agent’s subsequent unexpected encounter caused by the deceptive
container only when they were able to track the agent’s prior experiences.
Figure 3
The Number of Expressions Grouped by Condition and Group
Note. Error bars on the graph indicate the standard error for each category.
An exploratory analysis of the timing of the expressions was conducted to explore
whether the tension sensed by the participants could be reflected in the time when they expressed
suspense. Specifically, in the Deceptive Condition, since the content was not visible, Puppet A’s
expectations were retained until the end, while in the Transparent Condition, Puppet A, upon
returning, would recognize the change in the container’s contents, thus not holding any false
expectation. A GLME model was conducted with the time (in seconds) of the coded expressions’
appearance as outcome variable and condition and group, both contrast-coded, as independent
variables. The results revealed a significant interaction between group and condition, b = 1.35, se
= .49, p < .01, while the main effect of either group (b = .58, se = .49, p = .24) or condition (b = -
28
BELIEF UNDERSTANDING
.11, se = .49, p = .83) was not significant (see Figure 4). A post hoc analysis showed that, in the
Tracking Group, the timing of expressions in the Deceptive Condition (mean = 12.39 s, variance
= 74.15) was significantly later than the timing of expressions in the Transparent Condition
(mean = 9.91, variance = 53.38), b = 10.00, se = .76, p < .001. The difference in timing between
the two conditions in the No Tracking Group was not significant, b = - 1.46, se = .76, p = .11.
These results indicate that when children were able to track Puppet A’s prior interaction with the
food items from the container before replacement, they were more likely to sense the tension
building up as Puppet A approached the unexpected outcome. This further indicated their
anticipation of Puppet A’s upcoming actions or responses caused by Puppet A’s false expectation.
29
BELIEF UNDERSTANDING
Figure 4
The Timing of Expressions Grouped by Condition and Group
Note. The x-axis represents the timing of expressions in seconds, with “0” indicating the start of
the test phase. The scatter dots denote the timing of all expressions for each specific condition
across the groups. The whiskers extend from the box to depict the range of data outside the
middle 50%, while the line within each box signifies the median timing of expressions.
2.3.3 Discussion
In this experiment, we found that children showed more expressions in the Deceptive
Condition, where the container was a conventional opaque food box with deceptiveness that
could lead to false expectations, as opposed to the Transparent Condition with a see-through
container without such deceptiveness. The difference between conditions was limited to the
Tracking Group, indicating that tracking an agent’s experience is necessary for detecting that
agent’s expectations. It is by tracking another’s experience that young children, under age 4,
30
BELIEF UNDERSTANDING
come to understand that person’s expectation as well as the surprise, disappointment, and other
affective implications should that expectation turn out to be false
It is noteworthy that expressions were not limited to the Deceptive Condition, but that
some expressions were also observed in the Transparent Condition. To explain their presence, it
is possible that the children were curious to see how the agent would react to the new content.
They might also have anticipated the agent’s excitement upon indicating their intent to retrieve
the container. Such reactions were likely present in the Deceptive Condition as well; therefore, it
was in comparing the Deceptive and Transparent Conditions that we truly discerned signs of
children attributing a belief-like state to the agent and anticipating a violation of the agent’s
expectations.
As have previous studies using this method (Moll et al., 2016, 2017; Ni et al., 2023), the
results from this experiment confirmed the effectiveness of the expression-based measure. While
verbal measures reliably indicate poor performance on classic false belief tasks, the
expression-based technique indicates belief sensitivity by around age 3. However, in the spirit of
the middle position laid out above, the expression method detects important blind spots and
limits in children’s early theory of mind capacities. In this experiment, the pattern of children’s
expressions suggests that although 3-year-olds are able to detect others’ expectations, these are
limited to expectations grounded in prior, perceptual experience that the child was able to track.
Measuring expressions thus proved effective in capturing the subtle nuances of children’s
non-verbal responses, particularly in scenarios where verbal articulation may not adequately
reflect their internal states.
The results of the current experiment shed light on the significant role that tracking an
agent’s experience plays in helping children understand potential violations of the agent’s
31
BELIEF UNDERSTANDING
expectations. Notably, this experiment highlighted that when having the chance to track an
agent’s interactions with objects across time, children demonstrate a better grasp of the agent’s
forthcoming unexpected encounter. One important question that remains is whether
experience-tracking also enhances children’s explicit understanding of such scenarios, as
indicated by their verbal statements about what the returning agent expects to find in the
container This open question was addressed in Experiment 2, in which we asked whether
children’s experience tracking not only supports their implicit grasp of beliefs, as measured by
expressions, but also their explicit comprehension of beliefs, as measured by verbal statements.
2.4 Experiment 2
In this experiment, we used the same design of Experiment 1 varying group (Tracking
and No Tracking Groups) and condition (Deceptive and Transparent Conditions). However, this
time, we did not measure expressions as an indicator of an implicit grasp of beliefs but, instead,
verbal answers to questions as a sign of children’s explicit knowledge about beliefs. As the
puppet show story reached a point where the agent returned and stated her intention to get the
container, children were asked to articulate the agent’s beliefs (“What does Emma think is in the
box?”) To replicate the classic unexpected-content task, control questions were asked in between
parts of the puppet show story. An initial response question, “What do you think is in there?”
was asked as children first saw the container on screen. After the original food content was
shown, an original content question was posed: “Can you tell me what is in there now?”
Following the replacement of the food items with non-food items, an actual content question was
asked: “Can you tell me what is in there now?” We hypothesized that in the Deceptive Condition,
children would give false answers to the questions about the agent’s false belief, resulting in
lower accuracy compared to the Transparent Condition. This pattern was anticipated across both
32
BELIEF UNDERSTANDING
the Tracking and No Tracking groups, because while tracking the agent’s experience aids in
understanding true or false expectations, it may not necessarily manifest in the verbal articulation
of these insights. The study was pre-registered on the Open Science Framework
(https://osf.io/z9xgb/?view_only=6a25c0ad61d6438284f4c01ae9e95a19) to ensure
methodological transparency.
2.4.1 Methods
Participants
The sample size was the same as in Experiment 1, N = 60. A total of 63 three-year-olds
were recruited from social media, local preschools, and neighborhoods in the Greater Los
Angeles area. The data of three participants were excluded due to uncooperativeness.
All participants in the final sample were neurotypical and native English speakers. The
final sample included 60 (30 female) 3-year-olds (M = 38;23 months, range = 34;02-43;04
months). The sample’s ethnic composition included 15% Hispanic participants. The racial
composition was 63% White, 8% Asian, 3% Black, 10% multi-racial, 3% other, and 12% did not
disclose their race. Family household incomes within the sample varied widely, ranging from
less than $20,000 to over $120,000.
Materials
The materials used in this experiment were identical to those in Experiment 1.
Design
We adopted a mixed design identical to that of Experiment 1, with group (Tracking vs.
No Tracking) as between-subjects variable, and condition (Deceptive vs. Transparent) as
within-subjects variable. Each child received two stories within each condition.
Procedure
33
BELIEF UNDERSTANDING
The preparatory procedures in this experiment, including obtaining informed consent,
setting up the Zoom call, arranging the children’s seating, and conducting the warm-up, were
identical to those in Experiment 1.
The puppet show videos used here were mostly identical to those in the previous
experiment, with verbal control questions inserted in between parts of the puppet show story and
a verbal test question at the end. As shown in Figure 5, each story began with a picture of a
container (e.g., a cookie box) on the screen, prompting the child with the initial response
question, “Look! What do you think is in there?” After the child responded, E proceeded to play
the first part of the puppet show, saying, “Okay, now let’s put it here. Look who’s coming!”
In the Tracking Group, similar to the previous experiment, Puppet A entered the stage
and approached the container—either a conventional opaque food box (e.g., a cookie box) in the
Deceptive Condition, or a transparent box in the Transparent Condition. Next, Puppet A
interacted with the food items (e.g., cookies) and then left, just as in Experiment 1. Following
this, the picture of the container reappeared on the screen, and E asked the original content
question: “Can you tell me what is in there now?” After the child’s response, E said, “Okay, now
let’s see what will happen next,” and played the next part of the story. In this part, identical to
Experiment 1, Puppet B appeared and replaced the food items with non-food items. After Puppet
B’s exit, E showed the child the picture of the container once again and asked the actual content
question: “Can you tell me what is in there now?” Next, Puppet A returned saying, “Hi, I’m
back. Ooh! Alright! I will go get them now!” The puppet show ends as Puppet A began to
approach the container, and E posed the test question: “What does [Puppet A’s name] think is in
there?”
34
BELIEF UNDERSTANDING
In the No Tracking Group, children watched similar puppet shows, with the key
difference that Puppet A did not have prior experience with the container, and Puppet B opened
the box, took out the unexpected, non-food content, and held them in her hands saying to the
child “These look good!” Thus, for the No Tracking Group, only the initial response question
and the actual content question were asked, followed by the test question.
The session again lasted approximately 15 minutes. The participant’s family received a
$15 gift card to compensate for their time.
Figure 5
Procedural Steps in Experiment 2
Note. The upper panel illustrates the procedural steps for the Tracking Group, and the lower
panel depicts those for the No Tracking Group. Each line represents one condition, with the story
progressing temporally from left to right. “Initial Response Q”, “Original Content Q”, and
“Actual Content Q” represent the control questions asked at different times in the story. “Test Q,”
indicated in red text, denotes the test question asked at the end of the story.
35
BELIEF UNDERSTANDING
Coding and Scoring
Participants’ answers to the control questions (including the initial response question, the
original content question, and the actual content question) and test questions were scored by a
rater who was ignorant of the experimental condition or group. In the Deceptive Condition, the
correct response to the test question is the food items (e.g., “cookies” in the Cookies Story),
while in the Transparent Condition, the correct response is the non-food items (e.g., “blocks”).
Children received a score of 1 for a correctly answered question and a score of 0 if they
answered incorrectly or indicated that they did not know. Divergence from original labels was
tolerated; e.g., “popsicles”, or “ice bars” were coded as “ice cream” in the Ice Cream story. A
second coder, also unaware of the condition and group assignment, independently coded the
responses of 25% of the sample. Inter-rater reliability was perfect (Kappa = 1.00) for all the
control and test questions.
2.4.2 Results
Logistic generalized linear mixed-effects (GLME) models were conducted to assess the
potential effects of factors not related to our primary focus on participants’ scores in the test
questions. The results showed that there were no significant effects of age, gender, trial, or story,
ps > .11.
Trials from the Tracking Group were excluded if a child answered at least one control
question incorrectly, leaving 88 out of 120 trials after exclusion. In the No Tracking Group, trials
were excluded if a participant failed to identify the non-food contents of the container in
answering the actual content question, leaving 99 out of 120 trials after exclusion. In total, 187
trials from 59 participants were included in the subsequent data analyses.
36
BELIEF UNDERSTANDING
A logistic GLME model was conducted with group and condition as independent
variables (both contrast-coded) and children’s test question scores as outcome variable. As
depicted in Figure 6, the model revealed a significant main effect of condition, with children
answering the test question correctly more often in the Transparent than in the Deceptive
Condition, b = -3.47, se = .77, p < .001. There was no effect of group, b =.14, se = .70, p = .83,
nor an interaction effect between group and condition, b =- .77, se = .92, p = .42. This suggests
that being able to track the agent’s experience did not aid children in correctly identifying what
the agent falsely believed to be in the container. In other words, tracking experiences did not
support the explicit acknowledgment of the agent’s false beliefs.
Figure 6
Correct Response Rate to Test Question Broken Down by Condition and Group
Note. The response rate for each individual was calculated by dividing the number of correct
answers by the total number of trials. Error bars on the graph indicate the standard error for each
category.
37
BELIEF UNDERSTANDING
Among the 53 trials in which participants failed to recognize the change of content (for
the Tracking Group) or failed to identify the non-food content (for the No Tracking Group), only
3 trials were in the Transparent Condition and the remaining 50 trials were in the Deceptive
Condition. In 41 out of these 50 trials, participants responded by naming the food items in
response to the actual content question after the replacement or demonstration. For example, they
continued to answer “cookies” as the actual content of the container, even after the cookies had
been replaced with blocks (for the Tracking Group) or after Puppet B had demonstrated that
there were only blocks in the cookie box (for the No Tracking Group). It was possible that the
participants in these trials may not have been paying close attention to the replacement, or they
may have quickly forgotten about the new non-food items. However, the results suggested that it
was more possible that the conventional containers were highly deceptive–even after children
witnessed the change or reveal of content, they may have been influenced to answer based on the
appearance of the container when being asked about “what is in there now” as they were viewing
the container (without seeing the content) on the screen.
2.4.3 Discussion
We found that 3-year-olds were unable to correctly answer what the agent believed to be
in the container when its content conflicted with the agent’s expectation. This was the case
regardless of whether the children had witnessed the agent originally experiencing the original
content of the container or not. The ability to track the agent’s prior experience did not enhance
the accuracy of children’s identification of the agent’s belief, suggesting that while tracking
experiences is essential to implicit belief understanding, it does not support explicit belief
understanding.
38
BELIEF UNDERSTANDING
The design of this experiment, particularly the Deceptive Condition for the No Tracking
Group, closely mirrored the setup of the classic unexpected-content task. In such tasks, children
do not have the opportunity to track prior experiences the agent had with the content and the
container, and the container’s deceptive appearance due to its mismatch with the actual content
poses additional challenges. Our findings align with those from classic studies (Hogrefe et al.,
1986), demonstrating that children under age 4 struggle with tasks involving deceptive
appearances, suggesting a developmental stage where reality or their own belief heavily
influences their reasoning about another person’s different beliefs.
An interesting observation from this experiment is that, on several occasions, even after
children were shown that the container held non-food items (e.g., blocks in a cookie box), they
still answered with the food item when asked about the container’s contents. This response,
based on the appearance of the box, which was also what they viewed on the screen at that
moment, indicates the strong influence of the container’s appearance. Although the trials in
which children failed this control question were excluded from the analysis, this pattern
nonetheless highlights the high deceptiveness of conventional containers–some children were
misled again even after being shown the reality. This phenomenon might also contribute to
explaining why the classic unexpected-content task is particularly challenging for young
children.
A potential limitation of this experiment was the use of a narrative format in the puppet
show, along with verbal questioning, which could potentially disrupt the children’s immersive
experience. However, previous research (Lewis et al., 1994) indicates that narrative formats are
effective in engaging children and facilitating understanding. Thus, while this format might have
39
BELIEF UNDERSTANDING
influenced the children’s responses, it is unlikely to have compromised the validity of the
experiment’s findings.
2.5 General Discussion
How could children anticipate that an agent’s expectation will be violated when the actual
content of a container does not match its appearance? In two experiments, we investigated
whether 3-year-old children need to rely on tracking the prior agent-object interaction to
successfully anticipate the agent’s forthcoming unexpected encounter. Experiment 1 revealed
that children showed more expressions in response to the agent having a false expectation about
a conventional food container, but this heightened response was contingent on the children
having previously witnessed the agent’s earlier interaction with the container and its contents.
Experiment 2 used children’s verbal responses, instead of measuring expression, and found that
even when children tracked the agent’s initial interaction, they still failed to verbally answer the
question about the agent’s false belief. Collectively, these experiments demonstrate that while
tracking an agent’s experiences is necessary for 3-year-olds to form judgments about the agent’s
expectations, which manifest in their spontaneous emotional reactions, their understanding is still
limited: in the case where the container is deceptive, 3-year-olds are unable to explicitly
articulate the agent’s false belief.
The current study thus supports the middle position in the debate of theory of mind
development, suggesting that 3-year-olds are sensitive to others’ belief states, such as
expectations, that are incongruent with reality, yet they do not fully or explicitly understand
beliefs (Apperly & Butterfill, 2009; Moll et al., 2022). It provides more evidence to determine
the degree to which children understand how changes in the world affect an agent’s perspective
on a situation, particularly in 3-year-olds who are below the commonly recognized age threshold
40
BELIEF UNDERSTANDING
(4 to 5 years) for mental state understanding. More importantly, the current study identifies a key
limit in the mental state reasoning process of 3-year-olds–the necessity to track another person’s
experience with an object to attribute a mental state and reason about the person’s expectations.
This is consistent with the previous research that revealed 3-year-olds’ reliance on tracking
experiences (Ni et al., 2023). Another limit identified in the current study within the context of
the unexpected-content task is the deceptiveness of the container. When the container was a
transparent box and thus no longer deceptive, children demonstrated significantly higher
accuracy in their verbal responses, and they no longer sensed the tension as the agent approached
the container. This finding aligns with previous studies on the appearance-reality distinction
using deceptive objects, such as a chocolate-shaped eraser (Ni et al., 2023). To navigate the
deceptiveness of the container and understand how it leads to another person’s false belief,
children need to possess a mature ability to manage two conflicting perspectives simultaneously
and suppress one perspective while considering another (Moll & Tomasello, 2012).
The expression-based measure used in Experiment 1 and the verbal measure in
Experiment 2 tap into different forms of belief understanding. Children’s expressions, as an
anticipatory reaction embedded within another’s ongoing activity (Moll et al., 2022), indicate the
tension they felt while anticipating the agent’s forthcoming encounter with the unexpected
reality. Anticipation, in this context, is a simpler and more direct form of belief reasoning, one
that is supported by the fast-acting System 1 of the two mindreading systems (Apperly &
Butterfill, 2009; Ni et al., 2023). When given the chance to track agent-object relations,
3-year-olds were able to implicitly show signs of recognizing the difference between the
scenarios in which the agent acted following a false expectation versus a true expectation.
However, since System 2, which allows for more flexible reasoning, is not yet fully developed in
41
BELIEF UNDERSTANDING
3-year-olds, they lack the awareness that the agent was misled by the conventional container.
Thus, they could not explicitly articulate the agent’s false belief, which is a more theoretical and
complex level of belief understanding. The low performance in explicitly articulating false
beliefs in Experiment 2 also aligns with earlier findings that indicate a dissociation between
implicit, non-verbal tasks and classic verbal tasks (Clements & Perner, 1994; Moll et al., 2016).
Taken together, this study demonstrated that children under 4 years old, when observing
an agent naively approaching a conventional container with surprising contents, depend on
tracking the agent’s previous interactions to anticipate the agent’s forthcoming unexpected
encounter. At this age, they are not yet explicitly aware of how deceptive appearances of the
container can shape perceptions, and in turn, influence beliefs and actions. This finding also
opens new avenues for using expression-based measures, which are more sensitive than explicit,
verbal question-based measures, in understanding the developmental trajectory of theory of
mind.
42
BELIEF UNDERSTANDING
Chapter 3 Experiential Records: What Are They Made of?
To further explore what kinds of experience toddlers need to witness in order to attribute
informational states to an agent, in Study 2, we manipulated the initial interactions between an
agent and an object in a box. Experiment 1 compared the cases when children directly witnessed
the agent interact with the object perceptually and when children saw the agent learn about the
object’s existence through someone else’s testimonial information. Experiment 2 added the
condition where the agent perceptually interacted with the object but children had to infer what
the agent experienced due to limited visual access. We predicted that children need to directly
witness the agent’s perceptual interaction with the object in building experiential records.
3.1 Introduction
Study 1 (in Chapter 2) demonstrated that 3-year-olds need to track an agent’s experiences
to ascribe beliefs or belief-like states to them and anticipate the agent’s subsequent actions or
reactions. However, the role of experience tracking in understanding mental states remains
insufficiently clear and has not been systematically explored. A key question arises: What kinds
of experience do toddlers need to witness in order to attribute informational states to an agent?
Accounts that take the middle position in the theory of mind debate, such as the
two-systems account (Apperly & Butterfill, 2009) and the teleology-in-perspective account
(Perner & Roessler, 2012), agree on that children have to directly perceive an agent-object
interaction to be able to represent the agent as expecting that object thereafter. They cannot
ascribe beliefs or other subjective states and “get by” in indirect tests by tracking others’
perceptual experiences with objects.
Here, we spell out and test two of the claims that we think are implied by accounts in the
middle position. One claim is that children under 4 should not be able to represent an agent as
43
BELIEF UNDERSTANDING
expecting an object after she learned about it through testimony. This is because in testimony,
there is no perceptual contact with an object for the child to record. Toddlers should therefore not
pick up on others’ expectations that are grounded in testimony instead of perception. Another
claim that both accounts seem to endorse is that children have to directly perceive an
agent-object interaction to be able to represent the agent as expecting that object thereafter.
Seeing an agent encounter an object, with both parts or relata visible to the child (child sees
agent perceiving object) should be required for children to represent the agent as expecting the
object. If this is so, then children should fail to represent an agent’s object expectation when,
instead of directly seeing the encounter, children have to infer what the agent encountered. Think
of a person handling an object while it is inside an opaque container, so that the onlooking child
sees the agent interact with something she only gets to determine at a later time (when the object
is removed and replaced with a different one by another agent). It is our understanding that both
accounts would predict that the young children do not view the agent as expecting the object in
this situation, because they did not get to witness the whole encounter and was unable to record
perceptual experiences.
3.2 Current Study
In two experiments, we sought to determine the role of experience tracking in toddlers’
understanding of others’ expectations. We used the expression-based measure (Moll et al., 2016,
2017; Ni et al., 2023), in which children facially and bodily express tension when noticing that
an agent’s expectation is about to be violated. We focused on 3-year-olds because children at this
age, while reliably lacking full belief understanding (Baron-Cohen et al., 1985; Hogrefe et al.,
1986; Wimmer & Perner, 1983), are highly responsive in the expression-based measure.
44
BELIEF UNDERSTANDING
Children watched puppet shows in which an agent, depending on condition, learned that a
box contained a food item (e.g., blueberries) either by perceptually experiencing it or by
receiving testimonial information. The food was then swapped for an inedible object in the
agent’s absence, and children’s expressions were recorded as the agent returned. In Experiment
1, the same children received both conditions (Experience Condition and Information
Condition), and expressions were compared across conditions. In Experiment 2, a third
condition, the Inferred Experience Condition, was added in which children did not directly
witness the agent-object encounter: they instead saw the agent peeking and grasping inside a box
the content of which was only later revealed to the children. In this condition, children did not
directly perceive (as they did in the Experience Condition of Experiment 1 and the Direct
Experience Condition of Experiment 2) but had to retroactively infer the agent’s experience of
the object.
3.3 Experiment 1
The experiment was pre-registered on the Open Science Framework
(https://osf.io/uypfr/?view_only=a3b9e963a04949049035249e44361cbb).
3.3.1 Methods
Participants
An a priori power analysis with alpha set at .05 and a power of .80 (G*Power version 3.1)
showed that a sample of N = 30 was required to detect an effect with a size of .30. The final
sample included 30 (15 female) 3-year-olds (M = 37;07, range = 33;0–44;10 months). A total of
31 children were recruited from social media, local preschools, and neighborhoods in the Greater
Los Angeles area, and one of them was tested but excluded due to uncooperativeness. All
participants were neurotypical, native English speakers. Ten % were Latinx participants, and the
45
BELIEF UNDERSTANDING
racial composition was 70% White, 10% Asian, 3% Black, 10% multiracial, 7% other. Family
household incomes ranged widely (from < $20,000 to over $120,000), indicating various
socioeconomic backgrounds.
Materials & Design
Participants used computers or tablets with a minimum screen size of 11 inches. A
wooden puppet theater (71x74x14cm) and a different pair of hand puppets (25-35cm in height)
were used for four different pre-recorded puppet shows. Stimuli were chocolates and keys
(Chocolate Story), cookies and paper clips (Cookies Story), crackers and wooden blocks
(Cracker Story), and blueberries and puzzle pieces (Blueberry Story), respectively. Four opaque
cylindrical boxes (10 cm in height, 9 cm in diameter) of various colors were used as containers in
each story. A within-subjects design was applied, with children receiving two stories of the
Experience Condition and two stories of the Information Condition, presented in the order EIIE,
IEEI, EIEI, or IEIE (“E” here represents the Experience Condition and “I” represents the
Information Condition). Story order was counterbalanced.
Procedure
Contingent on parental consent, the child and parent joined a recorded Zoom session. The
parent was asked to enter full-screen mode, minimize video panels, and avoid interference. The
child sat at arm’s length (approximately 50cm) in front of the screen. After introducing herself,
the experimenter (E) shared her screen and started the puppet shows.
Stories in the Experience Condition began with Puppet A entering the stage from the left
and greeting the child (see Figure 7). Puppet A approached the box toward the right end of the
stage and took out the food item (e.g., blueberries) from the container. She uttered, e.g., “Hmm,
let me have some!”, tasted the food, and placed it back in the box. Puppet A then left, saying “I
46
BELIEF UNDERSTANDING
have got to go now. I will be back soon, bye!” During Puppet A’s absence, Puppet B appeared
with, e.g., wooden blocks (“Look at what I brought!”) from the right, replaced the food (“Let me
put these in here instead.”) and exited right. Puppet A returned with the intent to retrieve the
content, exclaiming “Let me go get them now.” She approached the box, picked it up, and left.
The test phase, which began with Puppet A’s return and ended with her exit, lasted
approximately 23s.
The Information Condition differed in that after Puppet A entered from the left, Puppet C
appeared from the right corner of the stage. Puppet A and Puppet C stood facing one another
with the closed box between them. Puppet C then told Puppet A what the box contained, e.g.,
“There are some blueberries in here”. Puppet A uttered “Okay” before taking leave using the
same words that she used in the Experience Condition. Puppet A and Puppet C then left the stage
by exiting left (Puppet A) or right (Puppet C).
Figure 7
Procedural Steps of Puppet Show Story in Experiment 1
Note. The story progresses temporally from left to right. Puppet A perceptually experiences (top
left, Experience Condition) or is informed (bottom left, Information Condition) by Puppet C that
a box contains food. Puppet A leaves, and Puppet B appears and replaces the food with an
inedible object. Later, Puppet A returns.
47
BELIEF UNDERSTANDING
Coding and Scoring
A coder, who was ignorant of the condition, coded children’s expressions in the test
phase of each trial. Using a modified version of Moll et al.’s (2016) coding scheme, the coder
judged the expressions a child showed in the test phase (see Appendix). A baseline of expressive
behavior was obtained from the initial 10s of the first puppet story that the child received;
expressions in the test phase were only coded if they were absent during baseline. As in prior
studies (Moll et al., 2016, 2017; Ni et al., 2023), common expressions included brow-raising,
lip-biting, and lip-tightening. A second coder, also unaware of the condition, independently
coded the expressions of 25% of the sample. Inter-rater reliability was good, Kappa = .70, with
disagreements being resolved through discussion.
3.3.2 Results
Generalized linear mixed-effects (GLME) models detected no effects of story, condition
order, or trial number on the number of children’s expressions, ps > .38. To test whether the
number of expressions differed by condition, a GLME model with condition (contrast coded) as
within-subject variable was run. As shown in Figure 8, children showed significantly more
expressions in the Experience Condition than in the Information Condition, b = .26, se = .003, p
<. 001. Children thus reacted with greater tension when an agent approached a container whose
swapped-out content she herself had experienced than when the agent had learned about the
original content through testimony.
48
BELIEF UNDERSTANDING
Figure 8
Number of Expressions Broken Down by Condition
Note. Error bars on the graph indicate the standard error for each category.
3.3.3 Discussion
The results are consistent with the shallow mindreading account, which states that
toddlers need to see agents directly interacting with objects to view these agents as expecting
these objects. An implication is that toddlers would not register an agent’s expectation resulting
from testimony. This is supported by our finding that children reacted with more tension after the
agent had engaged with the original, edible but not when the agent had received testimony about
the edible object.
However, these data lack volume and conclusiveness. One issue is that tracking
experiences might be only superior to tracking information exchanges, without it being
necessary that toddlers have records of others’ perceptual experiences. The experiment does not
49
BELIEF UNDERSTANDING
show that experiential records are essential in the sense that not witnessing others’ experiences
leaves toddlers unable to detect others’ expectations. This problem was addressed in the next
experiment by adding control trials in which the original content was not replaced.
Another problem concerns potential differences in the agent’s certainty about and desire
for what the box originally contained. In the Experience Condition, the agent saw, touched, and
tasted the object, thus having knowledge from three senses. In the Information Condition, the
agent only heard (single sense modality) about the object, potentially leaving her less certain
about what was in the container. Also, in stating “Hmm, let me have some” the agent expressed a
desire for the original object in the Experience Condition that was not matched in the
Information Condition. Although these confounds were not entirely removed (e.g., we saw no
way to match the sense modalities in the Experience and Testimony Conditions on multimodal
input), the issues were addressed by making small changes to existing conditions and adding a
third condition.
The next experiment also examined the so-far untested second prediction: that toddlers
can only represent an agent as expecting an object if they directly witnessed an encounter
between agent and object, but not if they have to infer that an encounter took place. A third
condition, the Inferred Experience Condition, was added in which children saw the agent looking
and grasping into a box whose content they themselves only saw when the object was later taken
out and replaced. Including this condition served to test the second prediction and to address the
mentioned confounds. The agent explored the object via the same senses (vision and touch, no
longer taste) and expressed the same desire for the object as she had in the Direct Experience
Condition.
50
BELIEF UNDERSTANDING
3.4 Experiment 2
This experiment expands on Experiment 1 in two main ways. First, control trials were
added to each condition, in which the original object was not replaced but remained in the
container. These trials served as a baseline to compare against experimental trials (in which food
items were replaced), permitting us to determine whether witnessing others’ object experiences
is necessary or only superior to witnessing them receive testimonial information. A response
pattern compatible with the view that witnessing others’ direct experiences is necessary would be
those children, in addition to showing more expressions in experimental trials of the Experience
Condition than in experimental trials of the Information Condition (replicating Experiment 1),
show more expressions in experimental than control trials in the Experience, but not in the
Information Condition. A response pattern compatible with the view that witnessing others’
object experiences is superior but not necessary, would be that, in addition to replicating
Experiment 1’s findings, children’s expressions in experimental trials outnumber those in control
trials in both, the Direct Experience and the Information Condition. This experiment is
pre-registered (https://osf.io/e6bkm/?view_only=521e715043354cc4a91e9736a1959cb8).
The second expansion was to test whether toddlers need to directly perceive the agent
encountering the object or whether children can also track experiences that they did not directly
witness but had to infer. In the Inferred Experience Condition, children did not directly witness
the agent experiencing the object (as she did in the Experience Condition of Experiment 1 and in
what we, in this experiment, call the “Direct Experience Condition”), but instead saw the agent
peeking and grasping inside a box whose content (e.g., blueberries) was only later revealed to the
children at the beginning. The Inferred Experience Condition matched the Direct Experience
Condition on the agent’s multisensory (vision and touch) experience of the object. Furthermore,
51
BELIEF UNDERSTANDING
the object was marked as desirable in all three conditions. In the Direct and Inferred Experience
Conditions, the agent herself proclaimed, prior to leaving, “These are great. I’m going to get
them later!” In the Testimony Condition, the testifier, after informing the agent of the box’s
content, added: “They are great. You can go get them later”. These procedural changes should
permit us to more sharply define the limits of toddlers’ mindreading.
3.4.1 Methods
Participants
An a priori power analysis revealed that a sample of N = 90 was needed to match the
statistical power of Experiment 1. The final sample included 90 (45 female) 3-year-olds (M =
37;10 range = 33;26–42;00 months), with four additional children who were recruited but
excluded due to uncooperativeness (3) or technical error (1). All participants were native English
speakers. Based on parents’ reports, 12% were Latinx participants, and the racial composition
was 64% White, 11% Asian, 5% Black, 11% multiracial, 9% other. Children were from diverse
socioeconomic backgrounds with household incomes varying between $20,000 to over
$120,000.
Materials & Design
The same puppet theater, boxes, and non-food items were used as in Experiment 1. The
food items were the same with the exception of the chocolates, which were replaced by larger
chocolate bars for easier identification. A 2x2 mixed design was used, with condition as
between-subjects variable and trial type (experimental vs. control) as within-subjects variable.
Children were randomly assigned to the Direct Experience Condition, the Information Condition,
or the Inferred Experience Condition. Each participant received four trials: two experimental and
52
BELIEF UNDERSTANDING
two control trials. Story order and trial type order (ECCE, CEEC, ECEC, CECE) were
counterbalanced.
Procedure
The recruitment and general testing procedures (video-recorded online testing with
computer screens >= 11’’) were the same as in Experiment 1.
As shown in Figure 9, in the Direct Experience Condition, Puppet A entered and removed
the food (e.g., blueberries) from the container, held the food, and said: “These are great! I’m
going to come back and get them later”. In experimental trials, Puppet B then appeared and
replaced the food with inedible objects (e.g., puzzle pieces), whereas in control trials, nothing
happened in Puppet A’s absence. Next, Puppet A returned, stated her intention to fetch the object,
“Hi, I’m back! Ooh - I’m going to get them now”, approached slowly, and finally grasped the
container. The duration of the test phase was identical to Experiment 1. In the Information
Condition, as in Experiment 1, another puppet, Puppet C, told Puppet A “There are some
blueberries in here for you”, and added “They are great! You can come back and get them later”.
The addition served to keep expressions of the object’s desirability similar between conditions.
In the Inferred Experience Condition, Puppet A first opened the container without removing the
items. She peeked and grasped inside the container and said “These are great! I’m going to come
back and get them later.” The remainder of the trials were as described for the experimental and
control trials in the Direct Experience Condition.
53
BELIEF UNDERSTANDING
Figure 9
Procedural Steps of a Puppet Show Story in Experiment 2
Note. The upper panel illustrates the procedural steps for the Direct Experience Condition, the
middle panel depicts those for the Information Condition, and the lower panel depicts those for
the Inferred Experienced Condition. Each line represents one trial type, with the story
progressing temporally from left to right.
Coding
The same coding and reliability test procedures were used as in Experiment 1. Inter-rater
reliability was good, Kappa = .81; disagreement was resolved by discussion.
54
BELIEF UNDERSTANDING
3.4.2 Results
A GLME model revealed that gender or story had no effect on the number of expressions
children showed, ps > .95. However, trial and age (in days) had an effect, with more expressions
found in later (rather than earlier) trials, b = .10, se = .04, p < .01, and in younger (rather than
older) children, b = - .001, se = .0007, p < .05.
We ran a GLME model with condition, trial type, and their interaction as predictors and
number of children’s expressions as dependent variable. As shown in Figure 10, there was an
effect of trial type, b = .58, se = .15, p < .001, with more expressions in experimental than control
trials. A significant interaction showed that the difference in expressions between experimental
and control trials was greater in the Direct Experience Condition than in the Information
Condition, b = .52, se = 22, p < .05, and than in the Inferred Experience Condition, b = .48, se =
.21, p < .05. The magnitude of the difference between experimental and control trials did not
differ between the Information and the Inferred Experience Conditions, b = .03, se = .22, p = .88.
Post-hoc analyses with Bonferroni-adjusted p-values showed significantly more expressions in
experimental than control trials in the Direct Experience Condition, b = .58, se = .15, p < .001,
but not in the other two conditions (Information Condition: b = .06, se = .16, p = .70; Inferred
Experience Condition: b = .09, se = .15, p = .54).
55
BELIEF UNDERSTANDING
Figure 10
Mean Number of Expressions by Condition and Trial Type in Experiment 2
Note. Error bars on the graph indicate the standard error for each category.
3.4.3 Discussion
The results support two claims by what we have called the shallow mindreading account
(see Apperly & Butterfill, 2009; Perner & Roessler, 2012). The first claim was that children
under age 4 are unable to grasp others as expecting objects unless they previously saw the agents
perceptually experiencing these objects. The second claim was that children of this age need to
directly witness the agent experiencing a specific object: seeing an agent interacting with
something hidden from the toddlers’ view should prevent the experiential record-building that is
deemed necessary for their detection of others’ object expectations. Both predictions were
confirmed. Only in the Direct Experience Condition, in which children saw how an agent looked
at and touched an object they could see, did children react with greater tension when the agent
56
BELIEF UNDERSTANDING
returned to a new (unexpected) object than when she returned to her original object. Children
who either watched the agent receive testimonial information or who witnessed the agent looking
at and grasping a hidden object (which they themselves only later got to see), did not show more
expressions in experimental than control trials.
3.5 General Discussion
In this study, we closely examined the conditions under which toddlers’ mindreading
“works”. Specifically, we tested two predictions derived from the middle position discussed in
the introduction (Apperly & Butterfill, 2009; Moll, Ni, & Stekeler, 2022; Roessler & Perner,
2012;) These predictions were that 1) an agent’s direct perceptual engagement with an object,
but not her obtaining testimony, would permit toddlers to grasp the agent’s object expectation,
and that 2) children need to fully witness agent-object interactions; inferentially piecing together
what an agent must have perceived should not lead to expectation encoding. The predictions
were derived from the claim, central to the middle position that children under age 4 or 5 only
view agents as expecting objects that they previously saw them experiencing.
Both predictions were supported by our experiments, in which children’s expressions of
tension were measured when an agent returned to a container whose content had been replaced.
In the first experiment, the same children showed stronger expressions when the agent had
perceptually experienced (Experience Condition) than when she had been informed (Information
Condition) about the content. Although this finding is congruent with the first prediction, its
interpretability was found to be limited. It did not clarify whether watching others experiencing
objects leads to deeper encoding expectations than watching others obtain information or
whether, as the two-systems and experiential accounts seem to imply, witnessing first-hand
experiences is necessary for toddlers to detect expectations. There were also confounds, as the
57
BELIEF UNDERSTANDING
agent’s access to the object was multimodal and its desirability made explicit (“Hmm, let me
have some”) in the Experience, but not the Testimony Condition. Some of these differences
could not be fully removed. For example, the Experience Condition required that the agent
remove the object from an opaque container (thus combining vision and touch), but testimony is
typically auditory, and we saw no straightforward way to make it multimodal.
Adding control trials and the Inferred Experience Condition in which multimodality
(vision and touch) and desire expression (“These are great. I’ll get them later”) were matched to
that of the Direct Experience Condition served to address the shortcomings and expand the
study’s scope. What was different in the Inferred Experience compared to the Direct Experience
Condition was that the child did not directly perceive but had to retroactively infer what the
agent experienced. In line with the middle position, toddlers did not encode the agent’s object
expectation in this situation. Not only did they show fewer expressions when the agent
approached the ersatz object than did children in the Direct Experience Condition, but their
expressions in this scenario also did not differ from those when the agent returned to the original,
non-replaced, object. As in Experiment 1, watching the agent receive testimonial information did
not permit children to represent the agent as expecting the object, either. Children’s expressions
in experimental trials of the Information Condition did not exceed baseline (obtained from
control trials in the same condition) and were significantly reduced compared to expressions in
experimental trials of the Direct Experience Condition.
Our experiments shed new light on toddlers’ mindreading and its limitations. There is
converging evidence that infants are neither entirely unable to access the minds of others (see
Perner & Ruffman, 2005; Ruffman, 2023, for a behavior reading interpretation of indirect test
results), nor do they possess, as others have claimed (Baillargeon et al., 2010; Carruthers, 2013)
58
BELIEF UNDERSTANDING
deep and flexible mindreading skills. Instead, as several recent studies have shown, infants and
toddlers have shallow and inflexible mindreading capacities that stand or fall by the availability
of experiential records.
To conclude, this study contributes to a better understanding of the foundation of
toddlers’ mindreading. We confirmed that toddlers use records of others’ experiences, and that
what goes into these records are directly witnessed perceptual interactions with objects. What
remains to be further explored is how to best conceptualize the cognition toddlers bring to bear
when tracking others’ expectations.
59
BELIEF UNDERSTANDING
Chapter 4 Backward Belief Reasoning: From Action to Belief and Experience
In this study with two experiments, we adopted a novel perspective by examining
backward reasoning to understand children’s grasp of the relationships between experience,
belief, and action. We compared two age groups, 3- and 5-year-olds, and further investigated the
correlation between children’s backward belief reasoning and forward belief reasoning. This
study aimed to demonstrate a growing understanding among children aged 3 to 5 of the
connections between experience, belief, and action.
4.1 Introduction
Most classic false belief tasks, including the change-of-location task (Baron-Cohen et al.,
1985; Wimmer & Perner, 1983) and the unexpected-content task (Hogrefe et al., 1986), require
children to understand the false belief that the agent has and then use this information to predict
the agent’s subsequent actions. In other words, children are assumed to perform forward
reasoning based on their prior understanding of the connection between belief and action, i.e.,
people will act following their specific beliefs. However, oftentimes we need to make judgments
on others’ beliefs and prior experiences based on their actions. For example, if Shirley is
surprised when she enters a room and sees a surprise party, we would infer that she must have
believed that there is no surprise party in the room. On the contrary, if she is not surprised, we
may infer that she must have already believed the existence of the surprise party. In this case,
backward reasoning is required in order to judge the belief a person has given their actions. It
involves the understanding of the connections between both prior experience and belief, as well
as belief and action.
Only a few studies have investigated how people engage in backward reasoning about
others’ mental states. For example, in an fMRI study with adults (Grèzes et al., 2004),
60
BELIEF UNDERSTANDING
participants watched an agent trying to lift a box; they were later asked if the agent had a correct
or false expectation about the weight of the box based on their lifting posture. The results
showed that participants judged the expectations of the agent correctly at a highly above-chance
level, suggesting that they were able to reason about the agent’s belief solely based on observing
the agent’s non-verbal actions. Additionally, a study by Chuey & Gweon (2021) looked at how
preschool-aged children infer a communicator’s knowledge by observing their impact on a
listener’s ability to use a toy. This study revealed that, around the age of 5, children are capable
of inferring the communicator’s epistemic state based on the outcomes of their communication,
whether it be success or failure. Another study that used an adapted version of the Sally-Anne
Task showed that when children were informed that Sally returned and expressed anger,
5-year-olds, and even some 4-year-olds, could infer that Sally had seen Anne moving the marble.
However, the 4-year-olds still could not accurately predict where Sally would go to find her
marble. This suggests that while 4-year-old children have the ability to infer that someone might
get angry when their belongings are moved by another person, they still lack the ability to use
this knowledge to link the agent’s prior experience with their future action (Wu & Schulz, 2021).
Heretofore, most of the studies on backward belief reasoning with child participants
focused on action explanation. In Bartsch and Wellman’s (1989) study, children were asked to
explain why an agent went to the wrong location searching for an object. More than half of the
3-year-olds provided belief- or desire-related explanations such as “she thinks the cat is under the
piano”. On the contrary, Moses and Flavell (1990) found that 3-year-olds did not perform well in
justifying the agent’s subsequent actions (e.g., going to the wrong location) or surprise reactions.
Furthermore, Wellman and Banerjee (1991) designed an experiment in which an agent opened a
container with unexpected content (e.g., a Band-Aid box with a balloon inside). After being
61
BELIEF UNDERSTANDING
informed that the agent felt surprised, 4-year-olds were more likely to interpret surprise as a
belief-dependent emotion, compared with 3-year-olds. This indicates that children at age 4 have
some understanding that a person will behave in a certain way as the outcome of holding a false
belief. The results were successfully replicated by a few studies (MacLaren & Olson, 1993;
Ruffman & Keenan, 1996) that also found that it was not until 4 to 5 years of age that children
could backward reason about belief in reference to actions.
The previous studies, however, may not fully capture children’s ability of backward
belief reasoning for the following reasons. First, it has been argued that there is a lag in
predicting surprise relative to understanding false beliefs because children also need to have
certain prior knowledge about how people react affectively when the reality is inconsistent with
their beliefs (Raffman & Keenan, 1996). Second, the action explanation questions were usually
open-ended (e.g., “Why do you think she is doing that?”). Considering toddlers’ limited ability to
express their thoughts verbally, simply coding their answers to the opened questions may fail to
capture their complete thinking process. Therefore, a more direct measurement of children’s
backward belief reasoning is needed.
Furthermore, the previous studies focused more on the connection between belief and
reaction (Bartsch & Wellman, 1989; Wellman & Banerjee, 1991), for example, “Shirley went to
the wrong location because she falsely believed that her marble was there,” instead of how belief
is formed from prior experience, for example, “Shirley falsely believed that her marble was at
that location because she put it there initially and did not see others moving it.” We argue that
3-year-olds are at a phase where their theory of mind is limited, and understanding how beliefs
are formed from prior experience is crucial for their mental state reasoning (as shown in Studies
1 and 2). Therefore, backward belief reasoning research should be extended to one step before
62
BELIEF UNDERSTANDING
belief, taking into account how prior experience leads to belief. In the present study, we thus aim
to address this gap by having children backward reasoning about an agent’s previous experiences
based on the agent’s subsequent actions.
Since both forward and backward reasoning are dependent on children’s understanding of
the connections between experience, belief, and action, it is reasonable to predict that their
performance in these two types of reasoning tasks might be correlated. Atance and O’Neill
(2004) have investigated children’s explanations of their own actions led by false beliefs. In their
experiments, 3-year-olds were first presented with a crayon box and asked what they thought was
inside. After they answered “crayons”, they were instructed to use the crayons to paint a piece of
paper. When children opened the box and found out that there were actually candles inside, they
were asked what they thought was in the box before it was opened and why they went to get the
paper. The results showed that children who failed to recall their previous false beliefs were also
unable to explain their actions, suggesting that the understanding of the connections between
experience, belief, and action might be the shared common base of forward and backward belief
reasoning.
4.2 Current Study
The questions addressed in the current study are (1) At what age can children reason back
from another’s intentional action to their prior experience? We examined whether children at age
3 and age 5 acquire the ability to infer what an agent must have seen or witnessed previously,
given the agent’s action. (2) Is children’s backward reasoning associated with their forward
reasoning? We aimed to uncover connections between these two facets of belief reasoning and
shed light on children’s understanding of the interrelation between experience, belief, and action.
63
BELIEF UNDERSTANDING
To measure backward belief reasoning, we created backward versions of the
change-of-location task (Experiment 1) and the unexpected-content task (Experiment 2). In
Experiment 1, children watched stories in which an agent placed an object at a location and then
sat down in a loungy position, facing away from the child. The object was then moved to a
second location by another agent. Children were told that the first agent either fell asleep or
stayed awake during the relocation. Later, the second agent approached one of the locations to
retrieve the object. Children were asked to infer whether the agent was asleep or awake during
the relocation. Two classic change-of-location tasks were added to assess the children’s forward
belief reasoning. In Experiment 2, Children were told that an agent, who sat facing away from
them in a lounge position, desired a specific food (e.g., cookies). Children learned that the agent
might fall asleep or stay awake. Next, a second agent arrived with a conventional container (e.g.,
a cookie box) that seemed to have the first agent’s favorite food inside but was shown to be filled
with non-food items (e.g., forks). Finally, the first agent got up and either happily approached the
container or disappointedly walked off. Children were asked whether the agent was awake or
asleep during the demonstration, based on the first agent’s reaction. Two classic
unexpected-content tasks were added to assess their forward belief reasoning.
For both experiments, we predicted that there would be an age difference in the backward
reasoning task–5-year-olds would be better at backward belief reasoning than 3-year-olds.
Children’s performance in the backward belief reasoning tasks was also predicted to be
positively correlated to their performance in the corresponding classic false belief tasks. The
study has received approval from the university’s Institutional Review Board.
64
BELIEF UNDERSTANDING
4.3 Experiment 1
This experiment has been pre-registered on the Open Science Framework
(https://osf.io/xnkpt/?view_only=5a85dfce343b4524aa6984ef6d81a3c5).
4.3.1 Methods
Participants
Participants were 3- and 5-year-olds recruited from social media, local preschools, and
neighborhoods in the Greater Los Angeles Area. An a priori power analysis revealed that a
sample size of N = 58 with alpha at .05 and a power of .80 (G*Power version 3.1) would detect
an effect of .30. Our final sample included 30 (15 female) 3-year-olds (M = 41;19 months, range
= 44;02–37;27 months), and 30 (15 female) 5-year-olds (M = 62;10 months, range =
58;00–68;17 months). The racial breakdown was 58% White, 20% Asian, 5% Black, 16%
multiracial (1% did not disclose their racial category); 7% identified as Latinx. Family household
incomes varied widely, ranging from less than $20,000 to over $120,000.
Materials
Witness Task. Four stories were presented to children in the format of slideshows. These were the
Apple Story, Popcorn Story, Ice-Cream Story, and Cupcake Story. There were 15 colorful
hand-drawn cartoon pictures in each story picturing a food item placed in one location by an
agent and then being moved to another location by a different agent. For example, as shown in
Figure 11, the Apple Story contained a boy as the first agent, his mother as the second agent, a
living room scene with a couch and two different colored cabinets (purple and orange).
Classic Change-of-Location Tasks. The Sally-Anne Task (Baron-Cohen et al., 1985) and the
Maxi-Chocolate Task (Wimmer & Perner, 1983) were presented to children using slideshows.
65
BELIEF UNDERSTANDING
There were 5 colorful hand-drawn cartoon pictures in the Sally-Anne Task and 6 colorful
hand-drawn cartoon pictures in the Maxi-Chocolate Task (Figure 13).
Figure 11
Four Stories in the Witness Task in Experiment 1
Note. The four stories used in the Witness Task of Experiment 1 are the Apple Story (upper left),
the Ice-Cream Story (upper right), the Popcorn Story (lower left), and the Cupcake Story (lower
right).
Design
A mixed design with age as between-subjects variable and condition as within-subjects
variable was used. In the Witness Task, both 3-year-olds and 5-year-olds experienced 4 stories of
which two stories were in the Witnessed Condition and the other two in the Unwitnessed
Condition. The order of story and condition, as well as how the object was relocated followed the
66
BELIEF UNDERSTANDING
counterbalance schedule. Two classic change-of-location tasks, the Sally-Anne task and the
Maxi-Chocolate task, were added to the end as a measure of children’s ability of forward belief
reasoning.
Procedures
The child and their parent were invited to an online meeting room on Zoom after the
parent gave consent. A female experimenter (E) instructed the parent to enter full-screen mode
and asked the child to sit in front of the screen at a distance of arm’s length. E then presented the
stories using screen-share and read the stories to the child while playing the slideshows. The
session was recorded.
Witness Task. The four stories had a similar structure (see Figure 12). In the beginning, Person A
(e.g., a boy cartoon figure) showed up in a scene with a food item (e.g. a red apple) and took a
bite of the food item. Person A then stated that they would eat the rest of the food item later, so
they went to one of the two containers placed on the two sides of the scene. Next, Person A
indicated that they would sit on the sofa/chair. The child was told that Person A may fall asleep
or may stay awake but so far it was unclear which one they would do (e.g., “Andy sits down on
the sofa. Maybe he will fall asleep, or maybe he will stay awake. We don’t know yet”). When
Person A was sitting on the sofa, a second character, Person B, appeared and opened both
containers. Person B took the food item out from the container, moved across the center of the
scene, and put it into the other container. Person B then closed both containers and left. After a
short pause, Person A stood up intending to retrieve and eat the rest of their food item. Next, in
the Unwitnessed Condition, Person A went to the first container where the food item was placed
initially (e.g., “So Andy goes to the purple shelf on the left”). In the Witnessed Condition, Person
A went to the second container which was the actual location of the food item (e.g., “So Andy
67
BELIEF UNDERSTANDING
goes to the orange shelf on the right”). Next, E called the child’s name and asked them about
Person A’s status during the relocation (e.g., “When mom took out the apple, was Andy asleep,
or was Andy awake?”). The order of the options provided (i.e., “awake” or “asleep”) was
counterbalanced as well to avoid potential effects caused by the child simply repeating the last
option.
Figure 12
Procedural Steps of the Witness Task in Experiment 1
Note. The figure contains selective pictures from the Apple Story to exemplify the procedural
steps. The story’s progression is shown temporally from left to right. The placement and
relocation of the object, as well as the part where Person A intended to retrieve the object, were
identical in both conditions. The only difference lies at the end of the story: in the Unwitnessed
Condition, Person A went to the first, original location, while in the Witnessed Condition, Person
A went to the second, actual location.
Classic Change-of-Location Tasks. The participant was presented with the Sally-Anne Task first
followed by the Maxi-Chocolate Task (see Figure 13). In the Sally-Anne Task, Sally put her red
marble in the basket and left. When Sally was away, Anne took the marble out of the basket and
put it into the box. Later, Sally returned and wanted to get her marble. The child was asked
where Sally would go to find the marble. In the Maxi-Chocolate Task, Maxi had a chocolate bar
and he put it in the green cupboard. When he went outside to play, his mom entered the kitchen
and moved the chocolate bar to the blue cupboard. When Maxi returned for a snack, the child
was asked where Maxi would go to find his chocolate.
68
BELIEF UNDERSTANDING
Figure 13
The Classic Change-of-Location Tasks in Experiment 1
Note. The upper graph (A) shows the pictures presented to the participant in the Sally-Anne
Task, and the lower graph (B) shows the pictures presented to the participant in the
Maxi-Chocolate Task.
At the end of the session, E briefly explained the research purpose to the parent and
answered any questions they may have. The family received a $15 gift card to compensate for
their time.
69
BELIEF UNDERSTANDING
Scoring
A primary coder, who was unaware of the group and condition assignments, coded
children’s verbal responses. In the Witness Task, the correct answer to the question is “asleep” in
the Unwitnessed Condition and “awake” in the Witnessed Condition. In the classic
change-of-location tasks, the correct answer is “basket” in the Sally-Anne Task, and “green
cupboard” in the Maxi-Chocolate Task. For all the questions, if a child answered correctly, they
received a score of 1 for that trial; if they answered incorrectly or said “I don’t know”, they
received a score of 0. If a child refused to answer, they would not receive a score, and the trial
was excluded from the analysis (this was the case for 1 trial). A secondary coder, who was also
unaware of the participants’ group and condition assignment, coded and scored 25% of the
responses using the same criteria. The interrater reliability for both the Witness Task and the
classic change-of-location tasks was exceptionally high, Kappa = 1.00, suggesting the scores
given by the secondary coder were fully identical to those given by the primary coder.
4.3.2 Results
Witness Task
Preliminary analyses using logistic Generalized linear mixed-effects (GLME) models
were run to test for potential effects of gender, trial, story, condition, location change pattern, and
question option order on children’s responses in the Witness Task. No such effects were found,
ps > .27. There was also no effect of condition, p = .09, indicating that children’s performance
did not differ between the situation when Person A was asleep during the relocation and thus
went to the old location and the situation when Person A was awake and went to the new
location.
70
BELIEF UNDERSTANDING
A logistic GLME model was run with age group as independent variable and scores on
the Witness Task as dependent variable. The results showed that 5-year-olds gave more correct
answers than did 3-year-olds, b = 1.12, se = .28, p < .001. As shown in Figure 14, on average,
5-year-olds’ performance significantly exceeded chance (set at a score of 2), t(29) = 4.97, p <
.001, Cohen’s d = 4.85, with an average score of 2.97 out of 4 (se = .20), while 3-year-olds’
scores did not differ from chance, t(29) = .01, p = .99, Cohen’s d = .00, with an average score of
2.00 out of 4 (se = .14).
Figure 14
Age Differences in the Witness Task in Experiment 1
Note. Error bars indicate standard errors.
Classic Chang-of-Location Tasks
A logistic GLME model with age group as independent variable and participants’ scores
in the classic change-of-location tasks as dependent variable did not detect a significant
difference between 3- and 5-year-olds, b = .78, se = .49, p =.11. When being asked where the
71
BELIEF UNDERSTANDING
character in the story would go to find the object, some of the 5-year-olds failed to understand
that the character had a false belief which was different from their own belief. The classic
change-of-location tasks may still be challenging for some 5-year-olds.
Correlation Between Backward and Forward Belief Reasoning
Table 1
Number of Children and Their Scores in Experiment 1 Tasks
Note. Each cell displays the number of children who received the score listed in the first column
for the Witness Task and the scores in the top row for the classic change-of-location task. A dash
“-” indicates that no participants fell into that category.
To examine the relationship between children’s performance in the Witness Task and the
classic change-of-location tasks, the scores of the classic change-of-location tasks were summed
up and added to the logistic GLME model predicting the scores in the Witness Task. Children’s
performance in the classic change-of-location tasks significantly predicted their performance in
the Witness Task, b = .67, se = .20, p < .01 (see Table 1). After adding both age and the scores in
the classic change-of-location tasks into the logistic GLME model as predictors, the scores in the
classic change-of-location tasks still predicted the scores in the Witness Task, b = .58, se = .22, p
< .01, so did age group, b = 1.00, se = .29, p < .001 (See Figure 15). This finding supports our
hypothesis that both reasoning about actions from experiences and reasoning from actions back
to prior experiences hang together and result from children’s grasp of the connection between
experience, belief, and action.
72
BELIEF UNDERSTANDING
Figure 15
Predicting Witness Task Scores using Classic Change-of-Location Tasks Scores Grouped by Age
Group
Note. Error bars indicate standard errors.
4.3.3 Discussion
This experiment revealed the significant age difference between 3- and 5-year-olds in
their backward belief reasoning. Older children, the 5-year-olds, were found to be better at
reasoning about an agent’s prior experiences based on their belief-based actions. This finding
aligns with and extends the findings of previous research that has consistently indicated that
belief reasoning skills continue to develop throughout early childhood (Bartsch & Wellman,
1989; Wellman & Banerjee, 1991; MacLaren & Olson, 1993; Ruffman & Keenan, 1996). At the
age of 3, which falls before the commonly agreed-upon age threshold for theory of mind
73
BELIEF UNDERSTANDING
understanding (around 4-5 years old), children have trouble understanding how an agent’s
actions stem from their beliefs and how these beliefs are formed through earlier experiences. In
contrast, 5-year-old children, having surpassed this developmental threshold, have a more
sophisticated ability to engage in complex belief reasoning tasks, such as inferring beliefs from
actions.
Another important finding of this experiment was the strong correlation between
children’s backward and forward reasoning skills. Children who excelled in backward belief
reasoning tasks also demonstrated competence in classic forward belief reasoning tasks. This
finding aligns with the previous study by Atance and O’Neill (2004), who demonstrated that
children’s ability to recall their own false beliefs was associated with their capacity to explain
their actions led by false beliefs. The results suggested that there might be a common ground, or
a shared cognitive foundation, underlying both types of belief reasoning: the understanding of
how experience, belief, and action are intricately connected. Although we were not able to
directly measure this shared foundation, it supports the reasoning from “what the person did,
saw, or missed” to “what the person must be thinking/believing”, and subsequently to “what the
person must be doing later.” As children mature, this shared cognitive foundation develops,
which manifests in their improved performance in belief-reasoning tasks such as the traditional
false belief tasks. If the common ground is not mature yet, such as for the 3-year-olds and some
5-year-olds in this experiment, their performance in both forward and backward belief-reasoning
tasks may be less proficient accordingly.
We found that the Maxi-Chocolate Task and the Sally-Ann Task were still challenging for
some of the 5-year-old children in this experiment. This could be the reason why we did not
observe significant age differences in children’s performance in these two tasks. Previous studies
74
BELIEF UNDERSTANDING
also showed that not all the 5-year-olds were guaranteed to be able to pass the Sally-Ann Task
(Baron-Cohen et al., 1985; Fabricius & Khalil, 2003; Schidelko et al., 2022; Takagishi et al.,
2010) and the Maxi-Chocolate Task (Fabricius & Khalil, 2003; Liu et al., 2008). Thus, it is safer
to say that ages 4 to 5 are when children begin to pass the classic, verbal response-based false
belief tasks. It is important to recognize that developmental milestones in belief reasoning are
gradual, and individual differences in cognitive development play a crucial role in determining a
child’s readiness for false belief tasks. Belief reasoning abilities can exhibit variability even
within the same age group. Some 5-year-olds may still be in the process of developing the
necessary cognitive skills for certain belief reasoning tasks, which lays the foundation for
understanding how experience, belief, and action are connected.
4.4 Experiment 2
Experiment 2 was conducted to explore backward belief reasoning within the context of
the unexpected-content task. Similar to Experiment 1, it involved comparing two age groups: 5-
and 3-year-olds. We developed a backward version of the unexpected-content task. Initially, the
agent’s status during the key event was ambiguous to the child, specifically, whether the agent
saw the demonstration of the non-food items in the food container. At the end of the story, the
child saw the agent’s action–either approaching the container to obtain the desired food or
walking away from it. Next, participants were asked about the agent’s status at the time of the
demonstration. Two classic unexpected-content tasks were included to assess the children’s
forward belief reasoning in a similar context. The experiment was pre-registered on the Open
Science Framework (https://osf.io/3hgjw/?view_only=0253fc86148944a99dbc11ac96b4a7cc).
4.4.1 Methods
Participants
75
BELIEF UNDERSTANDING
The sample size (N = 60) and recruitment strategies were the same as for Experiment 1.
The final sample included 30 (15 female) 3-year-olds (M = 41;00 months, range = 36;20–46;03
months), and 30 (15 female) 5-year-olds (M = 63;06 months, range = 58;05–66;27 months). The
sample’s ethnic composition was 15% Hispanic, and the racial demographics were 60% White,
13% Asian, 2% Black, 12% multiracial, 3% Other (10% did not disclose their racial category).
Family household incomes varied widely, ranging from under $20,000 to in excess of $120,000.
Materials
Witness Task. Four stories were presented to the children in the format of slideshows. These
stories were: the Cookies Story, the M&M’s Story, the Goldfish Story, and the Popcorn Story.
Each story contained 11 colorful, hand-drawn cartoon pictures depicting a food container with
non-food items being placed in a scene, either with or without an agent’s witness. For example,
in the Cookies Story, the illustrations included a boy cartoon figure, a cookie box as the
conventional container, some forks as the unexpected content, a living room background, a sofa,
and a mother cartoon figure (see Figure 16).
76
BELIEF UNDERSTANDING
Figure 16
Four Stories in the Witness Task in Experiment 2
Note. The four stories used in the Witness Task of Experiment 2 are the Cookies Story (upper
left), the Goldfish Story (upper right), the Popcorn Story (lower left), and the M&M’s Story
(lower right).
Classic Unexpected-Content Tasks. Two unexpected-content tasks (Hogrefe, Wimmer, & Perner,
1986) were presented to children using slideshows. Each task used pictures of a container, its
content, and a human figure (see Figure 17). Specifically, the Crayon Task featured pictures of a
crayon box, some puzzle pieces, and a boy, and the Ice-Cream Task used pictures of an ice-cream
box, four spoons, and a woman.
77
BELIEF UNDERSTANDING
Figure 17
The Classic Unexpected-Content Tasks in Experiment 2
Note. This figure illustrates the stimuli used in the two unexpected-content tasks of Experiment
2. The upper panel shows the crayon box, the puzzle, and the agent Bob in the Crayon Story. The
lower panel shows the ice-cream box, the spoons, and the agent Nancy in the Ice-Cream Story.
Design
The design of this experiment was identical to that of Experiment 1, with age group (3 vs.
5 years) as between-subjects variable and condition (Witnessed vs. Unwitnessed) as
within-subjects variable. The order of the story and condition were counterbalanced. Participants
first completed the Witness Task, followed by the two classic unexpected-content tasks.
Procedures
The procedure for obtaining parental consent and the setup were identical to those in
Experiment 1.
Witness Task. The four stories had a similar structure (see Figure 18). In the beginning, Person A
(e.g., a cartoon boy named Andy), appeared in a scene (e.g., a living room with a sofa).
78
BELIEF UNDERSTANDING
Participants were informed that Person A loved a particular food (e.g., cookies) and ate it all the
time. Person A then sat on a sofa chair with their back to the participant, preventing their face
from being seen. It was unclear to the participant at the moment whether Person A would fall
asleep or stay awake (e.g., “He sits down on the sofa. Maybe he will fall asleep, or maybe he will
stay awake. We don’t know yet.”) While Person A was seated, another character, Person B (e.g.,
a mother cartoon figure), entered with a food container (e.g., a cookie box) corresponding to
Person A’s desired food. Person B then revealed that the container actually held inedible items
(e.g., forks) instead of the expected food (e.g., cookies), as narrated to the participant (e.g.,
“Mom opens the box and takes everything out of the box. Look! There are no cookies, only
forks.”) After closing the container and placing it on a table, Person B left the scene. After a
short pause, Person A stood up wanting to eat their favorite food. Person A looked at the
container. Next, in the Unwitnessed Condition, Person A made a happy face, approached the
container on the table, and reached inside (e.g., “He makes a happy face and reaches inside”). In
the Witnessed Condition, Person A made a sad face and walked away towards the exit of the
room (e.g., “He makes a sad face and walks away”). E then asked the child about Person A’s
status during Person B’s demonstration (e.g., “When mom opened the box, was Andy asleep, or
was Andy awake?”), with the order of the options (“awake” or “asleep”) counterbalanced to
mitigate bias from the child repeating the last option heard.
79
BELIEF UNDERSTANDING
Figure 18
Procedural Steps of the Witness Task in Experiment 2
Note. The figure contains selective pictures from the Cookies Story to exemplify the procedural
steps. The story’s progression is shown temporally from left to right. The parts where Person A
expressed their desire, sat down, and Person B demonstrated the non-food items in the food box
were identical in both conditions. The only difference occurs at the story’s end: in the
Unwitnessed Condition, Person A reached into the container with a happy face, while in the
Witnessed Condition, Person A walked away with a sad face.
Classic Unexpected-Content Tasks. Participants were presented with the Crayon Task and the
Ice-Cream Task in counterbalanced order. In the Crayon Task, a crayon box appeared on the
screen, and the participant was asked, “Look! What do you think is in there?” Following their
response, E announced, “Let me open the box and take everything out.” The crayon box on the
screen rotated and shook, revealing puzzle pieces inside. E then stated, “Look! There are no
crayons, only puzzle pieces.” After putting the puzzle pieces back into the crayon box, E asked,
“Can you tell me what’s in the box?” Next, a boy named Bobby appeared beside the container. E
stated, “Here is Bobby. He didn’t see what we just did. He just got here and is looking at the
box.” and then asked: “What does Bobby think is in the box?” The child’s response was
recorded. The Ice-Cream Task followed the same structure with different materials.
At the end of the session, E briefly explained the research purpose to the parent and
answered any questions they may have. The family received a $15 gift card to compensate for
their time.
80
BELIEF UNDERSTANDING
Scoring
A primary coder, unaware of group or condition, coded 100% of children’s verbal
answers. Children received a score of 1 for a correct answer (“asleep” in Unwitnessed Condition
and “awake” in Witnessed Condition) and a score of 0 for an incorrect answer or responding
with “I don’t know.” A secondary coder, also unaware of group or condition, independently
coded a random 25% of the sample. The interrater reliability for both tasks was perfect (Kappa =
1.00).
4.4.2 Results
Witness Task
As a preliminary analysis, potential effects of gender, trial, story, and condition were
explored using logistic GLME models. The analysis found no effects of gender or trial on
children’s scores in the Witness Task, ps > .36. There was a significant effect of condition, with
children giving more correct answers when the correct answer was that Person A was “asleep”
(Unwitnessed Condition) than when the correct answer was that Person A was “awake”
(Witnessed Condition), p < .05. Children also gave more correct answers when the question was
“...was [Person A] awake or was [Person A] asleep?” than when the question ended with “...was
[Person A] asleep or was [Person A] awake?”, p < .01. Moreover, a likelihood ratio test was
conducted to compare the fit of the models with and without story as an independent variable,
revealing a significant effect of story, χ²(3) = 13.52, p < .01. (The raw probabilities of answering
correctly in the Cookies Story, the Goldfish Story, the M&M’s Story, and the Popcorn Story were
.68, .37, .45, and .52, respectively.) These unpredicted effects of condition and story were further
discussed in the discussion.
81
BELIEF UNDERSTANDING
To test for the hypothesized age effect, a logistic GLME model with age as independent
variable and children’s answers as dependent variable was run. As predicted, there was a
significant age difference, with 5-year-olds giving more correct answers than 3-year-olds, b =
.92, se = .27, p < .001. As shown in Figure 19, 5-year-olds’ answers exceeded chance, t(29) =
2.45, p < .05, Cohen’s d = 2.45 (average score = 2.47 out of 4), while 3-year-olds performed
significantly below chance, t(29) = -2.90, p < .01, Cohen’s d = -2.90 (average score = 1.57 out of
4). There was no significant interaction between age and condition, b = .51, se = .54, p = .34,
revealed by another logistic GLM.
Figure 19
Performance of the 3- and 5-Year-Olds in the Witness Task
Note. Error bars on the graph indicate the standard error for each category.
Classic Unexpected-Content Tasks
For the two classic unexpected-content tasks, a logistic GLME model showed that there
was no significant effect of gender, task, or trial on children’s scores, ps > .09.
82
BELIEF UNDERSTANDING
A logistic GLME model, with children’s scores in the classic unexpected-content tasks as
outcome variable and age group as independent variable, showed an age effect: 5-year-olds were
significantly more likely to answer correctly than 3-year-olds, b = 3.60, se = 1.11, p < .01.
Correlation Between Backward and Forward Belief Reasoning
To examine the relationship between children’s performance on the classic
unexpected-content tasks (forward reasoning) and the Witness Task (backward reasoning), the
scores of the two classic unexpected-content tasks were summed up and entered into the logistic
GLME model as a predictor of the Witness Task. The results showed that children’s performance
in the forward reasoning tasks significantly predicted their performance in the Witness Task, b =
.65, se = .16, p < .001. It suggested that children’s ability to reason about another person’s belief
and belief-based action was reflected in their performance in both the forward and backward
belief reasoning tasks. After adding age and the classic unexpected-content tasks’ scores into the
logistic GLME model as predictors, performance on the unexpected-content tasks still predicted
performance on the Witness Task, b = .49, se = .19, p < .01, while age did not, b = .50, se = .31, p
= .11. The number of children receiving a specific combination of scores on the two types of
tasks are shown in Table 2.
Table 2
Number of Children and Their Scores in Experiment 2 Tasks
Note. Each cell displays the number of children who received the score listed in the first column
83
BELIEF UNDERSTANDING
for the Witness Task and the scores in the top row for the classic unexpected-content tasks. A
dash “-” indicates that no children fell into that category.
4.4.3 Discussion
This experiment used a backward reasoning version of the unexpected-content task to test
the ability of 3- and 5-year-old children to reason about prior experiences based on consequential
belief-based actions towards a container with unexpected content. In line with the findings of
Experiment 1, 5-year-olds outperformed 3-year-olds at correctly inferring the agent’s prior
experience–whether the agent was asleep and therefore missed the demonstration of the actual
content, or whether the agent was awake and witnessed the demonstration. Including the classic
unexpected-content tasks, which measure forward belief reasoning, allowed us to assess the
association between backward and forward reasoning. Across both age groups, a correlation was
detected between children’s performance in backward belief reasoning and their performance in
forward belief reasoning. It suggests that what children come to acquire when first solving belief
problems of this kind is an overall comprehension of the connections between experience, belief,
and action, which tends to develop and strengthen with age.
Children’s performance in the classic unexpected-content tasks was consistent with
previous studies, which found that older children outperformed younger ones in reasoning about
an ignorant person’s false belief (de Villiers & Pyers, 2002; Hogrefe, et al., 1986). In our study,
we adapted the classic unexpected-content tasks to an online format, which retained the core
features of the classic task. Our results affirm the robustness of these tasks in assessing children’s
understanding of false beliefs, regardless of the medium used for their presentation.
A small difference in the findings concerns the effects of condition and story. In this
experiment, children were more likely to answer correctly in the Unwitnessed Condition (Person
A asleep) than in the Witnessed Condition (Person A awake), which was not observed in
84
BELIEF UNDERSTANDING
Experiment 1. Considering the detected effect of option order, it is possible that backward
reasoning in the unexpected content context is harder than the change-of-location task (as
discussed in Study 1). This complexity might lead children, when uncertain of the correct
answer, to default to guessing “asleep” due to the invisibility of Person A’s front face. This may
result in more “asleep” responses in trials where children were unsure, thereby increasing
accuracy in the Unwitnessed Condition where “asleep” is the correct answer.
Another interesting finding is the relative ease of the Cookies Story compared to the
M&M’s Story and the Goldfish Story. This could be attributed to children’s familiarity with these
containers. Although all four food containers are commonly known to children, a cookie box
might be more prevalent in family environments than the others, making it easier to recognize in
the slideshow. However, this aspect of container familiarity, while interesting, does not
substantially impact our primary conclusions. The key focus in our experiment was on children’s
recognition of the discrepancy between the agent’s desires and the actual contents of the box, a
concept that was consistently and clearly stated in each story.
4.5 General Discussion
When someone who is searching for cookies turns away from a cookie box with
disappointment, he must believe there are no cookies inside. Arriving at this inference involves a
complex backward reasoning process, from someone’s desires and actions to their beliefs. The
development of this process and its relation to children’s other theory of mind skills have been
under-researched but were targeted in this study.
Across two experiments with two age groups (3- and 5-year-olds), we investigated how
preschool-aged children use others’ belief-based actions to infer what previous experiences
others must have had. In Experiment 1, we used a backward version of the change-of-location
85
BELIEF UNDERSTANDING
task. Children watched as an agent placed an object in a location, then either witnessed or missed
its relocation, with their subsequent actions revealing their belief about the object’s location.
Experiment 2 explored backward belief reasoning in the unexpected-content context. Children
observed an agent’s reaction to a food container that was shown to contain non-food items,
allowing inference about the agent’s experience based on whether they approached or walked
away from the container. The results from both experiments showed that compared with
3-year-olds, 5-year-olds were better at inferring the agent’s prior experience based on their
subsequent actions. Moreover, children’s scores in the backward belief reasoning tasks were
predicted by their scores in the forward belief reasoning tasks, i.e., the classic false belief tasks.
This suggests a developmental progression in which the ability to engage in both forward and
backward belief reasoning strengthens concurrently.
The strong association between children’s backward and forward belief reasoning
suggests that what children at ages 4 to 5 are forming is not a highly specialized detection of
beliefs, but, by contrast, a more general and pervasive understanding of how experience, belief,
and action are interconnected. This notion aligns with the argument that children’s difficulty with
false beliefs reflects a more general inability to understand how psychological states are causally
related to physical events in the world (Leslie, 1992).
It was further argued that two key components of the causal role of beliefs involve
understanding how events in the world give rise to beliefs and how beliefs in turn can give rise to
actions in the world (Moses & Flavell, 1990). In other words, the fundamental understanding is
that beliefs are formed through perceptual experiences and that actions are guided by beliefs,
thereby linking prior experiences and subsequent actions through beliefs. To successfully
complete both forward and backward belief reasoning tasks, children require a comprehensive
86
BELIEF UNDERSTANDING
set of skills beyond mere “false belief understanding”, which previous research seems to often
isolate as a distinct social skill. For instance, these tasks demand a relatively mature executive
function set, including working memory and inhibitory control. Children must pay attention to
and remember the initial and subsequent locations of the object in the change-of-location tasks,
or the actual contents of the container in the unexpected-content tasks. Particularly in the
backward belief reasoning tasks, there is a period where the agent’s state of consciousness is
ambiguous, requiring children to hold two possibilities in mind simultaneously. This also
necessitates a basic understanding of possibilities, so they comprehend that statements like
“maybe Andy will fall asleep or maybe Andy will stay awake” represent two potential states
unknown to them. In the end, when answering questions about the agent’s prior state, they need
to select one possibility and suppress the other. Moreover, the tasks require basic inferential
abilities. For example, the line of reasoning in the backward task is “Sally went to the basket
searching for her marble, so she must believe it is there, but since the marble was relocated to the
box, she must have missed the relocation”. These skills extend far beyond simply understanding
that others may perceive the world differently. Therefore, the current study sheds light on the
holistic development of children, showing the importance of understanding children as whole
beings. It suggests that various domains of a child’s growth—physical, emotional, social, and
cognitive—are intertwined. These aspects of development cannot be fully comprehended in
isolation but rather should be understood as part of an interconnected whole (Katz et al., 2014;
Lally, 2023).
The current study used children’s age as a criterion to divide participants into two age
groups, the observed age differences align well with previous studies using false belief tasks
(Gopnik, 1993; Hogrefe et al., 1986; Wimmer & Perner, 1983; Wimmer & Weichbold, 1994).
87
BELIEF UNDERSTANDING
This supports the idea that the ages of 4 to 5 may represent a significant developmental threshold
in children’s theory of mind understanding. However, it’s important to recognize that age should
not be the sole marker of children’s theory of mind development. Particularly in our study, both
experiments revealed individual differences–even some 5-year-olds struggled with the forward
and backward tasks. This observation reinforces the notion that age, while a useful indicator, is
not an infallible predictor of children’s social and cognitive abilities. The underlying
fundamental understanding of the connection between experience, belief, and action, plays a
more crucial role in children’s ability to successfully pass these tasks.
A few minor differences in the results between Experiment 1 and Experiment 2 merit
discussion. Firstly, overall, children’s performance in both the backward and forward tasks in
Experiment 1 was better than in Experiment 2. Secondly, in Experiment 1, both the age group
and scores in the classic change-of-location tasks were predictors of performance in the Witness
Task. However, in Experiment 2, after including both age group and scores in the classic
unexpected-content tasks as predictors, only the scores in the classic unexpected-content tasks
predicted performance in the Witness Task, while age group was not a significant factor.
Although the children from Experiments 1 and 2 were different cohorts, thereby precluding a
direct comparison of their scores, this discrepancy might suggest that the change-of-location and
unexpected-content tasks may not present equal levels of challenge to children. It is possible that
the classic change-of-location tasks are generally easier than the unexpected-content tasks (as
discussed in Study 1), which might extend to their backward versions as well.
Due to time and recourse constraints, this study did not assess children’s executive
functions, their understanding of possibilities, or their inferential abilities, as these were not the
primary focus. Nonetheless, integrating assessments of these cognitive skills into the narratives
88
BELIEF UNDERSTANDING
of future tasks would be a valuable direction for further study. Furthermore, the current study
primarily relied on children’s verbal responses. Future research could profitably employ
non-verbal measures, such as action-based measures (Buttelmann et al., 2009; Allen, 2015), or
expression-based measures (Moll et al., 2016; 2017; Ni et al., 2023), which may uncover
additional evidence regarding backward belief reasoning.
Taken together, this study offers a fresh perspective on theory of mind in toddlers, with a
particular emphasis on backward belief reasoning. Through innovative adaptations of false belief
tasks, this study has revealed the key age differences, showing how children’s understanding of
the interconnectedness between experience, belief, and action develops over time. Although age
is a crucial developmental marker, the essence of children’s theory of mind ability lies in their
burgeoning capacity to synthesize and reason about others’ experiences, mental states, and
actions. The challenge of measuring this fundamental understanding is formidable, yet our work
provides an insightful glimpse into it and opens avenues for further exploration into how children
perceive and make sense of the world and people around them.
89
BELIEF UNDERSTANDING
Chapter 5 Concluding Discussion
5.1 Summary of Findings
Being in a social environment entails frequent interaction and communication with
people from diverse perspectives. Deciphering their thoughts, unspoken intentions, and swirling
emotions forms the cornerstone of social interaction, an area of research that remains particularly
enigmatic when considering young children. How, and to what extent, are toddlers capable of
navigating this complex web of mental state understanding? This dissertation focuses on the
scope and limits of young children’s belief understanding, examining its manifestations in verbal
and emotional expressions, and thus provides empirical evidence on children’s understanding of
the interconnection between experience, belief, and action.
Chapter 1 of this dissertation reviewed the existing evidence on young children’s theory
of mind, introduced the theoretical debates in the field, and proposed two limits in their
mindreading ability supporting the middle position. Study 1 (in Chapter 2), using an adapted
version of the unexpected-content task, confirmed that the first limit is that 3-year-olds can
anticipate an agent’s encounter with unexpected reality only when they have previously observed
the agent’s experience with the container and its original content. This anticipation is manifested
in children’s spontaneous suspenseful expressions. However, they are unable to explicitly
articulate the agent’s false belief about the conventional container, even with experience tracking.
Furthermore, 3-year-olds do not yet understand how the deceptiveness of an object, such as a
conventional food container, may lead to the agent’s false belief, which is another limit in their
theory of mind examined in Study 1.
Study 2 (in Chapter 3) built upon the findings of Study 1 and further explored the kinds
of experiences children need to witness, in order to register experiential records and then use
90
BELIEF UNDERSTANDING
them in understanding the mind. Using the expression-based measure, this study found that
3-year-olds recognized an agent as expecting an object only when they had directly witnessed the
agent’s perceptual experience with it. This was in contrast to when they witnessed the agent
receive non-experiential information or when they had to infer the agent’s experience. This study
contributed new evidence showing that toddlers’ capacity for mindreading is inflexible and
heavily dependent on the availability of directly tracking perceptual experiences.
Building on the findings from Studies 1 and 2 regarding the importance of experience
tracking in belief understanding and action prediction, Study 3 (in Chapter 4) took the opposite
direction and introduced a novel backward reasoning approach, in which children observed an
agent’s action and had to infer whether the agent had witnessed key events. In both Experiment 1
(the change-of-location task) and Experiment 2 (the unexpected-content task), 5-year-olds
outperformed 3-year-olds in backward belief reasoning. Additionally, children’s performance in
backward belief reasoning and forward belief reasoning were highly associated. This shows that
children between the ages of 3 and 5 are developing a more general understanding of how
experiences lead to beliefs and subsequent actions, and vice versa.
The three studies presented in this dissertation are connected, each building upon the
insights of the previous to deepen our understanding of young children’s theory of mind skills.
Study 1 laid the foundation by identifying two major limits in those younger than age 4: their
reliance on records of others’ experiences and a lack of understanding that deceptiveness can
lead to false expectations. Study 2 extended the previous findings showing that 3-year-olds need
to directly witness others as having perceptual experiences with objects (e.g., seeing and
handling objects). Study 3 showed how children’s understanding of beliefs and actions are
connected, and that forward reasoning from experiences to beliefs and backward reasoning (from
91
BELIEF UNDERSTANDING
beliefs to experiences) go hand in hand. The series of studies delineates a developmental
trajectory in theory of mind from ages 3 to 5, demonstrating a growth from an implicit to an
explicit understanding of others’ mental states.
5.2 Theoretical Implications
The dissertation has significant theoretical contributions to the research field of theory of
mind in early childhood, particularly in children younger than the commonly accepted age
threshold, age 4 to 5, for belief understanding. Our findings support the middle position in the
debate on theory of mind development. Particularly, Study 1 and Study 2 offer new evidence for
the view that children younger than 4 years have theory of mind skills that are limited in
systematic and predictable ways. They rely on directly tracking others’ perceptual experiences as
they make sense of others’ reactions to situations. Tracking experiences allows toddlers to
anticipate the return of an agent where she left an object or foresee her emotional response to a
salient change of reality.
Although tracking experiences does not necessarily mean that children automatically take
others’ perspectives (Ni et al., 2022), the availability of an agent’s prior experiences allows
children to register a record of the agent, the object, and their relationship. The role of tracking
experience is worth considering when revisiting many previous studies. For instance, in
anticipatory looking studies, children’s experience tracking may lead them to look toward where
the agent expects the object, aligning with the agent’s false belief based on the object’s original
location. Likewise, in looking-time experiments, such as those conducted by He et al. (2011) and
Kovács et al. (2010), infants tend to look longer at scenes involving false beliefs than those
involving true beliefs. This prolonged gaze may be attributed to the experience tracking enabling
them to vicariously experience the agent’s surprise. In Rubio-Fernández’s (2019) study,
92
BELIEF UNDERSTANDING
children’s performance in reasoning about the agent’s false belief improved after they initially
observed the agent perceptually interacting with plasters in a container. This highlights the
importance of considering the differences in task designs, particularly in terms of how and
whether children track others’ experiences, be they perceptual, testimonial, or of other types.
Such variations could lead to differences in children’s performance in tasks assessing belief
understanding.
Furthermore, the studies presented in this dissertation highlighted a relatively novel limit
in 3-year-olds’ theory of mind–they do not yet grasp that deceptiveness can induce false beliefs.
This revealed limit can help explain why children were found to perform worse in the
unexpected-content task than in the change-of-location task (Grosse Wiesmann et al., 2017;
Holmes et al., 1996), where no deceptiveness is involved. In addition, the dog robot with one red
side and one blue side used in Low & Watts’s (2013) study and the ambiguous rabbit-duck figure
in Low et al.’s (2014) study also possess deceptiveness, as these objects can be presented in more
than one mode visually. Children’s difficulty in handling deceptiveness also supports the
argument that they can only understand perspectives attached to separate objects, but struggle
with a single object that embodies two confronting perspectives (Moll & Tomasello, 2012).
Upon closer examination, young children’s difficulty in understanding different beliefs
about the same object or event seems to stem from their still-developing grasp of the relationship
between experience and belief. Young children might associate only one belief with an object
based on direct experience, such as believing a chocolate-shaped object is an eraser after
witnessing its use. However, they often fail to recognize that others might hold different beliefs
due to different experiences (e.g., missing the demonstration of the chocolate-shaped eraser and
93
BELIEF UNDERSTANDING
thus believing it is a real chocolate bar). This again shows beliefs’ “mind-to-world” direction of
fit (Searle, 1983), as beliefs represent the world entailing variety and the possibility of error.
The dissertation sheds light on the developmental trajectory of theory of mind from ages
3 to 5, revealing a maturation process where children’s understanding evolves from a practical to
a more theoretical and reflective state (Moll et al., 2022). In the early years of life, children
engage with the world and others primarily as agents, with an understanding of other minds
emerging through interactive experiences. This early sensitivity to mental states facilitates social
interactions and cooperation, as suggested by the Shared Intentionality Theory (Tomasello et al.,
2005; Tomasello & Moll, 2010).
As children grow, their cognitive abilities advance in tandem with biological maturation,
including language skills such as the use of embedded clauses, syntax, mental-state terms, and
perspective-shifting discourse (Astington, 2000; Astington & Pelletier, 1998). These language
features offer more flexible and diverse ways for children to express and comprehend nuanced
beliefs, intentions, and assumptions about their own mental states and those of others. The
development of executive function, including working memory, cognitive flexibility, and
inhibitory control, also allows children to hold and manipulate information about different
perspectives and to inhibit their own beliefs when necessary (Carlson et al., 2004). Furthermore,
increased complex and diverse social experiences through interactions with family, peers, and
strangers further contribute to this developmental journey. Through conversational exchanges,
pretend play, story-telling, collaborative problem-solving, and conflict resolution, they get better
at linking manifest actions and behaviors with mental states that are inaccessible to direct
observation (Nelson, 2005). With these practical experiences, their cognitive development
94
BELIEF UNDERSTANDING
gradually progresses toward a more sophisticated, meta-level understanding of one’s actions,
laying the groundwork for the refinement and expansion of mindreading capacities.
By the age of 4 to 5, children begin to develop a theoretical understanding of beliefs. This
milestone enables them to adopt a reflective stance, allowing for the juxtaposition of an
individual’s subjective perception of reality against other viewpoints, including objective ones.
Such an ability is important for participating in social discourse aimed at making sense of human
actions by examining people’s beliefs, assumptions, or intentions, and understanding how these
mental states inform and motivate specific behaviors (Moll et al., 2022). This progression from a
practical, implicit understanding to a more reflective, theoretical, and explicit understanding of
beliefs marks a significant phase in the lifespan development of theory of mind.
5.3 Methodological Implications
One of the key methodological implications of this research is the emphasis on the
affective aspect of theory of mind. By analyzing children’s emotional expressions, particularly in
response to scenarios involving false beliefs, this dissertation has highlighted the richness of
non-verbal indicators in understanding cognitive processes. The traditional theory of mind
measures have heavily relied on verbal responses, which may not fully capture the nuanced
understanding of young children, especially those who have limited verbal abilities. We thus
argue that emotional expressions offer an inclusive and potentially more sensitive measure.
Apart from eliminating the need for verbal articulation, another advantage of the
expression-based measure used in Study 1 and Study 2 is its reliance on children’s spontaneous
responses. Expressions are a natural, spontaneous process (Bar-On, 2013, 2017). In our studies,
we did not need to ask children to exhibit specific expressions or prompt them with cues. When
they observed an agent approaching an object with a false expectation, their emotional
95
BELIEF UNDERSTANDING
investment in the story and the characters naturally elicited responses, as they sensed the tension
building in the narrative. This contrasts with classic false belief tasks, which typically conclude
with a test question about what another person thinks (Hogrefe et al., 1986) or predicting the
person’s subsequent action (Wimmer & Perner, 1983). Even many action-based measures
involve some form of prompting. For example, in the active helping paradigm (Buttelmann et al.,
2009), the experimenter encouraged the infants to help the agent by saying, “Go on, you can help
him!” In our expression-based measure, no such prompts or questions were used. We truly
captured children’s spontaneous responses, providing a genuine insight into their theory of mind
capabilities without the influence of direct questioning or prompts.
The introduction of the backward reasoning task in Study 3 marks another
methodological innovation. Traditional approaches in this field have primarily focused on
forward reasoning. However, the backward reasoning task flips this approach by asking children
to infer past events or experiences based on the observed belief-based actions of agents. This
approach provides a deeper insight into how children understand the continuity and causality of
mental states over time. By observing how an agent reacts in a given situation, children are
challenged to deduce what the agent must have seen or experienced previously. This task
evaluates not only their ability to reason about others’ previous experiences but also enhances
our understanding of their fundamental grasp of how experience, belief, and action are
interconnected, as well as their comprehension of temporal sequences and causal relationships in
these connections. Such a methodological advancement opens up new avenues for exploring the
development of complex cognitive processes in young children and could potentially be applied
to various other aspects of cognitive and social development.
96
BELIEF UNDERSTANDING
5.4 Remaining Questions and Future Directions
Expressions have proven to be highly informative in providing insights into children’s
understanding of others’ expectations and mental states. A promising direction for future
research could involve a more detailed examination of the nature and timing of children’s
expressions. Observations of facial and bodily expressions during activities such as watching
puppet shows, playing, or engaging in conversations can be indicative of their real-time mental
state understanding. Mental state understanding is dynamic; individuals constantly revise their
beliefs and their understanding of others’ beliefs over short periods (Masnick et al., 2017).
Therefore, tracking children’s expressions in real time throughout a narrative or interactive
activity could yield valuable insights. For example, in the puppet show stories from Study 1,
children’s expressions of curiosity upon first encountering an object, excitement during an
interaction, or surprise when expectations are violated, all contribute to a deeper understanding
of the cognitive processes at play. It could also include the onset and cessation of expressions as
indicators of emerging emotional feelings or violated expectations, as well as their transitions in
relation to story events, such as changing from a shocked face to a smirk, which signifies shifts
in internal processing.
Another possibility is to investigate children’s predictions about others’ expressions in
anticipation of unexpected encounters. This could reveal whether children anticipate the affective
impact of these encounters. Using eye-tracking technology to observe gaze patterns, such as
whether children focus on the face or alternate between the face and the event of interest, could
provide insights into how children’s reasoning about others’ emotional or informational states is
reflected in their gaze behavior.
97
BELIEF UNDERSTANDING
Moreover, exploring whether children anticipate different expressions based on the nature
of an agent’s expectation violation presents another interesting question. Do children expect
distinct expressions when an agent’s expectation is negatively violated, such as shock in
discovering a fake object, or disappointment when expectations are unmet? Conversely, in
scenarios where expectations are positively violated, such as finding a ring in a cake, do they
anticipate expressions of pleasant surprise? By observing children’s reactions to such varied
scenarios, researchers can gain insights into their ability to differentiate and anticipate emotional
responses based on the outcomes of expectation violations.
In terms of the theoretical framework, this dissertation suggests that children develop an
understanding of how experiences, beliefs, and actions are interconnected. A remaining question
is how this understanding evolves into the mature state observed in adults. Specifically, children
may initially rely on tracking experiences to reason about mental states, manifesting in a
practical and implicit manner. As they grow older and accumulate more life experiences, they
gain a deeper comprehension of the formation and consequences of mental states. For example,
through increased interactions with various containers and their contents, or by observing others’
interactions, children learn about the conventionality of containers. They come to understand that
typically, a container’s appearance corresponds to its contents, leading people to form beliefs
based on the container’s appearance, even without prior interaction with its contents. They also
start to grasp why containers with mismatched contents are deceptive and can lead uninformed
individuals to form differing beliefs. Gradually, their understanding extends from situations
where experiential tracking is available to those where it is not. When their understanding
matures to the level of adults, children no longer need to rely on tracking experiences to reason
about others’ beliefs and can articulate their understanding more clearly. Future research could
98
BELIEF UNDERSTANDING
explore how this generalization occurs and identify factors influencing the process, as discussed
in the previous section. In this case, longitudinal studies, with both forward and backward
reasoning tasks, could be a powerful tool to examine the developmental trajectory of this general
understanding of mental states.
5.5 Concluding Remarks
This dissertation contributes to illuminating how far children between ages 3 and 5 have
progressed in their theory of mind development. Consistent with the middle position’s view that
children’s theory of mind skills are shallow until age 4 or 5, we identified two major limits in
3-year-olds’ reasoning about others’ belief states: 1) They fail to understand that an object’s
appearance can induce false beliefs or expectations (as shown in Study 1), and 2) the only kind
of belief-like states–we prefer to speak of expectations–children this young can grasp are those
that are grounded in direct prior experience of the object. Children can only attribute
expectations to others when they are able to track those prior experiences that lead to the
formation of these expectations.
By having identified these notable limits and rigidities in toddlers’ belief understanding,
this dissertation provides strong empirical support for the middle position on early theory of
mind skills. Taken together, the three studies (and 6 experiments) reported here confirm that
3-year-olds are skilled at tracking others’ perceptions and anticipating how agents will be
impacted or react when reality has changed for them in salient but unexpected ways. At the same
time, and in line with the middle position’s predictions, toddlers fail to understand deceptiveness
and the possibility of agents misidentifying objects or misrepresenting states of affairs. This,
however, is a crucial ingredient of representational states whose role is to accurately represent
reality and provide orientation in the world. In sum, by blending traditional and innovative
99
BELIEF UNDERSTANDING
methods, this dissertation has enriched our understanding of the steps by which young children,
between the ages of 3 and 5 years, come to deepen their understanding of the human mind. This
work can be used as background information, and perhaps at times as a foundation, for future
studies aimed at unraveling the complexities of early cognitive and social development.
100
BELIEF UNDERSTANDING
References
Allen, J. W. P. (2015). How to help: Can more active behavioral measures help transcend the
infant false-belief debate? New Ideas in Psychology, 39, 63–72.
https://doi.org/10.1016/j.newideapsych.2015.07.008
Apperly, I. A. (2010). Mindreaders: The cognitive basis of “theory of mind.” Psychology Press.
Apperly, I. A., & Butterfill, S. A. (2009). Do humans have two systems to track beliefs and
belief-like states? Psychological Review, 116(4), 953–970.
https://doi.org/10.1037/a0016923
Apperly, I. A., Warren, F., Andrews, B. J., Grant, J., & Todd, S. (2011). Developmental
continuity in theory of mind: Speed and accuracy of belief–desire reasoning in children
and adults. Child Development, 82(5), 1691–1703.
https://doi.org/10.1111/j.1467-8624.2011.01635.x
Aslin, R. N. (2007). What’s in a look? Developmental Science, 10(1), 48–53.
https://doi.org/10.1111/j.1467-7687.2007.00563.x
Astington, J. W. (2000). Language and metalanguage in children’s understanding of mind. In
Minds in the making: Essays in honor of David R. Olson (pp. 267–284). Blackwell
Publishing.
Astington, J. W., & Pelletier, J. (1998). The language of mind. In D. R. Olson & N. Torrance
(Eds.), The handbook of education and human development (pp. 569–593). John Wiley &
Sons, Ltd. https://doi.org/10.1111/b.9780631211860.1998.00027.x
Atance, C. M., & O’Neill, D. K. (2004). Acting and planning on the basis of a false belief: Its
effects on 3-year-old children’s reasoning about their own false beliefs. Developmental
Psychology, 40(6), 953–964. https://doi.org/10.1037/0012-1649.40.6.953
Baillargeon, R., Scott, R. M., & He, Z. (2010). False-belief understanding in infants. Trends in
Cognitive Sciences, 14(3), 110–118. https://doi.org/10.1016/j.tics.2009.12.006
Bar-On, D. (2013). Expressive communication and continuity skepticism. Journal of Philosophy,
110(6), 293–330. https://doi.org/10.5840/jphil2013110624
Bar-On, D. (2017). Communicative intentions, expressive communication, and origins of
meaning. In K. Andrews & J. Beck (Eds.), The routledge handbook of philosophy of
animal minds (1st ed., pp. 301–312). Routledge.
https://doi.org/10.4324/9781315742250-29
Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a “theory of
mind” ? Cognition, 21(1), 37–46. https://doi.org/10.1016/0010-0277(85)90022-8
101
BELIEF UNDERSTANDING
Bartsch, K., & Wellman, H. (1989). Young children’s attribution of action to beliefs and desires.
Child Development, 60(4), 946–964. https://doi.org/10.2307/1131035
Bernard, S., Proust, J., & Clément, F. (2015). Procedural metacognition and false belief
understanding in 3- to 5-year-old children. PLoS ONE, 10(10), e0141321.
https://doi.org/10.1371/journal.pone.0141321
Brown, J. R., Donelan-McCall, N., & Dunn, J. (1996). Why talk about mental states? The
significance of children’s conversations with friends, siblings, and mothers. Child
Development, 67(3), 836. https://doi.org/10.2307/1131864
Brüne, M., & Brüne-Cohrs, U. (2006). Theory of mind—Evolution, ontogeny, brain mechanisms
and psychopathology. Neuroscience & Biobehavioral Reviews, 30(4), 437–455.
https://doi.org/10.1016/j.neubiorev.2005.08.001
Burnside, K., Ruel, A., Azar, N., & Poulin-Dubois, D. (2018). Implicit false belief across the
lifespan: Non-replication of an anticipatory looking task. Cognitive Development, 46,
4–11. https://doi.org/10.1016/j.cogdev.2017.08.006
Buttelmann, D., Carpenter, M., & Tomasello, M. (2009). Eighteen-month-old infants show false
belief understanding in an active helping paradigm. Cognition, 112(2), 337–342.
https://doi.org/10.1016/j.cognition.2009.05.006
Butterfill, S. A., & Apperly, I. A. (2013). How to construct a minimal theory of mind. Mind &
Language, 28(5), 606–637. https://doi.org/10.1111/mila.12036
Carlson, S. M., Mandell, D. J., & Williams, L. (2004). Executive function and theory of mind:
Stability and prediction from ages 2 to 3. Developmental Psychology, 40(6), 1105–1122.
https://doi.org/10.1037/0012-1649.40.6.1105
Carruthers, P. (2013). Mindreading in infancy. Mind & Language, 28(2), 141–172.
https://doi.org/10.1111/mila.12014
Chuey, A., & Gweon, H. (2021). Preschool-aged children can use communicators’ influence on
others to infer what they know. Proceedings of the Annual Meeting of the Cognitive
Science Society, 43(43). https://escholarship.org/uc/item/2jg8796p
Clements, W. A., & Perner, J. (1994). Implicit understanding of belief. Cognitive Development,
9(4), 377–395. https://doi.org/10.1016/0885-2014(94)90012-4
de Villiers, J. G., & Pyers, J. E. (2002). Complements to cognition: A longitudinal study of the
relationship between complex syntax and false-belief-understanding. Cognitive
Development, 17(1), 1037–1060. https://doi.org/10.1016/S0885-2014(02)00073-4
102
BELIEF UNDERSTANDING
Dörrenberg, S., Rakoczy, H., & Liszkowski, U. (2018). How (not) to measure infant Theory of
Mind: Testing the replicability and validity of four non-verbal measures. Cognitive
Development, 46, 12–30. https://doi.org/10.1016/j.cogdev.2018.01.001
Egan, A. (2008). Seeing and believing: Perception, belief formation and the divided mind.
Philosophical Studies, 140(1), 47–63. https://doi.org/10.1007/s11098-008-9225-1
Fabricius, W. V., & Khalil, S. L. (2003). False beliefs or false positives? Limits on children’s
understanding of mental representation. Journal of Cognition and Development, 4(3),
239–262. https://doi.org/10.1207/S15327647JCD0403_01
Flavell, J. H. (1988). The development of children’s knowledge about the mind: From cognitive
connections to mental representations. In J. W. Astington, P. L. Harris, & D. R. Olson
(Eds.), Developing theories of mind (pp. 244–267). Cambridge University Press.
Flavell, J. H., Flavell, E. R., & Green, F. L. (1983). Development of the appearance-reality
distinction. Cognitive Psychology, 15(1), 95–120.
https://doi.org/10.1016/0010-0285(83)90005-1
Flavell, J. H., Flavell, E. R., & Green, F. L. (1989). A transitional period in the development of
the appearance-reality distinction. International Journal of Behavioral Development,
12(4), 509–526. https://doi.org/10.1177/016502548901200407
Flavell, J. H., Green, F. L., Flavell, E. R., Watson, M. W., & Campione, J. C. (1986).
Development of knowledge about the appearance-reality distinction. Monographs of the
Society for Research in Child Development, 51(1), i. https://doi.org/10.2307/1165866
Flavell, J. H., Lindberg, N. A., Green, F. L., & Flavell, E. R. (1992). The development of
children’s understanding of the appearance-reality distinction between how people look
and what they are really like. Merrill-Palmer Quarterly, 38(4), 513–524.
Garnham, W. A., & Perner, J. (2001). Actions really do speak louder than words-but only
implicitly: Young children’s understanding of false belief in action. British Journal of
Developmental Psychology, 19(3), 413–432. https://doi.org/10.1348/026151001166182
Gergely, G., Nádasdy, Z., Csibra, G., & Bíró, S. (1995). Taking the intentional stance at 12
months of age. Cognition, 56(2), 165–193.
https://doi.org/10.1016/0010-0277(95)00661-H
Gopnik, A. (1993). How we know our minds: The illusion of first-person knowledge of
intentionality. Behavioral and Brain Sciences, 16(1), 1–14.
https://doi.org/10.1017/S0140525X00028636
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(1), 26–37. https://doi.org/10.2307/1130386
103
BELIEF UNDERSTANDING
Grèzes, J., Frith, C. D., & Passingham, R. E. (2004). Inferring false beliefs from the actions of
oneself and others: An fMRI study. NeuroImage, 21(2), 744–750.
https://doi.org/10.1016/S1053-8119(03)00665-7
Grosse Wiesmann, C., Friederici, A. D., Disla, D., Steinbeis, N., & Singer, T. (2018).
Longitudinal evidence for 4-year-olds’ but not 2- and 3-year-olds’ false belief-related
action anticipation. Cognitive Development, 46, 58–68.
https://doi.org/10.1016/j.cogdev.2017.08.007
Grosse Wiesmann, C., Friederici, A. D., Singer, T., & Steinbeis, N. (2017). Implicit and explicit
false belief development in preschool children. Developmental Science, 20(5), e12445.
https://doi.org/10.1111/desc.12445
He, Z., Bolz, M., & Baillargeon, R. (2011). False-belief understanding in 2.5-year-olds: Evidence
from violation-of-expectation change-of-location and unexpected-contents tasks:
False-belief understanding in 2.5-year-olds. Developmental Science, 14(2), 292–305.
https://doi.org/10.1111/j.1467-7687.2010.00980.x
Hogrefe, J., Wimmer, H., & Perner, J. (1986). Ignorance versus false belief: A developmental lag
in attribution of epistemic states. Child Development, 57(3), 567–582.
https://doi.org/10.2307/1130337
Holmes, H. A., Black, C., & Miller, S. A. (1996). A cross-task comparison of false belief
understanding in a head start population. Journal of Experimental Child Psychology,
63(2), 263–285. https://doi.org/10.1006/jecp.1996.0050
Hrdy, S. B. (2009). Mothers and others. Harvard University Press.
Kampis, D., Kármán, P., Csibra, G., Southgate, V., & Hernik, M. (2021). A two-lab direct
replication attempt of Southgate, Senju and Csibra (2007). Royal Society Open Science,
8(8), 210190. https://doi.org/10.1098/rsos.210190
Katz, L. G., Chard, S. C., & Kogan, Y. (2014). Engaging children’s minds: The project approach.
Bloomsbury Publishing USA.
Knudsen, B., & Liszkowski, U. (2012a). 18‐month‐olds predict specific action mistakes through
attribution of false belief, not ignorance, and intervene accordingly. Infancy, 17(6),
672–691. https://doi.org/10.1111/j.1532-7078.2011.00105.x
Knudsen, B., & Liszkowski, U. (2012b). Eighteen- and 24-month-old infants correct others in
anticipation of action mistakes: Infants correct others’ action mistakes. Developmental
Science, 15(1), 113–122. https://doi.org/10.1111/j.1467-7687.2011.01098.x
Kovács, Á. M., Téglás, E., & Endress, A. D. (2010). The social sense: Susceptibility to others’
beliefs in human infants and adults. Science, 330(6012), 1830–1834.
https://doi.org/10.1126/science.1190792
104
BELIEF UNDERSTANDING
Kret, M. E., Prochazkova, E., Sterck, E. H. M., & Clay, Z. (2020). Emotional expressions in
human and non-human great apes. Neuroscience & Biobehavioral Reviews, 115,
378–395. https://doi.org/10.1016/j.neubiorev.2020.01.027
Kulke, L., Von Duhn, B., Schneider, D., & Rakoczy, H. (2018). Is implicit theory of mind a real
and robust phenomenon? Results from a systematic replication study. Psychological
Science, 29(6), 888–900. https://doi.org/10.1177/0956797617747090
Lally, A. (2023). Holistic development: Children’s health and well-being. In C. Hayes (Ed.), The
early years handbook for students and practitioners (2nd ed.). Routledge.
Leslie, A. M. (1992). Pretense, autism, and the theory-of-mind module. Current Directions in
Psychological Science, 1(1), 18–21. https://doi.org/10.1111/1467-8721.ep10767818
Lewis, C., Freeman, N. H., Hagestadt, C., & Douglas, H. (1994). Narrative access and
production in preschoolers’ false belief reasoning. Cognitive Development, 9(4),
397–424. https://doi.org/10.1016/0885-2014(94)90013-2
Liu, D., Wellman, H. M., Tardif, T., & Sabbagh, M. A. (2008). Theory of mind development in
Chinese children: A meta-analysis of false-belief understanding across cultures and
languages. Developmental Psychology, 44(2), 523–531.
https://doi.org/10.1037/0012-1649.44.2.523
Low, J., Drummond, W., Walmsley, A., & Wang, B. (2014). Representing how rabbits quack and
competitors act: Limits on preschoolers’ efficient ability to track perspective. Child
Development, 85(4), 1519–1534. https://doi.org/10.1111/cdev.12224
Low, J., & Watts, J. (2013). Attributing false beliefs about object identity reveals a signature
blind spot in humans’ efficient mind-reading system. Psychological Science, 24(3),
305–311. https://doi.org/10.1177/0956797612451469
MacLaren, R., & Olson, D. (1993). Trick or treat: Children’s understanding of surprise.
Cognitive Development, 8(1), 27–46. https://doi.org/10.1016/0885-2014(93)90003-N
Masnick, A. M., Klahr, D., & Knowles, E. R. (2017). Data-driven belief revision in children and
adults. Journal of Cognition and Development, 18(1), 87–109.
https://doi.org/10.1080/15248372.2016.1168824
Moll, H., Kane, S., & McGowan, L. (2016). Three-year-olds express suspense when an agent
approaches a scene with a false belief. Developmental Science, 19(2), 208–220.
https://doi.org/10.1111/desc.12310
Moll, H., Khalulyan, A., & Moffett, L. (2017). 2.5‐year‐olds express suspense when others
approach reality with false expectations. Child Development, 88(1), 114–122.
https://doi.org/10.1111/cdev.12581
105
BELIEF UNDERSTANDING
Moll, H., Ni, Q., & Stekeler-Weithofer, P. (2022). Ontogenetic steps of understanding beliefs:
From practical to theoretical. Philosophical Psychology, 1–25.
https://doi.org/10.1080/09515089.2022.2073211
Moll, H., & Tomasello, M. (2012). Three-year-olds understand appearance and reality—Just not
about the same object at the same time. Developmental Psychology, 48(4), 1124–1132.
https://doi.org/10.1037/a0025915
Moore, C., Pure, K., & Furrow, D. (1990). Children’s understanding of the modal expression of
speaker certainty and uncertainty and its relation to the development of a representational
theory of mind. Child Development, 61(3), 722–730.
https://doi.org/10.1111/j.1467-8624.1990.tb02815.x
Moses, L. J., & Flavell, J. H. (1990). Inferring false beliefs from actions and reactions. Child
Development, 61(4), 929–945. https://doi.org/10.2307/1130866
Nelson, K. (2005). Language pathways into the community of minds. In J. W. Astington & J. A.
Baird (Eds.), Why language matters for theory of mind (pp. 26–49). Oxford University
Press.
Ni, Q., Fascendini, B., Shoyer, J., & Moll, H. (2022). No signs of automatic perspective-taking
or its modulation by joint attention in toddlers using an object retrieval task. Royal
Society Open Science, 9(8), 220347. https://doi.org/10.1098/rsos.220347
Ni, Q., Shoyer, J., Bautista, Z., Raport, A., & Moll, H. (2023). Toddlers’ expressions indicate that
they track agent–object interactions but do not detect false object representations. Journal
of Experimental Child Psychology, 231, 105639.
https://doi.org/10.1016/j.jecp.2023.105639
Onishi, K. H., & Baillargeon, R. (2005). Do 15-month-old infants understand false beliefs?
Science, 308(5719), 255–258. https://doi.org/10.1126/science.1107621
Papafragou, A., Cassidy, K., & Gleitman, L. (2007). When we think about thinking: The
acquisition of belief verbs. Cognition, 105(1), 125–165.
https://doi.org/10.1016/j.cognition.2006.09.008
Parrott, W. G., & Gleitman, H. (1989). Infants’ expectations in play: The joy of peek-a-boo.
Cognition and Emotion, 3(4), 291–311. https://doi.org/10.1080/02699938908412710
Perner, J., Leekam, S. R., & Wimmer, H. (1987). Three-year-olds’ difficulty with false belief:
The case for a conceptual deficit. British Journal of Developmental Psychology, 5(2),
125–137. https://doi.org/10.1111/j.2044-835X.1987.tb01048.x
Perner, J., Mauer, M. C., & Hildenbrand, M. (2011). Identity: Key to children’s understanding of
belief. Science, 333(6041), 474–477. https://doi.org/10.1126/science.1201216
106
BELIEF UNDERSTANDING
Perner, J., Rendl, B., & Garnham, A. (2007). Objects of desire, thought, and reality: Problems of
anchoring discourse referents in development. Mind & Language, 22(5), 475–513.
https://doi.org/10.1111/j.1468-0017.2007.00317.x
Perner, J., & Roessler, J. (2010). Teleology and causal understanding in children’s theory of
mind. In J. H. Aguilar & A. A. Buckareff (Eds.), Causing human action: New
perspectives on the causal theory of action (pp. 199–228). The MIT Press.
Perner, J., & Roessler, J. (2012). From infants’ to children’s appreciation of belief. Trends in
Cognitive Sciences, 16(10), 519–525. https://doi.org/10.1016/j.tics.2012.08.004
Qureshi, A. W., Apperly, I. A., & Samson, D. (2010). Executive function is necessary for
perspective selection, not Level-1 visual perspective calculation: Evidence from a
dual-task study of adults. Cognition, 117(2), 230–236.
https://doi.org/10.1016/j.cognition.2010.08.003
Repacholi, B. M., & Gopnik, A. (1997). Early reasoning about desires: Evidence from 14- and
18-month-olds. Developmental Psychology, 33(1), 12–21.
https://doi.org/10.1037/0012-1649.33.1.12
Rubio-Fernández, P. (2019). Memory and inferential processes in false-belief tasks: An
investigation of the unexpected-contents paradigm. Journal of Experimental Child
Psychology, 177, 297–312. https://doi.org/10.1016/j.jecp.2018.08.011
Rubio-Fernández, P., & Geurts, B. (2016). Don’t mention the marble! The role of attentional
processes in false-belief tasks. Review of Philosophy and Psychology, 7(4), 835–850.
https://doi.org/10.1007/s13164-015-0290-z
Ruffman, T., & Keenan, T. R. (1996). The belief-based emotion of suprise: The case for a lag in
understanding relative to false belief. Developmental Psychology, 32(1), 40–49.
https://doi.org/10.1037/0012-1649.32.1.40
Saxe, R. (2006). Four brain regions for one theory of mind. In C. L. Pickett, J. T. Cacioppo, & P.
S. Visser (Eds.), Social neuroscience: People thinking about thinking people (pp.
83–102). MIT Press.
Schidelko, L. P., Huemer, M., Schröder, L. M., Lueb, A. S., Perner, J., & Rakoczy, H. (2022).
Why do children who solve false belief tasks begin to find true belief control tasks
difficult? A test of pragmatic performance factors in theory of mind tasks. Frontiers in
Psychology, 12. https://www.frontiersin.org/articles/10.3389/fpsyg.2021.797246
Schuwerk, T., Priewasser, B., Sodian, B., & Perner, J. (2018). The robustness and
generalizability of findings on spontaneous false belief sensitivity: A replication attempt.
Royal Society Open Science, 5(5), 172273. https://doi.org/10.1098/rsos.172273
107
BELIEF UNDERSTANDING
Scott, R. M. (2017). Surprise! 20-month-old infants understand the emotional consequences of
false beliefs. Cognition, 159, 33–47. https://doi.org/10.1016/j.cognition.2016.11.005
Scott, R. M., & Baillargeon, R. (2009). Which penguin is this? Attributing false beliefs about
object identity at 18 months. Child Development, 80(4), 1172–1196.
https://doi.org/10.1111/j.1467-8624.2009.01324.x
Scott, R. M., He, Z., Baillargeon, R., & Cummins, D. (2012). False-belief understanding in
2.5-year-olds: Evidence from two novel verbal spontaneous-response tasks: Toddlers pass
verbal spontaneous-response false-belief tasks. Developmental Science, 15(2), 181–193.
https://doi.org/10.1111/j.1467-7687.2011.01103.x
Searle, J. R. (1983). Intentionality: An essay in the philosophy of mind. Cambridge University
Press.
Senju, A., Southgate, V., Snape, C., Leonard, M., & Csibra, G. (2011). Do 18-month-olds really
attribute mental states to others?: A critical test. Psychological Science, 22(7), 878–880.
https://doi.org/10.1177/0956797611411584
Song, H., & Baillargeon, R. (2008). Infants’ reasoning about others’ false perceptions.
Developmental Psychology, 44(6), 1789–1795. https://doi.org/10.1037/a0013774
Song, H., Onishi, K. H., Baillargeon, R., & Fisher, C. (2008). Can an agent’s false belief be
corrected by an appropriate communication? Psychological reasoning in 18-month-old
infants. Cognition, 109(3), 295–315. https://doi.org/10.1016/j.cognition.2008.08.008
Southgate, V., Senju, A., & Csibra, G. (2007). Action anticipation through attribution of false
belief by 2-year-olds. Psychological Science, 18(7), 587–592.
https://doi.org/10.1111/j.1467-9280.2007.01944.x
Sprung, M., Perner, J., & Mitchell, P. (2007). Opacity and discourse referents: Object identity
and object properties. Mind & Language, 22(3), 215–245.
https://doi.org/10.1111/j.1468-0017.2007.00307.x
Surian, L., Caldi, S., & Sperber, D. (2007). Attribution of beliefs by 13-month-old infants.
Psychological Science, 18(7), 580–586.
https://doi.org/10.1111/j.1467-9280.2007.01943.x
Surian, L., & Geraci, A. (2012). Where will the triangle look for it? Attributing false beliefs to a
geometric shape at 17 months: Attribution of beliefs to a geometric shape. British Journal
of Developmental Psychology, 30(1), 30–44.
https://doi.org/10.1111/j.2044-835X.2011.02046.x
Takagishi, H., Kameshima, S., Schug, J., Koizumi, M., & Yamagishi, T. (2010). Theory of mind
enhances preference for fairness. Journal of Experimental Child Psychology, 105(1),
130–137. https://doi.org/10.1016/j.jecp.2009.09.005
108
BELIEF UNDERSTANDING
Tomasello, M. (1999). Social cognition before the revolution. In P. Rochat (Ed.), Early social
cognition: Understanding others in the first months of life (pp. 301–314). Lawrence
Erlbaum Associates Publishers.
Tomasello, M., Carpenter, M., Call, J., Behne, T., & Moll, H. (2005). Understanding and sharing
intentions: The origins of cultural cognition. Behavioral and Brain Sciences, 28(5),
675–691. https://doi.org/10.1017/S0140525X05000129
Tomasello, M., & Moll, H. (2010). The gap is social: Human shared intentionality and culture. In
P. M. Kappeler & J. Silk (Eds.), Mind the gap: Tracing the origins of human universals
(pp. 331–349). Springer. https://doi.org/10.1007/978-3-642-02725-3_16
Träuble, B., Marinović, V., & Pauen, S. (2010). Early theory of mind competencies: Do infants
understand others’ beliefs? Infancy, 15(4), 434–444.
https://doi.org/10.1111/j.1532-7078.2009.00025.x
Trautner, H. M., Gervai, J., & Németh, R. (2003). Appearance–reality distinction and
development of gender constancy understanding in children. International Journal of
Behavioral Development, 27(3), 275–283. https://doi.org/10.1080/01650250244000362
Wellman, H. M., & Banerjee, M. (1991). Mind and emotion: Children’s understanding of the
emotional consequences of beliefs and desires. British Journal of Developmental
Psychology, 9(2), 191–214. https://doi.org/10.1111/j.2044-835X.1991.tb00871.x
Wellman, H. M., Cross, D., & Watson, J. (2001). Meta-analysis of theory-of-mind development:
The truth about false belief. Child Development, 72(3), 655–684.
https://doi.org/10.1111/1467-8624.00304
Wellman, H. M., & Woolley, J. D. (1990). From simple desires to ordinary beliefs: The early
development of everyday psychology. Cognition, 35(3), 245–275.
https://doi.org/10.1016/0010-0277(90)90024-E
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining function
of wrong beliefs in young children’s understanding of deception. Cognition, 13(1),
103–128. https://doi.org/10.1016/0010-0277(83)90004-5
Wimmer, H., & Weichbold, V. (1994). Children’s theory of mind: Fodor’s heuristics examined.
Cognition, 53(1), 45–57. https://doi.org/10.1016/0010-0277(94)90076-0
Woodward, A. (1998). Infants selectively encode the goal object of an actor’s reach. Cognition,
69(1), 1–34. https://doi.org/10.1016/S0010-0277(98)00058-4
Woodward, A. (2009). Infants’ grasp of others’ intentions. Current Directions in Psychological
Science, 18(1), 53–57. https://doi.org/10.1111/j.1467-8721.2009.01605.x
109
BELIEF UNDERSTANDING
Wu, Y., & Schulz, L. (2021). Children can use others’ emotional expressions to infer their
knowledge and predict their behaviors in classic false belief tasks.
https://doi.org/10.31234/osf.io/w9d2c
Zillmann, D. (1981). Anatomy of suspense. In P. H. Tannenbaum (Ed.), The entertainment
functions of television. Psychology Press.
110
BELIEF UNDERSTANDING
Appendix
Coding Scheme
1. Facial Expressions
Lip bite: bites on the lower lip, with teeth visible or not
Lip pout: lower lip curls down, with or without ’orange peel chin’ appearing
Lip curl/lip tension: upper and/or lower lip tightens, or upper lip curls up
Furrowed eyebrow: eyebrows lower creating vertical lines and other wrinkles on the forehead
Raised brow: eyebrows move upward creating horizontal lines across the forehead
Mouth opens or closes: mouth suddenly opens or (if the baseline is an open mouth) suddenly
closes
Mouth tightening: muscles around mouth stretch and tighten, with lips slightly thinning
“Fish mouth”: pulls lips together to create a fish mouth
Smirk: smiles with a smirk
Hand(s) over/inside mouth: hands are brought to or inside the mouth
Hand(s) over ears: hands are brought to ears
Hand(s) out of mouth: hand moves out of mouth area suddenly and mouth appears (if hand
covered mouth at baseline)
Foot inside mouth: foot is placed inside mouth
Head shake: moves head left and right quickly
2. Bodily Expressions
Body tightening: bringing shoulders near chest while rising in chair
Pointing: gesturing excitedly toward the stage
Raised hand: raises one/two hands, similar to a hand gesture of stop
111
BELIEF UNDERSTANDING
Bouncing: moves body up and down or in a rocking motion, arms/legs may move as well
Shrugging shoulders: shoulders quickly move up and down
Neck tension: shows sudden neck muscle tightness with head slightly moving forward or
backward
Standing up: child suddenly stands up from chair
Sitting up: changes from slouched position to sitting up straight
Pushing back into chair: child pushes body back into the chair
112
Abstract (if available)
Abstract
This dissertation explored the scope and limitations of young children’s understanding of others’ mental states. In three consecutive studies, I investigated 3- to 5-year-olds’ understanding of others’ action-relevant beliefs or expectations, using a combination of classic (question-answer format) and novel (facial expressions) measures. The first study (N = 120) found that 3-year-olds can anticipate an agent’s false expectations about a container with unexpected contents only when they can track the agent’s prior experience with objects. This was revealed through their expressions. In verbal judgments, children showed no understanding of the agent’s false expectations. The second study (N = 90) further examined toddlers’ need to track others’ perceptual experiences. The third study (N = 120) introduced a novel approach to understanding the relationship between beliefs and actions by asking children to reason backward from observed actions to beliefs. It revealed a strong relation between backward and forward reasoning about beliefs, as well as a developmental progression between 3 and 5. Taken together, the findings show that 3-year-olds possess limited theory of mind skills: They fail to understand that an object’s appearance can induce false beliefs or expectations, and the only kind of belief-like states—expectations—that children this young can grasp are those grounded in direct prior experience with the object. This dissertation supports a middle position on the development of theory of mind, according to which a theory of mind is neither innately given nor late-emerging.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Do 3-year-olds understand that appearances can ‘deceive’ and lead to false beliefs?
PDF
Children's use of social and visual perspective-taking in reference resolution
PDF
Young children selectively transmit true information when teaching
PDF
Do you see what I see? Personality and cognitive factors affecting theory of mind or perspective taking
PDF
Social attributes of adolescents’ problem-talk partners predict changes in depression
PDF
Opportunity beliefs and behavioral outcomes in Latinx youth: an exploration of Ogbu’s cultural-ecological model
PDF
Young children’s demand for reciprocal engagement: evidence from typically-developing and autistic preschoolers
PDF
Children's ""mutuality demand"" in seeing, hearing, and speech
PDF
Zooming in on mindfulness and gaming: the application of a brief, standardized, zoom-based, mindfulness intervention on videogame play as compared to a treatment as usual approach
PDF
Anthropomorphic sociality theory: how connections to nonhumans connect us to humans
PDF
Reasoning with degrees of belief
PDF
Assessing the psychological correlates of belief strength: contributing factors and role in behavior
PDF
Understanding Native American women’s beliefs, values, and expectations for engagement with the Girl Scout organization
PDF
Understanding academic advisor training and professional development experiences
PDF
Beyond active and passive social media use: habit mechanisms are behind frequent posting and scrolling on Twitter/X
PDF
How misinformation exploits moral values and framing: insights from social media platforms and behavioral experiments
PDF
#BLM or #ALM: accessible perspective shapes downstream judgment even among people high in social dominance
PDF
Young children’s emerging ability to make false statements
PDF
The relationships between teacher beliefs about diversity and opportunities for culturally and linguistically diverse students, reflectiveness, and teacher self-efficacy
PDF
Development of error monitoring in preschool to 12th-grade students and relations in late childhood and adolescence to social-affective processing and emotions about math
Asset Metadata
Creator
Ni, Qianhui
(author)
Core Title
The scope and limitations of young children’s belief understanding
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Psychology
Degree Conferral Date
2024-05
Publication Date
04/03/2024
Defense Date
03/19/2024
Publisher
Los Angeles, California
(original),
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
belief understanding,OAI-PMH Harvest,perspective-taking,social cognition,theory of mind
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Moll, Henrike (
committee chair
), Hackel, Leor (
committee member
), Schachner, Adena (
committee member
), Schroeder, Mark (
committee member
), Schwartz, David (
committee member
)
Creator Email
niqianhui1208vicky@gmail.com,qni@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC113867466
Unique identifier
UC113867466
Identifier
etd-NiQianhui-12754.pdf (filename)
Legacy Identifier
etd-NiQianhui-12754
Document Type
Dissertation
Format
theses (aat)
Rights
Ni, Qianhui
Internet Media Type
application/pdf
Type
texts
Source
20240404-usctheses-batch-1135
(batch),
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
belief understanding
perspective-taking
social cognition
theory of mind