8) Concepts & Categories

Concepts = Categories

Natural Categories

Characteristics

Basic level categories

Family resemblance

Internal structure (typicality + family resemblance)

Fuzzy sets (ill-defined boundary)

Properties

Are they always preferred?

Semantic memory lecture (Patterson)

Concepts = internal representations

Categories = "out here in the world"

Useful

Hubs: efficent way to integrate our knowledge

Easier to detect semantic similarities across concepts that are different in modality specific attributes (prawns & scallops look different but are similar conceptually)

Category: A class of stimuli that are treated in an equivalent manner

Cut down diversity of objects & events that must be dealt with uniquely (eg. allergy to scallop generalised to allergy to seafood)

Generalise helps survival of an organism with limited capacity

Natural categories

Occur in natural language (eg. fruit, animal, tool, furniture, clothing)

Rosch's work= suggests a correlational structure

Unlike artificial categories: features are not combined arbitrarily

Artificial category learning study (Bruner 1956)

Shape (3), Colour (3), No of shapes (3), No of borders (3)

S Task: work out concept experimenter has in mind
(Eg. Black (colour) cards with circles (shape)

Concept universe: 4 feature dimensions (3 x 3 x 3 x 3 = 81 possible cards)

Features may be combined arbitrarily

Feature dimensions are not independent, they are correlated

If category member "Has feathers' it is also likely to "has a beak" but not "has petals"

Artificial categories: feature dimensions are independent & can be combined arbitrarily "black" with "circle" or "square" or "cross" equally

Typicality

(McCloskey & Glucksberg): Asked Ss to decide category membership for each exemplar-category pair twice a month apart

Some exemplars (eg. bookend-furniture) low between-subject agreement and within subject consistency

Artificial categories: boundaries are clearly defined

( Wittgenstein, 1953) "Game" no attributes that are shared by ALL members of the category

No single attribute shared by ALL members of a category
But each member has at LEAST 1 attribute in common with other members

Boards? (No, ball games)

Winners and losers (No, solitaire?)

Amusement? (No, serious games?)

(Rosch & Mervis, 1975) Eg. Different types of fruit rated for typicality on a (1-7) scale: typicality is not uniform

Orange (1.07), coconut (4.50), tomato (5.57)

Typicality + Family resemblance = high correlation

(Rosch & Mervis, 1978) Ss listed attributes possessed by each exemplar of a category (eg. bird)

Derived the family resemblence score: sum of the weighted scores of attributes

Greater weight given to attributes shared with other exemplars

Robin = 13, Swallow = 13, Penguin = 3

High correlation between rated typicality and family resemblance score

Eg. Robin (high) VS penguin (low scores): typical exemplars share more attributes in common with other members of the category

Prototype: member of a category having the highest family resemblance scores (eg. apple prototype of fruits)

Internal structure

Family resemblance: better exemplars share more attributes in common with other exemplars

A category is centred around a prototype

Typicality gradient: some members of a category are better exemplar (more typical) than others (eg. apple more typical than coconut)

Subordinate (furniture) --> Basic-level (chair, desk) ---> Subordinate (arm chair, office chair)

Basic level category: best balance between informativeness & distinctiveness

Many attributes are common to members within the category (subordinates)

Few attributes in common with members of other categories

Most general level: people use similar motor movements when interacting with category members (eg. chairs but not tables sitting)

Informative

Communication = effective (Chair broke, not a piece of furniture broke)

Learned first by children: car not vehicle

Expressed economically: short name (chair > arm chair)- simple sign language gestures

Experts prefer subordinate names

Faces: matched faster to subordinate level than basic level (Benedict Cumberbatch > a human face)

Dog experts: Beagle, Terrier, Labrador

Is expertise necessary for "subordinate shift" (Anaki & Bentin, 2009)

Down-shift from basic to subordinate level: familiar objects as well as faces/objects of expertise

Eg. towers, bridges that are familiar

Superordinate: Living/non-living, Basic: monkey/human (face), tower/bridge, Subordinate (exemplar) familiar faces & towers (eg. Benedict Cumberbatch, Tower of Pisa)

click to edit