PSYC 575 Cognitive Psychology

Overview

You will write a research paper that discusses and evaluates the current research in the field of false memories. As a formal research paper, it must be completely focused on the empirical evidence pertaining to the topic. Refrain from discussing your personal opinions or experiences. You must use scholarly sources for your references; do not use a textbook, website, or popular press as a source.

Instructions

The paper must include the following:

· 12–14 pages of content (not including title page, abstract, and reference page)

· Title, abstract and reference page

· Current APA formatting throughout

· At least 12 peer-reviewed journal articles

· A discussion of the important limitations of the evidence (studies) presented

· A discussion of conflicting evidence for any of the studies discussed

The organization of this paper must be as follows:

Title Page 1 page

Abstract 120–150 words

Introduction 1–2 paragraphs

Thesis statement: one sentence that states the focus of your paper

1. Key Points (include 3–5 main points)

a. Point 1

b. Point 2

c. Point 3

d. Point 4

e. Point 5

Body of Paper 8–10 pages

Topic sentence and supporting research for each of your key points. Use the same order as in the introduction.

1. Research Concept/Finding

a. Supporting evidence

b. Connect to next concept/finding

2. Research Concept/Finding

a. Supporting evidence
b. Connect to next concept/finding

3. Research Concept/Finding

a. Supporting evidence
b. Connect to next concept/finding

Summary 1–2 pages

Write a summary sentence that wraps up the concepts discussed in the paper.

1. Summary sentence must be followed by clear statements that summarize each of the main concepts/findings discussed in the body.

a. Summary of research Concept/Finding 1

b. Summary of research Concept/Finding 2

c. Summary of research Concept/Finding 3

d. Summary of research Concept/Finding 4

e. Summary of research Concept/Finding 5

Conclusion 1 paragraph

Final thoughts in a paragraph

References At least 12 peer-reviewed journal articles

Note: Your assignment will be checked for originality via the Turnitin plagiarism tool.

Page 1 of 2

Cognitive Psychology
and Its Implications
Eighth Edition

ANDERSON8e-FM.indd 1 13/09/14 10:04 AM

This page intentionally left blank

John R. Anderson
Carnegie Mellon University

w o r t h P u b l i s h e r s

A Macmillan Education Company

Cognitive Psychology
and Its Implications

Eighth Edition

ANDERSON8e-FM.indd 3 13/09/14 10:04 AM

u To Gordon Bower

Vice President, Editing, Design, and Media Production: Catherine Woods
Publisher: Rachel Losh
Associate Publisher: Jessica Bayne
Senior Acquisitions Editor: Christine Cardone
Marketing Manager: Lindsay Johnson
Marketing Assistant: Tess Sanders
Development Editor: Len Neufeld
Associate Media Editor: Anthony Casciano
Assistant Editor: Catherine Michaelsen
Director of Editing, Design, and Media Production: Tracey Kuehn
Managing Editor: Lisa Kinne
Project Editor: Kerry O’Shaughnessy
Art Director: Diana Blume
Cover Designer: Vicki Tomaselli
Text Designer: Dreamit Inc.
Illustration Coordinator: Janice Donnola
Illustrations: Dragonfly Media Group
Photo Editor: Bianca Moscatelli
Production Manager: Sarah Segal
Composition: MPS Ltd.
Printing and Binding: RR Donnelley and Sons
Cover Painting: Mario Colonel/Aurora/Getty Images

Library of Congress Control Number: 2014938514

ISBN-13: 978-1-4641-4891-0
ISBN-10: 1-4641-4891-0

© 2015, 2010, 2005, 2000 by Worth Publishers
All rights reserved.
Printed in the United States of America

First Printing

Worth Publishers
41 Madison Avenue
New York, NY 10010
www.worthpublishers.com

ANDERSON8e-FM.indd 4 13/09/14 10:04 AM

http://www.worthpublishers.com

John Robert Anderson is Richard King Mellon Professor of Psychology
and Computer Science at Carnegie Mellon University. He is known for developing
ACT-R, which is the most widely used cognitive architecture in cognitive science.
Anderson was also an early leader in research on intelligent tutoring systems,
and computer systems based on his cognitive tutors currently teach mathematics
to about 500,000 children in American schools. He has served as President of
the Cognitive Science Society, and has been elected to the American Academy
of Arts and Sciences, the National Academy of Sciences, and the American
Philosophical Society. He has received numerous scientific awards including the
American Psychological Association’s Distinguished Scientific Career Award, the
David E. Rumelhart Prize for Contributions to the Formal Analysis of Human
Cognition, and the inaugural Dr. A. H. Heineken Prize for Cognitive Science. He
is completing his term as editor of the prestigious Psychological Review.

v

About the Author

ANDERSON8e-FM.indd 5 13/09/14 10:04 AM

This page intentionally left blank

Preface xvii

Chapter 1
The Science of Cognition 1

Chapter 2
Perception 27

Chapter 3
Attention and Performance 53

Chapter 4
Mental Imagery 78

Chapter 5
Representation of Knowledge 97

Chapter 6
Human Memory: Encoding and Storage 124

Chapter 7
Human Memory: Retention and Retrieval 150

Chapter 8
Problem Solving 181

Chapter 9
Expertise 210

Chapter 10
Reasoning 237

Chapter 11
Decision Making 260

Chapter 12
Language Structure 281

Chapter 13
Language Comprehension 313

Chapter 14
Individual Differences in Cognition 338

Glossary 365
References 373
Name Index 393
Subject Index 399

vii

Brief Contents

ANDERSON8e-FM.indd 7 13/09/14 10:04 AM

This page intentionally left blank

Preface xvii

Chapter 1
The Science of Cognition 1

u Motivations for Studying Cognitive Psychology / 1
Intellectual Curiosity / 1
Implications for Other Fields / 2
Practical Applications / 3

u The History of Cognitive Psychology / 3
Early History / 4
Psychology in Germany: Focus on Introspective Observation / 4

Implications: What does cognitive psychology tell us about how to
study effectively? / 5

Psychology in America: Focus on Behavior / 6
The Cognitive Revolution: AI, Information Theory,
and Linguistics / 7
Information-Processing Analyses / 9
Cognitive Neuroscience / 10

u Information Processing: The Communicative Neurons / 10
The Neuron / 11
Neural Representation of Information / 13

u Organization of the Brain / 15
Localization of Function / 17
Topographic Organization / 18

u Methods in Cognitive Neuroscience / 19
Neural Imaging Techniques / 20
Using fMRI to Study Equation Solving / 22

Chapter 2
Perception 27

u Visual Perception in the Brain / 27
Early Visual Information Processing / 28
Information Coding in Visual Cells / 31
Depth and Surface Perception / 33
Object Perception / 34

ix

Contents

ANDERSON8e-FM.indd 9 13/09/14 10:04 AM

x / c o n t e n t s

u Visual Pattern Recognition / 35
Template-Matching Models / 36

Implications: Separating humans from BOTs / 37
Feature Analysis / 37
Object Recognition / 39
Face Recognition / 42

u Speech Recognition / 43
Feature Analysis of Speech / 44

u Categorical Perception / 45

u Context and Pattern Recognition / 47
Massaro’s FLMP Model for Combination of Context and Feature
Information / 48
Other Examples of Context and Recognition / 49

u Conclusions / 51

Chapter 3
Attention and Performance 53

u Serial Bottlenecks / 53

u Auditory Attention / 54
The Filter Theory / 55
The Attenuation Theory and the Late-Selection Theory / 56

u Visual Attention / 58
The Neural Basis of Visual Attention / 60
Visual Search / 61
The Binding Problem / 62
Neglect of the Visual Field / 65
Object-Based Attention / 67

u Central Attention: Selecting Lines of Thought to Pursue / 69
Implications: Why is cell phone use and driving a dangerous
combination? / 72

Automaticity: Expertise Through Practice / 72
The Stroop Effect / 73
Prefrontal Sites of Executive Control / 75

u Conclusions / 76

Chapter 4
Mental Imagery 78

u Verbal Imagery Versus Visual Imagery / 79
Implications: Using brain activation to read people’s minds / 81

u Visual Imagery / 82
Image Scanning / 84
Visual Comparison of Magnitudes / 85
Are Visual Images Like Visual Perception? / 86
Visual Imagery and Brain Areas / 87
Imagery Involves Both Spatial and Visual Components / 88
Cognitive Maps / 89
Egocentric and Allocentric Representations of Space / 91
Map Distortions / 94

u Conclusions: Visual Perception and Visual Imagery / 95

ANDERSON8e-FM.indd 10 13/09/14 10:04 AM

c o n t e n t s / xi

Chapter 5
Representation of Knowledge 97

u Knowledge and Regions of the Brain / 97

u Memory for Meaningful Interpretations of Events / 98
Memory for Verbal Information / 98
Memory for Visual Information / 99
Importance of Meaning to Memory / 101
Implications of Good Memory for Meaning / 103

Implications: Mnemonic techniques for remembering vocabulary
items / 104

u Propositional Representations / 104
Amodal Versus Perceptual Symbol Systems / 106

u Embodied Cognition / 108

u Conceptual Knowledge / 109
Semantic Networks / 110
Schemas / 112
Abstraction Theories Versus Exemplar Theories / 118
Natural Categories and Their Brain Representations / 120

u Conclusions / 122

Chapter 6
Human Memory: Encoding and Storage 124

u Memory and the Brain / 124

u Sensory Memory Holds Information Briefly / 125
Visual Sensory Memory / 125
Auditory Sensory Memory / 126
A Theory of Short-Term Memory / 127

u Working Memory Holds the Information Needed to Perform a
Task / 129
Baddeley’s Theory of Working Memory / 129
The Frontal Cortex and Primate Working Memory / 131

u Activation and Long-Term Memory / 133
An Example of Activation Calculations / 133
Spreading Activation / 135

u Practice and Memory Strength / 137
The Power Law of Learning / 137
Neural Correlates of the Power Law / 139

u Factors Influencing Memory / 141
Elaborative Processing / 141
Techniques for Studying Textual Material / 142
Incidental Versus Intentional Learning / 144

Implications: How does the method of loci help us organize
recall? / 145

Flashbulb Memories / 145

u Conclusions / 148

ANDERSON8e-FM.indd 11 13/09/14 10:04 AM

xii / c o n t e n t s

Chapter 7
Human Memory: Retention and Retrieval 150

u Are Memories Really Forgotten? / 150

u The Retention Function / 152

u How Interference Affects Memory / 154
The Fan Effect: Networks of Associations / 155
The Interfering Effect of Preexisting Memories / 157
The Controversy Over Interference and Decay / 158
An Inhibitory Explanation of Forgetting? / 159
Redundancy Protects Against Interference / 160

u Retrieval and Inference / 161
Plausible Retrieval / 162
The Interaction of Elaboration and Inferential Reconstruction / 164
Eyewitness Testimony and the False-Memory Controversy / 165

Implications: How have advertisers used knowledge of cognitive
psychology? / 166

False Memories and the Brain / 167

u Associative Structure and Retrieval / 169
The Effects of Encoding Context / 169
The Encoding-Specificity Principle / 172

u The Hippocampal Formation and Amnesia / 172

u Implicit Versus Explicit Memory / 174
Implicit Versus Explicit Memory in Normal Participants / 175
Procedural Memory / 177

u Conclusions: The Many Varieties of Memory in the Brain / 179

Chapter 8
Problem Solving 181

u The Nature of Problem Solving / 181
A Comparative Perspective on Problem Solving / 181
The Problem-Solving Process: Problem Space
and Search / 183

u Problem-Solving Operators / 186
Acquisition of Operators / 186
Analogy and Imitation / 188
Analogy and Imitation from an Evolutionary and Brain
Perspective / 190

u Operator Selection / 191
The Difference-Reduction Method / 192
Means-Ends Analysis / 194
The Tower of Hanoi Problem / 196
Goal Structures and the Prefrontal Cortex / 198

u Problem Representation / 199
The Importance of the Correct Representation / 199
Functional Fixedness / 201

u Set Effects / 202
Incubation Effects / 204
Insight / 206

ANDERSON8e-FM.indd 12 13/09/14 10:04 AM

c o n t e n t s / xiii

u Conclusions / 207

u Appendix: Solutions / 208

Chapter 9
Expertise 210

u Brain Changes with Skill Acquisition / 211

u General Characteristics of Skill Acquisition / 211
Three Stages of Skill Acquisition / 211
Power-Law Learning / 212

u The Nature of Expertise / 215
Proceduralization / 215
Tactical Learning / 217
Strategic Learning / 218
Problem Perception / 221
Pattern Learning and Memory / 223

Implications: Computers achieve chess expertise differently than
humans / 226

Long-Term Memory and Expertise / 226
The Role of Deliberate Practice / 227

u Transfer of Skill / 229

u Theory of Identical Elements / 231

u Educational Implications / 232
Intelligent Tutoring Systems / 233

u Conclusions / 235

Chapter 10
Reasoning 237

u Reasoning and the Brain / 238

u Reasoning About Conditionals / 239
Evaluation of Conditional Arguments / 240
Evaluating Conditional Arguments in a Larger Context / 241
The Wason Selection Task / 242
Permission Interpretation of the Conditional / 243
Probabilistic Interpretation of the Conditional / 244
Final Thoughts on the Connective If / 246

u Deductive Reasoning: Reasoning About Quantifiers / 246
The Categorical Syllogism / 246
The Atmosphere Hypothesis / 248
Limitations of the Atmosphere Hypothesis / 249
Process Explanations / 250

u Inductive Reasoning and Hypothesis Testing / 251
Hypothesis Formation / 252
Hypothesis Testing / 253
Scientific Discovery / 255

Implications: How convincing is a 90% result? / 256

u Dual-Process Theories / 257

u Conclusions / 258

ANDERSON8e-FM.indd 13 13/09/14 10:04 AM

xiv / c o n t e n t s

Chapter 11
Decision Making 260

u The Brain and Decision Making / 260

u Probabilistic Judgment / 262
Bayes’s Theorem / 262
Base-Rate Neglect / 264
Conservatism / 265
Correspondence to Bayes’s Theorem with Experience / 266
Judgments of Probability / 268
The Adaptive Nature of the Recognition Heuristic / 270

u Making Decisions Under Uncertainty / 271
Framing Effects / 273

Implications: Why are adolescents more likely to make bad
decisions? / 276

Neural Representation of Subjective Utility and Probability / 277

u Conclusions / 279

Chapter 12
Language Structure 281

u Language and the Brain / 281

u The Field of Linguistics / 283
Productivity and Regularity / 283
Linguistic Intuitions / 284
Competence Versus Performance / 285

u Syntactic Formalisms / 286
Phrase Structure / 286
Pause Structure in Speech / 287
Speech Errors / 288
Transformations / 290

u What Is So Special About Human Language? / 291
Implications: Ape language and the ethics
of experimentation / 293

u The Relation Between Language and Thought / 294
The Behaviorist Proposal / 294
The Whorfian Hypothesis of Linguistic Determinism / 295
Does Language Depend on Thought? / 297
The Modularity of Language / 299

u Language Acquisition / 300
The Issue of Rules and the Case of Past Tense / 303
The Quality of Input / 305
A Critical Period for Language Acquisition / 306
Language Universals / 308

ANDERSON8e-FM.indd 14 13/09/14 10:04 AM

c o n t e n t s / xv

The Constraints on Transformations / 310
Parameter Setting / 310

u Conclusions: The Uniqueness of Language: A Summary / 311

Chapter 13
Language Comprehension 313

u Brain and Language Comprehension / 314

u Parsing / 314
Constituent Structure / 314
Immediacy of Interpretation / 317
The Processing of Syntactic Structure / 318
Semantic Considerations / 320
The Integration of Syntax and Semantics / 321
Neural Indicants of Syntactic and Semantic Processing / 322
Ambiguity / 323
Neural Indicants of the Processing of Transient Ambiguity / 324
Lexical Ambiguity / 326
Modularity Compared with Interactive Processing / 326

Implications: Intelligent chatterboxes / 328

u Utilization / 329
Bridging Versus Elaborative Inferences / 329
Inference of Reference / 330
Pronominal Reference / 331
Negatives / 333

u Text Processing / 334

u Situation Models / 335

u Conclusions / 336

Chapter 14 Differences in C
Individual Differences in Cognition 338

u Cognitive Development / 338
Piaget’s Stages of Development / 340
Conservation / 341
What Develops? / 343
The Empiricist-Nativist Debate / 345
Increased Mental Capacity / 347
Increased Knowledge / 349
Cognition and Aging / 350
Summary for Cognitive Development / 353

u Psychometric Studies of Cognition / 353
Intelligence Tests / 353
Factor Analysis / 355

ANDERSON8e-FM.indd 15 13/09/14 10:04 AM

Implications: Does IQ determine success in life? / 356
Reasoning Ability / 358
Verbal Ability / 360
Spatial Ability / 361
Conclusions from Psychometric Studies / 362

u Conclusions / 363

3
Glossary 365
References 373
Name Index 393
Subject Index 399

xvi / c o n t e n t s

ANDERSON8e-FM.indd 16 13/09/14 10:04 AM

xvii

Preface

this is the eighth edition of my textbook—a new edition has appeared every
5 years. The first edition was written more than half of my life ago. In

writing this preface I thought I would take the opportunity to reflect on where
the field has been, where it is, where it is going, and how this is reflected in
the book. One piece of evidence to inform this reflection is the chart showing
number of citations to publication in each of the last 100 years. I have not felt
the need to throw out references to classic studies that still serve their purpose,
and so this provides one measure of how research over the years serves to
shape my conception of the field—a conception that I think is shared by many
researchers. There are a couple of fairly transparent historical discontinuities in
that graph and a couple of not so apparent changes:

● There are very few citations to papers before the end of World War II, and
then there is a rapid rise in citations. Essentially, the Greatest Generation
came back from the war, broke the behaviorist grip on psychology, and
started the cognitive revolution. The growing number of citations reflects
the rise of a new way of studying and understanding the human mind.

● The number of citations basically asymptotes about the time of the publi-
cation of the first edition of this textbook in 1980. Being a baby boomer,
when I came into the field, I was able to start with the framework that the
pioneers had established and organize it into a coherent structure that
appeared in the first edition.

● The relatively stable level of citations since 1980 hides a major development
in the field that began to really establish itself in the 1990s. Early research
had focused on behavioral measures because it seemed impossible to ethi-
cally study what was in the human brain. However, new techniques in neu-
ral imaging arose that allowed us to complement that research with neural
measures. This is complemented by research on animals, particularly
primates.

● There is a dip over the last 5 years. This reflects the need to properly digest
the significance of the most current research. I could be wrong, but I think
we are on the verge of significant change brought about by our ability to
mine large data sets. We are now able to detect significant patterns in the
huge amounts of data we can collect about people, both in terms of the
activity of their brains and their activities in the world. Some of this comes
out in the textbook’s discussion of the most recent research.

Each instructor will use a textbook in his or her own way, but when I teach
from this book, I impose the following structure on it:

● The introductory chapter provides a preparation for understanding what
is in the subsequent chapters, and the last chapter provides a reflection on
how all the pieces fit together in human cognition and intelligence.

ANDERSON8e-FM.indd 17 13/09/14 10:04 AM

xviii / P r e fa c e

● The meat of the textbook is the middle 12 chapters, and they naturally
organize themselves into 6 thematic pairs on perception and attention,
knowledge representation, memory, problem solving, reasoning and deci-
sion making, and language.

● There is a major break between the first three pairs and the last three pairs.
As I tell my class at that point: “Most of what we have discussed up to this
point is true of all primates. Most of what we are going to talk about is only
true of humans.”

u New in the Eighth Edition

This new edition discusses current and exciting themes in cognitive psychology.
One of these themes is the increasing cognitive capacity of modern tech-

nology. Chapter 1 opens with discussion of Watson’s performance on Jeopardy,
Apple’s Siri, and Ray Kurzwell’s prophesy of the impending Singularity. Chapter
2 discusses new technological developments in character and face recognition.
Chapter 4 describes new “mind-reading” research that uses fMRI to reconstruct
the thoughts and images of people.

A complementary theme explores the bounds on human intellectual capacity.
Chapter 5 describes new research on people with near-perfect autobiographical
memory, as well as everyone’s high capacity to remember images. Chapter 6
examines new research on the special benefits of self-testing, and new research
on flashbulb memories for 9/11. Chapter 8 describes new research on the role of
worked examples in acquiring problem-solving operators. Chapter 9 examines new
research on the general cognitive benefits of working-memory practice and video-
game playing, as well as the controversy surrounding these results. The final chapter
explores new theories of the interaction between genetic factors and environmental
factors in shaping intelligence.

A third theme is the increasing ability of neuroscience to penetrate the
mind. Chapter 3 describes research relating visual neglect to deficits in concep-
tual judgments about number order and alphabetical order. Chapter 5 discusses
the new work in neurosemantics. Chapter 6 describes new meta-analyses on the
regions of the brain that support working memory. Chapter 11 describes the
evidence connecting the response of the dopamine neurons to theories of rein-
forcement learning. Chapter 14 describes the research showing that single neu-
rons are tuned to recognize specific numbers of objects.

Then there are introductions to some of the new theoretical frameworks
that are shaping modern research. Chapter 7 describes the current state of
research on retrieval-induced interference. Chapter 10 describes dual-process
theories of reasoning. Bayesian analyses are playing an increasing role in our

35

30

25

20

15

10

5
Nu

m
be

r o
f c

ita
tio

ns

0

19
10

19
20

19
30

19
40

19
50

19
60

19
70

19
80

19
90

20
00

20
10

ANDERSON8e-FM.indd 18 13/09/14 10:04 AM

P r e fa c e / xix

field, and Chapter 12 describes one example of how the world’s kinship terms
are optimally chosen for communicative purposes. Chapter 13 describes the
role of situation models in text comprehension.

u New Teaching and Learning Resources

Our newest set of online materials, LaunchPad Solo, provides tools and
topically relevant content that you need to teach your class. LaunchPad Solo for
Cognitive Psychology includes 45 experiments that helped establish the core of
our understanding of cognitive functions. Taking the role of experimenter, you
will work in a first-of-its-kind interactive environment that lets you manipulate
variables, collect data, and analyze results.

Instructor resources include an Instructor’s Manual, computerized test bank,
and Illustration and Lecture slides.

Acknowledgments
There are three individuals who have really helped me in the writing of this
edition. In addition to all of her other responsibilities, my Senior Acquisitions
Editor Christine Cardone has provided a great set of reviews that helped me
appreciate both how others see the directions of the field and how others teach
from this text. The Development Editor, Len Neufeld, did a terrific job fact-
checking every bit of the book and providing it with a long overdue line-by-line
polishing. Finally, my son, Abraham Anderson, went through all of the text,
holding back no punches about how it registers with his generation.

In addition to Chris Cardone and Len Neufeld, I also acknowledge the
assistance of the following people from Worth: Kerry O’Shaughnessy, Project
Editor; Catherine Michaelsen, Assistant Editor; Sarah Segal, Production Manager;
Janice Donnola, Illustration Coordinator; Bianca Moscatelli, Photo Editor; Tracey
Kuehn, Director of Editing, Design, and Media Production; Anthony Casciano,
Associate Media Editor; Diane Blume, Art Director; and Vicki Tomaselli and
Dreamit Inc., who designed the cover and the interior, respectively.

I am grateful for the many comments and suggestions of the reviewers
of this eighth edition: Erik Altman, Michigan State University; Walter Beagley,
Alma College; Kyle Cave, University of Massachusetts; Chung-Yiu Peter Chiu,
University of Cincinnati; Michael Dodd, University of Nebraska, Lincoln; Jonathan
Evans, University of Plymouth; Evan Heit, University of California, Merced; Arturo
Hernandez, University of Houston; Daniel Jacobson, Eastern Michigan University;
Mike Oaksford, Birkbeck College, University of London; Thomas Palmeri, Vanderbilt
University; Jacqueline Park, Vanguard University; David Neil Rapp, Northwestern
University; Christian Schunn, University of Pittsburgh; Scott Slotnick, Boston College;
Niels Taatgen, University of Groningen; Peter Vishton, College of William & Mary;
and Xiaowei Zhao, Emmanuel College.

I would also like to thank the people who read the first seven editions of
my book, because much of their earlier influence remains: Chris Allan, Nancy
Alvarado, Jim Anderson, James Beale, Irv Biederman, Liz Bjork, Stephen
Blessing, Lyle Bourne, John Bransford, Bruce Britton, Tracy Brown, Gregory
Burton, Robert Calfee, Pat Carpenter, Bill Chase, Nick Chater, Micki Chi, Bill
Clancy, Chuck Clifton, Lynne Cooper, Gus Craik, Bob Crowder, Ann Devlin,
Mike Dodd, Thomas Donnelly, David Elmes, K. Anders Ericsson, Martha
Farah, Ronald Finke, Ira Fischler, Susan Fiske, Michael Gazzaniga, Ellen Gagné,
Rochel Gelman, Barbara Greene, Alyse Hachey, Dorothea Halpert, Lynn
Hasher, Geoff Hinton, Kathy Hirsh-Pasek, Buz Hunt, Louna Hernandez-Jarvis,

Cognitive Psychology

ANDERSON8e-FM.indd 19 13/09/14 10:04 AM

xx / P r e fa c e

Robert Hines, Robert Hoffman, Martha Hubertz, Lumei Hui, Laree Huntsman,
Lynn Hyah, Earl Hunt, Andrew Johnson, Philip Johnson-Laird, Marcel Just,
Stephen Keele, Walter Kintsch, Dave Klahr, Steve Kosslyn, Al Lesgold, Clayton
Lewis, Beth Loftus, Marsha Lovett, Maryellen MacDonald, Michael McGuire,
Brian MacWhinney, Dominic Massaro, Jay McClelland, Karen J. Mitchell, John
D. Murray, Al Newell, E. Slater Newman, Don Norman, Gary Olson, Allan
Paivio, Thomas Palmieri, Nancy Pennington, Jane Perlmutter, Peter Polson,
Jim Pomerantz, Mike Posner, Roger Ratcliff, Lynne Reder, Steve Reed,
Russ Revlin, Phillip Rice, Lance Rips, Roddy Roediger, Daniel Schacter, Jay
Schumacher, Miriam Schustack, Terry Sejnowski, Bob Siegler, Murray Singer,
Ed Smith, Kathy Spoehr, Bob Sternberg, Roman Taraban, Charles Tatum,
Joseph Thompson, Dave Tieman, Tom Trabasso, Henry Wall, Charles A.
Weaver, Patricia de Winstanley, Larry Wood, and Maria Zaragoza.

ANDERSON8e-FM.indd 20 13/09/14 10:04 AM

1

1
The Science of Cognition

Our species is called Homo sapiens, or “human, the wise,” reflecting the general
belief that our superior thought processes are what distinguish us from other

animals. Today we all know that the brain is the organ of the human mind, but the
connection between the brain and the mind was not always known. For instance, in
a colossal misassociation, the Greek philosopher Aristotle localized the mind in the
heart. He thought the function of the brain was to cool the blood. Cognitive psy-
chology is the science of how the mind is organized to produce intelligent thought
and how the mind is realized in the brain.

This chapter introduces fundamental concepts that set the stage for the rest of
the book by addressing the following questions:

● Why do people study cognitive psychology?
● Where and when did cognitive psychology originate?
● How is the mind realized in the body?

How do the cells in the brain process information?
What parts of the brain are responsible for different functions?
What are the methods for studying the brain?

◆ Motivations for Studying Cognitive
Psychology

Intellectual Curiosity
As with any scientific inquiry, the thirst for knowledge provides much of
the impetus to study cognitive psychology. In this respect, the cognitive
psychologist is like the tinkerer who wants to know how a clock works. The hu-
man mind is particularly fascinating: It displays a remarkable intelligence and
ability to adapt. Yet we are often unaware of the extraordinary aspects of human
cognition. Just as when watching a live television broadcast of a distant news
event we rarely consider the sophisticated technologies that make the broad-
cast possible, we also rarely think about the sophisticated mental processes that
enable us to understand that news event. Cognitive psychologists strive to un-
derstand the mechanisms that make such intellectual sophistication possible.

The inner workings of the human mind are far more intricate than the
most complicated systems of modern technology. For over half a century,
researchers in the field of artificial intelligence (AI) have been attempting to
develop programs that will enable computers to display intelligent behavior.
There have been some notable successes, such as IBM’s Watson that won over

Anderson_8e_Ch01.indd 1 13/09/14 9:32 AM

2 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

human contestants on Jeopardy and the iPhone personal assistant Siri. Still, AI
researchers realize they are a long way from creating a program that matches
humans in generalized intelligence, with human flexibility in recalling facts,
solving problems, reasoning, learning, and processing language. This failure of
AI to achieve human-level intelligence has become the cause of a great deal of
soul-searching by some of the founders of AI (e.g., McCarthy, 1996; Nilsson,
2005). There is a resurging view that AI needs to pay more attention to how
human thought functions.

There does not appear to be anything magical about human intelligence
that would make it impossible to model in a computer. Scientific discovery,
for instance, is often thought of as the ultimate accomplishment of human
intelligence: Scientists supposedly make great leaps of intuition to explain
a puzzling set of data. Formulating a novel scientific theory is supposed
to require both great creativity and special deductive powers. But is this
actually the case? Herbert Simon, who won the 1978 Nobel Prize for his
theoretical work in economics, spent the last 40 years of his life studying
cognitive psychology. Among other things, he focused on the intellectual
accomplishments involved in “doing” science. He and his colleagues (Langley,
Simon, Bradshaw, & Zytkow, 1987) built computer programs to simulate
the problem-solving activities involved in such scientific feats as Kepler’s
discovery of the laws of planetary motion and Ohm’s development of his law
for electric circuits. Simon also examined the processes involved in his own
now-famous scientific discoveries (Simon, 1989). In all cases, he found that
the methods of scientific discovery could be explained in terms of the basic
cognitive processes that we study in cognitive psychology. He wrote that
many of these activities are just well-understood problem-solving processes
(e.g., as covered in Chapters 8 and 9). He says:

Moreover, the insight that is supposed to be required for such work as
discovery turns out to be synonymous with the familiar process of rec-
ognition; and other terms commonly used in the discussion of creative
work—such terms as “judgment,” “creativity,” or even “genius”—appear
to be wholly dispensable or to be definable, as insight is, in terms of
mundane and well-understood concepts. (Simon, 1989, p. 376)

In other words, a detailed look reveals that even the brilliant results of human
genius are produced by basic cognitive processes operating together in complex
ways to produce those brilliant results.1 Most of this book will be devoted to de-
scribing what we know about these basic processes.

■ Great feats of intelligence, such as scientific discovery, are the result
of basic cognitive processes.

Implications for Other Fields
Students and researchers interested in other areas of psychology or social
science have another reason for following developments in cognitive psy-
chology. The basic mechanisms governing human thought are important in
understanding the types of behavior studied by other social sciences. For exam-
ple, an appreciation of how humans think is important to understanding why
certain thought malfunctions occur (clinical psychology), how people behave
with other individuals or in groups (social psychology), how persuasion works
(political science), how economic decisions are made (economics), why certain

1 Weisberg (1986) comes to a similar conclusion.

Anderson_8e_Ch01.indd 2 13/09/14 9:32 AM

T H e H i S TO r y O F C O G n i T i v e P S y C H O l O G y / 3

ways of organizing groups are more effective and stable than others (sociology),
and why natural languages have certain features (linguistics). Cognitive psy-
chology is thus the foundation on which all other social sciences stand, in the
same way that physics is the foundation for the other physical sciences.

Nonetheless, much social science has developed without grounding in
cognitive psychology, for two main reasons. First, the field of cognitive psy-
chology is not that advanced. Second, researchers in other areas of social
science have managed to find other ways to explain the phenomena in which
they are interested. An interesting case in point is economics. Neoclassical
economics, which dominated the last century, tried to predict the behavior of
markets while completely ignoring the cognitive processes of individuals. It
simply assumed that individuals behaved in ways to maximize their wealth.
However, the recently developed field of behavioral economics acknowledges
that the behavior of markets is affected by the flawed decision-making pro-
cesses of individuals—for example, people are willing to pay more for some-
thing when they use a credit card than when they use cash (Simester &
Drazen, 2001). In recognition of the importance of the psychology of deci-
sion making to economics, the cognitive psychologist Daniel Kahneman was
awarded the Nobel Prize for economics in 2002.

■ Cognitive psychology is the foundation for many other areas of
social science.

Practical Applications
Practical applications of the field constitute another key incentive for the study
of cognitive psychology. If we really understood how people acquire knowledge
and intellectual skills and how they perform feats of intelligence, then we would
be able to improve their intellectual training and performance accordingly.

While future applications of psychology hold great promise (Klatzky,
2009), there are a number of current successful applications. For instance,
there has been a long history of research on the reliability of eyewitness
testimony (e.g., Loftus, 1996) that has led to guidelines for law enforcement
personnel (U.S. Department of Justice, 1999). There have also been a number of
applications of basic information processing to the design evaluations of vari-
ous computer-based devices, such as modern flight management systems on
aircraft (John, Patton, Gray, & Morrison, 2012). And there have been a num-
ber of applications to education, including reading instruction (Rayner, Foor-
man, Perfetti, Pesetsky, & Seidenberg, 2002) and computer-based systems for
teaching mathematics (Koedinger & Corbett, 2006). Cognitive psychology is
also making important contributions to our understanding of brain disorders
that reflect abnormal functioning, such as schizophrenia (Cohen & Servan-
Schreiber, 1992) or autism (Dinstein et al., 2012; Just, Keller, & Kana, 2013).

At many points in this book, Implications boxes will reinforce the connec-
tions between research in cognitive psychology and our daily lives.

■ The results from the study of cognitive psychology have practical
implications for our daily lives.

◆ The History of Cognitive Psychology

Cognitive psychology today is a vigorous science producing many interesting
discoveries. However, this productive phase was a long time coming, and it is
important to understand the history of the field that led to its current form.

Anderson_8e_Ch01.indd 3 13/09/14 9:32 AM

4 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

Early History
In Western civilization, interest in human cognition can be traced to the
ancient Greeks. Plato and Aristotle, in their discussions of the nature and
origin of knowledge, speculated about memory and thought. These early
philosophical discussions eventually developed into a centuries-long debate
between two positions: empiricism, which held that all knowledge comes
from experience, and nativism, which held that children come into the world
with a great deal of innate knowledge. The debate intensified in the 17th, 18th,
and 19th centuries, with such British philosophers as Berkeley, Locke, Hume,
and Mill arguing for the empiricist view and such continental philosophers as
Descartes and Kant propounding the nativist view. Although these arguments
were philosophical at their core, they frequently slipped into psychological
speculations about human cognition.

During this long period of philosophical debate, sciences such as
astronomy, physics, chemistry, and biology developed markedly. Curiously,
however, it was not until the end of the 19th century that the scientific method
was applied to the understanding of human cognition. Certainly, there were no
technical or conceptual barriers to the scientific study of cognitive psychology
earlier. In fact, many cognitive psychology experiments could have been per-
formed and understood in the time of the ancient Greeks. But cognitive
psychology, like many other sciences, suffered because of our egocentric, mys-
tical, and confused attitudes about ourselves and our own nature, which made
it seem inconceivable that the workings of the human mind could be subjected
to scientific analysis. As a consequence, cognitive psychology as a science is less
than 150 years old, and much of the first 100 years was spent freeing ourselves
of the misconceptions that can arise when people engage in such an introverted
enterprise as a scientific study of human cognition. It is a case of the mind
studying itself.

■ Only in the last 150 years has it been realized that human cogni-
tion could be the subject of scientific study rather than philosophical
speculation.

Psychology in Germany: Focus on Introspective
Observation
The date usually cited as the beginning of psychology as a science is 1879,
when Wilhelm Wundt established the first psychology laboratory in Leipzig,
Germany. Wundt’s psychology was cognitive psychology (in contrast to other
major divisions, such as comparative, clinical, or social psychology), although
he had far-ranging views on many subjects. Wundt, his students, and many
other early psychologists used a method of inquiry called introspection,
in which highly trained observers reported the contents of their own
consciousness under carefully controlled conditions. The basic assumption was
that the workings of the mind should be open to self-observation. Drawing on
the empiricism of the British philosophers, Wundt and others believed that very
intense self-inspection would be able to identify the primitive experiences out
of which thought arose. Thus, to develop a theory of cognition, a psychologist
had only to explain the contents of introspective reports.

Let us consider a sample introspective experiment. Mayer and Orth (1901)
had their participants perform a free-association task. The experimenters spoke
a word to the participants and then measured the amount of time the partici-
pants took to generate responses to the word. Participants then reported all
their conscious experiences from the moment of stimulus presentation until the

Anderson_8e_Ch01.indd 4 13/09/14 9:32 AM

T H e H i S TO r y O F C O G n i T i v e P S y C H O l O G y / 5

moment of their response. To get a feeling for this method, try to come up with
an association for each of the following words; after each association, think
about the contents of your consciousness during the period between reading
the word and making your association.

coat book
dot bowl

In this experiment, many participants reported rather indescribable
conscious experiences, not always seeming to involve sensations, images,
or other concrete experiences. This result started a debate over the issue of
whether conscious experience could really be devoid of concrete content. As
we will see in Chapters 4 and 5, modern cognitive psychology has made real
progress on this issue, but not by using introspective methods.

■ At the turn of the 20th century, German psychologists tried to use
a method of inquiry called introspection to study the workings of the
mind.

What does cognitive
psychology tell us about
how to study effectively?

Cognitive psychology has identi-
fied methods that enable humans
to read and remember a textbook
like this one. This research will be
described in Chapters 6 and 13. The
key idea is that it is crucial to identify
the main points of each section of
a text and to understand how these
main points are organized. i have
tried to help you do this by ending
each section with a short summary
sentence identifying its main point.
i recommend that you use the fol-
lowing study technique to help
you remember the material. This
approach is a variant of the PQ4r
(Preview, Question, read, reflect,
recite, review) method discussed in
Chapter 6.

1. Preview the chapter. read
the section headings and
summary statements to get
a general sense of where
the chapter is going and
how much material will be
devoted to each topic. Try to
understand each summary
statement, and ask yourself

whether this is something you
knew or believed before read-
ing the text.

Then, for each section of the book,
go through the following steps:

2. For each section of the book,
make up a study question by
looking at the section heading
and thinking of a related ques-
tion that you will try to answer
while you read the text. For
instance, in the section intel-
lectual Curiosity, you might
ask yourself, “What is there to
be curious about in cognitive
psychology?” This will give you
an active goal to pursue while
you read the section.

3. read the section to under-
stand it and answer your
question. Try to relate what
you are reading to situations
in your own life. in the sec-
tion intellectual Curiosity, for
example, you might try to
think of scientific discoveries
you have read about that
seemed to require creativity.

4. At the end of each section,
read the summary and ask
yourself whether that is the
main point you got out of
the section and why it is the
main point. Sometimes you
may need to go back and
reread some parts of the
section.

At the end of the chapter, engage in
the following review process:

5. Go through the text, men-
tally reviewing the main
points. Try to answer the
questions you devised in
step 2, plus any other ques-
tions that occur to you.
Often, when preparing for
an exam, it is a good idea
to ask yourself what kind of
exam questions you would
make up for the chapter.

As we will learn in later chapters,
such a study strategy improves
one’s memory of the text.

I m p l I c a t I o n s
▼

Ha
nq

ua
n

Ch
en

/G
et

ty
Im

ag
es

▲

Anderson_8e_Ch01.indd 5 13/09/14 9:32 AM

6 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

Psychology in America: Focus on Behavior
Wundt’s introspective psychology was not well accepted in America. Early
American psychologists engaged in what they called “introspection,” but it was
not the intense analysis of the contents of the mind practiced by the Germans.
Rather, it was largely an armchair avocation in which self-inspection was casual
and reflective rather than intense and analytic. William James’s Principles
of Psychology (1890) reflects the best of this tradition, and many of the pro-
posals in this work are still relevant today. The mood of America was deter-
mined by the philosophical doctrines of pragmatism and functionalism. Many
psychologists of the time were involved in education, and there was a demand
for an “action-oriented” psychology that was capable of practical application.
The intellectual climate in America was not receptive to the psychology from
Germany that focused on such questions as whether or not the contents of
consciousness were sensory.

One of the important figures of early American scientific psychology
was Edward Thorndike, who developed a theory of learning that was directly
applicable to classrooms. Thorndike was interested in such basic problems
as the effects of reward and punishment on the rate of learning. To him, con-
scious experience was just excess baggage that could be largely ignored. Many
of his experiments were done on animals, research that involved fewer ethical
constraints than research on humans. Thorndike was probably just as happy
that such participants could not introspect.

While introspection was being ignored at the turn of the century in
America, it was getting into trouble on the continent. Various laboratories were
reporting different types of introspections—each type matching the theory of
the particular laboratory from which it emanated. It was becoming clear that
introspection did not give one a clear window into the workings of the mind.
Much that was important in cognitive functioning was not open to conscious
experience. These two factors—the “irrelevance” of the introspective method
and its apparent contradictions—laid the groundwork for the great behaviorist
revolution in American psychology that occurred around 1920. John Watson
and other behaviorists led a fierce attack not only on introspectionism but also
on any attempt to develop a theory of mental operations. Behaviorism held that
psychology was to be entirely concerned with external behavior and was not to
try to analyze the workings of the mind that underlay this behavior:

Behaviorism claims that consciousness is neither a definite nor
a usable concept. The Behaviorist, who has been trained always
as an experimentalist, holds further that belief in the existence of
consciousness goes back to the ancient days of superstition and magic.
(Watson, 1930, p. 2)

The Behaviorist began his own formulation of the problem of
psychology by sweeping aside all medieval conceptions. He dropped
from his scientific vocabulary all subjective terms such as sensation,
perception, image, desire, purpose, and even thinking and emotion as
they were subjectively defined. (Watson, 1930, pp. 5–6)

The behaviorist program and the issues it spawned pushed research on
cognition into the background of American psychology. The rat supplanted the
human as the principal laboratory subject, and psychology turned to finding
out what could be learned by studying animal learning and motivation. Quite
a bit was discovered, but little was of direct relevance to cognitive psychology.
Perhaps the most important lasting contribution of behaviorism is a set of
sophisticated and rigorous techniques and principles for experimental study in
all fields of psychology, including cognitive psychology.

Anderson_8e_Ch01.indd 6 13/09/14 9:32 AM

T H e H i S TO r y O F C O G n i T i v e P S y C H O l O G y / 7

Behaviorism was not as dominant in Europe. Psychologists such as
Frederick Bartlett in England, Alexander Luria in the Soviet Union, and
Jean Piaget in Switzerland were pursuing ideas that are still important in
modern cognitive psychology. Cognitive psychology was an active research topic
in Germany, but much of it was lost in the Nazi turmoil. A number of German
psychologists immigrated to America and brought Gestalt psychology with
them. Gestalt psychology claimed that the activity of the brain and the mind
was more than the sum of its parts. This conflicted with the introspectionist
program in Germany that tried to analyze conscious thought into its parts. In
America, Gestalt psychologists found themselves in conflict with behaviorism
on this point. However, they were also criticized for being concerned with men-
tal structure at all. In America, Gestalt psychologists received the most atten-
tion for their claims about animal learning, and they were the standard targets
for the behaviorist critiques, although some Gestalt psychologists became quite
prominent. For example, the Gestalt psychologist Wolfgang Kohler was elected
to the presidency of the American Psychological Association. Although not a
Gestalt psychologist, Edward Tolman was an American psychologist who did
his research on animal learning and anticipated many ideas of modern cogni-
tive psychology. Tolman’s ideas were also frequently the target for criticism by
the dominant behaviorist psychologists, although his work was harder to dismiss
because he spoke the language of behaviorism.

In retrospect, it is hard to understand how American behaviorists could
have taken such an anti-mental stand and clung to it for so long. The unreli-
ability of introspection did not mean that a theory of internal mental structure
and process could not be developed, only that other methods were required
(consider the analogy with physics, for example, where a theory of atomic
structure was developed, although that structure could only be inferred, not
directly observed). A theory of internal structure makes understanding human
beings much easier, and the successes of modern cognitive psychology show
that understanding mental structures and processes is critical to understanding
human cognition.

In both the introspectionist and behaviorist programs, we see the human
mind struggling with the effort to understand itself. The introspectionists held a
naïve belief in the power of self-observation. The behaviorists were so afraid of
falling prey to subjective fallacies that they refused to let themselves think about
mental processes. Current cognitive psychologists seem to be much more at ease
with their subject matter. They have a relatively detached attitude toward human
cognition and approach it much as they would any other complex system.

■ Behaviorism, which dominated American psychology in the first
half of the 20th century, rejected the analysis of the workings of the
mind to explain behavior.

The Cognitive Revolution: AI, Information Theory,
and Linguistics
Cognitive psychology as we know it today took form in the two decades
between 1950 and 1970, in the cognitive revolution that overthrew behavior-
ism. Three main influences account for its modern development. The first was
research on human performance, which was given a great boost during World
War II when governments badly needed practical information about how
to train soldiers to use sophisticated equipment and how to deal with prob-
lems such as the breakdown of attention under stress. Behaviorism offered no
help with such practical issues. Although the work during the war had a very

Anderson_8e_Ch01.indd 7 13/09/14 9:32 AM

8 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

practical bent, the issues it raised stayed with psychologists when they went back
to their academic laboratories after the war. The work of the British psychologist
Donald Broadbent at the Applied Psychology Research Unit in Cambridge was
probably the most influential in integrating ideas from human performance re-
search with new ideas that were developing in an area called information theory.
Information theory is an abstract way of analyzing the processing of informa-
tion. Broadbent and other psychologists, such as George Miller, Fred Attneave,
and Wendell Garner, initially developed these ideas with respect to perception
and attention, but such analyses soon pervaded all of cognitive psychology.

The second influence, which was closely related to the development of the
information-processing approach, was developments in computer science, par-
ticularly AI, which tries to get computers to behave intelligently, as noted above.
Allen Newell and Herbert Simon, both at Carnegie Mellon University, spent
most of their lives educating cognitive psychologists about the implications of
AI (and educating workers in AI about the implications of cognitive psychol-
ogy). Although the direct influence of AI-based theories on cognitive psychol-
ogy has always been minimal, its indirect influence has been enormous. A host
of concepts have been taken from computer science and used in psychological
theories. Probably more important, observing how we can analyze the intel-
ligent behavior of a machine has largely liberated us from our inhibitions and
misconceptions about analyzing our own intelligence.

The third influence on cognitive psychology was linguistics, which
studies the structure of language. In the 1950s, Noam Chomsky, a linguist
at the Massachusetts Institute of Technology, began to develop a new mode
of analyzing the structure of language. His work showed that language was
much more complex than had previously been believed and that many of the
prevailing behaviorist formulations were incapable of explaining these com-
plexities. Chomsky’s linguistic analyses proved critical in enabling cognitive
psychologists to fight off the prevailing behaviorist conceptions. George Miller,
at Harvard University in the 1950s and early 1960s, was instrumental in bring-
ing these linguistic analyses to the attention of psychologists and in identifying
new ways of studying language.

Cognitive psychology has grown rapidly since the 1950s. A milestone was
the publication of Ulric Neisser’s Cognitive Psychology in 1967. This book gave
a new legitimacy to the field. It consisted of 6 chapters on perception and atten-
tion and 4 chapters on language, memory, and thought. Neisser’s chapter divi-
sion contrasts sharply with this book’s, which has only 2 chapters on perception
and attention and 10 on language, memory, and thought. My chapter division
reflects a growing emphasis on higher mental processes. Following Neisser’s
work, another important event was the launch of the journal Cognitive Psychol-
ogy in 1970. This journal has done much to define the field.

In the 1970s, a related new field called cognitive science emerged; it at-
tempts to integrate research efforts from psychology, philosophy, linguistics,
neuroscience, and AI. This field can be dated from the appearance of the journal
Cognitive Science in 1976, which is the main publication of the Cognitive Science
Society. The fields of cognitive psychology and cognitive science overlap. Speak-
ing generally, cognitive science makes greater use of such methods as logical
analysis and the computer simulation of cognitive processes, whereas cognitive
psychology relies heavily on experimental techniques for studying behavior that
grew out of the behaviorist era. This book draws on all methods but makes most
use of cognitive psychology’s experimental methodology.

■ Cognitive psychology broke away from behaviorism in response to
developments in information theory, AI, and linguistics.

Anderson_8e_Ch01.indd 8 13/09/14 9:32 AM

T H e H i S TO r y O F C O G n i T i v e P S y C H O l O G y / 9

Information-Processing Analyses
The factors described in the previous sections of this chapter have converged
in the information-processing approach to studying human cognition, and
this has become the dominant approach in cognitive psychology. The infor-
mation-processing approach attempts to analyze cognition as a set of steps for
processing an abstract entity called “information.” Probably the best way to ex-
plain this approach is to describe a classic example of it.

In a very influential paper published in 1966, Saul Sternberg described an
experimental task and proposed a theoretical account of what people were doing
in that task. In what has come to be called the Sternberg paradigm, participants
were shown a small number of digits, such as “3 9 7,” to keep in mind. Then they
were shown a probe digit and asked whether it was in the memory set, and they
had to answer as quickly as possible. For example, 9 would be a positive probe
for the “3 9 7” set; 6 would be a negative probe. Sternberg varied the number
of digits in the memory set from 1 to 6 and measured how quickly participants
could make this judgment. Figure 1.1 shows his results as a function of the size
of the memory set. Data are plotted separately for positive probes, or targets,
and for negative probes, or foils. Participants could make these judgments quite
quickly; latencies varied from 400 to 600 milliseconds (ms)—a millisecond is a
thousandth of a second. Sternberg found a nearly linear relationship between
judgment time and the size of the memory set. As shown in Figure 1.1, partici-
pants took about 38 ms extra to judge each digit in the set.

Sternberg’s account of how participants made these judgments was very
influential; it exemplified what an abstract information-processing theory is
like. His explanation is illustrated in Figure 1.2. Sternberg assumed that when
participants saw a probe stimulus such as a 9, they went through the series
of information-processing stages illustrated in that figure. First the stimu-
lus was encoded. Then the stimulus was compared to each digit in the mem-
ory set. Sternberg assumed that it took 38 ms to complete each one of these
comparisons, which accounted for the slope of the line in Figure 1.1. Then
the participant had to decide on a response and finally generate it. Sternberg
showed that different variables would influence each of these information-
processing stages. Thus, if he degraded the stimulus quality by making the
probe harder to read, participants took longer to make their judgments. This
did not affect the slope of the Figure 1.1 line, however, because it involved only
the stage of stimulus perception in Figure 1.2. Similarly,
if he biased participants to say yes or no, the decision-
making stage, but not other stages, was affected.

It is worth noting the ways in which Sternberg’s
theory exemplifies a classic abstract information-
processing account:

1. Information processing is discussed without any
reference to the brain.

2. The processing of the information has a highly sym-
bolic character. For example, his theory describes the
human system as comparing the symbol 9 against the
symbol 3, without considering how these symbols
might be represented in the brain.

3. The processing of information can be compared to
the way computers process information. (In fact,
Sternberg used the computer metaphor to justify his
theory.)

4. The measurement of time to make a judgment is a
critical variable, because the information processing is

FIGURE 1.1 The time needed
to recognize a digit increases
with the number of items in the
memory set. The straight line
represents the linear function that
fits the data best. (Data from
S. Sternberg, 1969.)

400

450

500

600

650

550

0
1 2 3 4 5 6 7

Size (s ) of memory set

Time = 397 + 38 s

Re
ac

tio
n

tim
e

(m
s)

Targets
Foils

Sternberg Memory
Search

Anderson_8e_Ch01.indd 9 13/09/14 9:32 AM

10 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

conceived to be taking place in discrete stages. Flowcharts such as the one in
Figure 1.2 have been a very popular means of expressing the steps of infor-
mation processing.

Each of these four features listed above reflects a kind of narrowness in the
classic information-processing approach to human cognition. Cognitive psy-
chologists have gradually broadened their approach as they have begun to deal
with more complex phenomena and as they have begun to pay more attention
to the nature of information processing in the brain. For instance, this textbook
has evolved over its editions to reflect this shift.

■ Information-processing analysis breaks a cognitive task down into
a set of abstract information-processing steps.

Cognitive Neuroscience
Over the centuries there has been a lot of debate about the possible relationship
between the mind and the body. Many philosophers, such as Rene Descartes,
have advocated a position called dualism, which posits that the mind and the
body are separate kinds of entities. Although very few scientific psychologists
believe in dualism, until recently many believed that brain activity was too
obscure to provide a basis for understanding human cognition. Most of the re-
search in cognitive psychology had relied on behavioral methods, and most of
the theorizing was of the abstract information-processing sort. However, with
the steady development of knowledge about the brain and methods for studying
brain activity, barriers to understanding the mind by studying the brain are
slowly being eliminated, and brain processes are now being considered in almost
all analyses of human cognition. The field of cognitive neuroscience is devoted
to the study of how cognition is realized in the brain, with exciting new findings
even in the study of the most complex thought processes. The remainder of
this chapter will be devoted to describing some of the neuroscience knowledge
and methods that now inform the study of human cognition, enabling us to see
how cognition unfolds in the brain (for example, at the end of this chapter I will
describe a study of the neural processes that are involved as one solves a math-
ematical equation).

■ Cognitive neuroscience is developing methods that enable us to un-
derstand the neural basis of cognition.

◆ Information Processing: The Communicative
Neurons

The brain is just one part of the nervous system, which also includes the various
sensory systems that gather information from other parts of the body and the
motor systems that control movement. In some cases, considerable information
processing takes place outside the brain. From an information-processing point

FIGURE 1.2 Sternberg’s
analysis of the sequence
of information-processing
stages in his task.

9 Perceive
stimulus

Yes9 = 3? 9 = 9? 9 = 7? Make
decision

Generate
response

Anderson_8e_Ch01.indd 10 13/09/14 9:32 AM

i n F O r m AT i O n P r O C e S S i n G : T H e C O m m u n i C AT i v e n e u r O n S / 11

of view, neurons are the most important components of the nervous system.2
A neuron is a cell that receives and transmits signals through electrochemical
activity. The human brain contains approximately 100 billion neurons, each
of which may have roughly the processing capability of a small computer. A
considerable fraction of these 100 billion neurons are active simultaneously
and do much of their information processing through interactions with one
another. Imagine the information-processing power in 100 billion interacting
computers! On the other hand, there are many tasks, such as finding square
roots, at which a simple calculator can outperform all 100 billion neurons.
Comprehending the strengths and weaknesses of the human nervous system is
a major goal in understanding the nature of human cognition.

The Neuron
Neurons come in a wide variety of shapes and sizes, depending on their exact
location and function. (Figure 1.3 illustrates some of this variety.) There is,
however, a generally accepted notion of what the prototypical neuron is like,
and individual neurons match up with this prototype to greater or lesser
degrees. This prototype is illustrated in Figure 1.4. The main body of the neu-
ron is called the soma. Typically, the soma is 5 to 100 micrometers (μm) in
diameter. Attached to the soma are short branches called dendrites, and
extending from the soma is a long tube called the axon. The axon can vary in
length from a few millimeters to a meter.

Axons provide the fixed paths by which neurons communicate with one
another. The axon of one neuron extends toward the dendrites of other neu-
rons. At its end, the axon branches into a large number of arborizations.
Each arborization ends in terminal boutons that almost make contact with
the dendrite of another neuron. The gap separating the terminal bouton and
the dendrite is typically in the range of 10 to 50 nanometers (nm). This near
contact between axon and dendrite is called a synapse. Typically, neurons
communicate by releasing chemicals, called neurotransmitters, from the axon

2 Neurons are by no means the majority of cells in the nervous system. There are many others, such as glial
cells, whose main function is thought to be supportive of the neurons.

FIGURE 1.3 Some of the vari-
ety of neurons: (a) pyramidal
cell; (b) cerebellar Purkinje cell;
(c) motor neuron; (d) sensory
neuron.

Dendrite

Axon

Cell body

Dendrite

Axon

Cell body

Dendrite

Cell body

Myelin sheath

Schwann cell

Neuromuscular
junction

Node

Muscle

Receptor
cell

Peripheral
branch

Central
branch

Cell body

(a) (b) (c) (d)

Anderson_8e_Ch01.indd 11 13/09/14 9:32 AM

12 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

terminal on one side of the synapse; these chemicals act on the membrane of
the receptor dendrite to change its polarization, or electric potential. The in-
side of the membrane covering the entire neuron tends to be 70 millivolts (mV)
more negative than the outside, due to the greater concentration of negative
chemical ions inside and positive ions outside. The existence of a greater con-
centration of positive sodium ions on the outside of the membrane is particu-
larly important to the functioning of the neuron. Depending on the nature of
the neurotransmitter, the potential difference can decrease or increase. Synapses
that decrease the potential difference are called excitatory, and those that in-
crease the difference are called inhibitory.

The average soma and dendrite have about 1,000 synapses from other
neurons, and the average axon synapses to about 1,000 neurons. The change
in electric potential due to any one synapse is rather small, but the indi-
vidual excitatory and inhibitory effects will accumulate. If there is enough
net excitatory input, the potential difference in the soma can drop sharply.
If the reduction in potential is large enough, a depolarization will occur at
the axon hillock, where the axon joins the soma (see Figure 1.4). This de-
polarization is caused by a rush of positive sodium ions into the inside of
the neuron. The inside of the neuron momentarily (for a millisecond) be-
comes more positive than the outside. This sudden change, called an ac-
tion potential (or spike), will propagate down the axon. That is, the
potential difference will suddenly and momentarily change down the
axon. The rate at which this change travels can vary from 0.5 to 130 m/s,
depending on the characteristics of the axon—such as the degree to which the
axon is covered by a myelin sheath (the more myelination, the faster the trans-
mission). When the nerve impulse reaches the end of the axon, it causes neuro-
transmitters to be released from the terminal boutons, thus continuing the cycle.

To review: Potential changes accumulate on a cell body, reach a threshold,
and cause an action potential to propagate down an axon. This pulse in turn
causes neurotransmitters to be sent from the axon terminal to the body of a dif-
ferent neuron, causing changes in that neuron’s membrane potential. This se-
quence is almost all there is to neural information processing, yet intelligence
arises from this simple system of interactions. The challenge for cognitive neu-
roscience is to understand how.

The time required for this neural communication to complete the path
from one neuron to another is roughly 10 ms—definitely more than 1 ms and
definitely less than 100 ms; the exact speed depends on the characteristics of
the neurons involved. This is much slower than the billions of operations that

FIGURE 1.4 A schematic
representation of a typical
neuron.

Dendrites

Axon
Axon hillock

Myelin sheath

Arborizations

Terminal
boutons

Cell body
(soma)

Nucleus

Anderson_8e_Ch01.indd 12 13/09/14 9:32 AM

i n F O r m AT i O n P r O C e S S i n G : T H e C O m m u n i C AT i v e n e u r O n S / 13

a modern computer can perform in one second. However, there are billions of
these activities occurring simultaneously throughout the brain.

■ Neurons communicate by releasing chemicals, called neurotrans-
mitters, from the axon terminal on one side of the synapse, and these
neurotransmitters act on the membrane of the receptor dendrite to
change its electric potential.

Neural Representation of Information
Two quantities are particularly important to the representation of informa-
tion in the brain. First, as we just saw, the membrane potential can be more
or less negative. Second, the number of action potentials, or nerve impulses,
an axon transmits per second, called its rate of firing, can vary from very few
to upward of 100. The greater the rate of firing, the greater the effect the axon
will have on the cells to which it synapses. We can contrast information rep-
resentation in the brain with information representation in a computer, where
individual memory cells, or bits, can have just one of two values—off (0) or
on (1). A typical computer cell does not have the continuous variation of a
typical neural cell.

We can think of a neuron as having an activation level that corresponds
roughly to the firing rate on the axon or to the degree of depolarization on the
dendrite and soma. Neurons interact by driving up the activation level of other
neurons (excitation) or by driving down their activation level (inhibition). All
neural information processing takes place in terms of these excitatory and in-
hibitory effects; they are what underlies human cognition.

How do neurons represent information? Evidence suggests that individual
neurons respond to specific features of a stimulus. For instance, some neurons
are most active when there is a line in the visual field at a particular angle (as
described in Chapter 2), while other neurons respond to more complex sets of
features. For instance, there are neurons in the monkey brain that appear to be
most responsive to faces (Bruce, Desimone, & Gross, 1981; Desimone, Albright,
Gross, & Bruce, 1984; Perrett, Rolls, & Caan, 1982). It is not possible, however,
that single neurons encode all the concepts and shades of meaning we pos-
sess. Moreover, the firing of a single neuron cannot represent the complexity of
structure in a face.

If a single neuron cannot represent the complexity of our cognition, how are
complex concepts and experiences represented? How can the activity of neurons
represent our concept of baseball; how can it result in our solution of an algebra
problem; how can it result in our feeling of frustration? Similar questions can be
asked of computer programs, which have been shown to be capable of answering
questions about baseball, solving algebra problems, and displaying frustration.
Where in the millions of off-and-on bits in a computer program does the concept
of baseball lie? How does a change in a bit result in the solution of an algebra
problem or in a feeling of frustration? However, these questions fail to see the
forest for the trees. The concepts of a sport, a problem solution, or an emotion
occur in large patterns of bit changes. Similarly, human cognition is achieved
through large patterns of neural activity. One study (Mazoyer et al., 1993)
compared participants who heard random words to participants who heard
words that made nonsense sentences, to participants who heard words that made
coherent sentences. Using methods that will be described shortly, the researchers
measured brain activity. They found activity in more and more regions of the
brain as participants went from hearing words to hearing sentences, to hearing
meaningful stories. This result indicates that our understanding of a meaningful
story involves activity in many regions of the brain.

Anderson_8e_Ch01.indd 13 13/09/14 9:32 AM

14 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

It is informative to think about how the computer stores information.
Consider a simple case: the spelling of words. Most computers have codes by
which individual patterns of binary values (1s and 0s) represent letters. Table 1.1
illustrates the use of one coding scheme, called ASCII; it contains a pattern of 0s
and 1s that codes the words cognitive psychology.

Similarly, the brain can represent information in terms of patterns of neu-
ral activity rather than simply as cells firing. The code in Table 1.1 includes re-
dundant bits that allow the computer to correct errors should certain bits be
lost (note that each column has an even number of 1s, which reflects the added
bits for redundancy). As in a computer, it seems that the brain codes informa-
tion redundantly, so that even if certain cells are damaged, it can still determine
what the pattern is encoding. It is generally thought that the brain uses schemes
for encoding information and achieving redundancy that are very different
from the ones a computer uses. It also seems that the brain uses a much more
redundant code than a computer does because the behavior of individual neu-
rons is not particularly reliable.

So far, we have talked only about patterns of neural activation. Such pat-
terns, however, are transitory. The brain does not maintain the same pattern
for minutes, let alone days. This means that neural activation patterns cannot
encode our permanent knowledge about the world. It is thought that memo-
ries are encoded by changes in the synaptic connections among neurons. By
changing the synaptic connections, the brain can enable itself to reproduce spe-
cific patterns. Although there is not a great deal of growth of new neurons or
new synapses in the adult, the effectiveness of synapses can change in response
to experience. There is evidence that synaptic connections do change during
learning, with both increased release of neurotransmitters (Kandel & Schwartz,
1984) and increased sensitivity of dendritic receptors (Lynch & Baudry, 1984).
We will discuss some of this research in Chapter 6.

■ Information is represented by patterns of activity across many re-
gions of the brain and by changes in the synaptic connections among
neurons that allow these patterns to be reproduced.

TABLE 1.1 Coding of the Words COGNITIVE PSYCHOLOGY in 7-Bit
ASCii with even Parity

1 1 0 0 1 1 1 0 1
1 1 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0 0
0 0 0 0 0 1 0 1 0
0 1 0 1 1 0 1 0 0
0 1 1 1 0 1 0 1 1
1 1 1 1 0 0 0 1 0
1 1 1 0 1 0 1 0 1
0 0 0 1 0 1 1 1 0 0
1 1 1 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0 0 0
1 1 1 0 0 0 0 0 0 1
0 0 1 0 1 1 1 1 0 1
0 0 0 0 0 1 1 1 1 0
0 1 0 1 0 1 0 1 1 0
0 1 1 1 0 1 0 1 1 1

Anderson_8e_Ch01.indd 14 13/09/14 9:32 AM

O r G A n i z AT i O n O F T H e B r A i n / 15

◆ Organization of the Brain

The central nervous system consists of the brain and the spinal cord. The major
function of the spinal cord is to carry neural messages from the brain to the
muscles, and sensory messages from the body to the brain. Figure 1.5 shows a
cross section of the brain with some of the more prominent neural structures
labeled. The lower parts of the brain are evolutionarily more primitive. The
higher portions are well developed only in the higher species.

Correspondingly, it appears that the lower portions of the brain are respon-
sible for more basic functions. The medulla controls breathing, swallowing,
digestion, and heartbeat. The hypothalamus regulates the expression of basic
drives. The cerebellum plays an important role in motor coordination and vol-
untary movement. The thalamus serves as a relay station for motor and sensory
information from lower areas to the cortex. Although the cerebellum and thala-
mus serve these basic functions, they also have evolved to play an important
role in higher human cognition, as we will discuss later.

The cerebral cortex, or neocortex, is the most recently evolved portion of
the brain. Although it is quite small and primitive in many mammals, it accounts
for a large fraction of the human brain. In the human, the cerebral cortex can be
thought of as a rather thin neural sheet with a surface area of about 2,500 cm2.
To fit this neural sheet into the skull, it has to be highly convoluted. The large
amount of folding and wrinkling of the cortex is one of the striking physical dif-
ferences between the human brain and the brains of lower mammals. A bulge of
the cortex is called a gyrus, and a crease passing between gyri is called a sulcus.

The neocortex is divided into left and right hemispheres. One of the in-
teresting curiosities of anatomy is that the right part of the body tends to be
connected to the left hemisphere and the left part of the body to the right hemi-
sphere. Thus, the left hemisphere controls motor function and sensation in the
right hand. The right ear is most strongly connected to the left hemisphere. The
neural receptors in either eye that receive input from the left part of the visual
world are connected to the right hemisphere (as Chapter 2 will explain with re-
spect to Figures 2.5 and 2.6).

Brodmann (1909/1960) identified 52 distinct regions of the human cortex
(see Color Plate 1.1), based on differences in the cell types in various regions.
Many of these regions proved to have functional differences as well. The corti-
cal regions are typically organized into four lobes: frontal, parietal, occipital, and
temporal (Figure 1.6). Major folds, or sulci, on the
cortex separate the areas. The occipital lobe con-
tains the primary visual areas. The parietal lobe
handles some perceptual functions, including spa-
tial processing and representation of the body. It
is also involved in control of attention, as we will
discuss in Chapter 3. The temporal lobe receives
input from the occipital area and is involved in ob-
ject recognition. It also has the primary auditory
areas and Wernicke’s area, which is involved in
language processing. The frontal lobe has two
major functions: The back portion of the frontal
lobe is involved primarily with motor functions.
The front portion, called the prefrontal cor-
tex, is thought to control higher level processes,
such as planning. The frontal portion of the
brain is disproportionately larger in primates
than in most mammals and, among primates,

FIGURE 1.5 A cross-sectional
view of the brain showing some
of its major components.

Neocortex

Thalamus

Optic nerve

Pituitary

Hypothalamus

Midbrain
Cerebellum

Medulla

Pons

Anderson_8e_Ch01.indd 15 13/09/14 9:32 AM

16 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

humans are distinguished by having disproportionately larger anterior por-
tions of the prefrontal cortex (Area 10 in Color Plate 1.1—Semendeferi,
Armstrong, Schleicher, Zilles, & Van Hoesen, 2001). Figure 1.6 will be repeated
at the start of many of the chapters in the text, with an indication of the areas rel-
evant to the topics in those chapters.

The neocortex is not the only region that plays a significant role in higher level
cognition. There are many important circuits that go from the cortex to subcortical
structures and back again. A particularly significant area for memory proves to be
the limbic system, which is at the border between the cortex and the lower struc-
tures. The limbic system contains a structure called the hippocampus (located
inside the temporal lobes), which appears to be critical to human memory. It is not
possible to show the hippocampus in a cross section like Figure 1.5, because it is a
structure that occurs in the right and left halves of the brain between the surface
and the center. Figure 1.7 illustrates the hippocampus and related structures. Dam-
age to the hippocampus and to other nearby structures produces severe amnesia,

as we will see in Chapter 7.
Another important collection of subcorti-

cal structures is the basal ganglia. The critical
connections of the basal ganglia are illustrated
in Figure 1.8. The basal ganglia are involved
both in basic motor control and in the control of
complex cognition. These structures receive pro-
jections from almost all areas of the cortex and
have projections to the frontal cortex. Disorders
such as Parkinson’s disease and Huntington’s
disease result from damage to the basal ganglia.
Although people suffering from these diseases
have dramatic motor control deficits character-
ized by tremors and rigidity, they also have diffi-
culties in cognitive tasks. The cerebellum, which
has a major role in motor control, also seems to
play a role in higher order cognition. Many cog-
nitive deficits have been observed in patients
with damage to the cerebellum.

FIGURE 1.6 A side view of the
cerebral cortex showing the four
lobes—frontal, occipital, parietal,
and temporal—of each hemi-
sphere (blue-shaded areas) and
other major components of the
cerebral cortex.

FIGURE 1.7 Structures under the
cortex that are part of the limbic
system, which includes the hip-
pocampus. related structures are
labeled.

Broca’s
area

Prefrontal
association
cortex

Primary auditory
cortex

Wernicke’s
area

Sylvian fissure

Preoccipital notch

Primary visual
cortex

Parietal lobe

Motor cortex
Central sulcus

Cerebellum

Parietal-temporal-
occipital association
cortex

Primary somatic sensory cortex

Frontal lobe
Occipital
lobe

Temporal lobe

Basal
ganglia

Cerebral
cortex

Thalamus

Hippocampus

Amygdala

Anderson_8e_Ch01.indd 16 13/09/14 9:32 AM

O r G A n i z AT i O n O F T H e B r A i n / 17

■ The brain is organized into a number
of distinct areas, which serve different
types of functions, with the cerebral cor-
tex playing the major role in higher cog-
nitive functions.

Localization of Function

The left and right hemispheres of the cerebral
cortex appear to be somewhat specialized for
different types of processing. In general, the
left hemisphere seems to be associated with
linguistic and analytic processing, whereas
the right hemisphere is associated with per-
ceptual and spatial processing. The left and
right hemispheres are connected by a broad
band of fibers called the corpus callosum.
The corpus callosum has been surgically
severed in some patients to prevent epileptic
seizures. Such patients are referred to as split-
brain patients. The operation is typically suc-
cessful, and patients seem to function fairly
well. Much of the evidence for the differences
between the hemispheres comes from re-
search with these patients. In one experiment,
the word key was flashed on the left side of a screen the patient was viewing.
Because it was on the left side of the screen, it would be received by the right,
nonlanguage hemisphere. When asked what was presented on the screen, the
patient was not able to say because the language-dominant hemisphere did not
know. However, his left hand (but not the right) was able to pick out a key from
a set of objects hidden from view.

Studies of split-brain patients have enabled psychologists to identify the
separate functions of the right and left hemispheres. The research has shown a
linguistic advantage for the left hemisphere. For instance, commands might be
presented to these patients in the right ear (and hence to the left hemisphere)
or in the left ear (and hence to the right hemisphere). The right hemisphere
can comprehend only the simplest linguistic commands, whereas the left hemi-
sphere displays full comprehension. A different result is obtained when the
ability of the right hand (hence the left hemisphere) to perform manual tasks is
compared with that of the left hand (hence the right hemisphere). In this situa-
tion, the right hemisphere clearly outperforms the left hemisphere.

Research with other patients who have had damage to specific brain re-
gions indicates that there are areas in the left cortex, called Broca’s area and
Wernicke’s area (see Figure 1.6), that seem critical for speech, because dam-
age to them results in aphasia, the severe impairment of speech. These may
not be the only neural areas involved in speech, but they certainly are impor-
tant. Different language deficits appear depending on whether the damage is
to Broca’s area or Wernicke’s area. People with Broca’s aphasia (i.e., damage to
Broca’s area) speak in short, ungrammatical sentences. For instance, when one
patient was asked whether he drives home on weekends, he replied:

Why, yes . . . Thursday, er, er, er, no, er, Friday . . . Bar-ba-ra . . . wife
. . . and, oh, car . . . drive . . . purnpike . . . you know . . . rest and . . .
teevee. (Gardner, 1975, p. 61)

To motor cortex
and frontal areas

Thalamus

Subthalamic
nucleus

Substantia nigra Globus pallidus Putamen

Cerebral cortex

Caudate nucleus

FIGURE 1.8 The major structures
of the basal ganglia (blue-shaded
areas) include the caudate nu-
cleus, the subthalamic nucleus,
the substantia nigra, the globus
pallidus, and the putamen. The
critical connections (inputs and
outputs) of the basal ganglia are
illustrated. (After Gazzinga, Ivry, &
Mangun, 2002.)

Anderson_8e_Ch01.indd 17 13/09/14 9:32 AM

18 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

In contrast, patients with Wernicke’s aphasia speak in fairly grammatical sen-
tences that are almost devoid of meaning. Such patients have difficulty with
their vocabulary and generate “empty” speech. The following is the answer
given by one such patient to the question “What brings you to the hospital?”

Boy, I’m sweating, I’m awful nervous, you know, once in a while I get
caught up, I can’t mention the tarripoi, a month ago, quite a little, I’ve
done a lot well. I impose a lot, while, on the other hand, you know
what I mean, I have to run around, look it over, trebbin and all that
sort of stuff. (Gardner, 1975, p. 68)

■ Different specific areas of the brain support different cognitive
functions.

Topographic Organization
In many areas of the cortex, information processing is structured spatially in
what is called a topographic organization. For instance, in the visual area at
the back of the cortex, adjacent areas represent information from adjacent areas
of the visual field. Figure 1.9 illustrates this fact (Tootell, Silverman, Switkes,
& DeValois, 1982). Monkeys were shown the bull’s-eye pattern represented in
Figure 1.9a. Figure 1.9b shows the pattern of activation that was recorded on the
occipital cortex by injecting a radioactive material that marks locations of maxi-
mum neural activity. We see that the bull’s-eye structure is reproduced with only
a little distortion. A similar principle of organization governs the representation
of the body in the motor cortex and the somatosensory cortex along the cen-
tral fissure. Adjacent parts of the body are represented in adjacent parts of the
neural tissue. Figure 1.10 illustrates the representation of the body along the
somatosensory cortex. Note that the body is distorted, with certain areas receiving
a considerable overrepresentation. It turns out that the overrepresented areas
correspond to those that are more sensitive. Thus, for instance, we can make more
subtle discriminations among tactile stimuli on the hands and face than we can on
the back or thigh. Also, there is an overrepresentation in the visual cortex of the
visual field at the center of our vision, where we have the greatest visual acuity.

It is thought that topographic maps exist so that neurons processing similar
regions can interact with one another (Crick & Asanuma, 1986). Although there
are fiber tracks that connect different regions of the brain, the majority of the
connections among neurons are to nearby neurons. This emphasis on local con-
nections is driven to minimize both the communication time between neurons
and the amount of neural tissue that must be devoted to connecting them. The

FIGURE 1.9 evidence of
topographic organization. A
visual stimulus (a) is pre-
sented to a monkey. The
stimulus produces a pattern
of brain activation (b) in the
monkey that closely matches
the structure of the stimulus.
(From Tootell et al., 1982. Re-
printed with permission from
AAAS.)

(a) (b)

1 cm

Anderson_8e_Ch01.indd 18 13/09/14 9:32 AM

m e T H O d S i n C O G n i T i v e n e u r O S C i e n C e / 19

extreme of localization is the cortical minicolumn (Buxhoeveden & Casanova,
2002)—tiny vertical columns of about 100 neurons that have a very restricted
mission. For instance, cortical columns in the primary visual cortex are special-
ized to process information about one orientation, from one location, in one eye.

Neurons in a minicolumn do not represent a precise location with pin-
point accuracy but rather a range of nearby locations. This relates to another
aspect of neural information processing called coarse coding, which refers
to the fact that single neurons seem to respond to a range of events. For in-
stance, when the neural activity from a single neuron in the somatosensory
cortex is recorded, we can see that the neuron does not respond only when
a single point of the body is stimulated, but rather when any point on a large
patch of the body is stimulated. How, then, can we know exactly what point
has been touched? That information is recorded quite accurately, but not in
the response of any particular cell. Instead, different cells will respond to dif-
ferent overlapping regions of the body, and any point will evoke a different set
of cells. Thus, the location of a point is reflected by the pattern of activation,
which reinforces the idea that neural information tends to be represented in
patterns of activation.

■ Adjacent cells in the cortex tend to process sensory stimuli from ad-
jacent areas of the body.

◆ Methods in Cognitive Neuroscience

How does one go about understanding the neural basis of cognition? Much of
the past research in neuroscience has been done on animals. Some research
has involved the surgical removal of various parts of the cortex. By observing
the deficits these operations have produced, it is possible to infer the func-
tion of the region removed. Other research has recorded the electrical activity
in particular neurons or regions of neurons. By observing what activates these

FIGURE 1.10 A cross section
of the somatosensory cortex,
showing how the human body is
mapped in the neural tissue.

Lateral

Medial

Genitals

Foot

Toes

Leg

Hip

Trunk
Neck
Head

Shoulder
Arm

Elbow

ForearmW
rist

HandLittleRing

MiddleIndex

Thumb

Eye
Nose
Face

Upper lip

Lower lip

Teeth
Gums

Jaw

Tongue

Pharyn
x

Intra-
abdominal

Anderson_8e_Ch01.indd 19 13/09/14 9:32 AM

20 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

neurons, one can infer what they do. However, there is considerable uncertainty
about how much these animal results generalize to humans. The difference
between the cognitive potential of humans and that of most other animals is
enormous. With the possible exception of other primates, it is difficult to get
other animals even to engage in the kinds of cognitive processes that character-
ize humans. This has been the great barrier to understanding the neural basis of
higher level human cognition.

Neural Imaging Techniques

Until recently, the principal basis for understanding the role of the brain in
human cognition has been the study of patient populations. We have already
described some of this research, such as that with split-brain patients and
with patients who have suffered damages to brain areas that cause language
deficits. It was research with patient populations such as these that showed
that the brain is lateralized, with the left hemisphere specialized for language
processing. Such hemispheric specialization does not occur in other species.

More recently, there have been major advances in noninvasive methods
of imaging the functioning of the brains of normal participants engaged in
various cognitive activities. These advances in neural imaging are among the
most exciting developments in cognitive neuroscience and will be referenced
throughout this text. Although not as precise as recording from single neurons,
which can be done only rarely with humans (and then as part of surgical pro-
cedures), these methods have achieved dramatic improvements in precision.

Electroencephalography (EEG) records the electric potentials that
are present on the scalp. When large populations of neurons are active, this
activity will result in distinctive patterns of electric potential on the scalp. In
the typical methodology, a participant wears a cap of many electrodes. The
electrodes detect rhythmic changes in electrical activity and record them
on electroencephalograms Figure 1.11 illustrates some recordings typical of
various cognitive states. When EEG is used to study cognition, the partici-
pant is asked to respond to some stimulus, and researchers are interested in

FIGURE 1.11 eeG profiles
obtained during various
states of consciousness.
(Alila Medical Media/
Shutterstock.)

Anderson_8e_Ch01.indd 20 13/09/14 9:32 AM

m e T H O d S i n C O G n i T i v e n e u r O S C i e n C e / 21

discovering how processing this stimulus impacts general activity on the re-
cordings. To eliminate the effects not resulting from the stimulus, many trials
are averaged, and what remains is the activity produced by the stimulus. For
instance, Kutas and Hillyard (1980) found that there was a large dip in the wave
about 400 ms after participants heard an unexpected word in a sentence (this
is discussed further in Chapter 13). Such averaged EEG responses aligned to a
particular stimulus are called event-related potentials (ERPs). ERPs have very
good temporal resolution, but it is difficult to infer the location in the brain of
the neural activity that is producing the scalp activity.

A recent variation of ERP that offers better spatial resolution is magne-
toencephalography (MEG), which records magnetic fields produced by the
electrical activity. Because of the nature of the magnetic fields it measures,
MEG is best at detecting activity in the sulci (creases) of the cortex and is less
sensitive to activity in the gyri (bumps) or activity deep in the brain.

Two other methods, positron emission tomography (PET) and functional
magnetic resonance imaging (fMRI), provide relatively good information
about the location of neural activity but rather poor information about the time
course of that activity. Neither PET nor fMRI measures neural activity directly.
Rather, they measure metabolic rate or blood flow in various areas of the brain,
relying on the fact that more active areas of the brain require greater metabolic
expenditures and have greater blood flow. PET and fMRI scans can be conceived
as measuring the amount of work a brain region does.

In PET, a radioactive tracer is injected into the bloodstream (the radiation
exposure in a typical PET study is equivalent to two chest X rays and is not
considered dangerous). Participants are placed in a PET scanner that can de-
tect the variation in concentration of the radioactive element. Current methods
allow a spatial resolution of 5 to 10 mm. For instance, Posner, Peterson, Fox,
and Raichle (1988) used PET to localize the various components of the read-
ing process by looking at what areas of the brain are involved in reading a
word. Figure 1.12 illustrates their results. The triangles on the cortex represent
areas that were active when participants were just passively looking at concrete
nouns. The squares represent areas that became active when participants were
asked to engage in the semantic activity of generating uses for these nouns.
The triangles are located in the occipital lobe; the squares, in the frontal lobe.
Thus, the data indicate that the processes of visually perceiving a word take
place in a different part of the brain from the processes of thinking about
the meaning of a word.

The fMRI methodology has largely replaced PET. It offers even better
spatial resolution than PET and is less intrusive. fMRI uses the same MRI scan-
ner that hospitals now use as standard equipment to image various structures,
including patients’ brain structures. With minor modification, it can be used to
image the functioning of the brain. fMRI does not re-
quire injecting the participant with a radioactive tracer
but relies on the fact that there is more oxygenated
hemoglobin in regions of greater neural activity. (One
might think that greater activity would use up oxygen,
but the body responds to effort by overcompensating
and increasing the oxygen in the blood—this is called
the hemodynamic response.) Radio waves are passed
through the brain, and these cause the iron in the
hemoglobin to produce a local magnetic field that is
detected by magnetic sensors surrounding the head.
Thus, fMRI offers a measure of the amount of energy
being spent in a particular brain region: The signal is
stronger in areas where there is greater activity. Among

FIGURE 1.12 Areas in the lateral
aspect of the cortex activated
by visual word reading. Triangles
mark locations activated by the
passive visual task; squares mark
the locations activated by the
semantic task. (Task locations after
Posner et al., 1988.)

Anderson_8e_Ch01.indd 21 13/09/14 9:32 AM

22 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

its advantages over PET are that it allows measurement over longer periods be-
cause there is no radioacive substance injected and that it offers finer temporal
and spatial resolution. In the next section I will describe an fMRI study in detail to
illustrate the basic methodology and what it can accomplish.

Neither PET nor fMRI is what one would call a practical, everyday measure-
ment method. Even the more practical fMRI uses multimillion-dollar scanners
that require the participant to lie motionless in a noisy and claustrophobic
space. There is hope, however, that more practical techniques will become avail-
able. One of the more promising is near-infrared sensing (Strangman, Boas, &
Sutton, 2002). This methodology relies on the fact that light penetrates tissue
(put a flashlight to the palm of your hand to demonstrate this) and is reflected
back. In near-infrared sensing, light is shined on the skull, and the instrument
senses the spectrum of light that is reflected back. It turns out that near-infra-
red light tends not to be absorbed by oxygenated tissue, and so by measuring the
amount of light in the near-infrared region (which is not visible to human eyes),
one can detect the oxygenation of the blood in a particular area of the brain.
This methodology promises to be much cheaper and less confining than PET or
fMRI and does not require movement restriction. Even now it is used with young
children who cannot be convinced to remain still and with Parkinson’s patients
who cannot control their movements. A major limitation of this technique is that
it can only detect activity 2 or 3 cm into the brain because that is as far as the
light can effectively penetrate.

These various imaging techniques have revolutionized our understanding
of the brain activity underlying human cognition, but they have a limitation
that goes beyond temporal and spatial resolution: They provide only a lim-
ited basis for causal inference. Just because activity is detected in a region of
the brain during a task does not mean that the region of the brain is critical
to the execution of the task. Until recently researchers had to study patients
with strokes, brain injuries, and brain diseases to get some understanding of
how critical a region is. However, there are now methods available that allow
researchers to briefly incapacitate a region. Principal among these is a method
called transcranial magnetic stimulation (TMS), in which a coil is placed over
a particular part of the head and a pulse or pulses are delivered to that region
(see Figure 1.13). This will disrupt the processing in the region under the coil.
If properly administered, TMS is safe and has no lasting effect. It can be very
useful in determining the role of different brain regions. For instance, there is
activity in both prefrontal and parietal regions during study of an item that a
participant is trying to remember. Nonetheless, it has been shown that TMS to
the prefrontal region (Rossi et al., 2001) and not the parietal region (Rossi et al.,
2006) disrupts memory formation. This implies a more critical role of the pre-
frontal region in memory formation.

■ Techniques like EEG, MEG, fMRI, and TMS are allowing research-
ers to study the neural basis of human cognition with a precision
starting to approach that available in animal studies.

Using fMRI to Study Equation Solving

Most brain-imaging studies have looked at relatively simple cognitive tasks, as is
still true of most research in cognitive neuroscience. A potential danger of using
such techniques is that we will come to believe that the human mind is capable
only of the simple things that are studied with these neuroscience techniques.
However, it is possible to study more complex processes. For example, I will
describe a study—for which I was one of the researchers (Qin, Anderson, Silk,

Anderson_8e_Ch01.indd 22 13/09/14 9:32 AM

m e T H O d S i n C O G n i T i v e n e u r O S C i e n C e / 23

Stenger, & Carter, 2004)—that looked at equation solving by children aged 11 to
14 when they were just learning to solve equations. This research illustrates the
profitable marriage of information-processing analysis and cognitive neurosci-
ence techniques.

Qin et al. (2004) studied eighth-grade students as they solved equations
at three levels of complexity in terms of the number of steps of transformation
that were required:

0 step: 1x + 0 = 4
1 step: 3x + 0 = 12 or 1x + 8 = 12
2 steps: 7x + 1 = 29

Note that the 0-step equation is rather unusual, with the 1 in front
of the x and the + 0 after the x. This format reflects the fact that the
visual complexity of different conditions must be controlled to avoid
obtaining differences in the visual cortex and elsewhere just because
a more complex visual stimulus has to be processed. Students kept
their heads motionless while being scanned. They wore a response
glove and could press a finger to indicate the answer to the problem
(thumb = 1, index finger = 2, middle finger = 3, ring finger = 4, and
little finger = 5).

Qin et al. (2004) developed an information-processing model
for the solution of such equations that involved imagined trans-
formations of the equations, retrieval of arithmetic and algebraic
facts, and programming of the motor response. Figure 1.14 shows

FIGURE 1.13 TmS is
delivered by a coil on the
surface of the head, which
generates brief put powerful
magnetic pulses that induce
a temporary current in a
small area on the surface of
the brain. This current can
interfere with processing of
the brain with high temporal
and fair spatial precision.
(Boston Globe via Getty
Images.)

Key 4

Time (s) Visual Retrieval Motor

1.2
1.4
1.6
1.8
2.0
2.2
2.4
2.6

4.0

2.8
3.0
3.2
3.4
3.6
3.8

4.2
4.4
4.6
4.8
5.0

? = 29
? = 29

? 1 = 29
? + 1 = 29

Inverse of +

? = 28
7x = 28

x = 4

29 − 1 = 28

28 / 7 = 4

FIGURE 1.14 The steps of an
information-processing model for
solving the equation 7x + 1 = 29.
The model includes imagined
transformations of the equations
(visual processing), retrieval of
arithmetic and algebraic facts,
and programming of the motor
response.

Anderson_8e_Ch01.indd 23 13/09/14 9:32 AM

24 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

the sequencing of these activities. In line with existing re-
search, we would expect that:

FIGURE 1.15 regions of interest
for the fmri scan in the equation-
solving experiment. The imagined
transformations would activate a
region of the left parietal cortex;
the retrieval of arithmetic infor-
mation would activate a region
of the left prefrontal cortex; and
programming of the hand’s
movement would activate the left
motor and somatosensory cortex.

ParietalPrefrontal

Motor

1. Programming of the hand would be reflected in
activation in the left motor and somatosensory cortex.
(See Figure 1.10; participants responded with their
right hands, and so the left cortex would be involved.)

2. The imagined transformations of each equation would
activate a region of the left parietal cortex involved in
mental imagery (see Chapter 4).

3. The retrieval of arithmetic information would activate a
region of the left prefrontal cortex (see Chapters 6 and 7).

Figure 1.15 shows the locations of these three regions of
interest. Each region is a cube with sides of approximately

15 mm. fMRI is capable of much greater spatial resolution, but the application
within this study did not require this level of accuracy.

The times required to solve the three types of equation were 2.0 s for 0 step,
3.6 s for 1 step, and 4.8 s for 2 steps. However, after students pressed the appropriate
finger to indicate the answer, a long rest period followed to allow brain activity to
return to baseline for the next trial. Qin et al. (2004) obtained the data in terms of
the percentage increase over this baseline of the blood oxygen level dependent
(BOLD) response. In this particular experiment, the BOLD response was obtained
for each region every 1.2 s. Figure 1.16a shows the BOLD response in the motor
region for the three conditions. The percentage increase is plotted from the time
the equation was presented. Note that even though students solved the problem
and keyed the answer to the 0-step equation in an average of 2 s, the BOLD func-
tion did not begin to rise above baseline until the third scan after the equation was
solved, and it did not reach peak until after approximately 6.6 s. This result reflects
the fact that the hemodynamic response to a neural activity is delayed because it
takes time for the oxygenated blood to arrive at the corresponding location in the
brain. Basically, the hemodynamic response reaches a peak about 4 to 5 s after the
event. In the motor region (see Figure 1.16a), the BOLD response for a 0-step equa-
tion reached a peak at approximately 6.6 s, for a 1-step equation at approximately
7.8 s, and for a 2-step equation at approximately 9.0 s. Thus, the point of maximum
activity reflects events that were happening about 4 to 5 s previously.

The peak of a BOLD function allows one to read the brain and see when
the activity took place. The height of the function reflects the amount of activity
that took place. Note that the functions for motor activity in Figure 1.16a are
of approximately equal height in the three conditions because it takes the same
amount of effort to program the finger press, independent of the number of
transformations needed to solve the equations.

Figure 1.16b shows the BOLD responses in the parietal region. Like the
responses in the motor region, they peaked at different times, reflecting the
differences in time to solve the equations. They peaked a little earlier, how-
ever, because the BOLD responses reflected the transformations being made
to the mental image of the equation, which occurred before the response was
emitted. Also, the BOLD functions reached very different heights, reflecting
the different number of transformations that needed to be performed to solve
the equation. Figure 1.16c shows the BOLD responses in the prefrontal region,
which were quite similar to those in the parietal region. The important differ-
ence is that there was no rise in the function in the 0-step condition because it
was not necessary to retrieve any information in that condition. Students could
just read the answer from the mental representation of the original equation.

This experiment showed that researchers can separately track different
information-processing components involved in performing a complex task.

Anderson_8e_Ch01.indd 24 13/09/14 9:32 AM

m e T H O d S i n C O G n i T i v e n e u r O S C i e n C e / 25

Time (s)

In
cr

ea
se

in
B

O
LD

re
sp

on
se

(%
)

In
cr

ea
se

in
B

O
LD

re
sp

on
se

(%
)

In
cr

ea
se

in
B

O
LD

re
sp

on
se

(%
)

(a)

(c)

(b)

0 5 10 15 20
−0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4
Response 2.0 s
Peak 6.6 s

Response 3.6 s
Peak 7.8 s

Response 4.8 s
Peak 9.0 s

0 step
1 step
2 step

0 step
1 step
2 step

0 step
1 step
2 step

0.3

0.25

0.2

0.15

0.1

0.05

0

−0.05
0 5 10 15 20

Time (s)

Response 2.0 s
Peak 5.4 s

Response 4.8 s
Peak 7.8 s

Response 3.6 s
Peak 6.6 s

5 10 15 20

Time (s)

0

0

−0.05

0.05

0.10

0.15

0.20

0.25
Response 4.8 s
Peak 7.8 s

Response 2.0 s
No peak

Response 3.6 s
Peak 6.6 s

FIGURE 1.16 responses of the
three regions of interest shown in
Figure 1.14 for different equation
complexities: (a) motor region;
(b) parietal region; (c) prefrontal
region.

The fMRI methodology is especially appropriate for the study of complex cog-
nition. Its temporal resolution is not very good, and so it is difficult to study
very brief tasks such as the Sternberg paradigm (see Figures 1.1 and 1.2). On
the other hand, when a task takes many seconds, it is possible to distinguish
the timing of processes, as we see in Figure 1.16. Because of its high spatial
resolution, fMRI is able to separate out different components of the overall
processing. For brief cognition, ERP is often a more appropriate brain-imaging
technique because it can achieve much finer temporal resolution.

■ fMRI allows researchers to track activity in the brain of different
information-processing components of a complex task.

Anderson_8e_Ch01.indd 25 13/09/14 9:32 AM

26 / Chapter 1 T H e S C i e n C e O F C O G n i T i O n

Key Terms

This chapter has introduced quite a few key terms, most of which will reappear in later chapters:

action potential
aphasia
artificial intelligence (AI)
axon
basal ganglia
behaviorism
blood oxygen level

dependent (BOLD)
response

Broca’s area
cognitive neuroscience
cognitive psychology
corpus callosum
dendrite

electroencephalography
(EEG)

empiricism
event-related potential

(ERP)
excitatory synapse
frontal lobe
functional magnetic

resonance imaging
(fMRI)

Gestalt psychology
gyrus
hemodynamic response
hippocampus

information-processing
approach

inhibitory synapse
introspection
linguistics
magnetoencephalography

(MEG)
nativism
neuron
neurotransmitter
occipital lobe
parietal lobe
positron emission

tomography (PET)

prefrontal cortex
rate of firing
split-brain patients
Sternberg paradigm
sulcus
synapse
temporal lobe
topographic organization
transcranial magnetic

stimulation (TMS)
Wernicke’s area

Questions for Thought

The Web site for this book contains a set of questions
(see Learning Objectives and also FAQ) for each chap-
ter. These can serve as a useful basis for reflection—for
executing the reflection phase of the PQ4R method
discussed early in the chapter. The chapter itself will
also contain a set of questions designed to emphasize
the core issues in the field. For this chapter, consider
the following questions:

1. Research in cognitive psychology has been de-
scribed as “the mind studying itself.” Is this really
an accurate characterization of what cognitive
psychologists do in studies like those illustrated in
Figures 1.1 and 1.16? Does the fact that cognitive
psychologists study their own thought processes
create any special opportunities or challenges? Is
there any difference between a scientist studying a
mental system like memory versus a bodily system
like digestion?

2. Ray Kurzweil won the National Medal of Tech-
nology and Computation and is director of
engineering at Google. In his 2005 book, The Sin-
gularity Is Near, he predicted that by 2020 (5 years
from the publication of this edition of my text),
$1,000 will be able to buy a computer that can
emulate human intelligence. He projects further
development will lead to the Singularity in 2045,
when human life will be fundamentally trans-
formed. What do you think the growth of compu-
tation implies for your future life?

3. The scientific program of reductionism tries to
reduce one level of phenomena into a lower level.

For instance, this chapter has discussed how com-
plex economic behavior can be reduced to the de-
cision making (cognition) of individuals and how
this can be reduced to the actions of individual
neurons in the brain. But reductionism does not
stop here. The activity of neurons can be reduced
to chemistry, and chemistry can be reduced to
physics. When does it help and when does it not
help to try to understand one level in terms of a
lower level? Why is it silly to go all the way in a
reductionist program and attempt something like
explaining economic behavior in terms of particle
physics?

4. Humans are frequently viewed as qualitatively
superior to other animals in terms of their
intellectual function. What are some ways in
which humans seem to display such qualitative
superiority? How would these create problems in
generalizing research from other animals to
humans?

5. New techniques for imaging brain activity have
had a major impact on research in cognitive psy-
chology, but each technique has its limitations.
What are the limitations of the various techniques?
What would the properties be of an ideal brain-
imaging technique? How do studies that actually
go into the brain (almost exclusively done with
nonhumans) inform the use of brain imaging?

6. What are the ethical limitations on the kinds of
research that can be performed with humans and
nonhumans?

Anderson_8e_Ch01.indd 26 13/09/14 9:32 AM

27

2
Perception

Our bodies are bristling with sensors that detect sights, sounds, smells, and physical
contact. Billions of neurons process sensory information and deliver what they find

to the higher centers in the brain. This chapter will focus on visual perception and, to a
lesser extent, on the perception of speech—the two most important perceptual systems
for the human species. The chapter will address the following questions:

● How does the brain extract information from the visual signal?
● How is visual information organized into objects?
● How are visual and speech patterns recognized?
● How does context affect pattern recognition?

◆ Visual Perception in the Brain

Humans have a big neural investment in processing visual information. This is
illustrated in Figure 2.1, which shows the cortical regions devoted to processing
information from vision and hearing. This investment in vision is part of our
“inheritance” as primates, who have evolved to devote as much as 50% of their
brains to visual processing (Barton, 1998). The enormous investment underlies
the human ability to see the world.

This is vividly demonstrated by individuals with damage to certain brain
regions who are not blind but are unable to recognize anything visually, a
condition called visual agnosia. This condition results from neural damage.
One case of visual agnosia involved a soldier who suffered brain damage
resulting from accidental carbon monoxide poisoning. He could recognize
objects by their feel, smell, or sound, but he was unable to distinguish a picture
of a circle from that of a square or to recognize faces or letters (Benson &
Greenberg, 1969). On the other hand, he was able to discern light intensities
and colors and to tell in what direction an object was moving. Thus, his sen-
sory system was still able to register visual information, but the damage to his
brain resulted in a loss of the ability to transform visual information into per-
ceptual experience. This case shows that perception is much more than simply
the registering of sensory information.

Generally, visual agnosia is classified as either apperceptive agnosia or
associative agnosia (for a review, read Farah, 1990). Patients with apperceptive
agnosia, like the soldier just described, are unable to recognize simple shapes
such as circles or triangles, or to draw shapes they are shown. Patients with
associative agnosia, in contrast, are able to recognize simple shapes and can
successfully copy drawings, even of complex objects. However, they are unable to

Anderson_8e_Ch02.indd 27 13/09/14 9:36 AM

28 / Chapter 2 P e r C e P T i O n

recognize the complex objects. Figure 2.2 shows the original drawing of an anchor
and a copy of it made by a patient with associative agnosia (Ratcliff & Newcombe,
1982). Despite being able to produce a relatively accurate drawing, the patient could
not recognize this object as an anchor (he called it an umbrella). Patients with ap-
perceptive agnosia are generally believed to have problems with early processing of
information in the visual system. In contrast, patients with associative agnosia are
thought to have intact early processing but to have difficulties with pattern recogni-
tion, which occurs later. This chapter will first discuss the early processing of infor-
mation in the visual stream and then the later processing of this information.

Figure 2.3 offers an opportunity for a person with normal perception to
appreciate the distinction between early and late visual processing. If you have
not seen this image before, it will strike you as just a bunch of ink blobs. You
will be able to judge the size of the various blobs and reproduce them, just as
Ratcliff and Newcombe’s patient could, but you will not see any patterns. If
you keep looking at the image, however, you may be able to make out a cow’s
face (nose slightly to the left at the bottom). Now your pattern perception has
succeeded, and you have interpreted what you have seen.

■ Visual perception can be divided into an early phase, in which
shapes and objects are extracted from the visual scene, and a later
phase, in which the shapes and objects are recognized.

Early Visual Information Processing
Early visual information processing begins in the eye (see
Figure 2.4). Light passes through the lens and the vitre-
ous humor and falls on the retina at the back of the eye. The
retina contains the photoreceptor cells, which are made up
of light-sensitive molecules that undergo structural changes
when exposed to light. Light is scattered slightly in passing
through the vitreous humor, so the image that falls on the
back of the retina is not perfectly sharp. One of the functions
of early visual processing is to sharpen that image.

Photoreceptor cells in the retina contain light-sensitive
molecules that undergo structural changes when exposed to
light, initiating a photochemical process that converts light into
neural signals. There are two distinct types of photoreceptors in

FIGURE 2.2 A patient with asso-
ciative agnosia was able to copy
the original drawing of the anchor
at left (his drawing is at right), but
he was unable to recognize the
object as an anchor. (From Ellis
& Young, 1988. Human cognitive
neuropsychology. Copyright
© 1988 Erlbaum. Reprinted by
permission.)

FIGURE 2.1 Some of the cortical
structures involved in vision and
audition: the visual cortex, the
auditory cortex, the “where” visual
pathway, and the “what” visual
pathway.

“Where” visual pathway

“What” visual pathway

Auditory cortex

Visual cortex: Early visual processing

Brain Structures

Anderson_8e_Ch02.indd 28 13/09/14 9:36 AM

V i S u A l P e r C e P T i O n i n T H e B r A i n / 29

the eye: cones and rods. Cones are involved in color vision and produce high reso-
lution and acuity. Less light energy is required to trigger a response in the rods, but
they produce poorer resolution. As a consequence, they are principally responsible
for the less acute, black-and-white vision we experience at night. Cones are espe-
cially concentrated in a small area of the retina called the fovea. When we focus on
an object, we move our eyes so that the image of the object falls on the fovea, which
enables us to take full advantage of the high resolution of the cones in perceiving
the object. Foveal vision detects fine details, whereas the rest of the visual field—the
periphery—detects more global information, including movement.

The receptor cells synapse onto bipolar cells and these onto ganglion cells,
whose axons leave the eye and form the optic nerve, which goes to the brain.
Altogether there are about 800,000 ganglion cells in the optic nerve of each eye.
Each ganglion cell encodes information from a small region of the retina called
the cell’s receptive field. Typically, the amount of light stimulation in that region of
the retina is encoded by the neural firing rate on the ganglion cell’s axon.

FIGURE 2.3 A scene in which
we initially perceive just black
and white areas; only after
looking at it for some time is it
possible to make out the face of
a cow. (From American Journal of
Psychology. Copyright 1951 by the
Board of Trustees of the University
of Illinois. Used with permission
of the University of Illinois Press.
Adapted from Dallenbach, 1951.)

Cornea

Pupil

Lens

Vitreous humor

Retina

Optic nerve

Aqueous humor

Fovea

Iris

FIGURE 2.4 A schematic
representation of the eye.
light enters through the
cornea; passes through the
aqueous humor, pupil, lens,
and vitreous humor; then
strikes and stimulates the
retina. (After Lindsay & Norman,
1977.)

Anderson_8e_Ch02.indd 29 13/09/14 9:36 AM

30 / Chapter 2 P e r C e P T i O n

Figure 2.5 illustrates the neural pathways from
the eyes to the brain. The optic nerves from both eyes
meet at the optic chiasma, where the nerves from
the inside of the retina (the side nearest the nose)
cross over and go to the opposite side of the brain.
The nerves from the outside of the retina continue
to the same side of the brain as the eye. This means
that the right halves of both eyes are connected to the
right hemisphere. As Figure 2.5 illustrates, the lens
focuses the light so that the left side of the visual field
falls on the right half of each eye. Thus, information
about the left side of the visual field goes to the right
brain, and information about the right side of the vis-
ual field goes to the left brain. This is one instance of
the general fact, discussed in Chapter 1, that the left
hemisphere processes information about the right
part of the world and the right hemisphere processes
information about the left part.

Once inside the brain, the fibers from the
ganglion cells synapse onto cells in various subcorti-
cal structures. (“Subcortical” means that the struc-
tures are located below the cortex.) These subcortical
structures (such as the lateral geniculate nucleus in
Figure 2.5) are connected to the primary visual cortex
(Brodmann area 17 in Color Plate 1.1). The primary
visual cortex is the first cortical area to receive visual
input, but there are many other visual areas. Figure 2.6
illustrates the representation of the visual world in the

12

119

10

5

6

7

8

1
2

3
4

Fovea

Visual field

Right

Calcarine
fissure

Calcarine
fissure

Primary visual
cortex

Left

1

2

3

4
5

6

7

8
9

12 10
11

FIGURE 2.6 The orderly mapping
of the visual field (above) onto
the cortex. The upper fields are
mapped below the calcarine
fissure and the lower fields
are mapped above the fissure.
note the disproportionate
representation given to the fovea,
which is the region of greatest
visual acuity. (After Figure 29-7
in Kandel, E. R., Schwartz, J. H., &
Jessell, T. M. (1991). Principles of
neural science (3rd ed.). Copyright
© 1991 McGraw Hill. Reprinted by
permission.)

FIGURE 2.5 neural pathways
from the eye to the brain. The
optic nerves from each eye meet
at the optic chiasma. informa-
tion about the left side of the
visual field goes to the right brain,
and information about the right
side of the visual field goes to
the left brain. Optic nerve fibers
synapse onto cells in subcorti-
cal structures, such as the lateral
geniculate nucleus and superior
colliculus. Both structures are
connected to the visual cortex.

Eye

Optic
chiasm

Lateral
geniculate
nucleus

Superior
colliculus

Visual cortex

Anderson_8e_Ch02.indd 30 13/09/14 9:36 AM

V i S u A l P e r C e P T i O n i n T H e B r A i n / 31

primary visual cortex. It shows that the visual cortex is laid out topologically,
as discussed in Chapter 1. The fovea receives a disproportionate representation
while the peripheral areas receive less representation. Figure 2.6 shows that the
left visual field is represented in the right cortex and the right in the left cortex.
It also illustrates another “reversal” of the mapping—the upper part of the vis-
ual field is represented in the lower part of the visual cortex and the lower part
is represented in the upper region.

From the primary visual cortex, information tends to follow two path-
ways, a “what” pathway and a “where” pathway (look back at Figure 2.1).
The “what” pathway goes to regions of the temporal cortex that are specialized
for identifying objects. The “where” pathway goes to parietal regions of the brain
that are specialized for representing spatial information and for coordinating
vision with action. Monkeys with lesions in the “where” pathway have dif-
ficulty learning to identify specific locations, whereas monkeys with lesions
in the “what” pathway have difficulty learning to identify objects (Pohl, 1973;
Ungerleider & Brody, 1977). Other researchers (e.g., Milner & Goodale, 1995)
have argued that the “where” pathway is really a pathway specialized for action.
They point out that patients with agnosia because of damage to the temporal
lobe, but with intact parietal lobes, can often take actions appropriate to objects
they cannot recognize. For instance, one patient (see Goodale, Milner, Jakobson,
& Carey, 1991) could correctly reach out and grasp a door handle that she could
not recognize.

■ A photochemical process converts light energy into neural activity.
Visual information progresses by various neural tracks to the visual
cortex. From the visual cortex it progresses along “what” and “where”
pathways through the brain.

Information Coding in Visual Cells
Kuffler’s (1953) research showed how information is encoded by the ganglion
cells. These cells generally fire at some spontaneous rate even when the eyes are
not receiving any light. For some ganglion cells, if light falls on a small region of
the retina at the center of the cell’s receptive field, their spontaneous rates of firing
will increase. If light falls in the region just around this sensitive center, however,
the spontaneous rate of firing will decrease. Light farther from the center elicits
no change from the spontaneous firing rate—neither an increase nor a decrease.
Ganglion cells that respond in this way are known as on-off cells. There are
also off-on ganglion cells: Light at the center decreases the spontaneous rate of
firing, and light in the surrounding areas increases that rate. Cells in the lateral
geniculate nucleus respond in the same way. Figure 2.7 illustrates the receptive
fields of such cells (i.e., locations on the retina that increase or decrease the firing
rate of the cell).

Hubel and Wiesel (1962), in their study of the primary
visual cortex in the cat, found that visual cortical cells re-
spond in a more complex manner than ganglion cells and
cells in the lateral geniculate nucleus. Figure 2.8 illustrates
four patterns that have been observed in cortical cells. These
receptive fields all have an elongated shape, in contrast to the
circular receptive fields of the on-off and off-on cells. The
types shown in Figures 2.8a and 2.8b are edge detectors.
They respond positively to light on one side of a line and neg-
atively to light on the other side. They respond most if there
is an edge of light lined up so as to fall at the boundary point.
The types shown in Figures 2.8c and 2.8d are bar detectors.
They respond positively to light in the center and negatively

On-off cell Off-on cell

FIGURE 2.7 On-off and off-on
receptive fields of ganglion cells
and the cells in the lateral genicu-
late nucleus.

(a) (b) (c) (d)

FIGURE 2.8 response patterns
of cells in the visual cortex.
(a) and (b) are edge detectors,
responding positively to light on
one side of a line and negatively
to light on the other side.
(c) and (d) are bar detectors;
they respond positively to light in
the center and negatively to light
at the periphery, or vice versa.

Center-Surround
Receptive Fields and
Center-Surround
Illusions

Anderson_8e_Ch02.indd 31 13/09/14 9:36 AM

32 / Chapter 2 P e r C e P T i O n

to light at the periphery, or vice versa. Thus, a
bar with a positive center will respond most if
there is a bar of light just covering its center.
Figure 2.9 illustrates how a number of on-off
and off-on cells might combine to form a bar
or edge detector. Note that no single on-off
or off-on cell is sufficient to elicit a response
from a detector cell; instead, the detector cell
responds to patterns of input from the on-off
and off-on cells. Even at this low level, we see
the nervous system processing information
in terms of patterns of neural activation, a
theme emphasized in Chapter 1.

Both edge and bar detectors are spe-
cific with respect to position, orientation, and

width. That is, they respond only to stimulation in a small area of the visual
field, to bars and edges in a small range of orientations, and to bars and edges
of certain widths. Different detectors are tuned to different widths and orienta-
tions. Any bar or edge anywhere in the visual field, at any orientation, will elicit
a maximum response from some subset of detectors.

Figure 2.10 illustrates Hubel and Wiesel’s (1977) hypercolumn representa-
tion of cells in the primary visual cortex. They found that the visual cortex is
divided into 2 × 2 mm regions, which they called hypercolumns. Each hyper-
column represents a particular region of the visual field. As noted in Chapter 1,
the organization of the visual cortex is topographic, and so adjacent areas of the
visual field are represented in adjacent hypercolumns. Figure 2.10 shows that
each hypercolumn itself has a two-dimensional (2-D) organization. Along one
dimension, alternating rows receive input from the right and left eyes. Along
the other dimension, the cells vary in the orientation to which they are most
sensitive, with cells in adjacent rows representing similar orientations. This or-
ganization should impress upon us how much information is encoded about
the visual scene. Hundreds of regions of space are represented separately for
each eye, and within these regions many different orientations are represented.
In addition, different cells code for different sizes and widths of line (an aspect
of visual coding not illustrated in Figure 2.10). Thus, an enormous amount of
information has been extracted from the visual signal before it even leaves the
first cortical areas.

In addition to this rich representation of line orientation, size, and width,
the visual system extracts other information from the visual signal. For in-
stance, we can also perceive the colors of objects and whether they are moving.
Livingstone and Hubel (1988) proposed that the visual system processes these

various dimensions (form, color, and movement) separately.
Many different visual pathways and many different areas
of the cortex are devoted to visual processing (32 visual
areas in the count by Van Essen & DeYoe, 1995). Different
pathways have cells that are differentially sensitive to color,
movement, and orientation. Thus, the visual system ana-
lyzes a stimulus into many independent features in specific
locations. Such spatial representations of visual features are
called feature maps (Wolfe, 1994), with separate maps for
color, orientation, and movement. Thus, if a vertical red bar
is moving at a particular location, there are separate feature
maps representing that it is red, vertical, and moving in that
location. The maps for color, orientation, and movement
are separate.

R
L

R
L

FIGURE 2.10 representation of
a hypercolumn in the visual cor-
tex. The hypercolumn is organ-
ized in one dimension according
to whether input is coming from
the right eye or left eye. in the
other dimension, it is organized
according to the orientation of
lines to which the receptive cells
are most sensitive. Adjacent
regions represent similar orienta-
tions. (After Horton, 1984.)

(a) (b)

FIGURE 2.9 Hypothetical combi-
nations of on-off and off-on cells
to form (a) bar detectors and
(b) edge detectors.

Anderson_8e_Ch02.indd 32 13/09/14 9:36 AM

V i S u A l P e r C e P T i O n i n T H e B r A i n / 33

■ The ganglion cells encode the visual field by means of on-off and
off-on cells, which are combined by higher visual processing to form
various features.

Depth and Surface Perception
Even after the visual system has identified edges and bars in the environment,
a great deal of information processing must still be performed to enable vis-
ual perception of the world. Crucially, it is necessary to determine where
those edges and bars are located in space, in terms of their relative distance, or
depth. The fundamental problem is that the information laid out on the retina
is inherently 2-D, whereas we need to construct a three-dimensional (3-D)
representation of the world. The visual system uses a number of cues to infer
distance, including texture gradient, stereopsis, and motion parallax.

Texture gradient is the tendency of evenly spaced elements to appear more
closely packed together as the distance from the viewer increases. In the classic ex-
amples shown in Figure 2.11 (Gibson, 1950), the change in the texture gives the
appearance of distance even though the lines and ovals are rendered on a flat page.

Stereopsis is the ability to perceive 3-D depth based on the fact that each
eye receives a slightly different view of the world. The 3-D glasses used to view
some movies and some exhibits in theme parks achieve this by filtering the
light coming from a single 2-D source (say, a movie screen) so that different
light information reaches each eye. The perception of a 3-D structure resulting
from stereopsis can be quite compelling.

Motion parallax provides information about 3-D structure when a per-
son and/or the objects in a scene are in motion: The images of distant objects
will move across the retina more slowly than the images of closer objects. For
an interesting demonstration, look at a nearby tree with one eye closed and
without moving your head. Denied stereoscopic information, you will have
the sense of a very flat image in which it is hard to see the relative depths of
the leaves and branches. But if you move your head, the 3-D structure of the

FIGURE 2.11 examples of texture gradient. elements appear to be further away when
they are more closely packed together. (From Gibson, J. J. (1950). The perception of the
visual world. © 1950 Wadsworth, a part of Cengage Learning, Inc. Reproduced by permission.)

Size Constancy

Anderson_8e_Ch02.indd 33 13/09/14 9:36 AM

34 / Chapter 2 P e r C e P T i O n

tree will suddenly become clear, because the images of
nearby leaves and branches will move across the im-
ages of more distant ones, providing clear information
about depth.

Although it is easy to demonstrate the importance
to depth perception of such cues as texture gradient,
stereopsis, and motion parallax, it has been a chal-
lenge to understand how the brain actually processes
such information. A number of researchers in the area
of computational vision have worked on the problem.
For instance, David Marr (1982) has been influential in
his proposal that these various sources of information
work together to create what he calls a 2½-D sketch
that identifies where various visual features are located
relative to the viewer. While it required a lot of informa-
tion processing to produce this 2½-D sketch, a lot more
is required to convert that sketch into actual perception
of the world. In particular, such a sketch represents only
parts of surfaces and does not yet identify how these

parts go together to form images of objects in the environment (the problem we
had with Figure 2.3). Marr used the term 3-D model to refer to a later represen-
tation of objects in a visual scene.

■ Cues such as texture gradient, stereopsis, and motion parallax com-
bine to create a representation of the locations of surfaces in 3-D space.

Object Perception

A major problem in constructing a representation of the world is object seg-
mentation. Knowing where the lines and bars are located in space is not
enough; we need to know which ones go together to form objects. Consider
the scene in Figure 2.12: Many lines go this way and that, but somehow we put
them together to come up with the perception of a set of objects.

We organize objects into units according to a set of principles called the
gestalt principles of organization, after the Gestalt psychologists who first
proposed them (e.g., Wertheimer, 1912/1932). Consider Figure 2.13:

● Figure 2.13a illustrates the principle of
proximity: Elements close together tend to
organize into units. Thus, we perceive four
pairs of lines rather than eight separate lines.

● Figure 2.13b illustrates the principle of
similarity: Objects that look alike tend to be
grouped together. In this case, we tend to
see this array as rows of o’s alternating with
rows of x’s.

● Figure 2.13c illustrates the principle of good

FIGURE 2.12 An example of
how we aggregate the perception
of many broken lines into the
perception of solid objects. (From
Winston, P. H. (1970). learning
structural descriptions from
examples (Tech. Rep. No. 231).
Copyright © 1970 Massachusetts
Institute of Technology. Reprinted
by permission.)

(a)

A

BC

D

(c) (d)

(b)

FIGURE 2.13 illustrations of the
gestalt principles of organization:
(a) the principle of proximity,
(b) the principle of similarity,
(c) the principle of good con-
tinuation, (d) the principle of
closure.

continuation. We perceive two lines, one from
A to B and the other from C to D, although
there is no reason why this sketch could not
represent another pair of lines, one from A
to D and the other from C to B. However, the
lines from A to B and from C to D display bet-
ter continuation than the lines from A to D
and from C to B, which have a sharp turn.

Anderson_8e_Ch02.indd 34 13/09/14 9:36 AM

V i S u A l PAT T e r n r e C O g n i T i O n / 35

● Figure 2.13d illustrates the principles of closure and good form. We see the
drawing as one circle occluded by another, although the occluded object
could have many other possible shapes. The principle of closure means that
we see the large arc as part of a complete shape, not just as the curved line.
The principle of good form means that we perceive the occluded part as a
circle, not as having a wiggly, jagged, or broken border.

These principles will organize completely novel stimuli into units. Palmer
(1977) studied the recognition of shapes such as the ones shown in Figure 2.14.
He first showed participants stimuli (e.g., Figure 2.14a) and then asked them to
decide whether the fragments depicted in Figures 2.14b through 2.14e were part of
the original figure. The stimulus in Figure 2.14a tends to organize itself into a triangle
(principle of closure) and a bent letter n (principle of good continuation). Palmer
found that participants could recognize the parts most rapidly when they were the
segments predicted by the gestalt principles. So the stimuli in Figures 2.14b and
2.14c were recognized more rapidly than those in Figures 2.14d and 2.14e. Thus,
we see that recognition depends critically on the initial segmentation of the figure.
Recognition can be impaired when this gestalt-based segmentation contradicts the
actual pattern structure. FoRiNsTaNcEtHiSsEnTeNcEiShArDtOrEaD. The reasons
for this difficulty are (a) that the gestalt principle of similarity makes it hard to
perceive adjacent letters of different case as units and (b) that removing the spaces
between words has eliminated the proximity cues.

These ideas about segmentation can be extended to describe how more
complex 3-D structures are divided. Figure 2.15 illustrates a proposal by Hoff-
man and Richards (1985) for how gestaltlike principles can be used to seg-
ment an outline representation of an object into subobjects. They observed
that where one segment joins another, there is typically a concavity in the line
outline. Basically, people exploit the gestalt principle of good continuation: The
lines at the points of concavity are not good continuations of one another, and
so viewers do not group these parts together.

The current view is that the visual processing underlying the ability
to identify the position and shape of an object in 3-D space is largely innate.
Young infants appear to be capable of recognizing objects and their shapes and
where they are in 3-D space (e.g., Granrud, 1986, 1987).

■ Gestalt principles of organization explain how the brain segments
visual scenes into objects.

◆ Visual Pattern Recognition

We have now discussed visual information processing to the point
where we organize the visual world into objects. There still is a
major step before we see the world, however: We also must identify
what these objects are. This task is called pattern recognition. Much
of the research on this topic has focused on the question of how we
recognize the identity of letters. For instance, how do we recognize
a presentation of the letter A as an instance of the pattern A? We

(a) (b) (c) (d) (e)

FIGURE 2.15 Segmentation
of an object into subobjects.
The part boundary (dashed
line) can be identified with a
contour that follows points of
maximum concave curvature.
(From Stillings, N. A., Feinstein,
M. H., Garfield, J. L., Rissland, E. L.,
Rosenbaum, D. A., et al. (1987).
Cognitive Science: An introduction
(figure 12.17, page 495). Copyright
© 1987 Massachusetts Institute of
Technology, by permission of The
MIT Press.)

FIGURE 2.14 examples of stimuli
used by Palmer (1977) for study-
ing segmentation of novel figures.
(a) is the original stimulus that
participants saw; (b) through
(e) are the subparts of the
stimulus presented for recognition.
Stimuli shown in (b) and (c) were
recognized more rapidly than
those shown in (d) and (e).

Gestalt Psychology
Grouping Experiment

Anderson_8e_Ch02.indd 35 13/09/14 9:36 AM

36 / Chapter 2 P e r C e P T i O n

will first discuss pattern recognition with respect to letter identification and then
move on to a more general discussion of object recognition.

Template-Matching Models
Perhaps the most obvious way to recognize a pattern is by means of template
matching. The template-matching theory of perception proposes that a retinal
image of an object is faithfully transmitted to the brain, and the brain attempts
to compare the image directly to various stored patterns, called templates. The
basic idea is that the perceptual system tries to compare the image of a letter
to the templates it has for each letter and then reports the template that gives
the best match. Figure 2.16 illustrates various examples of successful and un-
successful template matching. In each case, an attempt is made to achieve a cor-
respondence between the retinal cells stimulated and the retinal cells specified
for a template pattern for a letter.

Figure 2.16a shows a case in which a correspondence is achieved and an A
is recognized. Figure 2.16b shows a case in which no correspondence is reached
between the input of an L and the template pattern for an A. But L is matched
in Figure 2.16c by the L template. However, things can very easily go wrong
with a template. Figure 2.16d shows a mismatch that occurs when the image
falls on the wrong part of the retina, and Figure 2.16e shows the problem when
the image is the wrong size. Figure 2.16f shows what happens when the image
is in a wrong orientation, and Figures 2.16g and 2.16h show the difficulty when
the images are nonstandard A’s.

Although there are these difficulties with template matching, it is one of the
methods used in machine vision (see Ullman, 1996), where procedures have been
developed for rotating, stretching, and otherwise modifying images to match.
Template matching is also used in fMRI brain imaging (see Chapter 1). Each
human brain is anatomically different, much as each human body is different.
When researchers claim regions like those in Figure 1.15 display activation pat-
terns like those in Figure 1.16, they typically are claiming that the same region
in the brains of each of their participants displayed that pattern. To determine
that it is the same region, they map the individual brains to a reference brain by
a sophisticated computer-based 3-D template-matching procedure. Although
template matching has enjoyed some success, there seem to be limitations to

FIGURE 2.16 examples of
attempts to match templates
to the letters A and L. The
little circles on the “input”
patterns represent the cells
actually stimulated on the
retina by a presentation
of the letter A or L, and
the little circles on the
“Template” patterns are the
retinal cells specified by a
template pattern for a letter.
(a) and (c) are successful
template-matching attempts;
(b) and (d) through (h) are
failed attempts.

TemplateTemplate

Input

Template

Input Input

Template

Input

Template

Input

(a) (b) (c) (d)

(e) (f) (g) (h)

Template

Input

Template

Input

Template

Input

Anderson_8e_Ch02.indd 36 13/09/14 9:36 AM

V i S u A l PAT T e r n r e C O g n i T i O n / 37

the abilities of computers to use template matching to recognize patterns, as
suggested in this chapter’s Implications Box on CAPTCHAs.

■ Template matching is a way to identify objects by aligning the stim-
ulus to a template of a pattern.

Feature Analysis
Partly because of the difficulties posed by template matching, psychologists
have proposed that pattern recognition occurs through feature analysis. In this
model, stimuli are thought of as combinations of elemental features. Table 2.1
from Gibson (1969) shows her proposal for the representation of the letters of
the alphabet in terms of features. For instance, the capital letter A can be seen
as consisting of a horizontal, two diagonals in opposite orientations, a line in-
tersection, symmetry, and a feature she called vertical discontinuity. So, some

Separating humans from BOTs

The special nature of human visual per-
ception motivated the development of
CAPTCHAs (Von Ahn, Blum, & langford,
2002). CAPTCHA stands for “Completely
Automated Public Turing test to tell Com-
puters and Humans Apart.” The motiva-
tion for CAPTCHAs comes from
real-world problems such as those faced
by YAHOO!, which offers free email ac-
counts. The problem is that automatic
BOTs will sign up for such accounts and
then use them to send SPAM. To test that
it is a real human, the system can present
images like those in Figure 2.17. use of
such CAPTCHAs is quite common on the
internet. Although template-based ap-
proaches may fail on recognizing such
figures, more sophisticated feature-based
character recognition algorithms have had
a fair degree of success (e.g., Mori &
Malik, 2003). This has led to more and
more difficult CAPTCHAs being used,
which unfortunately humans also have
great difficulty in decoding (Bursztein,
Bethard, Fabry, Mitchell, & Jurafsky,
2010). You can visit the CAPTCHA Web
site and contribute to the research at
http://w w w.captcha.net/.

I m p l I c a t I o n s
▼

▲

FIGURE 2.17 examples of CAPTCHAs that
humans can read but template-based com-
puter programs have great difficulty with.
(Staff KRT/Newscom.)

Anderson_8e_Ch02.indd 37 13/09/14 9:36 AM

http://w

38 / Chapter 2 P e r C e P T i O n

of these features, like the straight lines, can be thought of as outputs of the edge
and bar detectors in the visual cortex (see Figure 2.8).

You might wonder how feature analysis represents an advance beyond the
template model. After all, what are the features but minitemplates? The feature-
analysis model does have a number of advantages over the template model, how-
ever. First, because the features are simpler, it is easier to see how the system
might try to correct for the kinds of difficulties faced by the template-matching
model in recognizing full patterns as in Figure 2.16. Indeed, to the extent that
features are just line strokes, the bar and edge detectors we discussed earlier can
extract such features. Second, feature analysis makes it possible to specify those
relationships among the features that are most important to the pattern. For ex-
ample, in the case of the letter A, the critical point is that there are three lines
that intersect, two diagonals (in different directions) and one horizontal. Many
other details are unimportant. Thus, all the following patterns are A’s:
Finally, the use of features rather than larger patterns reduces the number of tem-
plates needed. In the feature-analysis model, we would not need a template for
each possible pattern but only for each feature. Because the same features tend to
occur in many patterns, the number of distinct entities to be represented would
be reduced considerably. Feature-based recognition is used in most modern
machine-based systems for character recognition such as those used on tablets
and smart phones. However, the features used by these machine-based systems
are often quite different than the features used by humans (Impedovo, 2013).

There is a fair amount of behavioral evidence for the existence of fea-
tures as components in pattern recognition. For instance, if letters have
many features in common—as C and G do, for example—evidence suggests
that people are particularly prone to confuse them (Kinney, Marsetta, &

TABLE 2.1 gibson’s Proposal for the Features underlying the recognition of letters

Features A E F H I L T K M N V W X Y Z B C D G J O P R Q S U

Straight

Horizontal + + + + + + + +

Vertical + + + + + + + + + + + + + +

Diagonal/ + + + + + + + + + + +

Diagonal\ + + + + + + + + + +

Curve

Closed + + + + + +

Open V + +

Open H + + + +

Intersection + + + + + + + +

Redundancy + + +

Cyclic
change

+ + + + +

Symmetry + + + + + + + + + + + + + + + +

Discontinuity

Vertical + + + + + + + + + + +

Horizontal + + + + +

+ indicates features for a particular letter.

Anderson_8e_Ch02.indd 38 13/09/14 9:36 AM

V i S u A l PAT T e r n r e C O g n i T i O n / 39

Showman, 1966). When such letters are presented for very brief intervals,
people often misclassify one stimulus as the other. So, for instance, partici-
pants in the Kinney et al. experiment made 29 errors when presented with
the letter G. Of these errors, there were 21 misclassifications as C, 6 misclas-
sifications as O, 1 misclassification as B, and 1 misclassification as 9. No other
errors occurred. It is clear that participants were choosing items with similar
feature sets as their responses. Such a response pattern is what we would ex-
pect if participants were using features as the basis for recognition. If partici-
pants could extract only some of the features in the brief presentation, they
would not be able to decide among stimuli that shared these features.

Another kind of experiment that yields evidence in favor of a feature-anal-
ysis model involves stabilized images. The eye has a very slight tremor, called
psychological nystagmus, which occurs at the rate of 30 to 70 cycles per second.
Also, the eye’s direction of gaze drifts slowly over an object. Consequently, the
retinal image of the object on which a person tries to focus is not perfectly con-
stant; its position changes slightly over time. This retinal movement is critical
for perception. When techniques are used to keep an image on the exact same
position of the retina regardless of eye movement, parts of the object start to
disappear from our perception. If the exact same retinal and nervous pathways
are used constantly, they become fatigued and stop responding.

The most interesting aspect of this phenomenon is the way the stabilized
object disappears. It does not simply fade away or vanish all at once. Instead,
different portions drop out over time. Figure 2.18 illustrates the fate of one of
the stimuli used in an experiment by Pritchard (1961). The leftmost item is
the image that was presented; the four others are various fragments that were
reported after the original image started to disappear. Two points are impor-
tant. First, whole features such as a vertical bar seemed to be lost. This find-
ing suggests that features are the important units in perception. Second, the
stimuli that remained tended to constitute complete letter or number patterns,
indicating that the remaining features are combined into recognizable patterns.
Thus, even though our perceptual system may extract features, what we actually
perceive are patterns composed from these features. The feature-extraction and
feature-combination processes that underlie pattern recognition are not avail-
able to conscious awareness; all that we are aware of are the resulting patterns.

■ Feature analysis involves recognizing first the separate features
that make up a pattern and then their combination.

Object Recognition
Feature analysis does a satisfactory job of describing how we recognize such
simple objects as the letter A, but can it explain our recognition of more com-
plex objects that might seem to defy description in terms of a few features?
There is evidence that similar processes might underlie the recognition of
familiar categories of objects such as horses or cups. The basic idea is that a
familiar object can be seen as a known configuration of simple components.
Figure 2.19 illustrates a proposal by Marr (1982) about how familiar objects can

FIGURE 2.18 The disintegration
of an image that is stabilized on
the eye. At far left is the original
image displayed. The partial
outlines to the right show various
patterns reported as the stabi-
lized image began to disappear.
(From Pritchard, 1961. Reprinted
by permission of the publisher.
© 1961 by Scientific American.)

Anderson_8e_Ch02.indd 39 13/09/14 9:36 AM

40 / Chapter 2 P e r C e P T i O n

be seen as configurations of simple pipelike components. For instance, an os-
trich has a horizontally oriented torso attached to two long legs and a long neck.

Biederman (1987) put forward the recognition-by-components theory.
It proposes that there are three stages in our recognition of an object as a
configuration of simpler components:

1. The object is segmented into a set of basic subobjects via a process that re-
flects the output of early visual processing, discussed earlier in this chapter.

2. Once an object has been segmented into basic subobjects, one can classify
the category of each subobject. Biederman (1987) suggested that there are
36 basic categories of subobjects, which he called geons (an abbreviation
of geometric ions). Figure 2.20 shows some examples. We can think of the
cylinder as being created by a circle as it is moved along a straight line (the
axis) perpendicular to its center. Other shapes can be created by varying the
generation process. We can change the shape of the object we are moving. If

Horse

Giraffe Ape Dove

Human Ostrich

FIGURE 2.19 Segmentation
of some familiar objects into
basic cylindrical shapes. Familiar
objects can be recognized as
configurations of simpler compo-
nents. (After Marr & Nishihara,
1978. © 1978 by the Royal
Society of London. Reprinted by
permission.)

Cone PyramidCylinder

HornFootball Wine glass

FIGURE 2.20 examples of Bieder-
man’s (1987) proposed geons,
or basic categories of subobjects.
in each object, the dashed line
represents the central axis of
the object. The objects can be
described in terms of the move-
ment of a cross-sectional shape
along an axis. Cylinder: A circle
moves along a straight axis. Cone:
A circle contracts as it moves
along a straight axis. Pyramid: A
square contracts as it moves along
a straight axis. Football: A circle
expands and then contracts as it
moves along a straight axis. Horn:
A circle contracts as it moves
along a curved axis. Wine glass: A
circle contracts and then expands,
creating concave segmentation
points, marked by arrows.

it is a rectangle rather than a circle that is moved
along the axis, we get a block instead of a cylin-
der. We can curve the axis and get objects that
curve. We can vary the size of the shape as we
are moving it and get objects like the pyramid
or wine glass. Biederman proposed that the 36
geons that can be generated in this manner serve
as an alphabet for composing objects, much
as letters serve as the alphabet for building up
words. Recognizing a geon involves recogniz-
ing the features that define it, which describe
elements of its generation such as the shape of
the object and the axis along which it is moved.
Thus, recognizing a geon from its features is like
recognizing a letter from its features.

3. Having identified the pieces from which the
object is composed and their configuration, one

Anderson_8e_Ch02.indd 40 13/09/14 9:36 AM

V i S u A l PAT T e r n r e C O g n i T i O n / 41

recognizes the object as the pattern formed by these pieces. Thus, recognizing
an object from its components is like recognizing a word from its letters.

As in the case of letter recognition, there are many small variations in the
underlying geons that should not be critical for recognition. For example, one
need only determine whether an edge is straight or curved (in discriminating,
say, a brick from a cylinder) or whether edges are parallel or not (in discrimi-
nating, say, a cylinder from a cone). It is not necessary to determine precisely
how curved an edge might be. Only very basic characteristics of edges are
needed to define geons. Color, texture, and small detail should not matter. If
this hypothesis is correct, schematic line drawings of complex objects that allow
the basic geons to be identified should be recognized just as quickly as detailed
color photographs of the objects. Biederman and Ju (1988) confirmed this hy-
pothesis experimentally: Schematic line drawings of such objects as telephones
provide all the information needed for quick and accurate recognition.

The crucial assumption in this theory is that object recognition is mediated
by the recognition of its components. Biederman, Beiring, Ju, and Blickle (1985)
performed a test of this prediction with objects
such as those shown in Figure 2.21. They pre-
sented these two types of degraded figures to
participants for various brief intervals and asked
them to identify the objects. In one type, whole
components of some objects were deleted; in
the other type, all the components were present,
but segments of the components were deleted.
Figure 2.22 shows that at very brief presenta-
tion times (65–100 ms), participants recognized
figures with component deletion more accu-
rately than figures with segment deletion, but
the opposite was true for the longer, 200-ms
presentation. Biederman et al. reasoned that at
the very brief intervals, participants were not
able to identify the components with segment

Complete Component
deletion

Midsegment
deletion

FIGURE 2.21 Sample stimuli
used by Biederman et al. (1985)
to test the theory that object
recognition is mediated by
recognition of components of the
object. equivalent proportions
either of whole components
or of contours at midsegments
were removed. results of
the experiment are shown in
Figure 2.22. (Adapted from
Biederman, I. (1987). Recognition-
by-components: A theory of
human image understanding.
Psychological review, 94, 115–147.
Copyright © 1987 American
Psychological Association. Adapted
by permission.)

FIGURE 2.22 results from the
test conducted by Biederman,
Beiring, Ju, and Blickle (1985) to
determine whether object recogni-
tion is mediated by recognition of
components of the object. Mean
percentage of errors of object
naming is plotted as a function
of the type of contour removal
(deletion of midsegments or
of entire components) and of
exposure duration. (Data from
Biederman, 1987.)

Midsegment deletion

Component deletion

40

30

20

10

10065 200

Exposure duration (ms)

M
ea

n
(%

) e
rro

r

Contour Deletion

Anderson_8e_Ch02.indd 41 13/09/14 9:36 AM

42 / Chapter 2 P e r C e P T i O n

deletion and so had difficulty in recognizing the objects. With 200 ms of exposure,
however, participants were able to recognize all the components in either condi-
tion. Because there were more components in the condition with segment dele-
tion, they had more information about object identity.

■ Complex objects are recognized as configurations of a set of subob-
jects defined by simple features.

Face Recognition
Faces make up one of the most important categories of visual stimuli, and
some evidence suggests that we have special mechanisms to recognize some-
one’s face. Special cells that respond preferentially to the faces of other mon-
keys have been found in the temporal lobes of monkeys (Baylis, Rolls, &
Leonard, 1985; Rolls, 1992). Damage to the temporal lobe in humans can
result in a deficit called prosopagnosia, in which people have selective
difficulties in recognizing faces. Brain-imaging studies using fMRI have
found a particular region of the temporal lobe, called the fusiform gyrus, that
responds when faces are present in the visual field (e.g., Ishai, Ungerleider,
Martin, Maisog, & Haxby, 1997; Kanwisher, McDermott, & Chun, 1997;
McCarthy, Puce, Gore, & Allison, 1997).

Other evidence that the processing of faces is special comes from re-
search that examined the recognition of faces turned upside down. In one of
the original studies, Yin (1969) found that people are much better at recogniz-
ing faces when the faces are presented in their upright orientation than they
are at recognizing other categories of objects, such as houses, presented in the
upright orientation. When a face is presented upside down, however, there is
a dramatic decrease in its recognition; this is not true of other objects. Thus,
it appears that we are specially attuned to recognizing faces. Studies have also
found somewhat reduced fMRI response in the fusiform gyrus when inverted
faces are presented (Haxby et al., 1999; Kanwisher, Tong, & Nakayama, 1998).
In addition, we are much better at recognizing parts of a face (a nose, say)
when it is presented in context, whereas recognizing parts of a house (for ex-
ample, a window) is not as context dependent (Tanaka & Farah, 1993). All this
evidence leads some researchers to think that we are specifically predisposed
to identify whole faces, and it is sometimes argued that this special capability
was acquired through evolution.

Other research questions whether the fusiform gyrus is specialized for just face
recognition and presents evidence that it is involved in making fine-grained dis-
tinctions generally. For instance, Gauthier, Skudlarski, Gore, and Anderson (2000)
found that bird experts or car experts showed high activation in the fusiform gyrus
when they made judgments about birds or cars. In another study, people given a lot
of practice at recognizing a set of unfamiliar objects called greebles (Figure 2.23)
showed activation in the fusiform gyrus. Studies like these support the idea that,
because of our great familiarity with faces, we are good at making such fine-grained
judgments in recognizing them, but similar effects can be found with other stimuli
with which we have had a lot of experience.

There have been rapid improvements in
face-recognition software, as most users of
Facebook are aware. In some circumstances
this software outperforms humans. This has
brought up concerns about privacy (see the
60 Minutes episode “A Face in the Crowd: Say
Goodbye to Anonymity,” which is available
online). Interestingly, these systems are quite

FIGURE 2.23 “greeble experts”
use the face area when recogniz-
ing these objects. (From Gauthier,
Tarr, Anderson, Skudlarski, & Gore,
1999. Reprinted by permission
from Macmillan Publishers Ltd.,
© 1999.)

Inversion Effect

Anderson_8e_Ch02.indd 42 13/09/14 9:36 AM

S P e e C H r e C O g n i T i O n / 43

specialized to perform only face recognition. So even though humans may not
have a specialized system for face recognition, modern computer applications do.

■ The fusiform gyrus, located in the temporal lobe, becomes active
when people recognize faces.

◆ Speech Recognition

Up to this point, we have considered only visual pattern recognition. An inter-
esting test of the generality of our conclusions is whether they extend to speech
recognition. Although we will not discuss the details of early speech processing,
it is worth noting that similar issues arise, especially the issue of segmentation.
Speech is not broken into discrete units the way printed text is. Although well-
defined gaps between words seem to exist in speech, these gaps are often an
illusion. If we examine the actual physical speech signal, we often find undi-
minished sound energy at word boundaries. Indeed, gaps in sound energy are
as likely to occur within a word as between words. This property of speech is
particularly compelling when we listen to someone speaking an unfamiliar for-
eign language. The speech appears to be a continuous stream of sounds with no
obvious word boundaries. It is our familiarity with our own language that leads
to the illusion of word boundaries.

Within a single word, even greater segmentation problems exist. These
intraword problems involve the identification of phonemes. Phonemes are
the basic units for speech recognition.1 A phoneme is defined as the minimal
unit of speech that can result in a difference in the spoken message. To
illustrate, consider the word bat. This word is composed of three phonemes:
/b/, /a/, and /t/. Replacing /b/ with the phoneme /p/, we get pat; replacing
/a/ with /i/ we get bit; replacing /t/ with /n/, we get ban. Obviously, a
one-to-one correspondence does not always exist between letters and
phonemes. For example, the word one consists of the phonemes /w/, /e/, and
/n/; school consists of the phonemes /s/, /k/, /ú/, and /l/; and knight consists of
/n/, /ī /, and /t/. It is the lack of perfect letter-to-phoneme correspondence that
makes English spelling so difficult.

A segmentation problem arises when the phonemes composing a spoken
word need to be identified. The difficulty is that speech is continuous, and pho-
nemes are not discrete in the way letters are on a printed page. Segmentation at
this level is like recognizing a written (not printed) message, where one letter
runs into another. Also, as in the case of writing, different speakers vary in the
way they produce the same phonemes. The variation among speakers is dra-
matically clear, for instance, when a person first tries to understand a speaker
with a strong and unfamiliar accent. Examination of the speech signal, however,
will reveal that even among speakers with the same accent, considerable vari-
ation exists. For instance, the voices of women and children normally have a
much higher pitch than those of men.

A further difficulty in speech perception involves a phenomenon known as
coarticulation (Liberman, 1970). As the vocal tract is producing one sound—
say, the /b/ in bag—it is moving toward the shape it needs for the /a/. As it is
saying the /a/, it is moving to produce the /g/. In effect, the various phonemes
overlap. This means additional difficulties in segmenting phonemes, and it also
means that the actual sound produced for one phoneme will be determined by
the context of the other phonemes.

1 Massaro (1996) presents an often proposed alternative that the basic perceptual units are consonant-
vowel and vowel-consonant combinations.

Anderson_8e_Ch02.indd 43 13/09/14 9:36 AM

44 / Chapter 2 P e r C e P T i O n

Speech perception poses information-processing demands that are in many
ways greater than what is involved in other kinds of auditory perception. Research-
ers have identified a number of patients who have lost just the ability to recognize
speech, as a result of injury to the left temporal lobe (see M. N. Goldstein, 1974,
for a review). Their ability to detect and recognize other sounds and to speak is in-
tact. Thus, their deficit is specific to speech perception. Occasionally, such patients
have some success if the speech they are trying to hear is very slow (e.g., Okada,
Hanada, Hattori, & Shoyama, 1963), which suggests that some of the problem
might lie in segmenting the speech stream.

■ Speech recognition involves segmenting phonemes from the contin-
uous speech stream.

Feature Analysis of Speech
Feature-analysis and feature-combination processes seem to underlie speech
perception, much as they do visual recognition. As with individual letters, in-
dividual phonemes can be analyzed into a number of features. These features
refer to aspects of how the phoneme is generated. Among the features of pho-
nemes are the consonantal feature, voicing, and the place of articulation
(Chomsky & Halle, 1968). The consonantal feature is the consonant-like qual-
ity of a phoneme (in contrast to a vowel-like quality). Voicing is a feature of
phonemes produced by vibration of the vocal cords. For example, the phoneme
/z/ in the word zip has voicing, whereas the phoneme /s/ in the word sip does
not. You can detect this difference between /z/ and /s/ by placing your fingers
on your larynx as you generate the buzzing sound zzzz versus the hissing sound
ssss. You will feel the vibration of your larynx for zzzz but not for ssss.

Place of articulation refers to the location at which the vocal tract is
closed or constricted in the production of a phoneme. (It is closed at some
point in the utterance of most consonants.) For instance, /p/, /m/, and /w/ are
considered bilabial because the lips are closed while they are being generated.
The phonemes /f/ and /v/ are considered labiodental because the bottom lip
is pressed against the front teeth. Two different phonemes are represented by
/th/—one in thy and the other in thigh. Both are dental because the tongue
presses against the teeth. The phonemes /t/, /d/, /s/, /z/, /n/, /l/, and /r/ are all
alveolar because the tongue presses against the alveolar ridge of the gums just
behind the upper front teeth. The phonemes /sh/, /ch/, /j/, and /y/ are all pal-
atal because the tongue presses against the roof of the mouth just behind the
alveolar ridge. The phonemes /k/ and /g/ are velar because the tongue presses
against the soft palate, or velum, in the rear roof of the mouth.

Consider the phonemes /p/, /b/, /t/, and /d/. All share the feature of being
consonants. The four can be distinguished, however, by voicing and place of
articulation. Table 2.2 classifies these four phonemes according to these two

features.
Considerable evidence exists for the role of such features in

speech perception. For instance, Miller and Nicely (1955) had par-
ticipants try to recognize phonemes such as /b/, /d/, /p/, and /t/
when they were presented in noise.2 Participants exhibited confu-
sion, thinking they had heard one sound in the noise when in re-
ality another sound had been presented. The experimenters were
interested in which sounds participants would confuse with which
other sounds. It seemed likely that they would most often confuse

2 Actually, participants were presented with the sounds ba, da, pa, and ta.

Place of
Articulation

Voicing

Voiced Unvoiced

Bilabial

Alveolar

/b/ /p/

/d/ /t/

TABLE 2.2 The Classification of /b/, /p/,
/d/, and /t/ According to Voicing and Place
of Articulation

Anderson_8e_Ch02.indd 44 13/09/14 9:36 AM

C AT e g O r i C A l P e r C e P T i O n / 45

consonants that were distinguished by just a single feature, and this prediction
was confirmed. To illustrate, when presented with /p/, participants more often
thought that they had heard /t/ than that they had heard /d/. The phoneme /t/
differs from /p/ only in place of articulation, whereas /d/ differs both in place of
articulation and in voicing. Similarly, participants presented with /b/ more often
thought they heard /p/ than /t/.

This experiment is an earlier demonstration of the kind of logic we saw in
the Kinney et al. (1966) study on letter recognition. When the participant could
identify only a subset of the features underlying a pattern (in this case, the pat-
tern is a phoneme), the participant’s responses reflected confusion among the
phonemes sharing the same subset of features.

■ Phonemes are recognized in terms of features involved in their pro-
duction, such as place of articulation and voicing.

◆ Categorical Perception

The features of phonemes result from the ways in which they are articulated.
What properties of the acoustic stimulus encode these articulatory features?
This issue has been particularly well researched in the case of voicing. In the
pronunciation of such consonants as /b/ and /p/, two things happen: The closed
lips open, releasing air, and the vocal cords begin to vibrate (voicing). In the
case of the voiced consonant /b/, the release of air and the vibration of the vocal
cords are nearly simultaneous. In the case of the unvoiced consonant /p/, the
release occurs 60 ms before the vibration begins. What we are detecting when
we perceive a voiced versus an unvoiced consonant is the presence or absence
of a 60-ms interval between release and voicing. This period of time is referred
to as the voice-onset time. The difference between /p/ and /b/ is illustrated in
Figure 2.24. Similar differences exist in other voiced-unvoiced pairs, such as /d/
and /t/. Again, the factor controlling the perception of a phoneme is the delay
between the release of air and the vibration of the vocal cords.

Lisker and Abramson (1970) performed experiments with artificial
(computer-generated) stimuli in which the delay between the release of air and
the onset of voicing was varied from –150 ms (voicing occurred 150 ms before
release) to +150 ms (voicing occurred 150 ms after release). The participant’s
task was to identify which sounds were /b/’s and which were /p/’s. Figure 2.25
plots the percentage of /b/ identifications and /p/ identifications. Throughout
most of the continuum, participants agreed 100% on what they heard, but there
was a sharp switch from /b/ to /p/ at about 25 ms. At a 10-ms voice-onset time,
participants were in nearly unanimous agreement that the sound was a /b/; at
40 ms, they were in nearly unanimous agreement that the sound was a /p/. Be-
cause of this sharp boundary between the voiced and unvoiced phonemes, per-
ception of this feature is referred to as categorical. Categorical perception is

�100 0

Lips released

Time (ms)

Voicing

Voicing

/b/

/p/

+60 +100

FIGURE 2.24 The difference be-
tween the voiced consonant /b/
and the unvoiced consonant /p/
is the delay in the case of /p/
between the release of the lips
and the onset of voicing. (Data
from Clark & Clark, 1977.)

Anderson_8e_Ch02.indd 45 13/09/14 9:36 AM

46 / Chapter 2 P e r C e P T i O n

the perception of stimuli as belonging in distinct categories and the failure to
perceive the gradations among stimuli within a category.

Other evidence for categorical perception of speech comes from discrimi-
nation studies (see Studdert-Kennedy, 1976, for a review). People are very poor
at discriminating between a pair of /b/’s or a pair of /p/’s that differ in voice-
onset time but are on the same side of the phonemic boundary. However, they
are good at discriminating between pairs that have the same difference in voice-
onset time but one item of the pair is on the /b/ side of the boundary and the
other item is on the /p/ side. It seems that people can identify the phonemic
category of a sound but cannot discriminate sounds within that phonemic cat-
egory. Thus, people are able to discriminate two sounds only if they fall on dif-
ferent sides of a phonemic boundary.

There are at least two views of exactly what is meant by categorical per-
ception, which differ in the strength of their claims about the nature of percep-
tion. The weaker view is that we experience stimuli as coming from distinct
categories. There seems to be little dispute that the perception of phonemes
is categorical in this sense. A stronger viewpoint is that we cannot discrimi-
nate among stimuli within a category. Massaro (1992) has taken issue with
this viewpoint, and he has argued that there is some residual ability to dis-
criminate within categories. While there is discriminability within catego-
ries, it is typical to find that people can better make discriminations that cross
category boundaries (Goldstone & Hendrickson, 2010). Thus, there is increased
discriminability between categories (acquired distinctiveness) and decreased
discriminability within categories (acquired equivalence).

Another line of research that provides evidence for use of the voicing fea-
ture in speech recognition involves an adaptation paradigm. Eimas and Corbit
(1973) had their participants listen to repeated presentations of the sound da,
which involves the voiced consonant /d/. The experimenters reasoned that, if
there were a voicing detector, the constant repetition of the voiced consonant
might fatigue it so that it would require a stronger indication of voicing. They
presented participants with a series of artificial sounds that spanned the acous-
tic range across distinct categories of phonemes that differed only in voicing—
such as the range between ba and pa (as in the Lisker & Abramson, 1970, study
mentioned earlier). Participants then indicated whether each of these artificial
stimuli sounded more like ba or more like pa. Eimas and Corbit found that some
of the stimuli participants would normally have called the voiced ba, they now
called the voiceless pa. Thus, the repeated presentation of da had fatigued the

Voice onset time (ms)

/b/ /p/

Id
en

tif
ica

tio
n

(%
)

0

20

40

60

80

100

�100 �50 0 +50 +100 +150

FIGURE 2.25 Percentage iden-
tification of /b/ versus /p/ as
a function of voice-onset time.
A sharp shift in these iden-
tification functions occurred
at about +25 ms. (Data from
Lisker & Abramson, 1970.)

Anderson_8e_Ch02.indd 46 13/09/14 9:36 AM

C O n T e x T A n D PAT T e r n r e C O g n i T i O n / 47

voiced feature detector and raised the threshold for detecting voicing in ba, mak-
ing many former ba stimuli sound like pa.

Although there is general consensus that speech perception is categorical
in some sense, there is considerable debate about what the mechanism is
behind this phenomenon. Some researchers (e.g., Liberman & Mattingly, 1985)
have argued that this reflects special speech perception mechanisms that en-
able people to perceive how the sounds were generated. Consider, for instance,
the categorical distinction between how voiced and unvoiced consonants are
produced—either the vocal cords vibrate during the consonant or they do not.
This has been used to argue that we perceive voicing by perceiving how the con-
sonants are spoken. However, there is evidence that categorical perception is not
tied to humans processing language but rather reflects a general property of how
certain sounds are perceived. For instance, Pisoni (1977) created nonlinguistic
tones that had a similar distinguishing acoustic feature as present in voicing—a
low-frequency tone that is either simultaneous with a high-frequency tone or lags
it by 60 ms. His participants showed abrupt boundaries like those in Figure 2.24
for speech signals. In another study, Kuhl (1987) trained chinchillas to discrimi-
nate between a voiced da and an unvoiced ta. Even though these animals do not
have a human vocal track, they showed the sharp boundary between these stim-
uli that humans do. Thus, it seems that categorical perception depends on neither
the signal being speech (Pisoni, 1977) nor the perceiver having a human vocal
system (Kuhl, 1987). Diehl, Lotto, and Holt (2004) have argued that the pho-
nemes we use are chosen because they match up with boundaries already present
in our auditory system. So it is more a case of our perceptual system determining
our speech behavior than vice versa.

■ Speech sounds differing on continuous dimensions are perceived as
coming from distinct categories.

◆ Context and Pattern Recognition

So far, we have considered pattern recognition as if the only information avail-
able to a pattern-recognition system were the information in the physical stim-
ulus to be recognized. This is not the case, however. Objects occur in context,
and we can use context to help us recognize objects. Consider the example in
Figure 2.26. We perceive the symbols as THE and CAT, even though the spe-
cific symbols drawn for H and A are identical. The general context provided
by the words forces the appropriate interpretation. When context or general
knowledge of the world guides perception, we refer to the processing as top-
down processing, because high-level general knowledge contributes to the in-
terpretation of the low-level perceptual units. A general issue in perception is
how such top-down processing is combined with the bottom-up processing of
information from the stimulus itself, without regard to the general context.

One important line of research in top-down effects comes from a series
of experiments on letter identification, starting with those of Reicher (1969)
and Wheeler (1970). Participants were presented very briefly with either a let-
ter (such as D) or a word (such as WORD). Immediately afterward, they were
given a pair of alternatives and instructed to report which alternative they had
seen. (The initial presentation was sufficiently brief that par-
ticipants made a good many errors in this identification task.)
If they had been shown the letter D, they might be presented
with D and K as alternatives. If they had been shown WORD,
they might be given WORD and WORK as alternatives. Note
that both choices differed only in the letter D or K. Participants

FIGURE 2.26 A demonstration
of context. The same stimulus
is perceived as an H or an A,
depending on the context. (From
Selfridge, 1955. Reprinted by per-
mission of the publisher. © 1955
by the Institute of Electrical and
Electronics Engineers.)

Anderson_8e_Ch02.indd 47 13/09/14 9:36 AM

48 / Chapter 2 P e r C e P T i O n

were about 10% more accurate in identifying the
word than in identifying the letter alone. Thus, they
discriminated between D and K better in the context
of a word than as letters alone—even though, in a
sense, they had to process four times as many letters
in the word context. This phenomenon is known as
the word superiority effect.

Figure 2.27 illustrates an explanation given by
Rumelhart and Siple (1974) and Thompson and
Massaro (1973) for why people are more accurate when
identifying the letter in the word context. The figure
illustrates the products of incomplete perception: Cer-
tain parts of the word cannot be detected—in part (a)
just the last letter is obscured, whereas in part (b) mul-
tiple letters are obscured. If the last letter were all that a
participant was shown, the participant would not be
able to say whether that letter was a K or an R. Thus, the
stimulus information is not enough to identify the letter.
On the other hand, the context is not enough by itself
either—although it is pretty clear in part (a) that the first
three letters are WOR, there are a number of four-letter

words consistent with a WOR beginning: WORD, WORE, WORK, WORM, WORN,
WORT. However, if the participant combines the information from the stimulus with
the information from the context, the whole word must be WORK, which implies K
was the last letter. It is not that participants see the K better in the context of WOR
but that they are better able to infer that K is the fourth letter. The participants are
not conscious of these inferences, however; so they are said to make unconscious
inferences in the act of perception. Note that participants given the alternatives
D and K must not have had conscious access to specific features such as the target
letter having a lower right diagonal, or they would have been able to choose correctly.
Rather, the participants have conscious access only to the whole word or whole let-
ter that the perceptual system has perceived. Note that this analysis is not restricted
to the case where the context letters are unambiguous. In part (b), the second letter
could be an O or a U and the third letter could be a B, P, or R. Still, WORK is the only
possible word.

This example illustrates the redundancy present in many complex stimuli
such as words. These stimuli consist of many more features than are required to
distinguish one stimulus from another. Thus, perception can proceed success-
fully when only some of the features are recognized, with context filling in the
remaining features. In language, this redundancy exists on many levels besides
the feature level. For instance, redundancy occurs at the letter level. We do not

need to perceive every letter in a string of words to be able
to read it. To xllxstxatx, I cxn rxplxce xvexy txirx lextex of
x sextexce xitx an x, anx yox stxll xan xanxge xo rxad xt—ix
wixh sxme xifxicxltx.

■ Word context can be used to supplement feature in-
formation in the recognition of letters.

Massaro’s FLMP Model for Combination
of Context and Feature Information
We have reviewed the effects of context on pattern recogni-
tion in a variety of perceptual situations, but the question of

WORK

WORK

(a)

(b)

FIGURE 2.27 A hypothetical set
of features that might be extracted
on a trial in an experiment of word
perception: (a) when only the last
letter is obscured; (b) when multi-
ple letters are obscured.

FIGURE 2.28 Contextual clues
used by Massaro (1979) to study
how participants combine stimu-
lus information from a letter with
context information from the sur-
rounding letters. (From Massaro,
D. W., Letter information and ortho-
graphic context in word perception,
Journal of experimental Psychology:
Human Perception and Perfor-
mance, 5, 595–609. Copyright ©
1979 American Psychological Asso-
cation. Reprinted by permission.)

Word Superiority

Anderson_8e_Ch02.indd 48 13/09/14 9:36 AM

C O n T e x T A n D PAT T e r n r e C O g n i T i O n / 49

1.0

.9

.8

.7

.6

.5

.4

.3

.2

.1

0
2 3 4 5

Stimulus value

Only c

Pr
ob

ab
ilit

y
of

e
re

sp
on

se

1
c

6
e

Both e and c

Only e
Neither
e nor c

FIGURE 2.29 Probability of an e
response as a function of the stimu-
lus value of the test letter and of
the orthographic context. The lines
reflect the predictions of Massaro’s
FlMP model. The leftmost line is for
the case where the context provides
evidence only for e. The middle line
is the same prediction when the
context provides evidence for both e
and c or when it provides evidence
for neither e nor c. The rightmost line
is for the case where the context pro-
vides evidence only for c. (Data from
Massaro, 1979.)

how to understand these effects still remains. Massaro
has argued that the perceptual information and the
context provide two independent sources of informa-
tion about the identity of the stimulus and that they
are just combined to provide a best guess of what the
stimulus might be. Figure 2.28 shows examples of the
material he used in a test of recognition of the letter c
versus the letter e.

The four quadrants represent four possibilities in
the amount of contextual evidence: Only an e can make
a word, only a c can make a word, both letters can make
a word, or neither can make a word. As one reads down
within a quadrant, the image of the ambiguous letter
provides more evidence for letter e and less for letter c.
Participants were briefly exposed to these stimuli and
asked to identify the letter. Figure 2.29 shows the results
as a function of stimulus and context information. As
the image of the letter itself provided more evidence for
an e, the probability of the participants’ identifying an
e went up. Similarly, the probability of identifying an e
increased as the context provided more evidence.

Massaro argued that these data reflect an inde-
pendent combination of evidence from the context and
evidence from the letter stimulus. He assumed that the
letter stimulus represents some evidence Lc for the letter c and that the context
also provides some evidence Cc for the letter c. He assumed that these evidences
can be scaled on a range of 0 to 1 and can be thought of basically as probabilities,
which he called “fuzzy truth values.” Because probabilities sum to 1, the evidence
for e from the letter stimulus is Le = 1 – Lc, and the evidence from the context is
Ce = 1 – Cc. Given these probabilities, then, the overall probability for a c is

p(c) 5
Lc 3 Cc

(Lc 3 Cc) 1 (Le 3 Ce)

The lines in Figure 2.29 illustrate the predictions from his theory. In gen-
eral, Massaro’s theory (called FLMP for fuzzy logical model of perception)
has done a very good job of accounting for the combination of context and
stimulus information in pattern recognition.

■ Massaro’s FLMP model of perception proposes that contextual
information combines independently with stimulus information to
determine what pattern is perceived.

Other Examples of Context and Recognition
Word recognition is one case for which there have been detailed analyses
of contextual influences, but contextual effects are ubiquitous. For instance,
equally good evidence exists for the role of context in the perception of speech.
A nice illustration is the phoneme-restoration effect, originally demonstrated
in an experiment by Warren (1970). He asked participants to listen to the sen-
tence “The state governors met with their respective legislatures convening in
the capital city,” with a 120-ms tone replacing the middle s in legislatures. Only
1 in 20 participants reported hearing the pure tone, and that participant was
not able to locate it correctly.

Anderson_8e_Ch02.indd 49 13/09/14 9:36 AM

50 / Chapter 2 P e r C e P T i O n

An interesting extension of this first study was an experiment by Warren and
Warren (1970). They presented participants with sentences such as the following:

It was found that the *eel was on the axle.
It was found that the *eel was on the shoe.
It was found that the *eel was on the orange.
It was found that the *eel was on the table.

In each case, the * denotes a phoneme replaced by nonspeech. For the four sen-
tences above, participants reported hearing wheel, heel, peel, and meal, depending
on context. The important feature to note about each of these sentences is that
they are identical through the critical word. The identification of the critical word
is determined by what occurs after it. Thus, the identification of words often is
not instantaneous but can depend on the perception of subsequent words.

Context also appears to be important for the perception of complex visual
scenes. Biederman, Glass, and Stacy (1973) looked at the perception of objects
in novel scenes. Figure 2.30 illustrates the two kinds of scenes presented to
their participants. Figure 2.30a shows a normal scene; in Figure 2.30b, the same
scene is jumbled. Participants viewed one of the scenes briefly on a screen, and
immediately thereafter an arrow pointed to a position on a now-blank screen
where an object had been moments before. Participants were asked to identify
the object that had been in that position in the scene. For example, the arrow
might have pointed to the location of the fire hydrant. Participants were consid-
erably more accurate in their identifications when they had viewed the coherent
picture than when they had viewed the jumbled picture. Thus, as with the pro-
cessing of written text or speech, people are able to use context in a visual scene
to help in their identification of an object.

One of the most dramatic examples of the influence of context on percep-
tion involves a phenomenon called change blindness. As I will discuss in detail in
Chapter 3, people are unable to keep track of all the information in a typical com-
plex scene. If elements of the scene change at the same time as some retinal distur-
bance occurs (such as an eye movement or a scene-cut in a motion picture), people
often fail to detect the change. The original studies on change blindness (McConkie
& Currie, 1996) introduced large changes in pictures that participants were viewing
while they were making an eye movement. For instance, the color of a car in the

(a) (b)

FIGURE 2.30 Scenes used by Biederman, glass, and Stacy (1973) in their study of
the role of context in the recognition of complex visual scenes: (a) a coherent scene;
(b) a jumbled scene. it is harder to recognize the fire hydrant in the jumbled scene.
(From Biederman, Glass, & Stacy, 1973. Reprinted by permission of the publisher. © 1973
by the American Psychological Association.)

Anderson_8e_Ch02.indd 50 13/09/14 9:36 AM

C O n C l u S i O n S / 51

picture might change and the change might not be noticed. Figure 2.31 illustrates a
dramatic instance of change blindness (Simons & Levin, 1998) where it seems con-
text is also promoting the insensitivity to change. The experimenter stopped pedes-
trians on Cornell University’s campus and asked for directions. While the unwit-
ting participant was giving the directions, workers carrying a door passed between
the experimenter and the participant and an accomplice took the place of the ex-
perimenter. Only 7 of the 15 participants noticed the change. In the scene shown
in Figure 2.31, the participants thought of themselves as giving instructions to a
student, and as long as the changed experimenter fit that interpretation, they did
not process him as different. In a laboratory study of the ability to detect changes
in people’s faces, Beck, Rees, Frith, and Lavie (2001) found greater activation in the
fusiform gyrus (see the earlier discussion of face recognition) when face changes
were detected than when they were not.

■ Contextual information biases perceptual processing in a wide
variety of situations.

◆ Conclusions

This chapter discusses how the neurons process sensory information and deliver it
to the higher centers in the brain, and how the information then becomes recog-
nizable as objects. Figure 2.32 depicts the overall flow of information processing in
the case of vision perception. Perception begins with light energy from the exter-
nal environment. Receptors, such as those on the retina, transform this energy into
neural information. Early sensory processing makes initial sense of the information
by extracting features to yield what Marr called the primal sketch. These features
are combined with depth information to get a representation of the location of sur-
faces in space; this is Marr’s 2½-D sketch. The gestalt principles of organization are
applied to segment the elements into objects; this is Marr’s 3-D model. Finally, the
features of these objects and the general context information are combined to rec-
ognize the objects. The output of this last level is a representation of the objects and
their locations in the environment, and this is what we are consciously aware of in
perception. This information is the input to the higher level cognitive processes.
Figure 2.32 illustrates an important point: A great deal of information processing
must take place before we are consciously aware of the objects we are perceiving.

FIGURE 2.31 An example of change blindness. Frames showing how one experimenter
switched places with an accomplice as workers carrying a door passed between the experi-
menter and an unwitting participant. Only 7 of the 15 participants noticed the change.

Light energy

Primal sketch

2 ½-D sketch

3-D model

Recognized
objects

Feature extraction

Depth information

Gestalt principles
of organization

Feature combination,
contextual information

FIGURE 2.32 How information
flows from the environment and
is processed into our perceptual
representation of recognized ob-
jects. The ovals represent differ-
ent levels of information in Marr’s
(1982) model and the lines are
labeled with the perceptual pro-
cesses that transform one level of
information into the next.

Anderson_8e_Ch02.indd 51 13/09/14 9:36 AM

52 / Chapter 2 P e r C e P T i O n

Questions for Thought

1. Figure 2.33a illustrates an optical illusion called
Mach Bands after the Austrian physicist and
philosopher, Ernst Mach, who discovered them.
Each band is a uniform shade of gray and yet it
appears lighter on the right side near the darker
adjacent band, and it appears darker on the left
side near the lighter band. Can you explain why,
using on-off cells, edge detectors, and bar detec-
tors in your explanation (see Figures 2.7 and 2.8)?

2. Use the gestalt principles to explain why we tend
to see two triangles in Figure 2.33b.

3. Rather than Biederman’s geon proposal
(see Figure 2.20), which involves recognizing

objects by recognizing abstract features of
their components, Ullman (2006) proposes we
recognize objects by recognizing fragments
like those in Figure 2.33c. What might be the
relative strengths of the geon theory versus the
fragment-based theory?

4. In Figure 2.21, we see that presented with the
stimulus “cdit,” there is an increased tendency
for participants to say that they have seen “edit,”
which makes a word. Some people describe this
as a case of context distorting perception. Do you
agree that this is a case of distortion?

FIGURE 2.33 (a) Mach bands; (b) demonstration of gestalt principles of organization; (c) fragments for recognizing a horse
from ullman (2006). (Epshtein, Lifshitz, & Ullman, 2008. Copyright 2008 National Academy of Sciences U.S.A.)

Key Terms

2½-D sketch
3-D model
apperceptive agnosia
associative agnosia
bar detectors
bottom-up processing
categorical perception
change blindness

consonantal feature
edge detectors
feature analysis
feature maps
fovea
fusiform gyrus
fuzzy logical model of

perception (FLMP)

geons
gestalt principles of

organization
phonemes
phoneme-restoration

effect
place of articulation
primal sketch

prosopagnosia
recognition-by-

components theory
template matching
top-down processing
visual agnosia
voicing
word superiority effect

Figures for Questions for Thought

(a)

(b)

(c)

Anderson_8e_Ch02.indd 52 13/09/14 9:36 AM

53

Chapter 2 described how the human visual system and other perceptual systems
simultaneously process information from all over their sensory fields. However,

we have limits on how much we can do in parallel. In many situations, we can
attend to only one spoken message or one visual object at a time. This chapter
explores how higher level cognition determines what to attend to. We will consider
the following questions:

● In a busy world filled with sounds, how do we select what to listen to?
● How do we find meaningful information within a complex visual scene?
● What role does attention play in putting visual patterns together as recognizable

objects?
● How do we coordinate parallel activities like driving a car and holding a

conversation?

◆ Serial Bottlenecks

Psychologists have proposed that there are serial bottlenecks in human in-
formation processing, points at which it is no longer possible to continue
processing everything in parallel. For example, it is generally accepted that
there are limits to parallelism in the motor systems. Although most of us can
perform separate actions simultaneously when the actions involve different
motor systems (such as walking and chewing gum), we have difficulty in
getting one motor system to do two things at once. Thus, even though we have
two hands, we have only one system for moving our hands, so it is hard to get
our two hands to move in different ways at the same time. Think of the familiar
problem of trying to pat your head while rubbing your stomach. It is hard to
prevent one of the movements from dominating—if you are like me, you tend
to wind up rubbing or patting both parts of the body.1 The many human motor
systems—one for moving feet, one for moving hands, one for moving eyes, and
so on—can and do work independently and simultaneously, but it is difficult to
get any one of these systems to do two things at the same time.

One question that has occupied psychologists is how early do the bot-
tlenecks occur: before we perceive the stimulus, after we perceive the stimu-
lus but before we think about it, or only just before motor action is required?
Common sense suggests that some things cannot be done at the same time.

1 Drummers (including my son) are particularly good at doing this—I definitely am not a drummer. This
suggests that the real problem might be motor timing.

3
Attention and Performance

Anderson_8e_Ch03.indd 53 13/09/14 9:37 AM

54 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

For instance, it is basically impossible to add two
digits and multiply them simultaneously. Still, there
remains the question of just where the bottlenecks in
information processing lie. Various theories about
when they happen are referred to as early-selection
theories or late-selection theories, depending on
where they propose that bottlenecks take place. Wher-
ever there is a bottleneck, our cognitive processes
must select which pieces of information to attend to
and which to ignore. The study of attention is con-
cerned with where these bottlenecks occur and how
information is selected at these bottlenecks.

A major distinction in the study of attention is be-
tween goal-directed factors (sometimes called endog-
enous control) and stimulus-driven factors (sometimes
called exogenous control). To illustrate the distinction,
Corbetta and Shulman (2002) ask us to imagine our-
selves at Madrid’s El Prado Museum, looking at the
right panel of Bosch’s painting The Garden of Earthly
Delights (see Color Plate 3.1). Initially, our eyes will

probably be drawn to large, salient objects like the instrument in the center
of the picture. This would be an instance of stimulus-driven attention—it is
not that we wanted to attend to this; the instrument just grabbed our atten-
tion. However, our guide may start to comment on a “small animal playing a
musical instrument.” Now we have a goal and will direct our attention over the
picture to find the object being described. Continuing their story, Corbetta and
Shulman ask us to imagine that we hear an alarm system starting to ring in the
next room. Now a stimulus-driven factor has intervened, and our attention will
be drawn away from the picture and switch to the adjacent room. Corbetta and
Shulman argue that somewhat different brain systems control goal-directed
attention versus stimulus-driven attention. For instance, neural imaging evi-
dence suggests that the goal-directed attentional system is more left lateralized,
whereas the stimulus-driven system is more right lateralized.

The brain regions that select information to process can be distinguished (to
an approximation) from those that process the information selected. Figure 3.1
highlights the parietal cortex, which influences information processing in regions
such as the visual cortex and auditory cortex. It also highlights prefrontal regions
that influence processing in the motor area and more posterior regions. These
prefrontal regions include the dorsolateral prefrontal cortex and, well below the
surface, the anterior cingulate cortex. As this chapter proceeds, it will elaborate on
the research concerning the various brain regions in Figure 3.1.

■ Attentional systems select information to process at serial bottle-
necks where it is no longer possible to do things in parallel.

◆ Auditory Attention

Some of the early research on attention was concerned with auditory attention.
Much of this research centered on the dichotic listening task. In a typical di-
chotic listening experiment, illustrated in Figure 3.2, participants wear a set of
headphones. They hear two messages at the same time, one in each ear, and are
asked to “shadow” one of the two messages (i.e., repeat back the words from
that message only). Most participants are able to attend to one message and
tune out the other.

FIGURE 3.1 A representation of
some of the brain areas involved
in attention and some of the per-
ceptual and motor regions they
control. The parietal regions are
particularly important in directing
perceptual resources. The pre-
frontal regions (dorsolateral
prefrontal cortex, anterior cingu-
late) are particularly important in
executive control.

Parietal cortex: attends
to locations and objects

Motor cortex:
controls hands

Dorsolateral prefrontal
cortex: directs central
cognition

Auditory cortex:
processes auditory
information

Extrastriate cortex:
processes visual
information

Anterior cingulate:
(midline structure)
monitors conflict

Brain Structures

Anderson_8e_Ch03.indd 54 13/09/14 9:37 AM

A u d I To r y AT T e n T I o n / 55

Psychologists (e.g., Cherry, 1953; Moray, 1959) have discovered that very
little information about the unattended message is processed in a dichotic
listening task. All that participants can report about the unattended message is
whether it was a human voice or a noise; whether the human voice was male
or female; and whether the sex of the speaker changed during the test. They
cannot tell what language was spoken or remember any of the words, even if
the same word was repeated over and over again. An analogy is often made
between performing this task and being at a party, where a guest tunes in to
one message (a conversation) and filters out others. This is an example of goal-
directed processing—the listener selects the message to be processed. However,
to return to the distinction between goal-directed and stimulus-driven processing,
important stimulus information can disrupt our goals. We have probably all
experienced the situation in which we are listening intently to one person and
hear our name mentioned by someone else. It is very hard in this situation to
keep your attention on what the original speaker is saying.

The Filter Theory
Broadbent (1958) proposed an early-selection theory called the filter theory
to account for these results. His basic assumption was that sensory infor-
mation comes through the system until some bottleneck is reached. At that
point, a person chooses which message to process on the basis of some phys-
ical characteristic. The person is said to filter out the other information. In a
dichotic listening task, the theory proposed that the message to each ear was
registered but that at some point the participant selected one ear to listen
with. At a busy party, we pick which speaker to follow on the basis of physical
characteristics, such as the pitch of the speaker’s voice.

A crucial feature of Broadbent’s original filter model is its proposal that we
select a message to process on the basis of physical characteristics such as ear
or pitch. This hypothesis made a certain amount of neurophysiological sense.
Messages entering each ear arrive on different nerves. Nerves also vary in
which frequencies they carry from each ear. Thus, we might imagine that the
brain, in some way, selects certain nerves to “pay attention to.”

People can certainly choose to attend to a message on the basis of its
physical characteristics, but they can also select messages to process on the

… and then John turned rapidly toward …

ran house ox cat

and, um, John turned . . .

FIGURE 3.2 A typical dichotic listening task. different messages are presented to the left
and right ears, and the participant attempts to “shadow” the message entering one ear.
(Research from Lindsay & Norman, 1977.)

Dichotic Listening

Attentional Filtering

Anderson_8e_Ch03.indd 55 13/09/14 9:37 AM

56 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

basis of their semantic content. In one study, Gray and Wed-
derburn (1960), who at the time were undergraduate stu-
dents at Oxford University, demonstrated that participants
can use meaningfulness to follow a message that jumps back
and forth between the ears. Figure 3.3 illustrates the par-
ticipants’ task in their experiment. In one ear they might be
hearing the words dogs six fleas, while at the same time hear-
ing the words eight scratch two in the other ear. Instructed to
shadow the meaningful message, participants would report
dogs scratch fleas. Thus, participants can shadow a message
on the basis of meaning rather than on the basis of what
each ear physically hears.

Treisman (1960) looked at a situation in which par-
ticipants were instructed to shadow a particular ear
(Figure 3.4). The message in the ear to be shadowed was
meaningful up to a certain point; then it turned into a
random sequence of words. Simultaneously, the mean-

ingful message switched to the other ear—the one to which the participant
had not been attending. Some participants switched ears, against instruc-
tions, and continued to follow the meaningful message. Others contin-
ued to follow the shadowed ear. Thus, it seems that sometimes people use
a physical characteristic (e.g., a particular ear) to select which message to
follow, and sometimes they choose semantic content.

■ Broadbent’s filter model proposes that we use physical features,
such as ear or pitch, to select one message to process, but it has been
shown that people can also use the meaning of the message as the
basis for selection.

The Attenuation Theory and the Late-Selection Theory
To account for these kinds of results, Treisman (1964) proposed a modification
of the Broadbent model that has come to be known as the attenuation theory.
This model hypothesized that certain messages would be attenuated (weak-
ened) but not filtered out entirely on the basis of their physical properties. Thus,
in a dichotic listening task, participants would minimize the signal from the un-
attended ear but not eliminate it. Semantic selection criteria could apply to all
messages, whether they were attenuated or not. If the message were attenuated,

it would be harder to apply these selection criteria, but
it would still be possible. Treisman (personal com-
munication, 1978) emphasized that in her experiment
in Figure 3.4, most participants actually continued to
shadow the prescribed ear. It was easier to follow the
message that is not being attenuated than to apply se-
mantic criteria to switch attention to the attenuated
message.

An alternative explanation was offered by
J. A. Deutsch and D. Deutsch (1963) in their late-
selection theory, which proposed that all the infor-
mation is processed completely without attenuation.
Their hypothesis was that the capacity limitation is in
the response system, not the perceptual system. They
claimed that people can perceive multiple messages
but that they can say only one message at a time. Thus,
people need some basis for selecting which message to

FIGURE 3.3 An illustration of the
shadowing task in the Gray and
Wedderburn (1960) experiment.
The participant follows the mean-
ingful message as it moves from
ear to ear. (Adapted from Klatzky,
1975.)

dogs six fleas . . .

. . . eight scratch two

dogs scratch fleas . . .

I SAW THE GIRL/Song was wishing . . .

The to-be-shadowed ear

I SAW THE GIRL JUMPING . . .

. . . me that bird
JUMPING IN THE STREET.

FIGURE 3.4 An illustration of the
Treisman (1960) experiment. The
meaningful message moves to
the other ear, and the participant
sometimes continues to shadow
it against instructions. (Adapted
from Klatzky, 1975.)

Anderson_8e_Ch03.indd 56 13/09/14 9:37 AM

A u d I To r y AT T e n T I o n / 57

shadow. If they use meaning as the criterion (either according to or in contra-
diction to instructions), they will switch ears to follow the message. If they use
the ear of origin in deciding what to attend to, they will shadow the chosen ear.

The difference between this late-selection theory and the early-selection
attenuation theory is illustrated in Figure 3.5. Both models assume that there
is some filter or bottleneck in processing. Treisman’s theory (Figure 3.5a) as-
sumes that the filter selects which message to attend to, whereas Deutsch and
Deutsch’s theory (Figure 3.5b) assumes that the filter occurs after the percep-
tual stimulus has been analyzed for verbal content. Treisman and Geffen (1967)
tested the difference between these two theories using a dichotic listening task
in which participants had to shadow one message while also processing both
messages for a target word. If they heard the target word, they were to signal
by tapping. According to the Deutsch and Deutsch late-selection theory, mes-
sages from both ears would get through and participants should have been able
to detect the critical word equally well in either ear. In contrast, the attenuation
theory predicted much less detection in the unshadowed ear because the mes-
sage would be attenuated. In the experiment, participants detected 87% of the
target words in the shadowed ear and only 8% in the unshadowed ear. Other
evidence consistent with the attenuation theory was reported by Treisman and
Riley (1969) and by Johnston and Heinz (1978).

There is neural evidence for a version of the attenuation theory that as-
serts that there is both enhancement of the signal coming from the attended
ear and attenuation of the signal coming from the unattended ear. The pri-
mary auditory area of the cortex (see Figure 3.1) shows an enhanced response
to auditory signals coming from the ear the listener is attending to and a de-
creased response to signals coming from the other ear. Through ERP recording,
Woldorff et al. (1993) showed that these responses occur between 20 and 50 ms
after stimulus onset. The enhanced responses occur much sooner in auditory
processing than the point at which the meaning of the message can be identi-
fied. Other studies also provide evidence for enhancement of the message in the
auditory cortex on the basis of features other than location. For instance, Za-
torre, Mondor, and Evans (1999) found in a PET study that when people attend

(a) (b)

Responses

Selection and
organization of
responses

Analysis of
verbal content

Perceptual
filter

Input messages

1 2

1 2

Responses

Selection and
organization of
responses

Analysis of
verbal content

Input messages

Response filter

1 2

1 2 FIGURE 3.5 Treisman and Geffen’s
illustration of attentional limita-
tions produced by (a) Treisman’s
(1964) attenuation theory and
(b) deutsch and deutsch’s (1963)
late-selection theory. (Data from
Treisman & Geffen, 1967.)

Anderson_8e_Ch03.indd 57 13/09/14 9:37 AM

58 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

to a message on the basis of pitch, the auditory cortex shows enhancement (reg-
istered as increased activation). This study also found increased activation in
the parietal areas that direct attention.

Although auditory attention can enhance processing in the primary audi-
tory cortex, there is no evidence of reliable effects of attention on earlier stages
of auditory processing, such as in the auditory nerve or the brain stem (Picton
& Hillyard, 1974). The various results we have reviewed suggest that the pri-
mary auditory cortex is the earliest area to be influenced by attention. It should
be stressed that the effects at the auditory cortex are a matter of attenuation and
enhancement. Messages are not completely filtered out, and so it is still possible
to select them at later points of processing.

■ Attention can enhance or reduce the magnitude of response to an
auditory signal in the primary auditory cortex.

◆ Visual Attention

The bottleneck in visual information processing is even more apparent than
the one in auditory information processing. As we saw in Chapter 2, the retina
varies in acuity, with the greatest acuity in a very small area called the fovea.
Although the human eye registers a large part of the visual field, the fovea reg-
isters only a small fraction of that field. Thus, in choosing where to focus our
vision, we also choose to devote our most powerful visual processing resources
to a particular part of the visual field, and we limit the resources allocated to
processing other parts of the field. Usually, we are attending to that part of the
visual field on which we are focusing. For instance, as we read, we move our
eyes so that we are fixating the words we are attending to.

The focus of visual attention is not always identical with the part of the visual
field being processed by the fovea, however. People can be instructed to fixate on
one part of the visual field (making that part the focus of the fovea) while attend-

ing to another, nonfoveal region of the visual field.2 In one
experiment, Posner, Nissen, and Ogden (1978) had par-
ticipants focus on a constant point and then presented them
with a stimulus 7° to the left or the right of the fixation point.
In some trials, participants were told on which side the stim-
ulus was likely to occur; in other trials, there was no such
warning. The warning was correct 80% of the time, but 20%
of the time the stimulus appeared on the unexpected side.
The researchers monitored eye movements and included
only those trials in which the eyes had stayed on the fixation
point. Figure 3.6 shows the time required to judge the stimu-
lus if it appeared in the expected location (80% of the time),
if the participant had not been given a neutral cue (50% of
the time on both sides), and if it appeared in the unexpected
location (20% of the time). Participants were faster when the
stimulus appeared in the expected location and slower when
it appeared in the unexpected location. Thus, they were able
to shift their attention from where their eyes were fixated.

Posner, Snyder, and Davidson (1980) found that peo-
ple can attend to regions of the visual field as far as 24°

Re
ac

tio
n

tim
e

(m
s)

Unexpected No expectation Expected

Condition

320

300

280

260

240

220

FIGURE 3.6 The results of an
experiment to determine how
people react to a stimulus that
occurs 7° to the left or right of
the fixation point. The graph
shows participants’ reaction times
to expected, unexpected, and
neutral (no expectation) signals.
(Data from Posner et al., 1978.)

2 This is what quarterbacks are supposed to do when they pass the football, so that they don’t “give away”
the position of the intended receiver.

Anderson_8e_Ch03.indd 58 13/09/14 9:37 AM

V I s u A l AT T e n T I o n / 59

from the fovea. Although visual attention can be moved without accompanying
eye movements, people usually do move their eyes, so that the fovea processes
the portion of the visual field to which they are attending. Posner (1988)
pointed out that successful control of eye movements requires us to attend to
places outside the fovea. That is, we must attend to and identify an interesting
nonfoveal region so that we can guide our eyes to fixate on that region to
achieve the greatest acuity in processing it. Thus, a shift of attention often
precedes the corresponding eye movement.

To process a complex visual scene, we must move our attention around in
the visual field to track the visual information. This process is like shadowing a
conversation. Neisser and Becklen (1975) performed the visual analog of the audi-
tory shadowing task. They had participants observe two videotapes superimposed
over each other. One was of two people playing a hand-slapping game; the other
was of some people playing a basketball game. Figure 3.7 shows how the situation
appeared to the participants. They were instructed to pay attention to one of the
two films and to watch for odd events such as the two players in the hand-slapping
game pausing and shaking hands. Participants were able to monitor one film
successfully and reported filtering out the other. When asked to monitor both films
for odd events, the participants experienced great difficulty and missed many of the
critical events.

As Neisser and Becklen (1975) noted, this situ-
ation involved an interesting combination of the use
of physical cues and the use of content cues. Partici-
pants moved their eyes and focused their attention in
such a way that the critical aspects of the monitored
event fell on their fovea and the center of their atten-
tive spotlight. The only way they could know where
to move their eyes to focus on a critical event was by
making reference to the content of the event. Thus,
the content of the event facilitated their processing
of the film, which in turn facilitated extracting the
content.

Figure 3.8 shows examples of the overlapping
stimuli used in an experiment by O’Craven, Downing,
and Kanwisher (1999) to study the neural conse-
quences of attending to one object or the other. Partici-
pants in their experiment saw a series of pictures that
consisted of faces superimposed on houses. They were
instructed to look for either repetition of the same face

(a) (b) (c)

FIGURE 3.7 frames from the two films used by neisser and Becklen in their visual
analog of the auditory shadowing task. (a) The “hand-game” film; (b) the basketball film;
and (c) the two films superimposed. (Neisser, U., & Becklen, R. (1975). Selective looking:
Attending to visually specified events. Cognitive Psychology, 7, 480–494. Copyright © 1975
Elsevier. Reprinted by permission.)

FIGURE 3.8 An example of
a picture used in the study of
o’Craven et al. (1999). When the
face is attended, there is activa-
tion in the fusiform face area,
and when the house is attended,
there is activation in the parahip-
pocampal place area. (Downing,
Liu, & Kanwisher, 2001. Reprinted
with permission from Elsevier.)

Spatial Cueing

Anderson_8e_Ch03.indd 59 13/09/14 9:37 AM

60 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

in the series or repetition of the same house. Recall from Chapter 2 that there
is a region of the temporal cortex, the fusiform face area, which becomes active
when people are observing faces. There is another area within the temporal cor-
tex, the parahippocampal place area, that becomes more active when people are
observing places. What is special about these pictures is that they consisted of
both faces and places. Which region would become active—the fusiform face area
or the parahippocampal place area? As the reader might suspect, the answer de-
pended on what the participant was attending to. When participants were looking
for repetition of faces, the fusiform face area became more active; when they were
looking for repetition of places, the parahippocampal place area became more
active. Attention determined which region of the temporal cortex was engaged in
the processing of the stimulus.

■ People can focus their attention on parts of the visual field and
move their focus of attention to process what they are interested in.

The Neural Basis of Visual Attention
It appears that the neural mechanisms underlying visual attention are very similar
to those underlying auditory attention. Just as auditory attention directed to one
ear enhances the cortical signal from that ear, visual attention directed to a spatial
location appears to enhance the cortical signal from that location. If a person
attends to a particular spatial location, a distinct neural response (detected using
ERP records) in the visual cortex occurs within 70 to 90 ms after the onset of a
stimulus. On the other hand, when a person is attending to a particular object
(attending to a chair rather than a table, say) rather than to a particular location
in space, we do not see a response for more than 200 ms. Thus, it appears to take
more effort to direct visual attention on the basis of content than on the basis of
physical features, just as is the case with auditory attention.

Mangun, Hillyard, and Luck (1993) had participants fixate on the center
of a computer screen, then judge the lengths of bars presented in positions dif-
ferent from the fixation location (upper left, lower left, upper right, and lower
right). Figure 3.9 shows the distribution of scalp activity detected by ERP when

P1 attention effect
(current density)

P1 P1 P1 P1

Stimulus

+ + + +

FIGURE 3.9 results from an experiment by mangun, Hillyard, and luck. distribution of scalp
activity was recorded by erP when a participant was attending to one of the four different
regions of the visual array depicted in the bottom row while fixating on the center of the
screen. The greatest activity was recorded over the side of the scalp opposite the side of the
visual field where the object appeared, confirming that there is enhanced neural processing
in portions of the visual cortex corresponding to the location of visual attention. (Mangun,
G. R., Hillyard, S. A., & Luck, S. J. (1993). Electrocortical substrates of visual selective attention.
In D. Meyer & S. Kornblum (Eds.), Attention and performance (Vol. 14, Figure 10.4 from
pp. 219–243). © 1993 Massachusetts Institute of Technology, by permission of The MIT Press.)

Anderson_8e_Ch03.indd 60 13/09/14 9:37 AM

V I s u A l AT T e n T I o n / 61

a participant was attending to one of these four different regions of the visual ar-
ray (while fixating on the center of the screen). Consistent with the topographic
organization of the visual cortex, there was greatest activity over the side of the
scalp opposite the side of the visual field where the object appeared. Recall from
Chapters 1 and 2 (see Figure 2.5) that the visual cortex (at the back of the head) is
topographically organized, with each visual field (left or right) represented in the
opposite hemisphere. Thus, it appears that there is enhanced neural processing in
the portion of the visual cortex corresponding to the location of visual attention.

A study by Roelfsema, Lamme, and Spekreijse (1998) illustrates the im-
pact of visual attention on information processing in the primary visual area of
the macaque monkey. In this experiment, the researchers trained monkeys to
perform the rather complex task illustrated in Figure 3.10. A trial would begin
with a monkey fixating on a particular stimulus in the visual field, the star in
part (a) of the figure. Then, as shown in Figure 3.10b, two curves would appear
that ended in blue dots. Only one of these curves was connected to the fixation
point. The monkey had to keep looking at the fixation point for 600 ms and
then perform a saccade (an eye movement) to the end of the curve that con-
nected the fixation (part c). While a monkey performed this task, Roelfsema
et al. recorded from cells in the monkey’s primary visual cortex (where cells
with receptive fields like those in Figure 2.8 are found). Indicated by the square
in Figure 3.10 is a receptive field of one of these cells. It shows increased re-
sponse when a line falls on that part of the visual field and so responds when
the curve appears that crosses it. The cell’s response also increased during the
600-ms waiting period, but only if its receptive field was on the curve that con-
nected to the fixation point. During the waiting period the monkey was shifting
its attention along this curve to find its end point and thus determine the des-
tination of the saccade. This shift of attention across the receptive field caused
the cell to respond more strongly.

■ When people attend to a particular spatial location, there is
greater neural processing in portions of the visual cortex correspond-
ing to that location.

Visual Search
People are able to select stimuli to attend to, either in the visual or auditory
domain, on the basis of physical properties and, in particular, on the basis of
location. Although selection based on simple features can occur early and

Fixation point

Receptive field

Stimulus (600 ms)Fixation (300 ms) Saccade

(a) (b) (c)

FIGURE 3.10 The experimental
procedure in roelfsema et al.
(1998): (a) The monkey fixates
the start point (the star).
(b) Two curves are presented,
one of which links the start point
to a target point (a blue circle).
(c) The monkey saccades to the
target point. The experimenter
records from a neuron whose
receptive field is along the curve
to the target point.

Anderson_8e_Ch03.indd 61 13/09/14 9:37 AM

62 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

quickly in the visual system, not everything people look for can be defined in
terms of simple features. How do people find more complex objects, such as
the face of a friend in a crowd? In such cases, it seems that they must search
through the faces in the crowd, one by one, looking for a face that has the de-
sired properties. Much of the research on visual attention has focused on how
people perform such searches. Rather than study how people find faces in a
crowd, however, researchers have tended to use simpler material. Figure 3.11,
for instance, shows a portion of the display that Neisser (1964) used in one of
the early studies. Try to find the first K in the set of letters displayed.

Presumably, you tried to find the K by going through the letters row by
row, looking for the target. Figure 3.12 graphs the average time it took partici-
pants in Neisser’s experiment to find the letter as a function of which row it ap-
peared in. The slope of the best-fitting function in the graph is about 0.6, which
implies that participants took about 0.6 s to scan each line. When people engage
in such searches, they appear to be allocating their attention intensely to the
search process. For instance, brain-imaging experiments have found strong ac-
tivation in the parietal cortex during such searches (see Kanwisher & Wojciulik,
2000, for a review).

Although a search can be intense and difficult, it is not always that way.
Sometimes we can find what we are looking for without much effort. If we
know that our friend is wearing a bright red jacket, it can be relatively easy to
find him or her in the crowd, provided that no one else is wearing a bright red
jacket. Our friend will just pop out of the crowd. Indeed, if there were just one
red jacket in a sea of white jackets, it would probably pop out even if we were
not looking for it—an instance of stimulus-driven attention. It seems that if
there is some distinctive feature in an array, we can find it without a search.

Treisman studied this sort of pop-out. For instance, Treisman and
Gelade (1980) instructed participants to try to detect a T in an array of 30 I’s
and Y’s (Figure 3.13a). They reasoned that participants could do this simply
by looking for the crossbar feature of the T that distinguishes it from all I’s
and Y’s. Participants took an average of about 400 ms to perform this task.
Treisman and Gelade also asked participants to detect a T in an array of I’s
and Z’s (Figure 3.13b). In this task, they could not use just the vertical bar
or just the horizontal bar of the T; they had to look for the conjunction of
these features and perform the feature combination required in pattern rec-
ognition. It took participants more than 800 ms, on average, to find the let-
ter in this case. Thus, a task requiring them to recognize the conjunction of
features took about 400 ms longer than one in which perception of a single
feature was sufficient. Moreover, when Treisman and Gelade varied the num-

ber of letters in the array, they found that participants
were much more affected by the number of objects in
the task that required recognition of the conjunction
of features (see Figure 3.14).

■ It is necessary to search through a visual array
for an object only when a unique visual feature
does not distinguish that object.

The Binding Problem
As discussed in Chapter 2, there are different types
of neurons in the visual system that respond to dif-
ferent features, such as colors, lines at various ori-
entations, and objects in motion. A single object in
our visual field will involve a number of features; for

0
0

10

20

30

40

10 20 30 40 50

Ti
m

e
(s

)

Position of critical item (line number)

FIGURE 3.12 The time required
to find a target letter in the array
shown in figure 3.11 as a function
of the line number in which it ap-
pears. (Data from Neisser, 1964.)

FIGURE 3.11 A representation of
lines 7–31 of the letter array used
in neisser’s search experiment.
(Data from Neisser, 1964.)

TWLN
XJBU
UDXI
HSFP
XSCQ
SDJU
PODC
ZVBP
PEVZ
SLRA
JCEN
ZLRD
XBOD
PHMU
ZHFK
PNJW
CQXT
GHNR
IXYD
QSVB
GUCH
OWBN
BVQN
FOAS
ITZN

Serial vs. Parallel
Search

Anderson_8e_Ch03.indd 62 13/09/14 9:37 AM

V I s u A l AT T e n T I o n / 63

instance, a red vertical line combines the vertical feature and the red fea-
ture. The fact that different features of the same object are represented by
different neurons gives rise to a logical question: How are these features put
back together to produce perception of the object? This would not be much
of a problem if there were just a single object in the visual field. We could
assume that all the features belonged to that object. But what if there were
multiple objects in the field? For instance, suppose there were just two ob-
jects: a red vertical bar and a green horizontal bar. These two objects might
result in the firing of neurons for red, neurons for green, neurons for verti-
cal lines, and neurons for horizontal lines. If these firings were all that oc-
curred, though, how would the visual system know it saw a red vertical bar
and a green horizontal bar rather than a red horizontal bar and a green ver-
tical bar? The question of how the brain puts together various features in
the visual field is referred to as the binding problem.

Treisman (e.g., Treisman & Gelade, 1980) developed her feature-integration
theory as an answer to the binding problem. She proposed that people must focus
their attention on a stimulus before they can synthesize its features into a pattern.
For instance, in the example just given, the visual system can first direct its atten-
tion to the location of the red vertical bar and synthesize that object, then direct its
attention to the green horizontal bar and synthesize that object. According to Tre-
isman, people must search through an array when they need to syn-
thesize features to recognize an object (for instance, when trying to
identify a K, which consists of a vertical line and two diagonal lines).
In contrast, when an object in an array has a single unique feature,
such as a red jacket or a line at a particular orientation, we can attend
to it without search.

The binding problem is not just a hypothetical dilemma—
it is something that humans actually experience. One source of
evidence comes from studies of illusory conjunctions in which
people report combinations of features that did not occur. For
instance, Treisman and Schmidt (1982) looked at what happens
to feature combinations when the stimuli are out of the focus of
attention. Participants were asked to report the identity of two
black digits flashed in one part of the visual field, so this was

(a)

(b)

FIGURE 3.13 stimuli used by
Treisman and Gelade to determine
how people identify objects in the
visual field. They found that it is
easier to pick out a target letter (T)
from a group of distracters if
(a) the target letter has a feature
that makes it easily distinguishable
from the distracter letters (I’s and
Y’s) than if (b) the same target
letter is in an array of distracters
(I’s and Z’s) that offer no obvious
distinctive features. (Data from
Treisman & Gelade, 1980.)

Array size (number of items)

T in I, ZRe
ac

tio
n

tim
e

(m
s)

1
0

400

800

1200

5 15 30

T in I, Y

FIGURE 3.14 results from the
Treisman and Gelade experiment.
The graph plots the average re-
action times required to detect
a target letter as a function of
the number of distracters and
whether the distracters contain
separately all the features of the
target. (Data from Treisman &
Gelade, 1980.)

Anderson_8e_Ch03.indd 63 13/09/14 9:37 AM

64 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

where their attention was focused. In an unattended part of the visual field,
letters in various colors were presented, such as a pink T, a yellow S, and a
blue N. After they reported the numbers, participants were asked to report
any letters they had seen and the colors of these letters. They reported see-
ing illusory conjunctions of features (e.g., a pink S) almost as often as they
reported seeing correct combinations. Thus, it appears that we are able to
combine features into an accurate perception only when our attention is fo-
cused on an object. Otherwise, we perceive the features but may well com-
bine them into a perception of objects that were never there. Although rather
special circumstances are required to produce illusory conjunctions in an or-
dinary person, there are certain patients with damage to the parietal cortex
who are particularly prone to such illusions. For instance, one patient stud-
ied by Friedman-Hill, Robertson, and Treisman (1995) confused which let-
ters were presented in which colors even when shown the letters for as long
as 10 s.

A number of studies have been conducted on the neural mechanisms
involved in binding together the features of a single object. Luck, Chelazzi,
Hillyard, and Desimone (1997) trained macaque monkeys to fixate on a cer-
tain part of the visual field and recorded neurons in a visual region called
V4. The neurons in this region have large receptive fields (several degrees
of visual angle). Therefore, multiple objects in a display may be within
the visual field of a single neuron. They found neurons that were specific
to particular types of objects, such as a cell that responded to a blue verti-
cal bar. What happens when a blue vertical bar and a green horizontal bar
are presented both within the receptive field of this cell? If the monkey at-
tended to the blue vertical bar, the rate of response of the cell was the same
as when there was only a blue vertical bar. On the other hand, if the mon-
key attended to the green horizontal bar, the rate of firing of this same cell
was greatly depressed. Thus, the same stimulus (blue vertical bar plus green
horizontal bar) can evoke different responses depending on which object is
attended to. It is speculated that this phenomenon occurs because attention
suppresses responses to all features in the receptive field except those at the
attended location. Similar results have been obtained in fMRI experiments
with humans. Kastner, DeWeerd, Desimone, and Ungerleider (1998) meas-
ured the fMRI signal in visual areas that responded to stimuli presented in
one region of the visual field. They found that when attention was directed
away from that region, the fMRI response to stimuli in that region de-
creased; but when attention was focused on that region, the fMRI response

was maintained. These experiments indicate en-
hanced neural processing of attended objects and
locations.

A striking demonstration of the effects of
sustained attention was reported by Simons and
Chabris (1999). They asked participants to watch
a video in which a team dressed in black tossed a
basketball back and forth and a team dressed in
white did the same (Figure 3.15). Participants were
instructed to count either the number of times
the team in black tossed the ball or the number
of times the team in white did so. Presumably, in
one condition participants were looking for events
involving the team in black and in the other for
events involving the team in white. Because the
players were intermixed, the task was difficult
and required sustained attention. In the middle of

FIGURE 3.15 This shows a single
frame from the movie used by
simons and Chabris to demon-
strate the effects of sustained
attention. When participants were
intent on tracking the ball passed
among the players dressed in
white T-shirts, they tended not
to notice the black gorilla walking
through the room. (Adapted from
Simons & Chabris, 1999.)

Anderson_8e_Ch03.indd 64 13/09/14 9:37 AM

V I s u A l AT T e n T I o n / 65

the game, a person in a black gorilla suit walked through the
room. Participants searching the video for events involving
team members dressed in white were so fixed on their search
that they completely missed an event involving a black object.
When participants were tracking the team in white, they no-
ticed the black gorilla only 8% of the time; when they were
tracking the team in black, they noticed it 67% of the time.
People passively watching the video never miss the black
gorilla. (You should be able to find a version of this video by
searching with the keywords “gorilla” and “Simons.”)

■ For feature information to be synthesized into a pat-
tern, the information must be in the focus of attention.

Neglect of the Visual Field
We have discussed the evidence that visual attention to a spa-
tial location results in enhanced activation in the appropriate
portion of the primary visual cortex. The neural structures that
control the direction of attention, however, appear to be lo-
cated elsewhere, particularly in the parietal cortex (Behrmann, Geng, & Shom-
stein, 2004). Damage to the parietal lobe (see Figure 3.1) has been shown to re-
sult in deficits in visual attention. For instance, Posner, Walker, Friederich, and
Rafal (1984) showed that patients with parietal lobe injuries have difficulty in
disengaging attention from one side of the visual field.

Damage to the right parietal region produces distinctive patterns of
deficit, as can be seen in a study of one such patient by Posner, Cohen, and
Rafal (1982). Like the participants in the Posner, Nissen, and Ogden (1978)
experiment discussed earlier, the patient was cued to expect a stimulus to the
left or right of the fixation point (i.e., in the left or right visual field). As in
that experiment, 80% of the time the stimulus appeared in the expected field,
but 20% of the time it appeared in the unexpected field. Figure 3.16 shows
the time required to detect the stimulus as a function of which visual field
it was presented in and which field had been cued. When the stimulus was
presented in the right field, the patient showed only a little disadvantage if
inappropriately cued. If the stimulus appeared in the left field, however, the
patient showed a large deficit if inappropriately cued. Because the right pa-
rietal lobe processes the left visual field, damage to the right lobe impairs its
ability to draw attention back to the left visual field once attention is focused
on the right visual field. This sort of one-sided
attentional deficit can be temporarily created in
normal individuals by presenting TMS to the
parietal cortex (Pascual-Leone et al., 1994—see
Chapter 1 for discussion of TMS).

A more extreme version of this attentional
disorder is called unilateral visual neglect.
Patients with damage to the right hemisphere
completely ignore the left side of the visual
field, and patients with damage to the left
hemisphere ignore the right side of the visual
field. Figure 3.17 shows the performance of a
patient with damage to the right hemisphere,
which caused her to neglect the left visual field
(Albert, 1973). She had been instructed to put
slashes through all the circles. As can be seen,

La
te

nc
y

(m
s)

1400

1200

1000

800

600

400
Left

Field of presentation

Right

Cued for right field
Cued for left field

FIGURE 3.16 The attention
deficit shown by a patient with
right parietal lobe damage when
switching attention to the left
visual field. (Data from Posner,
Cohen, & Rafal, 1982.)

FIGURE 3.17 The performance
of a patient with damage to the
right hemisphere who had been
asked to put slashes through all
the circles. Because of the dam-
age to the right hemisphere, she
ignored the circles in the left part
of her visual field. (From Ellis &
Young, 1988. Reprinted by permis-
sion of the publisher. © 1988 by
Erlbaum.)

Anderson_8e_Ch03.indd 65 13/09/14 9:37 AM

66 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

she ignored the circles in the left part of her visual field. Such patients will
often behave peculiarly. For instance, one patient failed to shave half of his
face (Sacks, 1985). These effects can also show up in nonvisual tasks. For
instance, a study of patients with neglect of the left visual field showed a
systematic bias in making judgments about the midpoint in sequences of
numbers and letters (Zorzi, Priftis, Meneghello, Marenzi, & Umiltà, 2006).
When asked to judge what number is midway between 1 and 5, they showed
a bias to respond 4. They showed a similar tendency with letter sequences—
asked to judge what letter was midway between P and T, they showed
tendency to respond S. In both cases this can be interpreted as a tendency
to ignore the items that were to the left of the point in the middle of the
sequence.

It seems that the right parietal lobe is involved in allocating spatial
attention in many modalities, not just the visual (Zatorre et al., 1999). For
instance, when one attends to the location of auditory or visual stimuli,
there is increased activation in the right parietal region. It also appears that
the right parietal lobe is more responsible for the spatial allocation of at-
tention than is the left parietal lobe and that this is why right parietal dam-
age tends to produce such dramatic effects. Left parietal damage tends to
produce a subtler pattern of deficits. Robertson and Rafal (2000) argue that
the right parietal region is responsible for attention to such global features
as spatial location, whereas the left parietal region is responsible for direct-
ing attention to local aspects of objects. Figure 3.18 is a striking illustra-
tion of the different types of deficits associated with left and right parietal
damage. Patients were asked to draw the objects in Figure 3.18a. Patients
with right parietal damage (Figure 3.18b) were able to reproduce the spe-
cific components of the picture but were not able to reproduce their spatial
configuration. In contrast, patients with left parietal damage (Figure 3.18c)
were able to reproduce the overall configuration but not the detail. Simi-
larly, brain-imaging studies have found more activation of the right parietal
region when a person is responding to global patterns and more activation
of the left parietal region when a person is attending to local patterns (Fink
et al., 1996; Martinez et al., 1997).

FIGURE 3.18 (a) The pictures
presented to patients with
parietal damage. (b) examples of
drawings made by patients with
right-hemisphere damage. These
patients could reproduce the
specific components of the picture
but not their spatial configuration.
(c) examples of drawings made
by patients with left-hemisphere
damage. These patients
could reproduce the overall
configuration but not the detail.
(After Robertson & Lamb, 1991.)

(a) (b) (c)

Anderson_8e_Ch03.indd 66 13/09/14 9:37 AM

V I s u A l AT T e n T I o n / 67

■ Parietal regions are responsible for the allocation of attention, with
the right hemisphere more concerned with global features and the left
hemisphere with local features.

Object-Based Attention
So far we have talked about space-based attention, where people allocate
their attention to a region of space. There is also evidence, for object-based
attention, where people focus their attention on particular objects rather
than regions of space. An experiment by Behrmann, Zemel, and Mozer
(1998) is an example of research demonstrating that people sometimes find
it easier to attend to an object than to a location. Figure 3.19 illustrates some
of the stimuli used in the experiment, in which participants were asked
to judge whether the numbers of bumps on the two ends of objects were
the same. The left column shows instances in which the numbers of bumps were
the same, the right column instances in which the numbers were not the same.
Participants made these judgments faster when the bumps were on the same ob-
ject (top and bottom rows in Figure 3.19) than when they were on different objects
(middle row). This result occurred despite the fact that when the bumps were on
different objects, they were located closer together, which should have facilitated
judgment if attention were space based. Behrmann et al. argue that participants
shifted attention to one object at a time rather than one location at a time. There-
fore, judgments were faster when the bumps were all on the same object because
participants did not need to shift their attention between objects. Using a variant
of the paradigm in Figure 3.19, Chen and Cave (2008) either presented the stimu-
lus for 1 s or for just 0.12 s. The advantage of the within-object effect disappeared
when the stimulus was present for only the brief period. This indicates that it takes
time for object-based attention to develop.

Other evidence for object-centered attention involves a phenomenon
called inhibition of return. Research indicates that if we have looked at a par-
ticular region of space, we find it a little harder to return our attention to that
region. If we move our eyes to location A and then to location B, we are slower
to return our eyes to location A than to some new location C. This is also true
when we move our attention without moving our eyes (Posner, Rafal, Chaote, &
Vaughn, 1985). This phenomenon confers an advantage in some situations: If
we are searching for something and have already looked at a location, we would
prefer our visual system to find other locations to look at rather than return to
an already searched location.

Tipper, Driver, and Weaver (1991) performed one demonstration of the
inhibition of return that also provided evidence for object-based attention. In
their experiments, participants viewed three squares in a frame, similar to what
is shown in each part of Figure 3.20. In one condition, the squares did not move
(unlike the moving condition illustrated in Figure 3.20, which we will discuss
in the next paragraph). The participants’ attention was drawn to one of the
outer squares when the experimenters made it flicker, and then, 200 ms later,
attention was drawn back to the center square when that square flickered. A
probe stimulus was then presented in one of the two outer positions, and par-
ticipants were instructed to press a key indicating that they had seen the probe.
On average, they took 420 ms to see the probe when it occurred at the outer
square that had not flickered and 460 ms when it occurred at the outer square
that had flickered. This 40-ms advantage is an example of a spatially defined in-
hibition of return. People are slower to move their attention to a location where
it has already been.

FIGURE 3.19 stimuli used in an
experiment by Behrmann, Zemel,
and mozer to demonstrate that it
is sometimes easier to attend to
an object than to a location. The
left and right columns indicate
same and different judgments,
respectively; and the rows from
top to bottom indicate the
single-object, two-object, and
occluded conditions, respectively.
(Behrmann, M., Zemel, R. S., &
Mozer, M. C. (1998). Object-based
attention and occlusion: Evidence
from normal participants and
computational model. Journal of
experimental Psychology: Human
Perception and Performance, 24,
1011–1036. Copyright © 1988
American Psychological Association.
Reprinted by permission.)

(a) (d)

(e)

(f)(c)

(b)

Anderson_8e_Ch03.indd 67 13/09/14 9:37 AM

68 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

Figure 3.20 illustrates the other condition of their experiment, in which the
objects were rotated around the screen after the flicker. By the end of the mo-
tion, the object that had flickered on one side was now on the other side—the
two outer objects had traded positions. The question of interest was whether
participants would be slower to detect a target on the right (where the flicker-
ing had been—which would indicate location-based inhibition) or on the left
(where the flickered object had ended up—which would indicate object-based
inhibition). The results showed that they were about 20 ms slower to detect an
object in the location that had not flickered but that contained the object that
had flickered. Thus, their visual systems displayed an inhibition of return to the
same object, not the same location.

It seems that the visual system can direct attention either to locations in
space or to objects. Experiments like those just described indicate that the
visual system can track objects. On the other hand, many experiments indicate
that people can direct their attention to regions of space where there are no ob-
jects (see Figure 3.6 for the results of such an experiment). It is interesting that
the left parietal regions seem to be more involved in object-based attention
and the right parietal regions in location-based attention. Patients with left
parietal damage appear to have deficits in focusing attention on objects (Egly,
Driver, & Rafal, 1994), unlike the location-based deficits that I have described

(a)

(b)

(c)

(d)

(e)

FIGURE 3.20 examples of frames used in an experiment by Tipper, driver, and Weaver
to determine whether inhibition of return would attach to a particular object or to its
location. Arrows represent motion. (a) display onset, with no motion for 500 ms. After
two moving frames, the three filled squares were horizontally aligned (b), whereupon
the cue appeared (one of the boxes flickered). Clockwise motion then continued, with
cueing in the center for the initial three frames (c–e). The outer squares continued to
rotate clockwise (d) until they were horizontally aligned (e), at which point a probe was
presented, as before. (© 1991 from Tipper, S. P., Driver, J., & Weaver, B. (1991). Short re-
port: Object-centered inhibition of return of visual attention. Quarterly Journal of experimental
Psychology, 43(section A), 289–298. Reproduced by permission of Taylor & Francis LLC,
http://www.tandfonline.com.)

Anderson_8e_Ch03.indd 68 13/09/14 9:37 AM

http://www.tandfonline.com

C e n T r A l AT T e n T I o n : s e l e C T I n G l I n e s o f T H o u G H T To P u r s u e / 69

in patients with right parietal damage. Also, there is greater activation in the
left parietal regions when people attend to objects than when they attend to
locations (Arrington, Carr, Mayer, & Rao, 2000; Shomstein & Behrmann,
2006). This association of the left parietal region with object-based attention is
consistent with the earlier research we reviewed (see Figure 3.18) showing that
the right parietal region is responsible for attention to global features and the
left for attention to local features.

■ Visual attention can be directed either toward objects independent
of their location or toward locations independent of what objects are
present.

◆ Central Attention: Selecting Lines of
Thought to Pursue

So far, this chapter has considered how people allocate their attention to pro-
cess stimuli in the visual and auditory modalities. What about cognition after
the stimuli are attended to and encoded? How do we select which lines of
thought to pursue? Suppose we are driving down a highway and encode the
fact that a dog is sitting in the middle of the road. We might want to figure
out why the dog is sitting there, we might want to consider whether there is
something we should do to help the dog, and we certainly want to decide how
best to steer the car to avoid an accident. Can we do all these things at once?
If not, how do we select the most important problem of deciding how to steer
and save the rest for later? It appears that people allocate central attention to
competing lines of thought in much the same way they allocate perceptual at-
tention to competing objects.

In many (but not all) circumstances, people are able to pursue only one
line of thought at a time. This section will describe
two laboratory experiments: one in which it appears
that people have no ability to overlap two tasks and
another in which they appear to have almost total
ability to do so. Then we will address how people can
develop the ability to overlap tasks and how they se-
lect among tasks when they cannot or do not want to
overlap them.

The first experiment, which Mike Byrne and I
did (Byrne & Anderson, 2001), illustrates the claim
made at the beginning of the chapter about it being
impossible to multiply and add two numbers at the
same time. Participants in this experiment saw a
string of three digits, such as “3 4 7.” Then they were
asked to do one or both of two tasks:

● Task 1: Judge whether the first two digits add up
to the third and press a key with the right index
finger if they do and another key with the left
index finger if they do not.

● Task 2: Report verbally the product of the first
and third numbers. In this case, the answer is 21,
because 3 × 7 = 21.

Figure 3.21 compares the time required to do
each task in the single-task condition versus the

Verify
addition

La
te

nc
y

(m
s)

Generate
multiplication

250

0

500

750

1000

1250

1500

1750

2000

2250
Stimulus: 3 4 7

Mean time to complete both tasks

Single task

Dual task

FIGURE 3.21 The results of
an experiment by Byrne and
Anderson to see whether
people can overlap two tasks.
The bars show the response
times required to solve two
problems—one of addition and
one of multiplication—when done
by themselves and when done
together. The results indicate that
the participants were not able to
overlap the addition and multipli-
cation computations. (Data from
Byrne & Anderson, 2001.)

Anderson_8e_Ch03.indd 69 13/09/14 9:37 AM

70 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

time required for each task in the dual-task condition. Participants took al-
most twice as long to do either task when they had to perform the other as
well. In the dual task they sometimes gave the answer for the multiplication
task first (59% of the time) and sometimes the addition task first (41%). The
bars in Figure 3.21 for the dual task reflect the time to answer the problem
whether the task was answered first or second. The horizontal black line
near the top of Figure 3.21 represents the time they took to give the both
answers. This time (1.99 s) is greater than the sum of the time for the verifi-
cation task by itself (0.88 s) and the time for the multiplication task by itself
(1.05 s). The extra time probably reflects the cost of shifting between tasks
(for reviews, see Monsell, 2003; Kiesel et al., 2010). In any case, it appears
that the participants were not able to overlap the addition and multiplication
computations at all.

The second experiment, reported by Schumacher et al. (2001), illustrates
what is referred to as perfect time-sharing. The tasks were much simpler than
the tasks in the Byrne and Anderson (2001) experiment. Participants simul-
taneously saw a single letter on a screen and heard a tone and, as in the first
experiment, had to perform two tasks, either individually or at the same time:

● Task 1: Press a left, middle, or right key according to whether the letter
occurred on the left, in the middle, or on the right.

● Task 2: Say “one,” “two,” or “three” according to whether the tone was low,
middle, or high in frequency.

Figure 3.22 compares the times required to do each task in the single-task
condition and the dual-task condition. As can be seen, these times are nearly un-
affected by the requirement to do the two tasks at once. There are many differ-
ences between this task and the Byrne and Anderson task, but the most apparent
is the complexity of the tasks. Participants were able to do the individual tasks
in the second experiment in a few hundred milliseconds, whereas the individual
tasks in the first experiment took around a second. Significantly more thought

Location
discrimination

Re
sp

on
se

ti
m

e
(m

s)

0

50

100

150

200

250

300

350

400

450

500

Tone
discrimination

Single task

Dual task

FIGURE 3.22 The results of
an experiment by schumacher
et al. illustrating near perfect
time-sharing. The bars show
the times required to perform
two simple tasks—a location
discrimination task and a tone
discrimination task—when done
by themselves and when done
together. The times were nearly
unaffected by the requirement
to do the two tasks at once,
indicating that the participants
achieved almost perfect time-
sharing. (Data from Schumacher
et al., 2001.)

Anderson_8e_Ch03.indd 70 13/09/14 9:38 AM

C e n T r A l AT T e n T I o n : s e l e C T I n G l I n e s o f T H o u G H T To P u r s u e / 71

was required in the first experiment, and it is hard for people to engage in both
streams of thought simultaneously. Also, participants in the second experiment
achieved perfect time-sharing only after five sessions of practice, whereas par-
ticipants in the first experiment had only one session of practice.

Figure 3.23 presents an analysis of what occurred in the Schumacher et al.
(2001) experiment. It shows what was happening at various points in time in
five streams of processing: (1) perceiving the visual location of a letter, (2) gen-
erating manual actions, (3) central cognition, (4) perceiving auditory stimuli,
and (5) generating speech. Task 1 involved visually encoding the location of the
letter, using central cognition to select which key to press, and then performing
the actual finger movement. Task 2 involved detecting and encoding the tone,
using central cognition to select which word to say (“one,” “two,” or “three”),
and then saying it. The lengths of the boxes in Figure 3.23 represent estimates
of the duration of each component based on human performance studies. Each
of these streams can go on in parallel with the others. For instance, during the
time the tone is being detected and encoded, the location of the letter is be-
ing encoded (which happens much faster), a key is being selected by central
cognition, and the motor system is starting to program the action. Although
all these streams can go on in parallel, within each stream only one thing can
happen at a time. This could create a bottleneck in the central cognition stream,
because central cognition must direct all activities (e.g., in this case, it must
serve both task 1 and task 2). In this experiment, however, the length of time
devoted to central cognition was so brief that the two tasks did not contend for
the resource. The five days of practice in this experiment played a critical role
in reducing the amount of time devoted to central cognition.

Although the discussion here has focused on bottlenecks in central
cognition, there can be bottlenecks in any of the processing streams. Earlier, we
reviewed evidence that people cannot attend to two locations at once; they must

Encode
letter

location
Vision

Manual action

Central cognition

Speech

Time (ms)

Streams of processing:

Ta
sk

2

Ta
sk

1

Audition

Select
action

Program key press

Detect and
encode tone

0 100 200 300 400

Generate speech

Select
action

FIGURE 3.23 An analysis of the timing of events in five streams of processing during
execution of the dual task in the schumacher et al. (2001) experiment: (1) vision,
(2) manual action, (3) central cognition, (4) speech, and (5) audition.

Anderson_8e_Ch03.indd 71 13/09/14 9:38 AM

72 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

shift their attention across locations in the visual array serially. Similarly, they
can process only one speech stream at a time, move their hands in one way at
a time, or say one thing at a time. Even though all these peripheral processes
can have bottlenecks, it is generally thought that bottlenecks in central cogni-
tion can have the most significant effects, and they are the reason we seldom
find ourselves thinking about two things at once. The bottleneck in central
cognition is referred to as the central bottleneck.

■ People can process multiple perceptual modalities at once or
execute actions in multiple motor systems at once, but they cannot
process multiple things in a single system, including central cognition.

Automaticity: Expertise Through Practice
The near perfect time-sharing in Figure 3.22 only emerged after 5 days of prac-
tice. The general effect of practice is to reduce the central cognitive component
of information processing. When one has practiced the central cognitive com-
ponent of a task so much that the task requires little or no thought, we say that
doing the task is automatic. Automaticity is a matter of degree. A nice example
is driving. For experienced drivers in unchallenging conditions, driving has
become so automatic that they can carry on a conversation while driving
with little difficulty. Experienced drivers are much more successful at doing
secondary tasks like changing the radio (Wikman, Nieminen, & Summala,
1998). Experienced drivers also often have the experience of traveling long
stretches of highway with no memory of what they did.

There have been a number of dramatic demonstrations in the psycho-
logical literature of how practice can enable parallel processing. For instance,
Underwood (1974) reports a study on the psychologist Neville Moray, who had
spent many years studying shadowing. During that time, Moray practiced shad-
owing a great deal, and unlike most participants in experiments, he was very
good at reporting what was contained in the unattended channel. Through a
great deal of practice, the process of shadowing had become partially automatic
for Moray, and he had capacity left over to attend to the unshadowed channel.

Spelke, Hirst, and Neisser (1976) provided an interesting demonstration of
how a highly practiced skill ceases to interfere with other ongoing behaviors. (This
was a follow-up of a demonstration pioneered by the writer Gertrude Stein when

Why is cell phone use
and driving a dangerous
combination?

Bottlenecks in information processing
can have important practical implica-
tions. A study by the Harvard Center
for risk Analysis (Cohen & Graham,
2003) estimates that cell phone
distraction results in 2,600 deaths,
330,000 injuries, and 1.5 million
instances of property damage in the

united states each year. strayer and
drews (2007) review the evidence
that people are more likely to miss
traffic lights and other critical informa-
tion while talking on a cell phone.
moreover, these problems are not
any better with hands-free phones. In

contrast, listening to a radio or books
on tape does not interfere with driv-
ing. strayer and drews suggest that
the demands of participating in a
conversation place more require-
ments on central cognition. When
someone says something on the cell
phone, they expect an answer and
are unaware of the current driving
conditions. strayer and drews note
that participating in a conversation
with a passenger in the car is not as
distracting because the passenger
will adjust the conversation to driv-
ing demands and even point out
potential dangers to the driver.

I m p l I c a t I o n s
▼

To
m

G
ril

l/C
or

bi
s.

▲

Anderson_8e_Ch03.indd 72 13/09/14 9:38 AM

C e n T r A l AT T e n T I o n : s e l e C T I n G l I n e s o f T H o u G H T To P u r s u e / 73

she was an undergraduate working with William James at Harvard University.)
Their participants had to perform two tasks simultaneously: read a text silently
for comprehension while copying words dictated by the experimenter. At first,
this was extremely difficult. Participants had to read much more slowly than nor-
mal in order to copy the words accurately. After six weeks of practice, however, the
participants were reading at normal speed. They had become so skilled at copying
automatically that their comprehension scores were the same as for normal read-
ing. For these participants, reading while copying had become no more difficult
than reading while walking. It is of interest that participants reported no awareness
of what it was they were copying. Much as with driving, the participants lost their
awareness of the automated activity.3

Another example of automaticity is transcription typing. A typist is simultane-
ously reading the text and executing the finger strokes for typing. In this case, we
have three systems operating in parallel: perception of the text to be typed, central
translation of the earlier perceived letters into keystrokes, and the actual typing of
the letters. It is the central processes that get automated. Skilled transcription typists
often report little awareness of what they are typing, because this task has become
so automated. Skilled typists also find it impossible to stop typing instantaneously.
If suddenly told to stop, they will hit a few more letters before quitting (Salthouse,
1985, 1986).

■ As tasks become practiced, they become more automatic and
require less and less central cognition to execute.

The Stroop Effect
Automatic processes not only require little or no central cognition to execute
but also appear to be difficult to prevent. A good example is word recognition for
practiced readers. It is virtually impossible to look at a common word and not
read it. This strong tendency for words to be recognized
automatically has been studied in a phenomenon known as
the Stroop effect, after the psychologist who first demon-
strated it, J. Ridley Stroop (1935). The task requires partici-
pants to say the ink color in which words are printed. Color
Plate 3.2 provides an illustration of such a task. Try naming
the colors of the words in each column as fast as you can.
Which column was easiest to read? Which was hardest?

The three columns illustrate three of the conditions in
which the Stroop effect is studied. The first column illus-
trates a neutral, or control, condition in which the words
are not color words. The second column illustrates the
congruent condition in which the words are the same as
the color of the ink they are printed in. The third column
illustrates the conflict condition in which there are color
words but they are different from their ink colors. A typical
modern experiment, rather than having participants read
a whole column, will present a single word at a time and
measure the time to name that word. Figure 3.24 shows
the results from such an experiment on the Stroop effect
by Dunbar and MacLeod (1984). Compared to the control
condition of a neutral word, participants could name the

3 When given further training with the intention of remembering what they were transcribing, partici-
pants were also able to recall this information.

Re
ac

tio
n

tim
e

(m
s)

900

800

700

600

500

400
Congruent Control Conflict

Color naming
Word reading

Condition

FIGURE 3.24 Performance data
for the standard stroop task. The
curves plot the average reaction
time of the participants as a func-
tion of the condition tested: con-
gruent (the word was the name
of the ink color); control (the
word was not related to color at
all); and conflict (the word was
the name of a color different
from the ink color). (Data from
Dunbar & MacLeod, 1984.)

Anderson_8e_Ch03.indd 73 13/09/14 9:38 AM

74 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

ink color somewhat faster in the congruent condition—when the word was the
name of the ink color. In the conflict condition, when the word was the name
of a different color, they named the ink color much more slowly. For instance,
they had great difficulty in saying “green” when the ink color of the word red was
green. Figure 3.24 also shows the results when the task is switched and partici-
pants are asked to read the word and not name the color. The effects are asym-
metrical; that is, individual participants experienced very little interference in
reading a word even if it was different from its ink color. This reflects the highly
automatic character of reading. Additional evidence for its automaticity is that
participants could read a word much faster than they could name its ink color.
Reading is such an automatic process that not only is it unaffected by the color,
but participants are unable to inhibit reading the word, and that reading can in-
terfere with the color naming.

MacLeod and Dunbar (1988) looked at the effect of practice on performance
in a variant of the Stroop task. They used an experiment in which the participants
learned the color names for random shapes. Part (a) of Color Plate 3.3 illustrates
the shape-color associations they might learn. The experimenters then presented
the participants with test geometric shapes and asked them to say either the color
name associated with the shape or the actual ink color of the shape. As in the
original Stroop experiment, there were three conditions; these are illustrated in
part (b) of Color Plate 3.3:

1. Congruent: The shape was in the same ink color as its name.
2. Control: White shapes were presented when participants were to say the

color name for the shape; colored squares were presented when they were
to name the ink color of the shape. (The square shape was not associated
with any color.)

3. Conflict: The random shape was in a different ink color from its name.

As shown in Figure 3.25, color naming was much more automatic than shape
naming and was relatively unaffected by congruence with the shape, whereas
shape naming was affected by congruence with the ink color (Figure 3.25a).

Re
ac

tio
n

tim
e

(m
s)

750

700

650

600

550

500

450
Congruent Control Conflict

Condition(a) (b)

Re
ac

tio
n

tim
e

(m
s)

750

700

650

600

550

500

450
Congruent Control Conflict

Condition

Color naming
Shape naming

Color naming
Shape naming

FIGURE 3.25 results from the experiment created by macleod and dunbar (1988) to
evaluate the effect of practice on the performance of a stroop task. The data reported
are the average times required to name shapes and colors as a function of color-shape
congruence: (a) initial performance and (b) after 20 days of practice. The practice made
shape naming automatic, like word reading, so that it affected color naming. (Data from
MacLeod and Dunbar, 1988.)

Stroop Effect

Anderson_8e_Ch03.indd 74 13/09/14 9:38 AM

C e n T r A l AT T e n T I o n : s e l e C T I n G l I n e s o f T H o u G H T To P u r s u e / 75

Then MacLeod and Dunbar gave the participants 20 days of practice at naming
the shapes. Participants became much faster at naming shapes, and now shape
naming interfered with color naming rather than vice versa (Figure 3.25b).
Thus, the consequence of the training was to make shape naming automatic,
like word reading, so that it affected color naming.

■ Reading a word is such an automatic process that it is difficult to
inhibit, and it will interfere with processing other information about
the word.

Prefrontal Sites of Executive Control
We have seen that the parietal cortex is important in the exercise of attention in the
perceptual domain. There is evidence that the prefrontal regions are particularly
important in direction of central cognition, often known as executive control. The
prefrontal cortex is that portion of the frontal cortex anterior to the premotor re-
gion (the premotor region is area 6 in Color Plate 1.1). Just as damage to parietal
regions results in deficits in the deployment of perceptual attention, damage to pre-
frontal regions results in deficits of executive control. Patients with such damage of-
ten seem totally driven by the stimulus and fail to control their behavior according
to their intentions. A patient who sees a comb on the table may simply pick it up
and begin combing her hair; another who sees a pair of glasses will put them on
even if he already has a pair on his face. Patients with damage to prefrontal regions
show marked deficits in the Stroop task and often cannot refrain from saying the
word rather than naming the color (Janer & Pardo, 1991).

Two prefrontal regions shown in Figure 3.1 seem particularly important in
executive control. One is the dorsolateral prefrontal cortex (DLPFC), which is
the upper portion of the prefrontal cortex. It is called dorsolateral because it is
high (dorsal) and to the side (lateral). The second region is the anterior cingu-
late cortex (ACC), which is folded under the visible surface of the brain along
the midline. The DLPFC seems particularly important in the setting of inten-
tions and the control of behavior. For instance, it is highly active during the
simultaneous performance of dual tasks such as those whose results are reported
in Figures 3.21 and 3.22 (Szameitat, Schubert, Muller, & von Cramon, 2002).
The ACC seems particularly active when people must monitor conflict between
competing tendencies. For instance, brain-imaging studies show that it is highly
active in Stroop trials when a participant must name the color of a word printed
in an ink of conflicting color (J. V. Pardo, P. J. Pardo, Janer, & Raichle, 1990).

There is a strong relationship between the ACC and cognitive control in many
tasks. For instance, it appears that children develop more cognitive control as their
ACC develops. The amount of activation in the ACC appears to be correlated with
children’s performance in tasks requiring cognitive control (Casey et al., 1997a). De-
velopmentally, there also appears to be a positive correlation between performance
and sheer volume of the ACC (Casey et al., 1997b). Weissman, Roberts, Visscher,
and Woldorff (2006) studied trial-to-trial variation in activity of the ACC when
participants were performing a simple judgment task. When there was a decrease
in ACC activation, participants showed an increase in time to make the judgment.
Weissman et al.’s interpretation was that lapses in attention are produced by de-
creases in ACC activation.

A nice paradigm for demonstrating the development of cognitive control in
children is the “Simon says” task. In one study, Jones, Rothbart, and Posner (2003)
had children receive instructions from two dolls—a bear and an elephant—such
as, “Elephant says, ‘Touch your nose.’ ” The children were to follow the instructions
from one doll (the act doll) and ignore the instructions from the other (the inhibit

Anderson_8e_Ch03.indd 75 13/09/14 9:38 AM

76 / Chapter 3 AT T e n T I o n A n d P e r f o r m A n C e

doll). All children successfully followed the act doll but many had diffi-
culty ignoring the inhibit doll. From the age of 36 to 48 months, children
progressed from 22% success to 91% success in ignoring the inhibit doll.
Some children used physical strategies to control their behavior such as sit-
ting on their hands or distorting their actions—pointing to their ear rather
than their nose.

Another way to appreciate the importance of prefrontal regions to
cognitive control is to compare the performance of humans with that of
other primates. As reviewed in Chapter 1, a major dimension of the evo-
lution from primates to humans has been the increase in the size of pre-
frontal regions. Primates can be trained to do many tasks that humans do,
and so they permit careful comparison. One such task involving a variant
of the Stroop task presents participants with a display of numerals (e.g., five
3s) and pits naming the number of objects against indicating the identity of
the numerals. Figure 3.26 provides an example of this task in the same form
as the original Stroop task (Color Plate 3.2): trying to count the number of
numerals in each line versus trying to name the numerals in each line. The
stronger interference in this case is from the numeral naming to the count-
ing (Windes, 1968). This paradigm has been used to compare Stroop-like
interference in humans versus rhesus monkeys who had been trained to as-
sociate the numerals with their relative quantities—for example, they had
learned that “5” represented a larger quantity than “2” (Washburn, 1994).
Both monkeys and humans were shown two arrays and were required to in-
dicate which had more numerals independent of the identity of the numer-
als (see Figure 3.27). Table 3.1 shows the performance of the monkeys and
humans. Compared to a baseline where they had to judge which array of
letters had more objects, both humans and monkeys performed better when

the numerals agreed with the difference in cardinality and performed worse when
the numerals disagreed (as they do in Figure 3.26). Both populations showed simi-
lar reaction time effects, but whereas the humans made 3% errors in the incongru-
ent condition, the monkeys made 27% errors. The level of performance observed in
the monkeys was like the level of performance observed in patients with damage to
their frontal lobes.

■ Prefrontal regions, particularly DLPFC and ACC, play a major
role in executive control.

◆ Conclusions

There has been a gradual shift in the way cognitive psychology has perceived
the issue of attention. For a long time, the implicit assumption was captured by
this famous quote from William James (1890) over a century ago:

5 5 5

1 1 1 1

2

3 3 3 3 3

4 4

5 5 5

4 4 4 4 4

5 5 5 5

3

4 4 4

2 2 2 2

3 3

4 4 4

1 1 1 1

3

2 2 2

FIGURE 3.26 A numerical stroop
task comparable to the color
stroop task (see Color Plate 3.2).

FIGURE 3.27 A monkey reaches
through its cage to manipulate
the joystick so as to bring the cur-
sor into contact with one of the
arrays. (From Washburn, 1994.)

Everyone knows what attention is. It is the taking posses-
sion by the mind, in a clear and vivid form, of one out of
what seem several simultaneously possible objects or trains
of thought. Focalization, concentration of consciousness
are of its essence. It implies withdrawal from some things in
order to deal effectively with others. (pp. 403–404)

Two features of this quote reflect conceptions once held about atten-
tion. The first is that attention is strongly related to consciousness—we
cannot attend to one thing unless we are conscious of it. The second is
that attention, like consciousness, is a unitary system. More and more,
cognitive psychology is beginning to recognize that attention operates

Anderson_8e_Ch03.indd 76 13/09/14 9:38 AM

C o n C l u s I o n s / 77

at an unconscious level. For instance, people often are not conscious of where they
have moved their eyes. Along with this recognition has come the realization that at-
tention is multifaceted (e.g., Chun, Golumb, & Turk-Browne, 2011). We have seen
that it makes sense to separate auditory attention from visual attention and atten-
tion in perceptual processing from attention in executive control from attention in
response generation. The brain consists of a number of parallel processing systems
for the various perceptual systems, motor systems, and central cognition. Each of
these parallel systems seems to suffer bottlenecks—points at which it must focus its
processing on a single thing. Attention is best conceived as the processes by which
each of these systems is allocated to potentially competing information-processing
demands. The amount of interference that occurs among tasks is a function of the
overlap in the demands that these tasks make on the same systems.

TABLE 3.1 mean response Times and Accuracy levels as a function of species
and Condition

Condition Accuracy (%) Response Time (ms)

Rhesus Monkeys (N = 6)

Congruent numerals 92 676

Baseline (letters) 86 735

Incongruent numerals 73 829

Human Participants (N = 28)

Congruent numerals 99 584

Baseline (letters) 99 613

Incongruent numerals 97 661

Questions for Thought

1. The chapter discussed how listening to one spo-
ken message makes it difficult to process a second
spoken message. Do you think that listening to a
conversation on a cell phone while driving makes
it harder to process other sounds, such as a car
horn honking?

2. Which should produce greater parietal
activation: searching Figure 3.13a for a T or
searching Figure 3.13b for a T?

3. Describe circumstances where it would be
advantageous to focus one’s attention on an
object rather than a region of space, and describe
circumstances where the opposite would be true.

4. We have discussed how automatic behaviors
can intrude on other behaviors and how some
aspects of driving can become automatic.
Consider the situation in which a passenger in a
car is a skilled driver and has automatic aspects
of driving evoked by the driving experience. Can
you think of examples where automatic aspects
of driving seem to affect a passenger’s behavior
in a car? Might this help explain why having a
conversation with a passenger in a car is not as
distracting as having a conversation over a cell
phone?

Key Terms

anterior cingulate cortex
(ACC)

attention
attenuation theory
automaticity
binding problem
central bottleneck

dichotic listening task
dorsolateral prefrontal

cortex (DLPFC)
early-selection theories
executive control
feature-integration

theory

filter theory
goal-directed attention
illusory conjunction
inhibition of return
late-selection theories
object-based attention
perfect time-sharing

serial bottleneck
space-based attention
stimulus-driven attention
Stroop effect

Anderson_8e_Ch03.indd 77 13/09/14 9:38 AM

78

Try answering these two questions:

● How many windows are in your house?
● How many nouns are in the American Pledge of Allegiance?

Most people who answer these questions have the same experience. For the first
question they imagine themselves walking around their house and counting win-
dows. For the second question, if they do not actually say the Pledge of Alliance out
loud, they imagine themselves saying the Pledge of Allegiance. In both cases they
are creating mental images of what they would have perceived.

Use of visual imagery is particularly important. As a result of our primate herit-
age, a large portion of our brain processes visual information. Therefore, we use
these brain structures as much as we can, even in the absence of a visual signal
from the outside world, by creating mental images in our heads. Some of human-
kind’s most creative acts involve visual imagery. For instance, Einstein claimed he dis-
covered the theory of relativity by imagining himself traveling beside a beam of light.

A major debate in cognitive psychology has been the degree to which the
processes behind visual imagery are the same as the perceptual and attentional
processes that we considered in the previous two chapters. Some researchers
(e.g., Pylyshyn, 1973, in an article sarcastically titled “What the Mind’s Eye Tells the
Mind’s Brain”) have argued that our perceptual experience when doing something
like picturing the windows in our house is an epiphenomenon; that is, it is a
mental experience that does not have any functional role in information processing.
The philosopher Daniel Dennett (1969) also argued that mental images are
epiphenomenal:

Consider the Tiger and his Stripes. I can dream, imagine or see a striped
tiger, but must the tiger I experience have a particular number of stripes? If
seeing or imagining is having a mental image, then the image of the tiger
must—obeying the rules of images in general—reveal a definite number of
stripes showing, and one should be able to pin this down with such ques-
tions as “more than ten?”, “less than twenty?” (p. 136)

Dennett’s argument is that if we are actually seeing a tiger in a mental image, we
should be able to count its stripes just like we could if we actually saw a tiger. If
we cannot count the stripes in a mental image of a tiger, we are not having a real
perceptual experience. This argument is not considered decisive, but it does illustrate
the discomfort some people have with the claim that mental images are actually
perceptual in character.

This chapter will review some of the experimental evidence showing the ways
that mental imagery does play a role in information processing. We will define

4
Mental Imagery

Anderson_8e_Ch04.indd 78 13/09/14 9:38 AM

V E r B A l I M A g E r y V E r S U S V I S U A l I M A g E r y / 79

mental imagery broadly as the processing of perceptual-like information in the
absence of an external source for the perceptual information. We will consider the
following questions:

● How do we process the information in a mental image?
● How is imaginal processing related to perceptual processing?
● What brain areas are involved in mental imagery?
● How do we develop mental images of our environment and use these to navi-

gate through the environment?

◆ Verbal Imagery Versus Visual Imagery

Cognitive neuroscience has provided increasing evidence that several dif-
ferent brain regions are involved in mental imagery. This evidence has come
from both studies of patients suffering damage to various brain regions and
studies of the brain activation of normal individuals as they engage in vari-
ous imagery tasks. In one of the early studies of brain activation patterns
during mental imagery, Roland and Friberg (1985) identified many of the
brain regions that have been investigated in subsequent research. The inves-
tigators measured changes in blood flow in the brain as participants either
mentally rehearsed a nine-word circular jingle or mentally rehearsed finding
their way around streets in their neighborhoods. Figure 4.1 illustrates the
principal areas they identified. When participants engaged in the verbal jin-
gle task, there was activation in the prefrontal cortex near Broca’s area and in
the parietal-temporal region of the posterior cortex near Wernicke’s area. As
discussed in Chapter 1, patients with damage to these regions show deficits
in language processing. When participants engaged in the visual task, there
was activation in the parietal cortex, occipital cortex, and temporal cortex.
All these areas are involved in visual perception and attention, as we saw in
Chapters 2 and 3. Thus, when people process imagery of language or visual
information, some of the same brain areas are active as when they process ac-
tual speech or visual information.

An experiment by Santa (1977) demonstrated the functional consequence
of representing information in a visual image versus representing it in a verbal
image. The two conditions of Santa’s experiment are shown in Figure 4.2. In
the geometric condition (Figure 4.2a), participants studied an array of three

R

R

R

R

J

J

Brain Structures FIGURE 4.1 results from roland
and Friberg’s (1985) study of
brain activation patterns during
mental imagery. regions of the
left cortex showed increased
blood flow when participants
imagined a verbal jingle (J) or a
spatial route (r).

Mental Imagery

Anderson_8e_Ch04.indd 79 13/09/14 9:38 AM

80 / Chapter 4 M E n TA l I M A g E r y

geometric objects, arranged with one object centered below the other two. As
can be seen without much effort, this array has a facelike property (eyes and
a mouth). After participants studied the array, it was removed, and they had
to hold the information in their minds. They were presented with one of sev-
eral different test arrays. The participants’ task was to verify that the test array
contained the same elements as the study array, although not necessarily in the
same spatial configuration. Thus, participants should have responded posi-
tively to the first two test arrays in Figure 4.2a and negatively to the last two.
The interesting results concern the difference between the two positive test ar-
rays. The first was identical to the study array (same-configuration condition).
In the second array, the elements were displayed in a line (linear-configuration
condition). Santa predicted that participants would make a positive identifi-
cation more quickly in the first case, where the configuration was identical—

because, he hypothesized, the mental image for
the study stimulus would preserve spatial infor-
mation. The results for the geometric condition
in Figure 4.3 confirm Santa’s predictions. Par-
ticipants were faster in their judgments when the
geometric test array preserved the configuration
information in the study array.

The results from the geometric condition
are more impressive when contrasted with the
results from the verbal condition, illustrated in
Figure 4.2b. Here, participants studied words ar-
ranged exactly as the objects in the geometric
condition were arranged. Because it involved
words, however, the study stimulus did not sug-
gest a face or have any pictorial properties. Santa

FIGURE 4.2 The procedure
followed in Santa’s (1977)
experiment demonstrating
that visual and verbal
information are represented
differently in mental images.
Participants studied an initial
array of objects or words and
then had to decide whether a
test array contained the same
elements. geometric shapes
were used in (a) and words
for the shapes in (b).

FIGURE 4.3 results from Santa’s
(1977) experiment. The data con-
firmed two of Santa’s hypotheses:
(1) In the geometric condition,
participants would make a positive
identification more quickly when
the configuration was identical
than when it was linear, because
the visual image of the study
stimulus would preserve spatial
information. (2) In the verbal con-
dition, participants would make a
positive identification more quickly
when the configuration was linear
than when it was identical, be-
cause participants had encoded
the words from the study array
linearly, in accordance with normal
reading order in English.

Geometric

Verbal

Re
ac

tio
n

tim
e

(s
) 1.25

1.15

Same
configuration

Linear
configuration

Study
array

arrays
Test

Test
arrays

Study
array

Identical,
same configuration

Same elements,
linear configuration

Different elements,
same configuration

Different elements,
linear configuration

Triangle Circle

Square

Triangle Circle

Square

Triangle Circle Square
Triangle Circle

Arrow

Triangle Circle Arrow

Identical,
same configuration

Same words,
linear configuration

Different words,
same configuration

Different words,
linear configuration

(a) Geometric condition

(b) Verbal condition

Anderson_8e_Ch04.indd 80 13/09/14 9:38 AM

V E r B A l I M A g E r y V E r S U S V I S U A l I M A g E r y / 81

speculated that participants would read the array left to right and top to bot-
tom and encode a verbal image with the information. So, given the study ar-
ray, participants would encode it as “triangle, circle, square.” After they stud-
ied the initial array, one of the test arrays was presented and participants had
to judge whether the words were identical. All the test stimuli involved words,
but otherwise they presented the same possibilities as the test stimuli in the
geometric condition. The two positive stimuli exemplify the same-configura-
tion condition and the linear-configuration condition. Note that the order of
words in the linear array was the same as it was in the study stimulus. Santa
predicted that, unlike the geometric condition, because participants had en-
coded the words into a linearly ordered verbal image, they would be fastest
when the test array was linear. As Figure 4.3 illustrates, his predictions were
again confirmed.

■ Different parts of the brain are involved in verbal and visual
imagery, and they represent and process information differently.

Using brain activation to read
people’s minds

Scientists are learning how to decode
the brain activity of people to deter-
mine what they are thinking. In one
of the most impressive examples of
this work, nishimoto et al. (2011)
reconstructed movies from the brain
activity of participants watching these
movies (the movie is on the left and
the reconstruction on the right). The
photos in this box shows examples
of the reconstructions—while blurry,
they capture some of the content
from the original videos. research-
ers have gone beyond this and
asked whether they can identify
participants’ internal thoughts. For
instance, is it possible to identify
the mental images a person is
experiencing? There has been some
success at this and, interestingly,
the brain areas involved seem to be
the same regions as are involved in
actual viewing of the images (Stokes,
Thompson, Cusack, & Duncan,
2009; Cichy, Heinzle, & Haynes,
2012). Other research has reported
success in identifying the concepts
participants are thinking about
(Mitchell et al., 2008) and what par-
ticipants are thinking while solving

an equation (J. r. Anderson, Betts,
Ferris, & Fincham, 2010). Could these
methods be used in interrogation
to determine what people are really
thinking and whether they are lying?
This question has been the subject
of debate, but the consensus is that
the methodology is a long way from
being reliable, and it has not been
allowed in court (read the Washington

Post article “Debate on Brain Scans
as lie Detectors Highlighted in Mary-
land Murder Trial”). not surprisingly,
such research has received a lot of
press—for instance, see the 60 Min-
utes report “reading your Mind” or
the PBS NewsHour report “It’s not
Mind-reading, but Scientists Explor-
ing How Brains Perceive the World,”
which you can find on youTube.

I m p l I c a t I o n s
▼

▲

Ni
sh

im
ot

o
et

a
l.,

20
11

. R
ep

rin
te

d
wi

th
p

er
m

iss
io

n
fro

m
E

lse
vie

r.

Anderson_8e_Ch04.indd 81 13/09/14 9:38 AM

82 / Chapter 4 M E n TA l I M A g E r y

◆ Visual Imagery

Most of the research on mental imagery has involved visual imagery, and this
will be the principal focus of this chapter. One function of mental imagery is
to anticipate how objects will look from different perspectives. People often
have the impression that they rotate objects mentally to change the perspective.
Roger Shepard and his colleagues were involved in a long series of experiments
on mental rotation. Their research was among the first to study the functional
properties of mental images, and it has been very influential. It is interesting to
note that this research was inspired by a dream (Shepard, 1967): Shepard awoke
one day and remembered having visualized a 3-D structure turning in space.
He convinced Jackie Metzler, a first-year graduate student at Stanford, to study
mental rotation, and the rest is history.

Their first experiment was reported in the journal Science (Shepard &
Metzler, 1971). Participants were presented with pairs of 2-D representations of
3-D objects, like those in Figure 4.4. Their task was to determine whether the
objects were identical except for orientation. In Figure 4.4a and Figure 4.4b, the
two objects are identical but are at different orientations. Participants reported
that to match the two shapes, they mentally rotated one of the objects in each
pair until it was congruent with the other object.

The graphs in Figure 4.5 show the times required for participants to decide
that the pairs were identical. The reaction times are plotted as a function of the
angular disparity between the two objects presented. The angular disparity is
the amount one object would have to be rotated to match the other object in
orientation. Note that the relationship is linear—for every increment in amount
of rotation, there is an equal increment in reaction time. Reaction time is plot-
ted for two different kinds of rotation. One is for 2-D rotations (Figure 4.5a),
which can be performed in the picture plane (i.e., by rotating the page); the
other is for depth rotations (Figure 4.5b), which require the participant to rotate
the object into the page. Note that the two functions are very similar. Processing
an object in depth (in three dimensions) does not appear to have taken longer
than processing an object in the picture plane. Hence, participants must have
been operating on 3-D representations of the objects in both the picture-plane
and depth conditions.

These data seem to indicate that participants rotated the object in a 3-D
space within their heads. The greater the angle of disparity between the two
objects, the longer participants took to complete the rotation. Though the par-
ticipants were obviously not actually rotating a real object in their heads, the
mental process appears to be analogous to physical rotation.

(b) (c)(a)

FIGURE 4.4 Stimuli in the Shepard and Metzler (1971) study on mental rotation. (a) The
objects differ by an 80° rotation in the picture plane (two dimensions). (b) The objects
differ by an 80° rotation in depth (three dimensions). (c) The objects cannot be
rotated into congruence. (From Shepard, R. N., & Metzler, J. (1971). Mental Rotation of
Three-Dimensional Objects. Science, 171. Copyright © 1971 American Association for the
Advancement of Science. Reprinted by permission.)

Mental Rotation

Anderson_8e_Ch04.indd 82 13/09/14 9:38 AM

V I S U A l I M A g E r y / 83

A great deal of subsequent research has examined the mental rotation of all
sorts of different objects, typically finding that the time required to complete a
rotation varies with the angle of disparity. There have also been a number of brain-
imaging studies that looked at what regions are active during mental rotation.
Consistently, the parietal region (roughly the region labeled R at the upper back
of the brain in Figure 4.1) has been activated across a range of tasks. This finding
corresponds with the results we reviewed in Chapter 3 showing that the parietal
region is important in spatial attention. Some tasks involve activation of other
areas. For instance, Kosslyn, DiGirolamo, Thompson, and Alpert (1998) found
that imagining the rotation of one’s hand produced activation in the motor cortex.

Neural recordings of monkeys have provided some evidence about neural
representation during mental rotation involving hand movement. Georgopoulos,
Lurito, Petrides, Schwartz, and Massey (1989) had monkeys perform a task in
which they moved a handle to a specific angle in response to a given stimulus. In
the base condition, monkeys just moved the handle to the position of the stimu-
lus. Georgopoulos et al. found cells that fired for particular positions. So, for in-
stance, there were cells that fired most strongly when the monkeys were moving
the handle to the 9 o’clock position and other cells that responded most strongly
when the monkeys moved it to the 12 o’clock position. In the rotation condition,
the monkeys had to move the handle to a position rotated some number of de-
grees from the stimulus. For instance, if the monkeys had to move the handle
90° counterclockwise from a stimulus at the 12 o’clock position, they would have
to move the handle to 9 o’clock. If the stimulus appeared at the 6 o’clock posi-
tion, they would have to move the handle to 3 o’clock. The greater the angle, the
longer it took the monkeys to initiate the movement, suggesting that this task in-
volved a mental rotation process. In this rotation condition, Georgopoulos et al.
found that various cells fired at different times during the transformation. At the
beginning of a trial, when the stimulus was presented, the cells that fired most
were associated with a move in the direction of the stimulus. By the end of the
trial, when the monkeys actually moved the handle, maximum activity occurred

Angle of rotation (degrees)

(a) (b)

0 40 80 120 160
0

1

2

4

3

5

M
ea

n
re

ac
tio

n
tim

e
(s

)

0

1

2

4

3

5

M
ea

n
re

ac
tio

n
tim

e
(s

)
16012080400

FIGURE 4.5 results of the Shepard and Metzler (1971) study on mental rotation. The
mean time required to determine that two objects have the same 3-D shape is plotted
as a function of the angular difference in their portrayed orientations. (a) Plot for pairs dif-
fering by a rotation in the picture plane (two dimensions). (b) Plot for pairs differing by a
rotation in depth (three dimensions). (Data from Metzler & Shepard, 1974.)

Anderson_8e_Ch04.indd 83 13/09/14 9:38 AM

84 / Chapter 4 M E n TA l I M A g E r y

in cells associated with the movement. Between the beginning and the end of a
trial, cells representing intermediate directions were most active. These results
suggest that mental rotation involves gradual shifts of firing from cells that en-
code the initial stimulus (the handle at its initial angle) to cells that encode the
response (the handle at its final angle).

■ When people must transform the orientation of a mental image to
make a comparison, they rotate its representation through the inter-
mediate positions until they achieve the desired orientation.

Image Scanning
Something else we often do with mental images is to scan them for critical in-
formation. For instance, when people are asked how many windows there are
in their houses (the task described at the beginning of this chapter), many re-
port mentally going through the house visually and scanning each room for
windows. Researchers have studied whether people are actually scanning per-
ceptual representations in such tasks, as opposed to just retrieving abstract in-
formation. For instance, are we really “seeing” each window in the room or are
we just remembering how many windows are in the room?

Brooks (1968) performed an important series of experiments on the scan-
ning of visual images. He had participants scan imagined diagrams such as the
one shown in Figure 4.6. For example, the participant was to scan around an
imagined block F from a prescribed starting point and in a prescribed direc-
tion, categorizing each corner of the block as a point on the top or bottom (as-
signed a yes response) or as a point in between (assigned a no response). In the
example (beginning with the starting corner), the correct sequence of responses
is yes, yes, yes, no, no, no, no, no, no, yes. For a nonvisual contrast task, Brooks
also gave participants sentences such as “A bird in the hand is not in the bush.”
Participants had to scan the sentence while holding it in memory, deciding
whether each word was a noun or not. A second experimental variable was how
participants made their responses. Participants responded in one of three ways:
(1) said yes or no; (2) tapped with the left hand for yes and with the right hand
for no; or (3) pointed to successive Y’s or N’s on a sheet of paper such as the one
shown in Figure 4.7. The two variables of stimulus material (diagram or sen-
tence) and output mode were crossed to yield six conditions.

Table 4.1 gives the results of Brooks’s experiment in terms of the mean
time spent in classifying the sentences or diagrams in each output mode.
The important result for our purposes is that participants took much longer
for diagrams in the pointing mode than in the other two modes, but this
was not the case when participants were working with sentences. Appar-
ently, scanning a physical visual array conflicted with scanning a mental
array. This result strongly reinforces the conclusion that when people are
scanning a mental array, they are scanning a representation that is analo-
gous to a physical picture.

One might think that Brooks’s result was due to the conflict between
engaging in a visual pointing task and scanning a visual image. Subsequent
research makes it clear, however, that the interference is not a result of the
visual character of the task per se. Rather, the problem is spatial and not spe-
cifically visual; it arises from the conflicting directions in which participants
had to scan the physical visual array and the mental image. For instance, in
another experiment, Brooks found evidence of similar interference when par-
ticipants had their eyes closed and indicated yes or no by scanning an array of
raised Y’s and N’s with their fingers. In this case, the actual stimuli were tac-
tile, not visual. Thus, the conflict is spatial, not specifically visual.

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

N

N

N

N

N

N

N

N

N

N

N

N

FIGURE 4.6 An example of a
simple block diagram that Brooks
used to study the scanning of
mental images. The asterisk and
arrow show the starting point and
the direction for scanning the
image. (From Brooks, 1968. Re-
printed by permission of the pub-
lisher. © 1968 by the Canadian
Psychological Association.)

FIGURE 4.7 A sample output
sheet of the pointing mode
in Brooks’s study of mental
image scanning. The letters
are staggered to force careful
visual monitoring of pointing.
(From Brooks, 1968. Reprinted by
permission of the publisher.
© 1968 by the Canadian Psycho-
logical Association.)

Anderson_8e_Ch04.indd 84 13/09/14 9:38 AM

V I S U A l I M A g E r y / 85

Baddeley and Lieberman (reported in
Baddeley, 1976) performed an experiment
that further supports the view that the
nature of the interference in the Brooks
task is spatial rather than visual. Par-
ticipants were required to perform two
tasks simultaneously. All participants
performed the Brooks letter-image task.
However, participants in one group simul-
taneously monitored a series of stimuli of
two possible brightness levels and had to
press a key whenever the brighter stimulus appeared. This task involved the
processing of visual but not spatial information. Participants in the other con-
dition were blindfolded and seated in front of a swinging pendulum. The pen-
dulum emitted a tone and contained a photocell and participants had to try to
keep the beam of a flashlight on the swinging pendulum. Whenever they were
on target, the photocell caused the tone to change frequency, thus providing
auditory feedback. This test involved the processing of spatial but not visual
information. The spatial auditory tracking task produced far greater impair-
ment in the image-scanning task than did the brightness judgment task. This
result also indicates that the nature of the impairment in the Brooks task was
spatial, not visual.

■ People suffer interference in scanning a mental image if they have
to simultaneously process a conflicting perceptual structure.

Visual Comparison of Magnitudes
A fair amount of research has focused on the way people judge the visual details
of objects in their mental images. One line of research has asked participants to
discriminate between objects based on some dimension such as size. This re-
search has shown that when participants try to discriminate between two ob-
jects, the time it takes them to do so decreases continuously as the difference in
size between the two objects increases.

Moyer (1973) was interested in the speed with which participants could
judge the relative size of two animals from memory. For example, “Which is
larger, moose or roach?” and “Which is larger, wolf or lion?” Many people report
that in making these judgments, particularly for the items
that are similar in size, they experience images of the two
objects and compare the sizes of the objects in their images.

Moyer also asked participants to estimate the abso-
lute size of these animals. Figure 4.8 plots the time required
to compare the imagined sizes of two animals as a function
of the difference between the two animals’ estimated sizes.
The individual points in Figure 4.8 represent comparisons
between pairs of items. In general, the judgment times de-
creased as the difference in estimated size increased. The
graph shows that judgment time decreases linearly with
increases in the difference between the sizes of the two ani-
mals. Note, however, that the differences have been plotted
logarithmically, which makes the distance between small
differences large relative to the same distances between large
differences. Thus, the linear relationship in the graph means
that increasing the size difference has a diminishing effect on
reaction time.

TABLE 4.1 results of Brooks’s (1968) Experiment Showing Conflict
Between Mental Array and Visual Array Scanning

Mean Response Time (s)
by Output Mode

Stimulus Material Pointing Tapping Vocal

Diagrams 28.2 14.1 11.3
Sentences 9.8 7.8 13.8

From Brooks, 1968. reprinted by permission of the publisher. © 1968 by
the Canadian Psychological Association.

1.5

1.4

1.3

1.2

1.1

1.0

0.9

0.8
0.10 1.10 2.10

Estimated difference in animal size

M
ea

n
re

ac
tio

n
tim

e
(s

)

FIGURE 4.8 results from
Moyer’s experiment demon-
strating that when people try to
discriminate between two objects
on the basis of size, the time it
takes them to do so decreases
as the difference in size between
the two objects increases. Par-
ticipants were asked to compare
the imagined sizes of two ani-
mals. The mean time required
to judge which of two animals is
larger is plotted as a function of
the estimated difference in size
of the two animals. The differ-
ence measure is plotted on the
abscissa in a logarithmic scale.
(Data from Moyer, 1973.)

Anderson_8e_Ch04.indd 85 13/09/14 9:38 AM

86 / Chapter 4 M E n TA l I M A g E r y

Significantly, very similar results are obtained when peo-
ple visually compare physical size. For instance, D. M. Johnson
(1939) asked participants to judge which of two simultaneously
presented lines was longer. Figure 4.9 plots participant judgment
time as a function of the log difference in line length, and again,
a linear relation is obtained. It is reasonable to expect that the
more similar the lengths being compared are, the longer percep-
tual judgments will take, because telling them apart is more dif-
ficult under such circumstances. The fact that similar functions
are obtained when mental objects are compared indicates that
making mental comparisons involves the same processes as those
involved in perceptual comparisons.

■ People experience greater difficulty in judging the
relative size of two pictures or of two mental images
that are similar in size.

Are Visual Images Like Visual Perception?
Can people recognize patterns in mental images in the same way that they rec-
ognize patterns in things they actually see? In an experiment designed to inves-
tigate this question, Finke, Pinker, and Farah (1989) asked participants to create
mental images and then engage in a series of transformations of those images.
Here are two examples of the problems that they read to their participants:

● Imagine a capital letter N. Connect a diagonal line from the top right cor-
ner to the bottom left corner. Now rotate the figure 90° to the right. What
do you see?

● Imagine a capital letter D. Rotate the figure 90° to the left. Now place a
capital letter J at the bottom. What do you see?

Participants closed their eyes and tried to imagine these transformations
as they were read to them. The participants were able to recognize their com-
posite images just as if they had been presented with them on a screen. In the
first example, they saw an hourglass; in the second, an umbrella. The ability to
perform such tasks illustrates an important function of imagery: It enables us
to construct new objects in our minds and inspect them. It is just this sort of
visual synthesis that structural engineers or architects must perform as they de-
sign new bridges or buildings.

Chambers and Reisberg (1985) reported a study that seemed to indicate dif-
ferences between a mental image and visual perception of the real object. Their re-
search involved the processing of reversible figures, such as the duck-rabbit shown
in Figure 4.10. Participants were briefly shown the figure and asked to form an
image of it. They had only enough time to form one interpretation of the picture
before it was removed, but they were asked to try to find a second interpretation.

Participants were not able to do this. Then they were asked to draw the im-
age on paper to see whether they could reinterpret it. In this circumstance,
they were successful. This result suggests that mental images differ from pic-
tures in that one can interpret visual images only in one way, and it is not
possible to find an alternative interpretation of the image.

Subsequently, Peterson, Kihlstrom, Rose, and Gilsky (1992) were able
to get participants to reverse mental images by giving them more explicit
instructions. For instance, participants might be told how to reverse
another figure or be given the instruction to consider the back of the head
of the animal in their mental image as the front of the head of another
animal. Thus, it seems apparent that although it may be more difficult

1.4

M
ea

n
re

ac
tio

n
tim

e
(s

)

Difference in line length (mm)

2.3

3.2

2.9

1.7

2.0

2.6

1 2 3 4 6 8

FIGURE 4.9 results from the
D. M. Johnson (1939) study in
which participants compared the
lengths of two lines. The mean
time required to judge which
line was longer is plotted as a
function of the difference in line
length. The difference measure is
plotted on the abscissa in a loga-
rithmic scale. These results, which
are very similar to the results of
the Moyer (1973) experiment
shown in Figure 4.8, demonstrate
that making mental comparisons
involves difficulties of discrimina-
tion similar to those involved in
making perceptual comparisons.

FIGURE 4.10 The ambiguous
duck-rabbit figure used in Cham-
bers and reisberg’s study of the
processing of reversible figures.
(From Chambers & Reisberg, 1985.
Reprinted by permission of the
publisher. © 1985 by the American
Psychological Association.)

Anderson_8e_Ch04.indd 86 13/09/14 9:38 AM

V I S U A l I M A g E r y / 87

to reverse an image than a picture, both can be reversed. In general, it seems
harder to process an image than the actual stimulus. Given a choice, people will
almost always choose to process an actual picture rather than imagine it. For
instance, players of Tetris prefer to rotate shapes on the screen to find an ap-
propriate orientation rather than rotate them mentally (Kirsh & Maglio, 1994).

■ It is possible to make many of the same kinds of detailed judgments
about mental images that we make about things we actually see,
though it is more difficult.

Visual Imagery and Brain Areas
Brain-imaging studies indicate that the same regions are involved in perception
as in mental imagery. As already noted, the parietal regions that are involved in
attending to locations and objects (see Chapter 3) are also involved in mental ro-
tation. O’Craven and Kanwisher (2000) performed an experiment that further il-
lustrates how closely the brain areas activated by imagery correspond to the brain
areas activated by perception. As discussed in Chapters 2 and 3, the fusiform face
area (FFA) in the temporal cortex responds preferentially to faces, and another
region of the temporal cortex, the parahippocampal place area (PPA), responds
preferentially to pictures of locations. O’Craven and Kanwisher asked participants
either to view faces and scenes or to imagine faces and scenes. The same areas
were active when the participants were seeing as when they were imagining. As
shown in Figure 4.11, every time the participants viewed or imagined a face, there
was increased activation in the FFA, and this activation went away when they

−0.5

0.0

Si
gn

al
c

ha
ng

e
(%

)
Si

gn
al

c
ha

ng
e

(%
)

0.5

1.0

1.5

2.0

−1.0
−0.5

0.0

0.5
1.0

1.5

2.0
2.5

3.0

Perception Imagery ImageryPerception

FFA

PPA

FIGURE 4.11 results from the O’Craven and Kanwisher study showing that visual im-
ages are processed in the same way as actual perceptions and by many of the same
neural structures. Participants alternately perceived (or imagined) faces and places, and
brain activation was correspondingly seen in the fusiform face area (FFA, upper panel) or
the parahippocampal place area (PPA, lower panel). (From O’Craven & Kanwisher, 2000.
Reprinted by permission of the publisher. © 2000 by the Journal of Cognitive neuroscience.)

Anderson_8e_Ch04.indd 87 13/09/14 9:38 AM

88 / Chapter 4 M E n TA l I M A g E r y

processed places. Conversely, when they viewed or imagined scenes, there was
activation in the PPA that went away when they processed faces. The responses
during imagery were very similar to the responses during perception, although a
little weaker. The fact that the response was weaker during imagery is consistent
with the behavioral evidence we have viewed suggesting that it is more difficult to
process an image than a real perception.

There are many studies like these that show that cortical regions involved in
high-level visual processing are activated during the processing of visual imagery.
However, the evidence is less clear about activation in the primary visual cortex
(areas 17 and 18) where visual information first reaches the brain. The O’Craven
and Kanwisher study did find activation in the primary visual cortex during
imagery. Such results are important because they suggest that visual imagery
includes relatively low-level perceptual processes. However, activation has not
always been found in the primary visual cortex. For instance, the Roland and
Friberg study illustrated in Figure 4.1 did not find activation in this region (see
also Roland, Eriksson, Stone-Elander, & Widen, 1987). Kosslyn and Thompson
(2003) reviewed 59 brain-imaging studies that looked for activation in early vis-
ual areas. About half of these studies find activation in early visual areas and half
do not. Their analysis suggests that the studies that find activation in these early
visual areas tend to emphasize high-resolution details of the images and tend to
focus on shape judgments. As an instance of one of the positive studies, Kosslyn
et al. (1993) did find activation in area 17 in a study where participants were
asked to imagine block letters. In one of their experiments, participants were
asked to imagine large versus small letters. In the small-letter condition, activity
in the visual cortex occurred in a more posterior region, closer to where the
center of the visual field is represented. This makes sense because a small image
would be more concentrated at the center of the visual field.

Imaging studies like these show that perceptual regions of the brain are
active when participants engage in mental imagery, but they do not estab-
lish whether these regions are actually critical to imagery. To return to the
epiphenomenon critique at the beginning of the chapter, it could be that the
activation plays no role in the actual tasks being performed. A number of experi-
ments have used transcranial magnetic stimulation (TMS—see Figure 1.13) to
investigate the causal role of these regions in the performance of the underlying
task. For instance, Kosslyn et al. (1999) presented participants with 4-quadrant
arrays like those in Figure 4.12 and asked them to form a mental image of the
array. Then, with the array removed, participants had to use their image to an-
swer questions like “Which has longer stripes: Quadrant 1 or Quadrant 2?” or
“Which has more stripes: Quadrant 1 or Quadrant 4?” Application of TMS to
primary visual area 17 significantly increased the time they took to answer these

questions. Thus, it seems that these visual regions do play a
causal role in mental imagery, and temporarily deactivating
them results in impaired information processing.

■ Brain regions involved in visual perception are
also involved in visual imagery tasks, and disrup-
tion of these regions results in disruption of the
imagery tasks.

Imagery Involves Both Spatial
and Visual Components
There is an important distinction to be made between the
spatial and visual attributes of imagery. We can encode
the position of objects in space by seeing where they are,

1

3

2

4

FIGURE 4.12 Illustration of stim-
uli used in Kosslyn et al. (1999).
The numbers 1, 2, 3, and 4 were
used to label the four quadrants,
each of which contained a set of
stripes. After memorizing the dis-
play, the participants closed their
eyes, visualized the entire display,
heard the names of two quad-
rants, and then heard the name
of a comparison term (for exam-
ple, “length”); the participants
then decided whether the stripes
in the first-named quadrant had
more of the named property
than those in the second.

Anderson_8e_Ch04.indd 88 13/09/14 9:38 AM

V I S U A l I M A g E r y / 89

by feeling where they are, or by hearing where they are. Such encodings use a
common spatial representation that integrates information that comes in from
any sensory modality. On the other hand, certain aspects of visual experience,
such as color, are unique to the visual modality and seem separate from spatial
information. Imagery involves both spatial and visual components. In the dis-
cussion of the visual system in Chapter 2, we reviewed the evidence that there is
a “where” pathway for processing spatial information and a “what” pathway for
processing object information (see Figure 2.1). Corresponding to this distinc-
tion, there is evidence (Mazard, Fuller, Orcutt, Bridle, & Scanlan, 2004) that the
parietal regions support the spatial component of visual imagery, whereas the
temporal lobe supports the visual aspects. We have already noted that mental
rotation, a spatial task, tends to produce activation in the parietal cortex. Simi-
larly, temporal structures are activated when people imagine visual properties
of objects (Thompson & Kosslyn, 2000).

Studies of patients with brain damage also support this association of spatial
imagery with parietal areas of the brain and visual imagery with temporal ar-
eas. Levine, Warach, and Farah (1985) compared two patients, one who suffered
bilateral parietal-occipital damage and the other who suffered bilateral inferior
temporal damage. The patient with parietal damage could not describe the loca-
tions of familiar objects or landmarks from memory, but he could describe the
appearance of objects. The patient with temporal damage had an impaired abil-
ity to describe the appearance of objects but could describe their locations.

Farah, Hammond, Levine, and Calvanio (1988) carried out more detailed
testing of the patient with temporal damage, comparing his performance on
a wide variety of imagery tasks to that of normal participants. They found
that he showed deficits in only a subset of these tasks: ones in which he had to
judge color (“What is the color of a football?”), sizes (“Which is bigger, a pop-
sicle or a pack of cigarettes?”), the lengths of animals’ tails (“Does a kangaroo
have a long tail?”), and whether two U.S. states had similar shapes. In contrast,
he did not show any deficit in performing tasks that seemed to involve a sub-
stantial amount of spatial processing: mental rotation, image scanning, letter
scanning (as in Figure 4.7), or judgments of where one U.S. state was relative to
another state. Thus, temporal damage seems to affect only those imagery tasks
that required access to visual detail, not those that required spatial judgments.

■ Neuropsychological evidence suggests that imagery of spatial
information is supported by parietal structures, and that imagery
of objects and their visual properties is supported by temporal
structures.

Cognitive Maps
Another important function of visual imagery is to help us understand and re-
member the spatial structure of our environment. Our imaginal representations
of the world are often referred to as cognitive maps. The connection between
imagery and action is particularly apparent in cognitive maps. We often find
ourselves imagining our environment as we plan how we will get from one loca-
tion to another.

An important distinction can be made between route maps and survey
maps (Hart & Moore, 1973). A route map is a path that indicates specific places
but contains no spatial information. It can even be a verbal description of a path
(“Straight until the light, then turn left, two blocks later at the intersection . . .”).
Thus, with a pure route map, if your route from location 1 to location 2 were
blocked, you would have no general idea of where location 2 was, and so you
would be unable to construct a detour. Also, if you knew (in the sense of a route

Anderson_8e_Ch04.indd 89 13/09/14 9:38 AM

90 / Chapter 4 M E n TA l I M A g E r y

map) two routes from a location, you would have no idea whether these routes
formed a 90° angle or a 120° angle with respect to each other. A survey map, in
contrast, contains this information, and is basically a spatial image of the en-
vironment. When you ask for directions from typical online mapping services,
they will provide both a route map and a survey map to support both mental
representations of space.

Thorndyke and Hayes-Roth (1982) investigated workers’ knowledge of the
Rand Corporation Building (Figure 4.13), a large, mazelike building in Santa
Monica, California. People in the Rand Building quickly acquire the ability to
find their way from one specific place in the building to another—for example,
from the supply room to the cashier. This knowledge represents a route map.
Typically, though, workers had to have years of experience in the building
before they could make such survey-map determinations as the direction of the
snack bar from the administrative conference room (due south).

Hartley, Maguire, Spiers, and Burgess (2003) used fMRI to look at differ-
ences in brain activity when people used these two representations. They had
participants navigate virtual reality towns under one of two conditions: route-
following (involving a route map) or way-finding (involving a survey map).
In the route-following condition, participants learned to follow a fixed path
through the town, whereas in the way-finding condition, participants first
freely explored the town and then had to find their way between locations.
The results on the experiment are illustrated in Color Plate 4.1. In the way-
finding task, participants showed greater activation in a number of regions
found in other studies of visual imagery, including the parietal cortex. There
was also greater activation in the hippocampus (see Figure 1.7), a region that
has been implicated in navigation in many species. In contrast, in the route-
following task participants showed greater activation in more anterior regions
and motor regions. It would seem that the survey map is more like a visual

Cognitive Mapping

Northwest
lobby

Computer center

Administrative
conference

room

East lobby

Cashier

Snack bar

Common
room

South
lobby

First-floor building plan
The Rand Corporation

Supply
room

Feet

500 100

N

S

EW

FIGURE 4.13 The floor plan for part of the rand Corporation Building in Santa Monica,
California. Thorndyke and Hayes-roth studied the ability of secretaries to find their way
around the building. (From Thorndyke & Hayes-Roth, 1982. Reprinted by permission of the
publisher. © 1982 by Cognitive Psychology.)

Anderson_8e_Ch04.indd 90 13/09/14 9:38 AM

V I S U A l I M A g E r y / 91

image and the route map is more like an action plan. This is a distinction that
is supported in other fMRI studies of route maps versus survey maps (e.g.,
Shelton & Gabrieli, 2002).

Landmarks serve as an important part of survey maps and enable flexible
action. Using a virtual environment navigation system, Foo, Warren, Duchon,
and Tarr (2005) performed an experiment that used the presence of landmarks
to promote creation of different types of mental maps. In the “desert” condition
(see Figure 4.14a) there were no landmarks and participants practiced navigat-
ing from a home position to two target locations. In the “forest” condition (see
Figure 4.14b) there were “trees” and participants practiced navigating from the
same home position to the same two target locations. Then they were asked
to navigate from one of the target locations to the other, having never done so
before. They were very poor at finding the novel path in the “desert” condition
because they had not practiced that path. They were much better in the “forest”
condition, where colored posts could serve as landmarks.

■ Our knowledge of our environment can be represented in either
survey maps that emphasize spatial information or route maps that
emphasize action information.

Egocentric and Allocentric Representations of Space
Navigation becomes difficult when we must tie together multiple different
representations of space. In particular, we often need to relate the way space
appears as we perceive it to some other representation of space, such as a cog-
nitive map. The representation of “space as we perceive it” is referred to as an
egocentric representation. Figure 4.15 illustrates an egocentric representation
that one might have when looking through the cherry blossoms at the Tidal Ba-
sin in Washington, D.C. Even young children have little difficulty understanding
how to navigate in space as they see it—if they see an object they want, they go
for it. Problems arise when one wants to relate what one sees to such representa-
tions of the space as cognitive maps, be they route maps or survey maps. Similar
problems arise when one wants to deal with physical maps, such as the map of

FIGURE 4.14 Displays used in the virtual reality study of Foo et al. (2005). The desert
world (a) consisted of a textured ground plane only, whereas the forest world (b) in-
cluded many colored posts scattered randomly throughout. The colored posts served as
potential landmarks. (Foo, Warren, Duchon, & Tarr, 2005. © American Psychological Associa-
tion, reprinted with permission.)

(a) (b)

Anderson_8e_Ch04.indd 91 13/09/14 9:38 AM

92 / Chapter 4 M E n TA l I M A g E r y

the park area in Figure 4.16. This kind of map is referred to as an allocentric
representation because it is not specific to a particular viewpoint, though, as
is true of most maps, north is oriented to the top of the image. Using the map
in Figure 4.16, assuming the perspective of the stick figure, try to identify the
building in Figure 4.15. When people try to make such judgments, the degree to
which the map is rotated from their actual viewpoint has a large effect. Indeed,
people will often rotate a physical map so that it is oriented to correspond to
their point of view. The map in Figure 4.16 would have to be rotated almost 180
degrees to be oriented with the representation shown in Figure 4.15.

When it is not possible to rotate a map physically, people show an effect
of the degree of misorientation that is much like the effect we see for mental
rotation (e.g., Boer, 1991; Easton & Sholl, 1995; Gugerty, deBoom, Jenkins, &

FIGURE 4.15 An egocentric view from the Tidal Basin. (Stock/360/Getty Images.)

FIGURE 4.16 An allocentric representation of Washington’s national Mall and Memorial
Parks. (National Park Service.)

Anderson_8e_Ch04.indd 92 13/09/14 9:39 AM

V I S U A l I M A g E r y / 93

Morley, 2000; Hintzman, O’Dell, & Arndt,
1981). Figure 4.17 shows results from a study
by Gunzelmann and Anderson (2002), who
looked at the time required to find an object
on a standard map (i.e., north oriented to the
top) as a function of the viewer’s location.
When the viewer is located to the south,
looking north, it is easier to find the object
than when the viewer is north looking south,
just the opposite of the map orientation. Some
people describe imagining themselves moving
around the map, others talk about rotating
what they see, and still others report using
verbal descriptions (“across the water”). The
fact that the angle of disparity in this task has
as great an effect as it does in mental rotation has led many researchers to believe
that the processes and representations involved in such navigational tasks are
similar to the processes and representations involved in mental imagery.

Physical maps seem to differ from cognitive maps in one important way:
Physical maps show the effects of orientation, and cognitive maps do not. For
example, imagine yourself standing against various walls of your bedroom, and
point to the location of the front door of your home or apartment. Most people
can do this equally well no matter which position they take. In contrast, when
given a map like the one in Figure 4.16, people find it much easier to point to
various objects on the map if they are oriented in the same way the map is.

Recordings from single cells in the hippocampal region (inside the tempo-
ral lobe) of rats suggest that the hippocampus plays an important role in main-
taining an allocentric representation of the world. There are place cells in the
hippocampus that fire maximally when the animal is in a particular location in
its environment (O’Keefe & Dostrovsky, 1971). Similar cells have been found in
recordings from human patients during a procedure to map out the brain before
surgery to control epilepsy (Ekstrom et al., 2003). Brain-imaging studies have
shown high hippocampal activation when humans are navigating their environ-
ment (Maguire et al., 1998). Another study (Maguire et al., 2000) showed that
the hippocampal volume of London taxi drivers was greater than that of people
who didn’t drive taxis. The longer they had been taxi drivers, the greater the vol-
ume of their hippocampus. It took about 3 years of hard training to gain enough
knowledge of London streets to be a successful taxi driver, and this training had
an impact on the structure of the brain. The amount of activation in hippocam-
pal structures has also been shown to correlate with age-related differences in
navigation skills (Pine et al., 2002) and may relate to gender differences in navi-
gational ability (Gron, Wunderlich, Spitzer, Tomczak, & Riepe, 2000).

Whereas the hippocampus appears to be important in supporting al-
locentric representations, the parietal cortex seems particularly important
in supporting egocentric representations (Burgess, 2006). In one fMRI study
comparing egocentric and allocentric spatial processing (Zaehle et al., 2007),
participants were asked to make judgments that emphasized either an allo-
centric or an egocentric perspective. In the allocentric conditions, partici-
pants would read a description like “The blue triangle is to the left of the green
square. The green square is above the yellow triangle. The yellow triangle is
to the right of the red circle.” Then they would be asked a question like “Is the
blue triangle above the red circle?” In the egocentric condition, they would
read a description like “The blue circle is in front of you. The yellow circle is
to your right. The yellow square is to the right of the yellow circle.” They would
then be asked a question like “Is the yellow square to your right?” There was

SE SENENNWSW WS
0.0

0.5

1.5

1.0

2.0

2.5

3.0

3.5

Ti
m

e
to

fi
nd

o
bj

ec
t o

n
m

ap
(s

)

Direction in which viewer is looking

FIGURE 4.17 results from
gunzelmann and Anderson’s
study to determine how much
effect the angle of disparity
between a standard map (looking
north) and the viewer’s view-
point has on people’s ability to
find an object on the map. The
time required for participants to
identify the object is plotted as
a function of the difference in
orientation between the map and
the egocentric viewpoint. (Data
from Gunzelmann & Anderson,
2002.)

Anderson_8e_Ch04.indd 93 13/09/14 9:39 AM

94 / Chapter 4 M E n TA l I M A g E r y

greater hippocampal activation when participants were answering questions in
the allocentric condition than in the egocentric condition. Although there was
considerable parietal activation in both conditions, it was greater in the ego-
centric condition.

■ Our representation of space includes both allocentric representa-
tions of where objects are in the world and egocentric representations
of where they are relative to ourselves.

Map Distortions
Our mental maps often have a hierarchical structure in which smaller regions
are organized within larger regions. For instance, the structure of my bedroom
is organized within the structure of my house, which is organized within the
structure of my neighborhood, which is organized within the structure of Pitts-
burgh. Consider your mental map of the United States. It is probably divided
into regions, and these regions into states, and cities are presumably pinpointed
within the states. It turns out that certain systematic distortions arise because
of the hierarchical structure of these mental maps. Stevens and Coupe (1978)
documented a set of common misconceptions about North American geography.
Consider the following questions taken from their research:

● Which is farther east: San Diego or Reno?
● Which is farther north: Seattle or Montreal?
● Which is farther west: the Atlantic or the Pacific entrance to the Panama

Canal?

The first choice is the correct answer in each case, but most people hold the
opposite opinion. Reno seems to be farther east because Nevada is east of Cali-
fornia, but this reasoning does not account for the westward curve in Califor-
nia’s coastline. Montreal seems to be north of Seattle because Canada is north of
the United States, but the border dips south in the east. And the Atlantic is cer-
tainly east of the Pacific—but consult a map if you need to be convinced about
the location of the entrances to the Panama Canal. The geography of North
America is quite complex, and people resort to abstract facts about relative
locations of large physical bodies (e.g., California and Nevada) to make judg-
ments about smaller locations (e.g., San Diego and Reno).

Stevens and Coupe were able to demonstrate such confusions with
experimenter-created maps. Different groups of participants learned the maps
illustrated in Figure 4.18. The important feature of the incongruent maps is
that the relative locations of the Alpha and Beta counties are inconsistent with
the locations of the X and Y cities. After learning the maps, participants were
asked a series of questions about the locations of cities, including “Is X east or
west of Y?” for the left-hand maps and “Is X north or south of Y?” for the right-
hand maps. Participants made errors on 18% of the questions for the congruent
maps, 15% for the homogeneous maps, but 45% for the incongruent maps.
Participants were using information about the locations of the counties to help
them remember the city locations. This reliance on higher order information
led them to make errors, just as similar reasoning can lead to errors in
answering questions about North American geography.

■ When people have to work out the relative positions of two loca-
tions, they will often reason in terms of the relative positions of larger
areas that contain the two locations.

Anderson_8e_Ch04.indd 94 13/09/14 9:39 AM

C O n C l U S I O n S : V I S U A l P E r C E P T I O n A n D V I S U A l I M A g E r y / 95

Dimension tested

Alpha
County

Beta
County

Congruent

Horizontal

X

Y

Z

Vertical

Alpha
County

Beta
County

X

Y

Z

X
Y

Z

X

Y

Z

X Y

Z

X

Y

Z

Alpha
County

Beta
County

Alpha
County

Beta
County

Homogeneous

Incongruent

W
S

E
N

W
S

E
N

W
S

E
N

W
S

E
N

W
S

E
N

W
S

E
N

◆ Conclusions: Visual Perception and
Visual Imagery

This chapter has reviewed some of the evidence that the same brain regions
that are involved in visual perception are also involved in visual imagery. Such
research has presumably put to rest the question raised at the beginning of the
chapter about whether visual imagery really had a perceptual character. How-
ever, although it seems clear that perceptual processes are involved in visual
imagery to some degree, it remains an open question to what degree the mech-
anisms of visual imagery are the same as the mechanisms of visual perception.

FIGURE 4.18 Maps studied by participants in the experiments of Stevens and Coupe,
which demonstrated the effects of higher order information (location of county lines) on
participants’ recall of city locations. (Data from Stevens & Coupe, 1978.)

Anderson_8e_Ch04.indd 95 13/09/14 9:39 AM

96 / Chapter 4 M E n TA l I M A g E r y

Evidence for a substantial overlap comes from neuropsychological patient stud-
ies (see Bartolomeo, 2002, for a review). Many patients who have cortical damage
leading to blindness have corresponding deficits in visual imagery. As Behrmann
(2000) notes, the correspondences between perception and imagery can be quite
striking. For instance, there are patients who are not able to perceive or image
faces and colors, but are otherwise unimpaired in either perception or imagery.
Nonetheless, there exist cases of patients who suffer perceptual problems but have
intact visual imagery and vice versa. Behrmann argues that visual perception and
visual imagery are best understood as two processes that overlap but are not iden-
tical, as illustrated in Figure 4.19. Perceiving a kangaroo requires low-level visual
information processing that is not required for visual imagery. Similarly, forming
a mental image of a kangaroo requires generation processes that are not required
by perception. Behrmann suggests that patients who suffer only perceptual losses
have damage to the low-level part of this system, and patients who suffer only im-
agery losses have damage to the high-level part of this system.

High-level
generation

ImageryPerception

Low-level
visual analysis

Intermediate
visual processing

FIGURE 4.19 A represen-
tation of the overlap in
the processing involved in
visual perception and visual
imagery.

Questions for Thought

1. It has been hypothesized that our perceptual sys-
tem regularly uses mental rotation to recognize
objects in nonstandard orientations. In Chapter 2
we contrasted template and feature models for ob-
ject recognition. Would mental rotation be more
important to a template model or a feature model?

2. Consider the following problem:
Imagine a wire-frame cube resting on a tabletop
with the front face directly in front of you and per-
pendicular to your line of sight. Imagine the long
diagonal that goes from the bottom, front, left-
hand corner to the top, back, right-hand one. Now
imagine that the cube is reoriented so that this
diagonal is vertical and the cube is resting on one
corner. Place one fingertip about a foot above the
tabletop and let this mark the position of the top
corner on the diagonal. The corner on which the
cube is resting is on the tabletop, vertically below
your fingertip. With your other hand, point to the
spatial locations of the other corners of the cube.

Hinton (1979) reports that almost no one is able
to perform this task successfully. In light of the
successes we have reviewed for mental imagery,
why is this task so hard?

3. The chapter reviewed the evidence that many
different regions are activated in mental imagery
tasks—parietal and motor areas in mental rota-
tion, temporal regions in judgments of object
attributes, and hippocampal regions in reasoning
about navigation. Why would mental imagery in-
volve so many regions?

4. Consider the map distortions such as the tendency
to believe San Diego is west of Reno. Are these
distortions in an egocentric representation, an al-
locentric representation, or something else?

5. The studies of the increased size of the hippocam-
pus of London taxi drivers were conducted before
the widespread introduction of GPS systems in
cars. Would the results be different for taxi drivers
who made extensive use of GPS systems?

Key Terms

allocentric representation
cognitive maps
egocentric representation

epiphenomenon
fusiform face area (FFA)
mental imagery

mental rotation
parahippocampal place

area (PPA)

route maps
survey maps

Anderson_8e_Ch04.indd 96 13/09/14 9:39 AM

97

Recall a wedding you attended a while ago. Presumably, you can remember who
married whom, where the wedding was, many of the people who attended,

and some of the things that happened. You would probably be hard pressed, how-
ever, to say exactly what all the participants wore, the exact words that were spo-
ken, or the way the bride walked down the aisle, although you probably registered
many of these details. It is not surprising that our memories lose information over
time, but what is interesting is that our loss of information is selective: We tend
to forget the less significant and remember the more significant aspects of what
happened.

The previous chapter was about our ability to form detailed visual images. It
might seem that it would be ideal if we had the capacity to remember such detail.
Parker, Cahill, and McGaugh (2006) describe a case of an individual with highly
detailed memory.1 She is able to remember many details from years ago in her
life but had difficulty in school and seems to perform poorly on tasks of abstract
reasoning such as processing analogies. A more recent study of 11 such individuals
(LePort et al., 2012) finds that although they can remember an enormous amount
of detail from their personal lives, they are no better than average on many
standard laboratory memory tasks. They probably would not do better than others
in remembering the information from a text like this. It seems like their memories
are bogged down in remembering insignificant details, without any special ability to
remember critical information.

In many situations, we need to rise above the details of our experience and get
to their true meaning and significance. Understanding how we do this is the focus of
this chapter, where we will address the following questions:

● How do we represent the significant aspects of our experience?
● Do we represent knowledge in ways that are not tied to specific perceptual

modalities?
● How do we represent categorical knowledge, and how does this affect the way

we perceive the world?

◆ Knowledge and Regions of the Brain

Figure 5.1 shows some of the brain regions involved in the abstraction of
knowledge. Some prefrontal regions are associated with extracting mean-
ingful information from pictures and sentences. The left prefrontal region is

5
Representation of
Knowledge

1 She has written her own biography, The Woman Who Can’t Forget (Price, 2008).

Anderson_8e_Ch05.indd 97 13/09/14 9:40 AM

98 / Chapter 5 R e P R e S e n TAT I o n o f K n o W L e D G e

more involved in the processing of verbal material and
the right prefrontal region is more involved in the pro-
cessing of visual material (Gabrieli, 2001). There is also
strong evidence that categorical information is repre-
sented in posterior regions, particularly the temporal
cortex (Visser, Jeffries, & Ralph, 2010). When this infor-
mation is presented verbally, there is also fairly consistent
evidence for greater activation throughout the left hemi-
sphere (e.g., Binder, Desai, Graves, & Conant, 2009).

At points in this chapter, we will review neurosci-
ence data on the localization of semantic information
in the brain, but our focus will be on the striking results
from behavioral studies that examine what people re-
member or forget after an event.

■ Prefrontal regions of the brain are associated with meaningful pro-
cessing of events, whereas posterior regions, such as the temporal cor-
tex, are associated with representing categorical information.

◆ Memory for Meaningful Interpretations
of Events

Memory for Verbal Information
A dissertation study by Eric Wanner (1968) illustrates circumstances in which
people do and do not remember information about exact wording. Wanner asked
participants to come into the laboratory and listen to tape-recorded instructions.
For one group of participants, the warned group, the tape began this way:

The materials for this test, including the instructions, have been
recorded on tape. Listen very carefully to the instructions because you
will be tested on your ability to recall particular sentences which occur
in the instructions.

The participants in the second group received no such warning and so had no
idea that they would be responsible for the verbatim instructions. After this
point, the instructions were the same for both groups. At a later point in the
instructions, one of four possible critical sentences was presented:

1. When you score your results, do nothing to correct your answers but mark
carefully those answers which are wrong.

2. When you score your results, do nothing to correct your answers but care-
fully mark those answers which are wrong.

3. When you score your results, do nothing to your correct answers but mark
carefully those answers which are wrong.

4. When you score your results, do nothing to your correct answers but
carefully mark those answers which are wrong.

Note that some sentences differ in style but not in meaning (sentences 1 and 2,
and 3 and 4), whereas other sentences differ in meaning but not in style (sen-
tences 1 and 3, and 2 and 4), and that each of these pairs differ only in the or-
dering of two words. Immediately after one of these sentences was presented, all
participants (warned or not) heard the following conclusion to the instructions:

To begin the test, please turn to page 2 of the answer booklet and judge
which of the sentences printed there occurred in the instructions you
just heard.

Prefrontal regions
that process pictures
and sentences

Posterior regions
that represent
concepts

Brain Structures

FIGURE 5.1 Cortical regions
involved in the processing of
meaning and the representation
of concepts.

Anderson_8e_Ch05.indd 98 13/09/14 9:40 AM

M e M o R Y f o R M e A n I n G f U L I n T e R P R e TAT I o n S o f e v e n T S / 99

On page 2, they found two sentences: the critical sentence
they had just heard and a sentence that differed just in
style or just in meaning. For example, if they had heard
sentence 1, they might have to choose between sentences
1 and 2 (different in style but not in meaning) or between
sentences 1 and 3 (different in meaning but not in style).
Thus, by looking at participants’ ability to discriminate
between different pairs of sentences, Wanner was able to
measure their ability to remember the meaning versus the
style of the sentence and to determine how this ability was
affected by whether or not they were warned.

The relevant data are presented in Figure 5.2. The
percentage of correct identifications of sentences heard is displayed as a func-
tion of whether participants had been warned. The percentages are plotted
separately for participants who were asked to discriminate a meaningful differ-
ence in wording and for those who were asked to discriminate a stylistic differ-
ence. If participants were just guessing, they would have scored 50% correct by
chance; thus, we would not expect any values below 50%.

The implications of Wanner’s experiment are clear. First, memory is
better for changes in wording that result in changes of meaning than for
changes in wording that result just in changes of style. The superiority of
memory for meaning indicates that people normally extract the meaning
from a linguistic message and do not remember its exact wording. Moreover,
memory for meaning is equally good whether people are warned or not.
(The slight advantage for unwarned participants does not approach statistical
significance.) Thus, participants retained the meaning of a message as a
normal part of their comprehension process. They did not have to be cued to
remember the sentence.

The second implication of these results is that people are capable of
remembering exact wording if that is their goal—the warning did have an effect
on memory for the stylistic change. The unwarned participants remembered
the stylistic change at about the level of chance, whereas the warned partici-
pants remembered it almost 80% of the time. Thus, although we do not nor-
mally retain much information about exact wording, we can do so when we are
cued to pay attention to such information.

■ After processing a linguistic message, people usually remember just
its meaning and not its exact wording.

Memory for Visual Information
Our memory for visual information often seems much better than our memory
for verbal information. Shepard (1967) performed one of the early experi-
ments comparing memory for pictures with memory for verbal material. In the
picture-memory task, participants first studied a set of magazine pictures one at
a time, then were presented with pairs of pictures consisting of one picture they
had studied and one they had not, and then had to indicate which picture had
been studied. In the sentence-memory task, participants studied sentences one
at a time and were similarly tested on their ability to recognize those sentences.
Participants made errors on the verbal task 11.8% of the time but only 1.5%
of the time on the visual task. In other words, memory for verbal information
was quite good, but memory for visual information was virtually perfect. Many
subsequent experiments have demonstrated our high capacity for remembering
pictures. For example, Brady, Konkle, Alvarez, and Oliva (2008) had partici-
pants first study a set of 2,500 pictures and then identify individual pictures

Memory for meaning

Memory for style

WarnedUnwarned

Co
rre

ct
(%

)

50

60

70

80

90

100

FIGURE 5.2 Results from
Wanner’s experiment to deter-
mine circumstances in which
people do and do not remember
information about exact wording.
The ability of participants to
remember a wording difference
that affected meaning versus one
that affected only style is plotted
as a function of whether or not
the participants were warned
that they would be tested on
their ability to recall particular
sentences. (Data from Wanner,
1968.)

Anderson_8e_Ch05.indd 99 13/09/14 9:40 AM

100 / Chapter 5 R e P R e S e n TAT I o n o f K n o W L e D G e

from the set when paired with a similar alternative (Color Plate 5.1 shows some
of these pairs). Participants were able to achieve almost 87.5% accuracy in mak-
ing such discriminations.

However, people do not always show such good memory for pictures—it
depends on the circumstances. Nickerson and Adams (1979) performed a clas-
sic study showing lack of memory for visual detail. They asked American stu-
dents to indicate which of the pictures in Figure 5.3 was the actual U.S. penny.
Despite having seen this object literally thousands of times, they were not able
to identify the actual penny. What is the difference between studies showing
good memory for visual detail and a study like this one, showing poor memory
for visual detail? It seems that the answer is that the details of the penny are not
something people attend to. In the experiments showing good visual memory,
the participants are told to attend to the details. The role of attention was con-
firmed in a study by Marmie and Healy (2004) following up on the Nickerson
and Adams study. Participants examined a novel coin for a minute and then,
a week later, were asked to remember the details. In this study, participants
achieved much higher accuracy than in the penny study.

How do people actually deploy their attention when studying a complex
visual scene? Typically, people attend to, and remember, what they consider to
be the meaningful or important aspects of the scene. This is illustrated in an
experiment by Mandler and Ritchey (1977) in which participants studied pic-
tures of scenes like the classroom scenes in Figure 5.4. After studying eight such
pictures for 10 s each, participants were presented with a series of pictures and
asked to identify the pictures they had studied. The series included the exact
pictures they had studied (target pictures) as well as distracter pictures, which
included token distracters and type distracters. A token distracter differed from
the target only in a relatively unimportant visual detail (e.g., the pattern of the
teacher’s clothes in Figure 5.4b is an unimportant detail). In contrast, a type

FIGURE 5.3 examples of the pennies used in the experiment by nickerson and Adams
(1979)—which is the real penny? (From Nickerson, R. S., & Adams, M. J. (1979). Long-term
memory for a common object. Cognitive Psychology, 11(3), 287–307. Copyright © 1979
Elsevier. Reprinted by permission.)

Anderson_8e_Ch05.indd 100 13/09/14 9:40 AM

M e M o R Y f o R M e A n I n G f U L I n T e R P R e TAT I o n S o f e v e n T S / 101

distracter differed from the target in a relatively important visual detail (e.g.,
the art picture in Figure 5.4c—instead of the world map in the target—is an im-
portant detail because it indicates the subject being taught). Participants rec-
ognized the original pictures 77% percent of the time and rejected the token
distracters only 60% of the time, but they rejected the type distracters 94% of
the time.

The conclusion in this study is very similar to that in the Wanner (1968) ex-
periment reviewed earlier. Wanner found that participants were much more sen-
sitive to meaning-significant changes in a sentence; Mandler and Ritchey (1977)
found that participants were more sensitive to meaning-significant changes in
a picture and not for details in the picture. This is not because they are inca-
pable of remembering such detail, but rather because this detail does not seem
important and so is not attended. Had participants been told that the picture il-
lustrated the style of the teacher’s clothing, the result would probably have been
quite different.

■ When people see a picture, they attend to and remember best those
aspects that they consider meaningful.

Importance of Meaning to Memory
So far we have considered memory for meaningful verbal and pictorial material.
However, what if the material is not meaningful, such as a hard-to-follow

(a)

(b) (c)

FIGURE 5.4 Pictures similar to those used by Mandler and Ritchey in their experiment
to demonstrate that people distinguish between the meaning of a picture and the physi-
cal picture itself. Participants studied the target picture (a). Later they were tested with a
series of pictures that included the target (a) along with token distracters such as (b) and
type distracters such as (c). (After Mandler & Ritchey, 1977. Adapted by permission of the
publisher. © 1977 by the American Psychological Association.)

Anderson_8e_Ch05.indd 101 13/09/14 9:40 AM

102 / Chapter 5 R e P R e S e n TAT I o n o f K n o W L e D G e

written description? Consider the following passage that was used in a study by
Bransford and Johnson (1972):

The procedure is actually quite simple. First you arrange items into
different groups. Of course, one pile may be sufficient depending
on how much there is to do. If you have to go somewhere else due
to lack of facilities that is the next step, otherwise you are pretty well
set. It is important not to overdo things. That is, it is better to do
too few things at once than too many. In the short run this may not
seem important but complications can easily arise. A mistake can be
expensive as well. At first the whole procedure will seem complicated.
Soon, however, it will become just another facet of life. It is difficult
to foresee any end to the necessity for this task in the immediate fu-
ture, but then one never can tell. After the procedure is completed

one arranges the materials into different groups again. Then they
can be put into their appropriate places. Eventually they will be
used once more and the whole cycle will then have to be repeated.
However, that is part of life. (p. 722)

Presumably, you find this description hard to make sense of; the
participants did, too, and showed poor recall on the passage. How-
ever, another group of participants were told before reading this
passage that it was about washing clothes. With that one piece of
information, which made the passage much more sensible, they
were able to recall twice as much as the uninformed group.

Similar effects are found in memory for pictorial material. One
study (Goldstein & Chance, 1970) compared memory for faces versus
memory for snowflakes. Individual snowflakes are highly distinct from
one another and more visually different than faces (see Figure 5.5).
However, participants do not know what sense to make of snowflakes,
whereas they are often capable of interpreting subtle differences in
faces. In a test 48 hours later, participants were able to recognize 74%
of the faces and only 30% of the snowflakes. In another study, provoca-
tively titled “Sometimes a Picture Is Not Worth a Single Word,” Oates
and Reder (2010) compared recognition memory for words with rec-
ognition memory for abstract pictures like those in Figure 5.6. They
found that recognition memory for these pictures was quite poor—
only half as good as their memory for words.

Bower, Karlin, and Dueck (1975) reported an amusing demonstra-
tion of the fact that people’s good memory for pictures is tied to their

FIGURE 5.5 examples of the snowflakes that Goldstein and Chance (1970) used in their
memory experiment. (Herbert/Stringer/Archive Photos/Getty Images)

FIGURE 5.6 examples of the
abstract pictures that participants
had a hard time remembering
in the experiment by oates and
Reder. (From Oates & Reder, 2010.
Copyright © 2010. Reprinted by
permission of Lynne Reder.)

Anderson_8e_Ch05.indd 102 13/09/14 9:40 AM

M e M o R Y f o R M e A n I n G f U L I n T e R P R e TAT I o n S o f e v e n T S / 103

ability to make sense of those pictures. Figure 5.7
illustrates some of the drawings they used, called
droodles. Participants studied the drawings, with
or without an explanation of their meaning, and
then were given a memory test in which they had
to redraw the pictures. Participants who had been
given an explanation when studying the pictures
showed better recall (70% correctly reconstructed)
than those who were not given an explanation (51%
correctly reconstructed). Thus, memory for the
drawings depended critically on participants’ ability
to give them a meaningful interpretation.

■ Memory is better for material if we are
able to meaningfully interpret that material.

Implications of Good Memory for Meaning
We have seen that people have relatively good memory for meaningful inter-
pretations of information. So when faced with material to remember, it will
help if they can give it some meaningful interpretation. Unfortunately, many
people are unaware of this fact, and their memory performance suffers as a
consequence. I can still remember the traumatic experience I had in my first
paired-associates experiment. It happened in a sophomore class in experi-
mental psychology. For reasons I have long since forgotten, we had designed
a class experiment that involved learning 16 pairs, such as DAX-GIB. Our
task was to recall the second half of each pair when prompted with the first
half and I was determined to outperform the other members of my class.
My personal theory of memory at that time, which I intended to apply, was
that if you try hard and focus intensely, you can remember anything well. In
the impending experimental situation, this meant that during the learning
period I should say (as loud as was seemly) the paired associates over and
over again, as fast as I could. I believed that this method would burn the
paired associates into my mind forever. To my chagrin, I wound up with the
worst score in the class.

My theory of “loud and fast” was directly opposed to the true means of im-
proving memory. I was trying to memorize a meaningless verbal pair. But the
material discussed in this chapter so far suggests that we have the best memory
for meaningful information. I should have been trying to convert my memory
task into something more meaningful. For instance, DAX is like dad and GIB
is the first part of gibberish. So I might have created an image of my father
speaking some gibberish to me. This would have been a simple mnemonic
(memory-assisting) technique and would have worked quite well as a means of
associating the two elements.

We do not often need to learn pairs of nonsense syllables outside the
laboratory. In many situations, however, we do have to associate various com-
binations that do not have much inherent meaning. We have to remember
shopping lists, names for faces, telephone numbers, rote facts in a college class,
vocabulary items in a foreign language, and so on. In all cases, we can im-
prove memory if we associate the items to be remembered with a meaningful
interpretation.

■ It is easier to commit arbitrary associations to memory if they are
converted into something more meaningful.

(a) (b)

FIGURE 5.7 Recalling “droodles.”
(a) A midget playing a trombone
in a telephone booth. (b) An early
bird that caught a very strong
worm. (From Bower, G. H., Karlin,
M. B., & Dueck, A. (1975). Compre-
hension and memory for pictures.
Memory & Cognition, 3, 216–220.
Copyright © 1975 Springer. With
kind permission from Springer
Science and Business Media.)

Anderson_8e_Ch05.indd 103 13/09/14 9:40 AM

104 / Chapter 5 R e P R e S e n TAT I o n o f K n o W L e D G e

◆ Propositional Representations

We have shown that in many situations people do not remember exact physical
details of what they have seen or heard but rather the “meaning” of what they
have encountered. In an attempt to become more precise about what is meant
by “meaning,” cognitive psychologists developed what is called a propositional
representation. The concept of a proposition, borrowed from logic and lin-
guistics, is central to such analyses. A proposition is the smallest unit of knowl-
edge that can stand as a separate assertion—that is, the smallest unit one can
meaningfully judge as true or false. Propositional analysis applies most clearly
to linguistic information, and I will develop the topic here in terms of such in-
formation.

Consider the following sentence:

Lincoln, who was president of the United States during a bitter war,
freed the slaves.

The information conveyed in this sentence can be communicated by the follow-
ing simpler sentences:

A. Lincoln was president of the United States during a war.
B. The war was bitter.
C. Lincoln freed the slaves.

Mnemonic techniques for
remembering vocabulary items

one domain where we seem to
have to learn arbitrary associations
is foreign language vocabulary. for
instance, consider trying to learn that
the Italian formaggio (pronounced
“for-MAH-jo”) means cheese. There
is a memorization technique, called
the keyword method, for learning
vocabulary items, which some stu-
dents are taught and others discover
on their own. The first step is to
convert the foreign word to some
sound-alike term in one’s native
language. for example, we might
convert formaggio into “for much
dough.” The second step is to create
a meaningful connection between
the sound-alike and the meaning.
for example, we might imagine ex-
pensive cheese being sold for much
money or “for much dough.” or con-
sider the Italian carciofi (pronounced

“car-CHoH-fee”), which means
artichokes. We might transform “car-
CHoH-fee” into “car trophy” and
imagine a winning car at an auto
show with a trophy shaped like an
artichoke. The intermediate sound-
alike term (e.g., “for much dough” or
“car trophy”) is called the keyword,
although in both of these examples
they are really key phrases. There
has been extensive research on
the effectiveness of this technique

(for a review, read Kroll & De Groot,
2005). The research shows that, as
with many things, one needs to take
a nuanced approach in evaluating
the effectiveness of the keyword
technique. There is no doubt that
it results in more rapid vocabulary
learning in many situations, but
there are potential costs. one might
imagine that having to go through
the intermediate keyword slows
down the speed of translation, and
the keyword method has been
shown to result in slower retrieval
times compared to retrieval of items
that are directly associated without
an intermediate. Moreover, going
through an intermediate has been
shown to result in poorer long-term
retention. finally, evidence sug-
gests that although the method
may help in passing an immediate
vocabulary test in a class and hurt
in a delayed test that we have not
studied for, its ultimate impact on
achieving real language mastery is
minimal. Chapter 12 will discuss
issues involved in foreign language
mastery.

I m p l I c a t I o n s
▼

Te
d

Ta
m

bu
ro

/G
et

ty
Im

ag
es

.

▲

Anderson_8e_Ch05.indd 104 13/09/14 9:40 AM

P R o P o S I T I o n A L R e P R e S e n TAT I o n S / 105

If any of these simple sentences were false, the complex sentence also would be
false. These sentences correspond closely to the propositions that underlie the
meaning of the complex sentence. Each simple sentence expresses a primitive
unit of meaning. Like these simple sentences, each separate unit composing our
meaning representations must correspond to a unit of meaning.

However, the theory of propositional representation does not claim that
a person remembers simple sentences like these when encoding the meaning
of a complex sentence. Rather, the claim is that the material is encoded in a
more abstract way. For instance, the propositional representation proposed by
Kintsch (1974) represents each proposition as a list containing a relation fol-
lowed by an ordered list of arguments. The relations organize the arguments
and typically correspond to the verbs (in this case, free), adjectives (bitter), and
other relational terms (president of ). The arguments refer to particular times,
places, people, or objects, and typically correspond to the nouns (Lincoln, war,
slaves). The relations assert connections among the entities these nouns refer
to. Kintsch represents each proposition by a parenthesized list consisting of a
relation plus arguments. As an example, sentences A through C would be rep-
resented by these following structures Kintsch called propositions:

A. (president-of: Lincoln, United States, war)
B. (bitter: war)
C. (free: Lincoln, slaves)

Note that each relation takes a different number of arguments: president of takes
three, free takes two, and bitter takes one. Whether a person heard the original
complex sentence or heard

The slaves were freed by Lincoln, the president of the United States
during a bitter war.

the meaning of the message would be represented by propositions a through c.
Bransford and Franks (1971) provided an interesting demonstration of the

psychological reality of propositional units. In this experiment, participants
studied 12 sentences, including the following:

The ants ate the sweet jelly, which was on the table.
The rock rolled down the mountain and crushed the tiny hut.
The ants in the kitchen ate the jelly.
The rock rolled down the mountain and crushed the hut beside the
woods.
The ants in the kitchen ate the jelly, which was on the table.
The tiny hut was beside the woods.
The jelly was sweet.

The propositional units in each of these sentences come from one of two sets of
four propositions. One set can be represented as

1. (eat: ants, jelly, past)
2. (sweet: jelly)
3. (on: jelly, table, past)
4. (in: ants, kitchen, past)

The other set of four propositions can be represented as

1. (roll down: rock, mountain, past)
2. (crush: rock, hut, past)
3. (beside: hut, woods, past)
4. (tiny: hut)

Anderson_8e_Ch05.indd 105 13/09/14 9:40 AM

106 / Chapter 5 R e P R e S e n TAT I o n o f K n o W L e D G e

Bransford and Franks looked at participants’ recognition memory for the
following three kinds of sentences:

1. Old: The ants in the kitchen ate the jelly.
2. New: The ants ate the sweet jelly.
3. Noncase: The ants ate the jelly beside the woods.

The first sentence was actually studied. The second sentence was not studied
but consists of a combination of propositions that occurred in the studied
sentences—that is, (eat: ants, jelly, past) and (sweet: jelly) from above. The third
sentence consists of words that were studied (beside, jelly, woods, past), but is not
composed from the propositions that were studied—for example, (beside, jelly,
woods) is a new proposition. Bransford and Franks found that participants had
almost no ability to discriminate between the first two kinds of sentences and
were likely to say that they had actually heard either. On the other hand, partici-
pants were quite confident that they had not heard the third, noncase, sentence.

The experiment shows that although people remember the propositions
they encounter, they are quite insensitive to the actual combination of proposi-
tions. Indeed, the participants in this experiment were most likely to say that
they heard a sentence consisting of all four propositions, such as

The ants in the kitchen ate the sweet jelly, which was on the table.

even though they had not in fact studied this sentence.

■ According to propositional analyses people remember a complex
sentence as a set of abstract meaning units that represent the simple
assertions in the sentence.

Amodal Versus Perceptual Symbol Systems
The propositional representations that we have just considered are examples of
what Barsalou (1999) called an amodal symbol system. By this he meant that
the elements within the system are inherently nonperceptual. The original stim-
ulus might be a picture or a sentence, but the representation is abstracted away
from the verbal or visual modality. Given this abstraction, one would predict
that participants in experiments would be unable to remember the exact words
they heard or the exact picture they saw.

As an alternative to such theories, Barsalou proposed a hypothesis that
he called the perceptual symbol system. This hypothesis claims that all
information is represented in terms that are specific to a particular perceptual
modality (visual, auditory, etc.). The perceptual symbol hypothesis is an ex-
tension of Paivio’s (1971, 1986) earlier dual-code theory that claimed that we
represent information in combined verbal and visual codes. Paivio suggested
that when we hear a sentence, we also develop a visual image of what it
describes. If we later remember the visual image and not the sentence, we will
remember what the sentence was about, but not its exact words. Analogously,
when we see a picture, we might describe to ourselves the significant features
of that picture. If we later remember our description and not the picture, we
will not remember details we did not think important to describe (such as the
clothes the teacher was wearing in Figure 5.4).

The dual-code position does not predict that memory for the wording
of a sentence is necessarily poor. The relative memory for the wording versus
memory for the meaning depends on the relative attention that people give to
the verbal versus the visual representation. There are a number of experiments
showing that when participants pay attention to wording, they show better
memory. For instance, Holmes, Waters, and Rajaram (1998), in a replication of

Anderson_8e_Ch05.indd 106 13/09/14 9:40 AM

P R o P o S I T I o n A L R e P R e S e n TAT I o n S / 107

the Bransford and Franks (1971) study that we just reviewed, asked participants
to count the number of letters in the last word of each sentence. This
manipulation, which increased their attention to the wording of the sentence,
resulted in an increased ability to discriminate sentences they had studied from
sentences with similar meanings that they had not—although participants still
showed considerable confusion among similar-meaning sentences.

But how can an abstract concept such as honesty be represented in a purely
perceptual cognitive system? One can be very creative in combining perceptual
representations. Consider a pair of sentences from an old unpublished study of
mine.2 We had participants study one of the following two sentences:

1. The lieutenant wrote his signature on the check.
2. The lieutenant forged a signature on the check.

Later, we asked participants to recognize which sentence they had studied.
They could make such discriminations more successfully than they could dis-
tinguish between pairs such as

1. The lieutenant enraged his superior in the barracks.
2. The lieutenant infuriated a superior in the barracks.

In the first pair of sentences, there is a big difference in meaning; in the second
pair, little difference. However, the difference in wording between the sentences
in the two pairs is equivalent. When I did the study, I thought it showed that
people could remember meaning distinctions that did not have perceptual
differences—the distinction between signing a signature and forging is not
in what the person does but in his or her intentions and the relationship
between those intentions and unseen social contracts. Barsalou (personal
communication, March 12, 2003) suggested that we represent the distinction
between the two sentences by reenacting the history behind each sentence. So
even if the actual act of writing and forging might be the same, the history of
what a person said and did in getting to that point might be different. Barsalou
also considers the internal state of the individual to be relevant. Thus, the
perceptual features involved in forging might include the sensations of tension
that one has when one is in a difficult situation.3

Barsalou, Simmons, Barbey, and Wilson (2003) cited evidence that when
people understand a sentence, they actually come up with a perceptual repre-
sentation of that sentence. For instance, in one study by Stanfield and Zwaan
(2001), participants read a sentence about a nail being pounded into either the
wall or the floor. Then they viewed a picture of a nail oriented either horizon-
tally or vertically and were asked whether the object in the picture was men-
tioned in the sentence that they just read. If they had read a sentence about a
nail being pounded into the wall, they recognized a horizontally oriented nail
more quickly. When they had read a sentence about a nail being pounded
into the floor, they recognized a vertically oriented nail more quickly. In
other words, they responded faster when the orientation implied by the sen-
tence matched the orientation of the picture. Thus, their representation of the
sentence seemed to contain this perceptual detail. As further evidence of the
perceptual representation of meaning, Barsalou et al. cited neuroscience stud-
ies showing that concepts are represented in brain areas similar to those that
process perceptions.

2 It was not published because at the time (1970s) it was considered too obvious a result given studies like
those described earlier in this chapter.
3 Perhaps it is obvious that I do not agree with Barsalou’s perspective. However, it is hard to imagine what
he might consider disconfirming data, because his approach is so flexible.

Anderson_8e_Ch05.indd 107 13/09/14 9:40 AM

108 / Chapter 5 R e P R e S e n TAT I o n o f K n o W L e D G e

■ An alternative to amodal representations of meaning is the view
that meaning is represented as a combination of images in different
perceptual modalities.

◆ Embodied Cognition

Barsalou’s perceptual symbol hypothesis is an instance of the growing emphasis
in psychology on understanding the contribution of the environment and our
bodies to shaping our cognition. As Thelen (2000) describes the viewpoint:

To say that cognition is embodied means that it arises from bodily in-
teractions with the world and is continually meshed with them. From
this point of view, therefore, cognition depends on the kinds of experi-
ences that come from having a body with particular perceptual and
motor capabilities that are inseparably linked and that together form
the matrix within which reasoning, memory, emotion, language and
all other aspects of mental life are embedded. (p. 5)

The embodied cognition perspective emphasizes the contribution of motor
action and how it connects us to the environment. For instance, Glenberg
(2007) argues that our understanding of language often depends on covertly
acting out what the language describes. He points to an fMRI study by Hauk,
Johnsrude, and Pulvermuller (2004), who recorded brain activation while peo-
ple listened to verbs that involved face, arm, or leg actions (e.g., to lick, pick, or
kick). They looked for activity along the motor cortex in separate regions as-
sociated with the face, arm, and leg (see Figure 1.10). Figure 5.8 shows that as
participants listened to each word, there was greater activation in the part of the
motor cortex that would produce that action.

A theory of how meaning is represented in the human mind must explain
how different perceptual and motor modalities connect with one another. For
instance, part of understanding a word such as kick is our ability to relate it to
a picture of a person kicking a ball so that we can describe that picture. As an-
other example, part of our understanding of someone performing an action is
our ability to relate to our own motor system so that we can mimic the action.
Interestingly, mirror neurons have been found in the motor cortex of monkeys;
these are active when the monkeys perform an action like ripping a paper or
see the experimenter rip a paper or hear the experimenter rip the paper without
seeing the action (Rizzolatti & Craighero, 2004). Although one cannot typically

FaceRegion:

M
R

sig
na

l c
ha

ng
e

(a
rb

itr
ar

y
un

its
)

Arm Leg

0.02

0

0.04

0.06

0.08

0.1

Leg words

Face words

Arm words

FIGURE 5.8 Brain activation in
different motor regions as par-
ticipants listen to different types
of verbs.

Anderson_8e_Ch05.indd 108 13/09/14 9:40 AM

C o n C e P T U A L K n o W L e D G e / 109

do single-cell recordings with humans, brain-imaging studies have found
increased activity in the motor region when people observe actions, particularly
with the intention to mimic the action (Iacoboni et al., 1999).

Figure 5.9 illustrates two conceptions of how mappings might take place
between different representations. One possibility is illustrated in the multimodal
hypothesis, which holds that we have various representations tied to different
perceptual and motor systems and that we have means of directly converting one
representation to another. For instance, the double-headed arrow going from
the visual to the motor would be a system for converting a visual representation
into a motor representation and a system for converting the representations
in the opposite direction. The alternative amodal hypothesis is that there is
an intermediate abstract “meaning” system, perhaps involving propositional
representations like those we described earlier. According to this hypothesis, we
have systems for converting any type of perceptual or motor representation into
an abstract representation and for converting any abstract representation into
any type of perceptual or motor representation. So to convert a representation
of a picture into a representation of an action, one first converts the visual
representation into an abstract representation of its significance and then converts
that representation into a motor representation. These two approaches offer
alternative explanations for the research we reviewed earlier that indicated people
remember the meaning of what they experience, but not the details. The amodal
hypothesis holds that this information is retained in the central meaning system.
The multimodal hypothesis holds that the person has converted the information
from the modality of the presentation to some other modality.

■ The embodied cognition perspective emphasizes that meaning is
represented in the perceptual and motor systems that we use to inter-
act with the world.

◆ Conceptual Knowledge

When we look at the picture in Figure 5.4a, we do not see it as just a collection
of specific objects. Rather, we see it as a picture of a teacher instructing a stu-
dent on geography. That is, we see the world in terms of categories like teacher,
student, instruction, and geography. As we saw, people tend to remember this
categorical information and not the specific details. For instance, the partici-
pants in the Mandler and Ritchey (1977) experiment forgot what the teacher
wore but remembered the subject she taught.

You cannot help but experience the world in terms of the categories you know.
For example, if you were licked by a four-legged furry object that weighed about
50 pounds and had a wagging tail, you would perceive yourself as being licked by

Multimodal Hypothesis Amodal Hypothesis

Visual

Other

Verbal

Motor

Visual

Other

Verbal

Motor

Meaning

(a) (b)

FIGURE 5.9 Representations
of two hypotheses about how
information is related between
different perceptual and motor
modalities. (a) The multimodal
hypothesis holds that there
are mechanisms for translating
between each modality. (b) The
amodal hypothesis holds that
each modality can be translated
back and forth to a central
meaning representation.

Anderson_8e_Ch05.indd 109 13/09/14 9:40 AM

110 / Chapter 5 R e P R e S e n TAT I o n o f K n o W L e D G e

a dog. What does your cognitive system gain by categorizing the object as a dog?
Basically, it gains the ability to predict. Thus, you can have expectations about what
sounds this creature might make and what would happen if you threw a ball (the
dog might chase it and stop licking you). Because of this ability to predict, categories
give us great economy in representation and communication. For instance, if you
tell someone, “I was licked by a dog,” your listener can predict the number of legs on
the creature, its approximate size, and so on.

The effects of such categorical perceptions are not always positive—for
instance, they can lead to stereotyping. In one study, Dunning and Sherman
(1997) had participants study sentences such as

Elizabeth was not very surprised upon receiving her math SAT score.

or

Bob was not very surprised upon receiving his math SAT score.

Participants who had heard the first sentence were more likely to falsely believe
they had heard “Elizabeth was not very surprised upon receiving her low math
SAT score,” whereas if they had heard the second sentence, they were more
likely to believe they had heard “Bob was not very surprised upon receiving
his high math SAT score.” Categorizing Elizabeth as a woman, the participants
brought the stereotype of women as poor at math to their interpretation of the
first sentence. Categorizing Bob as male, they brought the opposite stereotype
to their interpretation of the second sentence. This was even true among par-
ticipants (both male and female) who were rated as not being sexist in their
attitudes. They could not help but be influenced by their implicit stereotypes.

Research on categorization has focused both on how we form these
categories in the first place and on how we use them to interpret experiences.
It has also been concerned with notations for representing this categorical
knowledge. In this section, we will consider a number of proposed notations
for representing conceptual knowledge. We will start by describing two early
theories, one proposing semantic networks and the other proposing schemas.
Both theories have been closely related to certain empirical phenomena that
seem central to conceptual structure.

■ The categorical organization of our knowledge strongly influences
the way we encode and remember our experiences.

Semantic Networks
Quillian (1966) proposed that people store information about various categories—
such as canaries, robins, fish, and so on—in a network structure like that shown
in Figure 5.10. In this illustration, we represent a hierarchy of categorical facts,
such as that a canary is a bird and a bird is an animal, by linking nodes for the two
categories with isa links. Properties that are true of the categories are associated
with them. Properties that are true of higher-level categories are also true of lower
level categories. Thus, because animals breathe, it follows that birds and canaries
breathe. Figure 5.10 can also represent information about exceptions. For instance,
even though most birds fly, the illustration does represent that ostriches cannot fly.

Collins and Quillian (1969) did an experiment to test the psychological
reality of such networks by having participants judge the truth of assertions
about concepts, such as

1. Canaries can sing.
2. Canaries have feathers.
3. Canaries have skin.

Anderson_8e_Ch05.indd 110 13/09/14 9:40 AM

C o n C e P T U A L K n o W L e D G e / 111

Participants were shown these along with false assertions, such as “apples
have feathers,” and they had to judge which were true and which were false.
The false assertions were mainly to keep participants “honest”; Collins and
Quillian were really interested in how quickly participants could judge true
assertions like sentences 1 through 3, above.

Consider how participants would answer such questions if Figure 5.10
represented their knowledge of such categories. The information needed to
confirm sentence 1 is directly stored with canary. The information for sen-
tence 2, however, is not directly stored with canary; instead, the property of
having feathers is stored with bird. Thus, confirming sentence 2 requires
making an inference from two pieces of information in the hierarchy: a canary
is a bird and birds have feathers. Similarly, the information needed to confirm
sentence 3 is not directly stored with canary; rather, the property of having
skin is stored with animal. Thus, confirming sentence 3 requires making an
inference from three pieces of information in the hierarchy: a canary is a bird,
a bird is an animal, and animals have skin. In other words, to verify sentence 1,
participants would just have to look at the information stored with canary; for
sentence 2, participants would need to traverse one link, from canary to bird;
and for sentence 3, they would have to traverse two links, from canary to bird
and from bird to animal.

If our categorical knowledge were structured like Figure 5.10, we would ex-
pect sentence 1 to be verified more quickly than sentence 2, which would be
verified more quickly than sentence 3. This is just what Collins and Quillian
found. Participants required 1,310 ms to judge statements like sentence 1; 1,380
ms to judge statements like sentence 2; and 1,470 ms to judge statements like
sentence 3. Subsequent research on the retrieval of information from memory
has somewhat complicated the conclusions drawn from the initial Collins and
Quillian experiment. How often facts are experienced has been observed to
have strong effects on retrieval time (e.g., C. Conrad, 1972). Some facts, such

Level 1

Level 2

Level 3 Canary
Can sing

Is yellow

Bird
Has wings

Can fly
Has feathers

Animal

Has skin
Can move around
Eats

Breathes

Ostrich

Has long
thin legs
Is tall

Can’t fly

Shark

Can bite

Is dangerous

Fish

Has fins

Can swim
Has gills

Salmon

Is pink

Is edible

Swims
upstream
to lay eggs

FIGURE 5.10 A hypothetical memory structure for a three-level hierarchy using the exam-
ple canary. Quillian (1966) proposed that people store information about various catego-
ries in a network structure. This illustration represents a hierarchy of categorical facts, such
as that a canary is a bird and a bird is an animal. Properties that are true of each category
are associated with that category. Properties that are true of higher level categories are
also true of lower level categories. (Adapted from Collins, A. M., & Quillian, M. R. (1969).
Retrieval time from semantic memory. Journal of verbal Learning and verbal Behavior, 8,
240–247. Copyright © 1969 by Academic Press. Reprinted by permission.)

Anderson_8e_Ch05.indd 111 13/09/14 9:40 AM

112 / Chapter 5 R e P R e S e n TAT I o n o f K n o W L e D G e

as apples are eaten—for which the predicate could be stored with an intermedi-
ate concept such as food, but that are experienced quite often—are verified as
fast as or faster than facts such as apples have dark seeds, which must be stored
more directly with the apple concept. It seems that if a fact about a concept is
encountered frequently, it will be stored with that concept, even if it could also
be inferred from a more general concept. The following statements about the
organization of facts in semantic memory and their retrieval times seem to be
valid conclusions from the research:

1. If a fact about a concept is encountered frequently, it will be stored with
that concept even if it could be inferred from a higher order concept.

2. The more frequently a fact about a concept is encountered, the more
strongly that fact will be associated with the concept. The more strongly
facts are associated with concepts, the more rapidly they are verified.

3. Inferring facts that are not directly stored with a concept takes a relatively
long time.

■ When a property is not stored directly with a concept, people can
retrieve it from a higher order concept.

Schemas
Consider the many things we know about houses, such as

● Houses are a type of building.
● Houses have rooms.
● Houses can be built of wood, brick, or stone.
● Houses serve as human dwellings.
● Houses tend to have rectilinear and triangular shapes.
● Houses are usually larger than 100 square feet and smaller than

10,000 square feet.

The importance of a category is that it stores predictable information about
specific instances of that category. So when someone mentions a house, for
example, we have a rough idea of the size of the object being referred to.

Semantic networks, which just store properties with concepts, cannot cap-
ture the nature of our general knowledge about a house, such as its typical size
or shape. Researchers in cognitive science (e.g., Rumelhart & Ortony, 1976)
proposed a particular way of representing such knowledge that seemed more
useful than the semantic network representation. Their representational struc-
ture is called a schema. The concept of a schema was first articulated in AI and
computer science. Readers who have experience with modern programming
languages should recognize its similarity to various types of data structures.
The question for the psychologist is: What aspects of the schema notion are ap-
propriate for understanding how people reason about concepts? I will describe
some of the properties associated with schemas and then discuss the psycho-
logical research bearing on these properties.

Schemas represent categorical knowledge according to a slot structure, in
which slots are attributes that members of a category possess, and each slot is
filled with one or more values, or specific instances, of that attribute. So we have
the following partial schema representation of a house: