8. Knowledge 12/3/2012 2:14:00 PM
1) Why are categories important? Why don’t definitions work?
Categories > definitions
2) Comparison approach: How can we categorize by comparing to a
Categories determined by similarity: Prototypes/Exemplars
3) The Network Approach. What is a network, and how is info stored?
“Privileged” levels of categories?
4) Representing relationships between categories: Semantic Networks
5) Representing concepts between networks: The Connectionist Approach
6) Categories represented in the brain
1. Categories > definitions
A concept = a mental representation that is used for a variety of
cognitive functions (memory, reasoning, language)
Categorization = the process by which things are places into
Categories help us to understand objects in our environment = for
us to take mental shortcuts and use knowledge from our
Semantic memory = facts + knowledge
Categories are “pointers to knowledge”. Once you know something
is in a category, you know a lot of general things about it and can
focus your energy on specifying what is special about it.
The definitional approach to categorization doesn’t work because
most categories contain members that do not conform to the
o Many natural objects & man-made objects are hard to identify
with a strict definition due to their diversity.
The philosopher Wittgenstein proposed the idea of
family resemblances to deal with the fact that
definitions do not include all members of a category. His
solution expressed how similar characteristics can be
expressed and grouped in “family resemblances”. = less exclusive than definitions, which did not include
all members of a category.
FR = refer to the fact that things in a particular
category resemble one another in a number of ways,
which allows for some variation within a category
2. Categories determined by similarity: Prototypes/Exemplars
The idea behind the prototypical approach to categorization is that
we decide whether an object belongs to a category by deciding
whether it is similar to a standard representative of the category –
o A prototype is formed by averaging category members a
person has encountered in the past.
o Whereas the exemplar approach state that the standard is
created by considering a number of typical members of a
Prototypes: Is it SIMILAR to a standard of the category? – Similar but not
an actual member of the category = it is an average representation of the
Prototypicality is a term used to describe how well an object
resembles the prototype of a particular category.
Category of BIRD = HP = Sparrow. LP = Penguin
Category of FURNITIRE. HP = Chair, LP = Mirror
The following IS TRUE of high-prototypical objects: Is it a TYPICAL
member of the category?
i. They have high family resemblances
ii. Statements about them are verified rapidly.
(how rapidly people can answer Qs about an
objects category – “yes”/”no”, for Qs like “An
apple/ pomegranate is a fruit” = faster
response for HP objects like apple = typicality
iii. They are named first (BIRDS - sparrows before
penguins) iv. They are affected by priming (“Green”
prototype = matches the “good” green, but is
a poor match for the light green = priming
results in faster “same” judgments for
prototypical colours than for nonprototypical
Exemplars: also involved determining whether an object is similar to a
standard object. But the prototype is an “average” member of the category,
and an exemplar standard involves many real examples, each one called an
The exemplar approach to categorization involves determining
whether an object is similar to an exemplar.
o An exemplar is an actual member of a category that a person
has encountered in the past.
An advantage to the exemplar approach = it takes into account
individual cases (like birds that don’t fly), thus it doesn’t discard
info about atypical cases within a category, such as penguin in the
o The exemplar approach can also deal more easily with
categories that contain widely varying members, like games.
Researchers have concluded that people use both approaches to
Prototypes may be more important as people initially learn about
categories, (when it’s harder to learn about exceptions like
and then later exemplar info may become more important.
Exemplars may work best for small categories (e.g. US presidents),
and prototypes best for larger categories (e.g. birds).
Protoypes = learnt first, for BIG categories (Birds, Cars)
Exemplars = learnt after prototypes (more info), for SMALL
3. “Privileged” levels of categories? The kind of organization in which larger, more general categories
are divided into smaller, more specific categories is called
o Furniture Chairs (Kitchen chairs, Sofas, Stools etc)
Experiments by Rosch indicate that a “basic level” of categories
(such as guitar, as opposed to musical instrument/rock guitar) is a
„privileged‟ level of categorization that reflects peoples
Basic level is most important, as it is the level above which info is
lost, and below which little info is gained.
o Superordinate (Furniture)
Subordinate (Kitchen chair)
Experiments in which experts were tested show that the basic level
of categorization can depend on a persons degree of expertise.
4. Representing relationships between categories: Semantic
How hierarchies of information are organized in the mind –
explained with the semantic network approach
The semantic network approach proposes that concepts are
arranged in networks that represent the way concepts are
organized in the mind.
Collins and Quillian’s model on semantic networks: is a network that consists
of nodes that are connected by links.
Concepts and properties of concepts are located at the nodes.
o Skeleton nodes are connected by links
o Links between nodes are based on automatic associations =
e.g. canary & bird, bird & animal
Properties that hold for most members of a concept are stored at
higher-level nodes. E.g. Bird – “Can fly”, rather than having the
“can fly” by each exemplar birds node like by the word canary.
o Storing shared properties just once at a higher-level node is
called cognitive economy. We start by entering the network at the concept node
(e.g. for canary “can sing”, “is yellow”)
Concepts at nodes are most specific at the bottom, and
more general on top
Going up: Canary Bird Animal
To deal with exceptions exemplars of birds which can’t
fly like ostriches/penguins had “can’t fly” next to their
specific bird type.
SUPPORTED - Collins and Quillian’s model
o by the results of experiments using the sentence
verification technique (SVT) based on travelling time of
links between nodes
E.g. SVT - It will take longer to answer “yes” to a
canary is an animal, than a canary is a bird. As it takes
more time to travel along two links to get from canary
to animals, but only one to get from canarybird
o The spreading activation (SA) feature of the model is
supported by priming experiments.
SA – is activity that spreads out along any link that is
connected to an activated node (dashed lines alo