4/3
● What is cognitive science?
○ Interdisciplinary science of the mind
● Forces that led to cognitive science
○ Individual disciplines
○ Rejection of behaviorism
○ Advent of computer science
■ Both what mind is and how to test it
○ Chomsky’s argument for internal representations
● For any given topic
○ How it is represented and processed – this class
○ How it is learned by children
○ How it can be computationally modeled
○ How it is realized in the brain
● Topics covered in this class
○ Memory: short-term and long-term
○ Knowledge
○ Cognitive development
○ Learning
○ Reasoning → second half of class
■ Why are we sometimes smart and sometimes dumb?
○ Broader questions
■ learning/education → end of first half of class
■ Expertise
■ Animal cognition
● Do animals have symbol systems
■ Misinformation
■ Language and culture
4/8
● Incoming information → sensory memory attention → working memory/STM → encoding ←
retrieval LTM
● Working memory
○ Small capacity 7+- 2
○ Duration – about ½ minute without active processing
○ Active processing keeps information alive in WM and passes it into LTM
■ Rehearsal
■ Mental arithmetic
■ All kinds of active thinking
○ Stm = passive processing portion of working memory
● Sensory memory
○ Very brief
○ Requires ATTENTION to enter working memory
○ Ex: replaying a missed sentence
○ ½ second for visual, 5-10 seconds for auditory
● Short term/working memory
○ 30 second duration if you don’t rehearse
■ Maintenance rehearsal
■ ETC.
○ Chunking
○ Meaningfulness
● Baddeley’s working memory model
○ Central executive
■ Planning
■ Choosing what processes to carry out
■ Etc.
○ Phonological loop
■ Rehearsal or verbal info
■ Similar to overt speech
■ Phonological processing
■ 2 second principle
○ Visuospatial sketchpad
● Forms of recall → better
○ Free recall: reproduce
○ Cued recall: given cue to material
● Forms of recognition
○ yes/no recognition: was this in the studied material or not?
○ Force yes/no: which of these was in the studied material
● Bringing WM into LTM: encoding storage retrieval
○ Organizing incoming material
○ Integrating incoming material with longterm memory
○ Structure (such as a picture) aids comprehension by giving prior knowledge/advance structure
better to have BEFORE than AFTER
● Context effects in memory
○ Scuba diving study – match conditions (study on land, recall on land; study in water, recall in
water) or mismatch
● Depth in processing
○ Deeper, more elaborate processing means better memory retrieval
■ Conceptual as opposed to factual/surface-level
4/15
● classical/definitional view of categories
○ Category members must fit some necessary (must have these properties) and sufficient
properties (must be in this category if it has these properties)
○ Very clear-cut
○ Good Examples: even numbers (necessary and sufficient)
○ Bad example: bachelor (unmarried, adult, male)
● Eleanor rosch, prototype view
○ Prototype = most typical member - one that is most similar to all the others
■ Real exemplar or an abstraction that combines all the properties/averages them out
○ Membership is determined by similarity to prototype, leads to fuzzy boundaries
○ Bell curve as opposed to a solid-boundaried box
○ High-typicality
■ Categorized quickly
■ Named first
■ Learned first by children
■ Cognitive reference points (an oval is sort of like a circle, but a circle is not sort like an
oval)
○ Even classical concepts like even numbers can have prototypes
■ 4 less prototypical than 6 because roundness is associated with evenness
● Exemplar view
○ Challenge to prototype: prototyping varies with context
■ Chicken for “bird” v. “barnyard animals”
○ Not central rule or example
○ Similarity to exemplar
○ Problems
■ Storage: average horse - means you have to store every horse you’ve ever seen
■ Not clear what counts as an “instance’
● Solution:
○ High density of very similar items - average into a prototpye
○ Items that are less similar - stored as exemplars or distinct subcategories
● Theory view
○ Categorization is about knowledge and belief structures, not just properties
■ Birds can fly + birds have wings = wings enable birds to fly
■ Interconnectedness → causal knowledge and intuitive theories, relations between
features, explains concepts whose members don’t all share simple perceptual similarity
● Hierarchical organization
○ Superordinate: animal
○ Basic: bird
■ Intermediate level of abstraction
■ Used mostly in free-naming tasks
○ Subordinante: robin
● JW Tanaka and Taylor experiment
○ Nonexperts typically identify with basic level
○ Experts are much more likely to say the subordinate name
4/17
● How does human cognition arise developmentally?
○ Strong nativism: in-born
■ De cartes
■ Chomsky
○ Strong empiricism: experiential
■ Skinner
■ Locke
● 1940s/50s: behaviorism – b f skinner
● 1959: Chomsky argues that behaviorism isn’t enough to learn language
● Middle ground – learning is still essential, most of our intelligence is learned CONSTRUCTIVISM
○ Piaget: child as a little scientist
■ Stage theory of development
■ 0-2 sensory-motor (action schemas in first four months; a-not-b 8-12 months; , 2-6
preoperational thought (18 months will follow parents’ gaze; 3-4 years scale
errors/egocentricity, 7-11 concrete operations (5 years do not understand liquid
conversation), 12- formal operations (analogies - SOME COUNTERARGUMENTS
like tree-knee)
■ Great accomplishments
● Showed that thinking and learning happen even in young infants
● Traced the course of insights across development
● Invented many key techniques for investigating children’s ideas
● Pioneered ‘child as scientist’
○ Vygotsky: child as cultural apprentice, sociocultural
■ Scaffolding → independence
■ Social speech (“I want milk” – basic wants, outward-directed) → egocentric speech →
inner speech
■ Language first then cognition (language is a tool for cognition)
○ Verses piaget
■ Egocentric → social speech
■ Cognition first then language
● What causes developmental change?
○ 1) global stage changes in cognitive ability
■ Piaget
○ 2) maturational changes
○ 3) learning view
■ Shift from being a novice to an expert
■ Gains in knowledge drive cognitive development
● Evidence: contextually specific gains, influence of culture and language on
how children learn, gains in cognitive skills due to schooling, complementary
gains in processing ability, adults who lack knowledge are like children
■ Improvements in organizing knowledge
4/22
● Effects of schooling on children’s knowledge
○ Being in school october to april - big gains in knowledge
■ Gains in knowledge → gains in processing ability (“adults who lack knowledge are like
children”?)
● M. Chi: knowledge drives cognitive gains
○ Kids were much better at remembering chess plays (vice versa for numbers because adults are
experts in numbers because of more knowledge about numbers)
■ Crystalized knowledge = knowledge gained through culture
● Analogical thinking and learning: similarity is based in relational commonalities – objects don’t have to
be similar, but alignment is easier if they are – same respective relation in the target as in the base
○ Steps
■ Objects put into correspondence by relational alignment
■ Analogical processing can lead to new inferences
○ Central in higher-order cognition
■ Math / science
■ Social cognition
■ Everyday reasoning
■ Examples in education
● The heart is like a pump
● Literary metaphors
● Simple verbal analogy—Horse: Grass :: Robin : Worm
● Memes
○ Raven’s matrix task
■ Predictive of other intelligence tests
○ The relational-match-to-sample task (premack)
■ Experimented on chimpanzees/baboons
● Structure-mapping
○ Comparison can be used to foster relational learning
● Achilles heel of anatomy
○ Retrieving analogical examples from long-term memory
■ Dunker’s radiation problem
● Fortress story abets it
○ Overall similarity matches (relational commonality) are easier to retrieve
■ Proverbs situation
○ The inert knowledge problem
■ Students often fail to apply prior knowledge even if they're retained it
■ How can we improve relational transfer?
● Comparing analogous problems during encoding
○ Ex: learning negotiation strategies
4/24
● Children as learners
○ Powerful learning processes
■ Concept from example
■ Analogical comparison and generalization
● Relational shift
○ Raccoon v. skunk by 6 years old; why is a cloud like a sponge by 9
years old
○ Assumption that colored keys open colored doors
○ Novel spatial relations with animals – blicket thing (v. object
characteristics)
● Progressive alignment: going from near/easy look-alike comparisons to harder
ones
○ 4 year olds don’t understand relational choice if you go from
increasing size to increasing shading, but they can if its in the same
metric of increasing size
○ Intense social focus
■ Social learning
● Cooperation - toddler altruism
■ Joint attention
■ Imitation
● Initiates people like them and imitates exact actions, unlike apes
■ Language learning
■ Tomasello
● Social ability makes us very smart, not just cognitive/analogical
○ Language
■ Cognitive toolkit
● Spatial language production
● How do we know what babies learn?
○ Looking-time methods: if infants are surprised or interested in something, they will look more
○ Violation of expectation
■ Support events - no contact object will fall – 2.5 months, 4 months - 5 months (now
know that the sideways one is going to fall)
■ Occlusion then containment then covering - don’t understand containment until 7.5
months
■ Habituation with cats v. dogs, you’re actually teaching them in the lab
■ Dot patterns
● Social learning: theory of mind
○ What is ToM?
■ Your own mental states and those of others
■ The ability to infer and keep track of others’ goals, intentions, knowledge, beliefs,
desires (goals/desires easier to identify)
■ Recognizing false beliefs
■ Tasks
● Diverse desires
● Diverse beliefs
● Appearance-reality
● Location-change false belief
4/29
● Three components of wm
○ Phonological loop
○ Executive
○ Visuospatial sketch pad
● Ltm
○ Importance of connectivity and organization - depth of processing
○ Recall v. recognition - encoding specificity (context-specific encoding)
○ episodic v semantic memory - specific v general knowledge
○ Explicit v implicit memory -
■ Implicit: nonaware retrieval, no effect of depth of processing more important to just
do it a bunch of times, automaticity (=procedural memory)/priming
● Concepts and categories
○ Definitional - necessary and sufficient features
○ Prototype [Rosch] - graded membership
■ “It’s a perfect example of a train”
■ Average
○ Exemplar - concept as a set of exemplars
■ “It’s like every other train”
■ You compare to every other train
○ Theory-based - relational structure, not just features
● Hierarchical taxonomic structure of categories
● Representation of knowledge
○ Schema - general relational structures (wedding = bridge, groom, officiant)
○ Scripts - schema with timing information - sort of play by play
○ Analogy
■ Mapping as a way of focusing on common relational structure (easier than retrieval -
Duncker Radiation Problem, inert knowledge problem - can improve this by making
an abstraction)
■ Allows for relational abstractions and new inferences
■ Structure mapping process - evidence for structural alignment process
● Studies using cross-mapping
○ Mental models
● Cognitive development
○ Nativism v empiricism
○ Constructivism
■ Piaget and vyogtsky
○ Modern views
■ Tomasello - children’s imitation/prosocial behavior and altruism
■ Sociocultural views
■ Novice-expert shift
○ Learning is a major component of cognitive development
■ Chess - knowledge is part of what makes you a component cognition
○ Infant cognition
○ Conceptual development
■ Early focus on object properties
● Characteristic-to-defining shift
● Difficulty with relational mapping
■ Importance of language
● Ex: spaital language
■ Theory of mind
○ Cognitive development
■ Joint attention
■ Imitation
■ Prosocial behavior
● Mental models
○ Kind of schema
○ Common sense beliefs about physical world
■ Naive physics
■ Spontaneous analogy in forming wrong mental models
● Problem-solving
○ Well-structured
■ Means-ends
■ Problem space
○ Ill-structure
■ Insight - feeing of getting warmer
■ Functional fixedness
○ Novice-expert
Problem-solving
● Well-structured - you know what to do (finding area of parallelogram)
● Ill-structured - no clear solution paths (should I go to grad school? How can a doctor cure an
inoperable tumor? - latter insight)
● Four heuristics
○ Work forward
○ Work backward
○ Generate and test (trial and error)
○ Means-ends analysis
● Newell and simon
○ Much of human cognition can be seen as problem solving
○ Well-defined complex problems
■ Goals
■ Operators
■ All possible states
■ All possible legal paths through the states
○ Problem space
■ Triangle - one side is the quickest path to success
○ Tower of hanoi = acrobat problem
■ Content effects
○ ill-structured/insight problems
■ Window washing
● Set effect/content effect
■ Archimedes in the bath
● Use water to find volume displacement
■ Einstein’s theory of relativity
● A series of little insights
■ The monk and the hill problem
■ Janet feelings of warmth
● Mental models
○ Representation of a domain that supports understanding reasoning and prediction
○ Typically used for situations where there is a right answer - mental model of evaporation,
potato v potato chip
■ Allows to look at novice-expert differences
○ Analogical mental models can be wrong
■ Valve v threshold model
■ Illusion of explanatory knowledge
● Understanding that your knowledge is bad once you are asked to explain, but
an illusion that you do understand it before
■ Curved tube ball rolling - curvilinear v. tangent
■ Impetus theory - ex in cartoons
Questions
Diff between stm and wm
Diff between exemplar and prototype view
What does it mean to make something an abstraction in analogy?
How should we structure free response?
Is the basic level privileged?
Evidence: kids learn basic level terms first; when asked to name images, they name the basic level -
labrador as dog rather than mammal
5/15
● Learners benefit from comparison
○ Problem solving
○ Causal
○ Etc.
● Analogical comparison can lead to rapid change in understanding, particularly structure mapping
○ Including alignable differences
○ Helpful to compare more abstract concepts or concepts you are unable to see with something
you can see (ex: core of earth to lava lamp)
○ Novice learners
■ Surface features
○ Easier to identify differences between cat and tiger and cat and traffic light because the former
has more relational alignment (high-similarity pair)
○ The rich get richer effect
■ The more you know about the relational structure in the domain, the easier it is to
operate in the domain and learn more
● Skyscraper study
○ Figured out after wiggling the structures that the diagonal is stronger foundation because it
doesn’t wobble
■ Do they notice that it’s because of the diagonal brace? - when there’s higher
alignability, yes
■ Comparison > no training high align > low align boys > girls
● Structural alignment is essential - spatial juxtaposition and high similarity
help s
● What helps analogical processing?
○ Source visually presented
○ Source visible
○ Spatial alignment
○ Comparative gesture
○ Source familiarity
○ Imagery
● Language can support relational learning within a language
○ Common labels invite comparison and abstraction
○ Labels help retain learned relational structure and promote relational retrieval
○ Linguistic structure can invite conceptual structure over development
○ Study: does the amount of spatial language use/exposure predict nonverbal spatial tasks years
later?
■ All administered to children at 54 month visit
● Block design
● Spatial analogies
● Mental rotation
● → has to go through the children, no the parents
● Overcoming inert knowledge problem
○ Study
■ Tempting competitor (relational concept of reciprocity– social interactions, a favor is
likely to be returned; countries enter trade agreements that allow free trade v.
domain-similar but non-relational concept–governing body should only take as much
action as is necessary to achieve an objective)
■ Relational labels increase the likelihood of relational retrieval
○ Study
■ Does the amount of math talk from teacher matter in children’s learning of math?
■ Math gains greater with more math talk
● Class matters
● Kinds of cognitive engagement
○ Metacognition modeling: in the moment; metacognitive regulation: planning element
○ ICAP framework
■ High engagement to low engagement
● Interactive (talking to other people)
● Constructive (go beyond the information presented)
● Active (highlighting)
● Passive (reading and trying to memorize)
■ Study in psychology class
● Constructive helped
○ Self testing self explaining analogical comparison self monitoring
number of constructive strategies
● Active didn’t help improve test scores