By the end of this lecture, you will learn: How to see sound
and hear colors How to alter the perceptions of others How to know
what you dont know The most ingenious of Aristotles conclusions is
that metaphor is above all a tool of cognition - Umberto Eco, on
Aristotle Knowledge Representation Representations are Graded
Knowledge Acquisition
Slide 2
Knowledge Representation Units of knowledge allow intelligent
behavior they are an organizational unit of knowledge
Categories
Slide 3
Knowledge Representation What are categories really? Categories
are structured relationships between items. Structured on the basis
of: Shared features Shared purpose Shared context
Slide 4
Knowledge Representation Where are categories stored? Double
dissociations of category neurons Living things Nonliving
things
Slide 5
Knowledge Representation Where are categories stored? (cont.)
Category specific neurons Some neurons fire selectively to: Animals
Spiders Seals Horses Famous people Jennifer Aniston Halle Berry
Landmarks Tower of Pisa Sydney Opera House Etc
Slide 6
Knowledge Representation How are categories stored?
Definitional Approach What makes a game? Its an activity most often
practiced by children Its engaged in for fun Has certain rules
Involves multiple people Its in some ways competitive
Slide 7
Knowledge Representation How are categories stored? (cont.)
Similarity Approaches Prototype Theory But priming suggests its not
just one prototype First View pictures of Cujo Then youre faster to
recognize mean dogs as dogs than nice ones
Slide 8
Knowledge Representation How are categories stored? (cont.)
Similarity Approaches Exemplar Theory Instead of an idealized
member, there are many examples which are actual items encountered
in the past. Some research supports the idea that we store these
specific cases, other research doesnt
Slide 9
Knowledge Representation How are categories stored? (cont.)
Definitely not definitional A little of both prototypes and
exemplars Each category has multiple prototypes, which are in some
cases the actual items themselves Prototypes are updated through
experience Large categories vs. small categories
Slide 10
Knowledge Representation Structures of structures of structures
of Superordinate, Basic, & Subordinate
Representations are Graded Features of Typicality Typical
category members Have high family resemblance Can be verified as
members more quickly Are named first Are more subject to
priming
Slide 13
Representations are Graded Priming and Typicality Priming
effects depend on similarity of primed category to target category
If you think humans are comparable to animals If you think humans
are really different + + = = MEN ARE REALLY FAST TURTLES MEN ARE AS
SLOW AS TURTLES
Slide 14
Representations are Graded Typicality of Airplanes and
September 11 th Other contributions to typicality: Frequency of
exposure blah blah blah Airplane blah flight number blah blah Or
Frequency of instantiation? blah blah airplane as a weapon blah
blah blah airplane explode blah blah
Slide 15
Reasoning With Knowledge Inductive vs Deductive Inductive From
item to category Deductive From category to item
Slide 16
Knowledge Acquisition How did you learn the things you know?
There are some basic principlesbut we still dont know the whole
story There is agreement that it has something to do with
integrating and structuring information from multiple modalities,
in a flexible way Maybe WM span or maybe categories?
Slide 17
Knowledge Acquisition Thinking about what you know loud colors
of the early 90s The dark sounds of The Cure The sweet smells of
home The sharp crack of a whip
Slide 18
Knowledge Acquisition Do you actually know what you think you
know? Unskilled and Unaware of It The most incompetent people are
also the most unaware of their incompetence
Slide 19
Video of Dimensional Change Card Sort At the Cognitive
Development Center Yuko Munakata, Primary Investigator Knowledge
Acquisition What does it mean to know?
Slide 20
Explore on your own. Jobs Cognitive Engineering Knowledge
Engineering Knowledge Management Google search phrases Concept
Learning Project Halo or Cyc Expert Systems (Act-R)