-NOTE OVERHEADS

-A MEANS OF CLASSIFYING DIFFERENT ITEMS (INSTANCES OR EXEMPLARS) AS BELONGING TO A COMMON GROUPING:

THE CLASSIC APPROACH:

CONCEPT= A SET OF DISCRETE FEATURES "TIED" TOGETHER BY A RULE

EXAMPLE:

CONCEPT "BOY"=f(MALE) AND F(YOUNG) AND F(HUMAN)
        CONJUNCTION (AND) RULE

EXAMPLE
CONCEPT "STRIKE IN BASEBALL")= f(SWING BAT AND MISS) OR f(BALL THROWN IN LOCATION) OR f(HIT BALL FOUL)..
        DISJUNCTION (OR) RULE

CONCEPT LEARNING/IDENTIFICATION INVOLVES IDENTIFYING THE RELEVANT FEATURES AND THE RULE THAT BINDS THEM

USING ARTIFICIAL STIMULI

-THE TRADITION OF BRUNER, GOODNOW AND AUSTIN

PRESENT A SET OF CARDS ONE AT A TIME OR, IN A VARIANT, HAVE PARTICIPANTS CHOOSE WHICH CARD THEY WANT

GIVE FEEDBACK WHETHER OR NOT THE CARD IS OR IS NOT AN INSTANCE OF A CONCEPT (SORT OF A 20 QUESTIONS TYPE OF GAME)

EXAMPLE: CONSIDER A UNIVERSE OF STIMULI WHICH CONSIST OF THREE DIMENSIONS (NUMBER, COLOUR, AND SHAPE)

FOR EACH OF THESE DIMENSIONS WE HAVE THREE VALUES
NUMBER= 1,2, 0R 3
COLOUR,=RED, BLUE OR YELLOW
SHAPE=CIRCLE, TRIANGLE, OR SQUARE

CROSSING EACH DIMENSION, CHOOSING ONLY ONE VALUE PER DIMENSION WE HAVE A UNIVERSE OF 27 POSSIBLE COMBINATIONS
(EXAMPLE: 1 RED CIRCLE)

LET US ASSUME THAT YOU KNOW THAT THE CONCEPT TO BE DISCOVERED CONSISTS OF 1 VALUE FROM 2 OF THE DIMENSIONS

HOW DO PEOPLE GO ABOUT FINDING WHAT THAT CONCEPT IS:

EXAMPLE; YOUR FIRST CHOICE IS 1 RED CIRCLE
FEEDBACK: YES, THAT IS THE CONCEPT

WHAT DO YOU DO NEXT TO FIND THE CONCEPT?

CREATE A SET OF EXEMPLARS THE FOLLOWING WAY
GET A DOT PATTERN (LET IS SAY OF 8 DOTS IN A MATRIX; LET US CALL THIS THE PROTOTYPE)

USE A DISTORTION RULE (AN EXAMPLE MIGHT BE TO START WITH THE PROTOTYPE AND TO MOVE ANY 2 RANDOMLY CHOSEN DOTS TO A SPACE I CELL ADJACENT, WITH THE CELL BEING RANDOMLY CHOSEN FROM THE POSSIBILITIES); USING THIS RULE WOULD CREATE A SET OF DISTORTIONS FROM THE PROTOTYPE

COULD DO SO WITH A SECOND SET, CREATED FROM A DIFFERENT PROTOTYPE

NOW FROM THE TWO SETS OF EXEMPLARS SO CREATED, SELECT A SUBSET TO USE IN THE TRAINING PHASE:
TRAINING PHASE:  HAVE PEOPLE LEARN TO CLASSIFY THE DISTORTIONS AS BELONGING EITHER TO GROUP A OR B

NOTE; NEVER USE THE PROTOTYPE IN THE ThA1NING PHASE

PHASE 2: RECOGNITION MEMORY
PRESENT OLD DISTORTIONS (ITEMS USED IN THE TRAINING PHASE), NEW DISTORTIONS (OTHER ITEMS CREATED BY THE SAME DISTORTION RULE BUT NOT SEEN IN TRAINING) AND PROTOTYPE (NOT SEEN IN TRAINING); TASK; WHICH ITEMS WERE PRESENTED EARLIER? HOW CONFIDENT?
FIND: BETTER MEMORY FOR OLD OVER NEW DISTORTIONS;
MORE IMPORTANTLY: FALSE RECOGNITION (WITH HIGH CONFIDENCE) THAT PROTOTYPE WAS OLD (PRESENTED EARLIER); AFTER 1 WEEK FALSE MEMORY HIGHER THAN CORRECT RECOGNITION OF OLD DISTORTIONS!!!!

1. WITTGENSTEIN: AND THE CONCEPT OF FAMILY RESEMBLANCE
-WHAT ARE THE DEFINING FEATURES OF A CONCEPT SUCH AS "GAME"? NO DEFINING FEATURES, RATHER THE INSTANCES ARE TIED TOGETHER BY FEATURES CHARACTERISTIC OF THE CONCEPT, BUT ARE NEITHER NECESSARY NOR SUFFICIENT

2. EMPIRICAL WORK: ELEANOR ROSCH
-WHAT HAPPENS IF YOU RATE MEMBERS OF CATEGORIES FOR GOODNESS?
-EASY TO DO
-GET HIGHLY RELIABLE FINDINGS
-SOME ITEMS ARE SEEN AS GOOD EXAMPLES OF A CATEGORY OTHERS LESS SO

TYPICALITY EFFECTS: HOW COULD THIS EMERGE FROM THE CLASSIC VIEW'?

3. WHAT ABOUT THE CLASSICAL NOTION OF ALL-OR-NONE?
-SEE LABOV'S RESEARCH: PP. 324-327
        -note: no discrete boundary
        -modified by context

THIS SUGGESTS A PROTOTYPE MODEL

NOTE PROTOTYPES ARE AVERAGE REPRESENTATIONS (A SUMMARY DESCRIPTION)

1. DOES NOT PRESERVE ENOUGH ITEM INFORMATION (IF ALL WE STORE IS THE PROTOTYPE THEN CANNOT EXPLAIN THE FACT THAT PEOPLE ARE SENSITIVE TO;
-variability of instances
-correlations between features

2. CANNOT EXPLAIN CONTEXT EFFECTS;
-EG HARMONICA=MUSCICAL INSTRUMENT (CAMP FIRE CONTEXT) BUT IS NOT A MUSICAL INSTRUMENT (SYMPHONY ORCHESTRA CONTEXT)
-WOULD NEED A DIFFERENT PROTOTYPE FOR EACH POSSIBLE CONTEXT

NOTE: BOTH PROTOTYPE AND EXEMPLAR MODELS ARGUE THAT CATEGORIZATION IS BASED ON MATCHING AN INSTANCE TO SOME STORED COMPARISON ITEM (THE PROTOTYPE, OR A HIGHLY ACCESSIBLE EXEMPLAR)

EVIDENCE FOR SIMILARITY MATCHING IS BASED TO A LARGE EXTENT ON THE TYPICALITY EFFECT

BUT DOES TYPICALITY NECESSARILY IMPLY THAT SIMILARITY IS INVOLVED?

CONSIDER: DOING TYPICALITY RATINGS ON THE GOODNESS OF NUMBERS: HOW GOOD IS x OF AN EVEN NUMBER?

FIND TYPICALITY EFFECTS TN WHICH SOME NUMBERS ARE RELIABLY RATED AS MORE TYPICAL OF EVEN NUMBERS THAN ARE OTHERS

BUT NOTE: THERE IS A DISCRETE RULE THAT DETERMINES EVEN-NESS (IS IT DIVISIBLE BY "2")

SO WHAT ARE TYPICALITY RATINGS SHOWING?

DOES TYPICALITY REPRESENT AN HEURISTIC (RULE OF THUMB) THAT ALLOWS FOR A RAPID IDENT1FICATION TO A CATEGORY; BUT PERHAPS THERE IS A CORE OR ESSENCE TO A CATEGORY
THAT IS NOT BASED ON SOMETHING LIKE A PROTOTYPE OR A SET OF EXEMPLARS

CONCEPTS ARE REPRESENTED BY THE INDIVIDUAL INSTANCES THAT CONSTITUTE IT

DECIDING WHETHER AN ITEM BELONGS TO A CONCEPT, THE ITEM IS COMPARED WITH THE CONCEPT EXEMPLARS

• IF IT IS SIMILAR TO THE EXEMPLARS IT WILL BE CLASSIFIED AS A MEMBER OF THAT CATEGORY
• NOTE EXEMPLARS PRESERVE INFORMATION ABOUT
-variability
-correlation between features

-familiar examples are more likely to come to mind; used as the basis for category judgments
-therefore, there is a greater likelihood that there is a greater resemblance between exemplar used for comparison purposes and new item

-HOW DO WE EXPLAIN CONCEPT COHERENCE THAT IS CALLING ONE CONCEPT AS BEING MORE STRUCTURED THAN ANOTHER

-SOME CONCEPTS ARE BASED ARE BASED AROUND "IDEAL" PROPERTIES

-EXAMPLE: DIET FOODS==>ORGANIZED AROUND THE IDEAL OF "ZERO CALORIES"

PERHAPS THE ESSENCE OF A CONCEPT IS THE THEORY WE HOLD ABOUT THAT CONCEPT

EXAMPLE: WE INDUCE THAT A MAN IS DRUNK IF WE SEE HIM JUMP INTO A POOL FULLY CLOTHED
-IT WOULD BE THE CASE THAT OUR CATEGORIZATION IS NOT BECAUSE "JUMPS INTO POOLS CLOTHED" IS A FEATURE OF THE CONCEPT "DRUNK"
-RATHER IT IS BECAUSE PART OF OUR CONCEPT OF "DRUNK" INVOLVES A THEORY OF IMPAIRED JUDGMENT THAT SERVES TO EXPLAIN THE MAN'S BEHAVIOUR

HOW EXPLAIN TYPICALITY?
CONSIDER THE CONCEPT; "MOTHER"
MOTHER IS RELATED TO A SET OF THEORIES, INVOLVING
-WOMEN WHO RAISE CHILDREN
-WOMEN RESPONSIBLE FOR GENETIC INFORMATION OF CHILD'
-WOMEN MARRIED TO CHILD'S FATHER
etc

TYPICALITY EFFECTS ARE DUE TO THE NUMBER OF THEORIES RELATED TO INSTANCES
EXAMPLE; WHO IS THE BETTER MOTHER, x OR y
THE NUMBER OF THEORIES SATISFIED BY x (RELATIVE TO y) DETERMINES THE RELATIVE TYPICALITY SCORE (AND NOT SOME PROTOTYPICAL "MOTHER")