PSY270 Lecture 8
Knowledge & Categorization
•Concept: mental representation that is used for a variety of cognitive functions (memory, reasoning, language)
•Categorization: function of concepts, process by which things are placed into categories.
oEssential tool for understanding of the world, especially things we’ve never seen before.
oProvides information about item in category
How are objects placed into categories
Why definitions don’t work for categories
Definitio n al ap proa c h to c a teg o riz a tio n: we pla c e obje c ts into c a teg o ri e s b a s e d o n whether it
me e ts the d efinition of the c a teg o ry. (eg. sq u are – plane wit h 4 equal side s)
However, many n atural obje c ts an d h u man made obje c ts, d efinitio n s do not work well.
oChair: pie c e of furniture consistin g of s e a ts, legs, b a ck… most chairs h ave such, b u t there
Witt genstein : “family re s e mblanc e ” ide a : things in a p ar ticular c a teg o ry re s e mble o n e an o ther in
a n u mber of ways, which allows for v ari atio n.
o c a teg o riz ation is b a s e d o n d etermining h ow si mil ar an o bje c t is to some stan d ard
The prototype approach: finding the average case
prototy p e ap proa c h to c a teg oriz a tio n: membership in a c a teg o ry is d etermined by comparin g
o bjec t to a prototy p e (ty pic al member) that repre s e nts the c a teg ory.
Rosc h: propos e d that the ty pic al prototy p e is b a s e d o n an average of members of a c a teg ory
that are com m o nly ex p eri enc e d.
o“Bird” – eg. spar r ows, ro bins, blue jays
oPrototy pic ality: v ari atio n s wit hin c a teg o ri e s a s repre s e ntin g differenc e s (owls, p enguins)
oHigh prototy pic ality: c a teg o ry member clos ely re s e mble s c a teg o ry prototy p e
oLow: d o e s n o t clos ely re s e mble ty pic al member of c a teg ory.
oExp eri ment: pre s e nted p ar ticipants wit h c a teg o ry titl e “bird”/”furnit ure ” an d a list of
ab o u t 5 0 members of the c a teg ory.
Par ti cipants rate extent member repre s e nted c a teg o ry o n 7 p oint s c ale (1: hig h, 7:
Most agre e d o n “ s parr ow” an d “chair/sofa” a s the ty pic al member.
Most agre e d o n “bat” an d “telep h o n e ” a s least ty pic al member.
Prototypical objects have high family resemblance
oRosch & Mervis: asked participants to describe characteristics of “chair, sofa, mirror, telephone”
found that when item’s characteristics has a large overlap with other items of a category, meant family
resemblance is high. (little = low)
strong relationship b/w family resemblance & prototypicality.
Statements about prototypical objects are verified rapidly
oSmith et al : used procedure called “sentence verification technique” to determine how fast people could answer
questions about an object’s category.
given statements, say “yes” if agree, “no” if not.
Responded faster for objects high in prototypicality
• typicality effect – ability to judge highly prototypical objects rapidly
Prototypical objects are named first
oMervis et al : named objects high in prototypicality first when asked to list a category
Prototypical objects are affected more by priming
oPriming: presentation of one stimulus facilitates response to another usually followed closely in time
oRosch : demonstrated that prototypical members of a category are affected by a priming stimulus more than
Participants first given the prime (eg. color green), then 2 seconds later saw two colours side by side
and indicated by pressing a key as quickly as possible if colours were same or different.
PSY270 Lecture 8
•Side by side colours paired in 3 ways
oSame – good examples
oSame – poor examples
•Found that priming resulted in faster “same” judgments than different.
When hear “green”, they imagine a “good” (hyighly prototypical) green.
The Exemplar approach: thinking about examples
exemplar ap proa c h to c a teg oriz a tio n: involve s d eterminin g whether an o bje c t is si mil ar to a
oinvolve s many example s (u nlike prototy p e – o n e)
oExemplers: a c tual members of the c a teg o ry that a p erso n h a s encountered in the p a s t
Eg. if p erso n h a s encountered spar r ow/ro bins in the p a s t, the s e wo uld b e an
exemplar for the c a teg o ry “birds ”
oExplains the ty pic ality effe c t – o bje c ts that are like more of the exemplars are cla s sifi ed
Which works better – prototypes or exemplars?
Exemplars: by u sing re al example s, it c a n e a sily take into a c c ount aty pic al c a s e s (eg. fligh tle s s
birds) inste a d of comparin g a p enguin to an “ a verage ” bird.
oDoe s n’t dis c ard information that mi ght b e u s eful later
oRepre s e nts aty pic al birds (pen g uins, o striche s…)
oReq uir e s u s to remember only some of the s e v aryin g example s .
Stu die s: co n clud ed that p eo ple may u s e b o th ap proa c he s .
oWhen we initi ally le arn a c a teg o ry, may average exemplars into prototy p e s an d later
some of the exemplar information b e c ome s st ro n g er.
During e a rly le arnin g, p oor at takin g a c c ount “ exc e ptio n s” but later exemplars for
the s e c a s e s wo uld b e ad d ed to the c a teg ory.
oExemplar may work b e s t for small c a teg o rie s (eg. US pre sidents) an d prototy p e b e s t for
we kn ow g enerally what c a ts are (prototy p e), b u t we spe cific ally o u r own c a t the b e s t
oblen ding of prototy p e s & exemplars
Is there a psychologically “privileged” level of categories?
hierarchic al organiz a tio n: g eneral c a teg o ri e s are divided into small er, more spe cific c a teg o ri e s ,
cre a ti n g n u mber of levels of c a teg o ri e s
Rosch’s approach: what’s special about basic level categories?
glob al (su p erordinate) b a sic spe cific (subordinate)
furnit ure table kitchen table
reaso n ed that gre a ter n u mber of fe a ture s provide s more information ab o u t a c a teg ory
ofrom b a sic 9 com m o n fe a ture s
glob al – 3, los e a lot of informatio n
spe cific – 1 0 . 3 , g ain just a littl e information
How knowledge can affect categorization
Tanaka & Taylor: “ ex p er t ” ex p eri ment.
oExp er t s u s e d more spe cific c a teg o ri e s to n ame birds
oNon ex p er t s: u s e d b a sic c a teg o ri e s .
Peo ple wit h more ex p eri enc e an d familiarity wit h a p ar ticular c a teg ory ten d to focus on more
spe cific information wit h the spe cific level.
oAbility to c a teg oriz e is le arned from ex p erienc e.
Representing relationships between categories: semantic networks:
Bird e g . s p ar r ow s , ro bi n s , bl u e j ay s. Statements about prototypical objects are verified rapidly o. Smith et al questions about an object"s category. : used procedure called sentence verification technique to determine how fast people could answer given statements, say yes if agree, no if not. Responded faster for objects high in prototypicality: typicality effect ability to judge highly prototypical objects rapidly. Prototypical objects are named first: mervis et al. : named objects high in prototypicality first when asked to list a category. Prototypical objects are affected more by priming o: rosch. Priming: presentation of one stimulus facilitates response to another usually followed closely in time nonprototypical members. : demonstrated that prototypical members of a category are affected by a priming stimulus more than.