Main Sensation & perception

Sensation & perception

Year: 1989
Edition: 3rd
Publisher: Elsevier Australia
Language: english
Pages: 636
ISBN 10: 0155796496
ISBN 13: 978-0155796492
File: PDF, 35.79 MB
Download (pdf, 35.79 MB)
Preview
 
You can write a book review and share your experiences. Other readers will always be interested in your opinion of the books you've read. Whether you've loved the book or not, if you give your honest and detailed thoughts then people will find new books that are right for them.
1

The Little Book of Algorithms

Year: 2019
Language: english
File: PDF, 1.84 MB
2

Web

Year: 1980
Language: english
File: EPUB, 186 KB
SENSATION &
PERCEPTION

Stanley Coren
. Lawrence M. Ward

rvoT

do
-

SENSATION &
PERCEPTION
THIRD EDITION

SENSATION &
PERCEPTION
THIRD EDITION

Stanley Coren
University of British Columbia

Lawrence M. Ward
University of British Columbia

Based on previous editions
that included contributions by

Clare Porac
University of Victoria

Oil

Harcourt Brace Jovanovich, Publishers
San Diego New York Chicago Austin Washington, D.C.
London Sydney Tokyo Toronto

Cover photo: Eye, by Geoff Gove/Image Bank
Copyright © 1989, 1984 by Harcourt Brace Jovanovich, Inc.
All rights reserved. No part of this publication may be reproduced or
transmitted in any form or by any means, electronic or mechanical, including
photocopy, recording, or any information storage and retrieval system,
without permission in writing from the publisher.
Requests for permission to make copies of any part of the work should be
mailed to: Permissions, Harcourt Brace Jovanovich, Publishers, Orlando,
Florida 32887.
ISBN: 0-15-579647-X
Library of Congress Catalog Card Number: 88-81033
Printed in the United States of America
We wish to thank the following for permission to reprint photos:
For Chapter 1 opener: New Jersey State Museum Collection, Trenton,
Purchase FA 1968.73; Chapter 2: © George Hall/Woodfin Camp; Chapter 3:
Courtesy of Deric Bownds, Stan Carlson, University of Wisconsin; Chapter
4: Christian Delbert Photography/Picture Cube; Chapter 5: © Larry Lorusso/

Picture Cube; Chapter 6: © Henry Groskinsky; Chapter 7: © Roberta J.
Shefke; Chapter 8: © Margerin Studios/FPG; Chapter 9: Photograph by
L. L. Brown courtesy of The Institute of Psycho-Structural Balancing, San
Diego, CA; Chapter 10: The Soloman R. Guggenheim Museum; Chapter
11: © 1988 M. C. Escher c/o Cordon Art-Baam-Holland; Chapter 12: The

San Diego Opera; Chapter 13: Ed Carlin/Picture Cube; Chapter 14: E.
Nagle/FPG; Chapter 15: Aronson Photographers/Stock Boston; Chapter 16:
Elizabeth Crews/Stock Boston; Chapter 17: © Norman Snyder; Chapter 18:
United Nations Photo 36252. For Figure 10-1: © Margaret K. Porter; Figure
13-5: © Paulette M. Russo; Figure 15-5: © Paulette M. Russo.

Preface
Take away the sensations of softness, moisture,
redness, tartness, and you take away the
cherry. Since it is not a being distinct from
these sensations; a cherry, I say, is nothing but
a congeries of sensible impressions or ideas
perceived by various senses; which ideas are
united into one thing.
George Berkeley, 1713

Virtually everything we know about the world has
entered our minds through our senses. We all real¬
ize that without even one of our senses our experi¬
ences would be incredibly limited. Consider the
impossible problem of explaining the difference be¬
tween the color blue and the color green to a person
who has been blind since birth. And how would
you explain to a person who has no taste buds how
the taste of chocolate and vanilla differ from each
other? Such aspects of the world will never exist
for these individuals. For the blind person, salt and
pepper differ only in taste; for the person with no
ability to taste, salt and pepper differ only in color.
For those of us who have the senses of sight, hear¬
ing, taste, touch, and smell, our daily experience is
a continuous flow of changing percepts, with each
new sensation carrying information about the

have rewritten six of the chapters “from the ground
up” and reorganized or amalgamated materials
from other chapters. Every chapter has been up¬
dated to reflect the most recent literature. However,
we have retained all those features that instructors
felt made the Second Edition such a useful book.
For instance, concrete examples are used through¬
out the text in order to make the subject matter
“come alive” for students. Whenever possible,
common or natural instances of perceptual phe¬
nomena are described during the discussion of the
concepts underlying them. Each chapter is pre¬
ceded by an outline that serves as a preview to its
contents; the outlines also provide a structure to
guide students as they review the chapters.
Although terms are defined when they are in¬
troduced, a glossary is provided at the end of each
chapter as well. Any item printed in boldface in
the text is also listed in the chapter glossary. Stu¬
dents will find that these glossaries serve as a suc¬
cinct review and chapter summary, and can be used
for self-testing and study purposes.
One special feature of our book is the inclu¬
sion of 106 Demonstration Boxes. Each box de¬
scribes a simple demonstration designed to allow
the students to actually experience many of the per-

world.
Sensation and Perception, Third Edition, pro¬
vides an introduction to the study of our senses and
how we perceive through them. It has been revised
substantially since the Second Edition, and contains
over 45 percent new material. These changes re¬
flect many of the recent findings that have
emerged, or coalesced into meaningful patterns,
since the completion of the previous edition. We
v

VI

Preface

ceptual phenomena described in the text. Most re¬
quire only the stimuli in the box itself, or
commonplace items that can be found in most
homes or dormitory rooms. The majority of these
demonstrations require only a few moments of
preparation, which we feel is time well spent in
improving the understanding of the concepts under
discussion and in maintaining student interest.
Some instructors have reported that having students
perform the demonstrations in class has been very
useful. In such cases, the demonstrations may also
serve as the focal point for a lecture or for class¬
room discussion.
The book is designed to survey the broad
range of topics generally included under the head¬
ing of sensation and perception. The reader will
notice that no single theory of perception is cham¬
pioned. In general, we have attempted to be as
eclectic as we could, describing the various view¬
points in areas of controversy and attempting to
present a balanced overview so that instructors of
different opinions might be comfortable using the
book.
The topics in this book were selected on the
basis of our experience in teaching our own
courses; therefore, much of the material has already
been class tested. We have included three chap¬
ters—“Attention,” “Speech and Music,” and
“Individual Differences”—that are not often seen
in sensation and perception textbooks. These areas
have attracted a good deal of experimental work in
recent years, and they are sufficiently relevant to
many issues in perception that we felt students
should study them.
In order to keep the book to a manageable
size, we have occasionally been selective in our
coverage. Our first priority was to cover the central
concepts of each topic in enough detail to make the
material clear and coherent. To have included all
the topics ever classified as part of the field of sen¬
sation and perception, we would have had to pre¬
sent a “grocery list” of concepts and terms, each
treated superficially. Such an alternative was unac¬
ceptable to us.

Each of the chapters has been written so that
it is relatively self-contained and independent of the
other chapters, although this is not always com¬
pletely possible. Therefore, when material from
other places in the book is used in a discussion, the
location of that information is always cited. This
has been done to provide users with maximum flex¬
ibility as far as the sequence of chapter presentation
is concerned, thus permitting the instructor to im¬
press his or her orientation upon the material. A
brief appendix on some basic aspects of neurophys¬
iology has also been provided for the first time in
this edition.
The chapter sequence in this Third Edition is
quite different from that used in the previous edi¬
tions. At the request of many individuals who have
taught from the earlier versions, we have now or¬
ganized the book by sensory systems, with the first
half of the book covering the basic physiology and
sensory responses and the second half covering
those topics involving more complex and cognitive
interactions. Chapters 1 and 2 provide an introduc¬
tion to the problems of sensation and perception
along with methodological and theoretical aspects
of psychophysical measurement; Chapters 3,4, and
5 cover the physiology and basic sensory qualities
of vision; Chapters 6 and 7 do the same for audi¬
tion; and Chapters 8 and 9 cover the chemical and
mechanical senses. These first nine chapters thus
cover the major topics usually grouped together un¬
der the heading of sensation. Chapters 10 through
15 cover the perception of space, form, speech and
music, and time and motion, the perceptual con¬
stancies, and the perceptual aspects of attention;
and Chapters 16, 17, and 18 discuss how individual
variables such as age, experience, learning, gender,
culture, drugs, and personality may affect the per¬
ceptual response. These last nine chapters thus
cover the topics most frequently grouped together
as perception.
Those of you who have encountered earlier
versions of this book should know that Clare Porac
has retired from this project in order to pursue her
research and other writing projects. Clare’s contri-

Preface

butions were always organized, intelligent, and of
the highest professional quality. Although she did
not directly participate in this revision, we have
striven to retain the clarity in writing and wellstructured discussions that have characterized her
contributions to the first two editions.
In our attempts to collect and interpret the in¬
formation for this book, we have been assisted at
various stages by our colleagues. Some have read
preliminary versions of chapters and made useful
suggestions. We would like to specifically thank
Ray Corteen, Jim Enns, Ronnie Lakowski, Richard
Tees, and Janet Werker, all of the Psychology De¬
partment of the University of British Columbia. We
would also like to thank the personnel of the Hu¬
man Perception and Psychophysics Laboratories,
and especially Wayne Wong and Odie Geiger for
assisting with library work and all of the small but
necessary chores that eat up innumerable hours of
a textbook writer’s time.
In addition, we would like to thank Barry An¬
ton, of the University of Puget Sound, Robert

vn

Frank, of the University of Cincinnati, Robert
Levy, of Indiana State University, and Lyn Mowafy, of Vanderbilt University, for their helpful
comments and suggestions after reviewing the pre¬
vious edition.
Finally, the reader might notice that there is
no dedication page. This is not to say that we do
not wish to dedicate the book to anyone. It reflects
the fact that there are too many people who have
been important in our personal and professional
lives to list on any single page (no matter how
small the print). Perhaps it is best to simply dedi¬
cate this book to all of those researchers who have
provided the knowledge that we have attempted to
organize and review between these covers, and to
all of those researchers who will provide further
insights into sensation and perception for future au¬
thors to collate, review, digest, wonder at, and
learn from.
S.C.
L.M.W.

'

*

»

Contents
Preface

1
. . .
. . .

v

5

Sensation and Perception

1

Aspects of the Perceptual
Process

Color Stimulus

. . .

The Physiology of Color

. . .

. . .

Theories of Perception

. . .

The Plan of the Book

119

Sr.
. . .

9

Color

Vision

120

130

Color Perception

143

10
11

AUDITION
2

Psychophysics

. . .

Detection

. . .

Identification

. . .

Discrimination

. . .

Scaling

6

15

16
27
33

VISION
3 The Visual System
. .

Light

58

. .

The Structure of the Eye

. .

Neural Responses to Light

. .

The Visual Pathways

. .

The Visual Cortex

58

. .

The Structure of the Ear

. .

Electrical Activity of the Auditory

Photometric Units

Nerve

166

The Auditory Pathways

. .

The Auditory Cortex

Hearing

172

177

Detection of Sounds

. . .

Subjective Dimensions of
Sound

178

195

CHEMICAL AND MECHANICAL
SENSES

78

8

90

98

Cognitive Factors in Brightness

Taste and Smell

The Gustatory (Taste) Sense

212

. .

The Olfactory (Smell) Sense

222

Touch and Pain

237

. .

The Skin Senses

. .

Touch

238

245

Perception

103

. .

Warmth and Cold

Visual Acuity

104

. .

Kinesthesis

. .

Pain

108

211

. .

9

92

Spatial Frequency Analysis

169

. . .
. . .

157

71

Factors in Brightness
Brightness Contrast

152

. .

74

Brightness and Spatial
Frequency
89

Perception

Sound

7
57

151

. .

. .

40

The Auditory System

261

258

252

Contents

X

PERCEPTION
10
.
.
.
.
.
.
.
.

Space
.
.
.
.
.
.
.
.

Types of Depth and Distance
Perception
274
Pictorial Depth Cues
275
Physiological Cues for
Depth
284
Physiological Cues for
Direction
285
Binocular Depth Perception
289
Interaction of Cues
295
Development of Space Perception
299

. .
. .

11

Form

V;_y
. .
. .
. .
. .
. .

12
, . .
. .

273

Speech and Music

Time and Motion

. . .
, . .

Time
372
Motion
380

14

The Constancies

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

Lightness or Whiteness
Constancy
419
Color or Hue Constancy
Other Constancies
423

Attention

422

427

Varieties of Attention
Orienting
429
Filtering
433
Searching
440
Expecting
447
Theories of Attention

428

450

PERSONAL FACTORS AND CHANGE
16
.
.
.
.
.

.
.
.
.
.
.

371

403

The Task of Perception
Perceptual Constancies
Size Constancy
407
Shape Constancy
416

.
.
.
.
.

.
.
.
.
.

17

339

340
348

13

.
.
.
.

15

305

The Visual Field
306
Contour
307
The Perceptual Object
310
Object Recognition and
Identification
323
Theories of Object Identification
331

Music
Speech

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

18
404
405

.
.
.
.

.
.
.
.

.
.
.
.

Development

455

Perception in Infants
456
Perceptual Change through
Childhood
468
Perceptual Change in Adults
473

Learning and Experience

479

Experience and Development
480
Sensory-Motor Learning
487
Context and Meaning
494
Environmental and Life History
Differences
499

Individual Differences

511

Physiological Differences
512
Gender Differences
520
Personality and Cognitive Style
Differences
525

Contents

Appendix: Primer of
Neurophysiology
529
Neurons and the Nervous
System
530
The Nature of Neural
Activity
531

Techniques to Measure Neural
Function
533
References
537
Author Index
583
Subject Index
597

xi

'

CHAPTER

1
Sensation and
Perception
•

ASPECTS OF THE PERCEPTUAL PROCESS
THEORIES OF PERCEPTION
THE PLAN OF THE BOOK

1

2

C

Chapter 1/Sensation and Perception

an you answer the following questions? What
color is the sky? Which is warmer, fire or ice?
Which tastes sweeter, sugar or vinegar? Which has
a stronger smell, burning wood or burning rubber?
Which sounds louder, a chirping bird or the crack
of a rifle? Such questions probably seem quite triv¬
ial, and the answers obvious. Perhaps we should
phrase the questions differently. How do you know
what color the sky is? How do you know how hot
fire is relative to ice? How do you know that sugar
is sweet? Again, you might feel that the answers
are obvious. You see the color of the sky, you feel
the temperatures of a flame and an ice cube, and
you taste the sweetness of sugar—in other words,
the answers come through your senses.
Let us push our questioning one step further.
How do you know anything about your world? You
might say that you learn from books, television,
radio, films, lectures, or the actual exploration of
places. But how do you obtain the information
from these sources? Again, the answer is through
your senses. In fact, without your senses of vision,
hearing, touch, taste, and smell, your brain, the or¬
gan that is responsible for your conscious experi¬
ence, would be an eternal prisoner in the solitary
confinement of your skull. You would live in total
silence and darkness. All would be a tasteless, col¬
orless, feelingless, floating void. Without your sen¬
ses, the world would simply not exist for you. The
philosopher Thomas Hobbes recognized this fact in
1651 when he wrote, “There is no conception in
man’s mind which hath not at first, totally or by
parts, been begotten upon the organs of sense.’’
The Greek philosopher Protagoras stated the same
position around 450 B.c. when he said, “Man is
nothing but a bundle of sensations.”
You may protest that this is a rather extreme
viewpoint. Certainly, much of what we know about
the world does not arrive through our eyes, ears,
nose, and other sense organs. We have complex
scientific instruments, such as telescopes, that tell
us about the size and the shape of the universe by
analyzing images too faint for the human eye to
see. We have sonar to trace out the shape of the

sea bottom, which may be hidden from our eyes by
hundreds of feet of water. We have spectrographs
to tell us about the exact chemical composition of
many substances, as compared to the crude chemi¬
cal sensitivity of our noses and tongues.
Although such pieces of apparatus exist, and
measure phenomena not directly available to our
senses, this does not alter the fact that it is the per¬
ception of the scientist that constitutes the subject
matter of every science. The eye of the scientist
presses against the telescope or examines the pho¬
tograph of the distant star. The ear of the scientist
listens to the sound of sonar tracing out the size and
distance of objects, or his eyes read the sonograph.
Although the tongue of the scientist does not taste
the chemical composition of some unknown sub¬
stance, his eye, aided by the spectrograph, provides
the data for analysis. Really, the only data that
reach the mind of the scientist come not from in¬
struments but from the scientist’s senses. The in¬
strument he or she is looking at can be perfectly
accurate, yet if the scientist misreads a digital read¬
out, or does not notice a critical shift in the opera¬
tion of a measurement device, the obtained
information is wrong and the resulting picture of
the world is in error. The minds of the scientist,
the nonscientist, our pet dog sniffing about the
world, or a fish swimming about in a bowl, in fact,
the minds of all living, thinking organisms, are
prisoners that must rely on information smuggled
in to them by the senses. Your world is what your
senses tell you. The limitations of your senses set
the boundaries of your conscious existence.
Because our knowledge of the world is depen¬
dent on our senses, it is important to know how our
senses function. It is also important to know how
well the world created by our senses corresponds
to external reality (i.e., the reality measured by sci¬
entific instruments). At this point, you are probably
smiling to yourself and thinking, “Here comes an¬
other academic discourse that will attempt to make
something that is quite obvious appear to be com¬
plex.” You might be saying to yourself, “I see my
desk in front of me because it is there. I feel my

Sensation and Perception

chair pressing against my back because it is there.
I hear my telephone ringing because it contains a
bell that makes sounds. What could be more ob¬
vious?” Such faith in your senses is a vital part of
existence. It causes you to jump out of the way of
an apparently oncoming car, thus preserving your
life. It provides the basic data that cause you to
step back from a deep hole, thus avoiding a fall and
serious bodily harm.
Such faith in our senses is woven into the very
fabric of our lives. As the old saying goes, “Seeing
is believing.” Long before the birth of Christ, Lu¬
cretius stated this article of faith when he asked,
“What can give us surer knowledge than our sen¬
ses? With what else can we distinguish the true
form from the false?” Perhaps the most striking
example of this faith is found in our courts of law,
where people’s lives and fortunes often rest solely
on the testimony of the eyes and ears of witnesses.
A lawyer might argue that a witness is corrupt or
lying, or even that his memory has failed, but no
lawyer would have the audacity to suggest that her
client should be set free because the only evidence
available was what the witnesses saw or heard.
Certainly no sane person would charge the eye or
ear with perjury!
The philosophical position that perception is
an immediate, almost godlike knowledge of exter¬
nal reality has been championed not only by popu¬
lar sentiment but also by philosophers of the stature
of Immanuel Kant (1724-1804). Unfortunately, it
is wrong. Look at the drawings shown in Figure
1-1. Clearly, they are all composed of outlined
forms on various backgrounds. Despite what your
senses tell you, A, B, and C are all perfect squares.
Despite the evidence of your senses, D is a perfect
circle, the lines in E are both straight, and the lines
marked x and y in F are both the same length.
The ease with which we use our senses—
seeing, apparently through the simple act of open¬
ing our eyes, or touching, apparently by merely
pressing our skin against an object—masks the fact
that perception is an extremely sophisticated activ¬
ity of the brain. Perception calls on stores of mem¬

3

ory data. It requires subtle classifications and
comparisons, and myriad decisions before any of
the data in our senses becomes our conscious
awareness of what is “out there.” Contrary to what
you may think, the eyes do not see. There are
many individuals who have perfectly functioning
eyes yet have no sensory impressions. They cannot
perceive because they have injuries in those parts
of the brain that receive and interpret messages
from the eyes. Epicharmus knew this in 450 B.c.
when he said, “The mind sees and the mind hears.
The rest is blind and deaf.”
“So what?” you mutter to yourself. “So
sometimes we make errors in our perceptions, the
real point is that the senses simply carry a picture
of the outside world to the brain. The picture in the
brain represents our percept. Of course if we mess
up the brain we will distort or destroy perception.”
Again, this answer is too simple. If we look outside
and see a car, are we to believe that there is a pic¬
ture of a car somewhere in our brains? If we notice
that a traffic light is green, are we to believe that
some part of the brain has turned green? And sup¬
pose that there were such images in the brain, car¬
ried without distortion from the senses, would this
help us to see? Certainly, images in the brain
would only be of value if there were some other
eyes in the head, which would look at these pic¬
tures and interpret them. If this were the case, we
would be left with the question of how these inter¬
nal eyes see. Thus, we would eventually be forced
to set up an endless chain of pictures and eyes and
pictures and eyes, because the question of who is
perceiving the percept, and how, still remains.
If we are to understand perception we must
consider it in its natural context. Sensation and per¬
ception are some of the many complex processes
that occur in the continuing flow of individual be¬
havior. There is no clear line between perception
and many other behavioral activities. No perception
gives direct knowledge of the outside world, rather
such knowledge of the outside world is the end
product of many processes. The wet-looking black
spot on the edge of a desk could be the place where

4

Chapter 1/Sensation and Perception

FIGURE 1-1

Some instances where the senses tell lies.

Sensation and Perception

ink was spilled. Of course, this percept could be
wrong. The ink may be dry, or the spot might not
be there at all. The desk that is seen and touched,
might not really exist. We might be dreaming,
drugged, or hallucinating. Too extreme, you say?
Consider the following example that actually hap¬
pened to one of the authors. One night he walked
across the floor of his darkened home. In the dim
gloominess of the night, he saw his dog resting on
the floor, clearly asleep. When he bent to touch the
dog, he found that it was a footstool. He stepped
back, somewhat startled at his stupidity, only to
bump against the cold comer of a marble-topped
coffee table. When he reached back to steady him¬
self, he found that the comer of the table was, in
fact, his dog’s cold nose. Each of these percep¬
tions, dog, stool, table, and dog again, seemed,
when first received in consciousness, to be accurate

5

representations of reality. Yet, sensory data are not
always reliable. Sometimes they can be degraded
or not completely available. There seems to be no
clear distinction between perceiving or sensing an
object and guessing the identity of an object. In
some respects, we can say that all perception of
objects requires some guessing. Sensory stimula¬
tion provides the data for our hypotheses about the
nature of the external world, and these hypotheses
form our perceptions of the world.
Many human behaviors have been affected by
the fallible and often erroneous nature of our per¬
cepts. For example, the most elegant of the classic
Greek buildings, the Parthenon, is bent. The
straight clean lines, which bring a sense of simple
elegant grandeur, are actually an illusion. If we
schematically represent the east wall of the building
as it appears, it is square (as shown in Figure 1-2A).

FIGURE 1-2
(A) The Parthenon as it appears; (B) an illusion that should cause the Parthenon to appear
as C; (D) the way the Parthenon is built to offset the illusion.

6

Chapter 1/Sensation and Perception

Actually, the Parthenon was built in a distorted
fashion in order to offset a series of optical illu¬
sions. There is a common visual distortion in which
we find that placing angles above a line (much as
the roof is placed over the architrave) causes the
line to appear slightly bowed. One form of this il¬
lusion is shown as Figure 1-2B where the ends of
the horizontal line appear slightly higher than the
center. If the Parthenon were built physically
square, it would appear to sag as a result of this
visual distortion. This is shown in an exaggerated
manner in Figure 1-2C. But the sagging does not
appear because the building has been altered to
compensate for the distortion. Figure 1-2D illus¬
trates what an undistorted view of the Parthenon
would look like. The upward curvature is more
than 6 centimeters on the east and west walls and
almost 11 cm on the longer north and south sides.
The vertical features of the Parthenon (such as
the columns) were inclined inwards in order to cor¬
rect for a second optical illusion in which the fea¬
tures of rising objects appear to fall outward at the
top. Thus, if we projected all of the columns of the
Parthenon upward, they would meet at a point
somewhat less than 2 kilometers above the build¬
ing. Furthermore, the comer columns were made
thicker since it was found that when these columns
were seen against the sky, they appeared to be thin¬
ner than those seen against the darker background
formed by the interior wall.
These were conscious corrections made by
the Greek architects. To quote one of them, Vi¬
truvius, writing around 30 B.C.: “For the sight
follows gracious contours, and unless we flatter
its pleasure by proportionate alterations of these
parts (so that by adjustment we offset the amount
to which it suffers illusions) an uncouth and un¬
gracious aspect will be presented to the specta¬
tors.” In other words, the Parthenon appears to
be square, with elegant straight lines, because it
has been consciously distorted to offset perceptual
distortions. If it were geometrically square, it
would not be perceptually square.

It is amazing to discover the degree to which
our conscious experience of the world can differ
from the physical (scientific) reality. Although
some perceptual distortions are only slight devia¬
tions from physical reality, some can be quite com¬
plex and surprising, such as that shown in
Demonstration Box 1-1.
Such distortions, in the form of disagreements
between percept and reality, are quite common. We
call them illusions and they occur in predictable
circumstances for normal observers. The term illu¬
sion is drawn from the Latin root illudere, meaning
“to mock,” and in a sense they do mock us for
our unthinking reliance on the validity of our sen¬
sory impressions. Every sensory modality is subject
to distortions, illusions, and systematic errors that
misrepresent the outside environment to our con¬
sciousness. There are illusions of touch, taste, and
hearing, as well as visual illusions. Virtually any
aspect of perception you might think of can be
subject to these kinds of errors. For instance,
such basic and apparently simple qualities as the
brightness of an object or its color may be percep¬
tually misrepresented, as shown in Demonstra¬
tion Box 1-2.
Many perceptual errors are merely amusing,
such as that in Demonstration Box 1-1, or thoughtprovoking, as in Demonstration Box 1-2. Others
may lead to some embarrassment or annoyance,
such as might have been felt by the artisan who
created the picture frame shown as Figure 1-3A.
Although his workmanship is faultless, he has been
undone because the grain of the wood is too prom¬
inent. Despite the fact that the picture is perfectly
rectangular, it appears to be distorted. Unfortu¬
nately, some perceptual errors or illusions are quite
serious. In Figure 1-3B, we have shown a surgeon
probing for a bullet. She is using a fluoroscope,
which presents the outline of the patient’s ribs, and
her probe is positioned so that it is exactly on line
with the bullet lodged below the rib. As you can
see, it appears that she will miss and her probe will
pass above the bullet despite the fact that the probe

Sensation and Perception

DEMONSTRATION BOX 1-1.

Gears and Circles

The pattern shown in this box should be viewed in
motion. Move the book around so the motion resem¬
bles that which you would make if you were swirling
coffee around in a cup without using a spoon. Notice
that the six sets of concentric circles seem to show
radial regions of light and dark that appear to move
in the direction you are swirling. They look as though
they were covered by a liquid surface tending to swirl
with the stimulus movement.
A second effect has to do with the center circle that
seems to have gearlike teeth. As you swirl the array,
the center gear seems to rotate, but in a direction op¬
posite to that of the movement of the outer circles.
Some observers see it moving in a jerky, steplike
manner from one rotary position to another and other
observers see a smooth rotation. Of course, there is
no physical movement within the circles, and the
geared center circle is also unchanging, despite your
conscious impression to the contrary.

DEMONSTRATION BOX 1-2.

A Subjective Color Grid

The figure in this box consists of a series of thinly
spaced diagonal black lines alternating with white
spaces. Study this figure for a couple of seconds, and
you will begin to see faint, almost pastel streaks of
orange-red and other streaks of blue-green. For many
observers, these streaks tend to run vertically up and
down the figure crossing both white and black lines;
for others, they seem to form a random, almost fish¬
netlike pattern over the grid. These colors are not
present in the stimulus; hence they are subjective, or
illusory, colors.

7

8

Chapter 1/Sensation and Perception

is angled perfectly. Figure 1-3C shows an even
more disastrous occurrence of an illusion. It repre¬
sents a radar screen with various flight regions
marked across its face. The two oblique streaks
represent jet aircraft approaching the control re¬
gion, both flying at about 950 kilometers per hour.
The information displayed is similar to that which
an air traffic controller might use. From it he might
conclude that if these two aircraft continue in the
same direction they will pass each other with a safe
distance between them. At the moment repre¬
sented here, however, these aircraft are traveling
toward each other on the same line. If they are
flying at the same altitude it is very likely that
they will collide.
These examples illustrate how important dis¬
crepancies between perception and reality can be.
Therefore it becomes important for us to know how
our perceptions arise, how much we can rely on
them, under what circumstances they are most fal¬
lible, and under what conditions our perceptions
most accurately represent the world. An explora¬
tion of these questions is the purpose of this book.

C
FIGURE 1-3

Some perceptual distortions in common situations.

Aspects of the Perceptual Process

ASPECTS OF THE PERCEPTUAL
PROCESS

The study of perception is diverse. Partly this is the
result of the length of time that perceptual problems
have been studied. The Greek philosophers, the
pre-Renaissance thinkers, the Arabic scholars, the
Latin scholastics, the early British empiricists, and
the German physicists and physicians who founded
both physiology and psychology considered issues
in sensation and perception to be basic questions.
When Alexander Bain wrote the first English text¬
book on psychology in 1855 it was entitled The
Senses and the Intellect, with the most extensive
coverage reserved for sensory and perceptual func¬
tions. The major portion of both the theorizing and
the empirical work produced by Wilhelm Wundt,
who is generally credited with the founding of ex¬
perimental psychology, was oriented toward sen¬
sation and perception. In addition to the diversity
caused by a long and varied history, perception has
been affected by many “schools” of thought. Each
has its own major theoretical viewpoint and its own
particular set of methodological techniques. Thus,
we encounter psychophysicists, gestaltists, func¬
tionalists, analytic introspectionists, transactionalists, sensory physiologists, sensory-tonic theorists,
“new look” psychologists, efferent theorists, arti¬
ficial intelligence experts, and computational psy¬
chologists, to name but a few. There are even
theorists (such as some behaviorists) who deny the
existence of, or at least deny our ability to study,
the conscious event we call perception. Despite this
chorus of diverse voices and viewpoints, there
seems to be a consensus about the important as¬
pects of perceptual study.
Before we look at the major areas of emphasis
in the study of the perceptual process let us first
offer a disclaimer. We recognize that it is difficult,
perhaps impossible, and most certainly unwise to
attempt to draw sharp lines separating one field of
inquiry from another. However, there are certain

9

problem areas, or orientations, that characterize
certain groups of investigators, and these seem to
be definable. The study of sensation, or sensory
processes, is concerned with the first contact be¬
tween the organism and the environment. Thus,
someone studying sensation might look at the way
in which electromagnetic radiation (light) is regis¬
tered by the eye. This investigator would look at
the physical structure of the sense organ and would
attempt to establish how sensory experiences are
related to physical stimulation and physiological
functioning. These types of studies tend to focus on
less complex (although not less complicated) as¬
pects of our conscious experience. For instance,
these investigators might study how we perceive
brightness, loudness, or color; however, the nature
of the object having a given brightness, sound, or
color would not make much difference to them.
Someone who is interested in the study of
perception is interested in our conscious experi¬
ence of objects and object relationships. For in¬
stance, the sensory question would be “How bright
does the target appear to be?” whereas the percep¬
tual questions would be “Can you identify that ob¬
ject?” “Where is it?” “How far away is it?” and
“How large is it?” In a more global sense, those
who study perception are interested in how we
form a conscious representation of the outside
environment, and in the accuracy of that represen¬
tation. For those of you who have difficulty
in drawing a hard-and-fast line between the con¬
cepts of perception and sensation, rest easy.
Since Thomas Reid introduced the distinction in
1785, some investigators have championed its use
and others have totally ignored the difference,
choosing to treat sensation and perception as a
unitary problem.
Cognition is a term used to define a very ac¬
tive field of inquiry in contemporary psychology.
The word itself is quite old, probably first intro¬
duced by St. Thomas Aquinas (1225-1274). He di¬
vided the study of behavior into two broad
divisions, cognition, meaning how we know the

10

Chapter 1/Sensation

and

Perception

world, and affect, which was meant to encompass
feelings and emotions. Today’s definition of cog¬
nition is equally as broad as that of Aquinas. Al¬
though many investigators use the term to refer to
memory, association, concept formation, language,
and problem solving (all of which simply take the
act of perception for granted), other investigators
include the processes of attention and the conscious
representation and interpretation of stimuli as part
of the cognitive process. In other words, cognition
tends to be somewhere between the areas that were
traditionally called perception and learning; and it
incorporates elements of both. The similarity be¬
tween many of the problems studied by cognitive
psychologists and those studied by perceptual psy¬
chologists is best seen by the fact that both often
publish in the same journals and on similar topics.
Information processing is a relatively new
term. This approach emphasizes how information
about the external world is operated on (processed)
to produce our conscious percepts and guide our
actions. Information processing is typically as¬
sumed to include a registration or sensory phase,
an interpretation or perceptual phase, and a memoric or cognitive phase. Thus, rather than being a
separate subdiscipline, the information processing
approach attempts to integrate sensation, percep¬
tion, and cognition within a common framework.
It relies on a levels-of-processing analysis in
which each stage of processing, from the first reg¬
istration of the stimulus on the receptor to the final
conscious representation entered into memory, is
systematically analyzed.
None of these labels should be taken as rep¬
resenting inflexible, or completely separate, areas
of study. At a recent professional meeting one
well-known psychologist lamented, “When I first
started doing research, people said I studied per¬
ception. After a while, they said I studied cogni¬
tion. Now they say I’m studying human
information processing. I don’t know what’s going
on—I’ve been studying the same set of problems
for the last ten years!”

THEORIES OF PERCEPTION
In the same way that there are many aspects of per¬
ception, there are also many theoretical approaches
to perceptual problems. One important approach
may be called biological reductionism. It is based
on the presumption that for any given aspect of
the observer’s sensation there is a corresponding
physiological event. According to this approach,
the main goal of the perceptual researcher is to iso¬
late these underlying physiological mechanisms.
The search for specific neural units whose activity
corresponds to specific sensory experiences, char¬
acterized by researchers such as David Hubei
and Torston Wiesel (1979), is common to such
theories.
Other theoretical approaches are often less
bound to a specific class of mechanism. For ex¬
ample, direct perception involves a set of theories
that begins with the premise that all the information
needed to form the conscious percept is available
in the stimuli that reach our receptors, or in rela¬
tionships among these stimuli that are invariant
predictors of what is “out there” in the environ¬
ment. This theoretical position is characterized by
the work of J. J. Gibson (e.g., 1979), who argued
that certain aspects of the environment are imme¬
diately impressed on the observer and need no fur¬
ther computation or additional information based
on inferences or experience.
Recently, a number of perceptual theorists,
whose thinking has been influenced by develop¬
ments in artificial intelligence systems, have
adopted an alternative approach that contains some
of the same flavor of direct perception. Such theo¬
ries are usually presented in the form of computer
programs or computational systems that might al¬
low machines to directly interpret sensory infor¬
mation in the same manner that a human observer
might. Typical of such theorists is David Marr
(1982), who began with the general presumption
made in direct perception that all the information
needed is in the stimulus inputs, but added the sug-

The Plan of the Book

11

gestion that this interpretation might require the de¬
tection of fairly subtle dimensions in the stimulus
and might also require a number of computations
and several stages of analysis. This added feature
has resulted in the label computational approach
being applied to such theories.
A much older (but still active) theoretical ap¬
proach begins with the recognition that our percep¬
tual representation of the world is much richer and
more accurate than might be expected on the basis
of the information contained in the stimuli available
at any one moment in time. Theories to explain this
fact often begin with the suggestion that perception
is much like other logical processes. In addition to
the information available to our sense organs at the
moment, we can also use information based on our
previous experience, our expectations, and so
forth. This means, for example, that a visual per¬
cept may involve other sources of information,
some nonvisual in nature, some arising from our
past history and cognitive processing strategies.
The similarity of many of these mechanisms to rea¬
soning leads us to refer to this type of theory as
intelligent perception. This approach probably
originated with Helmholtz in 1867, and survives
today in the work of researchers such as Irving
Rock (1983) who have a more cognitive orienta¬
tion. These theories are also called constructive

ture, and a city planner might look at the same
bridge in terms of traffic flow. At first glance there
may seem to be very little overlap between the var¬
ious views, since the city planner does not care
about the specific shape of the bridge structure, and
the engineer cares only about the structural aspects
of the beams, not their specific alloy constituents.
Yet each level of analysis is valid for some specific
set of questions. This book addresses the problem
of how people build a conscious picture of their
environment through the use of information reach¬
ing their senses. We follow the lead of many con¬
temporary theorists and try to use data from all
levels of the perceptual process, and discussions in
terms of several different theoretical positions, in
order to give an integrated picture of the process of
perception. After all, the label we apply to our ap¬
proach is of considerably less importance than the
answer itself.

theories of perception, since our final conscious
impression may involve combining a number of
different factors to “construct” the final percept.
It is quite likely that each of these approaches
is useful in describing some aspects of the percep¬
tual process (see Coren & Girgus, 1978; Uttal,
1981); however, different orientations tend to lead
researchers in different directions, searching for

perceptual and sensory processes. In general, the
presentation of the material follows a levels-of-processing approach, in that the first half of the book
is concerned with the more basic sensory processes
and is organized around specific sensory systems,
such as vision or audition, and the second half of
the book is concerned with the more clearly per¬
ceptual processes that have strong cognitive influ¬
ences, and are often not bound to any single
sensory modality.
We have tried to make the individual chapters
relatively self-contained. We begin by explaining
how sensations and perceptions are measured
(Chapter 2). We then proceed with the physiologi¬
cal structures and the basic sensory capacities as-

different types of mechanism. Each approach is
likely to be valid for some parts of the problem and
irrelevant to others. This is a common occurrence
in many areas of endeavor. For instance, a metal¬
lurgist might look at a bridge and consider its ma¬
terial components, whereas a civil engineer might
look at the load-bearing capacity of the entire struc¬

THE PLAN OF THE BOOK
The orientation of this book is implicit rather than
explicit. Although theories are introduced and dis¬
cussed in the various chapters, no all-encompassing
theoretical position has been adopted. We have
chosen to be “militantly eclectic” in our orienta¬
tion. Thus, this text is mostly concerned with

12

Chapter 1/Sensation and Perception

sociated with vision (Chapters 3 through 5),
audition (Chapters 6 and 7), and the chemical and
mechanical senses (Chapters 8 and 9). For those
who feel a bit “rusty” about some of the very
basic physiological facts, we have also included a
“Primer of Neurophysiology” as an appendix.
Chapters 10 to 15 deal with those problems that
have traditionally been treated as part of classical
perception, our perceptual representation of space,
time, motion, form, and size. The more cognitive
aspects of perception are also introduced here in
those chapters that deal with the issues of music,
speech perception, and attention. The last three
chapters (16 to 18) deal with perceptual diversity,
which incudes many of the factors that make the
perceptual experience of one individual different
from that of another. These factors include the
changes that occur in the developing individual
because of the normal aging process, life history,
experience, learning, and personality factors, to
name a few.
You will notice that each chapter includes a
series of Demonstration Boxes. These are experi¬
mental demonstrations that you can perform for
yourself using materials that are easily found
around a house or other living quarters. They illus¬
trate many aspects of the perceptual process. Quite

often they demonstrate concepts that are very dif¬
ficult to put into words, but which, when experi¬
enced, are immediately understandable. You are
encouraged to try these demonstrations since they
are an integral part of the book. In the same way
that perception involves interaction with the world,
these demonstrations allow you to interact with
your senses in a controlled manner and to gain in¬
sight into yourself.
We hope this book will provide you with some
understanding of the abilities and the limits of your
senses. This knowledge should expand your com¬
prehension of many behavioral phenomena that de¬
pend on perception as a first step. Perception seems
to be the final judge of the truth or the falsity of
everything we encounter as part of our human ex¬
perience. How often have you heard the phrase
“Seeing is believing” or “I didn’t believe it until
I saw it with my own two eyes”? Yet you have
already seen in this chapter that such faith in the
truthfulness of our conscious percepts is often mis¬
placed. In 500 B.c., Parmenides considered how
perception can deceive us, summarizing his feel¬
ings in these words: “The eyes and ears are bad
witnesses when they are at the service of minds that
do not understand their language.” In this book,
we will try to teach you their language.

Glossary

13

GLOSSARY
The following definitions are specific to this book.
Biological reductionism
The theoretical premise
that each sensory experience is associated with particular
physiological events.

illusions
Distortions or incongruencies between per¬
cept and reality.
Information processing
The processes by which
stimuli are registered in the receptors, identified, and
stored in memory.

Cognition
The process of knowing, incorporating
both perception and learning.
Computational approach
Involves the presumption
that certain perceived qualities require computation and
that these computations can be precisely described math¬
ematically.

Intelligent perception
The theoretical presumption
that cognitive processes and experience can affect per¬
ception.
Levels-of-processing analysis
Analysis of the con¬
tribution of each stage of processing to the final percept,
beginning with the receptor and continuing through cog¬
nitive mechanisms.

Constructive theories
These maintain that percep¬
tion may involve the integration of several sources of
information, and may be affected by cognitive factors
and experience.

^

Direct perception
The theoretical position that all
the information needed for the final conscious percept is
in the stimulus array.

✓ Sensation
Simple conscious experience associated
with a stimulus.

Perception
The conscious experience of objects and
object relationships.

'

CHAPTER

2
Psychophysics
•

*

DETECTION
Method of Limits
Method of Constant Stimuli
Signal Detection Theory
IDENTIFICATION
Information Theory
Channel Capacity

•

•

•

•

DISCRIMINATION
Weber’s Law
Signal Detection Theory in
Discrimination
Reaction Time
SCALING
Indirect Scaling: Fechner’s Law
Direct Scaling
Category Judgment
Magnitude Estimation: Stevens’s Law
Cross-Modality Matching

Adaptation Level Theory

15

16

T

Chapter 2/Psychophysics

he ocean liner glides slowly through the thick
stormy night. Somewhere in the distance is
New York harbor. With the visibility near zero the
captain is forced to rely solely on the ship’s radar
system for information about the position of obsta¬
cles impeding the passage of his ship. The ship is
in a heavily traveled trade route, and the crew must
continually be alert for possible collisions with
other ships. The radar operator is watching her
screen intently, searching for a radar echo caused
by the presence of another ship nearby. Actually,
she is also wrestling with a basic sensory-percep¬
tual problem, that of detection. She is trying to
answer the question “Is there anything there?”
She is sure she sees an echo. Now the ques¬
tion becomes “What is it?” Is it an echo from an¬
other ship or just a “ghost,” a false echo often
encountered in stormy weather? The radar operator
is facing a second basic problem, identification.
We normally solve the detection and identification
problems quickly and automatically, since we gen¬
erally encounter stimuli that are so strong, and pro¬
vide so much information, that they pose little
problem for us. The complex nature of detection
and identification only emerges in the context of a
difficult or degraded stimulus situation.
The echo turns out to be just a “ghost” and
the order is given to maintain the previous heading
(compass direction). The helmsman has been given
the bearing and now holds the ship’s direction so
that the compass needle always points to the correct
place on the dial. At this moment, he is asking
himself, ‘ ‘Has the needle drifted slightly toward the
north?” If so, he must compensate by turning the
wheel so that the needle moves back to the desired
compass point. He is continuously concerned with
the problem of whether the compass needle is cen¬
tered on the desired heading. This task also evokes
an important perceptual process, called discrimi¬
nation. “Is this stimulus different from that one?”
is the general discrimination question.
Finally, through the clearing weather the en¬
trance to New York harbor appears. The ship is
taken in tow by a tugboat and maneuvered toward

its berth at the dock. The captain of the tugboat
peers from his bridge, carefully judging the dis¬
tance between the ship and the concrete wall of the
pier. He must continually ask himself, “How far
does the ship appear to be from the pier?” Such
questions are part of another sensory problem,
“How much of X is there?” This is the problem of
scaling.
These four problems, detection, identification,
discrimination, and scaling, are the central con¬
cerns of the area of perceptual psychology called
psychophysics. Psychophysics owes its name and
origin to Gustav Theodor Fechner (1801-1887), a
physicist and philosopher who set out to determine
the relationship between the magnitude of a sensa¬
tion registered in the mind and the magnitude of
the physical stimulus that gave rise to it. Hence,
the name psychophysics (from the Greek roots psy¬
che, or “mind,” and physike, which refers to nat¬
urally occurring phenomena). Fechner not only
established the philosophical rationale for studying
the relationship between sensations and physical
stimuli but also developed many of the experimen¬
tal methods still in use today. These methods of
collecting and analyzing data are employed in
every aspect of the study of sensation and percep¬
tion (see, e.g., Laming, 1986) and of many other
areas of psychology, including even social, person¬
ality, and clinical psychology (Baird & Noma,
1978; Grossberg & Grant, 1978; Wegener, 1982).

DETECTION
The basic task for any sensory system is to detect
the presence of energy changes in the environment.
Energy changes may take the form of electromag¬
netic (light), mechanical (sound, touch, movement,
muscle tension), chemical (tastes, smells), or ther¬
mal stimulation. The problem of detection is cen¬
tered around the problem of how much of such a
stimulus (relative to a zero energy level) is neces¬
sary for an individual to say that the stimulus is
heard, tasted, smelled, or felt. Classically, this

Detection

minimal amount of energy has been called the ab¬
solute threshold. In 1860, Fechner defined a
threshold stimulus as one that “lifted the sensation
or sensory difference over the threshold of con¬
sciousness.” The idea is that below some critical
value of a stimulus a person would not be expected
to detect that stimulus. As soon as this threshold
value is exceeded, however, we would expect the
observer to always detect its presence.
We can represent this relationship by a graph
on which we plot the percentage of time an ob¬
server would be expected to detect the presence of
a stimulus (values along the ordinate, or vertical
axis) against stimulus magnitude (values along the
abscissa, or horizontal axis). This has been done in
Figure 2-1 using arbitrary values for stimulus in¬
tensity. Notice that the percentage of time that the
stimulus is detected takes a sudden step up from 0
to 100 percent when the stimulus reaches a value
of 3.5. The absolute threshold is thus 3.5.

Method of Limits
How do we measure absolute thresholds? Let us
conduct a relatively simple but typical experiment
to measure the threshold of hearing. In this exper¬
iment an observer sits in a soundproof room wear¬
ing headphones. The experimenter presents a very
faint, undetectable tone of a particular and constant
frequency and increases its intensity in small steps
until the observer reports, “I hear it.” On alternate
trials the experimenter starts with a tone that can
easily be heard and decreases the intensity until the
observer reports, “I no longer hear it.” This
method of determining a threshold is called the
method of limits. Kraepelin gave it that name in
1891, because a stimulus series always ends when
the observer reaches a limit or a point of change in
his judgments. The two modes of presenting the
stimulus (increasing or decreasing) are usually
called ascending or descending stimulus series. A
sample of the kind of data such an experiment
might generate is shown in Table 2-1.
The first thing we notice about the data in Ta-

17

Threshold
stimulus levelX.
1.00
co
CD
CO

c=

o

Q.
CO
0)
CO
CD

>,

.50

o
c
o
o
CL
o

0

1

2

3

4

5

6

7

Stimulus intensity
FIGURE 2-1

Absolute threshold.

ble 2-1 is that the absolute threshold for hearing is
not a fixed value as we first proposed, but appears
to vary from trial to trial. For instance, in Trial 6
the observer could no longer detect the stimulus
when we presented a tone with an intensity of 8; in
Trial 4, a stimulus intensity of only 5 was detected.
Such data indicate that the absolute threshold is
anything but absolute. It seems that the threshold
varies from measurement to measurement, or mo¬
ment to moment. As early as 1888, Joseph Jastrow
speculated on the reason for this variability in the
threshold over time. He theorized that lapses of at¬
tention, slight fatigue, and other psychological
changes could cause the obtained fluctuations.
Demonstration Box 2-1 shows how you can expe¬
rience this threshold variability for yourself.
We can compute an estimate of the average
absolute threshold from the tabled data simply by
taking the average stimulus intensity at which a re¬
sponse shifted either from an “I hear it” to an “I
don’t hear it,” or from an “I don’t hear it” to an
“I hear it.” This gives us a threshold value of 6.65
intensity units. These computations are shown at
the bottom of the table. Table 2-1 also shows that
there is a slight difference in the threshold value
depending on whether it was computed from an as-

18

Chapter 2/Psychophysics

Table 2-1.

Determination of the Absolute Threshold of

Hearing by the Method of Limits
Trials

Sound intensity

f

(scale units)

1

I
2

16
15
14
13
12
11
10
9

5
4
3

T

I

5

6

+
+
+
+

+
+

+

7

6

1
4

+
+
+

+
+
+
+
+
+
+

8

f
3

+

+
+

-

+

+
+
+

+
+

+a
b

2
1
Threshold for series
computations

5.5

7.5

7.5

Mean descending threshold = 7'5
Mean ascending threshold
Mean absolute threshold

=

+

4.5

4-5

+

6.5

85

3
5.5 + 7.5 + 6.5

3
6.65
sound
units
=

=

8.5

6.8
o.5

a. “I hear it.”
b. “I don’t hear it.”

cending or a descending series of stimuli. Such dif¬
ferences may arise from observers continuing to
report yes in a descending series and no in an as¬
cending series, a tendency called the error of per¬
severation. It is also possible to have an error of
anticipation. Here an observer feels that she has
said yes too often and decides that it is time to say
no even though she still faintly hears the tone. To
balance out such possible constant errors we use
alternating ascending and descending stimulus se¬

ries, and we begin each series of the same kind at
different stimulus intensities.
The method of limits has been modified to
produce a different method for measuring absolute
thresholds called the staircase method. Here the
experimenter attempts to capture the absolute
threshold by changing the direction of the steps
whenever the observer changes her response. Thus,
we might increase the intensity of a tone, step by
step, until the observer reports that she hears it, and

Detection

DEMONSTRATION BOX

2-1.

19

The Variability of the Threshold

For this demonstration you will need a wristwatch or
an alarm clock that ticks. Place the clock on a table
and move across the room so that you can no longer
hear the ticking. If the tick is faint, you may accom¬
plish this merely by moving your head away some
distance. Now gradually move toward the clock. Note
that by doing this, you are actually performing a
method of limits experiment since the sound level
steadily increases as you approach the watch. At some

then start to decrease it, one step at a time, from
that level until it is no longer heard, and then again
start to increase it by steps. Notice that in this way
the value of the test stimulus flips back and forth
around the threshold value. The advantage of this
procedure is that it allows the experimenter to
“track” the threshold, even if sensitivity is contin¬
ually changing, such as after administration of
some drugs, or during adaptation to different back¬
ground stimuli (Bekesy, 1947; Jesteadt, 1980).
Why does the threshold seem to vary from
moment to moment? First we must recognize that
we have been assuming that the only stimulus pres¬
ent is the stimulus we are asking our observer to
detect. This is quite false. A constantly present and
ever-changing background of stimulation exists for
any signal we present. If you place both your hands
over your ears to block out the room noises, you
will hear a sound one observer poetically called
“the sound of waves from a distant sea” and an¬
other, somewhat less poetically, “the faint hissing
of radio static.” Similarly, if you sit in a com¬
pletely lightproof room in absolute darkness, you
do not see complete blackness. Your visual field
appears to be filled with a grayish mist (which has
been termed “cortical gray”) and occasionally you
can even see momentary bright pinpoint flashes
here and there. Any stimulus we ask an observer to
detect must force itself through this spontaneously

distance from the watch you will just begin to hear
the source of the sound. This is your momentary
threshold. Now hold this position for a few moments
and you will notice that occasionally the sound will
fade and you may have to step forward to reach
threshold, whereas at other times it may be noticeably
louder and you may be able to step back farther and
still hear it. These changes are a result of your chang¬
ing threshold sensitivity.

generated fluctuating background. It is as if every
stimulus to be detected is superimposed on a back¬
ground of noise generated within the observer.
As this endogenous, or internal, noise level
changes, so does our measured threshold, in the
same way that a person standing in the midst of a
noisy crowd must talk louder in order to be heard.
Some experimenters have resorted to the introduc¬
tion of experimentally controlled background noise
in order to achieve more constant conditions than
would be possible if they relied on the constancy
of internally generated noise. Under these circum¬
stances, the experimenter has a better idea of the
noise level with which the stimulus is competing.
Many of the experiments we will discuss have em¬
ployed such a controlled background noise level.
By noise we mean any background stimulus other
than the one to be detected. Of course, if we define
noise in this way we may have visual, chemical,
mechanical, and thermal, as well as auditory noise.

Method of Constant Stimuli
Discussion of another method will allow us to see
more clearly the nature of the absolute threshold.
This method is preferred when the threshold must
be measured precisely, but it is much more timeconsuming to use because it requires so many stim¬
ulus presentations and responses. Suppose we take

20

Chapter 2/Psychophysics

a set of stimuli ranging from clearly imperceptible
to clearly perceptible and present them, one at a
time, to our observer. We present each stimulus
many times in a prearranged irregular order. The
observer is simply required to respond yes when
she detects the stimulus and no when she does not.
This procedure is called the method of constant
stimuli, a name derived from the fact that a fixed
or constant set of stimuli is chosen beforehand and
presented a fixed or constant number of times to
each observer. Some typical data obtained with this
method are presented graphically in Figure 2-2.
We see in the figure that as the stimulus en¬
ergy increases, the relative number of times the ob¬
server says yes (meaning the stimulus was
perceived) gradually increases. It is not the single
jump we might have predicted from the definition
of absolute threshold illustrated in Figure 2-1.
These S-shaped curves, called ogives, are obtained
commonly with the method of constant stimuli in
all sensory systems.
What does the proportion of yes responses in¬
dicate in such experiments? One basic assumption
made by psychophysicists is that any type of be¬
havior, such as saying “Yes, I see it,” has some

Stimulus intensity

FIGURE 2-2

Typical data from method of
constant stimuli in detection.

strength. The strengths of various behaviors can be
represented by numbers, which indicate their rela¬
tive magnitudes. The measure that has found most
favor among contemporary workers in the field is a
numerical estimate of the likelihood that the partic¬
ular response in question will occur. We call this
likelihood the response probability. We can esti¬
mate the response probability for detecting the
stimulus, or more exactly for saying “Yes, I see
it,” by using the formula

Number of “Yes” Responses

p( Yes) = ---

Number “Yes” + Number “No”

The data in Figure 2-2 make it clear that the
probability that an observer will detect a stimulus
is not an “all or none” affair (as in Figure 2-1),
but rather changes gradually as the stimulus inten¬
sity increases. Then where is the absolute thresh¬
old? Here, as in many places, we must make a
somewhat arbitrary decision. The point usually
taken as the absolute threshold is that value
where the probability of saying yes is the same as
the probability of saying no. This is simply the
stimulus intensity that the subject claims she de¬
tected 50 percent of the time. In Figure 2-2 we
have indicated this threshold value by dotted
lines. The threshold is about 3.5 energy units for
this observer.
Here are some examples of approximate
threshold values as measured by these methods.
The visual system is so sensitive that a candle
flame can be seen from a distance of more than 48
kilometers on a dark clear night. In the auditory
system, we can detect the ticking of a wristwatch
in a quiet room at a distance of 6 meters—sensitiv¬
ity beyond this point would allow us to hear the
sound of air molecules colliding. As for our other
senses, we can taste 1 teaspoon of sugar dissolved
in 7VS liters of water and smell 1 drop of perfume
diffused through the volume of an average threeroom apartment (Galanter, 1962).

Detection

Signal Detection Theory
Some of you may have been bothered by one as¬
pect of the psychophysical measurement techniques
we have been discussing. We are supposedly
studying an observer’s sensory capacities, yet we
have not been talking about the probability that an
observer detects a stimulus, but rather the probabil¬
ity that he says “Yes, I hear (or see, or whatever)
it.” We can imagine that if an observer feels that
this is a “test” of some sort, where it would be
good for him to appear to be quite sensitive, he
might say yes on almost every trial. What is to pre¬
vent this from happening? Although we might ar¬
gue that people are basically honest, and would not
lie about whether or not they heard a stimulus, this
is not the sort of guarantee on which scientists
would like to rest their conclusions. We are not
criticizing the reliability of observers in psycho¬
physical experiments, for most are quite sincere
and honest. Rather, we are pointing out that at the
very low stimulus energies used in most detection
experiments an observer may be unsure about
whether a sensation has been experienced. This
may result in being unsure as to whether or not to
respond yes on any particular trial. Thus, on some
trials a “guess” response is made. Therefore, in
order to assess sensory capacities accurately, we
must take into account the observer’s decision¬
making behavior.
Experimenters became aware of this problem
early in the history of psychophysics. They first at¬
tempted to cope with it by inserting catch trials,
which were trials in which no stimulus was pre¬
sented. They reasoned that if observers were honest
in reporting what was detected, they would respond
no on these catch trials. If the yes response came
too frequently, the observer was warned by the ex¬
perimenter. Alternatively, an attempt was made to
adjust the calculated threshold to account for the
guesses, or the data were simply discarded. Over
many experiments, however, it became clear that
the observers were not trying to fool anyone.

21

Somehow their behavior was reasonable, although
it was not clear what they were doing.
If we now change our classical absolute
threshold experiment so that we can study not only
the observer’s ability to detect a stimulus when it
is there but also his guessing behavior as reflected
in a yes response when no signal is present, we
have entered the domain of signal detection theory
(see Baird & Noma, 1978; Egan, 1975; Green &
Swets, 1966). It is a mathematical, theoretical sys¬
tem, which recognizes that the observer is not
merely a passive receiver of stimuli but is also en¬
gaged in the process of deciding whether or not he
is confident enough that the stimulus was present
to say “Yes, I detected it.”
For the purposes of the following discussion
we shall list all the possible behaviors in a new
type of detection experiment and give them names.
Table 2-2 is a schematic representation of the stan¬
dard signal detection experiment. There are two
types of “stimulus” presentations (at the left of the
table). A signal absent presentation is like a clas¬
sical catch trial in which no stimulus is presented
and the observer sees or hears only the noise gen¬
erated by the sensory system. Signal present is a
trial in which the experimenter actually presents the
target stimulus (which is, of course, superimposed
on the endogenous noise in the sensory system).
There are also two possible responses in the exper¬
iment (at the top of the table). Yes indicates that
the observer thinks a stimulus was presented on a
particular trial (that is, signal present), and no in-

Table 2-2. Outcomes of a Signal Detection
Experiment
Response
Signal

Yes

No

Present
Absent

Hit
False alarm

Miss
Correct negative

22

Chapter 2/Psychophysics

dicates that the observer thinks the signal was ab¬
sent. The combination of two possible stimulus
presentations and two possible responses leads to
four possible outcomes on a given trial (the four
cells of the table). When the signal is present and
the response is yes the observer makes a hit. But
if the observer responds yes when the signal is ab¬
sent, then a false alarm is made. The other cells
are called misses and correct negatives. The rela¬
tionships among these responses depend not only
on the nature of the signal but also on the decision
processes occurring within the observer.
Consider a typical experiment as an example.
Suppose we want to measure an observer’s ability
to detect a tone. The tone for a given experiment
will be constant in intensity and frequency. After a
ready signal, the observer is required to respond by
pushing one button to indicate “Yes, the signal
tone was present” and a different button to signify
“No, it was not.” Let us also consider some dif¬
ferent experimental conditions that might be intro¬
duced. The first is one in which the signal was
presented in 50 percent of the trials, and no signal
was presented for the remaining 50 percent. A typ¬
ical set of data for one observer, expressing the pro¬
portions of trials on which the four possible out¬
comes occurred (the outcome matrix), is shown in
Table 2-3.
Notice that on 25 percent of the trials when
the signal was absent the observer responded “Yes,
the signal was present.” Why should the observer
report that a signal was present when it was not?
First, clearly he is not always sure that whatever he
heard was actually the signal. Because of this many

nonsensory aspects of the situation might influence
his pattern of responding. Consider the effect of his
expectations. If the observer knows that the signal
is present on almost every trial he might find him¬
self responding yes to even the faintest or most am¬
biguous of sensations (perhaps even generated by
endogenous noise in his own nervous system). This
is sensible behavior if the stimulus occurs most of
the time, because on these “doubtful” trials he will
quite often be correct. However, if the signal rarely
occurs, he would be less tempted by ambiguous,
faint sensations and might want to wait until he ex¬
perienced a stronger sensation before saying yes.
If our description of what the observer is
doing is correct, then we should be able to change
his response pattern by changing his expectations,
even though his sensitivity remains the same. Typ¬
ical results from the same observer are presented in
Table 2-4. In one case the signal was present in 90
percent of the trials and in the other only 10 percent
of the trials. Notice that when the signal is occur¬
ring frequently the observer says yes often. This
gives him many hits, but also many false alarms.
When he expects the signal only occasionally he

Table 2-4. Outcome Matrices for Two Different
Conditions
Stimulus present 90 percent of the time

Response
Signal

Yes

No

Present
Absent

0.95
0.63

0.05
0.37

Table 2-3. Outcome Matrix When Stimulus Is
Present 50 Percent of the Time

Stimulus present 10 percent of the time

Response

Response

Signal

Yes

No

Signal

Yes

No

Present
Absent

0.75
0.25

0.25
0.75

Present
Absent

0.35
0.04

0.65
0.96

Detection

says no more often, thus reducing the number of
false alarms, but also reducing the number of hits.
How, then, do we measure the observer’s sensitiv¬
ity? By our former definition of threshold (the point
at which a signal is detected 50 percent of the
time), the tone is clearly above threshold in the first
instance, whereas in the second it is clearly below
threshold. This does not make sense, since neither
the tone’s strength nor the observer’s sensitivity has
changed. We need some way of separating the ob¬
server’s sensitivity from his decision strategy.
We can approach such a method of analysis
by exploring how the observer’s responses change
for a particular signal strength if we vary only his
expectations by varying the relative frequency with
which the signal occurs. We will obtain proportions
of hits and false alarms for each different signal
probability, as we discussed above. If these pro¬
portions of hits and false alarms are plotted against
each other as in Figure 2-3, we obtain a receiver
operating characteristic curve (frequently abbre¬
viated ROC curve), which displays the relation¬
ship between proportions of hits and false alarms
as the likelihood of the signal changes. The termi¬
nology was inherited from the communications en¬
gineers who first developed signal detection theory.
A more descriptive term for those interested in per¬
ception would be isosensitivity curve, since the
curve represents the range of possible outcome
matrices for one level of sensitivity. As in the pre¬
vious example, Figure 2-3 shows that when the
signal is rare, the observer frequently says no even
when the signal is presented. At the high end of the
curve, where the signal occurs frequently, the ob¬
server says yes quite often even when the signal is
not there.
An ROC curve (in any modality) reflects an
observer’s response pattern for one signal strength.
If we increase the strength of the signal, we find
that the curve has a more pronounced bow, as
shown by the curved black line in Figure 2-3. If
we decrease the signal strength, the curve becomes
flatter and approaches the 45-degree line, which
represents chance responding. Thus, the amount of

23

ROC curves. Notice how the shape
of the curve changes for different levels of
sensitivity. The black dots on the white curve
represent results with indicated probability of
signal presentation.
FIGURE 2—3

bow in the curve can serve as a measure of the
perceived signal strength. An alternative way to in¬
terpret an ROC curve is in terms of variations in
the sensitivity of an observer to a signal of a partic¬
ular strength. Thus, the two curves in Figure 2-3
could also be interpreted as reflecting two different
sensitivities of a single observer (the more bowed
the curve the more sensitive) or the curves of two
different observers with different sensitivities to the
same signal strength.
We may also vary the observer’s response pat¬
tern, while holding the signal intensity constant, by
varying the importance or the payoff for a given
response. For instance, if we pay 10 cents for every
correct detection of the stimulus and do not penal¬
ize the observer for false alarms, the optimal strat¬
egy is to guess yes on every trial. This will
maximize the amount of money that can be earned
in the test situation. Contrast this to a situation
where we deduct 10 cents for each false alarm and
do not reward for correct detections. Here a reason¬
able observer would minimize the losses by saying

24

Chapter 2/Psychophysics

no on every trial. Actually, most situations fall
somewhere between these two extremes. For in¬
stance, we might pay our observer 10 cents for
every correct response and deduct 5 cents for every
wrong response. This situation may be represented
in a matrix of numbers as shown in Table 2-5.
Such a set of rewards and penalties is called the
payoff matrix. Changing the payoff matrix causes
changes in an observer’s response pattern in much
the same way that varying an observer’s expecta¬
tions concerning stimulus frequency would, so an
observer’s motives as well as expectations affect re¬
sponses during the detection experiment. Thus, by
systematically varying the payoff matrix of an ex¬
periment, we can vary an observer’s numbers of
hits and false alarms and produce an ROC curve
similar to that generated by varying the relative fre¬
quency of signals. Note that it is the observer’s re¬
sponse pattern (e.g., the overall number of yes
responses) that varies as the ROC curve is pro¬
duced, not the sensitivity to the stimulus. Because
the manipulation of motivation in this case is done
by varying the payoff matrix, and thus the amount
of money paid to an observer, this type of experi¬
ment has been given the snide name “sweatshop
psychophysics.”
Perhaps the theoretical and methodological
bases for signal detection will become clearer if we
look at the detection problem from a different con¬
ceptual angle. We have said that even when no
stimulus is present an observer’s sensory systems
are still active, generating sensory noise. The
amount of noise probably varies from moment to
moment. This fluctuation in noise level is probably

Table 2-5. A Typical Payoff Matrix for a
Psychophysical Experiment
Response
Signal

Yes

No

Present
Absent

100
-50

-50
100

caused by the operation of physiological, attentional, and other variables on the sensory and per¬
ceptual systems of the observer. Signal detection
theorists represent these fluctuations in the form of
a probability distribution, which is graphed in
Figure 2-4 as the “signal absent” curve. The ab¬
scissa is the amount of sensory activity (or sensa¬
tion level), and the ordinate can be thought of as
the likelihood of occurrence of any particular level
of sensation over a great many trials. This means
that even in the absence of any external signal, the
observer experiences some level of sensation that is
represented by a particular location along the ab¬
scissa. This level is experienced with a relative fre¬
quency represented by the height of the curve at
that point.
When a signal is actually presented it occurs
against this background of sensory noise. Of
course, the signal produces some sensory response
of its own, which then adds to whatever amount is
already present. The effect of this is the creation of
a new distribution of sensory activity, the “signal
present” curve. On average, the level of activity
elicited by the signal added to the sensory noise is
more intense than that of the noise alone. This is
shown by the fact that the mean of the signal pres¬
ent curve is shifted toward higher values of the sen¬
sory activity axis in Figure 2-4. When the signal is
weak, however, it will not add enough sensory ac¬
tivity to make the two distributions (signal absent
vs. signal present) completely distinct. The two
distributions in Figure 2-4 would overlap if drawn
on the same set of axes. You can see from Figure
2-4 that some levels of sensation could result either
from presentations of a signal or simply from noise
alone.
Imagine you are an observer sitting inside the
head trying to decide if a signal has been presented.
The only information you have is the intensity of
the sensation. Remember, however, that sometimes
the noise produces a sensation that is just as intense
as that produced by the signal, as shown in Figure
2-4. As a rational observer, you would probably
solve this problem by setting a criterion, or cutoff

Detection

25

Criterion level (p)

plus noise
distribution
Sensory activity level

Illustration of how signal absent and signal present distributions result in hits, misses,
false alarms, and correct negatives for a particular criterion setting. Notice that the two curves are
actually plotted on the same axes—they are separated for clarity. The curves would overlap if plotted
together.

FIGURE 2-4

point, for sensation level. This is the value you are
willing to accept as probably indicating that a sig¬
nal is present. If a sensation level is below the cri¬
terion (to the left in Figure 2-4), you respond no;
if it is above the criterion, you respond yes. This
simplifies the problem greatly, since you must
only decide, based on your motives and expecta¬
tions, where to put the criterion. From that point
on, the experienced level of sensation more or less
automatically determines the response. The crite¬
rion value is usually symbolized by the Greek let¬
ter P (beta).

If this is what the observer is doing, then we
can specify the proportions of hits and false alarms
we might expect, depending on where he places his
criterion. According to signal detection theory, the
proportions of the various outcomes observed in an
experiment (see Table 2-2) may be represented as
that proportion of the area under the appropriate
probability distribution curve to the right or left of
the criterion location. Thus, if Figure 2-4 repre¬
sents an actual situation, the proportion of signal
present trials on which a yes response would be
given (the proportion of hits) is represented by the

26

Chapter 2/Psychophysics

area under the signal present curve to the right of
the criterion, since the observer would say yes
whenever the sensation level was above, or to the
right of, the criterion. Similarly, the proportion of
false alarms is represented by the area under the
signal absent curve to the right of the criterion,
since that is the proportion of trials on which the
sensation level generated by the sensory system in
the absence of a signal exceeded the criterion level
set for the yes response. The other two possible
outcomes are also represented in Figure 2-4.
The motivation and expectation effects on an
observer’s response pattern in a detection experi¬
ment are now interpretable. Essentially, these vari¬
ables affect the placement of the criterion and,
hence, the proportion of hits and false alarms. For
instance, suppose that the observer is a radiologist
looking for a light spot as evidence of cancer in a
set of chest X rays (see e.g., Swensson, 1980). If
the radiologist thinks she has found such a spot,
she calls the patient back for additional tests. The
penalty for a false alarm (additional tests when no
cancer is present) only involves some added time
and money on the part of the patient, whereas the
penalty for a miss (not catching an instance of real
cancer) might be the patient’s death. Thus, the ra¬
diologist may set a criterion value that is quite low
(lax), not wanting to miss any danger signals. This
means she will have many hits and few misses, but
also many false alarms, a situation shown in Figure
2-5A. Conversely, if the observer is a radar oper¬
ator looking for blips on a screen signifying enemy
missiles, he might be much more conservative.
Here the penalty for a false alarm could be war,
whereas the penalty for a miss might be only a few
seconds lost in sounding the alarm. He would set a
high (strict) criterion in order to avoid false alarms,
but at the penalty of reducing the number of hits.
This would be equivalent to the situation shown in
Figure 2-5B. In this same manner, each point on
any given ROC curve simply represents a different
criterion setting.
Although we indicated that the location of the
criterion alters the pattern of response, we did not
mention the effect of criterion location on the sen-

Criterion (/3)

"1 Proportion of
J false alarms
Proportion

of hits

Criterion (0)

The effect of motives or
expectations on criterion placement and
proportion of hits and false alarms.

FIGURE 2—5

sitivity of the observer. That is because there is no
such effect. In signal detection theory, sensitivity
refers to the average amount of sensory activity
generated by a given signal as compared with the

Identification

average amount of noise-generated activity. This is
similar to the everyday use of the word sensitivity.
Thus, a radio receiver that produces a large electri¬
cal response that allows a weak signal to be heard
above the background static is more sensitive than
one that produces only a small electrical response
to that signal, which may then be obscured by
static and noise.
Within our present framework, the perceptual
analog of sensitivity is the distance between the
centers (means) of the signal absent and the signal
present distributions. This is merely a measure of
the difference in average sensation levels as a func¬
tion of the presence or absence of a signal. We call
this distance measure of sensitivity d' (see Figure
2-4). When the distributions are far apart, and
overlap very little, as in Figure 2-6B, d' is large
and the ROC curve is far from the diagonal and
sharply curved. When the distributions are close to¬
gether, and overlap to a great extent, d' is rela¬
tively small, as in Figure 2-6A. The corresponding
ROC curve is close to the diagonal, which you may
remember represents zero sensitivity. Signal detec¬
tion theory attempts to measure an observer’s sen¬
sitivity to a signal independently of his decision
strategy, while acknowledging that both might af¬
fect the actual responses made in the experimental
setting. Instructions for calculating d' and /3 using
proportions of hits and false alarms obtained from
any typical signal detection experiment (e.g., Ta¬
bles 2-3 and 2-4) can be found in Computation
Box 2-1.
This must seem like an unusually elaborate
procedure for investigating a seemingly simple
problem, namely, the determination of the minimal
amount of energy necessary for stimulus detection.
However, an observer is a living organism whose
expectations and motives affect his or her percep¬
tual behaviors and judgments nearly as much as
stimulus reception itself does. These nonperceptual
effects must be removed if we are to look at the
pure sensory responses. Our original notion of an
absolute threshold has proved to be too primitive.
The detection threshold is simply a convenient sta¬
tistically defined point. As an alternative we may

FIGURE 2-6

27

The effect of sensitivity and signal

strength on d'.

use the d' measure, which provides an index of the
observer’s sensitivity to stimuli, instead of the tra¬
ditional detection threshold measures.

IDENTIFICATION
The doctor listened very carefully, paused for a
moment to adjust the stethoscope to a more com¬
fortable position, and listened again to the sounds

28

Chapter 2/Psychophysics

COMPUTATION BOX

2-1

Calculating

d'

To calculate d' and p, first obtain the outcome matrix
from a signal detection experiment. Then find the
false alarm rate from the outcome matrix in the HIT/
FA column of the accompanying table. Read across
the table to the Z column (Z is the usual label of the
abscissa of the graph of the standard bell curve). Call
the value tabled there Z(FA) and write it down. Re¬
peat these operations for the hit rate, calling the ta¬
bled value Z(HIT) and writing it down. Be careful to
record the sign of the tabled Z values—Z(HIT) will
often be negative. Then to obtain d', plug Z(FA) and
Z(HIT) into the following equation:
d' = Z(FA) - Z(HIT)
Remember that subtracting a negative number, if

and p
Z(HIT) happens to be negative, is equivalent to add¬
ing a positive number, i.e., 2 — ( — 3) = 5. The
value of p can be obtained similarly, except that you
should use the ORD column (for ordinate, the height
of the bell curve) to obtain the values of ORD(HIT)
and ORD(FA), and then plug those numbers into the
following equation:
p = ORD(HIT)/ORD(FA)
If the exact values of the hit or false alarm rate do not
appear in the table, interpolate between the nearest
surrounding values that do appear, or simply round
the hit and false alarm rates to the closest number that
does appear. Your answer shouldn’t be too far from
the exact value of d’ or p

HIT/FA

z

ORD

HIT/FA

z

ORD

.01
.02
.03
.04
.05
.08
.10
.13
.15
.18
.20
.25
.30
.35
.40
.45
.50

2.33
2.05
1.88
1.75
1.64
1.40
1.28
1.13
1.04
0.92
0.84
0.67
0.52
0.38
0.25
0.12
0.00

0.03
0.05
0.07
0.09
0.10
0.15
0.18
0.21
0.23
0.26
0.28
0.32
0.35
0.37
0.39
0.40
0.40

.50
.55
.60
.65
.70
.75
.80
.82
.85
.88
.90
.92
.95
.96
.97
.98
.99

0.00
-0.12
-0.25
-0.38
-0.52
-0.67
-0.84
-0.92
-1.04
- 1.18
-1.28
-1.40
- 1.64
-1.75
- 1.88
-2.05
-2.33

0.40
0.40
0.39
0.37
0.35
0.32
0.28
0.26
0.23
0.20
0.18
0.15
0.10
0.09
0.07
0.05
0.03

emanating from the patient’s chest. The sounds
were quite clear and distinct. The problem was
simply to decide whether they indicated a normal
or a pathological heartbeat. This doctor is wrestling

with a problem that does not involve stimulus
detection, for the sounds are clearly above the
detection threshold. However, it does involve
identifying one of a number of possible alternative

Identification

stimuli. To identify a stimulus is one of the major
tasks the perceptual system is asked to perform.
The difficulty of any identification task de¬
pends, in part, on the number of possible stimulus
alternatives an observer is asked to distinguish
among. Consider an observer who claims she can
identify her favorite brand of cola. Suppose we
gave her two unmarked glasses of cola and asked
her to sample them and try to select her own favor¬
ite brand. If she did select the correct brand we
would not be very surprised, since she would be
expected to do so 50 percent of the time by chance
alone, even if her taste buds were nonfunctional. If
our “expert” selected her own brand out of 25
brands presented to her we would be much more
likely to take her claim seriously, since the prob¬
ability that she would by chance alone find her
brand out of 25 alternatives is only 1/25. Mea¬
sures of the difficulty of the identification task
must therefore take into account the number of
stimulus alternatives.

Information Theory
To solve the problem of specifying the difficulty of
an identification task, psychologists in the early
1950s turned to ideas arising from the efforts of
engineers to assess the performance of radio and
telephone communications systems. Books by
Shannon and Weaver (1949) and by Wiener (1961)
made it clear that the problems faced by the psy¬
chophysicist and by the communications engineer
were quite similar. The engineer deals with a mes¬
sage that is transmitted through a communication
channel and decoded by someone or something at
the receiver end. The degree to which the final de¬
coded message reflects the original message de¬
pends, in part, on the ability of the system to
transmit information without distortion (this is what
is meant by the fidelity of a system), and on the
complexity of the input. The psychophysicist has
an analogous problem. Stimulus information is
transmitted to an observer through a sensory sys¬
tem, and it is then decoded in the central nervous
system. The degree to which the observer’s identi¬

29

fication of the stimulus corresponds to the actual
stimulus input will be affected both by the ability
of the sensory system to handle the stimulus input
without distortion and by the complexity of the
input.
The quantitative system for specifying the
characteristics of the input message is known as in¬
formation theory. Information theory is not really
a theory at all, but rather a system of measurement.
The amount of information in a given stimulus dis¬
play is defined so that the nature of the object being
measured is irrelevant. What, then, do we mean by
information? We mean what the everyday use of
the word implies. If you tell us that this week will
contain a Sunday morning, you have conveyed
very little information, since we know that every
week contains a Sunday morning. If you tell us that
this Sunday morning there will be a parade in hon¬
our of Jiffy the Kangaroo, you have conveyed a
great deal of information because you have speci¬
fied which one out of a large number of possible
alternative events was about to occur.
One way to quantify information is to define
it in terms of the questions a person must ask to
discover which member of a stimulus set has oc¬
curred. Suppose we had only two possible alterna¬
tives, A or B, and you were to search for the target
among them. You need only ask “Is it A?” to de¬
termine unambiguously which alternative had been
selected as the target. If you receive an answer of
“No” you know immediately that B is correct.
Similarly, if you had to determine which of four
stimuli. A, B, C, or D, had been chosen as the
target, you could determine it with two questions.
The answer to the question “Is it A or 5?” reduces
your number of possible alternatives to two, since
a “No” answer reveals that it is either C or D,
whereas a “Yes” indicates that it is A or B. We
already know that only one more question is nec¬
essary in order to identify the correct item. Each
necessary question, structured to eliminate exactly
half of the alternatives, defines a bit of informa¬
tion. Bit is a contraction of the words binary digit
(which can be either a 0 or a 1, that is, there are
two possible digits).

Chapter 2/Psychophysics

30

Table 2-6.

Log2n for Selected Numbers

Number of stimulus
alternatives (n)

Number of bits (log2n)

2
4
8
16
32
64
128
256

1
2
3
4
5
6
7
8

The number of bits of information needed to
determine exactly one stimulus alternative is the
logarithm to the base 2 of the total number of pos¬
sible stimulus alternatives. The logarithm of a
number n to the base 2, which is written log2n, is
merely the power to which the number 2 must be
raised to equal n. Thus, if we have four alternatives
we must raise 2 to the second power (i.e., 22 =
2x2 = 4) and log24 = 2. Similarly, Table 2-6
gives the corresponding number of bits for n alter¬
natives (a more detailed table can be found in Gar¬
ner, 1962). Each time the number of stimulus
alternatives is doubled the amount of information
rises by 1 bit. Of course, for intermediate values
the number of bits will not be a whole number (for
example, seven alternatives gives 2.81 bits).

Channel Capacity
It is important at this point to define the concept of
information transmission. Let us consider an ob¬
server as a channel, in the way communications en¬
gineers do. Our observer may be represented as in
Figure 2-7. A stimulus is presented to the ob¬
server, who is asked to try to identify it. By iden¬
tification we mean giving a response that is the
correct, agreed-upon label for the particular stimu¬
lus presented. We can say that information is trans¬
mitted by the observer to the extent that the
responses given match the actual labels of the

Stimulus
input

FIGURE 2-7

Observer-information
transmission channel
(processing takes
place here)

Response
output

A human information channel.

stimuli presented. That is, if the observer cor¬
rectly identifies a stimulus, and gives the correct
label as a response, information (the correct label)
has been transmitted from one end to the other,
through the channel represented by the observer.
If the response matches the stimulus perfectly for
all stimuli, then the observer is a perfect informa¬
tion transmitter.
Consider an example in which we are calling
out alphabetic letters from a set containing eight
items: A, B, C, D, F, G, H, X. If the observer
correctly identifies (response) the letter we have
called out (stimulus) then she has transmitted 3 bits
of information (log28). Suppose identification is not
perfect. This means that only some of the stimulus
information is being transmitted. Thus, if the ob¬
server hears a faint “eee” sound, with the first part
of the letter cut off, she does not know exactly
which letter was called out. However, she can
eliminate A, F, H, and X, which have no “eee”
sound; hence, she has reduced the number of stim¬
ulus alternatives by half, and we would say that 1
bit of information has been transmitted. In general,
the greater the probability that the observer will
identify the stimulus—that is, the more she “picks
up” from the presentations—the more information
she is capable of transmitting.
Consider a hypothetical experiment in which
each of four stimuli are presented 12 times and ob¬
servers are asked to identify which stimulus was
presented. In Table 2-7, Observer A shows perfect
information transmission because every time Stim¬
ulus 1 is presented our observer correctly identifies
it, and every time 2 is presented it is named cor¬
rectly. Observer B shows poorer information trans¬
mission. Notice here that when Stimulus 2 is

Identification

Table 2-7.
Observers

Stimulus-Response Matrices for Three

Observer A: Perfect information transmission

Response
Stimulus

l

1

12

2

2

3

4

12

3
4

12
12
Observer B: Some information transmission

Response
Stimulus

l

2

1

8

4

2

2

3
4

3

4

8

2

2

8

2

4

8

Observer C: No information transmission

Response
Stimulus

l

2

3

4

1

3
3
3
3

3
3
3
3

3
3
3
3

3
3
3
3

2

3
4

presented, the observer calls it Stimulus 2 most of
the time; but sometimes he calls it Stimulus 1 and
sometimes he calls it Stimulus 3. When he does say
that it is Stimulus 2, however, there is a fair like¬
lihood that it is Stimulus 2. He is much better than
Observer C, who seems to be responding without
reference to the stimulus presented. Observer C is
transmitting none of the available stimulus infor¬
mation. Formulas for computing the amount of in¬

31

formation transmitted in such experiments may be
found in Gamer and Hake (1951).
How many bits of stimulus information can an
observer transmit perfectly? Let us first look at a
group of stimuli selected from a one-dimensional
physical continuum, such as sound or light inten¬
sity. The number of stimuli from one continuum
that a subject can identify perfectly has been found
to be surprisingly small. For the judgment of the
pitch of a tone, Pollack (1952) found it to be about
5 different pitches, which is equivalent to about 2.3
bits of stimulus information. Gamer (1953) found
much the same result for loudness, around 2.1 bits.
Eriksen and Hake (1955) measured several visual
continua and found information transmission to be
limited to 2.34 bits for brightness, 2.84 bits for
size, and 3.08 bits for hue. Overall, the number of
stimuli that may be perfectly identified on any sin¬
gle continuum turns out to be approximately seven
plus or minus two (7 ± 2), depending on the par¬
ticular stimulus continuum being tested (see Miller,
1956).
This limit is called (again using communica¬
tions theory terminology) the observer’s channel
capacity, and typical measurement of channel ca¬
pacity is shown in Figure 2-8. Notice that even
though we increase the amount of information
available in the display, our subject has reached his
limit of recognition (about 2.5 bits) and can trans¬
mit no more information.
Several theories have been proposed to explain
this general finding. In the most popular of these,
the limit reflects cognitive or response processes
(e.g., Durlach & Braida, 1969; Gravetter & Lockhead, 1973; Luce, Green & Weber, 1976; Marley
6 Cook, 1984). A less popular view is that the
limit is set by the response characteristics of sen¬
sory neurons, and is thus an absolute limit for a
single sensory continuum (Norwich, 1981).
Seven seems to be a very small number of
stimuli to be able to identify. Each of us knows that
singers, for example, seem able to identify (indeed
sing) hundreds of different songs. Every one of us
can certainly identify dozens of faces and thou-

32

Chapter 2/Psychophysics

capacity is so limited for any single stimulus di¬
mension, how can this occur? The answer involves
the number of dimensions along which the stimulus

Stimulus information available (bits)

FIGURE 2-8
Channel capacity. The straight
diagonal represents perfect information
transmission. The curve represents typical
performance. The dotted horizontal line is
channel capacity.

sands of words. How can this be, in light of our
inability to transmit more than about 3 bits of in¬
formation per stimulus dimension?
You might think that if the stimuli were more
widely spaced and discriminable, channel capacity
would be higher. The discriminability of stimuli,
however, has very little effect on identification per¬
formance (Pollack, 1952). Another possible expla¬
nation is that our everyday performance is
explained by practice or repetition. Except in ex¬
treme cases, where a person might have years of
intensive practice on a single dimension, the prac¬
tice effect is also not large enough to explain our
everyday performance. For instance, you hear a
new word today and now you can identify that
word with ease, even though you have only en¬
countered it once. You can also recognize that this
word is different from every other word in your
vocabulary. We are not at all surprised at such a
performance, yet this type of identification may in¬
volve the transmission of some 16 bits of informa¬
tion (or more, depending on the total number of
words in your vocabulary). Given that our channel

varies.
For example, Pollack (1953) found that if he
varied only pitch information transmission aver¬
aged about 1.8 bits, whereas if he varied only loud¬
ness information transmission was about 1.7 bits.
When both dimensions were varied simultaneously,
however, information transmission was 3.1 bits.
This is more than was obtained for either dimen¬
sion separately, although not the 3.5 bits expected
if the information transmission on the separate di¬
mensions were simply summed. Nonetheless, the
more dimensions the stimulus varies along, the bet¬
ter recognition performance is. Certain ways of
combining dimensions seem to produce better per¬
formance, by making stimuli “stand out” more
clearly, or capitalizing on the small gains obtain¬
able by familiarity (Fockhead, 1970; Monahan &
Lockhead, 1977). Thus, by proper selection of
stimulus dimensions, Anderson and Fitts (1958)
were able to obtain information transmission levels
of 17 bits on