Transcript No Slide Title

What Neurodevelopmental
Disorders Can Reveal about
Cognitive Architecture
Helen Tager-Flusberg
Boston University
Outline
• Discussion of whether neurodevelopmental
disorders has potential to inform debates
about cognitive architecture
(methodological issues and complications)
• Example of research on theory on mind in
NDD: autism and Williams syndrome
• Reinterpretation of the classical evidence
Defining Neurodevelopmental
Disorders (NDD)
• Focus on genetically-based NDD
• NDD divide into gene/chromosome
disorders (e.g., Down, Turner, Williams,
PKU) and polygenic/complex disorders
(e.g., autism, SLI)
• Genetic abnormalities (absence, mutation)
disrupt normal development of brain
• Lead to different ‘phenotypes’ that are
syndrome specific
Role of Disorders in Cognitive
Science
• Lengthy tradition of using data from people
with brain/cognitive damage to inform
cognitive theory, and organization of
cognitive system
• Examples from aphasia, amnesia, agnosia
etc.
• A for Acquired disorders - provides rich
information about dissociations between
cognitive modules from the “end state”
Can Individuals with Abnormal
Brain DEVELOPMENT Inform
Cognitive Science?
• Disruption in brain development affects the
ACQUISITION of cognitive systems
• Studies thus provide data on developmental
processes; can study end state of this
developmental process in adults with NDD
• Controversial whether NDD can provide
evidence about cognitive architecture, even
from a developmental perspective
Arguments Against: KarmiloffSmith (1998)
• Because brains develop DIFFERENTLY in
NDD
– But little evidence for major differences in
developmental processes
– Not a logical argument against using NDD
evidence
Arguments Against (contd 2)
• NDD does not create localized lesion; no
simple intact/impaired cortical circuits that
map directly onto cognitive modules
– May be true for some disorders but not others
– Depend on the NDD, the cognitive system, the
individual
Arguments Against (contd 3)
• Developmental changes in the phenotype
lead to fundamentally different
organization; plasiticity in structure
/function
– Data on phenotypic changes from infancy are
controversial and suspect
– Plasticity is highly constrained; the brain is not
a blank slate!
Alternative View • Arguments proposed by K-S not
compelling.
• May depend on which NDD investigate
• Striking difference in children with different
NDD - and subgroups within syndromes
• Depends on which cognitive module
• Identifying specific cognitive module that
has been disrupted in NDD has been elusive
Methodological Issues in
Studying NDD
• Heterogeneity in expression of phenotype
within each syndrome
– Individual differences related to genetic
variation related to NDD
– Individual differences related to genetic
variation NOT related to NDD
• Differences related to experience
– Age effects
– Remediation/Compensation
Effects of Retardation and
Language Limitations
• Task performance will be a function of
– MR (g - cognitive efficiency?) - non-modular
influences
– Language (comprehend instructions; task
content)
• These effects will potentially obsure
syndrome-specific effects on cognitive
module
• At the least, include subjects in limited age
range, measures of IQ and language ability
Example: Theory of Mind (ToM)
• Since ToM entered the cognitive ‘cannon’
evidence from NDD has been central in
arguments about its organization,
modularity and innateness
• Classic studies of autism; more recent
studies of other syndromes e.g., Williams
ToM in Autism
• Baron-Cohen, Leslie & Frith (1985)
demonstrated failure on FB
• Many studies replicate their findings on FB
and other related tasks; compared to control
subjects and control tasks
• Support for view of selective impairment to
ToM in autism
• Taken as evidence that ToM in unitary
module under genetic influence with unique
neurobiological substrate
Significance of ToM Hypothesis
of Autism
• Provides a unified explanation for core
social and communication symptoms in
terms of an underlying cognitive module
• Selectivity of cognitive deficit identifies the
specificity of impaired cognitive module
• Conforms with what is known about
neurobiological substrate
ToM in Williams Syndrome
• Contiguous gene deletion (~25-30 genes on
7q11.32)
• Constellation of physiological, cranio-facial
and cognitive profile (including MR)
forming a unique phenotype
• Unusual interest in people, sociable, good
language and face recognition
• Suggests spared ToM
• Karmiloff-Smith et al (1995) offer
supporting evidence
ToM in NDD
• Claim for double dissociation in ToM
• Autism:
impaired ToM module
• Williams:
spared ToM module
Problems with this Classic
Picture
• Criticisms of ToM hypothesis of autism
• Problems with the evidence of impairment
in autism and sparing in WMS
– Methodological issues
• Role of language and MR
• Question the basic view that ToM is directly
and selectively affected by NDD
Criticisms of ToM Hypothesis of
Autism
• ToM deficits are secondary to other domain
general deficits such as “executive
functions”
• Cannot explain full range of symptoms in
autism
• Symptoms of autism predate ToM
• Children with other NDD (e.g., Down
syndrome SLI) also fail ToM tasks
Problems with the Evidence
• Some children with autism PASS ToM tasks
(e.g., 20% in Baron-Cohen et al., 1985;
much higher in our studies)
• Our studies on ToM in young children (age
4-8) with Williams syndrome show that
they are NO BETTER (or worse) on FB and
related metarepresentational tasks than
control children matched on age, IQ, and
language
Reconciling the Evidence
• Performance on FB strongly related to
language level (not nonverbal IQ) in broad
range of children with NDD (and normal
children)
• Computing the contents of propositional
attitudes (e.g., beliefs) measures a
representational capacity that can be
acquired by language
ToM in Autism
• Fundamental impairments in the core
module that computes information about
mental states directly from perceptual inputs
(e.g., face, voice, gesture)
• Impaired in knowledge of concepts of
mental states
• IF have linguistic knowledge, then can
access metarepresentational level using this
alternative route
ToM in Williams Syndrome
• Fundamental sparing in the core module
that computes information about mental
states directly from perceptual inputs (e.g.,
face, voice, gesture)
• ? Relative sparing of knowledge of concepts
of mental states
• MR decreases cognitive efficiency
therefore, metarepresentational level also
depends on support from language, and
therefore not spared compared to controls
Language and FB
• Role of sentential/tensed complements in
the acquisition of FB
• Communication verbs - say, tell, whisper
• Cognition verbs - think, know, forget
– John said that he went shopping
– Mary thought that John was sleeping
– Larry believed that John was working
• John went to the movies!
Complements and ToM
• Parallels between sentential complements
and the representational requirements for
the contents of mental states (propositional
attitudes)
• Can mark embedded clause true/false
• Semantic parallels between mental state and
communication verbs and ToM (Pinker &
Bloom, 1990 - evolutionary link)
Evidence for Links between FB
and Sentential Complements
• Jill de Villiers et al. (1998) - longitudinal
study of preschoolers: complements
acquired prior to passing FB tasks
• Hale & Tager-Flusberg - training study:
training on complements (communication
verbs) led to changes on FB task
• Deaf children fail FB tasks - related to
language ability especially complements
(Peter de Villiers et al.)
Sentential Complements and
ToM in Autism
• Battery of ToM tasks (FB Smarties, FB
Sally-Anne, Perception/Knowledge)
• Extraction of say and think false
complements
• General language measure (PPVT+EVT)
• Executive functions (spatial and verbal
WM, WM+IC - stroop; planning: Tower)
• Annual Visits - Year 1 and Year 2
Subjects at Time 1
• ~40 children aged 5-14, all diagnosed with
autism using ADI and ADOS (Mean 8;3)
• IQ assessed using the Differential Ability
Scales
– VIQ Mean=82 sd=19
– NVIQ Mean=88 sd 21
• PPVT Mean=84 sd=20
• EVT Mean=79 sd = 19
Time 1 Variables Correlating
with Time 2 ToM
• Age and general language partialled out
– T1 Tower
– T1 Say complements
– T1 ToM
r=.51*
r=.56*
r=.62***
– NB: NOT Think complements or any other EF
measures
What Predicts ToM at Time 2
• Hierarchical regression analysis
• Step 1 - force in age, vocabulary and T1
ToM
– R2= .65***
• Step in the predictor variables (Say
complements and Tower) according to
significance level
– Say complements:
– Tower:
R2= .11**
R2= .01 ns
Summary of Autism Findings
• Ability to interpret sentential complements
FOR COMMUNICATION VERBS
predicted later ToM ability
• EF not related to ToM (so ToM isn’t
secondary to EF, as claimed)
• ToM at T1 does NOT predict EF or
sentential complements at T2
Implications
• Only linguistic knowledge for communication
verbs predicts passing FB in autistic children
• Communication verbs do not entail conceptual
understanding of mental states
• Passing FB does not entail ability to compute
mental state information from faces/voices
(evidence from Baron-Cohen, Klin, Meyer &
TF)
• Language is an alternative bootstrap into ToM