Page 240-262, 22 pages Page 1 of7
Chapter 9: Knowledge
HOW ARE OBJECTS PLACED INTO CATEGORIES?
• Concept: A mental representation that is used for a variety of cognitive functions, including memory,
reasoning, and using and understanding language. E.g. Concept of cats including what they are, what
they look like, etc.
• Categorization: The process of placing things into groups of categories. This is essential for helping
us understand individual cases we have never seen before, with categories acting as “pointers to
knowledge” that once something is categorized a lot of general knowledge can be applied to it while
energy can be focused on specifying what’s special about this particular object.
• Without categories, we would have to investigate each object or circumstance individually; this would
be very complicated and inefficient.
Why Definitions Don’t Work for Categories
• Definitional Approach to Categorization: Decide whether something is a member of a category by
determining whether it meets the definition of a category.
• This works well for some things, such as defining a square as “a plane figure with four equal sides”.
However, for most natural objects, this does not work as not all members have the same features.
• E.g. A chair as “a piece of furniture with a seat, 4 legs, and a
back” but there are many chairs which are not designed like this.
• Family Resemblance (Wittgenstein): The idea that things in a
particular category resemble one another in a number of ways,
allowing for some variation rather than setting definite criteria. E.g.
A place to sit, a way to support a person’s back for a chair but can
be many different sizes and shapes.
The Prototype Approach: Finding the Average Case
• Prototype Approach to Categorization: Membership in a
category is determined by comparing to a “typical”, representative
prototype member of the category. This is a statistical approach, based on the average/mean of many
• Rosch (1973): The prototype is based on an average of members of a category that are commonly
experienced, such as an average of common sparrows, robins, and blue jays for “birds” although the
prototype is an average and not an actual member.
• Prototypicality describes variations within a category, with high prototypicality meaning a member
highly resembles the prototype, while low prototypicality meaning it does not closely resemble the
• Subjects presented with a category title like “bird” and a list of 50 members, with task to rate the extent
to which each member represented the category title (1 meaning good example). Results included 1.18
for sparrow, 4.53 for penguin, and 6.15 for bat;
for “furniture”, 1.04 for chair and sofa, 2.59 for
china closet, 6.68 for telephone
Prototypical Objects Have High Family
• Rosch & Mervis (1975): For a group of common objects, list as many characteristics that you feel are
common to them, e.g. “bicycles” have two wheels, pedals, handlebars, are man-powered
• When an item’s characteristics have a large amount of overlap with the characteristics of many
other items in the category, they have high family resemblance and thus high prototypicality Page 240-262, 22 pages Page 2 of7
Statements about Prototypical Objects are Verified Rapidly
• Smith (1974): Sentence verification technique where subjects are presented with statements and
asked to answer “yes” if true, “no” if not, e.g. “An apple is a fruit”.
• Typicality Effect: Subjects respond faster for objects that are
high in prototypicality (apple for fruit), than objects low in
Prototypical Objects are Named First
• When subjects are asked to list as many objects in a category as
possible, they tend to list the most prototypical members of the
Prototypical Objects are Affected More by Priming
• Priming occurs when presentation of one stimulus facilitates the
response to another. Prototypical members of a category are
affected by a priming stimulus more than non-prototypical members.
• Rosch (1975): Subjects hear a prime, such as “green”. They then see a pair of coloured circles side by
side, and must indicate whether the two circles are the same colour or different. The two circles
presented are either 1) the same and good examples of the category (vivid green), 2) the same and poor
examples of the category (light green), or 3) different, from different categories (e.g. orange and blue)
• Priming with the word “green” resulted in faster “same” judgements for the prototypical good colours,
than the nonprototypical poor, light colours.
• This is because from priming works to facilitate a subject’s response to a stimulus by providing some of
the information they would later need to respond to the stimulus. So subjects imagine a prototypical
“good” green when they are primed with the word “green”; thus, prototypicality affects behaviour.
The Exemplar Approach: Thinking about Examples
• Exemplar Approach to Categorization: Involves determining whether an object is similar to a
standard object, but instead of a single “average” stereotype, it’s to exemplars which are actual
members of a category that a person has encountered in the past. E.g. Sparrows, robins, and blue jays
are each exemplars for “birds” – not a statistical approach with taking an mean
• The exemplar approach can explain the typicality effect with faster reaction times, since objects that are
more like exemplars are classified faster (family resemblance).
• One advantage is that since real examples are used, it can more easily take into account atypical
cases. E.g. Instead of comparing a penguin to an “average” bird, we remember there are examples of
flightless birds. Thus non-typical examples like ostriches and penguins can be used as exemplars rather
than being lost in an average.
• The exemplar approach also more easily deals with variable categories, like games – can be covered
by variety of exemplars, like football, computer games, marbles, and golf.
Which Approach Works Better: Prototypes or Exemplars?
• People may use both approaches. In early learning, we may use prototypes and then later take in more
exemplar information; in early learning we tend to be poor at taking into account exceptions like
• For small categories, exemplars may work best (e.g. US presidents), while prorotypes are more
efficient for larger categories.
IS THERE A PSYCHOLOGICALLY “PRIVILEGED” LEVEL OF
CATEGORIES? Page 240-262, 22 pages Page 3 of7
• Hierarchical Organization: Larger, more general categories are divided into smaller, more specific
• Although there may be one level that is more psychologically important than others, this may not be the
same level for everyone.
Rosch’s Approach: What’s special about Basic Level
• There are different levels of categories, ranging
from superordinate/global (“furniture”) to basic
(“bed”), to subordinate/specific (“double bed”).
• When asked to list characteristics of furniture, table, and kitchen table, subjects can only list a 3
features common to all furniture, 0 for all tables, and
10.3 for kitchen table
• There is great loss of information about a category
from the basic to the global level (3 to 9), and a gain of
only a little information from basic to specific (9 to 10.3).
Therefore the basic level is psychologically special.
• In a naming task when subjects must name 3 pictures,
most assign a basic level name, e.g. “guitar” rather than
“musical instrument” or “electric guitar”.
How Knowledge can affect Categorization
• Tanaka & Taylor (1991): Naming task performed on bird experts and nonexperts with objects from many
different categories, including of birds. Experts responded the majority of the time by giving specific
species names (robin), while nonexperts mostly responded by saying bird
• To fully understand categorization, must consider not only properties of the objects, but the learning
and experience of the people perceiving these objects. Thus, the level that is psychologically special
is not the same for everyone.
SEMANTIC NETWORKS: REPRESENTING RELATIONSHIPS BTWN
• Semantic Network Approach: Proposes that concepts are arranged in networks
Intro to Semantic Networks: Collins & Quillian’s Hierarchical Model
• The network consists of nodes that are connected by links. Each node represents a category or
concepts, and links connect related concepts; properties associated with each concept are also
• This is still a hierarchical model since it
consists of levels arranged so that more
specific concepts, such as “canary”, are at
the bottom and more general concepts,
such as “animal”, at the top.
• To retrieve properties of “canary”, enter at
the “canary” concept node. From this node,
we see canaries are yellow and can sing.
To access more information, more up the
link to learn a canary is a bird, and that Page 240-262, 22 pages Page 4 of7
birds have wings, can fly, and have feathers. To access even more, find canary is an animal, and thus
• Cognitive Economy: Storing shared properties just once at a higher level, rather than inefficiently for
each node. Exceptions are then encoded at lower nodes, e.g. for “ostrich”, one of its properties is “cannot
• This model is concerned with how concepts and properties are associated in the mind, not how they
actually correspond to specific structures, nerves, or locations in the brain.
• Using the sentence verificat