Mentalization the development of children's theory of mind

Document Sample
Mentalization the development of children's theory of mind Powered By Docstoc
					Learning II
Social learning

  Cognitive Psychology
  10 November, 2010
Topics for today
(1) Mentalization and false belief: some background
(2) Precursors of mental state attribution
(3) Constituents of the theory of mind and their development
(4) Varieties of the false belief task
(5) Theories of children‟s theory-of-mind development
(6) Mentalization in apes
1) The problem of other minds
How do we know that other people
- have mental states comparable to
- are conscious beings like us?

For now, put the philosophical
  paradox aside

 We can infer mental states from
  behavior: if John says ”I like pizza”
  then it is plausible to suppose that
  he likes pizza.
Background, continued
2) David Premack and Guy
    Woodruff (1978) on chimps‟
    theory of mind.
Theory of mind  the individual
    attributes mental states to
    oneself and others.
 This is a theory because
(i) it assumes states that are not
    directly observable              Sarah, looking at one of the tools
                                     that she could pick to solve the
(ii) it is used for predicting       problem in the video she saw
Childhood realism (Piaget)
"Do You know what it means to think?"
 "Yes"
  "Then think of your house. What do you think with?"
  "The mouth."
  "Can you think with the mouth shut?"
  "With the eyes shut?"
  "Now shut your mouth and think of your house. Are you
  "What did you think with?"
  "The mouth."
How far can we get with realism?
Attributing true belief
    There is a
    over there

-- Suppose there really is a storm
The attributor (on the right): he notices the same aspect of reality and that
the subject pointed it out to him.
”The subject‟s belief is a part/aspect of reality: the storm”
Point: The attributor need not attribute a mental state to the subject
How about attributing false belief?
     There is a golden                        what is he talking
    cube the size of a                            about?!
   cubic mile over there

There is a discrepancy between reality and what the subject is saying
The only way to interpret the subject is to attribute a thought to him
  (that is false, although meaningful).
This cannot happen unless reality and representation are thought of as
  distinct  realism is no longer an option
Forming an idea of someone else’s thought  a metarepresentation
When do children start attributing
false belief?
 Heinz Wimmer and Josef Perner
  (1983): Max and the chocolate
The Smarties box (Perner, Leekam
  és Wimmer, 1987 )

Results: 3-yr-olds cannot, whereas
  4-5-yr-olds can solve the task.
  i.e., attribute false belief to
Topics for today
(1) Mentalization and false belief: some background
(2) Precursors of mental state attribution
(3) Constituents of the theory of mind and their development
(4) Varieties of the false belief task
(5) Theories of children‟s theory-of-mind development
(6) Mentalization in apes
Some precursors of mental state
attribution in infancy
(1) Face recognition
 by the end of the first year, infants can even recognize some
   facial expressions (e.g., happiness).
 A cue in emotion attribution.
 The recognition of biological motion
 From about three months of age
 A cue in recognizing agents as opposed to inanimate objects
 Agents are capable of self-propelled motion
Joint attention
The ability to follow the caretaker‟s gaze and look where she
  is looking (~12 months)

Social reference:
The use of joint attention in word

On hearing a new word,
18-month-old infants attach it to
the object that the utterer is
looking at when using the word.
Protoimperative and protodeclarative
 Which is which on the

 Different implications for

 Other forms??
Proto-interrogative (Csibra &
  Gergely, 2010)
Topics for today
(1) Mentalization and false belief: some background
(2) Precursors of mental state attribution
(3) Constituents of the theory of mind and
  their development
(4) Varieties of the false belief task
(5) Theories of children‟s theory-of-mind development
(6) Mentalization in apes
Achievements in social cognition
around nine months (”The social-
cognitive revolution”)
 A new understanding of
  goal-directed action: ”the
  differentiation of means from
 Piaget already noted that at
  this age infants realize that
  different actions can lead to
  the same outcome.
 The jumping ball experiment
  (Gergely, Csibra et al.)
Means and ends at 14 months
 Turning on the night light
  with one‟s forehead –
  Meltzoff and Tomasello
 14 month-olds repeated
  the action after a delay of
  one week.
 This is an apparently
  irrational action.
 If infants can indeed
  differentiate goals from
  actions to achieve them at
  nine months, then…
 …why don‟t 14-month-old       Gergely et al., 2002
  subjects choose a ”more
  rational” action to turn on
  the night light?
Why separate the teleological stance
from the intentional one?
 Gergely:
(1) For reasons of parsimony  it is enough to attribute goals,
   externalistically specified, to the agent, in order to explain its

(2) At 9-12 months virtually no tests of mentalization come out
   positive (think of the Onishi-Baillargeon expt which
   demonstrated false belief understanding in dishabituation
   setting, with 15-month-olds – and that was already a little
Later achievements
Pretend play
Leslie (1987): production and
  understanding of pretence arise
  simultaneously (at age 2)

A complex activity
-- managing a ”double” criterion of
   truth (genuine and imaginary);
-- memorizing scenarios with
   imaginary elements,
-- coordinating action with peers, and
   so on.
Perception and perceptual error

A sponge that looks like a rock (Flavell and Green, 1983)
How does this task relate to the false belief task?
How does the false belief task relate to Gettier‟s
  counterexamples to Plato‟s definition of knowledge?
 (3) Desires
 Repacholi and Gopnik (1997): the
  broccoli-cookie exeriment 18-month-
  olds were able to pick the food that
  the model liked (even if they
  themselves disliked it).

 Two-year olds understand a lot about
  desires (Wellman, 1995)
Why does the understanding of false
belief come later than that of desire
and the appearance-reality
 Intense phenomenology (i.e., feeling
  or conscious experience) – Pillow,

 Desires are more directly connected
  with behavior (attempts, requests,

 Understanding belief presupposes
-- involvement in communication,
-- inference that others have thoughts. –
    Harris, 1996
(4) Intention

 Janet Astington: before age 4,
  children cannot differentiate
  intention (as a mental state) from
  the corresponding action and its
Topics for today
(1) Mentalization and false belief: some background
(2) Precursors of mental state attribution
(3) Constituents of the theory of mind and their development
(4) Varieties of the false belief task
(5) Theories of children‟s theory-of-mind development
(6) Mentalization in apes
False belief: varieties of the task
 The first experiments suggested
  that 3-yr-olds cannot attribute false

 It appears though that in easier
   versions they can.
-- performance limitation?

Karen Bartsch and Henry Wellman
  (1989): explanation vs. prediction

Michael Siegal and K. Beattie: asking
  a slightly different question helps at
  age 3
The role of performance limitations in
false-belief tasks
 How could we explain these findings?

- prediction (as opposed to explanation)
- handling linguistic complexity
- memory span
- the salience of the actual state of affairs (it is hard to suppress
    knowledge of the actual state of affairs)

These factors are not, strictly speaking, part of the
  understanding of false belief, still they affect performance in
  relevant tasks.
Implicit theory of mind in infants?
 Kristine Onishi and Renée Baillargeon (2005): Tested by the
  dishabituation method, 15-month-old infants exhibited false-
  belief attribution.
Belief induction
Recall – where is the toy?

True belief trials: the actor
  sees the move/transfer

False belief trials: The actor
  sees no transfer (first) or
  only the first of the two
Test trial
 Four belief conditions (A, B,
   C, D):
-- half of each received the
   green box test trial, the other
   half the yellow box trial.

What guided the infant‟s
-- her own belief?
-- The actor‟s belief?
TB green  infants are
  surprised by reach into the
  yellow box
TB yellow: green box seems

FB green: reach into the yellow
  box surprises the infant who
  is aware that the toy is in the
  yellow box
FB yellow: the same result
Explanation: can be based on
  the actor‟s belief understood
  by the infant
Sure 15-month-olds can attribute false
 There is a more parsimonious interpretation:
 Associative learning!
Consider this rule: IF the infant sees the actor and the toy move
  at the same time, THEN they subsequently expect the actor
  to search for the object where the toy‟s movement ended;
  ELSE they expect the actor to search at the previous
THAT IS, in the infant‟s memory, the actor is associated with
  that box at which they last saw together (i) the actor, and (ii)
  the toy disappear.

 This principle does not assume that infants at this age can
 attribute false belief.
Behaviorism across the board?
 Does this associationist interpretation apply to other forms of
  the FB task as well? Or is there a difference?
 Thank God, no – there is a difference.

 Preferential looking is a very simple sort of response. We
    have already seen its chief advantage: ... is unlikely to cause performance limitations.
 BUT it has a drawback too.
 Due to its simplicity, it conveys minimal information, hence it
    is little help in ruling out alternative interpretatiotns.
 Verbal responses (e.g., Bartsch & Wellman): the child
    explains the protagonist‟s behavior using words like „think‟
    and „believe‟, etc.  this makes the belief attribution
    interpretation a lot more plausible.
Topics for today
(1) Mentalization and false belief: some background
(2) Precursors of mental state attribution
(3) Constituents of the theory of mind and their development
(4) Varieties of the false belief task
(5) Theories of children’s theory-of-mind
(6) Mentalization in apes
(1) Theory-theory

In the course of development, children gradually develop a
   theory that serves social understanding.

Why call it a theory? Because it satisfies some general criteria
    for being a theory:
- It is about entities that are not directly accessible (i.e., mental
- It presupposes law-like connections between its entities
- It serves explanation and prediction of behavior
- It is constructed, and not directly experienced
- It is subject to modification or rejection in case of powerful
    evidence against it (defeasibility)
The development of children’s theory
of mind
 Children develop their theory of mind gradually

2 years: a non-representational conception of perception and
- desires are drives toward certain objects
- perception is a direct contact with objects (no appearance-
   reality distinction)
- desires play a central role in 2-yr-olds‟ psychology
3 years:
- extending mental vocabulary (know, dream, think, remember)
- understanding the distinction between dreams, imagination,
    and reality
- the appearance-reality distinction is still not in (Flavell‟s
    sponge and rock experiment)

- a preliminary notion of belief:
    a direct, “mirroring” relation between belief and reality
   the child cannot yet handle false belief – belief corresponds
   to reality
    3-yr-olds also don‟t understand that there can be different
   sources of belief (which would be another way to appreciate
   the separation between representation and reality)
Sources of belief and knowledge
 Direct perception; others‟ report;
  inference, etc.

 3-yr-old children do not
  understand these notions
 Gopnik and Graf (1988)

 Thought bubbles in comics – 3-4-
  yr-olds get the point (S. Hulme)
 Further theoretical development
4 to 5 years:
-- children come to understand       Morning star   Evening Star
   representational error – namely
   that what someone thinks or
   believes may not be true

-- Josef Perner views this as
   making the Fregean distinction
   between sense and reference:

-- error occurs when an object of
   reference is picked out by the
   wrong description
Sense and reference
Handling the false belief task by this distinction:
That thing (in the box) is candy
That place (where the chocolate is) is in the cupboard

Italic: reference; bold: the description applied to it.

Understanding that the same object can be described correctly
  and incorrectly is the key to solving the false belief task,
  according to Perner

In sum, we can see that there is substantial
  conceptual (theoretical) development wrt
(2) Simulation theory
 We have privileged access to our own mental states via
 We know how we would react in a range of situations

 When we see someone else in a situation, we imagine how
  we would react there, and our prediction of the model‟s
  behavior will be the output of this imaginary exercise, or
  “mental simulation”.
Examples of mental simulation
1. Judgments of grammaticality:
 The correctness of sentences is
   decided not via an explicit theory, but
   the speaker herself reacting to the
   input (running her own implicit program
   for processing grammar).

2. Emotions: seeing a model in a
   situation, we can imagine how we
   would be affected in that situation
Introspection in the theory-theory
 The ability to introspect is a matter of development, and first-
  person access to our mental states is, at least in part, a
  product of the development of third-person, social

 There is a parallel between understanding others vs.
  understanding ourselves.

 BTW, nor is adult-level introspection infallible. (Nisbett and
  Wilson, 1977).
Theories and simulation
There are pretty obvious cases of mental simulation.

Toward a consensus: mental simulation needs to be
    supplemented by inference and reasoning
-- see a suffering subject
-- feel awful (empathy)
-- INFER THAT the suffering subject feels similarly, i.e., awful.
(3) Nativism and domain specificity
 Jerry Fodor and Alan Leslie
Fodor: there is no conceptual gap between
  children‟s and adult‟s folk psychologies.

The difference between children and adults,
    on this view:
-- There occurs a differentiation of
    propositional attitudes (adults have hopes,
    fears, conjectures, suspicions, etc.)
-- Performance limitations are responsible for
    the relatively late success in the false
    belief task.
Domain specificity: an experiment
 A. Leslie (1994): there exists a special-purpose theory of
  mind mechanism that matures during the first two years of
  life and can be selectively damaged in disorders like autism.
 The Sally-Ann task:

   A change: Sally puts her marble in the basket, then goes out.
   A polaroid camera records the marble resting in the basket.
   Ann moves the marble over to the box
   Question: Where is the marble, in the Polaroid picture?
The camera task…
…includes a conceptual change from (false) belief to (out-of-
  date) photograph.
However, otherwise the logical structure remains the same.
Results:      3-yr-olds fail both tasks (FB; camera)
              4-yr-olds manage both tasks
              autistic children fail FB, but pass the camera task
Leslie‟s conclusion: two mechanisms at work: ToMM and a
  Selection Processor
                pretence                      standrad false
                           standard FB        representation

                                                                        ToMM   SP
                                         SP              Non-      3 yrs yes    no
                   ToMM                                 mental     4 yrs yes   yes
                                                       represen-   aut. no     yes
Apes and
Apes and mentalization tasks
 Complicated results; opinions divide whether chimps and
  orangutans understand false belief.

Call and Tomasello (1999): a standard non-verbal false belief
  task: chimps, orangutans, and children.

Hare and Tomasello (2004): chimps have difficulty using
  cooperative signs.

Chimps can understand what others can and cannot see/know
  (Hare et al., 2000).
Apes may be capable of complex social understanding skills
 Autism is a developmental disorder
  that appears to specifically affect
  theory-of-mind development
 Autism has various forms, but
  some areas of development are
  affected, primarily social behavior
  and communication

The diagnosis is difficult to establish
  before 3-4 yrs of age, although the
  lack of joint attention and pretend
  play at ~18 months are powerful
  predictors (Simon Baron-Cohen et
  al., 1996)
 Autistic children are good at certain tasks (rote memory;
  learning and applying mechanistic rules; some geometrical
 …but impaired on others (understanding verbal material;
  ordering the pictures of a comic strip)

 Although their digit span is good, autobiographical memory is
  less so

 Narrow attention span; shifts of attention are difficult to
Language: phonological processing is intact; syntax appears
  preserved as well
What suffers is communication, semantic processing (e.g.,
  mixing up personal pronouns), and metaphorical use.
A detour: savant syndrome
 Roughly one per cent of people living with autism have
  savant abilities (not only autistics can be savants though)

 Lower than average general intelligence accompanied by
  “islands of extraordinary ability” – performance in a narrow
  domain is high either relative to the subject‟s general
  intelligence or, in some cases, even relative to the general

 Memory skills; calendar calculation; mental calculation or
  “lightning calculation” (arithmetic; prime factorization);
 Music (memorizing pieces, playing an instrument – almost
  without exception, it is the piano)
 Painting, arts and crafts
Richard Wawro, Scottish painter

  James Henry Pullen, British
  carpenter (1835-1916)

Top left:
The Mystic Represen-
tation of the World as
A Ship
Top right:
A model of Isambard
Brunel‟s Great Eastern
Bottom: a collage of
cigar bands
The following factors have been invoked to account for different
  savant abilities:
 Eidetic imagery (calendar calculation?)
 Rote memory (remembering the weather, and other memory feats)
 Implicit rule learning – focusing attention on tasks that we would
  find boring, and doing so for an extended period (mental
 Like natural language syntax, which is a very complicated rule
  system, rules of calculation can also become a rapid automatized
  skill, if there is motivation and persistence on the part of the
 Arts and music presuppose more than these…
 Darold Treffert: Extraordinary People: Understanding Savant
Back to autism in general: three kinds
of explanation
(1) Impairment of social cognition
-- Limited ability to interpret facial expressions
-- Imitation is absent in infancy (Meltzoff and Gopnik)
-- Joint attention at 18 months is missing
-- Pretend play is a matter of controversy

Alan Leslie: understanding of mental representation is
  specifically affected – other representations are handled
  much better
Leslie and Thaiss (1992): autistic children who have trouble
  with false belief attribution, but they have no problem
  understanding outdated photographs
Problems with the social cognition
account of autism
P1. Using different methods, 20-
    50 per cent of autistic children
    pass the false belief test.
P2. The perseverative,
    stereotypical behavior of
    children with autism is not
    covered by the social cognition
What is definitely a problem in
    autism is things like
(i) inhibiting dominant but
    currently inadequate
(ii) action planning,
(iii) shifting attention
What such tasks require is…
 …frontal lobe involvement (e.g, according
   to fMRI studies, the DLPC was active
   during card sorting with WCST)
 So the second theory of autism is:
(2) Executive function disorder
Note that this theory has resources to account
   for failure in the false-belief task:
 The child has to inhibit her own knowledge
   of the actual location of the chocolate, say.
   If the inhibition of predominant and
   inadequate response tendencies is a key
   problem in autism, then failure in the
   original false-belief task is precisely what
   we should expect.
A related account
(3) Weak central coherence (Uta Frith and Francesca Happé)
Autistic subjects can remember incoherent lists of words and
   meaningful sentences equally well.
- They focus on details instead of “seeing the big picture”, or
   reconstructing the course of events in a story, for example
- More reliance on rote memory
- Perceptual and verbal-semantic tasks can both be used to
   demonstrate limitations of cognitive organization
The development of Self
and Agency
The self as a mental versus physical
 The way infants conceive of the physical self is that they
  develop an early understanding of the relation between their
  bodies, their actions, and the environment.
 Sometime between 2-6 months of age, infants differentiate
  their own bodies and the environment – more exactly,
  between stimuli that belong to themselves vs. to the external
 A sense of physical agency involves some representation of
  the causal relationship between the self and its actions on
  the one hand, and the actions and the environment on the
 The self as a mental agent is the result of even more
  complex developmental processes that eventually lead to a
  complex understanding in terms of mental state attribution.
The development of physical agency
 A key achievement on the part of the infant is to develop a
  sensitivity to contingencies

 Mobile kicking at 2 months
 John S. Watson and L. E. Bahrick (1985): the two-monitors

Self and other are differentiated based on perfect vs. imperfect
  stimulus contingency.
 A mechanism for contingency-detection (possibly a
  module?) develops by the fifth month of life
 According to Piaget‟s observations, when do secondary
  circular reactions occur?
The early development of social
 Communication between newborns and their mother

 Early sensitivity to faces
 Familiarization with the mother‟s voice in utero, and
  recognizing it after birth
 Neonatal imitation of facial expressions
 Motherese – ”caregiver-ese”
 Protoconversational turn-taking

 Affective regulation: vocal and facial mirroring of the infant‟s
Affect mirroring
 In interacting with their infant, parents display marked
  exaggerated, partially imitative gestures – reflections of the
  infant‟s affective behavior.
 The infant may well register these episodes as different from
  other expressions of the parent…
 …and come to discover her own contingent control over
  such parental displays.

 Special parental displays and control over them may then
  sensitize the infant to the internal cues that accompany her
  emotional state expressions.
 These external and internal correlates of the infant‟s initially
  reflexive emotional states are the first elements of
Achievements in social cognition
around nine months (”The social-
cognitive revolution”)
Can there be a dissociation between
the teleological stance and the
intentional one?
 There seems to be evidence for such a dissociation
(1) Autism. Frith et al. (2006). Children with autism and
   matched controls were presented with three kinds of
   animations involving abstract figures (triangles, etc.):
-- Random movement
-- Goal-directed movement (following, chasing, fighting)
-- Theory-of-mind interpretable events (hiding and surprising,
   coaxing, mocking, seducing)
Adults used adequate descriptions. Matched controls did too;
   they used mentalistic, teleological, and physical vocabulary
   as expected.
High-functioning autists showed a deficit in providing
   mentalistic descriptions, whereas they performed like the
   controls in the other two conditions.
(2) Enculturated apes. Chimpanzees raised in human families
   exhibited protoimperative pointing and imitative learning (as
   a means of acquiring novel actions).
However, protodeclarative pointing and intentional teaching of
   offspring were not observed in chimpamzees.
(The evidence for chimps teaching their children is
   controversial. Washoe was reported to teach her son Loulis
   sign language to some extent. Still it is a matter of debate
   whether this is comparable to the teaching by human

Autistic children also produce and understand protoimperative
  pointing, but fail either to produce or to understand
  protodeclarative pointing.
Intentional understanding in the
second year
 2-yr-olds use mental vocabulary in speech (mostly the word

The Repacholi-Gopnik experiment (broccoli vs. Goldfish
 18-month-old children gave the model the food she
 14-month-olds gave her the food that they themselves liked
  (i.e., crackers).
 Empathic reactions of concern that lead to prosocial acts like
  helping – emerge in the second year
 Certain inferences about emotions (e.g., an unfulfilled desire
  leads to sadness, vs. a fulfilled one to happiness) – come at
  the same time
Establishing the categorical self
 This is also an achievement of the second year
 Intentional actions and attitudes repeatedly expressed
  toward the child by caregivers and peers gradually become
  as the inferential basis for permanent intentional properties
  attributed to the self, in an attempt to rationalize the partners‟
  self-directed behavior.

 Self-recognition in mirror: children exhibit it from the end of
  the second year on
 However, since chimpanzees and autistic children show it as
  well, self-recognition in the mirror should not imply intentional
 Some additional factors are needed
Development beyond the second year
Correlated development of the following skills
 Understanding the representational nature of beliefs (~4 yrs.)
 Executive function tasks

 Causal self-referentiality (Perner, 2000): mental states are
    understood not only in terms of their content, but also in
    terms of their causal properties, namely that
(i) a belief represents a state of affairs, PLUS…
(ii) … it causes certain adequate actions.
Similarly, when recalling an event from memory, a proper
    understanding of the source of the memory includes the idea
    that the memory itself was caused by the event thus
 Remember: Perner demonstrated that children before 4-5 yrs
The autobiographical self
 The dual, causal-and-intentional understanding of mental
  states provides the ground for awareness of memories as
  personally experienced.
 Daniel Povinelli: before 4-5 yrs, children have serious
  difficulty organizing self-related experiences into a coherent
 This may be part of the reason for infantile amnesia (i.e. that
  we have virtually no conscious memories of our life prior to
  3-4 yrs of age.

 Delayed video feedback experiment

Shared By: