0% found this document useful (0 votes)

2 views

Chapter 9 - Conceptual Knowledge

The document discusses conceptual knowledge, focusing on how objects are categorized and the various approaches to understanding concepts and categories. It highlights the limitations of definitional approaches and introduces the idea of family resemblance in categorization. The text also outlines different models of categorization, including connectionist models and semantic networks, and explores how these concepts are represented in the brain.

Uploaded by

mollyty.henderson

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

2 views

Chapter 9 - Conceptual Knowledge

Uploaded by

mollyty.henderson

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 34

Conceptual Knowledge

9
Basic Properties of Concepts and Categories The Connectionist Approach
What Is a Connectionist Model?
How Are Objects Placed into Categories?
Why Definitions Don’t Work for Categories How Are Concepts Represented in a Connectionist Network?

The Prototype Approach: Finding the Average Case Representing a Canary

Training a Network
Prototypical Objects Have High Family Resemblance
➤ Demonstration: Family Resemblance ➤ TEST YOURSELF 9.2
Statements About Prototypical Objects Are Verified Rapidly How Concepts Are Represented in the Brain
➤ Method: Sentence Verification Technique
Prototypical Objects Are Named First Four Proposals About How Concepts Are
Represented in the Brain
Prototypical Objects Are Affected More by Priming
The Sensory-Functional Hypothesis
➤ The Exemplar Approach: Thinking About Examples
The Multiple-Factor Approach
Which Approach Works Better: Prototypes or Exemplars?
The Semantic Category Approach
Is There a Psychologically “Basic” The Embodied Approach
Level of Categories?
Summarizing the Approaches
Rosch’s Approach: What’s Special About Basic Level
Categories?
Something to Consider: The Hub
➤ Demonstration: Listing Common Features and Spoke Model
➤ Demonstration: Naming Things ➤ Method: Transcranial Magnetic Stimulation (TMS)
How Knowledge Can Affect Categorization ➤ TEST YOURSELF 9.3
➤ TEST YOURSELF 9.1

Network Models of Categorization

Representing Relationships Among Categories:
Semantic Networks
CHAPTER SUMMARY
Introduction to Semantic Networks: Collins and Quillian’s
Hierarchical Model THINK ABOUT IT
➤ Method: Lexical Decision Task KEY TERMS
Criticism of the Collins and Quillian Model COGLAB EXPERIMENTS

263

08271_ch09_ptg01.indd 263 4/18/18 4:52 PM

SOME QUESTIONS WE
WILL CONSIDER W hat do you think about when you think about knowledge? Facts you need to
know for the cognitive psychology exam? The names of people you know? The
location of that favorite hard-to-find restaurant? What a black hole is? The list of what you
◗ Why is it difficult to decide
if a particular object belongs know is vast, because you know about a lot of things. This way of thinking about knowledge
to a particular category, such is consistent with dictionary definitions of knowledge, such as the following:
as “chair,” by looking up its
definition? (267) Acquaintance with facts, truths, or principles, as from study or investigation.
—Dictionary.com
◗ How are the properties of
various objects “filed away” in Familiarity, awareness, or understanding of someone or something, such as facts, infor-
the mind? (276) mation, descriptions or skills, which is acquired through experience or education by
◗ How is information about perceiving, discovering, or learning.
—Wikipedia
different categories stored in
the brain? (285) Definitions such as these capture most people’s commonsense idea of what knowledge
is. It is, in brief, what we know. But as indicated by the title of this chapter, we will be consid-
ering a narrower conception of knowledge, which cognitive psychologists call conceptual
knowledge—knowledge that enables us to recognize objects and events and to make infer-
ences about their properties.

Conceptual knowledge involves answering questions such as the following:

➤ When we encounter a new item or event in the world, how do we come to know
what kind of thing it is?
➤ How do we tell which items in our environment are horses, bicycles, trees, lakes,
newspapers?
➤ How do we tell dolphins from sharks, or planets from stars? What makes a lemon
a lemon?
➤ What are the various kinds of ‘things’ in the world?” (Rogers & Cox, 2015).

We answer questions such as these all the time, usually without even realizing it. For
example, imagine that you find yourself in an unfamiliar town, where you have never been
before. As you walk down the street, you notice that many things are not exactly the same
as what you would encounter if you were in your own town. On the other hand, there are
many things that seem familiar. Cars pass by, there are buildings on either side of the street
and a gas station on the corner, and a cat dashes across the street and makes it safely to the
other side. Luckily, you know a lot about cars, buildings, gas stations, and cats, so you have
no trouble understanding what is going on.
This chapter is about the conceptual knowledge that enables you to recognize and
understand the objects in the street scene and the world. This knowledge exists in the form
of concepts. Concepts have been defined in a number of ways, including “the mental rep-
resentation of a class or individual” (Smith, 1989) and “categories of objects, events, and
abstract ideas” (Kiefer & Pulvermüller, 2012). To express this in concrete terms, we can say
that the concept “cat” is the answer to the question “What is a cat?” If your answer is that a
cat is an animal that is furry, meows, can be petted, moves, and catches mice, you will have
described some aspects of the concept “cat” (Kiefer & Pulvermüller, 2012).
Because we are interested in our knowledge about the world, we need to go beyond
cats! When we start expanding our scope to dogs, automobiles, can openers, radishes, and
264 roses, things start to become both more complicated and more interesting, because the

08271_ch09_ptg01.indd 264 4/18/18 4:52 PM

Conceptual Knowledge 265

question then becomes “How are all of these things organized in the mind?” One way we
organize concepts is in terms of categories.
A category includes all possible examples of a particular concept. Thus, the category
“cats” includes tabbies, Siamese cats, Persian cats, wildcats, leopards, and so on. Looked at
in this way, concepts provide the rules for creating categories. Thus, the mental represen-
tation for “cat” would affect what animals we place in the “cat” category. Because concepts
provide rules for sorting objects into categories, concepts and categories are often discussed
together, and a great deal of research has focused on the process of categorization—the
process by which things are placed in categories.
Categorization is something we do every time we place an object into a category, and
once we have assigned an object to a category, we know a lot about it. For example, being
able to say that the furry animal across the street is a “cat” provides a great deal of informa-
tion about it (Figure 9.1). Categories have therefore been called “pointers to knowledge”
(Yamauchi & Markman, 2000). Once you know that something is in a category, whether
“cat,” “gas station,” or “impressionist painting,” you can focus your energy on specifying
what’s special about this particular object (see Solomon et al., 1999).
Categorization not only helps understand what is happening in the environment, it also
plays an essential role in enabling us to take action. For example, to spread jam on bread, you
must recognize the jam jar, the bread, and the knife; realize their relevant properties (bread
is soft unless it is toasted; knives are rigid, and jam is sticky); and know how to grasp the
knife handle with the right grip to scrape the jam from its jar (Lambon Ralph et al., 2017).
Being able to place things in categories can also help us understand behaviors that we
might otherwise find baffling. For example, if we see a man with the left side of his face
painted black and the right side painted gold, we might wonder what is going on. However,
once we note that the person is heading toward the football stadium and it is Sunday
afternoon, we can categorize the person as a “Pittsburgh Steelers fan.” Placing him in that

Difficult to train
Catches mice

Has whiskers

Likes milk,
fish

Sleeps a lot,
but more active
at night

Likes to
rub up against
people and other
objects
Bruce Goldstein

Meows and purrs

A feline: related
to lions and tigers

➤ Figure 9.1 Knowing that something is in a category provides a great deal of information about it.

08271_ch09_ptg01.indd 265 4/18/18 4:52 PM

266 CHAPTER 9 Conceptual Knowledge

category explains his painted face and perhaps other strange behaviors that happen to be
normal on game day in Pittsburgh (Solomon et al., 1999).
These various uses of categories testify to their importance in everyday life. It is no
exaggeration to say that if there were no such thing as categories, we would have a very
difficult time dealing with the world. Consider what it would mean if every time you saw a
different object, you knew nothing about it other than what you could find out by investi-
gating it individually. Clearly, life would become extremely complicated if we weren’t able
to rely on the knowledge provided to us by categories.
Yes, categories are important. But what do we need to know in order to understand cat-
egories? The answer to this question isn’t obvious, because we routinely categorize things,
so it seems like an automatic process. It’s obvious that there’s a cat sitting in a chair, across
the room—cat, chair, and room being different categories. These things, and thousands of
others, are so easy to categorize that there seems to be no problem to solve.
But as with other cognitive abilities, just because it’s easy doesn’t mean it’s simple.
Categorization becomes more difficult if you encounter something unfamiliar. “What is
that over there? An aardvark? Very strange.” Or things become even more difficult when
a person suffers a brain injury that makes it difficult to identify different objects or know
what they are used for. Once we understand that there are situations in which categorization
becomes difficult, things become more interesting, because recognizing and understanding
these difficulties is the first step to uncovering the mechanisms of categorization.
This chapter considers, in three major sections, the difficulties of categorization and
the mechanisms involved in day-to-day categorization. Each of the three sections tells
a story that involves a different approach to categorization. First, in the section “Basic
Properties of Concepts and Categories,” we consider a behavioral approach that originated
with a series of experiments begun in the 1970s, which have helped us understand how
we place objects in different categories and which have shown that “not all objects are
created equal.” In the second section, “Network Models of Categorization,” we consider
the network approach to categorization that began in the 1960s, inspired by the emerging
field of computer science, which created computer models of how categories are repre-
sented in the mind. In the third section, “How Categories Are Represented in the Brain,”
we take a physiological approach, which looks at the relationship between categories and
the brain. We will see that each approach provides its own perspective on categorization,
and that all three together provide a more complete explanation of categorization than
any one approach alone.

Basic Properties of Concepts and Categories

We consider the following questions about the basic properties of categories:
➤ How are different objects, events, or ideas assigned to a particular category?
➤ How can categories be defined?
➤ Why do we say that “not all things in categories are created equal”?

How Are Objects Placed into Categories?

A time-honored approach to determining the characteristics of an object is to look up
its definition. We begin by describing how cognitive psychologists have shown that this
“definitional approach” to sorting objects into categories doesn’t work. We then consider
another approach, which is based on determining how similar an object is to other objects
in a category.

08271_ch09_ptg01.indd 266 4/18/18 4:52 PM

How Are Objects Placed into Categories 267

Why Definitions Don’t Work for Categories

According to the definitional approach to categorization, we can decide whether some-
thing is a member of a category by determining whether a particular object meets the defi-
nition of the category. Definitions work well for some things, such as geometric objects.
Thus, defining a square as “a plane figure having four equal sides, with all internal angles the
same” works. However, for most natural objects (such as birds, trees, and plants) and many
human-made objects (like chairs), definitions do not work well at all.
The problem is that not all of the members of everyday categories have the same fea-
tures. So, although the dictionary definition of a chair as “a piece of furniture consisting of a
seat, legs, back, and often arms, designed to accommodate one person” may sound reasonable,
there are objects we call “chairs” that don’t meet that definition. For example, although the
objects in Figures 9.2a and 9.2b would be classified as chairs by this definition, the ones in
Figures 9.2c and 9.2d would not. Most chairs may have legs and a back, as specified in the
definition, but most people would still call the disc-shaped furniture in Figure 9.2c a chair, and
might go so far as to say that the rock formation in Figure 9.2d is being used as a chair.
The philosopher Ludwig Wittgenstein (1953) noted this problem with definitions
and offered a solution:
Consider for example the proceedings we call “games.” I mean board-games, card-games,
ball-games, Olympic games, and so on. For if you look at them you will not see something
in common to all, but similarities, relationships, and a whole series of them at that. I can
think of no better expression to characterize these similarities than “family resemblances.”

(a) (b)
Bruce Goldstein

➤ Figure 9.2 Different objects, all possible “chairs.”

08271_ch09_ptg01.indd 267 4/18/18 4:52 PM

268 CHAPTER 9 Conceptual Knowledge

Wittgenstein proposed the idea of family resemblance to deal with the problem that
definitions often do not include all members of a category. Family resemblance refers to
the idea that things in a particular category resemble one another in a number of ways.
Thus, instead of setting definite criteria that every member of a category must meet, the
family resemblance approach allows for some variation within a category. Chairs may come
in many different sizes and shapes and may be made of different materials, but every chair
does resemble other chairs in some way. Looking at category membership in this way, we
can see that the chair in Figure 9.2a and the chair in Figure 9.2c do have in common that
they offer a place to sit, a way to support a person’s back, and perhaps a place to rest the arms
while sitting.
In a series of experiments beginning in the 1970s, Eleanor Rosch and coworkers used
the idea of family resemblance as a jumping off point for experiments that investigated the
basic nature of categories. One of the early ideas to emerge from these experiments is the
idea of prototypes.

The Prototype Approach: Finding the Average Case

According to the prototype approach to categorization, membership in a category
is determined by comparing the object to a prototype that represents the category. A
prototype is a “typical” member of the category.
What is a typical member of a category? Eleanor Rosch (1973) proposed that the
“typical” prototype is based on an average of members of a category that are commonly
experienced. For example, the prototype for the category “birds” might be based on some of
the birds you usually see, such as sparrows, robins, and blue jays, but doesn’t necessarily look
exactly like any one of them. Thus, the prototype is not an actual member of the category
but is an “average” representation of the category (Figure 9.3).
Of course, not all birds are like robins, blue jays, or sparrows. Owls, buzzards, and pen-
guins are also birds. Rosch describes these variations within categories as representing dif-
ferences in typicality. High typicality means that a category member closely resembles the
category prototype (it is like a “typical” member of the category). Low typicality means that
the category member does not closely resemble a typical member of the category. Rosch
(1975a) quantified this idea by presenting participants with a category title, such as “bird”
or “furniture,” and a list of about 50 members of the category. The participants’ task was
to rate the extent to which each member represented the category title on a 7-point scale,
with a rating of 1 meaning that the member is a very good example of what the category is,
and a rating of 7 meaning that the member fits poorly within the category or is not a
member at all.
Roger Tidman/Corbis Documentary/Getty Images

Gary W. Carter/Corbis Documentary/Getty Images

Paul Reeves Photography/Shutterstock.com

➤ Figure 9.3 Three real birds—a sparrow, a robin, and a blue jay—and a “prototype” bird that is the average
representation of the category “birds.”

08271_ch09_ptg01.indd 268 4/18/18 4:52 PM

How Are Objects Placed into Categories? 269

Results for some of the objects in two different categories

Bat Penguin Owl Sparrow
are shown in Figure 9.4. The 1.18 rating for sparrow reflects (6.15) (4.53) (2.96) (1.18)
the fact that most people consider a sparrow to be a good
example of a bird (Figure 9.4a). The 4.53 rating for penguin
and 6.15 rating for bat reflect the fact that penguins and bats 7 6 5 4 3 2 1
are not considered good examples of birds. Similarly, chair
(a) Category = birds
and sofa (rating = 1.04) are considered very good examples
of furniture, but mirror (4.39) and telephone (6.68) are poor Telephone Mirror China Chair,
examples (Figure 9.4b). The idea that a sparrow is a better (6.68) (4.39) closet sofa
(2.59) (1.04)
example of “bird” than a penguin or a bat is not very surpris-
ing. But Rosch went beyond this obvious result by doing a
series of experiments that demonstrated differences between 7 6 5 4 3 2 1
good and bad examples of a category. Poor Very
good
(b) Category = furniture
Prototypical Objects Have High Family Resemblance
How well do good and poor examples of a category compare
to other items within the category? The following demon- ➤ Figure 9.4 Results of Rosch’s (1975a) experiment, in which
stration is based on an experiment by Rosch and Carolyn participants judged objects on a scale of 1 (good example of a
category) to 7 (poor example): (a) ratings for birds; (b) ratings
Mervis (1975).
for furniture.

D E M O N S T R AT I O N Family Resemblance
Rosch and Mervis’s (1975) instructions were as follows: For each of the following
common objects, list as many characteristics and attributes as you can that you feel
are common to these objects. For example, common characteristics for bicycles
are two wheels, pedals, handlebars, you ride on them, they don’t use fuel, and so
on. Give yourself about a minute to write down the characteristics for each of the
following items:
1. chair 3. sofa
2. mirror 4. telephone

If you responded like Rosch and Mervis’s participants, you assigned many of the same
characteristics to chair and sofa. For example, chairs and sofas share the characteristics of
having legs, having backs, you sit on them, they can have cushions, and so on. When an
item’s characteristics have a large amount of overlap with the characteristics of many other
items in a category, this means that the family resemblance of these items is high. But when
we consider mirror and telephone, we find that there is far less overlap, even though they
were both classified by Rosch and Mervis as “furniture” (Figure 9.4b). Little overlap with
other members of a category means the family resemblance is low.
Rosch and Mervis concluded from their results that there is a strong relationship
between family resemblance and prototypicality. Thus, good examples of the category “fur-
niture,” such as chair and sofa, share many attributes with other members of this category;
poor examples, like mirror and telephone, do not. In addition to the connection between
prototypicality and family resemblance, researchers have determined a number of other
connections between prototypicality and behavior.

Statements About Prototypical Objects Are Verified Rapidly Edward Smith and
coworkers (1974) used a procedure called the sentence verification technique to deter-
mine how rapidly people could answer questions about an object’s category.

08271_ch09_ptg01.indd 269 4/18/18 4:52 PM

270 CHAPTER 9 Conceptual Knowledge

METHOD Sentence Verification Technique

Pomegranate The procedure for the sentence verification technique is
650
simple. Participants are presented with statements and are
asked to answer “yes” if they think the statement is true and
600 “no” if they think it isn’t. Try this yourself for the following
RT (ms)

two statements:

550 An apple is a fruit.

Apple
A pomegranate is a fruit.
500

When Smith and coworkers (1974) used this technique, they

found that participants responded faster for objects that are high
High Medium Low
in prototypicality (like apple for the category “fruit”) than they
Prototypicality
did for objects that are low in prototypicality (like pomegranate;
Figure 9.5). This ability to judge highly prototypical objects more
➤ Figure 9.5 Results of E. E. Smith et al.’s (1974) sentence rapidly is called the typicality effect.
verification experiment. Reaction time (RT) was faster for
objects rated higher in prototypicality. Prototypical Objects Are Named First When participants are
asked to list as many objects in a category as possible, they tend to list
the most prototypical members of the category first (Mervis et al., 1976). Thus, for “birds,”
sparrow would be named before penguin.

Prototypical Objects Are Affected More by Priming Priming occurs when presenta-
tion of one stimulus facilitates the response to another stimulus that usually follows closely
in time (see Chapter 6, page 182). Rosch (1975b) demonstrated that prototypical members
of a category are more affected by a priming stimulus than are nonprototypical members.
The procedure for Rosch’s experiment is shown in Figure 9.6. Participants first heard the
prime, which was the name of a color, such as “green.” Two seconds later they saw a pair of
colors side by side and indicated, by pressing a key as quickly as possible, whether the two
colors were the same or different.
The side-by-side colors that participants saw after hearing the prime were paired in
three different ways: (1) colors were the same and were good examples of the category
(primary reds, blues, greens, etc.; Figure 9.6a); (2) colors were the same but were poor
examples of the category (less rich versions of the good colors, such as light blue, light green,
etc.; Figure 9.6b); (3) colors were different, with the two colors coming from different cat-
egories (for example, pairing red with blue; Figure 9.6c).
The most important result occurred for the two “same” groups. In this condition,
priming resulted in faster “same” judgments for the prototypical (good) colors (reaction
time, RT = 610 ms) than for the nonprototypical (poor) colors (RT = 780 ms). Thus,
when participants heard the word green, they judged two patches of primary green as being
the same more rapidly than two patches of light green.
Rosch explains this result as follows: When participants hear the word green, they
imagine a “good” (highly prototypical) green (Figure 9.7a). The principle behind priming
is that the prime will facilitate the participants’ response to a stimulus if it contains some of
the information needed to respond to the stimulus. This apparently occurs when the good
greens are presented in the test (Figure 9.7b), but not when the poor greens are presented
(Figure 9.7c). Thus, the results of the priming experiments support the idea that partic-
ipants create images of prototypes in response to color names. Table 9.1 summarizes the
various ways that prototypicality affects behavior.
The prototype approach to categorization, and in particular Rosch’s pioneering
research, represented a great advance over the definitional approach because it provided a

08271_ch09_ptg01.indd 270 4/18/18 4:52 PM

How Are Objects Placed into Categories? 271

Reaction time
Prime Test for “same”

Same color — good examples

610 ms

(a)

Same color — poor examples

Hear
“green” 780 ms

(b)

Different color

(c)

➤ Figure 9.6 Procedure for Rosch’s (1975b) priming experiment. Results for the conditions
when the test colors were the same are shown on the right. (a) The person’s “green”
prototype matches the good green but (b) is a poor match for the light green; (c) shows
the condition in which colors were different.

Good match

(b)

Prototype
for green

(c)

Poor
match

“Green”

(a)

➤ Figure 9.7 How Rosch explains the finding that priming resulted in faster “same”
judgments for prototypical colors than for nonprototypical colors.

08271_ch09_ptg01.indd 271 4/18/18 4:52 PM

272 CHAPTER 9 Conceptual Knowledge

TABLE 9.1
Some Effects of Prototypicality
Effect Description Experimental Result
Family Things in a category resemble each Higher ratings for high-prototypical items
resemblance other in a number of ways. when people rate how “good” a member of
the category it is (Rosch, 1975a).

Typicality People react rapidly to members of Faster reaction time to statements like
a category that are “typical” of the “A _______ is a bird” for high-prototypical
category. items (like robin) than for low-prototypical
items (like ostrich) (Smith et al., 1974).

Naming People are more likely to list some High-prototypical items are named first
objects than others when asked to when people list examples of a category
name objects in a category. (Mervis et al., 1976).

Priming Presentation of one stimulus affects Faster same–different color judgments for
responses to a stimulus that follows. high-prototypical items (Rosch, 1975b).

wealth of experimental evidence that all items within a category are not the same. Another
approach to categorization, called the exemplar approach, also takes into account the wide
variation among items that belong to a particular category.

The Exemplar Approach: Thinking About Examples

The exemplar approach to categorization, like the prototype approach, involves deter-
mining whether an object is similar to other objects. However, whereas the standard for the
prototype approach is a single “average” member of the category, the standard for the exem-
plar approach involves many examples, each one called an exemplar. Exemplars are actual
members of the category that a person has encountered in the past. Thus, if a person has
encountered sparrows, robins, and blue jays in the past, each of these would be an exemplar
for the category “birds.”
The exemplar approach can explain many of Rosch’s results, which were used to support
the prototype approach. For example, the exemplar approach explains the typicality effect
(in which reaction times on the sentence verification task are faster for better examples of a
category than for poorer examples) by proposing that objects that are like more of the exem-
plars are classified faster. Thus, a sparrow is similar to many bird exemplars, so it is classified
faster than a penguin, which is similar to few bird exemplars. This is basically the same as
the idea of family resemblance, described for prototypes, which states that “better” objects
will have higher family resemblance.

Which Approach Works Better: Prototypes or Exemplars?

Which approach—prototypes or exemplars—provides a better description of how people
use categories? One advantage of the exemplar approach is that by using real examples, it
can more easily take into account atypical cases such as flightless birds. Rather than com-
paring a penguin to an “average” bird, we remember that there are some birds that don’t
fly. This ability to take into account individual cases means that the exemplar approach
doesn’t discard information that might be useful later. Thus, penguins, ostriches, and other
birds that are not typical can be represented as exemplars, rather than becoming lost in the
overall average that creates a prototype. The exemplar approach can also deal more easily
with variable categories like games. Although it is difficult to imagine what the prototype
might be for a category that contains football, computer games, solitaire, marbles, and golf,
the exemplar approach requires only that we remember some of these varying examples.

08271_ch09_ptg01.indd 272 4/18/18 4:52 PM

Is There a Psychologically “Basic” Level of Categories? 273

Some researchers have concluded that people may use both approaches. It has been
proposed that as we initially learn about a category, we may average exemplars into a pro-
totype; then, later in learning, some of the exemplar information becomes stronger (Keri
et al., 2002; Malt, 1989). Thus, early in learning, we would be poor at taking into account
“exceptions” such as ostriches or penguins, but later, exemplars for these cases would be
added to the category. We know generally what cats are—the prototype—but we know our
own specific cat the best—an exemplar (Minda & Smith, 2001; Smith & Minda, 2000).
A recent survey considering the virtues of both prototypes and exemplars ends with the
following conclusion: “The two kinds of information work together to produce our rich
store of conceptual knowledge allowing each kind of knowledge to explain the tasks that are
most suited for it” (Murphy, 2016).

Is There a Psychologically “Basic” Level

of Categories?
As we have considered the prototype and exemplar approaches, we have used examples of
categories such as “furniture,” which contains members such as beds, chairs, and tables. But,
as you can see in Figure 9.8a, the category “chairs” can contain smaller categories such as
kitchen chairs and dining room chairs. This kind of organization, in which larger, more
general categories are divided into smaller, more specific categories, creating a number of
levels of categories, is called a hierarchical organization.
One question cognitive psychologists have asked about this organization is whether
there is a “basic” level that is more psychologically basic or important than other levels. The
research we will describe indicates that although it is possible to demonstrate that there is a
basic level of categories with special psychological properties, the basic level may not be the
same for everyone. We begin by describing Rosch’s research, in which she introduced the
idea of basic level categories.

Rosch’s Approach: What’s Special About Basic

Level Categories?
Rosch’s research starts with the observation that there are different levels of categories,
ranging from general (like “furniture”) to specific (like “kitchen table”), as shown in
Figure 9.8, and that when people use categories, they tend to focus on one of these levels.
She distinguished three levels of categories: (1) the superordinate level, which we will
call the global level (for example, “furniture”); (2) the basic level (for example, “table”);
and (3) the subordinate level, which we will call the specific level (for example, “kitchen
table”). The following demonstration illustrates some characteristics of the different levels.

Furniture Global Vehicle

(Superordinate)

Chair Table Bed Basic Car Truck Bicycle

Kitchen Dining Kitchen Dining Single Double Specific Ford Chevy Pickup Van Road Trail
room room (Subordinate)
(a) (b)

➤ Figure 9.8 Levels of categories for (a) furniture and (b) vehicle. Rosch provided evidence for the idea that the basic level
is “psychologically privileged.”

08271_ch09_ptg01.indd 273 4/18/18 4:52 PM

274 CHAPTER 9 Conceptual Knowledge

D E M O N S T R AT I O N Listing Common Features

This demonstration is a repeat of the task you did in the Family Resemblance demon-
stration on page 269, but with different categories. For the following categories, list
as many features as you can that would be common to all or most of the objects in the
category. For example, for “table” you might list “has legs.”
1. furniture
2. table
3. kitchen table

If you responded like the participants in the Rosch and

NUMBER OF coworkers’ (1976) experiment, who were given the same task,
LEVEL EXAMPLE COMMON FEATURES you listed only a few features that were common to all furni-
ture but many features that were shared by all tables and by
Global Furniture 3
Lose a all kitchen tables. Rosch’s participants listed an average of 3
lot of common features for the global level category “furniture,” 9 for
information.
basic level categories such as “table,” and 10.3 for specific level
Basic Table 9
Gain just categories such as “kitchen table” (Figure 9.9).
a little Rosch proposed that the basic level is psychologically
information.
special because going above it (to global) results in a large loss of
Specific Kitchen table 10.3
information (9 features at the basic versus 3 at the global level)
and going below it (to specific) results in little gain of informa-
tion (9 features versus 10.3). Here is another demonstration that
➤ Figure 9.9 Category levels, examples of each level, and average
number of common features listed by participants in Rosch is relevant to the idea of a basic level.
et al.’s (1976) experiment.

D E M O N S T R AT I O N Naming Things
Look at Figure 9.10 and, as quickly as possible, write down or say a word that identifies
each picture.

What names did you assign to each object? When Rosch and coworkers (1976) did
a similar experiment, they found that people tended to pick a basic level name. They said
guitar (basic level) rather than musical instrument (global) or rock guitar (specific), fish
rather than animal or trout, and pants rather than clothing or jeans.

➤ Figure 9.10 Stimuli for the Naming Things demonstration.

08271_ch09_ptg01.indd 274 4/18/18 4:53 PM

Is There a Psychologically “Basic” Level of Categories? 275

In another experiment, Rosch and coworkers showed participants a category label,

such as car or vehicle, and then, after a brief delay, presented a picture. The participants’ task
was to indicate, as rapidly as possible, whether the picture was a member of the category.
The results showed that they accomplished this task more rapidly for basic level categories
(such as car) than for global level categories (such as vehicle). Thus, they would respond
“yes” more rapidly when the picture of an automobile was preceded by the word car than
when the picture was preceded by the word vehicle.

How Knowledge Can Affect Categorization

Rosch’s experiments, which were carried out on college undergraduates,
showed that there is a category level, which she called “basic,” that reflects Experts
college undergraduates’ everyday experience. This has been demonstrated Nonexperts
“Sparrow,”
by many researchers in addition to Rosch. Thus, when J. D. Coley and “robin,” etc. “Bird”
coworkers (1997) asked Northwestern University undergraduates to 100
name, as specifically as possible, 44 different plants on a walk around

Percent responses
campus, 75 percent of the responses used labels like “tree,” rather than
more specific labels like “oak.”
50
But instead of asking college undergraduates to name plants, what if
Coley had taken a group of horticulturalists around campus? Do you think
they would have said “tree” or “oak”? An experiment by James Tanaka and
Marjorie Taylor (1991) asked a similar question for birds. They asked bird 0
experts and nonexperts to name pictures of objects. There were objects from sic ific sic ific
Ba Spec Ba Spec
many different categories (tools, clothing, flowers, etc.), but Tanaka and Taylor Categories used to name
were interested in how the participants responded to the four bird pictures. pictures of birds
The results (Figure 9.11) show that the experts responded by spec-
ifying the birds’ species (robin, sparrow, jay, or cardinal), but the nonex- ➤ Figure 9.11 Results of Tanaka and Taylor’s (1991)
perts responded by saying “bird.” Apparently, the experts had learned to “expert” experiment. Experts (left pair of bars) used
pay attention to features of birds that nonexperts were unaware of. Thus, more specific categories to name birds, whereas
in order to fully understand how people categorize objects, we need to nonexperts (right pair of bars) used more basic
consider not only the properties of the objects but also the learning and categories.
experience of the people perceiving those objects (also see Johnson &
Mervis, 1997).
From the result of Tanaka’s bird experiment, we can guess that a horticulturist walking
around campus would be likely to label plants more specifically than people who had little
specific knowledge about plants. In fact, members of the Guatemalan Itzaj culture, who live
in close contact with their natural environment, call an oak tree an “oak,” not a “tree” (Coley
et al., 1997).
Thus, the level that is “special”—meaning that people tend to focus on it—is not the
same for everyone. Generally, people with more expertise and familiarity with a particular
category tend to focus on the more specific information that Rosch associated with the
specific level. This result isn’t surprising, because our ability to categorize is learned from
experience and depends on which objects we typically encounter and what characteristics
of these objects we pay attention to.

T E ST YOUR SELF 9.1

1. Why is the use of categories so important for our day-to-day functioning?
2. Describe the definitional approach to categories. Why does it initially seem like a
good way of thinking about categories, but it then becomes troublesome when
we consider the kinds of objects that can make up a category?

08271_ch09_ptg01.indd 275 4/18/18 4:53 PM

276 CHAPTER 9 Conceptual Knowledge

3. What is the prototype approach? What experiments did Rosch do that

demonstrated connections between prototypicality and behavior?
4. What is the exemplar approach to categorization? How does it differ from the
prototype approach, and how might the two approaches work together?
5. What does it mean to say that there are different levels within a category? What
arguments did Rosch present to support the idea that one of these levels is
“basic”? How has research on categorization by experts led to modifications of
Rosch’s ideas about which category is “basic”?

Network Models of Categorization

We consider the following questions about the network approach to categorization:
➤ How did an early network approach called the semantic network approach use
networks of connected concepts to explain how these concepts are organized in
the mind?
➤ How does a more modern network approach called connectionism describe how
networks can be “trained” to recognize specific objects?

Representing Relationships Among Categories:

Semantic Networks
We have seen that categories can be arranged in a hierarchy of levels, from global (at the
top) to specific (at the bottom). In this section, our main concern is to explain an approach
to categories that focuses on how categories or concepts are organized in the mind. The
approach we will be describing, called the semantic network approach, proposes that con-
cepts are arranged in networks.

Introduction to Semantic Networks: Collins and Quillian’s

Hierarchical Model
One of the first semantic network models was based on the pioneering work of Ross
Quillian (1967, 1969), whose goal was to develop a computer model of human memory.
We will describe Quillian’s approach by looking at a simplified version of his model pro-
posed by Allan Collins and Quillian (1969).
Figure 9.12 shows Collins and Quillian’s network. The network consists of nodes that
are connected by links. Each node represents a category or concept, and concepts are placed
in the network so that related concepts are connected. In addition, a number of properties
are indicated for each concept.
The links connecting the concepts indicate how they are related to each other in the
mind. Thus, the model shown in Figure 9.12 indicates that there is an association in the
mind between canary and bird, and between bird and animal (indicated by the dashes along
the links in Figure 9.12). It is a hierarchical model, because it consists of levels arranged
so that more specific concepts, such as “canary” and “salmon,” are at the bottom, and more
general concepts are at higher levels.
We can illustrate how this network works, and how it proposes that knowledge about
concepts is organized in the mind, by considering how we would retrieve the properties
of canaries from the network. We start by entering the network at the concept node for
“canary.” At this node, we obtain the information that a canary can sing and is yellow.

08271_ch09_ptg01.indd 276 4/18/18 4:53 PM

Representing Relationships Among Categories : Semantic Networks 277

living thing
can
is grow

living
IS A
IS AN

has roots move

can
plant
animal has
skin
IS A feathers IS A
bark petals leaves IS A scales
has has has swim can has
big is has fly can
tree is flower bird fish
has has has
branches pretty wings gills
IS A IS AN IS A IS A IS A IS A IS A IS A

daisy robin sunfish salmon

pine oak rose canary is
is has is is is can is is
green leaves red yellow red sing yellow yellow red

➤ Figure 9.12 Collins and Quillian’s (1969) semantic network. Specific concepts
are indicated in color. Properties of concepts are indicated at the nodes for each
concept. Additional properties of a concept can be determined by moving up
the network, along the lines connecting the concepts. For example, moving from
“canary” up to “bird” indicates that canaries have feathers and wings and can
fly. The dashed lines indicate the distance in the network from canary to bird and
from bird to animal.
(Source: Adapted from T. T. Rogers & J. L. McClelland, 2004)

To access more information about “canary,” we move up the link and learn that a canary is
a bird and that a bird has wings, can fly, and has feathers. Moving up another level, we find
that a canary is also an animal, which has skin and can move, and finally we reach the level
of living things, which tells us it can grow and is living.
You might wonder why we have to travel from “canary” to “bird” to find out that a
canary can fly. That information could have been placed at the canary node, and then we
would know it right away. But Collins and Quillian proposed that including “can fly” at the
node for every bird (canary, robin, vulture, etc.) was inefficient and would use up too much
storage space. Thus, instead of indicating the properties “can fly” and “has feathers” for
every kind of bird, these properties are placed at the node for “bird” because this property
holds for most birds. This way of storing shared properties just once at a higher-level node
is called cognitive economy.
Although cognitive economy makes the network more efficient, it does create a problem
because not all birds fly. To deal with this problem while still achieving the advantages of cog-
nitive economy, Collins and Quillian added exceptions at lower nodes. For example, the node
for “ostrich,” which is not shown in this network, would indicate the property “can’t fly.”
How do the elements in this semantic network correspond to the actual operation of
the brain? The links and nodes we have been describing do not necessarily correspond to
specific nerve fibers or locations in the brain. The Collins and Quillian model is not meant
to mirror physiology but to indicate how concepts and their properties are associated in the
mind, and to make predictions about how we retrieve properties associated with a concept.

08271_ch09_ptg01.indd 277 4/18/18 4:53 PM

278 CHAPTER 9 Conceptual Knowledge

Putting aside any possible connection between semantic

Property
networks and actual physiology, we can ask how accurate the
Category
model’s predictions are. One prediction is that the time it takes
A canary for a person to retrieve information about a concept should be
1500 has skin.
A canary determined by the distance that must be traveled through the
1400 A canary
can fly. network. Thus, the model predicts that when using the sentence
Mean response time (ms)

can sing. verification technique, in which participants are asked to answer

1300 A canary “yes” or “no” to statements about concepts (see Method: Sen-
A canary is an animal.
is a bird.
tence Verification Technique, page 270), it should take longer to
1200 answer “yes” to the statement “A canary is an animal” than to “A
canary is a bird.” This prediction follows from the fact, indicated
1100 A canary
is a canary. by the dashed lines in Figure 9.12, that it is necessary to travel
1000 along two links to get from “canary” to “animal” but only one to
get to “bird.”
900 Collins and Quillian (1969) tested this prediction by mea-
suring the reaction time to a number of different statements and
0 1 2 obtained the results shown in Figure 9.13. As predicted, state-
Levels to be traversed ments that required further travel from “canary” resulted in
longer reaction times.
Another property of the theory, which leads to further predic-
➤ Figure 9.13 Results of Collins and Quillian’s (1969) experiment
that measured reaction times to statements that involved
tions, is spreading activation. Spreading activation is activity that
traversing different distances in the network. Greater distances spreads out along any link that is connected to an activated node.
are associated with longer reaction times, when verifying For example, moving through the network from “robin” to “bird”
statements both about properties of canaries (top) and about activates the node at “bird” and the link we use to get from robin to
categories of which the canary is a member (bottom). bird, as indicated by the colored arrow in Figure 9.14. But accord-
(Source: A. M. Collins et al., 1969) ing to the idea of spreading activation, this activation also spreads
to other nodes in the network, as indicated by the dashed lines. Thus, activating the canary-to-
bird pathway activates additional concepts that are connected to “bird,” such as “animal” and
other types of birds. The result of this spreading activation is that the additional concepts that
receive this activation become “primed” and so can be retrieved more easily from memory.

Animal

Bird Spreading activation

Primed

Robin
Canary Ostrich

Spreading activation primes

canary, ostrich, and animal

➤ Figure 9.14 How activation can spread through a network as a person

searches from “robin” to “bird” (blue arrow). The dashed lines indicate
activation that is spreading from the activated bird node. Circled
concepts, which have become primed, are easier to retrieve from
memory because of the spreading activation.

08271_ch09_ptg01.indd 278 4/18/18 4:53 PM

Representing Relationships Among Categories : Semantic Networks 279

The idea that spreading activation can influence priming was studied by David Meyer
and Roger Schvaneveldt (1971) in a paper published shortly after Collins and Quillian’s
model was proposed. They used a method called the lexical decision task.

METHOD Lexical Decision Task

In the lexical decision task, participants read stimuli, some of which are words and
some of which are not words. Their task is to indicate as quickly as possible whether
each entry is a word or a nonword. For example, the correct responses for bloog
would be “no” and for bloat would be “yes.”

Meyer and Schvaneveldt used a variation of the lexical decision task by presenting par-
ticipants with pairs of words, one above the other, as shown below:

Pair 1 Pair 2 Pair 3 Pair 4

Fundt Bleem Chair Bread
Glurb Dress Money Wheat
The participants’ task was to press, as quickly as possible, the “yes” key when both
950
items were words or the “no” key when at least one item in the pair was a nonword.
Thus, pairs 1 and 2 would require a “no” response, and pairs 3 and 4 would require

Reaction time (ms)

a “yes” response.
The key variable in this experiment was the association between the pairs 850
of real words. In some trials, the words were closely associated (like bread and
wheat), and in some trials, they were weakly associated (chair and money). The
result, shown in Figure 9.15, was that reaction time was faster when the two words
750
were associated. Meyer and Schvaneveldt proposed that this might have occurred
because retrieving one word from memory triggered a spread of activation to other
nearby locations in a network. Because more activation would spread to words
Words Words not
that were related, the response to the related words was faster than the response to associated associated
unrelated words.

Criticism of the Collins and Quillian Model ➤ Figure 9.15 Results of Meyer and
Schvaneveldt’s (1971) experiment.
Although Collins and Quillian’s model was supported by the results of a number Participants responded faster for words that
of experiments, such as their reaction time experiment (Figure 9.13) and Meyer were more closely associated (left bar).
and Schvaneveldt’s priming experiment, it didn’t take long for other researchers to call the
theory into question. They pointed out that the theory couldn’t explain the typicality effect,
in which reaction times for statements about an object are faster for more typical members
of a category than for less typical members (see page 270; Rips et al., 1973). Thus, the state-
ment “A canary is a bird” is verified more quickly than “An ostrich is a bird,” but the model
predicts equally fast reaction times because “canary” and “ostrich” are both one node away
from “bird.”
Researchers also questioned the concept of cognitive economy because of evidence that
people may, in fact, store specific properties of concepts (like “has wings” for “canary”) right
at the node for that concept (Conrad, 1972). In addition, Lance Rips and coworkers (1973)
obtained sentence verification results such as the following:
A pig is a mammal. RT = 1,476 ms
A pig is an animal. RT = 1,268 ms
“A pig is an animal” is verified more quickly, but as we can see from the network in
Figure 9.16, the Collins and Quillian model predicts that “A pig is a mammal” should be

08271_ch09_ptg01.indd 279 4/18/18 4:53 PM

280 CHAPTER 9 Conceptual Knowledge

verified more quickly because a link leads directly from “pig” to “mammal,” but we
Animal
need to travel one link past the “mammal” node to get to “animal.” Sentence veri-
fication results such as these, plus the other criticisms of the theory, led research-
ers to look for alternative ways to using networks to describe how concepts are
organized (Glass & Holyoak, 1975; Murphy et al., 2012) and eventually, in the
1980s, to the proposal of a new approach to networks, called connectionism.
Bird Fish Mammal

The Connectionist Approach

Criticism of semantic networks, combined with advances in understanding how
Canary Salmon information is represented in the brain, led to the emergence of a new approach
Pig
to explaining how knowledge is represented in the mind. In two volumes,
➤ Figure 9.16 Semantic network that shows that both titled Parallel Distributed Processing: Explorations in the Microstructure of
“pig” is closer to “mammal” than to “animal.” Cognition (McClelland & Rumelhart, 1986; Rumelhart & McClelland, 1986),
James McClelland and David Rumelhart proposed a new approach called connec-
tionism. This approach has gained favor among many researchers because (1) it is inspired
by how information is represented in the brain; and (2) it can explain a number of findings,
including how concepts are learned and how damage to the brain affects people’s knowl-
edge about concepts.

What Is a Connectionist Model?

Connectionism is an approach to creating computer models for representing cognitive pro-
cesses. We will focus on connectionist models designed to represent concepts. These models
are also called parallel distributed processing (PDP) models because, as we will see shortly,
they propose that concepts are represented by activity that is distributed across a network.
An example of a simple connectionist network is shown
Output units in Figure 9.17. The circles are units. These units are inspired
by the neurons found in the brain. As we will see, concepts and
their properties are represented in the network by the pattern of
activity across these units.
The lines are connections that transfer information between
units and are roughly equivalent to axons in the brain. Like
neurons, some units can be activated by stimuli from the environ-
Hidden units
ment, and some can be activated by signals received from other
units. Units activated by stimuli from the environment (or stimuli
presented by the experimenter) are input units. In the simple
network illustrated here, input units send signals to hidden units,
which send signals to output units.
An additional feature of a connectionist network is connec-
Input units
tion weights. A connection weight determines how signals sent
from one unit either increase or decrease the activity of the next
➤ Figure 9.17 A parallel distributed processing (PDP) network unit. These weights correspond to what happens at a synapse
showing input units, hidden units, and output units. Incoming that transmits signals from one neuron to another (Figure 2.4,
stimuli, indicated by the arrows, activate the input units, and page 29). In Chapter 7 we saw that some synapses can transmit
signals travel through the network, activating the hidden and signals more strongly than others and therefore cause a high firing
output units. Activity of units is indicated by shading, with rate in the next neuron (Figure 7.12, page 209). Other synapses
darker shading and heavier connector lines indicating more
activity. The patterns of activity that occur in the hidden and
can cause a decrease in the firing rate of the next neuron. Connec-
output units are determined both by the initial activity of the tion weights in a connectionist network operate in the same way.
input units and by the connection weights that determine High connection weights result in a strong tendency to excite the
how strongly a unit will be activated by incoming activity. next unit, lower weights cause less excitation, and negative weights
Connection weights are not shown in this figure. can decrease excitation or inhibit activation of the receiving unit.

08271_ch09_ptg01.indd 280 4/18/18 4:53 PM

The Connectionist Approach 281

Activation of units in a network therefore depends on two things: (1) the signal that origi-
nates in the input units and (2) the connection weights throughout the network.
In the network of Figure 9.17, two of the input units are receiving stimuli. Activation
of each of the hidden and output units is indicated by the shading, with darker shading
indicating more activation. These differences in activation, and the pattern of activ-
ity they create, are responsible for a basic principle of connectionism: A stimulus pre-
sented to the input units is represented by the pattern of activity that is distributed
across the other units. If this sounds familiar, it is because it is similar to the distributed
representations in the brain we described in Chapters 2 (page 43) and 5 (page 154),
Now that we have used the simple network in Figure 9.17, we will consider how some
specific concepts are represented in the more complex connectionist network shown
in Figure 9.18.

Property
living thing
plant
animal
tree
flower
bird
flower
pine is a
Concept
Representation oak
pine Hidden
rose
oak daisy
rose robin
daisy canary
robin sunfish
canary salmon
sunfish
salmon pretty
tall
living
is
green
Relation red
IS A yellow
is
can grow
has move
swim can
fly
sing

bark
petals
wings
feathers
has
scales
gills
roots
skin

➤ Figure 9.18 A connectionist network. Activation of an item unit (“canary”) and a relation
unit (can) causes activity to travel through the network that eventually results in activation
of the property units grow, move, fly, and sing, associated with “canary can.” Shading
indicates the activity of the units, with darker shading indicating more activity. Note that
only a few of the units and connections that would be activated by “canary” and can are
shown as being activated. In the actual network, many more units and connections would
be activated.
(Source: T. T. Rogers & J. L. McClelland, 2004)

08271_ch09_ptg01.indd 281 4/18/18 4:53 PM

282 CHAPTER 9 Conceptual Knowledge

How Are Concepts Represented in a Connectionist Network?

The model in Figure 9.18 was described by James McClelland and Timothy Rogers (2003)
to show how different concepts and their properties can be represented in a connectionist
network. Although this model is more complex than the one in Figure 9.17, it has similar
components: units, links, and connection weights (although the connection weights are
not shown).
Representing a Canary Let’s first compare this model to the Collins and Quillian hi-
erarchical model in Figure 9.12. The first thing to notice is that both models are dealing
with the same concepts. Specific concepts, such as “canary” and “salmon,” shown in blue in
Figure 9.12, are represented on the far left as concept items in Figure 9.18. Also notice that
the properties of the concepts are indicated in both networks by the following four relation
statements: “is a” (A canary is a bird); “is” (A canary is yellow); “can” (A canary can fly); and
“has” (A canary has wings). But whereas the hierarchical network in Figure 9.12 represents
these properties at the network’s nodes, connectionist networks indicate these properties
by activity in the attribute units on the far right, and also by the pattern of activity in the
representation and hidden units in the middle of the network.
Figure 9.18 shows that when we activate the concept “canary” and a relation unit, can,
activation spreads along the connections from “canary” and can so that some of the repre-
sentation units are activated and some of the hidden units are activated. The connection
weights, which are not shown, cause some units to be activated strongly and others more
weakly, as indicated by the shading of the units. If the network is working properly, this
activation in the hidden units activates the grow, move, fly, and sing property units. What’s
important about all of this activity is that the concept “canary” is represented by the pattern
of activity in all of the units in the network.

Training a Network According to the above description, the answer to “A canary can …”
is represented in the network by activation of the property units plus the pattern of activa-
tion of the network’s representation and hidden units. But according to connectionism, this
network had to be trained in order to achieve this result.
We can appreciate the need for training by considering Figure 9.19, which indicates
how the network might have responded before training had occurred. In the untrained
network, stimulating the canary and can units sends activity out to the rest of the network,
with the effect of this activation depending on the connection weights between the units.
Let’s assume that in our untrained network, all of the connection weights are 1.0.
Because the connection weights are the same, activity spreads throughout the network,
and property nodes such as flower, pine, and bark, which have nothing to do with canar-
ies, are activated. For the network to operate properly, the connection weights have to be
adjusted so that activating the concept unit “canary” and the relation unit can activates only
the property units grow, move, fly, and sing. This adjustment of weights is achieved by a
learning process. The learning process occurs when the erroneous responses in the property
units cause an error signal to be sent back through the network, by a process called back
propagation (since the signals are being sent backward in the network starting from the
property units). The error signals that are sent back to the hidden units and the represen-
tation units provide information about how the connection weights should be adjusted so
that the correct property units will be activated.
To explain the idea behind activation and back propagation, let’s consider a behavioral
example. A young child is watching a robin sitting on a branch, when suddenly the robin flies
away. This simple observation, which strengthens the association between “robin” and can fly,
would involve activation. But if the child were to see a canary and say “robin,” the child’s parent
might correct her and say “That is a canary” and “Robins have red breasts.” The information
provided by the parent is similar to the idea of feedback provided by back propagation.

08271_ch09_ptg01.indd 282 4/18/18 4:53 PM

The Connectionist Approach 283

bark
petals
wings
feathers
has
scales
gills
roots
skin

➤ Figure 9.19 How the connectionist network in Figure 9.18 might respond before training,
when all of the connection weights are 1.0. See text for details.

Thus, a child’s learning about concepts begins with little information and some incor-
rect ideas, which are slowly modified in response both to observation of the environment
and to feedback from others. Similarly, the connectionist network’s learning about concepts
begins with incorrect connection weights that result in the activation shown in Figure 9.19,
which are slowly modified in response to error signals to create the correctly functioning
network in Figure 9.18.
Although this “educated” network might work well for canaries, what happens when a
robin flies by and alights on the branch of a pine tree? To be useful, this network needs to be
able to represent not just canaries but also robins and pine trees. Thus, to create a network
that can represent many different concepts, the network is not trained just on “canary.”
Instead, presentations of “canary” are interleaved with presentations of “robin,” “pine tree,”
and so on, with small changes in connection weights made after each presentation.
We can appreciate how this learning process occurs over many trials by looking at the
results of a computer simulation (McClelland & Rogers, 2003). The network in Figure 9.18
was presented with a number of different concepts and relation statements, one after another,
and the activity of the units and connection weights between units were calculated by the
computer. Figure 9.20 indicates the activation of the eight representation units in response to

08271_ch09_ptg01.indd 283 4/18/18 4:53 PM

284 CHAPTER 9 Conceptual Knowledge

Learning trials
0 250 750 2,500

Rose
Activity in hidden units

Daisy
Canary

Hidden units

➤ Figure 9.20 Learning in a connectionist network. Bars represent activity in the eight
representation units. Notice how the pattern of activation changes as learning
progresses.
(Source: Adapted from J. L. McClelland & T. T. Rogers, 2003)

the concepts “canary,” “rose,” and “daisy.” At the beginning of the process, the experimenter set
the connection weights so that activity was about the same in each unit (Learning trials = 0).
This corresponds to the initially weak and undifferentiated activation we discussed earlier.
As learning progressed, with each concept being presented one after another and the
computer changing the weights just slightly after each trial in response to error signals, the
patterns became adjusted, so by Trial 250, the patterns for “canary” and “daisy” begin to look
different. By Trial 2,500, it is easy to tell the difference between the patterns for “canary” and
“daisy,” while the two flowers, “daisy” and “rose,” have similar but slightly different patterns.
Although our description has been based on one particular connectionist network,
most networks have similar properties. Connectionist networks are created by a learning
process that shapes the networks so information about each concept is contained in the
distributed pattern of activity across a number of units.
Notice how different this operation of the connectionist network is from the operation
of Collins and Quillian’s hierarchical network, in which concepts and their properties are
represented by activation of different nodes. Representation in a connectionist network is
far more complex, involving many more units for each concept, but it is also much more like
what happens in the brain.
Because of the resemblance between connectionist networks and the brain, and the
fact that connectionist networks have been developed that can simulate normal cognitive
functioning for processes such as language processing, memory, and cognitive development
(Rogers & McClelland, 2004; Seidenberg & Zevin, 2006), many researchers believe that
the idea that knowledge is represented by distributed activity holds great promise. The fol-
lowing results also support the idea of connectionism:
1. The operation of connectionist networks is not totally disrupted by damage. Because
information in the network is distributed across many units, damage to the system
does not completely disrupt its operation. This property, in which disruption of
performance occurs only gradually as parts of the system are damaged, is called
graceful degradation. It is similar to what often happens in actual cases of brain
damage, in which damage to the brain causes only a partial loss of functioning.

08271_ch09_ptg01.indd 284 4/18/18 4:53 PM

Four Proposals About How Concepts Are Represented in the Brain 285

Some researchers have suggested that studying the way networks respond to dam-
age may suggest strategies for rehabilitation of human patients (Farah et al., 1993;
Hinton & Shallice, 1991; Olson & Humphreys, 1997; Plaut, 1996).
2. Connectionist networks can explain generalization of learning. Because similar
concepts have similar patterns, training a system to recognize the properties of
one concept (such as “canary”) also provides information about other, related
concepts (such as “robin” or “sparrow”). This is similar to the way we actually
learn about concepts because learning about canaries enables us to predict proper-
ties of different types of birds we’ve never seen (see McClelland et al., 1995).
While active research on connectionism continues in many laboratories, some research-
ers point out that there are limits to what connectionist networks can explain. Whatever
the final verdict on connectionism, this approach has stimulated a great deal of research,
some of which has added to our understanding of both normal cognition and how brain
damage affects cognition. In the next section, we will focus even more directly on the brain
by considering neuropsychological and brain imaging research on how concepts are repre-
sented in the brain.

T E ST YOUR SELF 9.2

1. What is the basic idea behind Collins and Quillian’s semantic network approach?
What is the goal of this approach, and how did the network created by Collins
and Quillian accomplish this goal?
2. What is the evidence for and against the Collins and Quillian model?
3. What is a connectionist network? Describe how a connectionist network learns,
considering specifically what happens to connection weights.
4. How does the way information is represented in a connectionist network differ
from the way it is represented in a semantic network?
5. How are connectionist networks affected by damage? How is this similar to what
happens in cases of brain damage?
6. How do connectionist networks explain generalization of learning?

How Concepts Are Represented in the Brain

We consider the following questions about how categories are represented in the brain:
➤ What do the results of neuropsychological research tell us about where different
categories are represented in the brain?
➤ How has this neuropsychological research led to a number of different models that
explain how categories are organized in the brain?
➤ What does brain imaging research tell us about how and where different categories
are represented in the brain?

Four Proposals About How Concepts

Are Represented in the Brain
Early research on how concepts are represented in the brain was based on studying patients
whose brain damage caused them to lose the ability to understand specific types of con-
cepts. This research led to the proposal of the sensory-functional hypothesis.

08271_ch09_ptg01.indd 285 4/18/18 4:53 PM

286 CHAPTER 9 Conceptual Knowledge

The Sensory-Functional Hypothesis

Nonanimals
In one of the classic papers in neuropsychology, Elizabeth Warrington and Tim
(nonliving things,
fruits, and vegetables)
Shallice (1984) reported on four patients who had suffered memory loss from
Animals encephalitis. These patients had a category-specific memory impairment—
100 an impairment in which they had lost the ability to identify one type of object
but retained the ability to identify other types of objects. Specifically, these
Percent correct performance

80 patients were able to identify nonanimals, like furniture and tools, as well as
fruits and vegetables, but had impaired ability to identify animals (Figure 9.21).
60 (As we discuss various cases next, we will use the term artifacts to refer to non-
living things, which would include furniture and tools.)
40
To explain why this selective impairment occurred, Warrington and Shal-
lice considered properties that people use to distinguish between artifacts and
20
living things. They noted that distinguishing living things depends on perceiv-
ing their sensory features. For example, distinguishing between a tiger and a
0
Patient K.C. Patient E.W. leopard depends on perceiving stripes and spots. Artifacts, in contrast, are more
likely to be distinguished by their function. For example, a screwdriver, chisel,
and hammer are all tools but are used for different purposes (turning screws,
➤ Figure 9.21 Performance on a naming task
scraping, and pounding nails) (Table 9.2).
for patients K.C. and E.W., both of whom
had category-specific memory impairment. The observation that living things are distinguished by sensory properties
They were able to correctly name pictures of and artifacts by functions led to the sensory-functional (S-F) hypothesis,
nonliving things (such as car and table) and which states that our ability to differentiate living things and artifacts depends
fruits and vegetables (such as tomato and pear) on a memory system that distinguishes sensory attributes and a system that dis-
but performed poorly when asked to name tinguishes functions.
pictures of animals. While the S-F hypothesis explained the behavior of Warrington and Shallice’s
(Source: B. Z. Mahon & A. Caramazza, 2009) patients, plus dozens of other patients, researchers began describing cases that
couldn’t be explained by this hypothesis. For example, Matthew Lambon Ralph and cowork-
ers (1998) studied a patient who had a sensory deficit—she performed poorly on perceptual
tests—yet she was better at identifying animals than artifacts, which is the opposite of what
the S-F hypothesis predicts. In addition, there are patients who are able to identify mechanical
devices even if they perform poorly for other types of artifacts. For example, Hoffman and
Lambon Ralph describe patients who have poor comprehension of small artifacts like tools
but better knowledge of larger artifacts, such as vehicles (Cappa et al., 1998; Hillis et al., 1990;
Warrington & McCarthy, 1987). Thus, “artifacts” are not a single homogeneous category as
hypothesized by the S-F hypothesis. Results such as these have led many researchers to con-
clude that many effects of brain damage can’t be explained by the simple distinction between
sensory and function (Hoffman & Lambon, 2013).

TABLE 9.2
The Sensory-Functional Hypothesis
Relevant Information Relevant to … Deficit in ability to process relevant
information causes…
Sensory Living things Problems identifying living things
(Ex: a tiger has stripes)

Functional Artifacts Problems identifying artifacts

(Ex: A hammer hits nails)

The Multiple-Factor Approach

The idea of distributed representation is a central feature of the multiple-factor approach,
which has led to searching for factors beyond sensory and functional that determine how
concepts are divided within a category. We can appreciate this approach by posing the

08271_ch09_ptg01.indd 286 4/18/18 4:53 PM

Four Proposals About How Concepts Are Represented in the Brain 287

TABLE 9.3
Sample Stimuli and Question Used in the Hoffman and Lambon Ralph (2013)
Experiment
A. A Few of the 160 Items Presented to Subjects
Mammal Machine Clothing

Pet Vehicle Weapon

Bird Furniture Tool

Door Fish Fruit

B. Question for Subjects

How much do you associate (insert item from list above) with a particular . . .

Color Taste

Visual form Smell

Motion Tactile (Touch)

Sound Performed action (in which you interact with the object)

following question: Assume that we start with a large number of items selected from lists of
different types of animals, plants, and artifacts. If you wanted to arrange them in terms of
how similar they are to each other, how would you do it? You could arrange them by shape,
but then items like a pencil, a screwdriver, a person’s finger, and a breakfast sausage might be
grouped together. Or considering just color, you could end up placing fir trees, leprechauns,
and Kermit the Frog together. Although it is true that members of specific categories do
share similar perceptual attributes, it is also clear that we need to use more than just one or
two features when grouping objects in terms of similarity.
Taking this idea as their starting point, researchers picked a number of different
features and had participants rate a large number of items with regard to these fea-
tures. This was the idea behind an experiment by Paul Hoffman and
Matthew Lambon Ralph (2013), who used 160 items like the ones shown
in Table 9.3a. The participants’ task was to rate each item on the features Animals
shown in Table 9.3b. For example, for the concept “door,” the participant Artifacts
would be asked “How much do you associate door with a particular color 1

(or form, or motion, etc.)?” Participants assigned a rating of 7 for “very

0.5
strongly” to 1 for “not at all.”
Mean rating

The results, shown in Figure 9.22, indicate that animals were more highly
0
associated with motion and color compared to artifacts, and artifacts were
more highly associated with performed actions (actions associated with using –0.5
or interacting with an object). This result conforms to the S-F hypothesis, but
when Hoffman and Lambon Ralph looked at the groupings more closely, they –1
found some interesting results. Mechanical devices such as machines, vehicles, Color Motion Performed
action
and musical instruments overlapped with both artifacts (involving performed
actions) and animals (involving sound and motion). For example, musical
instruments are associated with specific actions (how you play them), which ➤ Figure 9.22 How participants rated animals
and artifacts on color, motion, and performed
goes with artifacts, and are also associated with sensory properties (their visual
actions. Animals are rated higher on color
form and the sounds they create), which goes with animals. Thus, musical and motion; artifacts are rated higher on
instruments and some mechanical devices occupy a middle ground between performed actions.
artifacts and living things, because they can involve both action knowledge (Source: Based on data in P. Hoffman & M. A. Lambon
and sensory attributes. Ralph, 2013)

08271_ch09_ptg01.indd 287 4/18/18 4:53 PM

288 CHAPTER 9 Conceptual Knowledge

Bruce Goldstein
Albert Pego/Shutterstock.com
➤ Figure 9.23 Some animals and vehicles. Notice that the animals are more similar to each
other than are the vehicles. This higher similarity of animals is called crowding.

Another factor that researchers have proposed to differentiate between animals and
artifacts is crowding, which refers to the fact that animals tend to share many properties
(like eyes, legs, and the ability to move). In contrast, artifacts like cars and boats share fewer
properties, other than that they are both vehicles (Figure 9.23) (Rogers & Cox, 2015;
Tyler & Moss 2001). This has led some researches to propose that patients who appear
to have category-specific impairments, such as difficulty recognizing living things but not
artifacts, don’t really have a category-specific impairment at all. They propose that these
patients have difficulty recognizing living things because they have difficulty distinguishing
between items that share similar features. According to this idea, because animals tend to be
more similar then artifacts, these patients find animals harder to recognize (Cree & McRae,
2003; Lambon Ralph et al., 2007).

The Semantic Category Approach

The semantic category approach proposes that there are specific neural circuits in the
brain for some specific categories. According to Bradford Mahon and Alfonso Caramazza
(2011), there are a limited number of categories that are innately determined because of
their importance for survival. This idea is based on research that we described in Chapter 2,
which identified areas of the brain that respond to specific types of stimuli such as faces,
places, and bodies (page 42). In addition, we described an experiment by Alex Huth and
coworkers (2012), which resulted in the map in Figure 2.20 (page 43), showing where
different categories are represented in the cortex. This “category map” was determined by
measuring fMRI responses as participants were viewing films, and determining how indi-
vidual voxels responded to objects in the films. But semantic categories come into play not
only when we look at a scene but also when we listen to someone speaking. Understanding
spoken language involves not only knowing about concrete categories like living things,
food, and places, but about abstract concepts like feelings, values, and thoughts.
To create a map based on spoken language, Huth and coworkers (2016) used a pro-
cedure similar to their earlier experiment, but instead of having their participants view
films, they had them listen to more than two hours of stories from The Moth Radio Hour
broadcast (themoth.org) while in a brain scanner. Figure 9.24a shows a map that extends
over a large area of the cortex, which indicates were specific words activate the cortex.
Figure 9.24b zooms in to make some of the words easier to read. Figure 9.25 shows the

08271_ch09_ptg01.indd 288 4/18/18 4:53 PM

Four Proposals About How Concepts Are Represented in the Brain 289

Courtesy of Alexander Huth

(a) (b)

➤ Figure 9.24 Results of the Huth et al. (2016) experiment in which participants listened to
stories in a scanner. (a) Words that activated different places on the cortex. (b) Close-up
of a smaller area of cortex. Note that a particular area usually responded to a number of
different words, as indicated in Figure 9.25.

cortex color-coded to indicate where different categories of words activate the cortex.
For example, the light area in the back of the brain is activated by words associated with
violence. The words that activated a single voxel in that area are indicated on the right.
Another voxel, which is activated by words associated with visual qualities, is shown in the
green area near the top of the brain. An interesting aspect of Huth’s results is that the maps
were very similar for each of the seven participants.
While the semantic category approach focuses on areas of the brain that are specialized
to respond to specific types of stimuli, it also emphasizes that the brain’s response to items

ribbons sleeve
violence
colour satin
coloured tunic
social
body part
glossy colored strips person
visual
brown cap red mental

striped stripes silver place

inch gold
outdoor
leather yellow
thinner
believed
pregnant
judge guilty
death claims
innocent child
killed victim
murderer
stated
false birth
claim
believes
claiming
accusation
Courtesy of Alexander Huth

➤ Figure 9.25 More results from the Huth et al. (2016) experiment. Colors on the cortex
indicate where different categories of words caused activation, as indicated on the legend in
the upper right. The pink words activated a voxel that responded to words related to violence.
The green words activated a voxel that responded to visual qualities.

08271_ch09_ptg01.indd 289 4/18/18 4:53 PM

290 CHAPTER 9 Conceptual Knowledge

from a particular category is distributed over a number of different cortical areas (Mahon
et al., 2007; Mahon & Caramazza, 2011). Thus, identifying faces may be based on activity
in the face area in the temporal lobe (see Chapter 2, page 42), but it also depends on activity
in areas that respond to emotions, facial expressions, where the face is looking, and the face’s
attractiveness (see page 43).
Similarly, the response to a hammer activates visual areas that respond to the hammer’s
shape and color, but it also causes activity in areas that respond to how a hammer is used and
to a hammer’s typical motions. This idea that some objects, like hammers, cause activity in
areas of the brain associated with actions, brings us to the embodied approach to categorization.

The Embodied Approach

The embodied approach states that our knowledge of concepts is based on reactivation
of sensory and motor processes that occur when we interact with the object. According to
this idea, when a person uses a hammer, sensory areas are activated in response to the ham-
mer’s size, shape, and color, and, in addition, motor areas are activated that are involved in
carrying out actions involved in using a hammer. When we see a hammer or read the word
hammer later, these sensory and motor areas are reactivated, and it is this information that
represents the hammer (Barsalou, 2008).
We can understand the basis of the embodied approach by returning to Chapter 3,
where we described how perception and taking action interact, as when Crystal reached
across the table to pick up a cup of coffee (page 81). The important message behind
that example was that even simple actions involve a back-and-forth interaction between
pathways in the brain involved in perception and pathways involved in taking action
(see Figure 3.31, page 83) (Almeida et al., 2014).
Also in Chapter 3, we saw how mirror neurons in the prefrontal cortex fire both when a
monkey performs an action and when the monkey sees the experimenter performing the same
action (review pages 85–87; Figure 3.35). What do mirror neurons have to do with concepts?
The link between perception (a neuron fires when watching the experimenter pick up the food)
and motor responses (the same neuron fires when the monkey
picks up the food) is central to the embodied approach’s pro-
posal that thinking about concepts causes activation of percep-
O.Hauk, I. Johnsrude, & F. Pulvermuller, Somatopic representation of action words in human motor

Movements Action Words

tual and motor areas associated with these concepts. Evidence
for this link between perceptual and motor responses in the
human brain is provided by an experiment by Olaf Hauk and
and premotor cortex, Neuron, 41,301-307,2004, with permission from Elsevier.

coworkers (2004), who measured participants’ brain activity

using fMRI under two conditions: (1) as participants moved
their right or left foot, left or right index finger, or tongue;
(2) as participants read “action words” such as kick (foot
action), pick (finger or hand action), or lick (tongue action).
The results show areas of the cortex activated by the
actual movements (Figure 9.26a) and by reading the action
words (Figure 9.26b). The activation is more extensive for
actual movements, but the activation caused by reading the
Blue: Foot movements Blue: Leg words
Red: Finger movements Red: Arm words
words occurs in approximately the same areas of the brain. For
Green: Tongue movements Green: Face words example, leg words and leg movements elicit activity near the
(a) (b) brain’s centerline, whereas arm words and finger movements
elicit activity farther from the centerline. This correspon-
➤ Figure 9.26 Hauk et al. (2004) results. Colored areas indicate
dence between words related to specific parts of the body and
the areas of the brain activated by (a) foot, finger, and tongue the location of brain activity is called semantic somatotopy.
movements; (b) leg, arm, and face words. Although there is convincing evidence linking concepts
(Source: Hauk et al., 2004) and activation of motor areas in the brain, some researchers

08271_ch09_ptg01.indd 290 4/18/18 4:53 PM

Something to Consider: The Hub and Spoke Model 291

question whether the embodied approach offers a complete explanation of how the brain
processes concepts (Almeida et al., 2013; Chatterjee, 2010; Dravida et al., 2013; Goldinger
et al., 2016). For example, Frank Garcea and coworkers (2013) tested patient AA, who had
suffered a stroke that affected his ability to produce actions associated with various objects.
Thus, when AA was asked to use hand motions to indicate how he would use objects such
as a hammer, scissors, and a feather duster, he was impaired compared to normal control
participants in producing these actions. According to the embodied approach, a person
who has trouble producing actions associated with objects should have trouble recognizing
the objects. AA was, however, able to identify pictures of the objects. Garcea and cowork-
ers concluded from this result that the ability to represent motor activity associated with
actions is not necessary for recognizing objects, as the embodied approach would predict.
Another criticism of the embodied approach is that it isn’t well suited to explaining our
knowledge of abstract concepts such as “democracy” or “truth.” However, proponents of
the embodied approach have offered explanations in response to these criticisms (which we
won’t go into here; see Barsalou, 2005; Chatterjee, 2010).

Summarizing the Approaches

Our survey of how concepts are represented in the brain began with the sensory-functional
approach, based on neuropsychological studies begun in the 1980s. But once it became
clear that things were more complex than the distinction between sensory and functional,
research led in different directions that resulted in more complex hypotheses.
One thing that all of the approaches agree on is that information about concepts is
distributed across many structures in the brain, with each approach emphasizing different
types of information. The multiple-factor approach emphasizes the role of many differ-
ent features and properties. The category-specific approach emphasizes specialized areas
of the brain and networks connecting these areas, and the embodied approach emphasizes
activity caused by the sensory and motor properties of objects. It is likely that, as research
on concepts in the brain continues, the final answer will contain elements of each of these
approaches (Goldstone et al., 2012).

Something to Consider: The Hub and Spoke Model

The ideas we have been discussing about how concepts are represented in the brain have
been based largely on patients with category-specific memory impairments. However, there
is another type of problem, called semantic dementia, which causes a general loss of knowl-
edge for all concepts. Patients with semantic dementia tend to be equally deficient in iden-
tifying living things and artifacts (Patterson et al., 2007).
The generalized nature of the deficits experienced by semantic dementia patients, along
with the finding that the anterior temporal lobe (ATL) (purple area in Figure 9.27a) is
generally damaged in these patients, has led some researchers to propose the hub and spoke
model of semantic knowledge. According to this model, areas of the brain that are associated
with specific functions are connected to the ATL, which serves as a hub that integrates the
information from these areas. These functions, indicated in Figure 9.27a, include valence—
which is weak versus strong (yellow); speech (pink); auditory (red); praxis—which refers to
involving manipulation (dark blue); functionality (light blue); and visual (green).
Evidence supporting the idea of a hub with spokes is that damage to one of the spe-
cialized brain areas (the spokes) can cause specific deficits, such as an inability to identify
artifacts, but damage to the ATL (the hub) causes general deficits, as in semantic demen-
tia (Lambon Ralph et al., 2017; Patterson et al., 2007). This difference between hub and
spoke functions has also been demonstrated in non–brain-damaged participants using a
technique called transcranial magnetic stimulation (TMS).

08271_ch09_ptg01.indd 291 4/18/18 4:53 PM

292 CHAPTER 9 Conceptual Knowledge

TMS slowing (% baseline RT)

8
IPL TMS
4

–4
Manmade Living Low High
Category Item manipulability (Praxis)

TMS slowing (% baseline RT) 16

ATL TMS 8

–4
Manmade Living Low High
Category Item manipulability (Praxis)

(a) Neuroanatomical sketch of the (b) Effect of TMS stimulation on (c) Effect of TMS stimulation on
hub-and-spoke framework manmade and living things. manipulability Top: Parietal
Top: Parietal TMS; Bottom: ATL TMS. TMS; Bottom: ATL TMS.

➤ Figure 9.27 (a) The hub and spoke model proposes that areas of the brain specialized for different functions are linked to
the anterior temporal lobe (purple), which integrates the information from areas that serve the following functions: Valence
(yellow); speech (pink); auditory (red); praxis (dark blue); functionality (light blue); visual (green). The dark blue area is in the
parietal cortex. (b) Effect of TMS stimulation on man-made objects versus living things. Top: Parietal stimulation causes more
slowing of RT to man-made objects than to living things. Bottom: ATL stimulation causes the same effect on both. (c) Effect of
TMS stimulation on low versus high manipulability objects. Top: Parietal stimulation causes more slowing of RT to high
manipulability objects compared to low. Bottom: ATL stimulation causes the same effect on both.
(Source: Adapted from Lambon Ralph et al., 2017. Supplementary Figure 5. Based on data from Pobric et al., 2013.)

METHOD Transcranial Magnetic Stimulation (TMS)

It is possible to temporarily disrupt the functioning of a particular area of the human
brain by applying a pulsating magnetic field using a stimulating coil placed over the
person’s skull (Figure 9.28). A series of pulses presented to a particular area of the brain
for seconds or minutes temporarily interferes with brain functioning in that area. If a
particular behavior is disrupted by the pulses, researchers conclude that the disrupted
area of the brain is involved in that behavior.

Gorana Pobric and coworkers (2010) presented pictures of living things and artifacts to
participants and measured the response time for naming each picture. They then repeated
this procedure while TMS was being applied either to the ATL or to an area in the pari-
etal lobe that is normally activated when a person is manipulating an object. Figure 9.27b

08271_ch09_ptg01.indd 292 4/18/18 4:53 PM

Chapter Summary 293

indicates that parietal inactivation (top) slowed reaction times for manmade objects but
not for living objects, whereas ATL inactivation (bottom) caused the same effect for
both manmade and living objects. Figure 9.27c indicates that parietal activation caused a
large slowing in reaction time for highly manipulatable objects (like some tools) but not
for non-manipulatable objects (like furniture), whereas ATL inactivation affected both
types of objects equally. This result—a general effect of stimulating the “hub” (ATL),
but a more specific effect of stimulating an area that would be associated with one of the
“spokes” (parietal cortex)—supports the idea of a hub with general functions and spokes
with more specific functions ( Jefferies, 2013; Lambon-Ralph et al., 2017).
Most researchers agree that the ATL plays a role in integrating information from dif-
ferent areas. But it has also been suggested that other structures may be “hubs,” or that
the most important way concepts are represented is not by “hubs” but by the pattern of
connections formed between the “spokes” (Pulvermüller, 2013). Thus, as we noted at the
end of the last section, research on how concepts are represented in the brain is still a
“work in progress.”

T E ST YOUR SELF 9.3

1. What is the sensory-functional hypothesis of brain categorization? Describe the
➤ Figure 9.28 TMS coil positioned to
neuropsychological evidence that supports this hypothesis.
present a magnetic field to a person’s
2. Describe neuropsychological evidence that can’t be explained by the S-F head. The coil in this position is
hypothesis. stimulating the occipital lobe.
3. What is the multiple-factor approach? How did research on non–brain-damaged
participants support this approach?
4. What is crowding?
5. What is the semantic category approach? What do the results of Huth’s imaging
experiment in which participants had their brain scanned while listening to
stories indicate about how concepts are represented in the brain?
6. What is the embodied approach?
7. What are mirror neurons and how are they related to the embodied approach?
8. Describe Hauk’s brain imaging experiment that supports the embodied
approach and Garcea’s neuropsychology study that argues against the
embodied approach.
9. What is the hub and spoke model? Where is the hub?
10. How do the results of transcranial magnetic stimulation experiments provide
evidence for the hub and spoke model?

CHAPTER SUMMARY
1. Semantic memory is our memory for facts and conform to the definition. The philosopher Wittgenstein
knowledge. proposed the idea of family resemblances to deal with
2. Categories are “pointers to knowledge.” Once you know the fact that definitions do not include all members of
something is in a category, you know a lot of general things a category.
about it and can focus your energy on specifying what is 4. The idea behind the prototypical approach to categorization
special about this particular object. is that we decide whether an object belongs to a category by
3. The definitional approach to categorization doesn’t work deciding whether it is similar to a standard representative of
because most categories contain members that do not the category, called a prototype. A prototype is formed by

08271_ch09_ptg01.indd 293 4/18/18 4:53 PM

294 CHAPTER 9 Conceptual Knowledge

averaging category members a person has encountered in are organized in the mind. Collins and Quillian’s model
the past. is a network that consists of nodes connected by links.
5. Prototypicality is a term used to describe how well an object Concepts and properties of concepts are located at
resembles the prototype of a particular category. the nodes. Properties that hold for most members of a
concept are stored at higher level nodes. This is called
6. The following is true of high-prototypical objects: (a) they
cognitive economy.
have high family resemblance; (b) statements about them
are verified rapidly; (c) they are named first; and (d) they are 13. Collins and Quillian’s model is supported by the results of
affected more by priming. experiments using the sentence verification technique. The
spreading activation feature of the model is supported by
7. The exemplar approach to categorization involves
priming experiments.
determining whether an object is similar to an exemplar.
An exemplar is an actual member of a category that a person 14. The Collins and Quillian model has been criticized for
has encountered in the past. several reasons: It can’t explain the typicality effect, the
idea of cognitive economy doesn’t always hold, and it can’t
8. An advantage to the exemplar approach is that it doesn’t
explain all results of sentence verification experiments.
discard information about atypical cases within a category,
such as penguin in the “bird” category. The exemplar 15. The connectionist approach proposes that concepts are
approach can also deal more easily with categories that represented in networks that consist of input units, hidden
contain widely varying members, such as games. units, and output units, and that information about concepts
is represented in these networks by a distributed activation
9. Researchers have concluded that people use both approaches
of these units. This approach is also called the parallel
to categorization. Prototypes may be more important
distributed processing (PDP) approach.
as people initially learn about categories; later, exemplar
information may become more important. 16. Connectionist networks learn the correct distributed pattern
for a particular concept through a gradual learning process
10. The kind of organization in which larger, more general
that involves adjusting the weights that determine how
categories are divided into smaller, more specific categories
activation is transferred from one unit to another.
is called hierarchical organization. Experiments by Rosch
indicate that a basic level of categories (such as guitar, as 17. Connectionist networks have a number of features that
opposed to musical instrument or rock guitar) is a “basic” enable them to reproduce many aspects of human concept
level of categorization that reflects people’s everyday formation.
experience. 18. Four approaches to explaining how concepts are represented
11. Experiments in which experts were tested show that the in the brain are the sensory-functional hypothesis, the
basic level of categorization can depend on a person’s degree semantic category approach, the multiple-factor approach,
of expertise. and the embodied approach.
12. The semantic network approach proposes that concepts 19. The hub and spoke model proposes that different functions in
are arranged in networks that represent the way concepts the brain are integrated by the anterior temporal lobe (ATL).

THINK ABOUT IT
1. In this chapter, we have seen how networks can be people to name, as quickly as possible, three typical “birds”
constructed that link different levels of concepts. In or “vehicles” or “beverages.” What do the results of this
Chapter 7 we saw how networks can be constructed survey tell you about what level is “basic” for different
that organize knowledge about a particular topic (see people? What do the results tell you about the variability of
Figure 7.5). Create a network that represents the different people’s conception of categories?
material in this chapter by linking together things that 3. Try asking a number of people to name the objects pictured
are related. How is this network similar to or different in Figure 9.10. Rosch, who ran her experiment in the early
from the semantic network in Figure 9.12? Is your 1970s, found that the most common responses were guitar,
network hierarchical? What information does it contain fish, and pants. Notice whether the responses you receive
about each concept? are the same as or different from the responses reported
2. Do a survey to determine people’s conception of “typical” by Rosch. If they are different, explain why you think this
members of various categories. For example, ask several might have occurred.

08271_ch09_ptg01.indd 294 4/18/18 4:53 PM

CogLab Experiments 295

KEY TERMS
Anterior temporal lobe (ATL), 291 Exemplar, 272 Prototype approach to
Back propagation, 282 Exemplar approach to categorization, 268
Basic level, 273 categorization, 272 Semantic category approach, 288
Categorization, 265 Family resemblance, 268 Semantic dementia, 291
Category, 265 Global level, 273 Semantic network approach, 276
Category-specific memory Graceful degradation, 284 Semantic somatotopy, 290
impairment, 286 Hidden units, 280 Sensory-functional (S-F)
Cognitive economy, 277 Hierarchical model, 276 hypothesis, 286
Concept, 264 Hierarchical organization, 273 Sentence verification technique, 269
Conceptual knowledge, 264 Hub and spoke model, 291 Specific level, 273
Connection weight, 280 Input units, 280 Spreading activation, 278
Connectionism, 280 Lexical decision task, 279 Subordinate (specific) level, 273
Connectionist network, 280 Multiple-factor approach, 286 Superordinate (global) level, 273
Crowding, 288 Output units, 280 Transcranial magnetic stimulation
(TMS), 291
Definitional approach to Parallel distributed processing
categorization, 266 (PDP), 280 Typicality, 268
Embodied approach, 290 Priming, 270 Typicality effect, 270
Error signal, 282 Prototype, 268 Unit (in a connectionist network), 280

COGLAB EXPERIMENTS Numbers in parentheses refer to the experiment number in CogLab.

Lexical Decision (41) Absolute Identification (44) Prototypes (46)

08271_ch09_ptg01.indd 295 4/19/18 7:24 PM

dugdax/Shutterstock.com

“Visual imagery” occurs when a person sees in his or her mind something that isn’t physically
present. This picture symbolizes the finding that although visual perception and visual imagery
share many properties, experiences associated with visual imagery can be less detailed and more
fragile than experiences associated with visual perception.

08271_ch10_ptg01.indd 296 4/18/18 5:09 PM

HOW TO GET ANYTHING YOU WANT MAKE A MAGICK MIRROR! - Mike Nach PDF
96% (28)
HOW TO GET ANYTHING YOU WANT MAKE A MAGICK MIRROR! - Mike Nach PDF
70 pages
100 Lectii
No ratings yet
100 Lectii
100 pages
General Motors Regular Production Options (Rpo) Codes
100% (1)
General Motors Regular Production Options (Rpo) Codes
58 pages
COG-PSY-LEC-06
No ratings yet
COG-PSY-LEC-06
12 pages
Cognitive Psychology
No ratings yet
Cognitive Psychology
4 pages
Sternberg R.J. and Ben-Zeev T. 2001. Complex Cognition The Psychology of Human Thought
No ratings yet
Sternberg R.J. and Ben-Zeev T. 2001. Complex Cognition The Psychology of Human Thought
27 pages
Conceptual Knowledge
No ratings yet
Conceptual Knowledge
16 pages
Categories and Concepts _ Noba
No ratings yet
Categories and Concepts _ Noba
35 pages
Psychological Science Modeling Scientific Literacy 2nd Edition Krause Solutions Manual Download
100% (15)
Psychological Science Modeling Scientific Literacy 2nd Edition Krause Solutions Manual Download
78 pages
[Ebooks PDF] download Early Category and Concept Development Making Sense of the Blooming Buzzing Confusion 1st Edition David H. Rakison full chapters
100% (2)
[Ebooks PDF] download Early Category and Concept Development Making Sense of the Blooming Buzzing Confusion 1st Edition David H. Rakison full chapters
75 pages
semantica
No ratings yet
semantica
7 pages
Early Category and Concept Development Making Sense of the Blooming Buzzing Confusion 1st Edition David H. Rakison - Quickly download the ebook to never miss any content
100% (1)
Early Category and Concept Development Making Sense of the Blooming Buzzing Confusion 1st Edition David H. Rakison - Quickly download the ebook to never miss any content
50 pages
Cognitive Chap 8
No ratings yet
Cognitive Chap 8
63 pages
Concepts and Categorisation-2
No ratings yet
Concepts and Categorisation-2
60 pages
CATEGORISATION
No ratings yet
CATEGORISATION
32 pages
handout_Chptr 9_PSYF211_Knowledge
No ratings yet
handout_Chptr 9_PSYF211_Knowledge
21 pages
Concepts
100% (1)
Concepts
32 pages
8785Early Category and Concept Development Making Sense of the Blooming Buzzing Confusion 1st Edition David H. Rakison - Read the ebook now or download it for a full experience
100% (1)
8785Early Category and Concept Development Making Sense of the Blooming Buzzing Confusion 1st Edition David H. Rakison - Read the ebook now or download it for a full experience
47 pages
5 Knowledge2
No ratings yet
5 Knowledge2
11 pages
Chapter 4 Representation, Knowledge in Long-Term Memory
No ratings yet
Chapter 4 Representation, Knowledge in Long-Term Memory
45 pages
Cogpsych Midterms Reviewer
No ratings yet
Cogpsych Midterms Reviewer
6 pages
COGNITION
No ratings yet
COGNITION
16 pages
Knowledge (1)
No ratings yet
Knowledge (1)
11 pages
Categories and Concepts
No ratings yet
Categories and Concepts
1 page
Group 4 Conceptual Knowledge
No ratings yet
Group 4 Conceptual Knowledge
22 pages
Topic 7 - Concept Learning and Semantic Knowledge
No ratings yet
Topic 7 - Concept Learning and Semantic Knowledge
21 pages
Thinking 1
No ratings yet
Thinking 1
33 pages
CP#5 Knowledge
No ratings yet
CP#5 Knowledge
13 pages
Lecture 3 PPT
No ratings yet
Lecture 3 PPT
29 pages
General Psychology II: Unit I
No ratings yet
General Psychology II: Unit I
30 pages
Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual all chapter instant download
100% (6)
Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual all chapter instant download
42 pages
Building Object Categories in Developmental Time 1st Edition Lisa Gershkoff-Stowe - Download the ebook today to explore every detail
100% (1)
Building Object Categories in Developmental Time 1st Edition Lisa Gershkoff-Stowe - Download the ebook today to explore every detail
58 pages
Cognition
No ratings yet
Cognition
5 pages
Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manualpdf download
100% (3)
Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manualpdf download
53 pages
Free Access to Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual Chapter Answers
100% (15)
Free Access to Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual Chapter Answers
35 pages
D5S1
No ratings yet
D5S1
119 pages
Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual download
100% (2)
Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual download
48 pages
Get (Ebook) Early Category and Concept Development: Making Sense of the Blooming, Buzzing Confusion by David H. Rakison, Lisa M. Oakes ISBN 0195142934 free all chapters
No ratings yet
Get (Ebook) Early Category and Concept Development: Making Sense of the Blooming, Buzzing Confusion by David H. Rakison, Lisa M. Oakes ISBN 0195142934 free all chapters
82 pages
Categorization, Concept Learning, and Behavior Analysis: An Introduction
No ratings yet
Categorization, Concept Learning, and Behavior Analysis: An Introduction
12 pages
Conceptual Understanding and Achievements: Section Four
No ratings yet
Conceptual Understanding and Achievements: Section Four
49 pages
1. Notes - Thinking Part 1 - Tools
No ratings yet
1. Notes - Thinking Part 1 - Tools
13 pages
Cognitive Processing
No ratings yet
Cognitive Processing
8 pages
TimeAndSpace - Cognitive Science
No ratings yet
TimeAndSpace - Cognitive Science
80 pages
Concepts and Constructs
No ratings yet
Concepts and Constructs
8 pages
Categorizing Cognition Toward Conceptual Coherence in the Foundations of Psychology 1st Edition Graeme S. Halfordinstant download
100% (1)
Categorizing Cognition Toward Conceptual Coherence in the Foundations of Psychology 1st Edition Graeme S. Halfordinstant download
51 pages
Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual instant download
100% (3)
Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual instant download
67 pages
Chapter 9 Power Points
No ratings yet
Chapter 9 Power Points
29 pages
Big Book of Concepts
No ratings yet
Big Book of Concepts
9 pages
Categorizing Cognition Toward Conceptual Coherence in the Foundations of Psychology 1st Edition Graeme S. Halford - The 2025 ebook edition is available with updated content
No ratings yet
Categorizing Cognition Toward Conceptual Coherence in the Foundations of Psychology 1st Edition Graeme S. Halford - The 2025 ebook edition is available with updated content
83 pages
The Prefrontal Cortex: Categories, Concepts and Cognition: Earl K. Miller, David J. Freedman and Jonathan D. Wallis
No ratings yet
The Prefrontal Cortex: Categories, Concepts and Cognition: Earl K. Miller, David J. Freedman and Jonathan D. Wallis
14 pages
Access Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual All Chapters Immediate PDF Download
100% (9)
Access Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual All Chapters Immediate PDF Download
57 pages
ROSCH, Principles of Categorization
100% (2)
ROSCH, Principles of Categorization
22 pages
16254
No ratings yet
16254
56 pages
[Ebooks PDF] download Building Object Categories in Developmental Time 1st Edition Lisa Gershkoff-Stowe full chapters
100% (2)
[Ebooks PDF] download Building Object Categories in Developmental Time 1st Edition Lisa Gershkoff-Stowe full chapters
87 pages
Concept Formation and Categorization
No ratings yet
Concept Formation and Categorization
14 pages
Categorization and Concepts Presentation
No ratings yet
Categorization and Concepts Presentation
27 pages
Week 07_Concept Formation
No ratings yet
Week 07_Concept Formation
31 pages
Access Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual All Chapters Immediate PDF Download
100% (15)
Access Introduction to Psychological Science Canadian 1st Edition Krause Solutions Manual All Chapters Immediate PDF Download
66 pages
N. The Mental Representation of A Thing Or: Concept
No ratings yet
N. The Mental Representation of A Thing Or: Concept
2 pages
A Portfolio of Reflections: Reflection Sheets for Curriculum Areas
From Everand
A Portfolio of Reflections: Reflection Sheets for Curriculum Areas
Kelley Dos Santos Kremer
5/5 (1)
Brain Framing: Instructional Planning with the Brain in Mind
From Everand
Brain Framing: Instructional Planning with the Brain in Mind
Dr. Linda Karges-Bone
No ratings yet
Convey with Confidence: Powerful Techniques for Clear Communication
From Everand
Convey with Confidence: Powerful Techniques for Clear Communication
Sachin Naha
No ratings yet
Schaum's Quick Guide to Writing Great Research Papers
From Everand
Schaum's Quick Guide to Writing Great Research Papers
Laurie Rozakis
5/5 (4)
Derwent Brochure - 1-11final Final
No ratings yet
Derwent Brochure - 1-11final Final
25 pages
Learning and Coloring Volume1
No ratings yet
Learning and Coloring Volume1
26 pages
Ghent PDF Output Test V50 ALL X4
No ratings yet
Ghent PDF Output Test V50 ALL X4
6 pages
Grade: Technical Drawing 7
100% (1)
Grade: Technical Drawing 7
55 pages
Pygame
No ratings yet
Pygame
120 pages
Seeing The Aura - BSS.3
No ratings yet
Seeing The Aura - BSS.3
27 pages
All in One SCIENCE 10 Activity CH 10
No ratings yet
All in One SCIENCE 10 Activity CH 10
2 pages
LESSON 1 FINAL EXAM (Writing Test) Name
No ratings yet
LESSON 1 FINAL EXAM (Writing Test) Name
3 pages
RGB 10
No ratings yet
RGB 10
8 pages
Product Catalogue: Plastics
No ratings yet
Product Catalogue: Plastics
7 pages
Coco Jaunt 1728: 4th Main Road, Anna Nagar East, Chennai
No ratings yet
Coco Jaunt 1728: 4th Main Road, Anna Nagar East, Chennai
16 pages
Answer 2 Unit13 03 01
No ratings yet
Answer 2 Unit13 03 01
3 pages
Light Vein Viewer PDF
No ratings yet
Light Vein Viewer PDF
9 pages
Corduroy's Colors
100% (2)
Corduroy's Colors
7 pages
Module 1
No ratings yet
Module 1
11 pages
Name That Tone
No ratings yet
Name That Tone
12 pages
Color Grading
No ratings yet
Color Grading
15 pages
Return of The Incredible Machine Contraptions
No ratings yet
Return of The Incredible Machine Contraptions
112 pages
Chapter-Vii Rabindranath Tagore An-D The Aesthetic Imagination
No ratings yet
Chapter-Vii Rabindranath Tagore An-D The Aesthetic Imagination
20 pages
SurTec Technical Letter 4 - Service Life Extension of Trivalent Chromium Based Passivations For Zinc
No ratings yet
SurTec Technical Letter 4 - Service Life Extension of Trivalent Chromium Based Passivations For Zinc
7 pages
Color of Minerals
No ratings yet
Color of Minerals
3 pages
Jim Dine 1st Edtpa Format
No ratings yet
Jim Dine 1st Edtpa Format
19 pages
Emegeo 10
100% (1)
Emegeo 10
6 pages
Pottery Poetry Catalogue 2023 24
No ratings yet
Pottery Poetry Catalogue 2023 24
38 pages
Elements of Urban Design Image of The City
100% (1)
Elements of Urban Design Image of The City
27 pages
Colour Matching Triangle
No ratings yet
Colour Matching Triangle
3 pages
MODULE AR 182 Architecturalinteriors
No ratings yet
MODULE AR 182 Architecturalinteriors
57 pages