Presentation on theme: "Write down all the sports you can think of Provide a definition of the word ‘cup’ List items you would take out of your house in case of fire In the movie."— Presentation transcript:
Write down all the sports you can think of Provide a definition of the word ‘cup’ List items you would take out of your house in case of fire In the movie ‘Face off’, Nicholas Cage gets John Travolta’s face. Does he become John Travolta? (‘Big’) Is the next animal a bird? (yes/no)
–Name the race of the following celebrities Britney Spears Michael Jackson Ghandi Michael Jordan Yo Yo Ma Yao Ming
Course Overview Knowledge Types of Knowledge - Concepts & categories ch. 3: Vision. How are objects recognized? -It looks easy but it’s not Ch. 6-11: Memory - to know is to remember The Brain Acquisition (perception) ch.4: Attention. -Filters perceptual input ch. 5: Working Memory - Buffer for mental representations Use - Deficits & Errors Ch : Reasoning - inductive - deductive Problem Solving Ch 4:Executive Functions - Visual Knowledge - Language
Concepts & Categories What is a concept? Why are concepts useful? The structure of concepts (rules that guide how/why things are clustered into certain categories) –Similarity-based views The Classical view Probabilistic Views –Prototypes –Exemplars –Theory-based views Essence Errors and Stereotypes
What is a concept? Part of semantic memory (vs. episodic memory) A class of items that seem to belong together –‘dog’, ‘balloon’, ‘terrorist’ (things) –‘tall’, ‘ugly’ (properties) –‘walk’, ‘jump’ (actions) A mental construct (vs. the outside world) Abstract knowledge
Why are concepts useful? Functions Coding of experience: Classification of items as members of the same category –Reduces cognitive demands –Facilitates communication –Inductive Inferences Natural kinds vs. artifacts –Combines to create complex categories Contact lens Digital Camera
The structure of concepts How are things clustered into categories? Based on similarity Based on theories
What does it mean to know what a ‘dog’ is? Classical View: –To know what a ‘dog’ is, is to know its definition. –Dog = mammal, four legs, barks, wags tail –These properties are Singly necessary: every member must have them Jointly sufficient: everything that have them is a member –Categories have sharp boundaries Either you are in or you are out –Categories have a homogeneous space Everyone that is ‘in’ is equally good member of the category Classification => compare properties of instance to definition
Classical View: Problems Good definitions are hard to find! Example: A bachelor is an unmarried man –is my kid a bachelor? Adult –is the pope a bachelor? intention to get married Some members are more typical than others (Categories have internal structure)
Classical View: Problems Good definitions are hard to find! (games, Wittgenstein) Typicality effects (tea cup vs. Stanley cup) Solution : –There are no defining properties, but rather –properties characteristic of the group (typical features)
A prototype: –contains salient features that are true of most instances –is an abstract representation that could: Be the average of several instances Have the most frequent features, or Be the ideal Prototype Theory (Rosch) –specify the “center” of the category, –leave ‘fuzzy’ boundaries –graded category membership (tea cup vs. Stanley cup) Classification => comparison to prototype
Evidence Consistent with Prototype View 2. Sentence verification RT [Robin is bird] << RT [Penguin is bird ] 1. Production Robin mentioned before Penguin
Further Evidence for Prototype View 3. Picture Identification Sparrow identified faster than ostrich 4. Induction: Sparrow have X -> all birds have X but not Ostrichs have X -> all birds have X
Prototype theory ‘On the genesis of abstract ideas’ (Posner & Keele, 1968) Stimulus –Two displays of 25 dots each (Prototypes) –Variants on each of these two displays (10 dots are randomly relocated ) Training Phase: –Learn to classify variants into two categories –Items were variants of the prototypes Test phase –Old items –New Items Prototypes Variants
Prototype: Only shown during test phase Variants: departures from the Prototype Their average is the prototype Some displayed at study and test, others only at test
Training Phase Abstraction Prototype (not shown) Variants
Test Phase Prototype Old VariantsNew Variants Critical comparison: ?
Prototypes and Basic Level Physical Object Living Thing Animal Mammal Carnivore Canine Dog Australian Shepherd
Superordinate BASIC Subordinate
Superordinate: Low similarity within category (low coherence) Animals look different from each other BASIC level: -High similarity within category All fish look the same -Low similarity between categories Fishes look different from other animals Subordinate: High similarity between categories (low discriminability) Different types of fish look similar to each other Except to Experts
Properties of the Basic Level Categories –Maximize within-category similarity –Minimize between- Category similarity –Maximum level of abstraction while maintaining physical similarity –Shorter name
Basic Level “There is generally one level of abstraction at which the most basic category cuts can be made... …the basic level of abstraction in a taxonomy is the level at which categories carry the most information.” Rosch et al. 1976) One privileged level
The Exemplar View (Instance Theory) Similar to Prototype View –Representation is not a definition Different: Representation is not abstract –Descriptions of specific examples Store in memory numerous examples of each category. Categorization => comparison to stored exemplars To categorize X is to retrieve the example most similar to X
Exemplar-based explanations of typicality 2. Sentence verification RT [Robin is bird] << RT [Penguin is bird ] 1. Production Robin mentioned before Penguin More examples of typical birds in memory, so retrieval of an example similar to robin is more likely. Robins are encountered more frequently than penguins… thus, they are more frequently mentioned.
Exemplar-based explanations of typicality 3. Picture Identification Sparrow is identified faster than ostrich 4. Induction: Sparrows have X -> All birds have X but not Ostriches have X -> All birds have X Same logic here…more examples of birds will be similar to sparrow, promoting the likelihood that people will believe a fact about sparrows is true of the whole category. More instances of sparrows in memory than of ostriches, So match will occur fastest for sparrows
Exemplar-based explanations of other findings from Prototype theory 5. Prototypes are better categorized than new variants (Posner & Keele) Prototypes are ‘similar’ to variants studied in the training session. Thus, prototypes are better reminders than non-prototypical new variants
Prototype theory: Problems 1. Lack of sensitivity to within-category variability 12 inch 18 inch 8 inch 12 inch The next object is 18 inch long, is it a pizza or a ruler?
weight height Prototype Prototype theory: Problems 2. Lack information about correlated features
Prototype theory: Problems 3. Goal-directed & ad hoc categories: - “Pets, kids, money, photo albums” -‘Things to take out of the house in case of fire’ - category judgment depends on which exemplars come to mind. This, in turn, - depends on context and goals.
Prototype vs. Exemplars: The role of Expertise -Novice rely on Exemplars (okapi) -Experts rely on - prototypes (Posner & Keele), but also on - specific examples
Exemplar use by Experts Subjects: Experts (MDs) Task: Visual diagnosis of common skin conditions Study Phase –Incidental priming of certain exemplars See 30 slides, with correct diagnoses. Decide how typical it is Test Phase (2 weeks later) –Old Items (from same categories)best performance –New items from same categories (diseases) Different to items in the study phase similar to items in the study case*
Conclusions People probably often represent both prototypes & instances Experience may favor reliance on prototypes in some cases But: Instances continue to affect performance even after expertise has developed
Prototype and exemplar theories depend on the notion of similarity But Similarity depends on many factors: - Context: New York City vs. Memphis vs. Buenos Aires - Theories: - Accidental Features - Essential features And Similarity is not sufficient for categorization - Goal directed: things to take out in case of fire - similar things not always go together
Is Similarity sufficient for categorization? An object 3 inches in diameter. –more similar to a quarter –more likely to be a pizza
Theory-based Approach Folk theories highlight which perceptual features are important and which are accidental Objects are classified into concepts that best explain their attributes. Essence: Underlying non-obvious features –Dolphins and deer are both mammals, even if dissimilar ones
Psychological Essentialism People act as though things have essences (underlying natures) Change in surface features doesn’t change category membership Removing pigeon’s wings or feathers Change in essential features DO change category membership Changing a pigeon’s DNA structure Categorization ultimately based on essence –Category membership all-or-none Essentialist Heuristic: –Things that look alike tend to share deeper properties.
Three kinds of concepts –Natural Kinds (bird, fish, tree, gold) –Social Kinds (occupation, race, marriage) –Artifacts (furniture, vehicle, clothing)
Without counting, guess how many dots there are Less than 22? More than 22?
Concepts and Misconceptions
Arbitrary Categories –Over-estimate vs. under-estimate –In-group favoritism (US vs. them) Stereotypes (social categories) –Blacks, Republicans, Arab Nations –Stereotypes reduce complexity –The reduction in complexity leads to errors
Extreme examples of the category are more heavily weighted –same is true for some other categories, e.g. trees Within-category variability is underestimated –“all Bush supporters are the same” Insensitivity to disconfirmation –Members who challenge the stereotype are thought to be ‘special cases’ (poor examples of the category). Therefore –they are thought not to be diagnostic of the category. –“No women is a good soldier. A good female soldier, is less of a woman” Stereotypes are more stable than is warranted by evidence –test-retest reliability: after one week.94after 4 years.92 Illusory correlation –Distinctive behavior - Distinctive individuals are perceived to ‘go together’ even when they are independent (e.g., antisocial behavior - blacks) Misattribution (race – poverty – education- neighborhood) Racial stereotypes are thought as ‘essential’ categories Stereotypical biases
Race and essentialism Essence (natural kind) –An underlying fundamental property –common to all members of the category, and –only to members of the category Natural kind categories (e.g., ‘dog’, ‘daisy’) 1. Do not change 2. Lead to rich inductive inferences 3. Have an essence, even in the absence of a physical appearance (wolf in lamb skin) Race is a social category, but 1. Does not change (Michael Jackson) 2. Is thought to lead to rich inductive inferences race is perceived to be predictive of attributes and behavior 3. It’s thought (erroneously) to have an essence. Error: The absence of physical appearance does not negate the race (in nazi germany, Jews were Jews even if they looked Aryan)
Essentialism in other social categories ‘ Marriage is between a man and a woman’
Concepts & Categories What is a concept? Why are concepts useful? The structure of concepts –Similarity-based views The Classical view Probabilistic Views –Prototypes –Exemplars –Theory-based views Essence Errors and Stereotypes