Attention and Effort. (1973)

Attention
and Effort
DANIEL KAHNEMAN
The Hebrew University of Jerusalem
PRENTICE-HALL, INC.,
Englewood Cliffs, New Jersey
Library of Congress Cataloging in Publication Data

KAHNEMAN, DANIEL
Attention and effort.

(Prentice-Hall series in experimental psychology)
1. Attention. I. Title.
BF321.K26
155.7'33
73-337.1)
ISBN 0-13-050518-8
To
Irah, Michael and Lenore
1973 by PRENTICE-HALL INC., Englewood Cliffs, New Jersey
All rights reserved. No part of this book may be

reproduced in any form or by any means
without permission in writing from the publisher.
Printed in the United States of America
10 9 8 7 6 5 4 3 2
London
Sydney
CANADA, LTD., Toronto
INDIA PRIVATE LIMITED, New Delhi
JAPAN, INC., Tokyo
PRENTICE-HALL INTERNATIONAL, INC.,
PRENTICE-HALL OF AUSTRALIA, PTY. LTD.,

PRENTICE-HALL OF
PRENTICE-HALL OF
PRENTICE-HALL OJi"
Contents
ix
PREFACE
1 BASIC ISSUES IN THE STUDY OF ATTENTION

Selective Aspects of Attention, 2; Intensive Aspects of Attention, 3;
Bottleneck Models of Attention, 5; A Capacity Model of Attention, 7;
Review and Preview, 12
13
2 TOWARD A THEORY OF MENTAL EFFORT

The Mobilization of Effort, 13;
The Measurement of Effort by Arousal, 17;
Time-Pressure and Momentary Effort, 24; Review,
26
3 AROUSAL AND ATTENTION

Variants of High Arousal, 28;
The Yerkes-Dodson Law and the Effects of Noise on Performance,
Effects of Arousal on Allocation Policy, 37;
The Orientation Reaction, 42; Review, 49
28
33;
vi
CONTENTS
50
4 LOOKING
Spontaneous Looking, 52; The Acquisition of Visual Information,
Eye Movements and the Spatial Orientation of Thought, 60;
Review, 64
56;
5 ATTENTIO'N AND PERCEPTION
66
Stages of Perceptual Analysis, 66; Taxonomy of Selective Attention,

Unit Formation, 72; Figural Emphasis, 76;
.
The Control of Figural Emphasis in Search, 80;
The Organization of Recognition Units, 85;
Elementary Concepts of Signal Detection Theory, 87;
Perceptual Readiness as a Criterion Bias, 91; Review, 96
70;
ATTENTION TO ATTRIBUTES
98
Discrimination Learning, 98; Selective Report of Attributes, 103;

Speeded Classification, 105; Response Conflict: The Stroop Test, 109;
Review, 110
7 FOCUSED ATTENTION-FINDINGS AND THEORIES

Experimental Studies of Focused Attention, 113;
Broadbent's Filter Theory, 120;
Treisman's Filter-Attenuation Theory, 122;
The Deutsch-Norman Theory, 123; Neisser and Hochberg, 125;
A Theoretical Synthesis, 129; Groups or Channels, 132; Review,
112
135
8 ATTENTION DIVIDED AMONG INPUTS
136
Concurrent Messages'and the Split-Span Experiment, 137;

Failures of Divided Attention, 141;
Experimental Tests of the Deutsch-Norman Theory, 143;
Successful Division of Attention, 145;
Effort and the Allocation of Attention, 148;
Modality Effects in Divided Attention, 152; Review, 154
9 SPEEDED RESPONSES TO SIMULTANEOUS AND TO

IMMEDIATELY SUCCESSIVE SIGNALS
Multiple Responses and Multiple Tasks, 156;
Inter-Response Interval and the Psychological Refractory Period, 162;
Refractoriness and Effort, 170; Other Findings and Theories, 174;
Review, 176
156
Contents
10 ATTENTION AND TASK INTERFERENCE
vii
178
Capacity Interference, 178;

Decision Bottleneck or Competition for Effort, 182;
Probe Measures of Spare Capacity, 185; Perception and Effort, 188;
Set and Other Determinants of Effort Demands, 191;
Structural Interference, 196; Interference and Effort Theory, 199;
Review, 201
REFERENCES
203
NAME INDEX
238
SUBJECT INDEX
242
Preface
There are several distinct subcultures among students of attention: investigators whose background is in studies of audition, others who think
in terms of visual perception, others who are primarily interested in
speeded performance, and some who study physiological arousal and
its multiple psychological determinants. Each of these subcultures has
tended to evolve its own language, and its particular conventions concerning the choice of experimental manipulations arid of dependent variables. Each has also developed its own biases. I have attempted in this
book to incorporate findings and ideas from these disparate sources into
a coherent formulation of attention.
The book is intended for graduate students and for advanced undergraduates studying the role of attention in perception and in performance. It consists primarily of a review of the research areas that are
commonly grouped under the label 'attention.' While the book presents
a particular interpretation of this research, I hope it may be useful to
students and to teachers who do not share this interpretation.
As will be evident to the reader, I have learned much of what I
know about attention from Donald Broadbent and Anne Treisman. It
will also be obvious that I find Ulric Neisser's approach to perception
and to cognition very congenial. A less obvious but equally important
ix
PREFACE
intellectual debt is to the late David Rapaport. While serving as his research assistant for one summer many years ago, I was introduced to
the psychoanalytic view 'of attention as energy. Many years later, having
become (as I thought) a rather tough-minded experimental psychologist,
I was surprised to discover that my understanding of attention bears
the permanent imprint of that encounter.
This text owes its existence to Jacques Mehler, who suggested
several years ago that I write a chapter on attention, and who patiently
prodded me through these years, while a misshapen chapter finally
evolved into a book. The conception of the book was influenced by my
students and collaborators, Daniel Gopher and Anat Ninio, who insistently demanded a clarification of my own views, and who also contributed to that clarification. Frequent discussions and friendly disagreements with Michael Posner and Steven Keele during the year that I
spent in Oregon inspired much of the material in the present version. I
have benefited from their scholarship as well as from their intellectual
generosity. The text also bears the marks of comments by my wife Irah,
by Ulric Neisser, Paul Obrist, Anne Treisman, Barbara and Amos
Tversky.
So~e of the ideas in this book were shaped by the results of experiments carried out in my laboratory at the Hebrew University. I
learned much from the students who conducted several of these studies:
Uri Avner, Avishai Henik, Ditza Kafry, Nurit Lass, Rina Levy and
Eythan Weg. Several able assistants participated in the project: Absalom
Bauman, David Bigeliter, Itamar Gatti, Ruth Kimchi, Noa Klein and
David Shinar. Yitzchak Hadani provided the technical expertise that
made the experiments possible.
In the preparation of the book I had valuable bibliographical help
from Bernard Goitein and Ilan Shapiro, and help that went well beyond
the standard secretarial duties from Tamar Ziv, Nira Rebaisen and
Leila Berner at the Hebrew University, from Meredith Woodward and
Karon Johnson at the Oregon Research Institute.
The book was completed during a sabbatical year spent at the
Oregon Research Institute and it is indeed a pleasure to acknowledge
the marvelous hospitality and the intellectual stimulation of my colleagues in that institution.
Finally, it is a pleasant duty to admit that this work could not have
been completed withollt financial support from various sources: The
Center for the Study of Disadvantaged Children and the Central Research Authority at the Hebrew University of Jerusalem, and Grant
No.5 SOl RR 05612 to the Oregon Research Institute.
D.K.
Jerusalem, 1973
7
Basic Issues in
the Study of Attention
The concept of attention has had an uneven career in the history of

academic psychology. When that history began in the nineteenth century,
the study of the effects of attention was a favorite topic for introspection,
and Titchener (1908) could confidently assert that "... the doctrine of
attention is the nerve of the whole psychological system, and that as men
judge of it, so shall they be judged before the general tribunal of psychology [p. 173]." This was perhaps a valid judgment of the importance
of attention, but certainly a poor prediction of the development of scientific psychology. Within a few years of Titchener's pronouncement,
the most vital movements in psychology were the Gestalt and Behaviorist
schools, and both movements attempted to do without the concept of
attention-for essentially the same reason. Although differing in their
method of investigation and in the very aims of their research, the Behaviorists and Gestalt theorists shared the conviction that the operations
which relate output (response, or percept) to input (stimulus, or field)
conform to a simple and straightforward set of rules, such as isomorphism
or conditioning. The concept of attention was unpopular because it is
most applicable where simple rules break down. Only the functionalists,
who were more interested in describing behavior than in developing
1
ATTENTION AND EFFORT
theories about it, kept alive the concern with specific aspects of attention
such as the preparatory set and the span of ap-prehension. The term
"attention" was effectively banished from the vocabulary of scientific
psychology: the dominant theorists of the day found it useless, and the
empirically inclined functionalists were more concerned with the trees
than with the forest. -Thus, in 1953 Osgood published an important text
which covered the entire field of experimental psychology and mentioned
"attention" only once, in the discussion of a particular theory of discrimination learning.
By the end of the 1950s, the situation had altered radically, and
the newly legitimized concept of attention was a central topic in an
emergent cognitive psychology. The new Zeitgeist ascribed more spontaneity and autonomy to the organism than had the classical doctrines
of behaviorism, Gestalt theory, and psychoanalysis. Spontaneity and autonomy imply some degree of local unpredictability. Indeed, the main
function of the term "attention" in post-behavioristic psychology is to
provide a label for some of the internal mechanisms that determine the
significance of stimuli and thereby make it impossible to predict behavior
by stimulus considerations alone.
SELECTIVE ASPECTS OF ATTENTION
The existence of mechanisms that control the significance of stimuli

can hardly be denied. For example, a pigeon may learn to favor a red
triangle over a green circle. On a subsequent transfer test, will the
pigeon favor a red circle over a green triangle, or will he prefer the
triangle? The behavior of different pigeons leads to different answers;
the psychologist is tempted to state-not very helpfully-that some pigeons
attend to shape while others attend to color. A sailor of the British Royal
Navy enduring a period of servitude in a psychological laboratory is presented with two simultaneous instructions on different loudspeakers; he
obeys one and is apparently oblivious to the other. A Harvard sophomore
is trained to locate specific letters in a large array, and he eventually
reports that whatever letter is designated as target seems to erupt spontaneously from an indistinct background. In a Russian laboratory, a dog
is strapped and harnessed in front of a speaker and a tone is sounded at
regular intervals. When a tone of different pitch is inserted in the series,
the dog catches its breath, moves its eyes, and pricks its ears. Recordings
of autonomic activity reveal that a complex yet orderly sequence of vascular and electrodermal changes follows the presentation of the novel
tone.
In all these situations and in many others, the organism appears
Basic Issues in the Study of Attention
to control the choice of stimuli that will be allowed, in turn, to control

its behavior. The organism selectively attends to some stimuli, or aspects
of stimulation, in preference to others.
There are many variants of selective attention. The present work
borrows a taxonomy of selective operations suggested by Treisman (1969).
Attention tasks are classified according to what they require the subject
to select: inputs (or stimuli) from a particular source; targets of a particular type; a particular attribute of objects; outputs (or responses) in a
particular category. There is growing agreement that these varieties of
selective attention are governed by different rules and are to be explained by different mechanisms.
INTENSIVE ASPEGrS OF ATTENTION
There is more to attention than mere selection. In everyday language, the term "attention" also refers to an aspect of amount and
intensity. The dictionary tells us that to attend is to apply oneselfpresumably to some task or activity. Selection is implied, because there
are always alternative activities in which one could engage, but any
schoolboy knows that applying oneself is a matter of degree. Lulled into
a pleasant state of drowsiness by his teacher's voice, the schoolboy does
not merely fail to pay attention to what the teacher says; he has less
attention to pay. A -schoolboy who reads a detective story while his
teacher speaks is guilty of improper selection. On the other hand, the
drowsy schoolboy merely suffers from, or perhaps enjoys, a generally
low level of attention.
A comprehensive treatment of the intensive aspect of attention was
offered by Berlyne (1960). He suggested that the intensity of attention
is related to the level 'of arousal, that arousal can be measured with the
aid of electrophysiological techniqlIes, and that it is largely controlled by
the properties of the stimuli to which the organism is exposed. Berlyne
(1951, 1960, 1970) also pioneered in the study of collative properties,
sllch as novelty, complexity, and incongrllity, which cause some stimuli
to be more arousing than others. He observed that the more arousing
stimuli generally tend to capture the control of behavior in situations of
response conflict.
Berlyne was mainly concerned with i.nvolllntary attention. The collative properties that he studied control an involuntary selective process
and they elicit an involuntary surge of arousal. A cognitive psychology,
however, is not congenial to studies of involuntary behavior. Perhaps as
a result, the line of investigation which Berlyne opened has not been
followed very actively,. In contrast, the study of voluntary selective atten-
tion has become one of the central topics of experimental psychology. In

voluntary attention the subject attends to stimuli because they are relevant to a task that he has chosen to perform, not because of their arousing quality. The modern study of voluntary selective attention has
therefore been' conducted with little or no reference to arousal or to the
intensive aspect of attention.
The present work contends that intensive aspects of attention must
be considered in dealing with voluntary as well as with involuntary selection. For this integration to be possible, however, the intensive aspect of
attention must be distinguished from the more inclusive concept of
arousal. Thus, the schoolboy who pays attention is not merely wide
awake, activated by his teacher's voice. He is performing work, expending his limited resources, and the more attention he pays, the harder he
works. The example suggests that the intensive aspect of attention corresponds to effort. rather than to mere wakefulness. In its physiological
manifestations effort is a special case of arousal, but there is a difference
between effort and other varieties of arousal, such as those produced by
drugs or by loud noises: the effort that a subject invests at anyone time
corresponds to what he is doing, rather than to what. is happening to him.
The identification of attention with effort suggests a reinterpretation of the correlation between arousal and involulltary attention. Novel
and surprising stimuli which spontaneously attract attention also require
a greater effort of processing than do more familiar stimuli. The surge
of arousal that follows a novel stimulus represents, at least in part, a
surge of mental effort. In this view, voluntary attention is an exertion
of effort in activities which are selected by current plans and intentions.
Involuntary attention is an exertion of effort in activities which are selected by more enduring dispositions.
As will be shown in Chapter 2, mental effort is reflected in manifestationsof arousal, such as the dilation of the pupil of the eye or the
electrodermal response. Furthermore, these measures follow second by
second the fluctuations of effort. Finally, the transient variations in the
effort that a subject invests in a task determine his ability to do something else at the same time. For example, imagine that you are conducting a conversation while driving an automobile through city traffic. As
you prepare to turn into the traffic, you normally interrupt the conversation. Physiological measures would certainly indicate a surge of arousal
at the same time, corresponding to the increased demands of the driving task.
A valid physiological measure of effort could contribute to the
solution of a basic problem of experimental psychology: the measurement of various types of mental work in common units. The problem is
indeed formidable: what common units can be applied to such activities
as conversing, driving a car, memorizing lists, and observing pictures?
Basic Issues in the Study of Attention 5
There has been one major attempt to solve this problem by using
the terms and measures of a branch of applied mathematics called the
theory of information (Attneave, 1959; Garner, 1962). This theory provides a measure of the complexity and unpredictability of both stimuli
and responses, the "bit" of information. In the context of the theory, man
is viewed as a communication channel that transmits information. The
capacity of such a channel is given in bits/ second, reHecting the rate at
which information is transmitted through it. Channel capacity has been
measured in human activities such as reading, driving a car, or playing
the piano, as well as in the operation of systems such as telephone links
or television sets. Unfortunately, estimates of human channel capacity
in different tasks, or at different stages of practice, have been too inconsistent to be useful. Indeed, the variables of stimulus discriminability
and stimulus-response compatibility are ij10re powerful determinants of
the speed and quality of performance than are the variables suggested
by the information analysis (Fitts & Posner, 1967). As cognitive psychology abandoned the measures of information theory, it was left without
a meaningful common unit to compare different tasks, and without a
valid approach to the measurement of human capacity. Physiological
measures of effort could contribute to fill these gaps.
BOTTLENECK MODELS OF ATTENTION
One of the classic dilemmas of psychology concerns the division

of attention among concurrent streams of mental activity. Whether attention is unitary or divisible was hotly debated by introspectionists in
the nineteenth century, by experimentalists since 1950, and the question
is still unanswered. Much of the research that will be reviewed in this
book was concerned directly or indirectly with this issue.
Two common observations are pertinent to the question of the
unity of attention, but the answers they suggest are contradictory. The
first of these observations is that man often performs several activities
in parallel, such as driving and talking, and apparently divides his attention between the two activities. The second basic observation is obtained
when two stimuli are presented at once: often, only one of them is perceived, while the other is completely ignored; if both are perceived, the
responses that they elicit are often made in succession rather than simultaneously. The frequent occurrence of suppression or queuing in the
organization of behavior suggests the image of a bottleneck, a stage of
internal processing which can only operate 011 one stimulus or one response at a time.
Man's sensory and motor performance is obviously constrained by
some bottlenecks in his biological constitution. Thus, man is equipped
with only a narrow beam of clear and sharp vision, and he is therefore
dependent on sequential scanning for a comprehensive look around him.
He is also equipped with a single tongue and must therefore arrange his
verbal responses in sequence. Attention theorists are concerned with the
possibility that there are similarly limited stages in the central nervous
system, which would make man unable to think, remember, perceive, or
decide more than one thing at a time.
As Chapters 7 and 8 will show, the modern study of attention has
been dominated by theories which assume a bottleneck stage somewhere
in the system, but the locus of the bottleneck has been controversial. To
introduce this issue, Figure 1-1 presents a crude outline of two models
of selective attention, in which the bottleneck is located at different
stages.
( A )
Stimulus
Stimulus
......
,,-
SENSORY
REGISTRATION
. . . AND STORAGE
.,'-
PERCEPTUAL
ANALYSrS
'"""
RESPONSE
SELECTION
~7
( B)
Stimulus
Stimulus
.....
1
-?
"
......
......
SENSORY
REGISTRATION
AND STORAGE
'"
7'
;'
......
PERCEPTUAL
ANALYSIS
RESPONSE
SELECTION
"
FIGURE I-I
Two models of selective attention.
Model A illllstrates son1e central aspects of the filter theory first

proposed by Broadbent (1957a, 1958). This theory assumes a bottleneck
at or just prior to the stage of perceptual analysis, so that only one stimulus at a time can be perceived. When two stimuli are presented at once,
one of them is perceived immediately, while the sensory information that
corresponds to the other is held briefly as an unanalyzed echo or image.
The observer can attend to such echoes and images and perceive their
content, but only after the perceptual analysis of the first message has
been completed. In this model, attention controls perception.
In model B, which is associated with the names of Deutsch and
Deutsch (1963), the bottleneck is located at or just prior to the stage of
response selection. According to this model, the meanings of all concur-
rent stimtIli are extracted in parallel and without interference. The bottleneck that imposes sequential processing is only encountered later: it
prevents the initiation of more than one response at a time, and selects
the response that best fits the requirements of the situation.
As an example of the questions to which the two models provide
different answers, consider a person at a eocktail party who actively participates in one of the many loud conversations that take place in the
room. Assun1ing that the sensory messages that correspond to several of
these conversations reach the central nervous system of the listener, we
may ask: at what point is the attended conversation favored over the
others? To what stages of perceptual analysis do the unattended messages penetrate? According to filter theory (model A) the unattended
messages are never decoded in perceptual analysis. In effect they are not
"heard.~' According to model B, all the conversations are heard, but only
one is responded to. The interested student who ponders Figure 1-1 will
probably be able to invent several of the experiments which have been
designed to a~swer such questions, and which will be discussed in some
detail in Chapters 7 and 8.
The evidence of these studies indicates that selective attention to
inputs affects perceptual analysis. This is contrary to model B. However,
man is also capable of dividing his attention between concurrent messages. This is contrary to model A. Thus, one of the main conclusions
of research on attention is that man's cognitive operations are far more
flexible than either of these bottleneck theories would suggest.
While the allocation of attention is flexible and highly responsive
to the intentions of the moment, there are pre-attentive mechanisms that
operate autonomously, outside voluntary control (Neisser, 1967). These
provide a preliminary organization to perception by a process of grouping and segmentation. The objects of perception are defined at that
stage, and subsequent processes of selective attention operate on these
objects. The general rule is that it is easy to focus attention exclusively
on a single object and difficult to divide attention amollg several objects.
Conversely, it is easy to notice several aspects or attributes of an object,
but it is difficult or impossible to prevent the perceptual analysis of
irrelevant attributes. Thus, we seem tInable to see the shape of an
object without seeing its color as well.
A CAPACITY MODEL OF ATTENTION
A capacity theory of attention provides an alternative to theories

which explain man's limitations by assuming the existence of structural
bottlenecks. Instead of such bottlenecks, a capacity theory assumes that
there is a general limit on man's capacity to perform mental work. It

also assumes that this limited capacity can be allocated with considerable freedom among concurrent activities (Moray, 1967). A capacity
theory is a theory of how one pays attention to objects and to acts. In
the present work, the terms "exert effort" and "invest capacity" will often
be used as synonymous for "pay attention."
Prior to the introduction of a capacity model, it may be useful to
briefly consider the question of how a mental activity is to be represented in a cognitive theory. As an example, consider such activities as
"recognizing the visual word CAT," "rehearsing the word BLUE," or "deciding to press the right-hand key in the display." Theories of cognitive
function usually assume that to each such activity there corresponds a
hypothetical structure, and that the activity occurs when the state of the
structure is temporarily altered. For example, many theorists would
agree that there is a structure corresponding to the word CAT: it has
been called a trace, a category state (Broadbent, 1971), a dictionary unit
(Treisman, 196,0), or a logogen (Keele, 1973; Morton, 1969a). Something
happens in that structure whenever the word CAT is presented and
recognized. The structure is specific, and its activation depends on the
presence of the appropriate specific input.
It is already known that much of the basic sensory analysis of
stin1uli proceeds in this manner. Thus, there may be one or several
neurons in the visual cortex which shift into a characteristic state of
activity whenever any conceivable visual stimulus is presented, e.g., a
corner-shape moving from left to right in a particular region of the
retina.
The recognition of specific stimuli by specialized detectors provides
an attractive model for a more general theory of the activation of cognitive
structures. Indeed, it is tempting to think of the hypothetical structure which "recog11izes" the input CAT as basically similar to a cornerdetector. In such a system, the appropriate input (from the outside world
or from the activity of other neural structures) serves as a key which
releases some of the energy contained in the structure and causes it to
generate outputs to serve as keys for other structures, and so forth. Because the structures do not share a con1mon source of energy, considerations of overall capacity are not necessary. to describe the system. Only
the structural connections betw~en components and the thresholds for
the activation of each need to be specified. Structural models of the type
illustrated in Figure 1-1 are most easily justified in such a view of information-processing.
Two observations of the present chapter suggest that such a description of information-transfer in man may be inadequate. First, it was
noted that momentary variations in the difficulty of what a subject is

trying to do are faithfully reflected in variations of his arousal level.
There would seem to be little reason for such arousal variations if energy
transfer plays no significant role in the system. The second observation
was that the ability to perform several mental activities concurrently
depends, at least in part, on the effort which each of these activities demands when performed in isolation. The driver who interrupts a conversation to make a turn is an example.
These observations suggest that the completion of a mental activity
requires two types of input to the corresponding structure: an information input specific to that strllcture, and a nonspecific input, which may
be variously labeled "effort," "capacity," or "attention." To explain man's
limited ability to carry Ollt multiple activities at the same time, a capacity
theory assumes that -the total amount of attention which can be deployed
at any time is limited. .
Not all activities of information-processing require an input of attelltion. The early stages of sensory analysis do not, since such elements
as corner detectors can be activated by sensory inputs alone. Subsequent
stages of perceptual analysis appear to demand some effort, because
they are subject to interference by intense involvement in other mental
activities. However, as Posner and Keele (1970) have noted, the demands
for effort increase as one approaches the response-end of the system. It
will be shown in Chapter 2 that covert activities such as rehearsal or
mental arithmetic are highly demanding, as are all activities which are
carried out under pressure of time.
A model of the allocation of capacity to mental activity is shown
in Figure 1-2. The model should be read beginning with the boxes
labeled Possible Activities. These boxes correspond to structures that
have received an information inpllt (not shown in the model). Each such
structure can now be "activated," i.e., each of the possible activities can
be made to occur, by an additional input of attention or effort from the
limited capacity. Unless this additional input is supplied, the activity
cannot be carried out. Any type of activity that demands attention would
be represented in the model, since all such activities compete for the
limited capacity. Activities that can be triggered by an information inptlt
alone are not considered in the model.
Different mental activities impose different demands on the limited
capacity. An easy task demands little effort, and a difficult task demands
mllch. When the supply of attention does not meet the demands, per-
formance falters, or fails entirely. According to the model, an activity
can fail, either because there is altogether not enough capacity to meet
its demands or becallse the allocation policy channels available capacity
lOA TTENTION
AND EFFORT
MISCELLANEOUS
DETERMINANTS
MISCELLANEOUS
.......~ MANIFESTATIONS
OF AROUSAL
AROUSAL
rlJvv'L
ENDURING
DISPOSITIONS
!AVAILABLE;
:CAPACITY:
I
MOMENTARY
INTENTIONS
POSSIBLE ACTIVITIES
RESPONSES
FIGURE 1-2
A capacity model for attention.
to other activities. In addition, of course, an action can fail because the

input of relevant information was insufficient. Thus, we may fail to detect or recognize a signal because we were not paying attention to it.
But there are signals so faint that no amount of attention can make them
plain.
A capacity theory must deal with three central questions: (1) What
makes an activity more or less demanding? (2) What factors control the
total amount of capacity available at any time? (3) What are the rules
of the allocation policy? These questions will be considered in detail in
Chapter 2" and occasionally in subsequent chapters. Figure 1-2 merely
illustrates some of the interactions between elements of the model that
will be llsed in that analysis.
The key observation that variations of physiological arousal accompany variations of effort shows that the limited capacity and the arQusal
system must be closely related. In Figure 1-2, a wavy line suggests that
capacity and arousal vary together in the low range of arousal levels.
In addition, arousal and capacity both increase or decrease according
to the changing demands of Cllrrent activities.
11
The two central elements of the model are the allocation policy
and the evaluation of demands on the limited capacity. The evaluation
of demands is the governor system that causes capacity (or effort) to be
supplied, as needed by the activities that the allocation policy has selected. The policy itself is controlled by four factors: (1) Enduring dispositions which reflect the rules of involuntary attention (e.g., allocate
capacity to any novel signal; to any object in sudden motion; to any
conversation in which one's name is mentioned); (2) Momentary intentions (e.g., listen to the voice on the right earphone; look for a redheaded
man with a scar); (3) The evaluation of demands: there appears to be
a rule that when two activities demand more capacity than is available,
one is completed (see Chap. 8); (4) Effects of arousal: systematic
changes of allocation policy in high arousal will be discussed in Chapter 3.
The capacity model of Figllre 1-2 is intended to complement rather
than supersede models of the structure of information-processing such
as those illustrated in Figure 1-1. The two figures, in fact, belong to different types: the models of Figure 1-1 are schematic flow-charts that
describe the sequence of operations that are applied to a set of simultaneous stimuli. In contrast, Figure 1-2 is a control diagram that describes
the relations of influence and control between components of a system.
For example, Figure 1-2 implies that a state of overload in which the
demands of ongoing activities exceed available capacity will induce a
compensatory increase of both arousal and capacity.
The present chapter has illustrated two types of attention theories,
which respectively emphasize the structural limitations of the mental
system and its capacity limitations. Both types of theory predict that
concurrent activities are likely to be mutually interfering, but they
ascribe the interference to different causes. In a structural model, interference occurs when the same mechanism is required to carry out two
incompatible operations at the same time. In a capacity model, interference occurs when the demands of two activities exceed available capacity. Thus, a structural model implies that interference between tasks
is specific, and depends on the degree to which the tasks call for the
same mechanisms. In a capacity model, interference is nonspecific, and
it depends only on the demands of both tasks. As Chapters 8 and 10 will
show, both types of interference occur. Studies of selective and divided
attention indicate that the deployment of attention is more flexible than
is expected under the assumption of a structural bottleneck, but it is
more constrained than is expected under the assumption of free allocation of capacity. A comprehensive treatment of attention must therefore
incorporate considerations of both structure and capacity.
12
REVIEW AND PREVIEW
The major themes of this book have been outlined in the present
chapter. The n10st important of these themes is an attempt to integrate
the intensive and selective aspects of attention. The intensive aspect of
attention is identified with effort, and selective attention is viewed as the
selective allocation of effort to some mental activities in preference to
others. Because of the connection between effort and arousal, physiological measures of arousal can be used to measure the exertion of effort.
Some types of information-processing activities can be triggered solely
by an input of information. Others require an additional input of attention or effort. Because the total quantity of effort which can be
exerted at anyone time is limited, concurrent activities which require
attention tend to interfere with one another"
A contrast was drawn between a structural model, in which cognitive activity is limited by a bottleneck, or station at which parallel processing is impossible (see Fig. 1-1), and a capacity model in which the
limited capacity determines which activities can be carried out together
(see Fig. 1-2). Neither model is adequate alone, but each captures some
important aspects of cognitive activity.
'
These major concepts should serve as background for the study of
subsequent chapters, which review some central areas of research in attention. Chapters 2 and 3 discllss some intensive aspects of attention and
elaborate the capacity model of attention and mental effort. Chapter 4 is
devoted to looking behavior. Some variants of selective attention are discussed in Chapter 5, which presents a model of the role of attention in
perception. A brief review of attention to attributes in Chapter 6 is followed by a more thorough review of focused and divided attention with
simultaneous inputs (Chaps. 7 and 8). The division of attention between
simultaneous or immediately successive speeded responses is discussed
in Chapter 9. Chapter 10 returns to the concept of effort and its measurement by task interference.
The interested student will find additional relevant material in
several recent texts (Broadbent, 1971; Keele, 1973; Moray, 19'69a, 1969b;
Norman, 1969a). A vast amount of research relevant to attention is conveniently available in special volumes of the journal Acta Psychologica,
published in 1967, 1969, and 1970. Kornblum (1972) has edited an additional volume in this series. For a humbling look at what was known
about attention at the turn of the century, a text by Pillsbury (1908)
should be consulted. Woodworth (1938) also reviews much research
which remains relevant and illteresting, although it is rarely cited in recent work.
2
Toward a Theory
of Mental Effort
This chapter elaborates the capacity model that was introduced in Figure 1-2. The first section is concerned with the control of effort by the
feedback loop leading from the Evaluation of Demands on Capacity to
the Arousal-Capacity system. The second section summarizes the evidence that arousal varies with momentary changes in the load' imposed
by mental' activity. Some determinants of the effort requirements of
various activities are discussed in the final section.
THE MOBILIZATION OF EFFORT
The capacity model shown in Figure 1-2 assumed that the capacity
which can be allocated to various activities is limited. It also assumed
that the limit varies with the level of arousal: more capacity is available
when arousal is moderately high than when arousal is low. Finally, it
assumed that momentary capacity, attention, or effort (the three terms are
interchangeable in this context) is controlled by feedback from the execution of ongoing activities: a rise in the demands of these activities
causes an increase in the level of arousal, effort, and attention.
13
14
The key observations suggesting this model will be discussed in detail in the next section, where it will be shown that physiological arollsal
varies second by second when a subject is engaged in a task, and that
these variations correspond to momentary changes in the demands imposed by the task. Thus, aro,usal and effort are usually not determined
prior to the action: they vary continuously, depending on the load which
is imposed by what one does at any instant of time.
A crude physical analogy may help clarify these ideas. When you
push a slice 0. bread into the toaster, this increases the load on the general electric supply. Witholtt a countervailing change, the new, load
would cause the voltage supplied to all users to drop. However, the
generator that supplies the current is equipped with a governor system
which immediately causes more fuel to be burned to restore the constant
voltage. In this manner, the total power that the generator supplies
varies continuously as a function of the load which is imposed by the
momentary choices of the consumers of electricity.
The analogy can be pursued further. Note that, as a user of electric
power, you rarely control the amount of power that you require in a continuous or graded fashion. All you decide is that a certain aim is to be
achieved, whether it be toasting a bun or illuminating a room. How
much power is drawn depends on the structure of the elements that you
switch on. As a first approximation, the same rule applies to mental work
as well. In general, we merely decide what aims we wish to achieve. The
activities in which we then engage determine the effort that we exert.
An important observation in studies of physiological arousal and
performance is that arousal varies with the difficulty, of different tasks, as
measured by error rate. This apparently reasonable finding is actually
quite puzzling. At an intermediate level of difficulty, the subject makes
a significant number of errors. Yet he does not work as hard as he can,
since he exerts greater effort when difficulty is further increased. Why,
then, does the subject not work harder at the initial level of difficulty,
and avoid all errors?
The answer appears to be that the subject simply cannot try as
hard in a relatively easy task as he does when the task becomes more demanding. The reader may wish to confirm this by an armchair experiment. First, try to mentally multiply 83 by 27. Having completed this
task, imagine that you are going to be given four numbers, and that
your life depends on your ability to retain them for ten seconds. The
numbers are seven, two, five, nine. Having completed the second task,
it may appear believable that) even to save one's life, one cannot work as
hard in retaining four digits as one must work to complete a mental
multiplication of two-digit numbers.
In an attempt to study this question experimentally, subjects were
Toward a Theory of Mental Effort
15
asked to perform an easy and a relatively difficult task separately, under

varying conditions of monetary incentive and risk (Kahneman, Peavler &
Onuska, 1968). We did not threaten our subjects' lives but merely rewarded or penalized the'ill ten cents on so-called High-Incentive trials
and two-cents on Low-Incentive trials. The diameter of the pupil of the
eye was recorded. The incentive had a marginal effect on this manifestation of arousal in the easy task condition, and no effect whatever in the
more difficult task. The major determinant of arousal was the difficulty
of the task.
This study of incentives is far from conclusive. However, it is consistent with the general .hypothesis that the effort invested in a task is
mainly determined by the intrinsic demands of the task, and that voluntary control ove'r effort is quite limited. Of course, voluntary control of
stop-or-gochoices is retained: we can stop working at any time, and
often do. How hard we work, when we do, seems to depend primarily
on the nature of the activity in which we choose to engage. The tentative conclusion, then, is that the p.erformance of any activity is associated
with the .allocation of a certain amount of effort. This standard allocation
does not yield errorless performance. Allocating less effort than the standard probably will cause a deterioration of performance. Allocating more
than the standard seems to be beyond our ability.
Consider again the electrical analogy. In that analogy, the concept
of a limited capacity has a precise meaning. The generator can only supply a certain amount of power. When the demands exceed that amount,
the addition of one more toaster or .air conditioner to the circuit no longer
results in a corresponding increase of electrical output. In some systems,
overload actually causes the total power supplied by the source to
decrease.
Supply =
Demand /
'y/
/
Total capacity / /
o&. "
,,"
,'-
~r~---
..".",.""""/
/
/
~
Capacity supplied
to pri mary task
Capac ity demanded by pri mary task
FIGURE 2-1
Supply of effort as a function of demands of a primary task.
16
A hypothesis concerning the human response to demands on effort

is illustrated in Figure 2-1. The illustration refers to a situation in which
the subject engages in a particular activity as his primary task. The allocation of effort to that task and the total effort allocated to all activities
are shown as a function of the demands of the primary task. Figure 2-1
.suggests that capacity (effort) increases steadily with increasing demands
of the primary task. However, the increase is insufficient to maintain performance at a constant level of speed and quality. As the demands of
the task increase, the discrepancy between the effort demanded and the
effort actually supplied increases steadily.
An additional suggestion in Figure 2-1 is that some effort is exerted
even when task demands are at zero. The continuous monitoring of our
surroundings probably occupies some capacity even in the most relaxed
conscious state. This is labeled spare capacity. The figure illustrates the
hypothesis that spare capacity decreases as the effort invested in the
primary task increases: attention is withdrawn from perceptual monitoring and concentrated on the main task. According to a hypothesis stated
by Easterbrook (1959), such a change of allocation occurs whenever
arousal is high (see pp. 37-42).
A measure of spare capacity can be obtained by studying the response to a probe signal, which is shown to the subject at an unpredictable time during the performance of the primary task (e.g., Kahneman,
1970; Kahneman, Beatty & Pollack, 1967; Posner & Boies, 1971; Posner & Keele, 1968; Posner & Klein, 1972; Shulman & Greenberg, ,1971).
As will be shown in Chapter 10, a failure to identify a signal that is normally identified with ease or an unusually slow response provides evidence that spare capacity is reduced by task performance. The logic of
these methods is that they indicate how much attention was deployed
in monitoring at the instant of signal presentation. A failure of attention
at that time necessarily causes a slowing of the response, and it may
cause a failure to identify a target, if the target is removed before attention can be drawn to it.
Interference between tasks is due to the insufficient response of the
system to demands, and to the narrowing of attention when effort is
high. Interference will occur even when the total load on the system is far below total capacity. However, the amount of interference
is an increasing function of load. At low values of load, the response of
the system is approximately linear, and there may be little or no interference between tasks in that region.
It is sometimes assumed that all the capacity of the individual is
applied to a primary task, and the occurrence of errors in that task is
used as evidence that such is the case (e.g., Shiffrin & Gardner, 1972).
The reasoning seems to be that if the individual had more capacity at
17
his disposal, he would surely use it to reduce his error-rate. This view assumes that effort is maximal whenever a well-motivated subject engages
in a task in which he makes some errors, regardless of how difficult the
task is. In fact, tasks at different levels of complexity elicit different degrees of arousal and demand different amounts of attention and effort.
The present section has elaborated the connection between two elements of the capacity model that was introduced in Figure 1-2: the
Evaluation of Demands on Capacity and the Arousal-Capacity system.
The main assumption of the model is that the mobilization of effort in a
task is controlled by the demands of the task, rather than by the performer's intentions. In addition, the system response is assumed to be insufficient, with an increasing gap between demand and supply when
overload is approached. Finally, it is assumed that the spare capacity
which is devoted to continuous activities of perceptual monitoring decreases with increasing involvement in a primary task.
THE MEASUREMENT OF EFFORT BY AROUSAL
According to the capacity !ll0del introduced in the first chapter, the

level of arousal is controlled by two sets of factors: (1) the demands
imposed by the activities in which the organism engages, or prepares to
engage; and (2) miscellaneous determinants, including the prevailing intensity of stimulation and the physiological effects of drugs or drive
states. Thus, as illustrated in Figure 2.-2, a state of high arousal may reflect what the subject is doing and the effort he is investing, or it may
reflect what is happening to the subject, and the stress to which he is
exposed. The fundamental difficulty in the use of physiological techniques to measure effort is caused by the similarity between the physiological responses to mental effort and to stress.
There have been some attempts to identify distinctive physiological
concommitants of effort, but the search for such measures has not been
very successful. One index that appears promising is a reduction of sinus
arrythmia: irregularities of heart rate tend to disappear during the performance of continuous tasks (Kalsbeek & Ettema, 1963, 1964). Porges
(1972) reported that subjects who show the greatest reduction of cardiac
variability during a task also tend to have the fastest RT's. The reduction
of autonomic variability during task performance is apparently a general
effect: rhythmic contractions and dilations of the pupil, which are prevalent at rest, are virtually abolished during the performance of mental
arithmetic (Kahneman & Beatty, 1966, unpublished observations), and
Thackray (1969) has found an inhibition of variability in other measures
of autonomic activity during task performance. While promising, th~se
18
MISCELLANEOUS SOURCES
OF AROUSAL:
anxi ety f fear/anger f
sexual excitement,
muscular strain, effects of
drugs, intense stimulation, etc.
AROUSAL
r-Lrvv--L
I
:AVAILABLE:
!CAPACITY:
MISCELLANEOUS
.......... MANIFESTATIONS OF AROUSAL:
pupillary dilation,
inc reased sk in
conductance, fast
pulse etc.
ALLOCATION
POLICY
/ / \
...----'//
available capacity and

arousal increase to meet
demands for processi ng
capacity
~---
D11~~ ~~_~D
POSSIBLE ACTIVITIES
RESPONSES
FIGURE 2-2
Effort and other determinants of arousal.
specific measures of effort have had little application, and two standard
measures of sympathetic activity remain the most useful autonomic indications of effort: dilation of the pupil is the best single index and an
increase of skin conductance provides a related, but less satisfactory
measure (Colman & Paivio, 1969; Kahneman, Tursky, Shapiro & Crider,
1969). A third measure of sympathetic dominance, increased heart rate,
cannot be used as a measure of effort, for reasons that will be described
in Chapter 3.
A useful physiological measure of mental effort must be sensitive
to both between-tasks and within-task variations. That is, it should order
tasks by their difficulty, since more difficult tasks usually demand greater
effort. It should also reflect transient variations of the subject's effort during the performance of a particular task. A perfect measure of mental
effort would also reflect between-subject differences, i.e., differences in
the amount of effort that different people invest in a given task. There is
19
little evidence concerning the third point (Kahneman & Peavler, 1969;
Peavler, 1969), but measurements of pupil diameter appear to meet the
first two requirements, and they provide a sensitive indication of both
between-tasks and within-task variations of effort (see Goldwater, 1972,
for a comprehensive review).
The claim that pupillary dilations indicate mental effort was made
by Hess and Polt (1964; Hess, 1965), who observed a striking correspondence between the difficulty of mental arithmetic problems and the
n1agnitllde of the dilation during the solution period. The correspondence between cognitive load and pupillary dilation was later confirmed
in many contexts: arithmetic (Bradshaw, 1968b; Payne, Perry & Harasymin, 1968); short-term memory tasks of varying load (Kahneman &
Beatty, 1966); pitch discriminations of varying difficulties (Kahneman &
Beatty, 1967); standard tests of "concentration" (Bradshaw, 1968a); sentence comprehension (Wright & Kahneman, 1971); paired-associate learning (Colman & Paivio, 1970; Kahneman & Peavler, 1969); imagery tasks
with abstract and with concrete words (Paivio & Simpson, 1966, 1968;
Simpson & Paivio, 1968), and the emission of a freely selected motor response instead of an instructed response (Simpson & Hale, 1969). In all
these situations, the amount of dilation increases with task demand or
difficulty. The relation between attention and pupillary dilation is maintained even in the absence of specific task instructions: Libby, Lacey,
and Lacey (1973) observed dilations of the pupil when the subject
merely looked at pictures. The largest dilations occurred while looking at
"interesting" and "attention-getting" pictures (see Fig. 3-1 on p. 30).
Pratt (1970) also observed that the pupillary dilation varied with the
unpredictability of random shapes to which subjects were exposed. Evidently, complex and interesting pictures, like difficult tasks, attract attention and demand a relatively large investment of effort.
The second test of an adequate measure of effort is within-task
sensitivity. Several studies have confirmed the suggestion (Hess, 1965)
that the size of the pupil at any time during performance reflects the
subject's momentary involvement in the task. Indeed, the fidelity of the
pupil response permits a second-by-second analysis of task-load and
effort. Kahneman and Beatty (1966), for example, showed that the presentation of each successive digit in a short-term memory task is accompanied by a dilation of the pupil. The increase in pupil diameter
corresponds to the increasing rate of rehearsal which is imposed by the
presentation of the additional digit. This pattern of rehearsal can be
altered by presenting the items in several groups, separated by pauses.
Then, a brief dilation of the pupil occurs after the presentation of each
group, corresponding to the spurt of rehearsal during each pause (Kahne-'
man, Onuska & Wolman, 1968). Finally, when a subject is informed that
20
he need no longer retain the digits he has heard, his pupil briefly dilates,
then constricts, as he ceases to rehearse (Johnson, 1971).
The pupillary dilation is a relatively fast response, and major dilations can occur within one second after the presentation of a demanding
stimulus. Thus, Beatty and Kahneman (1966) showed that the pupil
dilates about 10 percent of base diameter during the first second following the presentation of a familiar name, when the subject must respond
by the appropriate telephone number. Similarly, in a pitch discrimination task, the diameter of the pupil reaches a maximum within one second of the presentation of the critical tone; the size of the pupil at that
time faithfully reflects the difficulty of the discrimination (Kahneman &
Beatty, 1967). When subjects are required to produce an image that corresponds to a particular word, pupil diameter reaches its maximal value
faster with concrete than with abstract words (Colman & Paivio, 1969;
Paivio & Simpson, 1968; Simpson, Molloy, Hale & Climan, 1968). A plausible explanation of this finding is that the visual image is produced
sooner for concrete than for abstract words.
To further test the validity of the pupillary measure of effort, a
behavioral measure of spare capacity was introduced. Subjects were
required to perform two tasks simultaneously. The primary task involved
the transformation of a digit string: the subject heard a series of four
digits (e.g., 3916) at a rate of one digiti second, and he was instructed to
pause for a second, then to respond with a transform of that series
(4027), adding 1 to each digit of the original set. In addition, the subjects
performed a secondary task. In one experiment (Kahneman, Beatty &
Pollack, '1967), a series of letters was flashed in quick succession, and
the subjects monitored the display for the occurrence of a "K." In
another experiment (Kahneman, 1970), the subjects were briefly shown a
single letter, which was to be reported after the completion of the digittransformation task. The payoff structure in these experiments was designed to ensure priority for the digit-transformation task: the subject
was paid for the visual task only if he had performed the transformation
task adequately.
Figure 2-3 shows the results of these studies. It includes four
curves: (1) a typical pupillary response to the digit-transformation task;
(2) the average percentage of missed K's as a function of the time of
their presentation; (3) the average percentage of incorrectly reported
letters as a function of the time of their presentation; and (4) the average
percentage of failures in the digit-transformation task, as a function of
the time of presentation of the visual letter.
The most important feature of Figure 2-3 is that ~he pupillary response and two different behavioral measures of spare capacity show
similar trends, although the pupil appears to lag slightly. As a first ap-
0----0 Pupil response

0--- ....0
Detection of 'K I
Letter
...---.. identification
. - -... Digit task
70
0.7
60
. 0.6
50
0.5
,
en
(l)
L.
.~
:::>
.....
c
(l)
u
L.
(l)
0-
\,
40
0.4
\,
\
30
z
a
'"0
>-..
a'-
0.3 .0..
::>
a..
\
\
20
.",.
........
---- ........, "
0.2
'-...... 'e
0.1
10
Report
o
Time (Sec)
FIGURE 2-3
Two measures of perceptual deficit and the pupillary response to a digittransformation task. Also shown, the probability. of success in the transformation task as a function of the time of occurrence of the visual target. (Sources:
Kahneman, Beatty & Pollack, 1967; Kahneman, Tursky, Shapiro & Crider,
1969; Kahneman, 1970, with permission).
21
22
proximatioll, a decrement of 10 percent in the likelihood of detecting a

K is associated with an increase of 0.2 mm in pupil diameter. Thedecrement of performance is not caused by the dilation of the pupil, however,
since similar decrements are observed when the subject sights the target
throllgh an artificial pupil. Thus, the physiological and behavioral measlIres are indepelldent indices of the momentary effort invested in the
primary task. Another significant feature is that performance of the primary task appears to be completely independent of the timing of the
critical visual event. In the experiments summarized in the figure, a
letter that could interfere with the main task was simply not seen, and
the performance of the primary task was thereby protected. This strategy, however, is readily altered by modifying the payoffs (Kahneman,
1970). Finally, Figure 2-3 shows that visual performance was severely
impaired during the pause between the two parts of the digit-transformation task, a time at which the subject was engaged neither in listening
nor in speaking. This observation indicates that mental effort, rather than
involvement ill either perception or overt respons,e, was the cause of the
perceptual deficit. Thus, the results of Figure 2-3 prOVide support for
three central themes of the present chapter: (1) there is a limited capacity for effort, which can' be allocated to different tasks; (2.) the subject's intentions govern the allocation of this capacity in a highly flexible
manner; (3) physiological variables, such as pupil size, provide a useful
measure of the momentary exertion of effort.
An additional methodological point should be noted: the pupillary'
method yields a reliable effort curve of the type illustrated in Figure 2-3
in two or three trials, because the entire response is measuted on each
occasion. In c011trast, dozens of trials are needed to obtain equally reliable results by a behavioral method, in which a single temporal position
is probed on each trial.
These demonstrations leave little doubt that pupillary dilations reflect effort. However, much to the chagrin of the student of effort, dilations also occur in other psychological states. As Figure 2-2 indicated,
there are many determinants of arousal which all affect autonomic func..
tions in similar ways (Nunally, Knott, Duchnowsky & Parker, 1967). In
order to ascribe a particular autonomic change to mental effort, the investigator must therefore assume the burden of proving that this change
is not due to such miscellaneous determinants of arousal as muscular
strain or anxiety.
Fortunately, the evidence suggests that these contaminating factors
play a relatively small part in arousal variations that occur during the
performance of mental tasks. The issue of muscular strain arises, for
example, whenever a subject must verbalize his responses, but verbalization as such has little effect on the pupil. Figure 2-4 shows the results
23
4.70
Say Add 1
Think Add 1
4. 50
..... - - -- Say Add 0

0- - -
E
E
c
-0
Think Add 0
4.30
s-
Q)
Q)
E
c
-0
4.10-
.0.
:>
0..
3.90
1 2
10
12
14
16
18
20
22
24
26
Ti me (seconds)
FIGURE 2-4
Pupillary responses to two tasks under instructions to say response twice
(say), or to think response and then say it, (think). (From Kahneman, Peavler
& Onuska, 1968, with permission.)
of an experiment in which subjects heard a string of four digits and were

instructed either to repeat the string (Add 0) or to transform the string
by adding 1 to each digit (Add 1). In the "Say" condition, they repeated
the response twice. In the "Think" condition, they were instructed to
"think" their answer first, in time with recorded beats, then to say it.
The subjects were given the task instructions (e.g., Say-Add-1) on
seconds 4-6 of the trial; then they heard the digits, said or "thought"
their answer; and always said the answer on seconds 20-23. The anticipation of the "Say" task caused the pupil to be larger when the presentation of the digits began, and there were other significant effects of
verbalization during the task, but these effects were slight in comparison
to the effect of task difficulty.
It is often suggested that observed autonomic responses indicate
anxiety rather than effort, because it seems reasonable that difficult tasks
are associated with high levels of test anxiety. However, this hypothesis
would also imply a substantial difference between the conditions of
"Think" and "Say" in Figure 2-4, and this was not found. There is other
24
evidence tllat momentary fluctuations of anxiety play a limited role in

determining the pupillary responses in task situations (Kahneman &
Peavler, 1969'; Kahneman, Peavler & Onuska, 1968; Kahneman & Wright,
1971). Specifically, the anxiety hypothesis would predict a high level of
arousal not only at the instant of effort, but also in anticipation of failure
and immediately following failure. In fact, the pupil is always largest
dllring the performance of the task, rather than earlier or later. Furthermore, the pupillary dilations which accompany correct responses are
often larger than the dilations which accompany failures. Thus, neither
muscular strain nor anxiety can aCCOllnt for most of the pupillary changes
that occur during mental activity. Nevertheless, the possibility of confounding effects must be cautiously considered in each experiment which
relies on measures of arousal to study mental effort (Kahneman &
Wright, 1971).
The reader may wish to confirm some of the previous conclusions
for himself, and this is easily done. Face a mirror, look at your eyes
and invent a mathematical problem, such as 81 times 17. Try to solve
the problem and watch your pupil at the same time, a rather difficult
exercise in divided attention. After a few attempts, almost everyone is
able to observe the pupillary dilation that accompanies mental effort,
in a situation which elicits neither overt responses nor test anxiety.
TIME-PRESSURE AND MOMENTARY EFFORT
The studies which validated the pupillary measure of effort usually

compared several tasks of the sanle type, but of different levels of difficulty. Almost invariably, the most difficult version of a task caused the
largest pupillary dilation. Among tasks of the same type, it is usually
easy to determine a ranking of difficulty by considering the complexity
of each task, the speed at which it can be performed, or the probability
of failure. It is far more difficult to compare tasks of different types,
since neither complexity, speed, nor errors retain a common significance
in such comparisons.
The study of pupillary responses, or of other physiological measures of effort, could contribute to such comparisons between tasks of
different types and structllres. Some rather puzzling results are already
available, which must be considered in a theory of effort. In' studies of
paired-associate learning, for example, the dilation which occurs when
the subject's recall is tested may be four to six times as large as the
dilation which OCCllrs when the subject attempts to memorize an item
(Kahneman & Peavler, 1969). Can it be inferred that learning requires
much less effort than recall? Large pupillary responses accompany other
tasks that could be considered "easy," such as the prompted recall of
25
thoroughly overlearned information: one's telephone number or one's

(Beatty & Kahneman, 1966; Schaefer, Ferguson, Klein & Rawson,
1968). Similarly, retaining five digits for immediate recall is considered
since it is a task in which we rarely fail. Nevertheless, larger dilaoccur in this simple task than in an apparently more complex task,
subjects .are required to listen to a long message and comprehend
it (Carver, 1971).
It is apparent from these observations that the intuitive notion of
difficulty is not sufficient to determine the amount of effort that a
task demands. The problem arises at least in part because of the vague11esS of the notion of difficulty. Thus, difficulty is often identified with the
likelihood of error. By this definition, retaining nine digits in a test of
short-term memory is extraordinarily difficult. By the same definition,
crossing out every letter A in this book is also very difficu~t, since a few
will almost certainly be missed. However, retaining nine digits and crossing out A's impose different demands at any instant in time.
The momentary effort that a task demands must be distinguished
from the total amount of work that is required to complete that task. The
momentary effort exerted in running the 60-yard dash is greater than
the effort exerted in walking two miles at a comfortable pace, although the
total expenditure of energy is surely greater in the second task. In the
terms of this analogy, much of our mental life appears to be carried out
at the pace of a very sedate walk. When one reads a book or listens to a
lecture, for example, effort is minimal because the material is not actively rehearsed, and because the redundancy of the message reduces
any sense of time-pressure.
Furthermore, the amount of genuinely new information acquired
per unit time in such situations is probably small. Murdock (1960) estimated that subjects presented with a long list of unrelated words transfer information into long-term memory at t~e strikingly slow rate of 3.6
words/minute. Memory for connected discourse appears to be better
only because of the effects of prior knowledge and redundancy. Thus,
it is not inconceivable that continuous mental activities, such as reading,
tax our capacity only rarely. We cover great distances by such mental
walking, with only minimal effort.
This conception of mental work suggests that time-pressure must
be an important determinant of effort. This is a familiar idea in the context of physical exertion: anyone who has tried jogging knows that even
a small increase of speed beyond the relatively effortless "natural" speed
causes a disproportionate increase in the sense of strain.
Time-pressure is often involved in mental tasks. It is sometimes imposed by explicit instructions to hurry and sometimes by demand characteristics of the task. For example, Simpson and Paivio (1966, 1968)
asked subjects to produce images to words, and they observed particu__ ._r.:t.T"LJJ
26
larly large pupillary dilations when the subject was also asked to indicate the instant at which he achieved the image. Since the occurrence of
an overt response is neither a necessary nor a sufficient condition for
large pupillary dilations, it seems likely that the instruction to report the
achievement of an image induced time-pressure, and thereby increased
effort.
The most important type of time-pressure is that which is inherent
in the structure of the task. Thus, severe time-pressure necessarily arises
in any task which imposes a significant load on short-term memory, because the subject's rate of activity must be paced by the rate of decay of
the stored elements. In mental arithmetic, for instance, one must keep
track of the initial problem, of partial results already obtained, and of
the next step. Stopping or slowing even for an instant usually forces one
to return to the beginning and start again. In tests of short-term recall,
the increasing number of items that must be rehearsed causes a rapid
buildup of time-pressure, which is also reflected in autonomic measures
of arousal. Time is also critical in a pitch-discrimination task with brief
tones, where rapidly decaying traces must be quickly evaluated. In all
these tasks, large pupillary dilations occur.
Some problems are difficult because the elements that are essential
to the solution are relatively inaccessible to retrieval from memory.
Other prob,lems are difficult because they impose severe time-pressure.
The indications are that effort is less closely related to the dimension of
accessibility than to the dimension of time-pressure. During paired-associate learning, for example, the pupillary response at recall decreases
quite slowly with increasing familiarity (Kahneman & Beatty, unpublished observations). Bradshaw (1968b) has reported that the size of
pupillary dilations does not vary with the difficulty of word-construction
problems, although. it varies consistently with the difficulty of arithmetic
problems. The difference could be due to the differing roles of storage
and rehearsal in the two tasks. The more difficult arithmetic problems require more storage and rehearsal than do easier problems, and therefore impose more time-pressure. In contrast, a word problem is difficult
only because correct answers are few and inaccessible; it imposes neither
more load on storage nor more time-pressure than an easy problem, and
it does not elicit greater effort.
REVIEW
The approach to the concept of effort that was developed in this

chapter assumes that effort is mobilized in response to the changing
demands of the tasks in which one engages, and that there is a standard
27
allocation of effort for each task. The investment of less than this stan'effort causes a deterioration of performance, but in most tasks it is
impossible to completely eliminate errors by a voluntary increase of
ort beyond the standard. As a result, the voluntary control of effort is
limited in scope. It was assumed that the increased allocation of effort
difficult tasks-does not suffice to maintain performance at a constant
and that the spare capacity that remains available for perceptual
monitoring decreases with increasing involvement in a primary task.
Evidence was presented that transient variations of arousal during
performance of a mental task correspond to transient changes in the
demands of the task and to temporary decrements in behavioral measures of spare capacity. However, the measurement of effort by physiological indications of arousal such as the pupillary dilation is complicated
the fact that the manifestations of arousal are not specific to effort.
Finally, the concept of momentary effort was distinguished from
the probability of failure in a task and from the total amount of work
required by that task. Much mental activity appears to occur without
the exertion of substantial effort. Time-pressure is a particularly important determinant of momentary effort. Tasks that impose a heavy load
on short-term memory necessarily impose: severe time-pressure.
3
Arousal and Attention
The first section of this chapter describes the autonomic manifestations

of two attentional states: a state of motor activation and active manipulation of information, and a state of acceptance of sensory information
and inhibition of response. Subsequent sections are devoted to the
Yerkes-Dodson law, which describes the effects of arousal on performance, and to Easterbrook's hypothesis that high arousal causes an alteration in the allocation of attention. The final section describes the
orientation reaction, which comprises some aspects of the involuntary
allocation of attention to novel stimuli.
VARIANTS OF HIGH AROUSAL
In the preceding chapters, the concept of arousal was treated as a

unitary dimension, as if a subject's arousal state could be completely
specified by a single measurement such as the size of his pupil. This,
however, is an oversimplification. Although the idea of a dimension of
general arousal is useful, some important qualifications must be con28
Arousal and Attention 29
As this section will show, there are at least two distinctively

~ ~11 states of high arousal.
Manifestations of sympathetic dominance have traditionally been
to identify arousal level. Indeed, pulse rate, pupil diameter, and
conductance usually increase in arousing conditions. However,
(1959, 1967) has pOinted out that the concept of a unitary dimenarousal implies that the correlations among these measures should
high: if an individual is more aroused in one situation than in anall indices of sympathetic dominance should reflect this fact. The
correlations, however, are often quite low. Furthermore, systematic discrepancies between measures occur under different types of
: different stressors elicit different patterns of autonomic activity, as
as different degrees of sympathetic dominance.
In some situations, one autonomic variable may indicate sympathetic dominance even as another variable displays a typical parasympathetic response. Lacey (1967) has coined the term directional
fractionation for such discrepant patterns.
An important instance of directional fractionation was first described by Davis (1957). He observed conditions in which most indices
of sympathetic dominance rose while the pulse slowed down. Davis
found it easy to produce this response, which he labeled the P-pattern,
by showing male students pictllres of female nudes, but the effect is not
restricted to such stimuli. Thus, the presentation of visual stimuli to infants also causes a very n1arked cardiac deceleration, which is sufficiently
reliable to provide a useful index of attention (Kagan, 1972; Kagan &
Lewis, 1965; Lewis, Kagan, Campbell & Kalafat, 1966; Lewis & Spauldi11g, 1967).
Figure 3-1 illustrates directional fractio11ation in a study by Libby,
Lacey, and Lacey (1973). They allowed subjects to look at 30 pictures
for 15 seconds each, without any specific task instruction. The figure
shows pupillary and cardiac responses for pictures rated low, medium, or
high on a factor of Attention-Interest. Directional fractionation is clearly
evident, since pupil size increases while the heart slows. Furthermore,
the amount of fractionation depends on how interesting the pictures are:
the largest dilations and the lowest pulse are obtained for the most interesting pictures.
In more complex tasks, directional fractionation occurs if the subject is allowed to passively observe the stimuli. Lacey, Kagan, Lacey,
and Moss (1963) measllred cardiac responses of subjects in a series of
one-minute tasks. They found deceleration and directional fractionation
in tasks of passive observation, and generalized sympathetic-like responses in problem-solving tasks. Intermediate results were obtained
when both task components were involved.
_
.......
....
II II
Attention
5. 1
- - - - - -.-- - - - --e Lo w
. - - - . - - - . Med
5.0
High
~
~
u.J
Iu.J
4.9
<{
0
....J
a:
=>
CL
4.8
4.7
PUPIL DIAMETER
75
74
..
- I
..., \.
'\
"
., '.'.....
73
,, ,
\.
\
\
u.J
'~-.\ ......
~\
~-.
,
.i
~,J
'e
f~
<{
UJ
:c
72
HEART RATE
71
10
15
Seconds after presentation
FIGURE 3-1
Cardiac and pupillary responses to pictorial stimuli at three attention
values (adapted from Libby, Lacey & Lacey, 1973, with permission).
30
Directional fractionation occurs when the subject is merely exto information, but it is replaced by the common arousal pattern
the subject starts to manipulate the information in a task. Thus,
Schwartz, and Crider (1970) asked subjects to listen to a string
four digits and to subsequently report a transform of this list (Add 1).
... ,..~_.,..,. the subject was listening to the digits, there was a marked cardiac
deceleration accompanied by a rise in skin conductance. Later, the heart
accelerated as the subject prepared and rehearsed his response.
Lacey (1967) proposed that generalized sympathetic dominance occurs when the individual resists stimulation, either because it is aversive
continuous pain) or because it is distracting (e.g., stimulation that
interrupts problem-solving activity). He suggested that directional fractionation with cardiac deceleration occurs in states of attentive acceptance of external stimulation, and that heart rate plays a causal role
a feedback loop which controls attention.
While the existence of directional fractionation is not in doubt,
Lacey's original interpretation of acceleration as a correlate of stimulusrejection was probably incorrect. In the data of Libby, Lacey, and Lacey
(1973), for example, the largest cardiac decelerations were observed for
the most unpleasant stimuli. The evidence supports an alternative formulation, that momentary heart rate reflects the current degree of motor
tension or mobilization for action.
Directional fractionation and cardiac deceleration reliably occur
under conditions of "waiting for something to happen." The two waiting
situations that have been investigated most thoroughly are: (1) the foreperiod between an alerting signal and the stimulus in a reaction-time
experiment (Chase, Graham & Graham, 196'8; Connor & Lang, 1969;
Coquery & Lacey, 1966; Lacey & Lacey, 1964, 1966; Obrist, Webb &
Sutterer, 1969; Obrist, Webb, Sutterer & Howard, 1970b; Webb & Obrist,
1970); and (2) the interval between a neutral conditioned stimulus (CS)
and an aversive unconditioned stimulus (DeS) in a classical conditioning paradigm (Deane, 1961; Hastings & Obrist, 1967; Jenks & Deane,
1963; Notterman, Schoenfeld & Bersh, 195,2; Obrist, 1968; Obrist, Wood
& Perez-Reyes, 1965; Wilson, 1964).
The waiting paradigm has been studied carefully by Obrist, who
found that cardiac deceleration is typically accompanied by a marked
reduction of irrelevant movement, and by the steady fixation of an unblinking eye (Obrist, Webb & Sutterer, 1969; Webb & Obrist, 1970). This
pattern is adaptive: any subject in a reaction-time experiment soon discovers that a high level of motor tension during the foreperiod yields a
slow RT. A relatively relaxed posture, in which ongoing 'activity is inhibited, tends to be optimal. Indeed, Obrist, Webb, Sutterer, and
Howard (1~70b) confirmed the correlation between inhibition of irrele-
32
vant activity and the sllbseqllent RT. They also found that the correlation is maintained even when the cardiac deceleration is. prevented by
atropine, a result which provides decisive evidence against Lacey's suggestion that cardiac deceleration plays a callsal role in the control of
attentional patterns. Rather, the decrease in heart rate is simply a manifestation of a general inhibitio11 pattern. A detailed discussion of the
physiological n1echa11isms controlling the cardiac response has been offered by Gbrist, Webb, Sutterer, and Howard (1970a). Fllrther evidence
for Gbrist's analysis was offered by Cohen and Johnson (1971), who observed highly significant correlatio11s between heart rate and electro-
myographic measures of muscle tension, both within each subject's data
(over successive measurements) and between subjects: the most relaxed
sllbjects had the slowest pulse.
The inhibition of movement during the RT foreperiod has correlates in the measurable activity of the brain. The alerting stimulus of the
RT paradigm is normally followed by a very consistent change in the
EEG, known as the CNV, or continge11t .negative variation (Walter,
Cooper, Aldridge, McCalltlm & Winter, 1964). The CNV, sometimes
called the expectancy wave, is a sllstained change of baseline potential
which is contingent upon the expectation of a subsequent significant
stimulus (Cohen, 1969; Tecce, 1972). The occurrence of a CNV tends to
be associated with a slow heart rate during the foreperiod, and with a
fast RT (Connor & Lang, 1969; Hillyard, 1969).
Elliott (1969) observed directional fractionation in a new and rather
llnexpected situation, and he proposed an interpretation of the cardiac
response which was quite similar to Obrist's view. He studied autonomic
respo11ses i11 the conflict situation induced by the Stroop test, in which
the sllbject lTIUst read the colors in which color names are printed and
refrain from reading the ,vords themselves (see p.109). In this conflict
situation the sllbject may be noticed "reading . . . with almost emphatically deliberate pace, holding himself back from a speed that
might produce c011fusion and error [Elliott, 1969, p. 218]." This inhibitory
pattern is accompanied by a slowing of the heart. Elliott concluded that
cardiac deceleration is associated with the inhibition of responses, and
that cardiac acceleration accompanies the instigation, anticipation, and
initiatio11 of responses. In a fllrther test of this hypothesis, Elliott, Bankart,
and Light (1970) measured heart rate and palmar conductance for the
three conditions of the Stroop test (word, color, and word-color interference) and found that heart rate fell as the difficulty of the test condition increased, while palmar condllctance rose.
111 c011clusion, consideration of the cardiac response and of the nattlre of the task situation pern1its two, and perhaps three states of high
arollsal to be distingllished:
(1)
(2)
(3)
A pattern of motor inhibition. The state of generalized alertness

which is induced by a warning signal (Posner & Boies, 1971)
probably consists of a combination of inhibition and increased
arousal. The inhibition serves to clear the system for an anticipated
stimulus (e.g., the foreperiod situation), or to cope with potentially
disruptive response conflict (e.g., the Stroop test). Elliott's observations with the Stroop test show that an inhibitory tendency is
sufficient to cause cardiac slowing, even while the subject is
verbalizing and showing considerable evidence of motor tension
(for further detail on this point, see p. 109).
A pattern of relaxed acceptance of external stimulation. Whether
this pattern must be distinguished from the first is currently not
clear.
The standard pattern of generalized sympathetic dominance', which
invariably occurs both in situations of physical strain or effort and
in problem-solving. In these situations, a tendency toward verbalization and motor response is sufficient to cause a cardiac acceleration. Thus, a pronounced cardiac acceleration may occur while a
subject attends to an external source of stimulation, if he is engaged in preparing a verbal response (Campos & Johnson, 1966,
1967; Johnson & Campos, 1967).
The evidence of this section disproves the early idea that arousal
can be identified with sympathetic dominance. Subtypes of arousal fl1USt
be distinguished. However, the suggestion that the concept of arousal
should be abandoned appears too extreme. A concept of arousal is
needed to differentiate the state of the subject in a task situation from
his state at rest. While solving a problem, looking at a picture, or reading
a Stroop card, the subject is more active and alert, in short more aroused,
than he is at rest. Arousal can be measured, since there are at least two
indices of autonomic activity, skin conductance and pupil size, which
appear to-increase monotonically with attention in all task situations.
THE YERKES-DoDSON LAW AND THE EFFECTS OF NOISE ON PERFORMANCE
The capacity model introduced in Chapter 2 indicated a mutual

relation between attention and arousal. Variations of attention demands
cause corresponding variations of arousal, but variations of arousal also
affect the pOlicy by which attention is allocated to different activities.
The fundamental law that relates performance to arousal is the YerkesDodson law, which states that the quality of performance on any task
is an inverted U-shaped function of arousal, and that the range over
which performance improves with increasing arousal varies with task
34
complexity (Yerkes & Dodson, 1908). These relations are schematically illustrated in Figure 3-2.
High
A simple task
Q)
o
E
so
"+-
A complex
task
s-
Q)
0..
"+-
....>-.
o
~
Low
High
Low
Arousal level
FIGURE 3-2
The Yerkes-Dodson law.
The Yerkes-Dodson law was initially formulated in the context of

animal discrimination leaI'lling. Yerkes and Dodson discovered that increasing the intensity of a shock administered to mice facilitated the
learning of a brightness discrimination, up to a point. Further increases
of shock intensity caused learning to deteriorate. Yerkes and Dodson also
discovered that the effects of shock were more pronounced in difficult
discriminations, and that the optimum level of shock was higher in easy
discriminations. These concltlsions, initially drawn from a crude experiment with a few dozen mice, appear to be valid in an extraordinarily
wide range of situations (Broadhurst, 1957, 1959; Duffy, 1957; Malmo,
1959; Schlosberg, 1954; Stennett, 1957).
An interestillg application of the Yerkes-Dodson law concerns the
effects of IOlld background noise on human performance (Hockey, 1969,
1970a). Generally, this type of distraction is resisted very well (Broadbent, 1957b, 1958), bllt long periods of exposure to noise do affect per-
in several ways. Continuous noise increases arousal level

1968; Hockey, 1969), and the effects on performance of this
: __,"D~~.:.tJ of arousal depend on the nature of the task, as predicted by the
Yerkes-Dodson law. In general, the presentation of loud background
causes an improvement in the performance of easy tasks and a deterioration when the tasks are more complex (Boggs & Simon, 1968;'
Broadbent, 1954b; Hockey, 1970a; Houston, 1968). This conclusion is
collsistent with the hypotheses represented in Figure 3-2.
The Yerkes-Dodson law provides an elegant explanation for cases
which the effects of concurrent stresses appear to be non-additive. For
example, both loud background noise and lack of sleep are detrimental to
performance in a complex serial-reaction task, but the combination of
the two stresses, lack of sleep and loud noise, is less detrimental than
lack of sleep alone. On the other hand, the detrimental effects of loud
110ise are aggravated by giving the subject full knowledge of results
(KR) concerning the quality of his performance, although KR alone normally improves performance (Wilkinson, 1963). The results are explained
by assuming that both loud noise and knowledge of results increase
arousal, and that together they raise it excessively.
The Yerkes-Dodson law also explains some puzzling differences in
the response to noise stress shown by introverts and extroverts. Although
extroverts are more lively than introverts, research evidence suggests that
they are chronically less aroused (Corcoran, 1965; Eysenck, 1967). Correspondingly, the gradual deterioration of performance in continuous
watch-keeping, called the vigilance decrement, is normally more severe
for extroverts than for introverts (Bakan, Belton & Toth, 1963; Broadbent, 1963). As may be expected from this analysis, extroverts engaged in
a watch-keeping task benefit more from the presentation of noise than
do introverts (Davies & Hockey, 1966; Davies, Hockey & Taylor, .1969).
Presumably, the arousal level of extroverts tends to be suboptimal, and it
is restored by the presentation of lloise.
The results discussed in this section are interpretable within a capacity model. Figure 3-3 includes the elements of that model which are
relevant to the Yerkes-Dodson law. The figure suggests that the detrimental effects of low and high arousal are due to different mechanisms.
The failure of the under-aroused subject is most easily explained
by assuming that the effort exerted in the task is insufficient. Why is this
so? The answer is surely not that arousal cannot increase to meet task
demands: perhaps the most striking conclusion of research on sleep
deprivation is that the sleepless subject, normally under-aroused, can
perform almost allY task at a normal level when highly motivated (Wilkinson, 1962, 1965). Illdeed, sleepless and fatigued individuals who must
11evertheless perform a task show evidence of high physiological arousal
36
MISCELLANEOUS
DETERMI NANTS
AROUSAL
high arousal
disrupts allocation
rl.rvvL
when evaluation
is incorrect
a ro usa I fa i Is to
rise sufficientl y
for good
performance
:AVAILABLE:
i CAPACITY:
I
I
I
MOMENTARY
INTENTIONS
EVALUATION
OF DEMANDS
ON CAPACITY
""",.....------ .......... ,
1/
(
\
,
"
low motivation
"\
or low arousal may
\
prevent adoption
)
of task set
/
--- - -
- ----
",,------- ........
ONGOING ACTIVITIES
/
/'
",.,
RESPONSES
\
\
"-
fatigue
\
and low arousal
\
may interfere with }
the evaluation
/
of performance /
'.........
..........
---."", ,,//
FIGURE 3-3
Effects of high and low arousal on attention and performance.
(Malmo, 1965; Malmo & Surwillo, 1960). Fatigue and sleep deprivation
apparently increase the difficulty of continuous performance, and the
motivated subject compensates for the added difficulty by increasing his
effort.
In general, then, if an initially drowsy person is given a task, he
will wake up and perform. In the terms of an effort model, it may be
more accurate to say that the person will perform and wa~e up, since
it is the demand of the performance that causes the increase of arousal
and of capacity. The occasional failures of this feedback system are best
explained by motivational factors (Broadbent, 1971). The fatigued or
sleepy subject may (1) fail to adopt a task set; or (2) fail to evaluate the
quality of his own performance. Figure 3-3 indicates that, if failures in
a task are not detected, the system will reach equilibrium at low levels
of both arousal and performance. A motivational interpretation explains
both the dramatic effect of knowledge of results (KR) on the performance of sleep-depriv,ed subjects (Wilkinson, 1961, 1963), and the original
Yerkes-Dodson discovery that the speed of discrimination learning depends on the intensity of the stress that motivates learning.
The detrimental effects of over-arousal must be explained in other
terms, however. The allocation of capacity appears to change systematically when arousal is high, and this change causes a decrement in the
performance of certain tasks. The next section reviews the evidence for
this conclusion.
EFFECTS OF AROUSAL ON ALLOCATION POLICY
Easterbrook (1959) presented a theory which was intended to explain both the decrement of task performance with increasing arousal,
and the observation that this decrement occurs sooner in complex tasks
than in simple ones. He proposed that an increase of arousal causes a
restriction of the range of cues that the organism uses in the guidance of
action.
This hypothesis explains the Yerkes-Dodson law as follows: consider a task which requires the simultaneous processing of a certain number of cues. When arousal is low, selectivity is also low, and irrelevant
cues are accepted uncritically. When arousal increases, selectivity increases also, and performance improves because irrelevant cues are more
likely to be rejected. With further increases of arousal, however, the
continuing restriction of the range of usable cues eventually causes relevant cues to be ignored, and performance deteriorates again, in accordance with the Yerkes-Dodson law. With the additional assumption that
the range of necessary cues is narrower for simple than for complex tasks,
this argument implies that the optimal level of arousal should be relatively high in simple tasks. It also implies~ that chronically over-aroused
individuals should perform poorly in complex tasks and relatively better
in simple tasks. There is considerable evidence that both conclusions are
valid.
Easterbrook marshalled much research support for the narrowing
of attention under high arousal. For example, he cited an experiment by
Bahrick, Fitts, and Rankin (1952) in which subjects were engaged in two
tasks: continuous tracking of a target, and monitoring the occurrence of
occasional signals in the visual periphery. When the incentive pay for
both tasks was increased, performance of the central task improved, and
performance of the peripheral task deteriorated. Similar findings were
also described by Bursill (1958), who manipulated arousal by making his
subjects work under conditions of extreme heat and humidity. The balance of attention to central and peripheral tasks was altered in conditions of high arousal.
38
Related results have been reported by Callaway (1959; Callaway &

Stone, 1960; Callaway & Thompson, 1953), who manipulated arousal by
means of drugs. He concluded that atropine, which decreases arousal,
tended to improve the registration of peripheral cues, whereas amphetamine had the opposite effect.
The results of these studies do not imply that peripheral vision is
impaired by drugs, stress, or incentives. What happens in high arousal
is a change in the rules of allocation of attention and effort. Thus, Bursill
(1958) noted that the decrement of peripheral detection in high arousal
did not occur when the peripheral task was emphasized. Cornsweet
(1969) also found that peripheral vision was ullimpaired when the competition between peripheral and central tasks ,vas removed. Indeed, she
found that a peripheral cue anticipating a central signal was used more
effectively when the central signal was associated with shock (high
arousal) than when it was not (low arousal). In a dual task situation similar to Bursill's, Hockey (1970c) observed that the relative preference for
central targets is reduced under sleep deprivation (i.e., low arousal) and
enhanced under noise stress (Hockey, 19.70a). However, he was able to
show that the neglect of peripheral targets under stress is due to the low
probability of detecting such targets, which reduces their importance
(Hockey, 1970b). In the terms introduced earlier, the detection of rare
events in the presence of a primary task depends on the allocation of
spare capacity to perceptual monitoring, which diminishes when effort
and arousal increase.
This research demonstrates that high arousal causes attention to be
concentrated on the dominant aspects of the situation at the expense of
other aspects. As Easterbrook noted, such a change of allocation policy
will disrupt any performance in which attention must be deployed over
a wide range of cues. Complex tasks often require attention to varied
cues, and are therefore performed poorly when arousal is high.
High arousal apparently causes an increased tendency to focus on a
few relevant cues. However, the selection of relevant cues often involves
a discrimination between these cues and others. A state of high arousal
tends to impair such discriminations, with a consequently reduced ability
to focus on the relevant cues. Thus, although subjects spontaneously become more selective when highly aroused, the effectiveness of their
selections is likely to deteriorate, if the selection requires a fine discrimination.
Broadbent (1971, p. 430) has studied selective attention in high
arousal. In one experiment his subjects were shown tachistoscopic exposures of word pairs, with one word in heavy print and the other less
visible. On different trials the subject was required to identify one or the
other of these words, and the duration of exposure was gradually in-
creased until he could do so. The task was performed both in quiet and
tillder intellse continllous noise. The noise had a slight beneficial effect
on the detection of the heavily printed word. Of course, when that word
was at threshold, the Inore finely printed word paired with it was hardly
visible. When the subject was required to identify the less visible word,
however, the word in heavy print must have been clear and obvious. The
"pull" of that stimulus was apparently harder to resist in noise than in
quiet, because the identification of the faint word was significantly impaired by the presence of noise.
Another experiment reported by Broadbent (1971, p. 430) also suggests that the ability to select relevant stimuli is impaired by arousal.
Subjects were briefly shown an array of red and white digits and were
asked to report as many digits of one specified color as they could. Performance in this selective task deteriorated under loud noise. In contrast, noise was associated with a slight performance improvement when
subjects were told to write as many digits as pOSSible, regardless of color.
In contrast to the impairment of effective selection in Broadbent's
experiments, there are situations in which selective attention to relevant
stimuli appears to improve under noise stress. One of these situations is the
rod and frame test, in which a subject in a darkened room is to adjust a
faintly luminous rod to the vertical. The rod is enclosed within a tilted
luminous frame which suggests a false orientation. To determine the true
vertical, the subject must ignore this visual cue and rely on kinesthetic
sensations. Performance in this task improves in the presence of loud
background noise (Oltman, 1964).
Another task in. which performance improves in noise is the .Stroop
test, which will be discussed in more detail in C~apter 6. In the most
difficult condition of this test, subjects are shown a card in which names
of colors are printed in inks of different colors. They are required to report the color of the ink in which each word is printed, suppressing the
tendency to read the word itself. Performance in this condition improves
in loud noise (Agn.ew & Agllew, 1963; C'allaway & Stone, 196;0; Houston,
1969; Houston & Jones, 1967). What improves in noise is specifically the
ability to control interference. Thus, there is no interference when subjects merely name the colors of neutral symbols, and performance in this
easier condition is actually worse in noise than in quiet (Houston, 1969;
Houston & Jones, 1967).
Although these results appear to support Easterbrook's hypothesis
that high arousal enhances selectivity, Houston and Jones (1967) found
reasons to doubt this interpretation. Noting that drug stimulants and
noise do not produce identical effects on the Stroop test (Callaway &
Stone, 1960; Quarton & Talland, 1962), they suggested that it is the struggle to inhibit irrelevant responses to the noise which enhances the sub-
40
ject's ability to inhibit the irrelevant Stroop responses. This attractive

hypothesis is consistent with physiological evidence of an inhibitory set
in the Stroop test (see p. 32). The same hypothesis would also explain
the improvement of performance on the rod and frame test under noise.
The methodological implication is clear: a change of behavior can be
explained as a consequence of arousal only if it occurs in several conditions of high arousal, e.g., both in noise and after the ingestion of stimulant drugs. These converging operations have not been used in all the
situations to which Easterbrook's hypothesis has been applied.
This note of caution notwithstanding, the weight of the evidence
does favor the conclusion that high arousal restricts the range of cues
among which attention may be divided, and also disrupts the control of
selective attention. In the terms of a capacity model, the allocation of capacity becomes both more uneven and less precise when arousal is high.
Consequently, performance is impa~red in tasks that require either the
deployment of attention over a broad range of information-processing
activities, or the control of selection by fine discriminations.
Easterbrook (1959) has attempted to account for all the effects subslllned llllderthe Yerkes-Dodson law by the single hypothesis that cue
utilization is narrowed by increasing arousal. However, this theory appears inadequate on several grounds. First, it implies the unlikely idea
that the difficulties of the under-aroused, drowsy subject result from an
excessive openness to experience. Second, it suggests that concentration
is highest when arousal is high. This is contrary to everyday observation,
which indicates that a state of high arousal is associated with high distractibility.
The apparent paradox that rigidity and lability of attention both
rise with arousal has often been noted (Callaway & Stone, 1960). To resolve this parado.x a distinction must be drawn between the breadth of
allocation of attention at anyone time and the stability of allocation
over time. Wachtel (1967) emphasized this distinction. He quoted Hernandez-Peon's (1964) description of attention as a "beam of light in which
the central brilliant part represents the foclls, surrounded by a less intense fringe," and noted that such a beam has two characteristics that
could define breadth of attention: its width, and the extent to which it
roams in scanning the field of stimulation. These two characteristics are
conceptually independent, and a proper definition of breadth of attention must explicitly separate the width of the beam from its stability.
Wachtel defines beam-width in terms of the number and range of cues
that are integrated in a judgment or percept. One example of narroW
beam-width is the consistent failure of children in tests of conservation
of number or quantity, which Piaget has attribllted to excessive concentration on one salient attribllte at a time.
The evidence presented earlier regarding Easterbrook's hypothesis

is consistent with the idea that l1igh arousal narrows the attentional
beam. The evidence concerning the effects of arousal on scanning is far
less conclusive, but clinical observation suggests that extremely high
arousal may lead to an increase in scanning, with a corresponding increase in distractibility (Korchin, 1964) and a consequent disorganization of behavior. It is perhaps relevant that a very high saccadic rate is
observed after the ingestion of LSD-25 (Kohn & Bryden, 1965). If it is
true that the allocation of attention becomes both narrower and more
labile under high arousal, the disruption of complex performance is
inevitable.
Variations of arousal have other effects on performance. For example, Broadbent (1971) has discussed the effects of arousal in a prolonged vigilance task, where subjects are to detect occasional signals and
report their confidence in each detection. Under the effect of noise,
fewer responses are made at an intermediate level of confidence: subjects tend to be either very sure or very unsure of their detections.
Arousal also affects the speed-accuracy tradeoff, i.e., the balance
which subjects spontaneously adopt between speed and the avoidance
of errors. In reviewing the literature on effects of noise, Broadbent
(1957b) noticed several studies which suggested that work tends to be
faster but less accurate under noise.
Posner, Klein, Summers, and Buggie (1973) have reported a detailed analysis of speed-accuracy tradeoff in choice reaction-time as a
function of the duration of the foreperiod. It is well known that RT is
reduced if the stimulus is preceded by a warning signal, but the facilitating effect of the warning signal requires a foreperiod of 0.5 seconds
to develop fully. Posner and Boies (1971) had interpreted this effect as
a preparatory rise in alertness. Subsequently, Posner, Klein, Summers,
and Buggie (1973) showed that the effect of the foreperiod in reducing
RT is sometimes associated with an increase in the number of errors.
Apparently, the surge of arousal which is caused by the warlling signal
does not simply improve the overall effectiveness with which the task is
performed, but it alters some aspects of the subject's strategy in dealing
with the task.
Some aspects of learning also show systematic changes with arousal.
Several investigators have reported that items which elicit a large galvanic skin"response (GSR) at presentation are retained better than other
items after a long retention interval, but worse than other items if the
retention test is immediate (Kleinsmith & Kaplan, 1963, 1964; Walker &
Tarte, 1963; Corteen, 1969). Similar results have been obtained by presenting bursts of noise during the learning period: immediate recall was
often impaired, but forgetting was slower (Berlyne, Borsa, Craw, Gelman
42
& Mandell, 1965; Berlyne, Borsa, Hamacher & Koenig, 1966; Berlyne &
Carey, 1968). Thus, short-term memory appears to be impaired by high
arousal (Easterbrook, 1959), while long-term memory improves. Further
evidence for the dependence of short-term memory on arousal was obtained in studies of the diurnal rhythm. Physiological arousal level is
known to increase gradually during the day (Kleitman, 1963), and performance in most tasks shows a corresponding improvement. Immediate
memory, however, shows a significant decrease between morning and
afternoon (Baddeley, Hatter, Scott & Snashall, 1970; Blake, 1967). The
interpretation of these results has usually been in terms of a direct effect
of arousal level on the consolidation of memory traces. An alternative hypothesis is that subjects engage in more active rehearsal when highly
aroused, and that the effects of such rehearsal are beneficial for longterm retention and detrimental for short-term recall.
In summary, the evidence reviewed in these sections suggests that
a state of high arousal is associated with the following effects: (1) narrowing of attention; (2) increased lability of attention; (3) difficulties in
controlling attention by fine discriminations; and (4) systematic changes
of strategy in various tasks. On the other hand, a state of extremely low
arousal may cause: (1) a failure to adopt a task set; (2) a failure in the
evaluation of one's performance, resulting in an insufficient adjustment
of the investment of capacity to the demands of the task (see Fig. 3-3).
THE ORIENTATION REACTION
The model of attention which has been developed in these chapters

assumes that the allocation of capacity is determined principally by two
sets of factors: the momentary task intentions of voluntary attention and
the more enduring dispositions which control involuntary attention. These
enduring dispositions cause us to pay more attention to some stimuli
than to others. Novel stimuli, in particular, are favored in the allocation
of capacity.
The pattern of physiological responses which is elicited by novel
stimuli is variously named the orientation reaction, response, or reHex
(OR). It was first discovered and described by Russian physiologists
(Pavlov, 1927; Sokolov, 1963, 1965), and the experimental results available up to 1965 originated almost entirely in Russian laboratories (Lynn,
1966). Subsequent studies in the West have usually confirmed the conclusions of the Russian investigators.
The OR and states of high arousal, such as pain or fear, share several components: EEG desynchronization (alpha blocking) and manifestations of sympathetic dominance, including the galvanic skin respollse
(GSR) and the dilation of the pupil. However, Sokolov has distinguished
the orientation reaction to novel stimuli from the defensive reaction to
aversive and painful stimuli. The arousal pattern is commonly identified
with the defensive reaction. The most important difference between
orientation and defense is that the OR is characterized by vasoconstriction
in the limbs and vasodilation in the head, while the defensive reaction includes generalized vasoconstriction. Sokolov considered this dissociation of the vascular response so important that he tended to use it
as an operational definition of the OCCllrrence of an orientation reaction.
This usage h~s not generally been adopted in the West, where several
experimenters have confirmed that peripheral vasoconstriction follows
novel stimuli (e.g., Unger, 1964; Zimny & Miller, 1966), while others have
failed (Cohen & Johnson, 1971; Keefe & Johnson, 1970; Raskin, Kotses &
Bever, 1969a, b). Western investigators tend to use the GSR (e.g., Germana, 1968; Maltzman & Raskin, 1965) or a transient desynchronization
of the EEG (Berlyne & Borsa, 1968) as measures of the OR. Unfortunately, these measures do not distinguish the specific OR pattern from
related states, such as emotional arousal and mental effort.
The OR precedes and dominates other responses to the same stimulus (Sokolov, 1963; Zimny & Kienstra, 1967; Zimny & Miller, 1966). This
is shown most dramatically by the reaction to sudden immersion of the
hand in hot water. Although the adaptive reaction to this stimulus is a
peripheral vasodilation which facilitates heat loss, the initial response to
immersion in hot water is a typical OR, complete with peripheral vasoconstriction. Only after a few seconds does the adaptive response predominate. The OR similarly dominates early responses to aversive stimuli,
such as electric shock, which later elicit the defensive pattern of
vasoconstriction in both head and limbs. After a few repetitions the OR
diminishes in both extent and duration, until it eventually vanishes completely and only those adaptive or defensive reactions remain that are
appropriate to the stimulus.
Habituation with repetition is the most important characteristic of
the OR. For example, when a subject is instructed to "listen to tones,"
the first and perhaps the second tones elicit very substantial GSR's, but
a low steady state is reached with a few presentations (Uno & Grings,
1965). The habituation of the OR does not imply that the stimulus is no
longer registered or analyzed. Rather, the subject has learned to expect
the stimulus, and the OR is only released when the characteristics of
the stimulus violate expectations. Sokolov (1969) has provided a compelling demonstration of this expectation effect: when a single flash of
light is omitted from a regular series, a major OR occurs soon after the
time at which the omitted light was due. Similar results have been described by Badia and Defran (19'70), among others.
44
Sokolov has presented a "neuronal model theory" to account for

such results. According to his theory, an incoming sensory message
reaches analyzers at the cortex which match its features to neuronal
models constructed by previolls experience. A mismatch between stimulus and model triggers an orientation reaction, which is controlled by
subcortical centers. An interesting feature of this theory is the idea that
stimuli are analyzed at the corti~al level before the decision is made to
activate the system by an OR: the cortex appears to be the only structure capable of performing the precise analyses that determine if a stimulus is familiar or novel. However, the elicitation of the OR also influences
the subsequent activity of the cortex itself. The processing of a. novel
stimulus is therefore recursive: the output of a preliminary analysis at
the cortex is eventually fed back to control subsequent cortical activity.
Figure 3-4 explains in terms of a recursive process the detailed and
intense study of a novel stimulus which is one of the salient manifestations of the OR. The figure illustrates the two types of input, information
and effort, which affect perceptual processing in a capacity model. In
this example, the information from preliminary analyses of a novel stimulus causes the allocation of greater effort to elaborate the analysis of
that stimulus. Although several stages are indicated, the entire cycle can
probably be completed within 150-200 milliseconds. Chapter 4 will show
that the decision to fixate on a particular area in visual ~earch involves
similar considerations, and such. decisions are made several times every
second.
Sokolov's concept of neuronal models raises interesting problems:
what are the characteristics of such models, and how are they constructed? In general, of course, the neuronal model is set by repetition
of the same stimulus, but it has been shown that repetition is not essential. A definite expectation, whatever its source, sets up a neuronal
Effort
Processing of
Information - -...
novel
stimulus
Improved
perceptual
analysis
FIGURE 3-4
Recursiveness in the processing of a novel stimulus.
and the violation of an expectation elicits an OR. Thus, Unger

presented a series of numbers in ascending order, and he found
orderly disappearance of the OR to successive numbers, and an obOR to a number presented out of sequence (e.g., 11, 12, 13, 17).
results have been reported by Maltzman and Raskin (1965), and
conclusively by Zimny, Pawlick, and Saur (1969).
In some situations, however, the neuronal model and the subject's
COI1scious expectancies probably do not coincide. Thus, Maltzman, Harris,
and Wolff (1971) exposed subjects to a constant level of illumifor ten minutes. The illumination was then changed, then rein a series of regular alternations. Although the subjects surely
realized the repetitive pattern of events, they continued to show larger
OR's when the illumination was changed from the initial adapting level
when it was restored to that level. The initial adaptation period
apparently established a neuronal model which defined some changes
illllmination as deviations from the standard, and others as restorations of standard conditions.
Furedy and Scull (1971) exposed subjects to a random sequence of
possible events. They noted that repetition of the same event caused
a smaller OR than alternation. This result is particularly interesting because the verbal expectations of subjects in such situations usually show
a I1egative recency effect, favoring alternation over repetition. The neuronal model apparently represents a more primitive type of "expectation" than is normally verbalized. In a very different context, Epstein
and Rock (1960) had also observed a marked positive recency effect
\vhich overshadowed cognitive expectancies. In a series of trials, they
alternated presentation of two profiles. On a test trial, two profiles were
combined into an ambiguous figure, and the subjects were asked what
they "saw." There was a marked tendency to see the profile that had
been shown on the preceding trial rather than the profile that was consciously "expected."
Neuronal model theory describes a system which monitors the fit
of events to some type of expectation and triggers an OR whenever the
fit is poor. In this view the neuronal model is an automatic novelty detector, and novelty is a sufficient condition for an OR. This is not, however, strictly true: novelty is only one important contributing factor to
the elicitation of an OR, but it is neither a sufficient nor a strictly necessary condition.
The significance of the stimulus to the organism is a second major
determinant of the OR. Razran (1961, p. 118) describes an experiment
Biryukov in which fox cubs were exposed to the squeaks of mice. The
OR's to these squeaks soon extinguished. If the cubs were allowed to
eat the mice, however, a single meal sufficed to make the OR essentially
,.c---._-..
"
II:1..1.,&,--
III1LL ....... " ' ......
46
permanent. The continued effectiveness of a signal stimulus, one's own

name, for example, also demollstrates the role of significance. Such a
stimulus, although hardly novel, is a potent elicitor of the OR.
Bernstein (19'69) has stressed the marked dependence of the OR on
stimulus intensity as another indication that novelty cannot be the sole
determinant of the OR. He suggested a two-stage model in which the
novelty and the potential significance of a stimulus are both evaluated
before an OR is released. Bernstein et ale (1971) further supported this
view by showing that apparent motion of a pattern toward the subject
(looming) elicits a larger OR than apparent motion into the distance.
Pavlov (1927) viewed the autonomic changes of the OR as part
of a more general pattern, which he called the "what is it?" reflex. This
pattern includes' various adjustments which facilitate sensory registration, notably a marked increase in the rate of eye movements. However,
the functional significance of other manifestations of the OR has not
been established.
Sokolov (1963) emphasized the idea that a major function of the
OR is to improve sensory receptivity, and he claimed that thresholds
are lowered during the OR, both by peripheral adjustments, such as
pupil dilation, and by central sensitivity changes. Although plausible,
this idea should not be accepted uncritically. The argument that a large
pupil enhances sensitivity is doubtful, for example, because gains in
sensitivity to light are probably offset by a loss in the quality of the
retinal image. The observation that weak or ambiguous stimuli elicit
a large OR (Sokolov, 1963, 1965) may well represent the subject's effort
to process these stimuli rather than any enhancement of peripheral sensitivity. When subjects are instrllcted to make discriminations about stimllli, large pupillary responses occur on the presentation of weak (Hakerem
& Sutton, 1966) or ambiguous stimuli (Kahneman & Beatty, 1967), because the analysis of such stimuli demands much effort. Similar responses
probably occur when the organism spontaneously engages in detailed
processing of an alerting stimulus. In this interpretation, a weak stimulus
does not directly elicit a large autonomic response. It elicits a surge of
effort, which is accompanied by autonomic manifestations of arousal.
It has sometimes been suggested that the facilitation of learning
is one of the functions of the OR. Indeed, the first phase of classical
conditioning is normally the, development of a marked OR to the conditioned stimulus. Conversely, when repeated presentation has thoroughl)T
habituated the OR to a stimulus, that stimulus becomes ineffective, and
it is very difficult to attach any response to it (Sokolov, 1963, p. 244).
Whell one no 1011ger pays attention to the occurrence of an event, it is
difficult to learn anything new abollt it.
The OR cannot be likened to a stereotyped reflex because the corre-
lations among its various manifestations are not very high. It is better
viewed as a set of independently controlled changes, which usually
OCCllr together becallse they are often adaptive on the satre occasions.
Figure 3-5 presents such a view of the OR.
MISCELLANEOUS
DETERMI NANTS
MISCELLANEOUS
........---. MANIFESTATIONS
OF AROUSAL
AROUSAL
,I
~
ENDURING
DISPOSITIONS
1 ) Enhance processing
of novel stimulus
2) Inhibit ongoing
activi ty
3) Orient and prepare
lAVAILABLE:
: CAPACITY:I
I
4) Increase arousal
to meet current
or anticipated
demands
ALLOCATION
POLICY
/ I \
r-----/ /
//
~--_
D1J~~
"-
~=D
POSSIBLE ACTIVITIES
EVALUATION
OF DEMANDS
ON
CAPACITY
RESPONSES
FIGURE 3-5
Components of the orientation reaction.
The figure distinguishes four components of the OR.

(1) A transient effort to process and analyze the alerting stimulus.
The effort component of the OR will be most significant when the alerting stimulus is novel, complex, or barely discernible. The effort required
to process a stimulus is probably much greater on the first presentation
of a stimulus than on subsequent occurrences: with the construction of
a neuronal model, processing effort habituates. Conversely, if rapid
habituation of the OR is observed, as in the repeated presentation of a
tone, we can infer that effort was a dominant element in the originaIOR's.
The present analysis of effort in the OR applies to cases in which
the allocation of attention to the stimulus is involuntary, governed by
enduring dispositions. It seems important to distinguish these cases from
48
others, in which the sllbject is instructed to make decisions and execute

specific responses to stimuli. Autonomic responses in these situations of
voluntary effort should not be identified as OR's. Much confusion has
been caused by inconsistent usage in this context.
(2) Inhibition of ongoing activity. This is a. very salient feature of

the response to novel stimuli in normal situations. The inhibitory aspect
of the OR is often obscured in the laboratory situation, which is typically
rigged to prevent or minimize spontaneous activity at the time of stimulus presentation. The presence of a pronounced deceleration component
in the cardiac OR, which consists of a succession of accelerations and
decelerations (Chase, Graham & Graham 1968; Connor & Lang, 1969;
Graham & Clifton, 1966), probably reflects this inhibitory aspect. The
inhibition of irrelevant activity facilitates a rapid and effective response
to the novel stimulus and to other significant stimuli that may follow it.
(3) An orientation toward probable sources of future significant information. The increase in ocular motion, and the pricking of the ears
in cats and dogs are components of the OR, which are obviously related
to an enhanced readiness for relevant stimulation. These adjustments are
not random or diffuse. An alerting sound from a particular direction
reliably elicits postural changes to facilitate the registration of additional
stimuli from the same direction. There is also an immediate increase in
the readiness to respond in the direction of an eliciting stimulus. Simon
(1969; Craft & Simon, 1970; Simon, Craft & Small, 1970) has studied the
effects of sounds delivered from various directions on the performance of
speeded directional responses. This work demonstrated "a potent natural
tendency to react toward the major source of stimulation [Simon,Craft
& Small, 1970, p. 63]." Unfortunately, experimental arrangements often
minimize the postural aspect of the OR: stimuli are typically presented
in the frontal plane, or through headphones. However, any comprehensive description of the response to unexpected stimuli must emphasize
the !postural adjustments that serve orientation.
(4) A transient increase of arousal. It has been emphasized that
the cardiac and vascular components of the OR differentiate it from
other cases of heightened arousal. These specific aspects of the OR are
probably related to its inhibitory and preparatory functions, but the OR
also represents an increase in arousal level. This increase of arousal does
not appear to be controlled by feedback from ongoing activity. Rather, it
seems that when the system can anticipate a particular activity, it allocates capacity to that activity in advance of relevant stimulation. An
anticipatory mobilization of capacity may well occur in other situations,
such as the foreperiod in a reaction-time task. However, the evidence
of the foreperiod effect suggests that the anticipatory surge of arousal

cannot be effectively sustained ove'r a long period.
In summary, the OR should be viewed as a loosely organized set
of physiological changes, each independently controlled by some aspect
of the stimulus situation and of the response to that situation. The OR
consists of an effort to analyze the alerting stimulus, and of a complex
pattern of preparation for future stimuli and responses.
REVIEW
The chapter was concerned with several connections between

arousal and attention,1t
The first section showed that the nature of the task situation determines the pattern of autonomic responses. At least two states of intermediate or high arousal must be distinguished. The standard arousal
pattern is associated with active processing and with the performance of
motor responses. An inhibitory pattern is adopted in passive acceptance
of stimuli, in states of waiting, and in states of response conflict. This
pattern is characterized by directional fractionation: the heart slows
down even as other indices suggest an increase of arousal level.
The Yerkes-Dodson law states that performance is an inverse
U-shaped function of arousal level. The failures of under-aroused subjects were interpreted in motivational terms. The failures of over-aroused
subjects were interpreted in terms of Easterbrook's hypothesis: in high
arousal, attention tends to be concentrated on the dominant and most
obvious aspects of the situation. In addition, high arousal impairs the
ability to discriminate relevant from irrelevant aspects, and increases
the lability of the allocation policy. These changes cause a decrement of
performance in high arousal, which is most obvious in tasks that require
a wide range of cues or fine discriminations.
An enduring disposition causes a specific allocation policy to be
adopted when novel and significant stimuli are detected. Some aspects
of this policy are manifest in the orientation response (OR), which consists of an inhibition of ongoing activity, intense processing of the novel
stimulus, and various preparations for future stimuli and responses.
4
Looking
The allocation of attention has both instantaneous and sequential aspects.

At any point in time, attention can be divided among several activities.
In addition, the focus of attention changes. from instant to instant, in an
organized fashion. The act of looking provides a basic example of this
sequential organization of selective attention. The world extends 360
degrees around us, our field of vision spans about 210 degrees, but vision
is sharp only within a small foveal region of about 2 degrees, and the
rate at which this narrow beam of sharp vision can be moved is limited
to about 3-5/ second. The question of where to direct this beam is obviously of great adaptive significance, and the mechanisms that have
evolved for the control of eye position and eye movements are of exquisite precision.
A description of the physiology of oculomotor control is beyond
the scope of this book (the interested reader should consult a review by
Alpern, 1971). It is sufficient to note that two distinct mechanisms control two major types of eye movements. Most eye movements are saccades, very rapid movements which are planned in advance and are
executed withollt continuous control during the movement itself. This
so-called ballistic character of the saccade permits it to be executed at
50
Looking 51
high speed. The other type of eye movement is pursuit, a smooth motion
which occurs only when the eye fixates a moving object. Smooth motions
of the eye do not occur in the absence of a moving object in the field,
and the eye cannot be moved slowly from one locus to another, except
in pursuit.
Looking is obviously under voluntary control, because one can decide where to fixate, but conscious and deliberate control of fixation is
actually infrequent. As with other highly skilled components of voluntary performance, such as walking or the maintenance of balance, looking is controlled by a general intention, and consciousness plays a minor
role in the execution of the intended sequence of fixations. The processes
that determine the locus of individual fixations are psychologically silent,
and their feedback is so poor that people do not usually know precisely
where they are looking.
The precise measurement of eye movements and eye position by
photography or electrophysiological measures is a laborious procedure
which requires sophisticated equipment. However, because people are
unaware of the precise locus of their fixation at any time, a simple alternative technique can be used to provide crude estim-ates of fixation
tendencies. Kaufman and Richards (1969) have recently reintroduced
this technique, which was known in the nineteenth century. The equipment consists of a slide or movie projector, fitted with a blue filter and
with a polarizing filter which can be made to rotate at slow speed. A
non-depolarizing projection screen is also required. The observer sees
whatever image is projected on the. screen. In addition, he sees a small
fuzzy line whirling on the screen whenever the polarizing filter rotates.
The line is actually the shadow cast on the observer's fovea by crystalline structures in his eye (called the Haidinger brush). This shadow is
normally invisible because it is stabilized on the retina, but the rotation
of the polarizing filter causes the shadow to disappear and reappear
intermittently, and this makes it visible to the observer. Naive observers,
however, are invariably convinced that the whirling shape on the screen
has been projected by the experimenter, and they can be asked occasionally to indicate the position of the shadow on a map, thus providing a
record of where they are looking at the time. Kaufman and Richards
(196-9) have documented the fact that subjects are often ignorant of the
true locus of their fixation. Moreover, they can be exposed to the Haidinger brush repeatedly before discovering its relationship to their eye.
Looking behavior is never random. When one's activities require
the intake of visual information, the movements of the eye adjust to that
function. In the absence of a specific task set, the control of fixation is
handled by enduring dispositions and standard routines of "spontaneous
looking." These routines, many of which are probably innate, tend to
52
select stimuli that are ecologically likely to be significant. The enduring

dispositions that control spontaneous looking include all the dispositions
that call for spontaneous attention in the orientation reaction, as well as
additional factors which affect looking without affecting autonomi,c
activity. Finally, looking is closely involved in cognitive activities that
have little or nothing to do with visual intake: highly consistent patterns of eye movements accompany various types of mental activity, and
various tasks of selective attention with auditory stimuli.
The determinants and manifestations of spontaneous visual attention are discussed in the next section. Sllbsequent sections deal with the
deliberate intake of visual information, and with situations in which the
content and direction of mental activity are the main determinants of
looking behavior.
SPONTANEOUS LOOKING
In the absence of a specific instruction to search for visual information, spontaneous looking is controlled by enduring dispositions that determine which parts of the field of view should attract and hold the gaze.
Berlyne (1960, 1966) has distinguished two classes of stimuli that attract
spontaneous attention: physical properties, such as the presence of many
contours, and collative properties, such as novelty, complexity, or significance.
The distinction between physical and collative properties is not
sharp, however, because collative properties, in their most elementary
form, can be reduced to physical properties. Novelty and complexity are
important collative properties that control spontaneous attention in the
adult. The inf.ant is already very responsive to the most elementary level
of complexity-an isolated figure in a blank field-and to the most elementary level of novelty-movement (Gesell & Ilg, 1949). Human infants
who are given a choice of two patterns in their visual field (Fantz, 1958)
show an immediate preference for relatively complex stimuli: patterned
stimuli are preferred to homogeneous gray patches (Fantz, 1965a; .Hershenson, Munsinger & Kessen, 1965), and within the first few months
there is a gradual development of preferences for complex random arrays
over simpler displays (Fantz, 1967), for radial over linear patterns, and
for solid over two-dimensional figures (Fantz, 1965b).
Some of the collative factors that attract the spontaneous attention
of human adults have similar effects on the behavior of lower animals.
The effect of novelty on viewing time, for example, is not restricted to
humans: Berkson (1965) has shown a very rapid decrease of viewing
time with repeated exposures in infant chimpanzees. Complexity also
Looking 53
controls attention similarly in humans and monkeys: Brown and Gregory

(1968) found that the number of sides in a visual pattern (a measure of
complexity) affects the visual attention of adult humans and determines
how easily a visual discrimination can be learned by squirrel monkeys.
Novelty and incongruity are defined by a mismatch between stimulation and a neuronal model of expectations. For example, Berlyne
(1957, 1958; Berlyne & McDonnell, 1965) showed that the attention of
adult subjects is attracted by incongruous pictures such as a camel with
a lion's head. The neuronal models of the adult, of course, are vastly
more elaborate than those of the infant, who would surely not respond
to such manipulations of incongruity. However, the rule that novel and
complex stimuli attract spontaneous attention is valid at all stages of
development. Moreover, the adult's ability to develop highly sophisticated patterns of expectations merely supplements the innate rules of
spontaneous attention without replaci.ng them. Adults retain an extremely
powerful tendency to direct their gaze toward moving objects and to
scan contours, and they tend to fixate an isolated object in the field even
when it carries no information.
Adult subjects also respond very consistently to trivial manipulations of Visual complexity such as the number of sides of a shape or the
variability of their length (Brown & Gregory, 1968). An important observation here is that the function which relates spontaneous attention
to complexity has the shape of an inverted U. Excessively complex stimuli are treated as irrelevant noise and no longer attract attention.
Spontaneous attention can be measured in several ways, and the
same stimulus properties invariably dominate results. Novel, complex,
and incongruous objects are always fixated in preference to others (e.g.,
Berlyne, 1958; Day, 1965), and subjects also spend more time lookillg at
such stimuli when given control of a device which presents pictures successively (Berlyne, 1957; Berlyne & Lawrence, 1964). The same collative
properties also determine which of several concurrent stimuli will control behavior in a situation of conflict. Berlyne has introduced an experimental paradigm to study this type of stimulus choice (Berlyne, 1950,
1951, 1967, 1970; Berlyne & Lawrence, 1964; McDonnell, 1967, 1970):
the subject fixates a mark located at an equal distance from several windows. Under each window there is a response button, which the subject
presses whenever a picture appears. When several pictures are shown
simultaneously, the subject is to press only one of the buttons, and the
aim, of course, is to discover which he will push. A plausible assumption
is that one presses the button corresponding to the most attractive or
compelling stimulus. Results show that the same collative and physical
variables that determine the choice of fixation and the duration of free
viewi11g time also control immediate and undeliberate choice in a con-
54
flict situation (Berlyne, 1966). The next chapter will show that related
factors also determine which area in the visual field spontaneously
emerges as figure over the background (see Fig. 5-3 on p. 77).
A stimulus which is novel, complex, or incongruous certainly demands greater processing effort than a stimulus distinguished by none
of these properties. Thus, a basic rule of the allocation policy appears
to be that perceptual activities which demand much capacity are favored
over less demanding activities. This rule already controls the looking behavior of infants, and it remains valid in adults.
Although pleasure-seeking is often seen as a basic principle of
behavior, spontaneous looking does not seem to conform to this principle. The best evidence has been obtained in the free-viewing paradigm,
in which subjects are given control of the time they spend viewing a
series of abstract pictures. The behavior of subjects who are given no
specific instructions tends to be similar to the behavior of subjects instructed to linger on "interesting" stimuli, and quite different from that
of subjects who follow a "pleasingness" set. Three-sided shapes, for instance, are judged more pleasant than nine-sided shapes, but they are
looked at less, in the absence of special instructions (Brown & Farha,
1966). Berlyne and Lawrence (1964) and Day (1966) also found a negative correlation between free viewing time and verbal preference for
irregularity of shape. Observations of this kind suggest that the enduring
dispositions which control spontaneous attention reflect epistemic motivation, the need to perceive clearly and to reduce uncertainty.(Berlyne,
1960, 1965; Durham, Nunnally & Lemond, 1971; Nunnally, Faw & Bashford, 1969; Woodworth,1958).
It must be mentioned, however, that spontaneous looking is not
always controlled by epistemic motivation. The widespread use of female
beauty in advertising does not appeal to collative variables. More generally, the needs and values of individuals determine what they find interesting, and what they prefer to look at. Extroverts, for example, prefer
to look at a picture of a party than at a picture of a lone man reading
a book (Bakan & Leckart, 1966). And a subject whose Rorschach responses classify as a "repressor" may studiously avoid a bare-breasted
woman in a picture, concentrating instead on a man reading a newspaper (Luborsky, Blinder & Schimek, 1965).
Yarbus (1967) has studied patterns of fixations during prolonged
observation of pictures. Figure 4-1 presents two of his pictures and typical records of fixation sequences. A striking feature of these records is
their repetitive nature: in the left panel, fixations repeatedly travel back
and forth between the girl's eyes, and in the right panel they repeatedly
climb the trees. The right panel also illustr'1tes the roles of both phYSical
and collative properties in. the control of spontaneous looking: the selec-
Looking
55
FIGURE 4-1
Records of eye movements during the continuous observation of two pictures (from
Eye Movements & Vision by A. L. Yarbus, with permission).
tion of the trees that are often fixated is controlled by the physical
property of brightness contrast, but the large numbers of fixations on
the human figure must be attributed to collative factors.
56
THE ACQUISITION OF VISUAL INFORMATION
Lookil1gat simple figures is governed by simple rules, of which

several are innate. The rules of fixation guide the eye to areas which are
ecologically likely to be most informative. Thus, Salapatek and Kessen
(1966) have shown that infants typically follow contours in scanning a
triangle; contollrs, of course, define the shape of objects. Brightness contrast is also significant, and for sound ecological reasons: a sharp contrast between adjacent areas of the two-dimensional scene is likely to
represent a demarcation between two objects which are located at different distances from the observer. These rules govern infants' fixations
and are retained in the adult. For example, Kaufman and Richards
(1969) found that the fixations of adlllts hover near a vertical contour
separating a field into a dark and a lighter area, with a bias in favor
of the lighter area. With figures that span more than 5 degrees, the eye
tends to stay near the center of the figure.
The innate routines which control the infant's fixations are the
precursors of the more complex operations which direct the eye of an
older person to the most informative areas of a scene. In scanning a picture of a family scene, for example, a sophisticated observer will look
at the furniture to answer a question abollt the family's financial circumstances and he will look at the children when attempting to guess how
long the parents have beel1 married. This type of information search has
been documented by Yarblls (1967).
Mackworth and Bruner (1970) studied the eye movements of both
children and adults who were attempting to recognize an object in a
blurred picture. The area of the picture was divided into 64 squares,
and each square was rateCl for informativeness by independent judges.
The most informative areas attracted more fixations (Mackworth &
Bruner, 1970; Mackworth & Morandi, 1967), although the correlation was
far from perfect. Mackworth and Morandi (1967) reported the important
finding that il1formative areas are identified very early in the observation
of pictllres; the average informativeness of fixated areas is already high
in the first two seconds of observation. Similar conclllsions have been
reported in other studies of eye movements (De Groot, 1966; Jongman,
1968; Yarbus, 1967).
Although an adlllt can very quickly decide where to look next, this
decision often depends on a highly sophisticated weighting of many
factors. Thus, the control of the eye is one of man's most accomplished
skills.
A fixation is often determined on the basis of informatiol1 previ-
Looking
57
ously acquired in peripheral vision. Williams (1966) studied this process

in an elegant experiment: his subjects were shown a large array of
figures, each including a number. The figures varied in color, size, or
shape. On each trial the subject was told to look for a particular number,
and he was given some information about the figure containing the target number, e.g., its color or size. Subjects found it easy to restrict their
fixations to figures of the designated color, thus red,ucing search time.
They could also use the size cue to some extent, but they were essentially unable to control search by shape cues.
An instructive example of the skilled control of eye movements was
reported by Gould and Schaffer (1965). Their subjects were shown an
array of digits and were required to count the occurrences of a particular
target digit. Subjects most easily detected the digit 7 and found 9 and 2
the most difficult. They did not always fixate on targets that they recognized, but their fixations tended to come closer to the difficult 2's and 9's
than to the easier 7's. In addition, the difficult targets were more often
fixated directly. Evidently, subjects often decided to have a closer look
at the more difficult targets. An impressive aspect of decisions like these
is their speed: the decision must depend on an evaluation of the information acquired during the current fixation, yet the whole cycle is
normally complete within 1/3 second.
Sanders (1963, 1970) studied the conditions under which a subject
moves his eyes or head in order to obtain adequate information from a
display. In some of his experiments the subject's task was to decide
whether two simultaneously presented digits were identical. Sanders
distinguished three ranges in the functional visual field: (1) The stationary field, in which the task can be carried out without scanning eye
movements, i.e., when all relevant information is presented to central
vision. (2) The eye field, in which the visual information acquired in a
single glance only suffices to construct a hypothesis concerning . the nature
of the objects presented to peripheral vision. With a display that lies
within the eye field, the subject has the two options of directing a fixation to the object originally seen in the periphery, or of acting' on an
unverified hypothesiS. This decision to move the eyes or not is responsive
to instructions, payoffs, and the 'nature of the display. (3) The head field,
in which a head movement is usually needed to collect the necessary
information. Again, the decisions appear to be made very rapidly and
with little deliberation.
When a decision to redirect search must be made deliberately, it is
slower and less effective. Neisser and Stoper (196'5) trained subjects to
perform a task of visual search over successive lines of printed material,
with occasional marks indicating to the subject that he could safely skip
a given number of lines. These marks were often ignored, apparently
58
because the deliberate control of search required more time and effort
than it could save.
A particularly interesting type of looking decision controls fixations in a monitoring task with multiple targets. Thus, an airline pilot
faces more dials than he can see at a glance, and he must distribute his
looking to maximize the probability that any significant event will be detected soon after it occurs. The probability that a dial contains new information is a function of at least two factors: the overall rate at which
information is conveyed on that dial, and the elapsed time since the
last look at it. Senders (1965) has used the mathematics of information
theory to derive an optimal policy that determines the sequential allocation of looks among dials. He found that trained observers closely approximate optimal looking behavior. Evidently, the system which allocates
looks is able to consider the changing probabilities that each of many
dials will merit a look.
Senders (personal communication) has di~cussed the example of an
adult reading a newspaper beside a swimming pool, while a baby is randomly cra'Yling about in the area. At what intervals will the adult look
up to check the baby's positiol1? The interval will depend on where the
baby was when last seen, al1d on its direction and rate of progress at
the time. It will also depend on the depth of the pool, and on whether
or not it is filled with water.
The duration of fixations appears to be less responsive to momentary
fluctuations of attention than is their location. Although it is possible for
an observer to maintain steady fixation for several seconds, the rate of
fixations is normally much faster, and the fixations are usually rhythmic.
Thus, a conclusion of the study of reading is that the difficulty of reading
material has a greater effect on the number of fixations per line than
on the duration of individllal fixations, which averages 200-225 milli~
seconds (Morton, 1964; Tinker, 1947, 1958). In the observation of pictllres, the common duration of fixation is 300-350 milliseconds (Mackworth
& Bruner, 1970; Yarbus, 1967). Rhythmic motion occurs during uninstructed observation (see Figs. 4-1 and 4-2). Of course, ocular motion is
l1ecessary to prevent loss of vision from retinal stabilization (Riggs, Ratliff, Cornsweet & Cornsweet, 1953), but the spontaneous rate appears to
be considerably faster than is necessary to avoid fading.
Under some cOl1ditions, however, the duration of an individual
fixation may correspond to the attention that the object of fixation reqllires. An interesting pattern of results was reported in a series of studies
by COllld. In the initial experiment, subjects scanned a display of numerals for occurrences of a target numeral (Could & Schaffer, 1965).
Although targets were fixated more often than non-targets, the duration
of fixations on targets and on non-targets was the same. Different results
Looking 59
were obtained in more complex tasks of pattern-matching (Gould, 1967;

Gould & Dill, -1969; Gould & Peeples, 19'70). Subjects in these studies
were shown a nonsense pattern in the center of the display (the standard)
surrounded by other nonsense patterns in the periphery. They were required to count as fast as possible the number of peripheral forms identical to the central standard. The initial study of the standard pattern
was typically done in one prolonged fixation. During the search, fixations
on target items were longer by about 80 milliseconds than fixations on
non-targets. F'urthermore, the duration of a fixation on a non-target pattern varied with the similarity between that pattern and the standard.
In general, the same factors that deter,mined the probability that a parti~ular pattern would be fixated at all, also determined the duration of
fixation on jt, and the probability that it would be refixated. The difficulty of discriminating a pattern from the standard had a significant
effect on all three measures. For fixation duration to reflect attentional
demands, an extremely rapid decision process is required. Since fixations
as short as 220-250 milliseconds were sometimes observed (typically on
easily discriminable non-targets) it appears that the computation of
whether a longer look was required must have been completed within
about 150 milliseconds from the instant of fixation.
While prolonged fixations reflected visual attention in Gould's
studies, they can also indicate inattention to the visual channel. Thus,
Furst (1971) noted a progressive reduction of saccadic rate during the
observation of a single picture, as well as an increasing stereotypy of
fixation sequences and an increasingly steady rhythm. This pattern of
habituation to a picture was reversed when the picture was repeated
after an interval, in~icating spontaneous recovery of visual attention.
A marked prolongation of fixations with continued exposure was
found in subjects exposed to a Rorschach card (Thomas, 1963) and in
radiologists studYIng an X-ray plate (Thomas & Lansdown, 1963). It is
not altogether clear whether this effect was due to increased visual difficulties, or to inattention.
In general, the studies reviewed in this section indicate a fair correspondence between what the eye does and the demands of a set to
search or recognize. There is direct evidence, however, that the linkage
between fixation and attention is optional rather than obligatory. As
Helmholtz already knew, one can look at one object and attend to another, and such attending can alter visual perception. Thus, Fraisse,
Ehrlich, and Vurpillot (1956) demonstrated that the apparent size of
objects to which one pays attention increases even when the attended
objects are not fixated directly, and Grindley and Townsend (1968) have
shown that deliberate attention to a peripheral area increases acuity in
that area. The figure-ground effect, which will be discussed in the next
60
chapter, is another illustration of an attentional effect that is not controlled by fixation, althollgh we usually fixate the figure rather than the
background.
It will be useful to summarize the information presented in this
section and the preceding one, concerning the factors which control
looking.
When fixation is governed by a visual task, the locus of fixation is
determined by an assessment of the probabilities that relevant information will be acquired, and that the acquired information will be useful.
Many factors can contribute to the assessment of these probabilities:
, known base rates for a particular area, and the elapsed time since the
area was scanned (Senders, 1965); the detection of some features of a
possible target, such as its color (Williams, 1966); a preliminary identification of a target (Gould, 1965; Sanders, 196;3); and general knowledge
about the strllcture of the situation (Yarbus, 1967).
When fixation is not controlled by a specific task set, it reverts to
the control of enduring dispositions. Looking and attention are then
spontaneous rather than deliberate. Two sets of enduring dispositions
may be distinguished: innate routines which are triggered by specific
physical features, and more elaborate responses which are mostly triggered by the mismatch between a stimulation and a neuronal model of
expectations, i.e., by collative features.
The more elementary dispositions are usually overridden by a task
set. However, a sudden change of the visual scene will usually elicit
both an orientation reaction and a fixation toward the locus of change,
even when one is engaged in a task. Thus, enduring dispositions and
task set can override one another.
EYE MOVEMENTS AND THE SPATIAL ORIENTATION OF THOUGHT
Let the reader attempt to think of an object in the room, and he

will soon become aware of a tendency to look at that object. When one
person in a group conversation mentions the name of one of the people
present, the collective gaze of the group is immediately drawn to the
person mentioned. There seems to be a strong tendency to look where
one thinks.
Eye movements of this kind represent a general orientation toward
the object of thought. They occur even when the resulting visual stimulation is not useful. In experiments by Kahneman and Lass (1971), subjects were shown an array of four schematic line drawings of objects
(automobile, person, tree, and airplane) and were asked questions such
as "What makes of automobiles can you remember?" The Haidinger
Looking
61
brush technique (Kaufman & Richards, 1969; see p. 51) was used to
detect fixation tendencies. The eye quite regularly fixated the task-related
object, although. that object could provide no useful information. The
hierarchy of the processes that control looking was clearly evident. When
a single object was shown in the field, subjects almost invariably looked
at it, regardless of whether it was relevant or not. When more than one
object was shown, the relevant object was fixated. Finally, when the
subject was questioned about a picture that was no longer present, he
usually fixated that area of the blank screen where the relevant information had been shown.
Are these orientations helpful to performance of the task? In an
attempt to find out, Kahneman and Lass compared subjects' performance
when a relevant object was shown, when an irrelevant object was shown,
and when the screen was blank. No significant effects were found. For
example, subjects produced as many words with a specified letter in the
third position when that letter was shown on the screen and when an
irrelevant letter was projected. The irrelevant letter was fixated, but it
did not interfere. In this study, at least, the preference for fixation on
the relevant letter appeared to serve no purpose. In general, however,
the correspondence of orientation to thought is adaptive, because it ensures that relevant information will be quickly acquired.
An impressive demonstration. of the association between eye movements and internal processing was provided by Bryden (1961) and by
Crovitz and Daves (1962), who showed that the locus of greatest accuracy
in tachistoscopic recognition is related to the direction of eye movements
following the stimulus exposure: for example, when the eye moves to the
right, the far right figures are likely to be reported accurately. This is
true although the eye movement occurs after the exposure and cannot
affect sensory registration in any way.
A series of studies of eye movements during paired-associate learning provides further evidence of the correspondence between the locus of
fixation and the focus of internal processing. The studies also provide a
demonstration of the use of eye movements to test theories about internal
events. A prevalent theory of paired-associate learning suggests that its
first phase is response consolidation (Underwood, Runquist & Schultz,
1959; Underwood & Schultz, 1960). Correspondingly, during early phases
of learning, the eye typically fixates the stimulus, then the response (S-R),
and it lingers on the response item (McCormack & Haltrecht, 1966; McCormack, Haltrecht & Hannah, 1966, 1967). The tendency to fixate the
response item is further enhanced when that item is low in meaningfulness
(McCormack & Hannah, 1967; McCormack & Moore, 1969). The second
stage of paired-associate learning has been identified as a phase of stimulus-response association. In that stage, the typical sequence of fixations
62
is S-R-S, and the eye spends more time fixating the stimulus than the
response. The patterns of fixatioll observed in these ,experiments certainly represent processing effort rather than information intake, because
the exposure time (typically two seconds) is more than sufficient to
acquire two visual nonsense syllables.
There has been much interest in the eye movements of chess players.
De Groot (1966) and Jongman (1968) described the fixation patterns of
master players, who were allowed to study a complex chess situation for
five seconds in order to later reproduce it from memory. Many of the
players immediately perceived the best possible moves for both opponents. Indeed, the master's eye quickly finds the area of main tension
of the game, and the first fixation after the presentation of the display is
already highly selective. Generally, however, the correspondence between the verbalizations of a chess player and his recorded looking behavior is far from perfect, and the master player often fails to fixate a
piece about which he is much concerned. In the initial study of a position, possible sequences of moves are clearly perceived, but rarely mirrored by eye movements. When a player is allowed a longer period to
study the board, the correspondence between fixations and the moves
considered apparently improves after the first 10 or 15 seconds (Simon &
Barenfeld, 1969; Tikhomirov & Poznyanskaya, 1966). In general, these
investigations suggest that the movements of the "mind's eye" are correlated with those of the physical eye, but also that the correlation is
optional rather than obligatory.
Another observation which demonstrates the optional nature of
the correspondence between eye and thought was obtained by Kaplan
alld Schoenfeld (1966). They showed subjects five-letter anagram problems which could all be solved by the same transposition of the order
of the five letters. Those subjects who discovered the rule usually announced their response after fixating each of the letters exactly once, in
the order of their position in the solution. But other subjects were able
to solve the anagrams withollt discovering the rllle, and their sequence
of fixations did not correspond to the solution.
Gopher (1971) studied the patterns of eye movements accompanying different tasks of auditory attention. The auditory messages were
presented by earphones in his experiments, and eye movements could
serve no function of sensory acquisition. Neve~theless, highly consistent
patterns were observed. Eye movements were markedly inhibited when
sllbjects listened to a monallral message (i.e., a message presented to a
single ear), or when they were exposed to dichotic messages (i.e., different messages to the two ears) and were told to focus attention on one
and ignore the other. When focusing attention, subjects almost always
make a large saccade at the beginning of the message, invariably in the
Looking
63
direction of the relevant ear and they maintain their fixation in that
direction during the entire message. Dividing attention elicits a different
pattern of eye movements. Gopher studied a task in which the subject is
instructed to listen to both messages and to repeat target words that can
be presented to either ear. When the target words are distinguished by
a physical property (e.g., a word spoken by a male voice inserted in a
message spoken by a female voice), subjects primarily fixate ahead, although they make an eye movement whenever a critical word is heard.
When the critical word is defined by a semantic property (e.g., an animal
name) the rate of eye movements doubles, and rhythmic alternations of
small saccades become very freqllent.
Gopher (1971) found that deliberate fixation to right or left can
alter performance in a task of divided auditory attention. His subjects
were asked to monitor dichotic messages for the occurrence of semantically defined target words. Occasionally, two target words were presented simultaneously. As will be shown in Chapter 8, subjects often
detect only one member of such a simultaneous pair, and right-handed
subjects most often respond only to the word presented to their right
ear. This pattern was even more pronounced when subjects were instructed to fixate 20 degrees to the right of center. When they were
instructed to fixate left of center, the imbalance between the ears vanished. The result could reflect either a shift in the spatial focus of attention or a temporary alteration of the normal pattern of cerebral dominance (Kinsbourne, 1970, 1972).
The relation between auditory attention and the direction of the
gaze is an important source of cues in situations of social interaction
(Argyle & Dean, 1965; Exline, 1963, 1971; Kendon, 1967; Strongman,
1970). People are extremely sensitive to eye-to-eye contact and show unusually high acuity in judging whether someone else is gazing directly
at them (Gibson & Pick, 1963). The listener in a conversation tends to
gaze directly at the speaker, and this gaze, which conveys continued
interest, provides support for the speaker. The listener normally averts
his gaze when he prepares to speak, probably indicating that he is turning attention to the preparation of his own message. The listener's averted
gaze is often accepted as a tacit instruction for the speaker to fall silent.
The direction of the gaze aversion which accompanies the onset
of active thought is highly consistent for different individuals and for
different classes of problems. About half the population initially look
to the right when they begin to think about a verbal problem, and the
other half look to the left (Day, 1964, 1967a, b). Kinsbourne (1972) noted
that movements to the left are relatively more frequent when subjects
solve spatial problems than when they solve verbal problems, and he
also,showed Significant differences between right-handed and left-handed
,/
64
subjects. Both observations suggest that the lateral movement may itldicate a temporary preponderance of activity in one or the other hemisphere of the brain. Other investigators have studied various correlates
of the preferred direction of the lateral eye movement. Thus, Bakan
(1971) reported that left-movers are more hypnotizable (the correlation
was 0.44), and he confirmed Day's observation that the EEG of leftmovers has a prominent alpha component. Bakan and Shotland (1969)
also showed that right-movers read significantly faster than left-movers
and are less prone to interference on the Stroop test. Perhaps most surprising, a significant negative correlation was fOtlnd between the eye
movement tendencies of spouses (Day, 1967a); right-movers tend to
marry left-movers!
Turning inward to think is also associated with a dramatic increase
in the rate of eye movements (Lorens & Darrow, 1962), which contrasts
with the inhibition of eye movements during attentive listening (Gopher,
1971). Antrobus, Antrobus, and Singer (1964) noted that both active
thinking and deliberate attempts to suppress a conscious wish or fantasy are associated with a very high saccadic rate. Relaxed, passive, or
wish-fulfilling thought leads to reduced motility. The frequency of blinks
follows similar rules. These results suggest that changes of fixation and
blinks punctuate changes of mental content, a conclusion which is also
consistent with the observation th3:t LSD causes a very high saccadic
rate (Kohn & Bryden, 1965). It is notable that mental work increases
ocular motility even in congenitally blind subjects (Amadeo & Gomez,
1966).
In summary, the involvement of eye movements in mental processes
attests to the linkage between the eye and the focus of attention. Thoughts
often "move" over a representation of space, and the position of the eyes
tends to reflect the current direction of attention. Eye position also
serves to label the direction of sensory attention, even in the absence of
visual input, and this pattern of selective orientation may affect the
allocation of auditory attention. Finally, the rate of eye movements often
corresponds to the rate of thinking, even in the absence of any spatial
component.
REVIEW
The three sections of this chapter were devoted to three types of

eye movements, which were distinguished by the situations in which
they occur: spontaneous looking in the ahsence of a specific task set;
looking that serves to acquire task-relevant information; and looking that
accompanies internal processing events.
Looking
65
Spontaneous looking is controlled by collative features of stimuli,

such as novelty, complexity, and incongruity. The antecedents of these
enduring dispositions are found in innate dispositions to orient toward
contours and toward moving objects. The enduring dispositions that control spontaneous looking serve the function of information-seeking, rather
than the function of pleasure-seeking.
Task-relevant looking was viewed as an allocation problem. Because the area of sharp vision is narrow, it must be directed to those
portions of the field which are likely to be richest in relevant information. The decisions often require a sophisticated weighting of many
factors, and they are made quickly, for the eye changes positions 3-5
times a second. The sequential allocation of glances is a highly skilled
performance. The system generally makes decisions about the locus of
individual fixations rather than about their duration, which is often
quite stable. In complex visual discriminations, however, the duration
of individual fixations may vary, within rather narrow limits, according
to the demands of the task.
Finally, eye movements are a salient manifestation of the changing
orientations which occur whenever the focus of thought refers to a direction in space. This orientation occurs even when it cannot possibly aid
in the acquisition of new information. Movements of the eye also accompany, and perhaps influence, the balance of activity between the cerebral
hemispheres, and the rate of eye movements often corresponds to the
rate of mental activity.
5
Attention and Perception
The preceding chapter was concerned with overt orientations that reflect the allocation of attention. We now turn to the study of central
mechanisms of selection and allocation. The present treatment identifies
the allocation of attention to perceived objects with the figural process,
which selects certain areas of the field as figure and relegates others to
the background.
The first section outlines a model of some stages of perceptual
analysis, and the second introduces a taxonomy of attention tasks. The
remainder of the chapter is concerned with the explication of the processes of unit formation, figural emphasis, recognition, and perceptual
interpretation, and with the relation of these processes to selective
attention.
STAGES OF PERCEPTUAL ANALYSIS
The analysis of attention in this and in subsequent chapters will

assume the model of perception illustr<;lted in Figure 5-1. The figure
traces the vicissitudes of a pattern of stimulation to which an observer
66
SENSORY
REGISTRATION
AND STORAGE
ALLOCATION
POLICY
(Attention and effort)
Perceptual
Readiness
UNIT FORMATION
Preliminary
segregation of
perceived objects
FIGURAL EMPHASIS
Attention is allocated
to objects
ACTIVATION OF
RECOGNI TION UNITS
Possible
interpretations
for some objects
SELECTION OF
INTERPRETATIONS
Perceptual
in te rp re ta t ions
are se Iec ted
RESPONSE SELECTION
FIGURE 5-1
Schematic model for perception and attention.
68
is exposed, from an initial stage of sensory registration and temporary

storage in sensory memory, through a final stage at which a response
may be selected. The model assumes that an early stage of Unit Formation partitions the stimulus field into segments, or groups. The Gestalt
laws of grouping describe the operation of this stage on visual stimuli.
Similar rules of grouping operate in audition: for example, successive
sounds that originate in the same place are more likely to be grouped
as a unit than sounds from different places. These rules produce perceptual units that have a high probability of corresponding to distinct
objects in the scene: a robot programmed to apply Gestalt laws of grouping to a photograph will usually segregate real objects. Units have both
spatial and temporal aspects: grouping over space yields perceived ob- .
jects; grouping over time yields perceived events.
Attention enters at the next stage, where some of the units isolated
earlier receive greater Figural Emphasis than others. The decision made
at this stage involves choosing the size of the relevant unit, and selecting
the unit or units of that size which should be emphasized. Thus, the
page, the line, the word, or the individual letter may be the relevant
units, among which we select that word or that letter to which most
attention will be paid.
The amount of attention that is allocated to a perceived object or
event at this stage affects subsequent processing in several ways. Attended events are more likely to be perceived consciously, and more
likely to be perceived in detail. They have a higher probability of eliciting and controlling responses, and they are more likely to be stored in
permanent memory in a manner that permits intentional retrieval.
The next stage of processing is the activation of Recognition Units.
These hypothetical structures are activated by the occurrence of a stimulus that possesses certain critical features. The activation of a recognition unit is a matter of degree. Activation is highest for a stimulus which
has all the critical features, is presented at high intensity, and is attended.
Inattention, degraded presentation, and a mismatch between the features
of the stimulus and those of the recognition unit cause activation to
decrease.
The graded output of recognition units is fed to a stage which
selects Perceptual Interpretations for some of the perceived objects or
events. Recognition units and interpretations are organized in dimensions and sets. The stage of selection of interpretations guarantees that
no more than one interpretation is assigned to each object in each set or
dimension. Thus, a homogeneous patch of color is not seen as both red
and yellow, nor is it seen as both a square and a circle. A perceived object
is normally assigned values on the dimensions of size, color, distance,
Attention and Perception 69
direction and velocity of motion, and so forth. In addition, it may be

assigned a meaning. Thus, the full perceptual interpretation of an object
or event consists of a bundle of partial interpretations.
Selection of an interpretation is required because stimulation is
normally ambiguous. Any stimulus event probably activates several recognition units in each set or dimension, although. to different degrees.
In addition, there are different degrees of Perceptual Readiness to make
each of the possible interpretations at any instant in time. The interpretation which is selected is that for which the sum of readiness and activation is highest.
It is useful to assume a threshold below which no interpretation
is made. Thus, a stimulus may fail to be fully interpreted if it was faint
or did not activate any recognition for which tllere was sufficient readiness. Interpretations serve as input for subsequent stages of processing,
including storage in permanent memory and the selection and control of
responses. An uninterpreted event will have little or no effect on these
stages.
The last stage shown in Figure 5-1 is that of Response Selection. In
many experimental studies of attention, one of the multiple interpretations that is attached to an attended object controls the choice of a
response. The subject in such experiments is usually constrained to make
a response of a particular class, e.g., name a digit, identify a word, or
evaluate the length of a line. These instructions induce a state of Response Readiness, making the appropriate responses more easily available. In addition, there may be differences in the degree of readiness for
p,ossible responses within each set.
The model shown in Figure 5-1 is not intended as a complete
"model of the mind." It does not refer explicitly to various storage systems, and it does not deal with th.e initiation and control of covert and
overt responses. It only distinguishes a few stages and operations which
are essential to a treatment of selective attention in perception.
The allocation of attention affects, events at two stages in the sequence of the information-processing chain in Figure 5-1. At the stage of
figural selection, paying attention to some perceived objects in preference
to others facilitates the activation of recognition units. At the stage of
response selection, effort and attention are allocated to some responses
in preference to others.
Two recursive paths are indicated in the model. The path leading
from the stage of Activation of Recognition Units to the Unit Formation
stage indicates that tentative recognitions can affect the segmentation of
objects ~f perception. Another important path leads from the Activation
of Recognition Units to the Allocation Policy and eventually back to
70
affect Figural Emphasis. This recursive path /was already mentioned in

the context of the orientation reaction. It appears to play an important
role in search tasks that will be discussed later in this chapter.
Conscious perception can be identified with the selection of interpretations. This stage is sometimes bypassed in the control of action.
For example, there is suggestive evidence that the latency of conscious
perception is about the same as the latency of overt responses in a simple
reaction-time task (Kahnenlan, 1968). If this is the case, then simple responses cannot be dependent on prior conscious perception (Fehrer &
Raab, 1962). The possibility of bypassing the stage of conscious perception is indicated in Figure 5-1 by the arrow leading directly from the
Activation of Recognition Units to Response Selection.
TAXONOMY OF SEL.ECTlVE ATTENTION
According to the model shown in Figure 5-1, attention is allocated

at two stages: figural emphasis and response selection. The two possibilities are related to a distinction drawn by Broadbent (1970, 1971) between stimulus set and resp'onse set. Stimulus set defines the relevant
stimuli by a physical characteristic, which permits these stimuli to be
analyzed in more detail than other stimuli. Response set restricts the
vocabulary of possible responses. When a subject is instructed to read
words printed in r~d and ignore other words, he adopts a stimulus set.
When instructed to read digits and ignore other words, he adopts a
response set.
A more elaborate classification scheme for attention tasks was proposed by Treisman (1969), who distinguished four types of selection: of
inputs, targets, analyzers (or attributes), and outputs. Table 5-1 illustrates
this scheme by examples of four tasks that a subject may be asked to
perform, given a particular stimulus array.
(1)
(2)
Selection of inputs. The relevant and irrelevant stimuli are discriminated by an obvious physical characteristic, allowing the
subject to adopt a stimulus set. Broadbent (1958, 1971) calls this
type of early selection filtering. An auditory example of inputselection task could be: "Listen to the message that comes from
the left; ignore other messages." According to the model introduced in the preceding section, the selection of inputs is mediated
by the allocation of attention to the relevant inputs at the stage of
figural emphasis
Selection of targets. Here the subject is instructed to search for a
designated target. The distinction between selection of inputs and
selection of targets is' that the relevant items are rare and relatively
71
diffi~ult to find in the latter task. However, the mechanism of selection appears to be similar in the two cases.
(3)
Selection of outputs of perceptual analysis. In the example of

Table 5-1, the numerals are not distinguished from other words by
any obvious physical characteristics. Consequently, the relevant
items can only be selected after they are interpreted in perception.
In Broadbent's terms, this task involves a response set, since the
relevant items are defined by a common category of responses
rather than by a shared physical attribute.
(4)
Selection of attributes. In the example of Table 5-1, the relevant

attribute is letter-type. This task involves response set, since the
vocabulary of allowable responses is sharply limited. In the model
of Figure 5-1, the task is performed by allocating attention to one
of the responses elicited by each item (describing type) in preference to other responses (e.g., reading the word).
Discriminations are made at several stages of perceptual processing:

pre-attentive discriminations control unit formation and figural emphasis
(Neisser, 1967). Additional discriminations, achieved at the level of perceptual interpretations, guide the selection of responses. Most tasks involve discriminations at both levels (von Wright, 1970). F'or an example,
consider task 1 in Table 5-1. One must first find the capitalized words,
then read them. Figural selection (finding) is guided by a discrimination
of letter size, while response selection is controlled by discriminations of
letter shape.
For another example, consider the two questions "What is the botton1 word in the array of Table 5-1?" and "Where is the word five?" Both
TABLE
5-1
A classification of attention tasks.
Stimulus
Task
Labels
cat EIGHT TABLE seven

TWO dog BAT chair
BOOK egg time PLANT
soon fish PIANO
door FOUR
five
(1) Read the capitalized words.

(2) If the word "egg"
is in the array, say
it aloud.
(3) Describe the type
in which each word
is printed.
(4) Read the digits.
Selection of inputs; filtering; stimulus set.

Selection of targets or
tests.
Selection of analyzers;
attention to attributes.
Selection of outputs;
response set.
72
questions eventually refer to the same object, the word-five-that-isprinted-at-the-bottom, but the sequence of operations that lead to this
object are different in the two tasks.
The reader will probably agree that it is easier"to find the word at
the bottom and read it than to find the "five" and report its location. The
sequence of operations on attributes is important because the attributes
which allow the most effective control of figural emphasis are not the
same as the attributes to which responses are most easily attached. The
example illustrates the general rule that it is easy to direct attention by
the attribute of location, and easy to control the final response by the
attribute of shape. It is also easy to control visual attention by the
attribute of color (Uleman & Reeves, 1971; von Wright, 1970; Williams,
1966).
UNIT FORMATION
Some examples of unit formation in vision are shown in Figllre 5-2.

The most compelling of these examples is (A), which is interpreted unambiguously as an object over a background. Note that a grouping
process is required to segregate the object as a single unit. Note also
the hierarchy of the grouping organization: the object in panel A is a
unit, which is included in the larger unit of the panel, which is included
in turn within the larger unit of the entire figure, and so on. Within
panel A, the background may be considered as a group, but each of the
small objects within it provides another natural unit.
Other panels of Figure 5-2 illustrate various determinants of grouping, which differ in their effectiveness. In panel B, most observers see
rows, rather than columns, by an effect of similarity. In panel C, columns
are seen, because proximity overcomes the effect of similarity. Panels D
and E show that the discriminability of elements determines the quality
of grouping. Grouping is distinctly "better" and more definite in panel D
than in panel E.
Finally, compare panels F, G, and H. In these panels, the elements
in three segments of the circle share a feature that could distinguish
them from the elements in three other segments. A rather clear organization emerges in panel F, but less in panels G and H. The variable of
shape in panel G and the variable of letter orientation in panel H do not
suffice to integrate the three similar segments into a single form. The
grollpings that spontaneously emerge in these examples are more restricted, and even an intelltional effort to "see" the larger pattern generally fails.
There are other differences among attributes in the degree to which
(- A )
( B)
,-:"1-') )'
r( "
) .,})
.,') J
~ <.
l :J ) ) ~ \ ) "-(.
Jl1~()
\., (
) J '- "'\ '\ J
(C )
( D)
TTTTTTTO
TTTTTTTT
TT
TTOOOO
TTTTTOOO
TTTTTTOO
( E)
T T TTL
TTTTTLLL
TTTTTTLL
T T
T T
T TTL
T T
T T T
T T T
(G)
FIGURE
5-2
Determinants of grouping.
73
74
they allow for similarity grollping: similarity of shape and similarity of

slope, for example, are less effective than similarity of colo~ or brightness. The properties that provide strong units also allow for the effective
control of attention, because attention is most easily directed toward a
natural perceptual unit. Thus, Williams (1966) and von Wright (1968,
1970) noted informally that effective search is possible within the visual
field only when all potential targets share a physical characteristic (e.g.,
color) which permits them to be segregated into a figural group.
An experimental demonstration of the relation between grouping
and search was offered by Beck (1972). He showed subjects an array consisting of a majority of elements of one kind, and of scattered elements
of another kind. Some subjects were asked to count the minority elements. Other subjects rated the ease with which the minority elements'
were segregated as a perceptual group. As expected, it was easy to count
elements that made up "good" groups. In agreement with earlier work
(Beck, 1966, 1967; Olson & Attneave, 1970), a difference in overall slope
(e.g., tilted T's in a field of upright T's) provided a better basis for grouping than differences in line arrangement with constant slope (e.g., L's in
a field of T's). Correspondingly, the tilted T's were also easier to find and
count. The relation between grouping and counting is strong: it is
easier to count the O's in panel D of Figure 5-2 than the L's in panel E.
Beck (1972; Beck & Ambler, 1972) proposed that grouping is often
based on the detection of differences between elements in peripheral
vision, prior to a focusing of attention. Furthermore, he suggested that
pre-attentive and attentive discriminations follow different rules. Thus,
sensitivity to differences in overall slope is relatively greater in preattentive discriminations, or when attention is divided among many objects than when attention is focused. A tilted T is more discriminable
from an upright T than is all L, but only when several background stimuli are shown. When a single form is shown in peripheral vision, the
tilted T and the L are equally discriminable from an upright T. Further,
when an array of letters is shown briefly, then masked, the tilted T is
more discriminable than the L when the masking stimulus quickly
erases the array, but not when the subject is given more time to redirect
his attention (Beck & Ambler, 1972).
Beck's work suggests that the grouping process is controlled primarily by the detection of similarities and differences among the elements simultaneously present in the field. His analysis helps explain the
observation noted earlier, that different attributes are most effective in
controlling attention and in controlling responses. The discrepancy
could be related to the relative ease with which one makes simultaneous
or sllccessive discriminations. Attributes that allow for easy simult~neous
discriminations will be effective in controlling attention, because simul-

taneo~s
75
discriminations are involved in unit formation and in figural

emphasis. Attributes that permit accurate successive discriminations
should be most effective in the selection of responses.
The effects discovered by Beck provide strong support for Neisser's
(1967) suggestion that pre-attentive mechanisms carry out the task of
sorting and organizing the field prior to the operation of focal attention.
These discriminations are not guided by the perceiver's intentions. It is
not known whether they vary with available capacity. Pre-attentive discriminations refer to obvious physical features, but it is inappropriate to
assume, as some authors have done, that discriminations of physical features are always pre-attentive, and that only higher-order properties are
analyzed attentively (Ellis & Chase, 1971). Beck's work indicates that
discriminations of physical features occur at both the pre-attentive and
attentive levels, but follow different rules at the two levels.
Neisser (1967) suggested that pre-attentive discriminations are relatively crude. Indeed, the strongest g-rouping effects are controlled by
proximity or by similarity of obvious physical features. There are indications, however, that perceived units sometimes depend on more complex
analyses of the stimulus, even including semantic decoding. Speech, for
example, is perceived as consisting of discrete words, although a physical
analysis of the sounds often reveals no pause between the end of one
word and the beginning of the next. Fixate above the central 0 in
THEDOGATE and you may see figural areas of different size, but most
often the meaningful unit "dog" (Osgood, 1953). Finally, although the
THE of OFTHEOX -makes a poorer perceptual unit than the XXX in
--XXX--, it is easier to segregate than the THE in BATHERE, because
of the competition of coding responses in the latter case. In the model
of Figure 5-1, these effects are represented by the arrow leading from
the recognition units to the grouping stage.
Treisman (1970) has reported an auditory experiment in which a
. 11hysical cue (ear of origin) becomes effective in segregating a perceptual
unit only when it interacts with existing language habits. She presented
pairs of computer-synchronized auditory items, either binaurally (both
sounds to both ears) or dichotically (one sound to each ear), and her
subjects often responded with a mixture of phonemes from the two stimuli,
even in dichotic presentation (e.g., the response TEV to the stimuli TAV
and SEM). Surprisingly, the frequency of these confusions was about
equal in dichotic and in binaural presentation when the two stimuli were
both nonsense syllables. However, dichotic presentation did reduce confusions when a nonsense syllable was presented to one ear and a digit to
the other. The very precise synchronization of dichotic inputs apparently
provides a powerful stimulus for fusion, which can only be overcome
when one of the separate inputs activates a recognition unit.
76
So far, this discussion of unit formation has reflected the historical

emphasis on spatial grouping in vision. However, grouping processes
operate in other sense modalities, and grouping occurs over time as' well
as in space. Temporal grouping isolates events, rather than objects.
Michotte (1963) and Heider and Simmel (1944) have described some
compelling examples of such grouping. Imagine a scene with two white
squares separated by a gap. The left square starts to move to the right;
it reaches the right square and stops; within 100 milliseconds, the right
square starts to move to the right. This sequence is almost invariably
perceived as a single event, in which the left square "hits" the right
square and causes it to move. Audition, of course, is a temporal sense
and auditory grouping is largely temporal grouping. The musical or the
verbal phrases function as perceptual units. Some factors of grouping>
such as proximity and similarity, are common to the formation .of spatial
units in vision and of temporal units in audition: sounds tend to be
grouped if they originate from the same location, or if they share certain
physical characteristics (Broadbent, 1971, C'h.ap. 4).
The basic identity of grouping processes over space and time- has
not always been recognized, probably because it is difficult to speak of
events and of objects in the same terms. As a consequence, a harmful
distinction has been introduced between closely related variants of
attention. Thus, visual attention is often described as the selection of
stimuli or objects, while auditory attention is commonly described as the
selection of a "channel." Treisman (1969) has attempted to overcome
this difficulty by using the neutral term "selection of inputs." The terms
of the present treatment, units and figures, suggest visual images-but
the concepts are more abstract, and they can be applied alike to different
modalities, and to units over time and space.
FIGURAL EMPHASIS
The spontaneous division of the field into figure and ground is a

basic fact of perceptual experience. 'It is also a prototype of a purely
central process of selection which does not depend on orienting movements of the head or eyes. When part of a flat picture "stands out" in
perception, it is seen as, figure over its background. The subjective experience of attention is often described in the same terms: the attended
object "stands out."
Panels A-D of Figure 5-3 illustrate the familiar distinction between
figure and ground. In all panels there is a clear organization of the field
into segments or groups, of which one predominates and is seen as
figure. The main manifestation of the figural character of an area is that
( B)
( A)
~
r~
) (J
>'
\ - ,,)
r ,""\ I
' . ) J'\
) ) . . ~ ) "'l
f
\ 1''- (,) ')') I

,..,
J
J
'-
'J
( D)
(C )
( E)
A
A
A
A
A
( F)
A
A
xxx XXX XX
A
A
XXOXXXXX
XXXXXXXX
A A A A A A A A A
A A A A A A A
A A
A
A
A
A A
A
A
( G )
ABCDEFGJ
KLPQRIUV
YABCDEFG
FIGURE 5-3
Determinants of figural emphasis.
77
78
its bounding contours are perceived as belonging to it rather than to the

background. In addition, there is a depth effect: the figure appears closer
than the ground. Finally, the figure is more impressive than the background: it has "thing-character," whereas the background has "stuffcharacter." Note this change of character when the figural organization
in pallel B reverses.
In view of these differences between figure and ground, it is not
surprising that elements and aspects of a picture are more likely /to!> be
noted and remembered if they belong to the figure than if they belong
to the ground (Weitzman, 1963). Similarly, Luria (1961) reports that
children find it much easier to respond to information conveyed by the
figure than to information conveyed by the background. In general, then,
the figure is what perception is about.
In the age of introspection in psychology, attensity, or vividness
were the terms for the attribute that a sensation gains when it is attended
to (Titchener, 1908, 1915). Titchener identified attensity with clearness,
and sharply distinguished this attribute from intensity. Thus, a baby's
whimper can be heard with high attensity over the roar of a storm. It
requires little introspective indoctrination to agree that the figures in
panels A-D are perceived with much greater attensity than are the backgrounds.
In each of the other panels of Figure 5-3, there is also one area of
the field that has greater attensity, or greater Figural Emphasis than
others. The illustrations provide examples of several determinants of
. figural emphasis. These determinants can be divided into three sets, similar to the sets of factors that control eye movements: (1) innate dispositions that operate on physical characteristics; (2) collative factors; and
(3) selective intentions.
(1)
In general, as shown in panel C of Figure 5-3, the smaller object

tends to be seen as figure, and the larger object as ground (Koffka,
1935). There also appears to be a strong tendency to favor objects
in warm colors, such as red or yellow, over cool colors, such as blue
or green. Moving objects are particularly likely to be seen as
figure. This combination of rules must have helped our distant
forebears perform the vital task of detecting prey and predator
on a background of sky and vegetation. Contour-rich (panel D) and
isolated stimuli (panel E) are also favored, as are bright objects.
Because the same features of stimuli control both the fixation of the
eye and the selection of a figure, the figure is usually fixated in
preference to the background. Figural selection and fixation are
functionally independent, however, since a deliberate fixation on
the ground does not always reverse the prevailing figural organization.
(2)
(3)
79
Collative factors are illustrated in panels F and G;, where the odd
element tends to "stand out," as does the CAPITALIZED word in
this sentence. There may be an impression of three-dimensionality,
although not as pronounced as in panels A-D.
The effect of intentions is easily observed by deliberately selecting an element in any of the panels, e.g., a particular letter in
panel G. One may choose a larger unit, but at a certain cost in the
experience of attensity. It is this feature of the experience of attention which suggests the frequently used metaphors of "a beam
of light of varied width" (Hernandez-Peon, 1964) or "a lens of
variable power" (Eriksen & Rohrbaugh, 19-70; Eriksen & Hoffman,
1972).
The sllbjective experience of selective attention to inputs is closely

related to the experience of figural emphasis. It is therefore reasonable
to describe selective attention as a consistent emphasis on a class of
perceived objects or perceived events in preference to others. Thus,
paying attention to the red objects in a scene means that these objects
will be seen as figures, all together or one at a time. Listening to the
radio while one's children are fighting means that the announcer's voice
must be heard as figure, the children's screams as background.
In general, we succeed superbly in such tasks. Indeed, the main
limitation on our ability to control attention occurs at the stage of unit
formation. If a list of digits and a list of letters are recorded in such
manner that one hears- both messages in the same voice over the same
speaker, it is virtually impossible to attend selectively to the digits, because successive digits do not form a distinct unit, while the simultaneous
digit and letter tend to fuse. The same experiment may be carried out
in vision: if a mixed array of digits and letters is briefly exposed, it is
almost impossible to read only the digits, again because of a failure of
the digits to constitute a group. In contrast, it is possible to read items
printed in red or listen to a message spoken by a woman and ignore a
simultaneous message spoken by a man. Finally, it is quite easy to attend
to the voice heard from the right, or to the top row of a visual array.
Where the unit formation stage provides several "good" groups or units,
it is usually possible to deliberately select one of them for the role of
figure.
The decision to select some stimulus for special emphasis can be
made before the stimulus is actually shown, with immediate effects on
how the stimulus is perceived. Perhaps the most compelling evidence for
this conclusion is the phenomenon of prior entry. When a subject is told
that he will see a flash of light and hear a tone at about the same time,
and that he is to attend especially to one of them, the perception of simultaneity is biased. The stimulus that is attended to is perceived as
80
occurring relatively sooner than the other. Consequently, the two appear
to be simultaneous when the attended stimulus is actually shown later
than the other (see p. 137).
THE CONTROL OF FIGURAL EMPHASIS IN SEARCH
Eriksen and Collins (1969a) have recently described an impressive

demonstration of a figural effect in a search task. They used a visual display to present the digits 1-9 in rapid succession, omitting one digit of
the sequence on half the trials. On each trial the subject reported
whether the sequence was complete. Two conditions were compared:
a precuing condition in which the subject was told in advance which
digit, if any, would be omitted, and a postcuing condition in which
the same information was given only after the exposure. Performance
was vastly better in the precuing condition, where subjects set themselves to look only for the designated target. As a result of this set, the
target digit "stood out" clearly in perception whenever it was shown.
The effect was so strong that subjects confidently asserted that the target
had not been shown whenever it failed to "stand out" perceptually, and
they were usually right, even when the digits were presented at the fast
rate of one item per 50 milliseconds. In the postcuing condition, on the
other hand, subjects achieved perfect performance only if they could
perceive and identify each successive digit, and this required at least
200 milliseconds for each item.
The experience that a designated target tends to become figural
over an indistinct background has been described in other studies of
visual search. In a paradigm developed by Neisser (1963, 196:5; Neisser,
Novick & Lazar, 1963), subjects are instructed to look for a particular
letter or for anyone of several letters in an array. The subject scans the
array, line after line, and the time that he needs to decide that a line
does not include the target is measured by plotting the latency of the
detection of the critical line against the position of that line on the page.
For example, if a target on line 20 is detected in 16.5 seconds, and a
target on line 30 is detected in 24.5 seconds, then the time-per-line must
be 0.8 seconds. In some of Neisser's experiments, highly practiced subjects could scan a line of four letters in, as little as 0.1 second, but this
high speed was achieved at the cost of many errors of omission. Under
these conditions, Neisser's subjects reported that the non-target items
were seen as a mere blur, while target letters appeared to jump from
the line.
As might be expected, the speed of search depends on the ~ase
with which the target can be discriminated from its background. Even
81
. afte:r; prolonged practice Neisser's subjects found it more difficult to

search for a Q in an array of rounded letters (C, G, 0, etc.) than in an
array of angular letters (X, M, K, etc.). The difficulty of discriminating
target from background is also increased when the background elements
are heterogeneous. This effect was observed in a series of studies by
Gordon (1969; Gordon, Dulewicz & Winwood, 1971).
Skill in the performance of search tasks is acquired slowly (Rabbitt,
1964, 1967). When searching for multiple targets, the observer gradually
develops the ability to respond to critical features that the targets share.
Thus, Rabbitt (1967) showed that subjects trained to discriminate the
two target letters C and
from the irrelevant set (A, E, F, H, I, K, L)
transferred readily to a discrimination with the same targets and a new
irrelevant set (M, N, T, V, W, X, Y), because both irrelevant sets consisted of letters with no curved segments. Negative transfer was obtained
when the new irrelevant set consisted of curved letters (B, D, G, P,
Q, S). Rabbitt agreed with Neisser that irrelevant items are not analyzed
in as much detail as are the relevant targets. In the terms of this chapter,
the targets are emphasized more than other items.
It is not clear whether the figural emphasis on the target occurs
directly, or through the mediation of recognition units. Under some conditions, the stage of figural emphasis can be preset so that a stimulus
which possesses certain features will gain emphasis. This mechanism is
involved in the selection of inputs. Alternatively, a stimulus may first
activate the recognition unit for a target, and this tentative recognition
would cause the emphasis. The recursive path of attention control was
discussed earlier in the context of Sokolov's neuronal model theory of the
orientation response. Collative properties such as novelty or incongruity
can only affect perception through such a path, because these properties
arise from a mismatch between stimuli and expectations: the comparison
of stimuli to expectations requires the participation of recognition units.
A similar mechanism appears to be involved in the control of eye movement: as was shown in Chapter 4, the decision to have a closer look at
a target is triggered by a rapid evaluation of how much information was
acquired in the last fixation. An evaluation of information can also lead
to the eye lingering slightly on a single fixation. These decisions which
occur between saccades must be completed within 125-175 milliseconds,
although they require complex computations in the recognition system.
In general, then, the reCllrsive path of attention control is involved
when the initial analysis of a stimulus does not yield a sufficiently detailed and complete perceptual interpretation. This may occur with novel
or ""incongruous stimuli, and also in some search tasks. Neisser (1967)
mentioned that subjects who were looking for several targets at once
often became aware that they had detected a target before they knew
SENSORY
REGISTRATION
00
t..o
ENDURING
DISPOSITIONS:
( 1 ) Standard rules:
moving objects, warm
colors, etc.
( 2 ) Collative variables,
,.,---.........
//
"\
Rules of grouping: \
similarity,
proximity, etc.
novel ty, etc.
""
"'- - - " " .,,/
UNIT FORMATION
/"_
FIGURAL EMPHASIS
I/
(
\
MOMENTARY
INTENTIONS
(e. g., search for
blue objects, listen
to man on the right,
etc. )
/'
--- ----- "
Tentative
\
'\
recognition of \
signi ficant obj ect.
\
Violation of
/
expectations
\
""'------"'"
,/
FIGURE 5-4
Attention, grouping, and figural selection.
ACTIVATION OF
RECOGNITION UNITS
I.'
I.'
..
-........,
Effects of
coding
availability
"
/
--'"
"
\
what it was. This experience would be expected if a recursive path is

involved. Thus, the observation that a target "jumps" from the background does not prove that the target was detected_pre-attentively
(Cavanagh & Chase, 1971). It is possible that the rapid rate of scanning
in search provides a degraded visual input which is sufficient to weakly
activate some recognition units, but is not sufficient to provide complete
perceptual interpretations. The activation of the recognition unit corresponding to a target could cause additional capacity to be allocated to
the relevant object, which would then become figural.
A study by Lawrence (1971) provides suggestive evidence that it
may take appreciable time for a stimulus to call attention to itself. Subjects were shown a series of words which were successively presented
in the same place. They were instructed to detect and read the one
word in the series that was printed in capitals. Although subjects were
usually very confident of their answers, they erred quite frequently by
reporting a word that actually appeared after the target. This result is
consistent with the assumption of a recursive loop. If the redirection of
attention takes time, the word that initially called for attention will have
been replaced by another before the cycle can be completed. However,
the recursive loop could not have been involved ill the experiment of
Eriksen and Collins (1969a) that was described earlier, because the rate
of presentation was too fast. At present there is simply not enough information about the conditions under which the control of attention is direct
or recursive, although there appears to be sufficient evidence that both
modes of control are sometimes adopted.
It appears reasonable to assume that the case in which a stimulus
calls attention to itself is not fundamentally different from the case in
which attention is directed by a cue. In a classic study, Averbach and
Coriell (1961) analyzed the effect of a cue which indicated the location
of a target in a complex tachistoscopic display. The warning cue was
fully effective only when it preceded the display by 100-200 milliseconds
or more. Thus, the redirection of attention appears to require that
amount of time. Eriksen and Collins (1969b) later confirmed this measurement, with a procedure which eliminated possible artifacts of eye
movements.
Space is involved, as well as time, in the control of attention. It is
difficult to direct attention to a specific target in a crowded field. When
a complex display is shown in the tachistoscope, a common type of error
is the report of an item adjacent to the designated target (Eriksen &
Rohrbaugh, 1970). This adjacency effect is much greater when the indicator is presented simultaneously with the target than when it precedes
the target by 200 milliseconds. When the presentation of the array and
the indicator is simultaneous, 100-200 milliseconds may be needed to
84
direct attention to the target. By that time the display has been removed,
and attentioll must be directed toward a fading representation. The frequency of adjacency errors indicates that spatial confusions are likely to
occur in that case. These difficulties in "addressing" a target are increased
both by crowding the items and by presenting a large number of items.
Differences in reaction time as a function of addressing difficulties can
be found even with clearly visible stimuli (Eriksen & Hoffman, 1972).
A particularly impressive observation was reported by Snyder (1972),
who found adjacency errors even when the relevant target called attention to itself. Subjects were briefly shown a display of 12 different letters.
One of these differed from the others, in color, in orientation, or by
being fragmented. The subjects were required to report the odd letter.
In many cases they erred and reported one of the letters adjacent to the
target. This result provides strong evidence for recursiveness in the control of attention.
The examples that have been discussed so far were all visual, but
the concepts of grouping and figural selection apply to other modalities
as well. There is a clear experience of grouping in audition, both in
space and over time. There is also an experience of figure-ground organization: as we listen to a concerto, the soloist often provides the figure,
and we can also deliberately choose to attend to one group of instruments when the orchestra is playing.
The main conclusions of the preceding discussion of unit formation
and figural emphasis are illustrated in Figure 5-4. The figure suggests
that the process of figural emphasis should be viewed as the allocation
of effort, capacity, or attention to the perceptual elaboration of some perceptual units in preference to others. The allocation policy is governed
by enduring dispositions and by momentary intentions. As was the case
in the control of eye movements, the enduring dispositions that control
figural selection are of two types: standard rules that allocate attention
whell certain physical characteristics are detected, and collative features
such as novelty or significance. The standard rules can be applied before
the stimulus information makes contact with recognition units. Collative
variables, on the other hand, can only affect figural selection through a
recursive path.
The analysis of figural emphasis as a special instance of allocation
of attention implies that perceptual processing draws on the limited
capacity system that was discussed in Chapters 2 and 3. This conclusion
is supported by the frequent observations of perceptual deterioration in
a crowded and -complex field (Eriksen & Lappin, 1967; Eriksen & Rohrbaugh, 1970; Keeley, 1969; Mackworth, 1965; Rummelhart, 1970). In an
important study, Sperling et ale (1971) showed that information is extract~d
at approximately the same overall rate from arrays of varying complexity,
85
and also that the processing of different elements in an array is carried

out in parallel. Thus, the same total processing capacity is allocated, in
different ways, to simple and to cOluplex arrays. The relation between
perceptual processing and the non-specific capacity is most clearly established by the observation that perception is impaired during mental
effort (Broadbent & Gregory, 1963; Kahneman, 1970; Kahneman, Beatty
& Pollack, 1967; Shulman & Greenberg, 1971). The detection and identification of brief or faint stimuli deteriorate when attention is withdrawn
from perceptual elaboration to other activities. The vulnerability of perception to the competition of other activities indicates that all draw on
a common pool of capacity-or attention.
THE ORGANIZATION OF RECOGNITION UNITS
The concepts of recognition units and perceptual interpretations

are applied in a very broad sense in the present treatment. They refer to
the perception of features of objects, such as size, shape, or color, as well
as to the recognition and implicit naming of objects. The recognition system includes many functionally independent replicas of each recognition
tInit. This duplication is illustrated by our ability to see the pattern
XOXXX. If there were a single X-detector, we would see an "0," paired
with a single "X" of unusually high intensity! Some recent theories of
speech-recognition, such as the logogen model (Morton, 1970a) do not include this duplication feature, and propose that a single logogen corresponds to each word-meaning. This assumption mayor may not be
valid with respect to the recognition of meanings. It is clearly not valid
with respect to visual recognitions.
The recognition units appear to be organized in dimensions and
in levels. The presentation .o f a stimulus normally causes activation of
several units in each of these sets, and the role of the selection stage is to
choose no more than one interpretation from each set. Thus, a single siz~~
color, shape, and semantic meaning will eventually be perceived, although many more possible percepts may have been implicitly considered and rejected by the system.
Recognition units at several levels can participate in perceptual interpretation. A salient example is the effect of word context on letter
recognition (Reicher, 1969; F. Smith, 1969; Wheeler, 1970): when a subject is shown the word WORK in the tachistoscope, and is asked if the
last letter of the word was D or K, he does better than if the initial
sttmulus was CORK. The effect is particularly surprising, of course, because both WORD and WORK are words. How could the redundant
ilJ-itialletters aid in the discrimination?
86
A possible illterpretation of this result is that sensory information

activates recognition llnits both at the level of individual letters and at
the level of words. Because the signal is faint and the recognition system
is noisy, it may happen that the recognition unit for WORK is activated
more thall the recognition unit for WORD, while the separate units for
the letters K and D are activated eqllally. If the recognition units at the
word level are in contact with the selection of interpretations for individual letters, the recognition of a letter within a word will be superior
to the recognition of the same letter in a nonsense syllable. In addition/to
the recognition units for letters and for words, there is evidence for
llnits at the level of the spelling pattern (Gibson, 1965), the syllable
(Smith & Haviland, 1972), and perhaps at other levels in the parsing of
printed words (Spoehr & Smith, 1972). Recognition units at all these'
levels could collaborate in the selection of an interpretation at any level,
e.g., reading a whole word or identifying a single letter. There is no compelling reason to assume that the units at the various levels are arranged
in series, so that the output of one is th'e input for the other. An essentially parallel organization appears more plausible, in an overlearned
skill such as reading.
The organization of recognition units has been extensively studied
in two major paradigms: search and speeded judgments of sameness or
difference. Posner (1969, 1970; Posner & Mitchell, 1967; Posner, Lewis
& Conrad, 1972) has. contributed several detailed analyses of this problem. The main tool that he employed is a same-different judgment with
different rules. For example, a subject maybe instructed to press the
"same" key only when two stimuli have physical identity..He then responds "same" when exposed to the pair of letters a-a, but he responds
"different" for the pair a-A. In the condition of name-identity, the correct
response Jor a-A would be "same." In the condition of rule-identity, the
correct response for a-U could be "same," because both are vowels,
while the correct response for a-B would be "different."
A qllestion of central interest is whether the different "codes" for a
stimllllls are elicited in parallel or serially. For example, the three instrllctioI1S call for different codes of the stimulus "a": as a visual shape
(for the detection of physical identity); as a letter name (for name iden~
tity); and as a vowel (for rule identity). Are the three codes generated in
sequence? Note that the inferences that may be drawn from reactiontime data are not symmetric: if judgments under two. instructions are
equally fast, this provides evidence for independence of the corresponding codes. On the other hand, a difference of reaction-time does not
provide equally strong evidence for a serial-dependent production of
the codes because of the possibility that the two codes are generated ~n
parallel, but at differellt speeds. Posner (1969) presented considerable
evidence that the visual code and the name code for letters are in fact
produced in parallel, although the production of the name code is
slower. In the terms of the present chapter, this means that the visual
letter "a" simultaneously activates at least two recognition units: a unit
that is specific to the lower-case "a," and a unit for which both the
lower-case and the capital forms are appropriate stimuli.
The conclusion that two physically distinct'stimuli such as "a" and
"A" can make contact with a common recognition unit without the mediation of prior recognitions is of fundamental importance, because it contradicts a common view of perception as consisting of the sequential
production of increasingly abstract codes for a stimulus.
From the conclusion that "a" and "A" activate a common recognition unit, there is .but a small step to the idea that all numerals may activate a recognition unit, and all letters another. Evidence from both reaction time (Posner, 1970) and visual search (Brand, 1971) indicates that
this is the case. In contrast, there seems to be no common recognition
unit for vowels (Posner & Mitchell, 1967), and the identification of a
letter as a vowel or consonant therefore requires the elicitation of the
name code for that letter. As Posner (1970) pointed out, the availability
of an immediate code common to all letters or all digits, and the absence
of such a code for vowels, reflects the manner in which these materials
are learned. With. very prolonged practice, a common recognition unit
can apparently be formed even for totally arbitrary collections (Rabbitt,
1967).
The very slow process in which recognition units evolve has been
described in an important text by Gibson (1969). In a recent treatment
of attention, Broadbent (1971) spoke of a slow process of categorizing,
by which different stimulus configurations which are associated with the
same response (e.g., the letter "a" in different handwritings) eventually
come to elicit the same category state (here called perceptual interpretation). This type of perceptual learning results in a recognition system
which is both highly refined and well adapted' to the requirements of the
environment.
ELEMENTARY CONCEPTS OF SIGNAL-DETECTION THEORY
According to the model introduced earlier in this chapter, the

activation of a recognition unit depends on the match between sen~sory data and the specific features to which the unit responds, and on a
variable that was labeled perceptual readiness. At anyone time we are
more ready to recognize some events than others, and a sensory signal
. that will enable us to recognize a familiar event with confidence may not
88
suffice to identify a less familiar event. Bruner (1957) has offered a classic
treatment of this perceptual readiness. He described the main manifestations of readiness by the example of a man who is peculiarly ready to
see apples: "The apples will be more easily and swiftly recognized, a
wider range of things will be identified or misidentified as apples, and in
consequence the correct or best fitting identity of these other inputs will
be masked [Bruner, 1957, p. 130]." The main determinants of the perceptual readiness for a particular stimulus are the past frequency of its
occurrence, its probability of occurrence in the momentary context, and
its present significance to the individual.
Recent treatments of perceptual readiness have increasingly used
the tools and concepts of signal-detection theory. Some of the same concepts have also become central to theoretical treatments of attention'
(Broadbent, 1971; Norman, 1968; Treisman & Geffen" 1967). It will therefore be useful to briefly introduce some essential terms of signal-detection theory (for a more detailed, highly readable treatment, see Coombs,
Dawes & Tversky, 1971).
Signal detection theory was originally developed to account for
studies of detection and discrimination with a yes-no response (Green &
Swets, 1966; Tanner & Swets, 1954), in the general paradigm illustrated in
Table 5-2. In this paradigm the experimenter presents the target .stimulus
on some trials but not on others, and the observer indicates on each
trial whether he believes the target was present or absent. The four
entries in Table 5-2 represent the possible outcomes of such a trial.
The most obvious observation in this situation is the variability of
the subject's behavior on repeated occurrences of the same condition.
He sometimes says "Yes" and sometimes says "No" both when the stimulus
has occurred and when it has not. Signal detection theory explains this
unreliability by assuming the existence of internal noise, which causes
the value of a hypothetical sensory magnit~de to vary randomly over
time, even in the absence of a signal. When a signal is shown, the sensory magnitude increases by a certain amount, depending on the intensity of the Signal. If the signal is weak, it is possible that the sensory
magnitude produced by the combination of signal and noise is less than
TABLE 5-2
The basic structure of the signal-detection paradigm.
Response
Present
No
Miss
Yes
Hit
Correct
Rejection
False
Alarm
Stimulus
Absent
values that occasionally occur by noise alone. The subject, of course, cannot know whether the sensory magnitude that he experiences on a given
trial is due to noise /alone or to a combination of noise and signal. What
he does know is that higher values of sensory magnitude are more likely
to occur when the signal was presented than when it was not. Under
these circumstances, a rational observer will adopt a criterion, i.e., determine in advance a critical value of sensory magnitude. On any trial he
will say "Yes" if the sensory magnitude exceeds that criterion, and "No"
otherwise. The vallIe of the criterion is often labeled by the Greek letter
Beta.
Figure 5-5 illustrates these concepts by two elementary examples.
In both panels A and B, the distribution at the left represents the probability that a value of sensory magnitude will arise from the internal
noise of the system. The distribution of noise is assumed to be normal,
and it has been standardized so that its mean is zero and its standard deviation is one. The two panels also present the hypothetical distributions
of sensory magnitude for trials on which the signal is shown. The
illustrations refer to the simplest possible situation, where the variance
of the distribution is not affected by the introduction of a signal. The
same considerations apply to the more realistic models, which assume
that the signal causes both a shift of the distribution to the right and an
increased variance. The signal in panel A is weak, and it causes the distribution of values to shift by only half a standard deviation, relative
to the noise distribution. The distance between the two distributions,
in standard units, is the sensitivity parameter of the theory. Sensitivity is commonly denoted by the symbol d'; in panel A, d' == 0.5. In
both panels A and B, two values of the criterion are indicated, at values
of 0.0 and 2.0 on the scale of sensory magnitude. A subject who adopts a
criterion of 0.0 will say "Yes" if the value of sensory magnitude exceeds
the mean of the noise distribution. A Beta of 2.0 signifies that the ob..
. server says "Yes" only if sensory magnitude is higher than the mean of
the noise distribution by two staIldarddeviations or more.
The four panels of Table 5-3 present the expected performance of an
observer in the four situations illtIstrated in Figure 5-5 (d' == 0.5 and
TABLE 5-3
Distribution of responses in four signal-detection problems.
Beta == 0.0
d' == 0.5
Beta == 2.0
d'== 1.5
Beta == 0.0
d' == 1.5
Beta == 2.0
"Response
No
Yes
No
Yes
No
Yes
No
Yes
Present
,Absent
.308
.500
.692
.500
.933
.976
.067
.024
.067
.500
.933
.500
.692
.976
.308
.024
d'- 0.5
. Stimulus
Beta = 0.0
( A)
No
Yes
+-- --+
Yes
~
Signal and noise
-2.0
-1.5 -1.0
-0.5
0.5
1.0
1.5
2.0
'---y--J
d'
Sensory magn itude
( B)
Beta = 0.0
No
Beta = 2.0
No
Yes
+--
Yes
+-
Signal and noise
-2.0 -1.5
-1.0 -0.5
0.5
1.0
1.5
2.0
d'
Sensory magnitude
FIGURE 5-5
Constructions from signal-detection theory.
d' ==.1.5; Beta == 0.0 and Beta == 2.0). Note that performance in. the
absence of a stimulus depends only on the criterion, while performance
when the stimulus is present depends on both the criterion (Beta) and
the discriminability of the signal (d').
In a real experiment, of course, the data are obtained in the form
illustrated by Table 5-3, and the structure illustrated in Figure 5-5 is inferred from these data. Given the four entries in a table of experimental
results, d' and Beta can be calculated, by using the assumptions of. the
90
91
model: i.e., normality of the distributions, and equality of the variances

of the noise and signal-and-noise distributions. Furthermore, these assumptions can be checked and altered if needed, by measuring detection
performance for the same signal (i.e., at a constant value of d') under
several values of the criterion. The criterion is most simply manipulated
by instructing the observer to indicate his confidence in each response:
any observer spontaneously adopts a lower criterion for the response
-"Perhaps there was a signal" than for the response 4:'There certainly was a
signal." Accordingly the ratio of hits to false alarms is expected to be
higher when the subject expresses high confidence than when confidence
is low. The theory provides a precise prediction of this ratio. The criterion can also be altered by other experimental manipulations. For
example, an observer normally adopts a lower criterion if signals are frequent than if they are rare, and he also adopts a lower criterion when
penalized for. misses than when penalized for false alarms. The detailed
predictions that signal detection theory entails for the effects of these
manipulations have often been spectacularly confirmed.
The two parameters of the theory, sensitivity and the criterion, provide a much needed tool in the analysis of many situations. For example,
observers in a vigilance task fail more often to respond to signals at the
end of a tedious session than at the beginning. It is natural to ask
whether this vigilance decrement is due to an impaired ability to detect
the signal (lower d'), or to an increasing unwillingness to respond to signals (higher Beta). Broadbent (1971) and Mackworth (1969, 1970) have
provided detailed treatments of this issue. Similarly, when a subject attends to one message and fails to respond to another, it is possible to determine whether d' or Beta has been altered by the lack of attention to
the rejected message (Broadbent & Gregory, 1963; Moray & O'Brien,
1967).
PERCEPTUAL READINESS AS A CRITERION BIAS
The distinction between sensitivity and criterion suggests an elegant approach to the fascinating and intractable question of perception
vs. 'response. It is tempting to identify d' as a measure of perceptual
efficiency arid Beta as a measure of response readiness. Indeed, the ease
with which Beta can be altered in the detection paradigm appears to
support such an identification. In the context of recognition, however, a
low value of Beta for a particular recognition response has gentline
effects on perception, as shown in the following ~xample.
Consider the picture of the room in panel A of Figure 5-6. All ob. servers see it as a normal rectangular room, but in fact it is not, as shown
92
FIGURE 5-6
Interior and exterior views of a distorted room (Ittelson, 1952, with
permission).
by the exterior view of panel B. The distorted room was carefully constructed so that when it is viewed or photographed from a particular
spot, the image that it casts on the retina, or on the photographic plate,
is identical to the image cast by a rectangular room (Ittelson & Kil-
p_atrick, 1951). The photograph is ambiguous, because it provides no

clue to distinguish the veridical distorted room from a standard room,
or from an unlimited number of other distorted rooms that could be
constructed to cast the same image. The ambiguity is not reflected in
perception, which adopts an unequivocal decision in favor of the standard room. Now consider this example in the terms of signal-detection
theory: the room was constructed to provide no signal that would be
relevant to the decision, so that d' == O. Thus, the perceptual decision
that the room is rectangular rather than distorted is made in the absence
of a relevant signal. As Table 5-3 showed, decisions made in the absence
of a signal provide a pure measure of criterion bias. It follows that our
seeing the rectangular room represents a criterion bias. The reader is
therefore in a position to observe the powerful effects that such a bias
can have in perception. Bruner's term of perceptual readiness is justly
applied to such effects.
The well-known word-frequency effect is a controversial instance
of perceptual readiness. The identification thresholds of frequent words
are markedly lower than those of words which are rare in the language
(Howes & Solomon, 1951). Frequent words are identified at a lower
loudness than rare words in auditory presentation and at a shorter
duration of exposure . in visual tachistoscopic presentation. The elementary concepts of signal-detection theory cannot be applied to this situation, because the response vocabulary consists of the entire language in
word identification, and only of two responses in the detection tasks that
were initially treated in the theory. However, signal-detection theory can
be adapted to the identification situation. Broadbent (1967, 1971) and
Morton (1968, 1969a) carried out this task, and they derived testable
consequences from several possible models of the word-frequency effect.
The detailsof the mathematical analysis exceed the scope of the prese11t
review, but the Havor of the approach is conveyed by an illustration that
. Broadbent (1967) provided, of "... a vast array of test tubes, each partly
full of water and each corresponding to a word in the language. The
choice of one tube corresponds to perception of a word, and the probability of choice in any tube is greater when the water level in it is higher
[p. 3]." In this model, the presentation of a word causes the level of
water to rise in the appropriate tube. The .amount by which the level
rises corresponds to the sensitivity parameter of detection theory, and the
initial level of water in each tube .represents the level of. the criterion
for the recognition of "its" word.
In the terms of this analogy, Broadbent (1967) and Morton (1968)
concluded that the presentation of a word does not raise the water level
by a greater amount if the word is of high frequency than if it is rare.
. The word-frequency effect is entirely due to the initial level of water in
the tubes, i.e., to a criterion difference. The main reason for this conclu-
94
sion is that subjects' incorrect responses in the word-recognition task include large numbers of frequent words (Broadbent, 1967; Brown &
Rubenstein, 1961; Pollack, Rubenstein & Decker, 1960)~ In the absence of
stimulus information the subject is more likely to guess that a highfrequency word has been presented, which is clear evidence for a criterion bias. For a more detailed discussion of this conclusion, see Catlin
(1969) and M. Treisman (1971).
Perceptual readiness probably mediates the context effects that play
a crucial role in our ability to recognize events on the basis of impovern
ished and degraded cues. Thus, the recognition unit for "your is almost
certainly activated to some degree whenever we are exposed to the word
"year." Nevertheless, mistakes of interpretation will be rare because of
the probable presence of contextual cues which increase the readiness
to recognize one of these words ("please give me y-r coat") or the other
("he will graduate next y-r").
In Figure 5-7, the concepts of signal-detection theory are related
to the information-processing sequence which is the topic of this chapter.
The figure suggests that the sensitivity (d') and the criterion (Beta) for
any response are each affected by events at several stages of the
sequence.
Sensitivity (d') is affected by the quality of the information that is
delivered to the recognition units. Sensitivity is high if the initial signal
was loud and clear. There'is also evidence to suggest that sensitivity is
high for an object to which we pay attention, and which has been selected at the earlier stage of figural emphasis (Broadbent & Gregory,
1963; Kahneman, Beatty & Pollack, 1967; Moray & O'Brien, 1967; Treisman & Geffen, '1967). In addition, sensitivity is affected by the availability
of recognition units: if an American and a Chinese adult are compared in their ability to discriminate Chinese characters, the outcome
will surely be a vast superiority of d' favoring the person who has had
lifelong experience with these characters.
The criterion level (Beta) is determined by events at two different
stages of the sequence. A state of perceptual readiness affects the selection of interpretations, in the manner illustrated by the example of the
distorted room. In addition, a criterion bias' may operate at the subsequent stage of response selection. A subject in a tachistoscopic experiment, having tentatively identified a briefly exposed word as WHORE
may nevertheless opt for WHOLE as a s~fer response, lest his mind be
thought dirty.
It is assumed in Figure 5-7 that the response system is itself noisefree. Both d' and Beta will be altered if there is unreliability in the
selection or execution of responses. Imagine,' for example, that the observer indicates a yes-no response by 'pressing one of two keys, which
Quality of
Sensory Signal
SENSORY
REGISTRATION
increases d '
UNIT FORMATION
'V
Allocation
of Attention
increases d '
\/
\V
'li
FIGURAL EMPHASIS
I
I
\{J
\1/
Availability
of
Recognition
Units
\ It
ACTIVATION OF
RECOGNITION UNITS
increases d '
\ II
, I
1.,
\11
W
Perceptual
Readiness
SELECTION OF
INTERPRETATIONS
lowers Beta
I I
I I
I I
I I
Response
Readiness
lowers Beta
SELECTION OF
RESPONSES
FIGURE 5-7
Determinants of sensitivity and criterion level.
95
96
..
are so close together that he presses the wrong key on a significant number of trials. Such response noise invariably lowers d', and usually alters
Beta as well. It is probably safe to ignore the effects of response noise
in most psychophysical experiments. When signal-detection methods are
applied to studies of attention, however, the possible occurrence of careless responses should not be neglected.
Two of the five processes mentioned in Figure 5-7 are neither new
nor controversial: the quality of sensory registration surely affects d'
while response readiness is reflected in the value of Beta. The three
remaining processes . deserve a final 'comment.
The model of Figure 5-7 assumes that the allocation of attention to
an object enchances the sensitivity of the system (d') in dealing with that
object. This view is similar to the treatment of focal attention by Neisser
(1967). Other authors have adopted the position that the withdrawal of
attention from a stimulus causes a lowering of sensitivity, equivalent to
an attenuation of the input (Broadb'ent, 1971; Broadbent & Gregory,
1963; Treisman, 1960; Treisman & Geffen, 1967). But Norman (1968) has
proposed that all effects of selective attention can be explained by rapid
alterations of criterion biases. We shall be concerned with this issue in
Chapters 7 and 8.
The idea that the availability of recognition units is an important
determinant of sensitivity was emphasized by Broadbent (1971). In the
absence of appropriate recognition units, the selection of an appropriate
interpretation becomes impossible.
Broadbent's term for perceptual readiness is pigeonholing,which
he defines as "the process by which the nervous system adjusts so as to
allocate larger or smaller numbers of states of evidence to each category
state [p. xi]." In his terms, for example, the category state "rectangular
room" is a pigeonhole to which states of evidence (corresponding to
stimulus events, except for the effects of "noise" in the nervous system)
are very liberally assigned. Pigeonholing is reflected in the setting of the
criterion for a particular recognition. Pigeonholing, or perceptual readiness can affect the experience of perception. By suitable analyses, however, it is possible to distinguish between perceptual changes which
represent a shift of criteria and other perceptual changes which represent alterations of the sensitivity of perceptual analysis.
REVIEW
This chapter has described some perceptual processes which must

be considered in an analysis of attention. Perception was described as
the achievement of a set of interpretations. These interpretations are at-
tached to perceived objects or events which are segregated at an early

stage of Unit Formation. The allocation of attention to some of these
objects in- preference to others at the stage of Figural Emphasis alters
the quality of the information which is delivered to subsequent stages.
Figural emphasis represents an allocation of attention, which is guided
by the same enduring dispositions and momentary intentions that also
guide the allocation of eye movements. A perceived object may attract
attention because of a prior setting of figural selection. Alternatively, a
recursive alteration of the allocation policy may follow either a tentative
recognition that the object is significant-or a failure to establish an adequate interpretation of the object.
Recognition units are organized by sensory dimensions, and also
by level of analysis. Units at several levels can collaborate in achieving
an interpretation at one specified level: for example, the activation of a
recognition unit for a word may facilitate the recognition of a letter
in that word. In the terms of signal-detection theory, the availability of
reco-gnition units increases tIle sensitivity of the system (d'). The allocation of attention to an object was also assumed to affect d'e The criterion
parameter of the theory (Beta) is determined by two types of readiness:
perceptual readiness, which affects the selection of interpretations in
subjective perceptual experience; and response readiness, which affects
the selection of responses at a post-perceptual stage.
6
Attention to Attributes
In everyday communication we often use expressions such as "look at

the shape of this vase," or "look at the color of that shirt," which direct
the observer to attend to a particular attribute. The present chapter is
concerned with the processes that permit us to obey such instructions.
The first section reviews a few results from the vast literature of discrimination learning. Subsequent sections discuss the verbal report of attributes, speeded tasks of classification, and the Stroop test.
DISCRIMINATION LEARNING
The relation between attention and discrimination learning, origi11ally stated by Lashley (1942; Lashley & Wade, 1946), was rediscovered
and enthusiastically studied in the 1960s (e.g., Fellows, 1968; Lovejoy,
1968; Sutherland & Mackintosh, 1971; Trabasso & Bower, 1968). In a
typical discrimination problem the human or animal subject is faced
with stimuli that differ in many attributes, such as shape, color, size, and
number. The subject must learn to respond to a particular class of stimuli, defined by a simple rule which he must discover, e.g., "all large ob~)
98
99
jects are positive." In this example the size attribute or dimension is

relevant, and all other attributes can be ignored. There is much evidence
that learning in this situation occurs in two stages:
(a)
(b)
The subject first learns to "attend" to the relevant dimension.

He then attaches the positive response to the appropriate value of
the relevant dimension.
Dramatic examples of discontinuity between these two successive

stages of learning were described by Zeaman and House (1963), who
studied discrimination learning in retarded subjects. Their data indicate
that performance on a discrimination problem may remain at chance
level over several hundred trials. Once learning starts, however, criterion
is reached fairly quickly. The duration of the initial stage depends on
the salience of the relevant dimension and on the intelligence of the
subject. Zeaman and House observed that the relevant dimension was
discovered sooner by subjects of higher mental age. Once in stage (b),
however, retarded and normal subjects learned at about the same rate
(see Fig. 6-1).
Early versions of a discontinuity theory of discrimination learning
were stated by Krechevsky (1932, 1938) in terms of hypotheses and by
Lashley (1942; Lashley & Wade, 1946) in terms of attention. The major
assumption of discontinuity theories was that attention to a stimulus
dimension is on an all-or-none basis: the subject either attends to the
relevant dimension or he does not, and his performance on the discrimination task must remain at chance level as long as he attends to irrelevant dimensions. In the terms that Sutherland (1959) introduced, the
relevant analyzer must be "switched on" and irrelevant analyzers must
be switched off before learning can occur.
A modified discontinuity theory (Mackintosh, 1965; Sutherland,
1964) argues more moderately that some dimensions are vastly more
salient than others, and that the most salient dimension tends to dominate performance. This modification of the original discontinuity theory
is necessary to explain the fact that animals do learn something about
the (jrelevant dimension even while their performance is dominated by
another, irrelevant dimension (Mackintosh, 1965). The modified theory
retains the essential idea that the organism in a discrimination situation
does not associate a response to the physical stimulus but rather, in
Lawrence's (1963) phrase, to a stimulus-as-coded (S-A-C). Learning to
produce the appropriate code, or to attend to the relevant dimension, is
distinguished frqm learning the overt response.
Several formal models of learning (Lovejoy, 1966, 1968; Sutherland,
1964; Shepp, Kernler & Anderson, 1972; Trabasso & Bower, 1968; Zea-
100
100
Color-form obiect
Learners, MA 2-4
N = 21
Learners, MA 4-6
N = 20
90
80
(1)
Nonlearners, MA 2-4
(N = 19)
70
-c
0
.....
0
60
(1)
s..
s..
.....
50
Chance
(1)
0
s..
(1)
a..
40
Non Iearners, MA 4-6

(N = 11)
(1)
30
20
10
20
40
60
80
100
120
140
160
180
200
220
240
Trials ( T )
FIGURE 6-1
Effects of intelligence on discrimination learning are shown in the average
performance of four groups classified by mental age and achievement. From
Handbook of Mental Deficiency, edited by N. R. Ellis. Copyright 1963 by
McGraw-Hill, Inc. Used with permission of McGraw-Hill Book Co.
man & House, 1963) incorporate a two-stage notion.. These models have
much in common, although they differ in important details, such as the
number of dimensions that can be attended on each trial, and the precise
effects of nonreinforcement on attention and on the overt response. All
attention models can account for either continuity or discontinuity in
learning. Continuous learning is typically observed when the relevant
dimension is very obviolls, so that the subject may be in stage (b) from
the very first trial. On the other hand, sharply discontinuous learning is
predicted if the relevant dimension is obscure or if the learner is very
slow (see Fig. 6-1). Formal theories of discrimination learning are concerned with the details of several effects that Sutherland (1964) related
to his two-stage model: the transfer of discrimination training along a
dimension (Sutherland, Mackintosh & Mackintosh, 1965); the reversal
shift effect (Kendler & Kendler, 1962, 1970a, 1970b); and the overtraining
101
reversal effect (Lovejoy, 1966, 1968; Mackintosh, 1964, 1965; Wolford &
Bower, 1969).
Allattentional theories of discrimination learning assume that the
various attributes of objects are not equally attended. In the simplest
version of such a theory, only one dimension is attended when the discrimination is learned. For example, a pigeon may be selectively reinforced for pecking at a green circle, rather than at a red triangle (Jones,
1954). Shape and color are both relevant to this discrimination, and it is
of interest to discover what the animal learned. Does it now peck at circles, or at green objects, or at both? The answer is obtained by studying
transfer to a situation in which one of the relevant cues is kept constant
while the other is varied. There is general agreement that individual subjects in animal experiments typically learn only one of the relevant cues
and respond well to that cue and poorly to all others. The selected cue,
however, may be different for different subjects or classes of subjects
(Jones, 1954; Reynolds, 1961; Sutherland & Holgate, 1966; Sutherland &
Mackintosh, 1964).
The dominance of a single cue in a complex of relevant cues is not
restricted to lower animals. Trabasso and Bower (1968) illustrated various manifestations of cue dominance in a series of experiments with
student subjects. Thus, subjects who have solved the problem on one
cue often fail to notice any change when an initially irrelevant cue is
made relevant (and redundallt) in a later stage of the experiment. The
results of transfer tests are usually consistent with.a dominance hypothesis, although some subjects do solve discrimination problems on several
cues. Trabasso and Bower (1968) found no consistent differences in
learning rate between those subjects who solved a problem by two cues,
and those who solved it by only one.
Some factors that make a particlllar cue more salient than others
have been identified. Discriminability is such a factor. For example, if
ellipses are presented which vary greatly in overall size and only slightly
in eccentricity, alld both size ,and shape are relevant, then size rather
than shape will' dominate behavior in a concept-identification task
(Archer, 1962; Imai & Garner, 1965; Trabasso, 1963). The corresponding
effect in animal learning is called overshadowing (Mackintosh, 1971).
Prior learning is also important. A dimension that has been successfully attended in one discrimination problem tends to dominate performance in subsequent discrimination learning (Lawrence, 1949, 1950). On
the other,hand, there is evidence that prior experience in which a cue is
irrelevant significantly retards learning when that cue is eventually made
relevant (Goodwin & Lawrence, 1955; Levine, 1962; Lovejoy, 1968;
Mackintosh, 1964; Trabasso & Bower, 1968).
In addition, there are consistent individual differences (Shepard,
102
1964) as well as significant developmental changes in the relative salience

of cues. The prevalellce of mirror writing in children and its stubborn
resistance to training indicate that orientation is not a very salient dimension at an early age (Fellows, 1968). Suchman and Trabasso (1966a,
1966b) studied the relative salience of the dimensions of color and
shape for children of different ages. They used two experimental situations: jlldgments of similarity and discrimination learning. Similarity
judgments were dominated by similarity of color at age three and a' half;
at age six, similarity of shape prevailed (Suchman & Trabasso, 1966a).
This transition from color to form dominance is an important aspect of
general cognitive development (Kagan & Lemkin, 1961). The dimension
that dominates the similarity judgments of a particular child also determines his performance in discrimination learning (Suchman & Trabasso,
1966b). When both form and color are relevant, a child usually learns
only one Clle, and in most cases this is the same cue that also dominates
his similarity judgments.
A significant theoretical question concerns the locus of the determinants of discrimination learning in the sequence of stages of information-processing. The question was foreshadowed by Krechevsky's usage
of "hypothesis" for the same concept that Lashley later labeled attention.
"Attention" suggests an effect on perception while "hypothesis" does not.
Sutherland's concept of analyzer also implied an, operation at an early
stage of perceptual processing, and Treisman (1969) adopted that term
in proposing a general approach to attention. When the subjects are
rats, these distinctions have little operational significance, but the situation may be different with human subjects.
The general similarity of the results of human adults and lower
animals suggests that Krechevsky's term "hypothesis" may have been
more appropriate than "attention." If discrimination behavior reflects the
activity of analyzers, then the appearance of the objects of discrimination
would be expected to change in the course of learning. If sllbjective
reports can be accepted as evidence, this is simply not the case: the perceptual world of a college student hardly changes when he discovers
that color, rather than shape, is the relevant dimension in a conceptformation task. What the human subject learns is to attach the control
of responses to one or the other of a set of unchanging perceptual interpretations. The term code is often used in this context. It is the coding
of stimuli that is altered in most discrimination learning. The frequent
finding that a single cue is learned when many are available suggests the
important conclusion that subjects tend to develop the simplest possible
codes that will suffice for the task at hand.
Discrimination training probably alters perceptual interpretations
only when the relevant dimension must be discovered for the first time.
Evidence cited by Gibson (1969) suggests that the intensive experience
~
103
of first graders with the cues relevant to reading may have generalized
effects on the perception of objects and on the ease with which they can
be discriminated from one another. The slow process by which recognition units develop has been labeled categorizing (Broadbent, 1971).
SELECfIVE REPORT OF ATTRIBUTES
The process that Treisman called selection of an analyzer has been

studied in an experimental situation originally introduced by Kiilpe in
1904. In this experiment, the subject is instructed to attend to a particular attribute of a briefly presented object, such as its shape or color. The
subject reports the designated attribute on every trial, but he is also occasionally required to report some other features of the object. Observers
in such experiments often make errors when reporting attributes to
which they did not attend at the time of presentation, and introspection
suggests that the character of the perceptual experience may be altered
by the instruction to attend to a specific stimulus dimension. It is easy to
convince oneself that listening to the loudness of a varying tone can provide a different experience than listening to its pitch. Introspection is a
poor source of evidence, however, and Kiilpe's task obviously confounds
perceptual and response variables. Wilcocks (1925) raised the question
of whether Kiilpe's results represented an alteration of perception at the
time of exposure, or merely a failure to recall the neglected attributes.
The controversy over Kiilpe's effect has been well reviewed by
Haber (1966) and Egeth (1967), and a detailed description is unnecessary
here. Briefly, there are three main explanations for the superiority of
attended over neglected attributes:
(1)
(2)
(3)
According to the perceptual tuning hypothesis, the selected dimension "stands out" in perception at the time of presentation.
According to the response hypothesis, the attended attribute suffers less forgetting, because it is rehearsed more effectively and is
reported first (Lawrence & La Berge, 1956).
According to the encoding hypothesis, the attended and unattended attributes are treated differentially only at the point of
transition from sensory memory to the verbally encoded representation of the stimulus (Haber, 1964a, b; 1966; Harris & Haber,
1963). The attended attribute is likely to be encoded first, thus
gaining the advantage of primacy in recall.
Haber (1964b) also suggested that there may be no opportunity to

encode some of the attributes of a briefly presented object: encoding the
relevant attributes takes time, and the sensory memory decays very
rapidly (Averbach & Sperling, 1961). By the time the first attribute has
been encoded, information concerning other attributes may be lost.
104
The evidence for the encoding hypothesis was derived from an

experimental situation initially introduced by Lawrence and La Berge
(1956). On each trial the subject is shown two cards from the Wisconsin
Card Sorting Task. The stimuli on each card can be described by the
three attributes of numerosity, color, and shape (e.g., two red circles).
The subject reports all the information of the display (three attributes
for each of two objects), but he is sometimes instructed that one of the
attributes is more important than the others (emphasis instructiol,1).
Lawrence and La Berge (1956) had noted that the emphasized attribute
is usually the first to be reported, and they believed that the order of
report accounts for the effect of emphasis on accuracy.
Harris and Haber (1963) introduced the hypothesis that the order
of covert encoding may be more important than the order of the overt
report in determining the emphasis effect. They instructed subjects to
adopt one of two encoding strategies. Object Coding corresponds to the
structure of English syntax. An example is: two red squares; three blue
triangles. In Dimension Coding the information about the two cards is
organized by dimensions, e.g., red, blue; square, triangle; two, three. The
two strategies are often spontaneously adopted by uninstructed subje9ts.
The crucial difference between them is that the sequence of dimension
coding can readily be altered, whereas the sequence of object coding is
fixed.
In the experiments, Harris and Haber (1963; H~ber, 1964a) independently prescribed emphasis and order of report. They found that
emphasis had an effect on accuracy even with order of report controlled.
The most important result was that emphasis instructions altered the
sequence of covert responses of subjects who were using a dimension'
code: they usually encoded the emphasized dimension first. Object
coders could not do this. As a result, the effect of emphasis on the relative accuracy of report for the different dimensions was very pronounced
for dimension coders, negligible for object coders.
On the whole, however, object coding was the more effective strategy. Object coders are faster (Haber, 1964b) and generally more accurate.
This is an important result, which reflects a tendency to encode experience in terms of objects rather than in terms of dimensions. This
tendency has other manifestations; in another experimental situation
Lappin (1967) found that observers are much more accurate in reporting
three attributes of a single object than in reporting one attribute for
three objec~s. He also noted that order of report has a pronounced effect
on accuracy in the latter case. Lappin (1967) concluded that the processing of a single dimension with multiple objects is necessarily serial at
some stage, while the processing of several dimensions of a single object
may be parallel.
Attention to Attributes 105
Treisman (1969) accepted Lappin's conjecture. She explained his

findings and other results by assuming the existence of a set of analyzers.
Processing is necessarily serial within' each analyz~r, but it may occur in
parallel in different analyzers. Since the various dimensions of an object
are processed by different analyzers, they can be processed in parallel.
To explain why parallel processing is impOSSible with one dimension and
several objects, Treisman apparently assumed that there is only one
, analyzer of each kind. This is certainly incorrect: we are obviously caP1,lble of seeing more than one color at a time.
The model that was developed in the preceding chapter suggests a
different interpretation. The process of figural selection allocates attentionto objects rather than to dimensions. Analyses of all features of an
object, are facilitated whe~ that object is figural. When the observer is
required to report on attributes of three objects he must either extend the
figural 'area at some cost in detail, or else deal with the objects in
sequence.':,
In sUtl)mary, the study of Kiilpe's effect did not provide compelling
evidence that" attention to a dimension alters perception. The intention to
pay attention 'to a particular attribute appears to have its effects by increasing response readiness for a category of responses (e.g., color
names), and by controlling the quality and the sequence of encoding
and the order of report. This interpretation does not violate naive introspection, as you, can probably confirm for yourself. Listen to a brief tune,
while trying to pay special attention to the attribute of loudness. Now
listen to a tune and attend to pitch and melody. How did you interpret
the instruction to attend to Oll,e or the other attribute? You may find that
you acted as if you were preparing to recall the designated attribute
with special accuracy, after the termination of the tune. If this was the
case, did you adopt different strategies to store the two attributes? Could
the different experiences of listening to loudness and to pitch arise from
different modes of rehearsal? The encoding hypothesis is compatible
with such an account of the phenomenology of Kiilpe's task.
I
SPEEDED
CLASSIFICATION
Attention to attributes has been extensively studied in reaction-time

and speeded-classification tasks (Egeth, 1967). In a typical speededclassification experiment the subject is given a deck of cards, each containing a deSign; he is to sort the cards into piles according to some
attribute (e.g., all red objects into one pile, all blue objects into another).
As in the case of discrimination learnillg, the other attributes of the design maybe correlated with the designated relevant attribute (e.g., all
106
the red objects may be small circles printed near the top right corner of ,
the card, while all the blue objects are large triangles printed near the
bottom left corner). Alternatively, the other attributes may be orthogonal
to the relevant dimension (e.g., the red objects that are to be sorted together may vary in shape, size, vertical and horizontal position).
The basic design can be modified in a variety of ways (Posner,
1964). The most elementary version, in which a single attribute is relevant, is called a gating task. The term "filtering" has sometimes been
used in this context. In this book, however, filtering refers to the selectionof inputs .which share a particular attribute (e.g., all words printed
in red ink in a page of text), but call for distinct responses (e.g., reading
each word). In the gating variant of speeded classification, on the other
hand, the sa;me response is made to all the stimuli that share the criterial
attribute.
The condensation task is a more complex variant of speeded classification. Here the stimuli requiring a single response are defined either by
a disjunctive rule applicable to one dimension (e.g., red or blue stimuli
vs. green or yellow), or by a rule involving several dimensions (e.g., red
squares or blue circles vs. red circles or blue squares). Keele (1970) observed that the second variant of condensation is far more difficult than
the first. Accordingly, Gottwald and Garner (1972) referred to the easier
variant as grouping, and retained the term condensation for the harder
task. The dependent variable in speeded-classification tasks is the speed
achieved in sorting cards. Alternatively, the cards may be successively
presented in the tachistoscope, and the subject may be required to indicate his response by pressing an appropriate key as quickly as possible.
A vast amount of research has been devoted to the questions of
whether the attributes of a stimulus are interrogated sequentially or in
parallel, and of whether this interrogation is exhaustive or self-terminating. The results of this research are confusing and contradictory (e.g.,
Biederman, 1972; Biederman & Checkosky, 1970; Garner, 1970; Rabbitt,
1971; Smith, 1968). This outcome should not be surprising. Some of the
complex tasks that have been studied certainly involve covert verbal encoding of the various stimulus attributes prior to the selection of a response. In such tasks the processing of the relevant dimensions will
appear to be serial, because the encoding is serial. Furthermore, verbal
encoding is relatively flexible. The order of encoding can be altered, as
was shown in the preceding section, and the subject may be able to
terminate the encoding as soon as he accumulates sufficient information
(e.g., Biederman, 1972). In other situations the subject's task is simpler,
and he soon learns to dispense with verbal encoding. In those situations,
analyses of multi-attribtlte discrimination problems provide evidence of
parallel processing (e.g., Biederman & Checkosky, 1970; Hawkins, 1969).
Furthermore, the disappearance of verbal encoding is certainly gradual,
f
Attentioflt to Attributes 107
and the prevalence of serial or parallel processing varies with practice

(Marcel, 1970).
There is no reason to believe that subjects always process information in one manner, either serially or in parallel. As Garner (1970, p. 350)
pointed out: "Why must the organism do one or the other? Very probably it can do either, depending on the task and the stimuli. And even
as likely is that the organism frequently does both, not in the sense of
doing first one and then the other, but in the sense of doing both simultaneously." In a comprehensive review of modern reaction-time research,
Rabbitt (1971, p. 262) stated incisively: "My view is that the evidence
leads to no conclusion, but rather to doubt about the value of trying
to distinguish between serial and parallel processing as a guide to the
development of models and to experiments."
A more fruitful question about speeded classification concerns the
efficacy of gating. When the subject is told to sort stimuli according to
one attribute-e.g., color-does he in fact ignore variations in other at tributes? There are two ways of studying this problem: when another
dimension is correlated with the relevant dimension (e.g., the red objects
are always large and the blue objects are small), classification may be
facilitated by a redundancy gain. When the two dimensions vary independently, there may be interference. If the subject strictly obeys the
gating instruction, however, neither facilitation nor interference should
occur.
The analysis of the perceptual sequence presented in the preceding
chapter emphasized the importance of the initial processes of unit formation. From that analysis it is obvious that the hue and the size of a
single object (or group) are much more likely to interact than are the
hue of one object and the size of another. Similar considerations led
Lockhead (1966a,b) to a distinction between integral and nonintegral
dimensions. Integral dimensions are those which are presented simultaneously and at the same place. In short, they are the dimensions of a
single object: "Phenomenologically, it is difficult for a normal person to
look at a lighted and colored incandescent bulb without being awareat one time-of its hue, brightness, size and form [Lockhead, 1966a,
p. 103]." Garner (1970, p. 354), carefully avoiding phenomenology, offered a very similar definition: "Two dimensions are integral if in order
for a level on one dimension to be realized, there must be a dimensional
level specified for the other. For example, a visual stimulus must have
a brightness and a hue and a saturation and a size and a form. That fact
makes any pair of these dimensions integral."
Garner and Felfoldy (1970) conducted speeded-classification experiments to study the effects of integrality. When two integral dimensions were perfectly correlated in a deck of cards, and the subject sorted
cards according to one of these dimensions, sorting was faster than in
108
control sitllations in which the irrelevant dimension was not varied. This
finding was obtained with the hue and brightness (chroma and value)
of a color chip and with the horizontal and vertical position of a dot.
Conversely, interference was observed when the irrelevant dimension in
an integral pair was randomly varied. There was neither facilitation nor
interference when the dimensions were assigned to distinct objects, except in th~ rather trivial case where the subject found it more convenient
to concentrate on the "irrelevant" object, which provided a better clue
than the relevant one (Felfoldy & Garner, 1971).
Garner (1970) related the effects of integrality in speeded classifications to some observations obtained in the scaling of similarity. Thus,
Shepard (1964) had subjects judge the similarity of circles, each containing a single radius; the size of the circle and the orientation of the radius
were varied. These two dimensions are not integral, because they refer
to separable objects. As might be expected, subjects encounter severe
difficulties in evaluating the similarity of such compound stimuli (Eisler
& Knoppel, 1970). Which two objects are more similar: two circles of the
same size with different radii? or two circles of different sizes with identically oriented radii? Shepard (1964) noted that some subjects' judgments
were more affected by circle size, while others were more affected by the
orientation of the line, and he attributed these differences to attep.tion.
More important, he concluded that the pattern of similarity judgtnents
was intermediate between the pattern predicted by a "city-block" tnodel
(in which the "distance~' between stimuli corresponds to the sum ,of distances on two dimensions) and a pattern predicted by a Euclidean model
(in which distance is measured along the shortest path between two
points). Hyman and Well (1968). performed a similar experiment with
decidedly nonintegral attributes: the hue of one color chip and the
brightness of another. Their similarity judgment~ confolined to a cityblock model. Garner and Felfoldy (1970) used the stitI?uli of these two
experiments in their study of speeded classification, al1d they found no
interference between nonintegral dimensions. Thus, Garner (1970)
concluded that integral dimensions have three characteristics:
(1)
(2)
(3)
They lead to a Euclidean metric in direct distance scaling.

When correlated in a classification task, they yield a redundancy
gain.
When varied orthogonally they cause interference in classification
(E,geth & PacheIla, 1969; Garner, 19'69).
The last characteristic is not always found. Subjects can sometimes use
the redundancy of integral dimensions when it is available, but also
avoid interference when there is no redundancy (Felfoldy & Garner,
1971; Garner, 1970). However, the rules of integrality provide a useful
109
first approximation. In general, these findings confirm the conclusion that

the perceiver has little control over the initial stage of perceptual analysis, which determines the effective stimuli for subsequent stages. As Rabbitt (1971, p. 263) concluded: "... what is perceived as 'a stimulus'
depends critically on the organization of the nervous system rather than
on the arbitrary intuitions of experimenters who select and define component 'attributes' or 'dimensions' of display in terms of semantic convenience or ease of preparation of stimulus material."
RESPONSE CONFLICT: THE STROOP TEST
Man's ability to "switch off analyzers" or gate irrelevant attributes

can be studied by attaching conflicting responses to several attributes of
an object. In the test of selective efficiency, the subject is required to
respond to one of these attributes, and to ignore the others. A perfect
selection device would simply prevent the analysis of all irrelevant attributes, and thereby avoid response-conflict at the source, but man is
not endowed with such a device. Thus, Egeth's (1967) review of filtering
in speeded-classification tests concluded that subjects in such tasks can
ignore irrelevant stimulus attributes (Morin, Forrin & Archer, 1961; Fitts &
Biederman, 1965; Imai & Garner, 1965), but only when no conflicting
responses have been attached to these attributes (Montague, 1965). When
the competing responses are weak, interference may be slight -or absent
altogether (Well, 197.1), but when the responses are overlearned, some
interference always occurs.
Conflict between responses to differellt attributes of the same object has been extensively studied in an experimental situation devised
by Stroop (1935) (see p. 32). The stimulus materials consist of three
types of cards: On card W, the subject must read a set of color names;
the relevant attribute here is letter shape. On card C, he must name the
colors of a set of color patches. On card CW, he Illustname the colors
in which a set of words are printed. The relevant attribute is color, but
the words on card CW are themselves color names. Thus, the subject
may see the word "red" printed in orange, and he must respond
"orange." The most dramatic finding with this test is the difficulty of
card CW. Jensen and Rohwer (1966) describe the behavioral effects of
this card: subjects "... become more tense, they strain forward, they
take on the expression of eyestrain, they gesture with arms and hands,
and occasionally they stamp their feet. Exaggerated vocal emphasis is
also characteristic. . . . Repeated testing decreases these overt signs of
stress, though subjects never come to regard the CW task with the same
bored equanimity that they finally show toward cards C and W [p. 59]."
The relative difficulty of this task resists extended practice (Jensen,
110
1965). Subjects only improve by the adoption of such techniques as

squinting or deaccommodation to make the words illegible.
Interference in card CW occurs even when the words whose colors
are to be named are not themselves color names. Klein (1964) suggested
that it is harder to say the colors of any nameable symbol than it is to
say the colors of a set of asterisks, and that the amount of interference
follows a gradient of color-relatedness. For example, color-related words
such as "sky" or "lemon" tend to cause greater interference than nonsense syllables. Similarly, Morton (1969b) observed that it is difficult to
sort cards by the numerosity of the symbols printed on them, if these
symbols happen to be other digits. Fox, Shor, and Steinman (1971) replicated these results, and they .also reported interference when direction
names (Up, Down, Right, Left) appeared in incongruent positions.
Klein's conclusion that any nameable symbol will cause interference was not supported in several subsequent studies (Egeth, Blecker &
Kamlet, 1969; Pritchatt, 1968; Keele, 1972). It seems that the interference
effect occurs primarily when the printed word elicits a coding response
which is relevant to the task. Thus, the interference in the Stroop test
is a result of competition at the level of encoding: some responses are
"primed" by the task, and the elicitation of these responses by an irrelevant stimulus causes interference. It follows from this analysis that facilitation could be produced if the responses to the relevant and irrelevant
attributes are congruent. This was found to be true in a reaction-time
experiment, where the subject had to name the color of a tachistoscopically presented color word (Hintzman et al., 1972). There was interference when the word and the color were different, but facilitation when
they were the same. Similar results were obtained by Morton (1969c),
in a card-sorting task.
These results illustrate the concept of Response Readiness. Responses appear to be organized in sets. When a set of responses is relevant to the task (e.g., color names), these responses are readily elicited
even by inappropriate stimuli. Facilitation arises if the responses elicited
are compatible, and interference if the responses are competing.
In addition, the findings obtained in the Stroop paradigm strongly
support the general conclusion of this chapter: subjects cannot prevent
the perceptual analysis of irrelevant attributes ofan attended object.
REVIEW
When an object is perceived, many perceptual interpretations are

made, apparently in parallel. The various attributes of the perceived
object correspond to these interpretations. There is little evidence that
Attention to Attributes 111
"- an intention to attend to a particular dimension of experience can prevent

the perceptual interpretation of other dimensions. Attention to attributes
affects the post-perceptual stage of Response Selection by increasing the
readiness to produce codes of the relevant dimension (e.g., color words),
and the tendency to attach overt responses to such codes.
Studies of discrimination learning indicate that responses are often
attached to a single attribute even when several are relevant. The salience
of the attributes and the prior learning history of the organism determine
which of the attributes will control behavior. These observations, among
others, have suggested models in which analyzer systems can be
switched on or off, depending on circumstances. However,. the fact that
similar rules apply to concept learning with adult humans indicates that
discrimination learning can occur without perceptual modification. Similarly, the results in Kiilpe's paradigm are explained satisfactorily as an
effect of instructions on encoding. Studies in this paradigm have shown
that object coding is common and generally efficient.
Studies of speeded classification have shown that certain pairs of
attributes are integral. It is relatively difficult to ignore an irrelevant
attribute which is integral with the relevant attributes. Attributes of separate objects are not integral, but attributes of a single object often are.
Thus it is easy to ignore an irrelevant object but considerably harder to
ignore irrelevant attributes of an attended object. Results in several variants of the Stroop test confirm the conclusion that irrelevant dimensions
cannot be switched off at will. Responses associated with irrelevant attributes interfere most severely with performance if they belong to the
same set as the relevant responses.
>
7
Focused AttentionFindings and Theories
Debate about the nature of selective attention has centered on tasks that
require the subject to select inputs, or filter informatron. The classic
example of input selection is the situation that Cherry (1957, p. 278)
described as the cocktail-party problem: a guest at a cocktail party usually listens to one conversation and ignores all others, regardless of how
loud they may be. In general, a person is ,said to select inputs when he
focuses attention exclusively on stimuli that originate from a particular
source or share some other characteristic feature.
Experimental studies of input selection have. typically used auditory stimuli. Broadbent (1958) defended the choice of the auditory
modality for the study of attention on the grounds th~t auditory attention can be studied without the encumbrance of the orientation movements which dominate visual attention. When a medley of auditory
messages is fed through headphones, the listener must rely on central
selective mechanisms to isolate the relevant message and ignore the
others, whereas the selection of relevant visual stimuli is usually carried
out by eye movements. To obtain a pure measure of central processes
of visual selection, the experimenter is therefore compelled to present
brief stimuli which are removed before eye movements can occur. This
112
Focused Attention-Findings and Theories
113
tachistoscopic situation is exceedingly contrived, and the auditory task

with multiple messages is clearly more natural and ecologically representative. As this chapter will show, however, .the emphasis on audition
in the study of selective attention has limited the theoretical treatment
of the problem in several ways.
The performance of a listener who selectively attends to a relevant
message in the presence of an irrelevant message can be evaluated by two
sets of questions: (1) How effective is the processing of the relevant message? Is comprehension impaired relative to a control situation in which
that message is presented alone? (2) How effective is the rejection of the
irrelevant message? In what ways, and at what stages, are the selected
and rejected messages treated differently?
The first section of this chapter summarizes experimental findings
in studies of focused attention. Subsequent sections review several
theories that have been proposed to explain these findings.
EXPERIMENTAL STUDIES OF FOCUSED ATTENTION
Man's notable ability tores_ist distraction is a manifestation of selective attention. The success with which distraction can be resisted was
documented in a series of early studies reviewed by Woodworth (1938).
On the average, measures of intellectllal functions were barely impaired
by intense irrelevant stimulation. However, distraction is resisted ata
cost: motor tension and autonomic manifestations of arousal are higher
than normal. Thus, one is much more likely to break one's pencil while
Writing an examination in a noisy room than when the room is quiet. One
is rarely justified, however, in attributing failure in a test to the presence
of distracting !conditions.
In the early studies of distraction, the subject's attention was focused on his mental activities, but modern studies of selection typically
deal with the ability to select a relevant input in the presence of others.
Many studies have used the shadowing task, in which the listener follows
a message by repeating every word, and attempts to ignore other messages to which he is simultaneollsly exposed. Cherry (1953; Cherry &
Taylor, 1954) established that the presence of a distracting message
barely impairs shadowing performance when the rejected and attended
messages are distinguished by an obvious physical characteristic, such as
spatial origin. In some of these experiments Cherry used the method of
dichotic presentation, in which two messages are presented by earphones
to different ears. He observed that subjects are always aware of the
presence of the rejected message on the unattended ear, but know
114
virtually nothing about it when subsequently questioned, not even the

guage in which it was spoken. They also fail to detect a switch to
verted speech on the rejected channel. However, they are
aware of the sex of the voice on the rejected channel and easily ....."'....... _.
any major physical change, such as a change of voice, a switch from
voice to tone, or an isolated sound (Lawson, 1966; Treisman & Riley,
1969). Shadowing is most effective when both the relevant and the distracting stimuli are unambiguously labeled. Treisman (1964b) found
marked interference when a subject shadowed one continuous message
while simultaneously exposed to two distracting messages from different
sources.
Spatial position is the most effective attribute for identifying the
selected message. It is relatively easy to attend to a position, both with
auditory stimuli (e.g., Poulton, 1953; Spieth, Curtis & Webster, 1954;
Treisman, 1964b) and in tachistoscopic visual presentation (e.g., Sperling,
1960). In the case of audition, selection by location can be precluded
by presenting several messages which originate from the same position.
When several messages are presented in this manner, subjects are able to
isolate the relevant message 'by its pitch or loudness, but only with great
difficulty (Treisman, 1964b).
Selection of inputs can be almost perfectly effective when guided
by an appropriate cue. This is true both in th~ shadowing situation and
in other tasks. In tachistoscopic present~tions of complex arrays, for example, the subject can be set to select items in a particular row (Sperling, 1960) or items of a particular color (von Wright, 1968, 1970), and he
performs almost as well as if the irrelevant material had not been present. Monitoring an auditory message for critical items is almost as effective in the presence of a competing message to the other ear as without
that message (Moray & O'Brien, 1967), and the covert rehearsal of a
memorized list is barely affected by the presentation of loud rhythmic
music (Kahneman, 1970).
Selection is effective only when the relevant and irrelevant inputs
differ in obvious physical characteristics. In Broadbent's terms, selection
by stimulus set is effective, selection by response set is not. Thus, the
relevant items in a tachistoscopic presentation can easily be selected
by spatial location, but it is essentially impossible to selectively attend
to the digits in a brief exposure of a mixed array of digits and letters
(Sperling, 1960). Similarly, it is exceedingly difficult to isolate ~n auditory
message in English from a simultaneous message spoke'n by the same
voice in French (Treisman, 1964a).
Recent studies of auditory attention have used tasks other than
shadowing. In the monitoring task, the subject is exposed to a continuous message but responds only to occasional target items. Monitoring
115
a list of letters for occasional digits is not seriously impaired by the presentation of an irrelevant message to the other ear (Moray & O'Brien,
1967; Underwood & Moray, 1972). Similarly, a subject instructed to press
a key as soon as he hears an animal name in a recorded message responds as fast in the presence of an irrelevant message as when that
message is absent (Ninio & Kahneman, 1973).
The main difference between shadowing and monitoring is that
the former task requires continuous overt responses, while the latter does
not. Several experiments in my laboratory have investigated a recognition
task that requires no immediate response (Henik, 1972; Kahneman, 1970;
Levy, 1971). Two messages, each consisting of 31 unrelated words, are
presented dichotically and the subject subsequently attempts to recognize some of the words that were presented to the right ear. The recognition choices also include an equal number (eight) of words presented
to the left ear, and of words that were not presented at all. The critical feature of the design is that the subject is penalized for recognizing words
that were presented to his left ear.
Three experiments using a fast rate of presentation (two words/ sec
in each ear) compared recognition in focused attention and in a control
condition where the relevant message was presented alone. The presentation of an irrelevant message caused a decrement in the recognition of .
relevant items (from 61 percent to 54 percent). The percentage of false
recognitions of unpresented words was 32 percent in both conditions,
and 37 percent of the left-ear words were judged to be familiar. Thus,
selectivity was high, though far from perfect. Selectivity was only slightly
poorer ata slower rate of presentation (one word pair/ 1.5 sec). Finally,
subjects were able to prevent a high rate of intrusions even when the
words were presented to the two ears in alternation at the comfortable
rate of one word every .75 seconds. The results show that a listener
could usually refrain from paying attention to the irrelevant items, even
when no relevant word was presented at the same time. However, intrusions of left-ear items were more frequent in an experiment (Henik,
1972) where only a few such words were presented. It appears that the
continuity of the irrelevant message is important in permitting that
message to be ignored.
There is no doubt that selective attention was less effective in our
recognition. experiments than in studies of shadowing. In one of these
studies subjects who shadowed a message on one ear later failed to
recognize a phrase that had been repeatedly presented to the other ear
(Moray, 1959). This result was obtained although the subjects were not
specifically enjoined not to listen to the irrelevant ear. In our studies sub~
jects were penalized for listening to the left ear but they nevertheless did
so occasionally. The difference is due in part to the shadower's own voice,
116
which functions as a source of interference (Underwood & Moray, 1972).

In addition, limitations of capacity are probably involved. Shadowing is
more demanding than monitoring, and it leaves less spare capacity to be
captured by the irrelevant message.
An ingenious study.by Zelnicker (1971) provides further evidence
for the role of capacity demands in focused attention. Three groups of
four auditory digits were presented in rapid succession (e.g., 3256-81296543). There were two experimental conditions, which may be labeled
Easy and Hard. In the Easy condition the subject repeated the first
group of digits twice (3256-3256), synchronizing his responses with the
second and third groups heard on the tape. In the Hard condition he
repeated the first group while hearing the second, and he repeated the
second while hearing the third. The correct response in the example would
be: 3256-8129. In that condition, it was necessary to say 3256 while listening to 8129, which was the set to be reported later.
In both conditions the subject was also exposed to a playback of
his own voice, which was delayed by 0.2 seconds. Such delayed auditory
feedback (DAF) often causes stuttering. The amount of st~ttering was
compared in the first group of digits that the subject reported. (3256, in
both conditions). There was less stuttering in the 'Hard condition. Attempting to listen to the second group of digits while speaking made it
easier to ignore the DAF.Since DAF is an extremely unpleasant experience, the subjects must have been motivated to ignore it under both
experimental conditions. It is consistent with a notion of limited capacity
that they were more successful when engaged in a demanding task.
The evidence reviewed thus far is generally consistent with predictions from Broadbent's (1957a, 1~58) filter theory. The theory assumes
that a filter sorts simultaneous stimuli by obvious physical characteristics,
such as position, voice quality, or. color. Further perceptual analyses are
applied only to stimuli which share the property that defines the relevant
"channel" or message, e.g., words presented to the right ear or letters
printed in blue. Other stimuli are rejected and filtered out. Irrelevant
sensory information is stored momentarily as an "unanalyzed tape recording" (Treisman, 1969), but is permanently lost unless a shift of the
filter retrieves it from sensory storage..Thus, the material presented to an
irrelevant channel. is not analyzed in perception, beyond a few tests on
physical features. Specifically, filter theory implies that speech messages
on an irrelevant channel are not analyzed as speech.
Strong evidence was advanced against filter theory soon after it
was formulated. Thus, although the theory accounts for the cocktailparty phenomenon of selective attention, it fails to explain another common experience of cocktail parties: the detection of one's own name as
117
soon as it is mentioned in an otherwise ignored conversation. Moray

(1959) documented this everyday experience in the shadowing situation.
He observed that subjects were much more likely to notice a message on
the rejected ear if it was preceded by their own name than if it was not.
Moray's results are incompatible with Broadbent's assumption that the
sounds arriving at the rejected ear are not analyzed as speech.
Neisser (1969) developed a visual analogue to the auditory shadowing situation, and he obtained results very similar to Moray's. He required subjects to read coherent text aloud and to ignore words printed
in red under each line of the selected text. Subjects can do this very well.
The situation is similar to ordinary reading, where the lines just above
and below the attended line do not intrude. Neisser also showed that
subjects do not recognize the words presented on the ignored lines, even
when the same word is repeated several times'. Two-thirds of his subJects, however, noticed their own name on a rejected line.
There is much additional evidence that, even in the shadowing situation, the message on the rejected ear is analyzed as speech. Treisman
(1960) occasionally switched messages from one ear to the other, usually at
a point of high redundancy in the relevant shadowed message. On a
substantial number of instances, subjects followed the attended message
into the incorrect ear for one or two words before reverting to the
designated ear. Such transitions were most likely to occur if the shadowed message was connected prose. Most of her subjects were unaware
of their transition errors. Treisman's results demonstrate that continuity
of meaning can briefly overcome the effect of channel selection in determining the subject's shadowing. response. These finding~ are incompatiblewith Broadbent's early version of filter theory, because they show
that the message to the neglected ear is not necessarily rejected at an
early stage of processing.
Another experiment of Treisman's (1964c) demonstrates an important effect of selective attention. and an important difficulty for filter
theory. She studied a situation originally devised by Cherry (1953), in
which identical messages, one lagging behind the other, are presented
on the two ears. The subject is to shadow what he hears on one ear, and
he is not told that the two messages are actually identical. The lag between the messages is gradually redu~ed until the subject comments on
their identity. Treisman (1964c) repeated and extended these observations. She found that subjects recognize the identity of the two messages
when the lag is about five seconds, but ,only if the relevant message
leads. When the neglected message leads, identity is recognized only at
an interval of one or two seconds. These results show that the trace of
the shadowed message persists longer than that of the rejected message,
118
just as filter theory would predict. Contrary to filter theory, however, the
rejected message is apparently analyzed as speech: subjects realize the
identity of messages even when they are spoken in different voices; and
bilingual listeners often recognize the identity of a message and its translation. Neither result should occur if the rejected message is not analyzed
as speech. At the very least, these findings show that some verbal analysis of the rejected message sometimes occurs.
There is additional support for this conclusion. Lewis (1970) recorded latencies for shadowing unrelated words and found that the shadowing latency for a word is significantly increased by simultaneously
presenting its synonym to the other ear. Evidently both words must be
recognized for this effect to occur. However, Treisman (unpublished)
observed that this synonym effect occurs only at the beginning of the
message. Selectivity improves within a few seconds and the content
of the irrelevant message no longer affects shadowing latency. An intriguing result was reported by Corteen and Wood (1972). They first
associated an electric shock to the presentation of city names in a word
list. Later, city names which were included in the rejected message in a
dichotic shadowing task often elicited a galvanic skin response, although
they were never consciously identified and did not interfere with the
shadowing performance.
Selectivity with auditory stimuli appears to be generally poor when
the messages are brief. Thus, Brown (1970) instructed subjects to attend
to one ear and then presented a single dichotic pair of words. Precuing
the relevant ear did not improve the subject's ability to recognize a word
presented on that ear. With somewhat longer messages, however, such
precuing is very helpful (Broadbent, 1952; Spieth, Curtis & Webster,
1954). These results indicate that focusing attention takes time.
Greenwald (19'70a, b) described another instance of a failure to
filter a very brief message. He simultaneously presented a visual and an
auditory digit and recorded subjects' reaction times for reading the visual digit. The subjects were unable to reject the irrelevant auditory digit;
their RT was slower when this digit was not the same as the visual digit.
Greenwald also reported an important interaction between the modality
of the interfering stimulus and the modality of the response: interference
from the auditory item was more severe when the subject had to say the
visual digit than when he wrote it (Greenwald, 1970a,.1970c).
In response to the suggestion that the failures of selection in his
experiment were due to the brevity of the messages, Greenwald (1970b)
showed that a spoken digit delays RT to a relevant visual digit even
with successive stimuli presented at the rapid rate of one item/ second.
However, this serial RT task cannot be considered a truly continuous
119
performance, since it is carried out in a series of discrete, speeded acts.

Perhaps selection becomes most effective only when the primary task is
coherent, i.e., involves preview of future stimuli and serial grouping of
both stimuli and responses (Kahneman, 1970).
Gopher and Kahneman (1971) have documented the importance of
a distinction between reorientation and maintenance of attention in auditory monitoring. Monitoring a continuous list of words for the occasional
occurrence of digits is an easy task even in the presence of a competing
message to the other ear. It is also easy to report a short list of digits
that is presented to one ear, and ignore digits presented to the other ear.
Both tasks are combined in our experiments. The subjects first monitor
one of two dichotic lists of words and digits for several seconds, reporting the digits heard on the relevant ear, then they hear a cue which defines the relevant ear for the second part of the task. Shortly after that
cue, short lists of digits are presented to the two ears. The reorientation
of attention after a period of selective listening is quite difficult. Subjects are prone to intrusions and confusions for a few seconds after the
reorientation cue. There are pronounced individual differences in the
rate of these errors. The lability of selective attention after a reorientation cue is negatively correlated with the proficiency of military pilots
(Gopher & Kahneman, 1971) and with the safety record of bus drivers
(Kahneman, Ben-Ishai & Lotan, 1973), while the rate of errors in steadystate monitoring is consistently less valid as a predictor of the same
criteria.
These observations indicate that it takes some time to change from
one selective set to another. Gopher (1971), in a study of eye movements,
confirmed the importance of this distinction between orientation (from a
neutral, uncommitted state) and reorientation. Reorientation was accompanied by a much larger eye movement than was the initial adoption of an orientation.
In summary, although the selection of inputs is highly effective,
it is imperfect. A relevant input on which attention is focused can be
processed effectively even in the presence of irrelevant stimulation..
However, focusing attention on one message does not completely prevent the processing of stimuli on irrelevant channels. There is much evidence that at least some of these stimuli are analyzed for content. Thus,
a stimulus for which there is high readiness will probably be recognized.
In addition, any obvious change on any sensory channel will be detected.
A few seconds are apparently required for the focusing of auditory attention to become fully effective.
A brief survey of the main theories advanced to account for these
facts will now be presented.
120
BROADBENT'S FILTER THEORY
Broadbent's filter theory is the natllral starting point for any discussion of modern theories of attention. Some of the main features of
this theory have already been noted, as well as some of the evidence
that shows it to be inadequate. Briefly, Broadbent assumed a sequence
of three elements: a short-term store (S-system), -a selective filter, and a
limited capacity challnel (P"7system). Concurrent stimuli enter into the
S-system in parallel, and they are analyzed there for physical features,
such as location or tonal quality. There is no definite limit on the capacity of the S-system. The selective filter allows those ~timuli that arrive
on a designated "channel" into the P-system. A channel is defined by any
physical characteristic for which the filter can be set. Thus, location or
pitch could both define a channel in audition. Color or size could define
a channel in vision.
More elaborate perceptual analyses are carried out in the P-system.
This system deals serially with accepted stimuli,. and the time spent on
each stimulus depends on the amount of information that the stimulus
conveys. When the P~system has cleared, the filter allows a new stimulus
to enter. Thus, when two stimuli are presented simultaneously, .they can
be handled successively, but only if the processing of the first is completed before the record of the other in the S-system has decayed. This
feature of Broadbent's theory explains the common experience of the
"double take," in which one returns to a stimulus that was ignored or not
fully processed at the instant of its presentation. Such is the experience
of the husband, deeply engrossed in his paper, who first exclaims,
"What?" and then, without waiting for an answer, goes on to say, "No,
I'm not hungry," as he retrieves his wife's query from an -echoic memory.
Filter theory interprets focused attention as setting the filter -to
select a certain class of stimuli' and to reject all others. Irrelevant messages are simply allowed to decay in the S-systemwithout undergoing
more advanced processing in the P-system. Therefore, attention is most
effectively focused by a stimulus set, in which the relevant stimuli are
distinguished by one of the simple operations that the filter can perform, e.g., discriminations of location, pitch, and speech-like quality' in
sounds. Selection is difficult or impossible in the absence of a clear physical distinction between relevant and irrelevant stimuli. Filter theory is
supported by the finding that subjects cannot focus attention solely on
digits when a mixed array of digits and letters is briefly presented (Sperling, 1960). Similarly, bilingual subjects cannot separate a message in
English from a simultaneous message in French if the two messages are
121
spoken in the same voice and originate at the same location (Treisman,
1964a). Selection by semantic class, or by language, requires the .subject
to adopt a response set (Broadbent, 1970, 1971), because the relevant
items are defined by a common set of responses rather than by a common stimulus feature. Although Broadbent's (1970) elaboration of his
original theory acknowledged that selection by response set is sometimes
possible, he presented evidence that response set is generally much less
effective than stimulus set.
Filter theory implies that attention cannot be divided, because the
P-system performs no' parallel processi;ng of discrete stimuli. According.
to the theory, the apparent division of attention in the performance of
concurrent activities is mediated by alternation between channels or
between acts, and the rate of alternation is slow. Broadbent (1958) assumed that the minimum dwell-time of the filter is about 300-500 milliseconds. The processing of simultaneous complex messages fails when
the processing of the first message which enters the P-system is so prolonged that the traces of the other message decay in the S-system before
they can be retrieved.
As initially stated, filter theory was wrong. It will be shown in
Chapter 8 that parallel processing of simultaneous stimuli does occur in
divided attention. Furthermore, the evidence of the preceding section
demonstrates that the content of an irrelevant message is identified, at
least dimly and at least some of the. time, even when the subject attempts to ignore it. Finally, the idea of a slow-moving filter that selects
one stimulus at a time is not viable. Thus, virtually all the predictions of
filter theory about what people cannot do have been disproved. However, filter theory provides a useful approximation to what people usually do. In addition, it has the unique distinction among attention
theories of being sufficiently precise to be definitely disproved.
Many of the terms and concepts of filter theory have been widely
applied. In particular, the image of filtering as an operation that opens
one channel and closes others has been very influential. This image,
however, was derived from the study of auditory attention and of the
dichotic case in particular. It is not easily applied to visual attention. For
example,. what defines the channel selected when one reads a book?
The analysis of attention presented in Chapter 5 proposes the concept
of perceptual unjt,or group, as an alternative to the concept of channel.
Another influential idea of filter theory was the concept of a preperceptual memory (the S-system). The temporary storage of unanalyzed
sensory information has acquired many names from numerous investigators. Sperling (1960) spoke of a visual image, which he later (Sperling,
1963) renamed Visual Information Storage (VIS), and to which he added
an Auditory Information Storage (AIS) (Sperling, 1967). Crowder and
122
ATTENTION. AND EFFORT
Morton (1969) and Morton (1970a) described a Precategorical Acoustic

Storage (PAS), and Neisser (1967) introduced the terms echoic and
iconic memory for the auditory and visual stores.
There is general agreement, however, that the precategorical or
iconic stores must be distinguished from various forms of post-perceptual
short-term memory (e.g., Atkinson & Shiffrin, 1968; Broadbent, 1971). In
addition, the basic assumption that unanalyzed material can be stored
for several seconds has been questioned (Massaro, 1972; Norman,
1969b). Thus, Norman (1969b) required subjects to shadow one message
of a dichotic pair and tested their memory for items presented in
the relevant and the irrelevant messages immediately before the interruption of shadowing. There were no important differences between the
retention of relevant and irrelevant items, and Norman inferred that both
classes of items have access to the same systems of post-perceptual
memory. The relation between attention and memory will be discussed
again in the next chapter.
TREISMAN'S FILTER-ATTENUATION THEORY
In an attempt to accommodate the evidence against filter theory,

Treisman (1960; 1964d) proposed a modification of that theory which
Broadbent (Broadbent & Gregory, 1964) subsequently accepted. The
modification was simply that filtering is not all-or-none: the rejected
message is merely attenuated, not eradicated.
According to Treisman (1960), a sensory message activates hypothetical "dictionary units" in memory. Each unit has a threshold which must
be exceeded for perception to, occur. The thresholds for highly significant stimuli, such as one's name, are permanently lowered. The threshold
for a word which the context makes probable is lowered temporarily. Because of these variations of .thresholds, a word of high significance or
high probability which is presented in an irrelevant channel can be perceived in spite of attenuation. The assumption of lowered thresholds for
Significant stimuli was intended to explain Moray's (1959) discovery that
subjects often respond to their name spoken on one ear while they
shadow a message on the other ear. Temporary alterations of threshold
explain the effects of context on the recognition of degraded stimuli
(Morton, 1969b; Tulving & Gold, 1963), and also explain Treisman's
(1960) finding that shadowing subjects occasionally follow the content
of a message which is suddenly switched from one ear to the other. In
the terms of Signal-detection theory, these effects are mediated by rapid
and short-lived criterion changes.
Treisman's modification of filter theory retained the essential idea
123
that attended and unattended stimuli are treated differentially from a very
early stage of perceptual analysis. This differential treatment causes a reduction of sensitivity (d') for unattended stimuli. In general, unattended
items do not activate the corresponding dictionary units, ~xcept when the
threshold of one of these units is exceptionally low.
Treisman (1969) later presented a more inclusive treatment of the
entire field of selective attention. Two observations were basic to that
theory: (1) people can easily focus attention on one input (e.g., the voice
on the right), while the.y have great difficulty in dividing attention between two inputs; and (2) people can easily divide their attention between the various aspects or attributes of a particular input (La Berge &
Winokur, 1965; Lappin, 1967), but they encounter great difficulty in focusing on one aspect of a stimulus and ignoring the others (Stroop, 1935;
Treisman & Fearnley, 1969).
As was mentioned in the preceding chapter, Treisman (1969) proposed that a single input can be processed by several analyzers in parallel,
while the processing of two inputs by the same analyzer is necessarily serial. In a major departure from filter theory, she concluded that
divided attention and parallel processing are possible for two simultaneous inputs, but only if they do not reach the same analyzers. Serial
processing is mandatory, however, whenever a single analyzer must
operate on two inputs.
The main implication of this new theoretical idea concerns divided
attention: unlike filter theory, Treisman's analyzer theory permits parallel processing, e.g., of information presented to different modalities. This
issue will be considered in detail in the next chapter.
Treisman (1969) retained the filter-attenuation approach to focused
attention. She used the concept of analyzer only to explain why any major physical change in the characteristics of a rejected message is invariably recognized (Lawson, 1966; Treisman & Riley, 1969). Such a stimulus
is easily detected because it reaches analyzers that are not occupied by
the relevant message.
THE DEUTSCH-NoRMAN THEORY
An important alternative. to filter-attenuation theory was formulated

by Deutsch and Deutsch (1963). The evidence that had led Treisman to
a moderate revision of Broadbent's theory brought Deutsch and Deutsch
to the nlore radical conclusion that "a message will reach the same perceptual and discriminatory mechanisms whether attention is paid to it
or not [p. 83]." They postulated central structures, equivalent to Treisman's dictionary units, but proposed that attention does not affect the
124
degree to which these structures are activated by s.ensory stimulation.

However, each central structure has a preset weighting of importance,
which reflects momentary intentions (e.g., animal names are now relevant) or enduring dispositions (e.g., my own name is always relevant).
Among concurrently active central structures the one with the highest
weighting of importance is selected to control awareness and response.
In the terms of signal-detection theory, the importance parameter is a
criterion bias favoring the relevant items.
The Deutsch and Deutsch theory locates the transition from parallel to serial processing closer to the ultimate response than does filter
theory (see Fig. I-Ion p. 6). The distinction between the tp.eories is
sharpest in the context of divided attention. Filter theory asserts that
division of attention among concurrent stimuli is simply impossible,
since attention can only be directed to one channel at a time. Deutsch
and Deutsch, on the other hand, imply that detection of a relevant signal
should be easy wllether or not the observer is currently attending to the
channel on which the signal is presented. As will be shown in the next
chapter, this prediction is not confirmed.
An obvious deficiency of the formulation proposed by Deutsch and
Deutsch (1963) is its failure to account for the facts of focused attention
which filter theory w.as designed to explain. They assumed a" system that
can be preset in advance to favor the recognition of certain stimuli,
such as animal names. However, such a system cannot be preset in
advance to favor words that ,viII be heard on the right ear, since it has
no knowledge of what those words will be. It can only favor a right-ear
word after all concurrent stimuli have activated the central structures to
which they correspond. Thus, the process of selection by stimulus features appears to be more complex than selection by response class. The
added complexity should probably make selection by phYSical features
relatively difficult. The evidence of focused attention, however, indicates
that stimulus set is far more efficient than response set.
Norman (1968) attempted to reformulate the Deutsch and Deutsch
theory to overcome this deficiency. He assumed central units which accept two types of inputs: (1) sensory inputs; and (2) pertinence inputs.
The latter are equivalent to the importance weighting proposed by
Deutsch and Deutsch. The magnitude of the pertinence input reflects
the criterion level for the elicitation of activity in each central unit. At
any moment of time the unit with the highest total of sensory and pertinence inputs dominates .perception, awareness, and memory.
Norman (1968) explained the operation of stimulus set by assuming
that the activation of a recognition unit is a gradual and recursive process. A central unit which is activated by a stimulus on the relevant channel "knows" this fact at an early stage in the process of recognition, and
Focused Attention---Findings and Theories
125
this information causes the pertinence of the unit to increase. With this
assumption of recursiveness, Norman's theory explained why stimulus set
need . not be substantially more difficult than response set. It still failed
to explain, however, why stimulus set is actually easier.
Norman (1968, p. 528) emphasized the contrast between his view
and Treisman'sfilter-attenuation theory. Both theories account for the
effects of context and word significance in selection by criterion bias.
However, Norman also explains filtering as a criterion effect, whereas
Treisman implies that discriminability (d') is reduced for items rejected
by the filter.
An experiment by Mor(ly and O'Brien (1967) appears to provide a
test of ~orman's predictions. Subjects were exposed to a dichotic message consisting of letters and digits; they were to attend only to the- right
ear, and to press a key with the right hand whenever they heard a letter
on that ear. Although instructed to ignore the message on the left ear,
they were to tap a key with the left hand whenever they happened to
hear a letter on that ear. The signal-detection analysis of the results was
not entirely conclusive, because of the very low false alarm rate, but it
suggested that the criterion for left-hand responses was lower than the
criterion, for right-hand responses. That is, the number of false alarms on
the irrelevant channel was greater than Norman's theory wouldperdict.
In addition, d' was much lower on that channel. Other experiments (e.g.,
Broadbent & Gregory, 1963; Kahneman, Beatty & Pollack, 1967) have
also supported the conclusion that selective attention affects discriminability, contrary to the position of Deutsch and Deutsch and Norman.
One could perhaps attempt to dismiss these results by invoking a
distinction between two types of criterion effects, which operate respectively on recognition and on the overt response. In the experiment by
Moray and O'Brien, for example, the criterion for recognizing irrelevant
words on the left ear could be high (low pertinence), while the criterion
for making responses with the left hand could be low (careless responses). This distinction has some intuitive appeal. If it is accepted,
however, the claim that pertinence affects the criterion is robbed of any
operational consequences.
NEISSER AND HOCHBERG
Other alternatives to filter-attenuation theory have been proposed

by Neisser (1967, 1969) and Hochberg (1970). Neisser's (1967) important
text generalized to all areas of perception a theory originally developed
to account for the perception of speech (Liberman, Cooper, Shankweiler
& Studdert-Kennedy, 1967). According to Neisser's theory, perception is
126
an active process of analysis by synthesis. Thus, one understands a spoken

message by covertly reproducing it, and visual percepts are produced
by a similar activity of synthesis. Perception is an act of construction, and the role of attention is to select the percepts that will be constructed or synthesized. "On this hypothesis, to 'follow' one conversation
in preference to others is to synthesize a series of linguistic units which
match it successfully. Irrelevant, unattended streams of speech are
neither filtered out nor attenuated; they fail to enjoy the benefits of
analysis by synthesis [Neisser, 1967, p. 213]."
Elsewhere, Neisser (1969) summarized his point of view by an image: "If a man picks up a sandwich from a dozen offered to him on a
tray we do not ordinarily say that he has blocked or attenuated the
others; he simply hasn't picked them up. Naturally he finds out a good
deal more about the one he has selected, because he must shape his hand
to fit it, to hold it together and so on." In addition, ". . . we might think
of him keeping his fingers lightly on the other sandwiches, both before
and during his activities with the one he selects, to make sure that
nothing untoward is going on." The two passages illustrate the essential
point that there is no evidence for the negative view of attention implied
by the concept of filtering. Selective attention consists of the allocation
of a limited capacity to the processing of chosen stimuli and, to the
preparation of chosen responses. '
In addition to the active process of analysis by synthesis, Neisser
assumed the existence of passive systems to perform a preliminary sorting and organization of sensory data. These are "silent" systems whose
operation is not represented in awareness. They are responsible for
grouping and localization and they routinely watch for critical features
of stimulation that may require a redirection of focal attention. The sudden motion of an object is such a feature, and the responsiveness to it is
probably innate. In addition, special tests are constructed to detect significant and recurrent stimuli, such as the listener's own name. It is worth
noting that the stimuli which most easily redirect focal attention are also
those which reliably elicit an orienting response (Lynn, 1966; Sokolov,
1963).
Neisser's theory provides an adequate account of focused attention.
It implies a process which selects the relevant stimuli that deserve the effort of perceptual synthesis. Although Neisser objected to the image of
a filter, the selection of messages for synthesis is u-ndistinguishable from
the operation of a filter. His theory attributes the effects of significance
and context to the role of expectations in the process of synthesis, and
it assumes a crude and global analysis of rejected messages. Thus, there
seem to be no predictions to separate 'Neisser's view from Treisman's attenuation theory. Indeed, the single difficulty which Neisser conceded in
127
comparing his theory to Treisman's could easily be avoided by a slight

reformulation of his position.
Treisman (1964b) had found that a subject can shadow a message
to the right ear more easily in the presence of a single competing message to the left ear than with two competing messages, one on the left
and one heard in the middle of the head. Furthermore, a pair of interfering messages caused less interference when they were superimposed on
a single channel than when they were presented on distinct channels.
After describing these findings,Neisser (1967, p. 217) wrote: "While
the filter theory can probably accommodate this result rather comfortably,
I would not have predicted it from considerations of analysis-by-synthesis. If unattended messages are simply remaining unsynthesized, it is
not obvious why a spatial separation between them should make a difference of any kind." In fact, this finding poses no difficulty for Neisser's
theory. The theory implies that the effectiveness of selective attention depends on the ability of the pre-attentive mechanisms to segregate the
relevant from the irrelevant messages. It is plaUSible that this task is
more difficult when there are two distinct irrelevant messages than when
there is only one. On the other hand, two messages originating in the
same location are heard as noisy gibberish, which is easily distinguished
from the relevant message.
Neisser's theory elegantly dismisses the issue of perception versus response by the simple assertion that the two are undistinguishable, because
perception is enactive. In addition, it suggests the interesting possibility that pre-attentive processes and focal processes may follow different rules. The work of Beck (1972; Beck & Ambler, 1972) supports this
idea by showing that the relative difficulty of discrimination problems
may change in different states of attention (see above, p. 74). The distinction between pre-attentive and focal processes may be related to a
distinction recently proposed between two functional visual systems:
an orientation system concerned with the perception of space and with
the detection of significant events in the periphery of the field; and a
central system concerned with fine discriminations (Held, 1968; Ingle,
1967; Schneider, 1967; Trevarthen, 1968). PhYSiological and comparative
analyses of visual function in various animals provide much support for
this distinction.
There remains a significant difficulty in Neisser's treatment of focal
and pre-attentive processes. He identified detailed perceptual analysiS
with focal attelltion, and focal attention with awareness. This is implausible, since cO'mplex psychomotor skills, such as driving, are often performed with little awareness, although they certainly require detailed
perceptual analysis.
Hochberg (1970) presented a similar view of selective attention
128
which could avoid this difficulty. He described perception as the confirmation of a changing set of expectations, concerning future phonemes
when one listens to speech, or the foveal image that would be produced
by possible movements of the eye when one looks at a picture. He also
assumed that the perceiver normally stores in memory only sets of expectations that have been confirmed. Stimuli that are not matched to
prior expectations are very rapidly forgotten, unless they are exceptionally salient. An intention to focus attention on one message causes detailed expectations to be produced for that message alone. Irrelevant
messages are not expected in detail, and are forgotten almost as soon as
they are heard. The production of expectations, of course, is very similar
to Neisser's active synthesis.
Hochberg's approach is similar to Neisser's, but he implies a separation of detailed perceptual analysis from awareness. Detailed perception depends on the generation of confirmed expectations, but awareness
of what one perceives also depends on whether the results of perceptual
analysis are stored inrpemory. If a stimulus is anticipated, but immediately forgotten, there will be no awareness although perception may be
detailed. Thus, Hochberg would probably describe driving as a case of
detailed anticipation with immediate forgetting. This description appears
more appropriate than a statement that driving is controlled by' crude
and global mechanisms, as implied by Neisser's theory. Another heuristic
advantage of Hochberg's formulation over Neisser's is that the concept
of expectation is more readily translated into the language of signal-detection theory than analysis-by-synthesis. However, Treisman (personal
communication) has observed that detailed expectations cannot be quite
as important as Hochberg's treatment would suggest. Thus, it is possible
to shadow a message even if it consists of unrelated words, precluding
the formation of expectations.
It may be noted that Freud's analysis of attention in the famous
seventh chapter of The Interpretation of D'reams was somewhat similar
to Hochberg's proposal. Freud discussed the attachment of attentioncathexis to objects of perception or to objects of thought, and the hypercathexis that allows them into consciousness. Freud adopted a positive
view of focused attention, in which selective attention is the active
elaboration of chosen ideas, rather than the inhibition of others (Freud,
1900; Rapaport, 1967; Schwartz & Schiller, 1967,1970).
An important notion in Freud's view was that the total quantity of
attention cathexis available at anyone time is limited, and that the
amount of attention demanded by an object of thought or perception
depends on how it is elaborated in cognitive activity. This view implies
that the limitation on what man can perceive depends on how he perceives, and on what he does" with his percepts. Freud's theory of attention was an effort theory.
129
THEORETICAL SYNTHESIS
In the present section the analysis of attention that was developed

in Chapter 5 (see Fig. 5-1) is reviewed and related to the theories discussed in preceding sections.
Unit Formation
The array of stimulation is sorted into integral units, which maintain their identity through subsequent stages of perceptual analysis. Subsequent operations are applied to these units: units are allocated capacity
at the stage of figural emphasis, and units or features of units activate the recognition stage. An operation at one of these later stages can
fail because the earlier grouping stage did not isolate the relevant unit.
The suffix effect, which will be discussed below, is an example of a
failure of selection due to grouping.
The idea of an initial grouping stage is adopted from Neisser's notion of pre-attentive mechanisms. It applies both to vision and audition.
For example, letters printed in red may form a natural group within a
larger array of letters printed in black. Similarly, a phrase spoken by a
particular voice and originating in a particular location constitutes a natural auditory unit in the cocktail-party situation. In both vision and
audition location in space is the primary determinant of unit formation:
sounds that originate in a particular location tend to be grouped, as do
clustered visual objects.
According to this analysis, attention is focused by selecting among
available perceptual units (objects or events) those units to which most
capacity should be allocated. By measuring grouping, we may be able to
predict the outcome of selective attention (Beck & Ambler, 1972; von
Wright, 1968, 1970; Williams, 1966). Furthermore, a careful study of the
laws of unit formation is needed to overcome a serious weakness of filter
theory: its failure to explain why certain physical features of stimuli are
effective in defining "channels," while others are not.
Figural Emphasis
Capacity is allocated in graded fashion to various groups. The
frequent demonstrations that selective atteation usually results in attenuation rather than in total blocking suggest that figural selection is not allor-none. Parallel processing of different units is possible, but perception
draws ona common pool of capacity, and the ability to carry out detailed
analyses of several units is limited.
130
Broadbent's theory assigns the functions of both grouping and selection to the filter. However, it appears essential to separate these functions, since they follow different rules. A major difference is the degree
of voluntary control over the two stages: with rare exceptions, unit formation is largely controlled by involuntary and psychologically silent
processes, while the allocation of capacity is immediately responsive to
momentary intentions.
Although the allocation of capacity is generally effective, it is not perfect. Some capacity is allocated to the processing of irrelevant stimuli,
and the processing of a selected stimulus is rarely as effective in the
presence of other stimulation as when the same stimulus is shown alone.
Recognition and Interpretation

The present treatment sllggests a distinction between an early stage
at which sensory information makes contact with recognition units, and a
subsequent stage at which a coherent set of interpretations is selected
for some of the objects in the field. It was assumed that the interpretation stage has a threshold. If no recognition unit is sufficiently activated,
there may be no interpretation for a particular object. The hypothesis
that figural emphasis controls the quality of the input to recognition units
implies that most items on an irrelevant "channel" will remain uninterpreted. However, if the readiness for a particular item is particularly
high, that item is likely to be consciously recognized even when it was
not initially favored at the stage of figural selection. Hearing one's name
mentioned in a neighboring conversation is an example. In addition, the
activation of a recognition unit can have behavioral consequences even
when it does not yield a perceptual interpretation. Thus, significant
words sometimes elicit emotional responses without attracting attention
and without gaining conscious interpretation (Corteen & Wood, 1972).
Responses to the Rejected Channel

According to the theory of effort presented in Chapter 2, the subject in a task of focused attention allocates only spare capacity to the
continuous monitoring of irrelevant inputs. The ~lmount of spare capacity
varies inversely with the demands of the primary activity. Consequently,
the rejection of unwanted inputs should be most effective when the primary task requires great effort. Zelnicker (1971) has confirmed this conclusion in her study of delayed auditory feedback.
The results of Corteen and Wood (1972) and Lewis (1970) indicate
that a stimulus on a rejected "channel" may affect some aspects of be:havior if it activates recognition units. In addition, such a stimulus may
131
be consciously perceived if it causes a reorientation of attention. The

conditions for such a reorientation have been described in the context
of the orienting response. A sudden and intense change will elicit an OR.
In addition, the activation of certain recognition units usually causes an
orientation. The effectiveness of a subject's name as an elicitor of OR's
is well documented in the Rllssian work. Stimuli that will cause an OR
when presented on an attended channel are also the most likely to be
detected when presented on a rejected channel.
Responses to an item on a rejected channel are not always associated with an OR. Sometimes, an item on a neglected channel is perceived because it conforms to expectations, not because it violates them.
An example is Treisman's' (1960) observation that subjects will folIowan attended message when it is switched from one earphone to
the other. This cannot be explained by the occurrence of an OR. Here, a
stimulus that was expected is perceived by a grouping effect. An important difference between this case and the responses to orientation stimuli
on the rejected channel is that Treisman's subjects were usually unaware
that they" had switched channels. On the other hand, a subject who detects a tone or a mention of his own name on the previously unattended
channel is immediately aware of the reorientation of attention.
Consciousness and Expectations

According to the present model, we are aware of perceptua,l interpretations, but we are aware neither of the activity nor of the output of
earlier stages of processing. Since it is assumed that attention to a stimulus object increases the likelihood that the object will be fully interpreted in perception, it follows that perceptual effort and awareness
should be correlated.
Stimuli on the rejected channel do not attract much effort. Consequently, the perceptual interpretations that correspond to these stimuli
are impov.erished, and the awareness of them is slight. Thus, we consciously perceive only very few of the events on a rejected channel. Since
the conditions for awareness and for storage in long-term memory are
closely related (Posner & Warren, 1972), we also remember very little of
what happened on that channel.
Contrary to the position taken by Hochberg (1970), it may be argued tha~ it is the violation of expectations, rather than their confirmation, which promotes conscious experience. We soon lose our awareness
of the ticking of the clock, although the expectation of continued ticking
is continuously confirmed. It is the stopping of the clock of which we become aware. Similarly, we are keenly aware of driving a car only when
expectations are violated .or when the situation is changing so rapidly
132
that no reliable expectations can be formed. These are conditions that

require considerable effort.
The present approach to focused attention entails a number of predictions that distinguish it from other theoretical positions. It shares with
Neisser's theory a common emphasis on the role of pre-attentive mechanisms that constrain the subsequent allocation of attention. An experiment
that illustrates these effects is discussed in some detail in the next section.
GROUPS OR CHANNELS
An instructive failure of selective attention can be observed in the

following experimental situation: a subject hears seven relevant digits,
preceded by the irrelevant and redundant digit "zero." He is to repeat
only the seven relevant digits. In spite of the instruction to ignore the
"zero," the memory for the relevant digits is markedly impaired (Dallett,
1964). This impairment has been called the stimulus prefiX effect. There
is also a stimulus suffix effect, where memory is impaired by the presentation of an irrelevant and redundant "zero" at the end of the list.
The two effects represent a clear failure of selection: the subject is forewarned, but he is nevertheless incapable of rejecting the interfering' irrelevant stimulus.
The boundary conditions for the occurrence of the suffix effect have
been investigated in a series of careful studies (Crowder, 1967, 1969,
1971; Morton, 1970a, b; Morton, Crowder & Prussin, 1971; Morton &
Holloway, 1970). It was found that memory can be disrupted by a suffix
that is not a digit, or even a meaningful word: any speech sound uttered
in the same voice as the relevant mes~age causes a suffix effect.
Several manipulations were found which abolished or reduced the
suffix effect: a visual stimulus does not cause the effect, nor does a sound
that is not speech-like. The disruptive effect is reduced when the suffix
is spoken by a different voice than the relevant list, or when it appears
to originate from a different location.
Morton and' Crowder interpreted these results by a theory of the
precategorical acoustic storage, or PAS. The ineffectiveness of a visual
suffix was accepted as evidence that interference occurs within an auditory system. The unimportance of semantic content indicated that this
system is precategorical, i.e., located upstream' of the word recognition
system. The role of similarity between suffix and list was interpreted in
the context ofa filter theory: items that arrive on the same channel enter
the same system of storage, but the filter can be set to preclude entry
of stimuli that arrive on other channels. CO'ntrary! to Broadbent's assumption that the filter is encountered after all'stimuli'are accepted into the
133
S-system, the observation that the suffix effect can be prevented .led to
the conclusion that PAS is located after the filter (Morton, 1970a, b).
The stimulus prefix effect was addressed by Neisser in very different terms (Neisser, Hoenig & Goldstein, 1969). In the context of a theory
of analysis-by-synthesis, the rhythm of the presentation was assumed
to dominate the perceived organization of the digit list: "The stimulus
string consists of eight digits and is heard as such; all eight take up space
in the resulting construction-even if one was redundant-because they
were heard as a single utterance [Neisser, Hoenig & Goldstein, 1969, p.
425]." It follows that the prefix effect should be eliminated by altering the
perceived structure of the string. This was successfully achieved by presenting the prefix in a different voice than the list. Moreover, there was
no disruptive effect when the prefix consisted of the sequence "zero,
zero, zero." This prefix constitutes a group, which can be easily segregated from the relevant material.
This study illustrates the superiority of a formulation of selective
attention as an operation on perceptual units rather than on channels.
It is surely unreasonable to assume that the triad "zero, zero, zero" defines a channel, whereas a single "zero" does not. The triad, however,
provides an adequate group.
The theory of attention that was summarized in the preceding section explains the results of this study by the operation of a pre-attentive
process of unit formation: interference occurs only within a perceptual
unit, and it can be prevented if the potentially interfering material is included within a unit of its own. This interpretation applies to the suffix
effect investigated by Crowder and Morton, as wel~ as to the prefix effect
studied by Neisser et ale Furthermore, the same rule is expected to apply
in vision as well as in audition. In contrast, the Crowder-Morton hypothesis explains only auditory suffix. effects, in terms of interference
with a precategorical acoustic storage.
To test this conception, Ulric Neisser and I tried to obtain a visual
equivalent of the suffix effect, and also to reduce that effect by a manipulation of grouping structure.
Subjects (N == 64) were shown a clearly legible array of six relevallt digits for half a second, and they immediately wrote the digits they
could recall. A visual "suffix" was shown on most trials, next to the far
right item. The 'digit "zero" appeared only asa suffix, and the subjects
were given advance exposure to all the suffixes that were used. Figure
7-1 shows several of the displays and indicates the number of errors that
were made in the fifth and sixth positions for each of these displays. As
in the case of the auditory suffix, the detrimental effect of the visual suffix
was most pronounced in the two positions closest to the interfering item.
Figure 7-1 shows that interference is pronounced when the suffix
134
Percent errors
Percent errors
Positions
Positions
54
57
45
49
43
42
(B)
( A)
976543
41
38
( D)
(C )
1375260
53
69
( E)
857146
000
000
726439000
000
000
418359
( F)
48
47
842915.
FIGURE 7-1
Effects of different visual suffixes.
is embedded within the perceptual group of relevant material. Interference can be prevented or reduced by removing the suffix from the relevant group, as in E, or by embedding it into another group, as in D.
This experiment has been discussed in detail because it provides a
suggestive prototype of focused attention tasks. As in other attention
tasks, the subject is instructed to respond to some stimuli and ignore
others. Whether he can do so depends on a grouping process which
precedes and constrains the allocation of attention. Attention operates
by emphasis rather than by filtering: the suffix is always "seen," but it is
seen as background rather than as figure.
The results of Figure 7-1 demonstrate the futility of attempts to
explain all effects of attention by a bias on the control of responses. The
pertinence of all suffixes was surely very low, yet some caused interference while others did not.
It is also difficult to analyze this visual example of focused attention in terms of a filter that selects among channels. The difficulty illustrates the prevalence of auditory concepts in modern discussions of
135
attention: the phrase "attend to a channel" suggests a sense of temporal

continuity which is inappropriate to th.e perception of a ,stationary visual
scene. In the context of attention, the most important difference between
vision and audition is that auditory perception requires spatiotemporal
grouping while the visual analysis of unmoving objects involves only
spatial grouping. Auditory attention to one message in a medley is
analogous to visual attention to one dancer in an ensemble, a vastly more
complex case than that of our experiment.
It is tempting to speculate that the modern study of attention could
have taken a different course if Broadbent (1958) had been concerned
with how one sees dancers rather than with how one hears messages.
Since it is surely possible to see many dancers while attending to one,
the concept of a filter that allows inputs into perception in single file
might 'not have been proposed. Deutsch and Deutsch (1963), on the
other hand, might not have argued that attention does not alter perceptual analysis, because the difference between the perception of the prima
ballerina and of lesser dancers is too obvious to be ignored. Finally, the
traditional emphasis on spatial organization in vision would have led
much sooner to a discussion of the pre-attentive mechanisms that control attention.
REVIEW
Selective attention to inputs is the allocation of capacity to the

processing of certain perceptual units in preference to others. The focusing of attention is very effective in preventing irrelevant stimuli from
interfering with the primary task, but there is evid~nce that irrelevant
stimuli are sometimes processed at least up to the level of recognition
units. In addition, one often perceives such stimuli, if they tend to be
grouped with the message, if they represent obvious physical changes,
or if they are both fanliliar and highly significant. These observations
are consistent with the hypothesis that spare capacity is continuously allocated ,to the processing of perceptual units that are not emphasized-.
The present theory assumes a mechanism of unit formation, which
performs some of the functions that Neisser attributed to pre-attentive
mechanisms. The stage of figural emphaSiS selects some of the units for
especially detailed processing, much in the manner of Broadbent's filter.
The emphaSis on the selected messages is a matter of degree, as suggested by Treisman's concept of attenuation. The distinctive predictions
of the present the~ry are that the effectiveness of selection depends on the
ease with which relevant stimuli can be segregated at the stage of unit
formation, and that the effectiveness of rejection of irrelevant stimuli
depends on the amount of capacity demanded by the primary task.
8
Attention Divided Among Inputs
The class~c question of whether attention is unitary can be rephrased in

modern terms: can two simultaneous but unrelated inputs be processed
at the same time? The various theories of attention reviewed in the prece~ing chapter imply different answers to .this question.
Broadbent's filter theory proposed that inputs are processed in
parallel in the S-systemof .sensory registration and preliminary storage
and at the elementary level of analysis which controls the setting of the
filter. Beyond the filter, -inputs are handled serially. Two simultaneous inputs may both be perceived and responded to, but not at th.e . same time.
One is processed first, and the filter only later retrieves the other item
from the S-system of storage. If processing the first stimulus in the Psystem takes too long, the second willbe lost from the S-system.
Deutsch and Deutsch (1963) and Norman (1968) did not deal directly with the divided attention issue. Their treatment of focused attention assumed that parallel processing normally occurs at all levels of
perceptual analysis, with .a bottleneck that controls entry to awareness,
res-ponse selection, and Ile:rmanent memory. Subsequently, Norman supported the id'ea ofa limited capacity for the: systems of perceptual analy136
Attention" Divided Among Inputs
137
sis (Norman & Rummelhart, 1970) and memory (Lindsay & Norman,
1969).
Neisser (1967) assumed parallel processing at the pre-attentive
level, but he treated focal attention as unitary. Furthermore, the idea
that speech is analyzed by synthesis appears to imply that only one verbal input can be synthesized at a time.
Treisman's (1960) attenuation concept implied that some parallel
processing of concurrent inputs occurs even when attention is deliberately focused on one. input. Later she argued that parallel processing of
simultaneous stimuli is possible in different analyzers, while serial
processing is necessary within \a single analyzer (Treisman, 1969).
The discussions of figural emphasis and of effort in previous chapters suggests that parallel processing of simultaneous inputs is possible.
However, para1lel processes that impose heavy demands on the limited
capacity are likely to interfere with one another.
We will first examine some experiments in which the processing of
simultaneous stimuli was either seriously impaired or obviously serial,
then turn to other experiments in which processing was demonstrably
parallel. Effects of effort and modality are discussed in the final sections.
CONCURRENT MESSAGES AND THE SPLIT-SPAN EXPERIMENT
A classic case of inability to process simultaneous sensory messages

in parallel was the "personal equation" of nineteenth-century astronomers
(Boring, 1950). Astronomers who endeavored to time the crossing of stars
by coordinating a visual event to the beat of a clock discovered that they
could not agree on their judgments. They eventually traced the difficulty to
the irresistible tendency of any observer to focus his attention primarily
either on the star or on the beats of the clock. Simultaneous stimuli on the two lTIodalities were not perceived Simultaneously: the attended one was perceived as having come sooner. This phenomenon became known as the 1aw of prior entry, and it played an important role
in early experimental studies of attention (James, 1890, Chap. 13; Titchener, 1908, p. 251). This phenomenon was used to demonstrate the
unitary character of attention and the impossibility of dividing it among
concurrent events. A careful recent experiment has confirmed the existence of the effect (Sternberg, Knoll & Bates, 1971).
There have been many other recent studies of apparent simultaneity (see Ladefoged & Broadbent, 1960; Fodor & Bever, 1965; Kristofferson, 1967; Moray, 1969b; Reber & Anderson, 1970). The common
interpretation of systematic errors in judgments of simultaneity continues
to be in terms of limitations of) human capacity which impose serialproc-
138
essing at some stage of analysis or decision. The results and the models,
however, vary widely (see Sternberg & Knoll, 1972, for a detailed review).
Prior entry was originally investigated because the demands on the
nineteenth-century astronomer appeared to exceed his capacity. The
modern investigations of the problem were prompted by the difficulties
of 'another overloaded functionary: the air-traffic controller (Broadpent,
1952, 1954a; Mowbray, 1953, 1954; Poulton, 1953; Spieth, Curtis, & Webster, 1954; Webster & Solomon, 1955). The common finding in these
studies was that listeners either completely fail to deal with simultaneous
messages or at best handle them successively.
Broad"bent (1954a) inve!1ted an experimental task which he used to
demonstrate the successive handling of simultaneous stimuli. The subject
in the split-span experiment is presented with two lists of digits simultaneously and reports what he recalls from the presentation. In a dichotic presentation, for example, he may receive the sequence 7-2-8 to one
. ear, and 9-4-5 to the' other. In other variants of the split-span design,
auditory items may be presented simultaneously on two external speakers, or an auditory item may be paired with a visual item (Broadbent
& Gregory,1961, 1965; Madsen, Rollins & Senf, 1970). When the presentation rate is faster than about one pair/second, these designs produce
the same result: subjects' reports tend to group all items that have arrived on one channel (defined by ear, location, voice, or modality), followed by the items from the other channel. Subjects who are required to
report the items in pairs (e.g., 79-42-85 in the example above) make more
errors than when they are allowed to report channel-by-channel. Order
information, in particular, is often lost in pair-wise recall (Bryden, 1962,
1964; Moray & Barnett, 1965). The difficulty of pair-wise recall decreases
markedly with prolonged practice (Moray & Jordan, 1966), but the task
always remains difficult.
The split-span experiment has produced three main findings. (1)
The task can be performed. Although subjects' performance with such
short lists is impaired, it is often adequate. (2) Successive, not simultaneous items tend to be grouped in recall. (3) In particular, items presented
on the same channel or in the same modality tend. to be grouped. These
findings suggested to Broadbent the image of a filter which selects a
channel, stays on that channel until the termination of its message, and
then switches to accept the second message, which was stored meanwhile in the S-system.
An alternative interpretation is that the preferred order of report
is determined by .perceptual grouping. If items are perceptually grouped
by source or modality, the most advantageous order of report will be
compatible with this spontaneous organization. The tendency to group
successive rather than simultaneous items also represents a law of group-
139
ing. Savin (1967) presented subjects with two successive pairs of digits,
all four digits spoken in the same voice and, originating at the same
place. His subjects almost invariably grouped successive rather than
simultaneous items in their reports. Thus, the normal mode of organization for auditory stimuli is sequential. Although this was not its original
purpose, the split-span experiment has provided valuable information
concerning, Gestalt-like rules in auditory perception.
The tendency to report items by channel can be overcome by other
grouping 'factors. This was first shown by Gray and Wedderburn (1960).
They presented the three syllables of a word (e.g., extirpate) or the three
words of a brief phrase (e.g., mice eat cheese) in alternation to the two
ears (e.g., right-Ieft-right), and simultaneously alternated a list of three
digits (left-right-Ieft). They found that subjects' reports followed content
rather than ear of arrival. Similar results have been obtained in many
other experiments (e.g., Bartz, Satz & Fennell, 1967; Broadbent & Gregory, 1964; Yntema & Trask, 1963). This effect of content is limited to
dichotic presentation; when the series are presented on different modalities, report by content almost never occurs (Madsen, Rollins & Senf,
1970).
The content effect in the split-span design is probably related to
the finding that subjects who are instructed to shadow b'y ear are nevertheless affected by the continuity of content (Treisman, 1960). It has already been mentioned that semantic factors can affect perceptual organization. In addition, groupillg by content in the split-span design also
facilitates retrieval. Subjects who attempt to report by channel must
produce awkward sequences such as mice-three-cheese, six-eat-two, and
organization by content could be imposed at retrieval to avoid such sequences (Broadbent & Gregory, 1964; Sanders & Schroots, 1968).
The question of how attention affects storage in the split-span
design has been studied by Bryden (1971). His subjects were to actively
rehearse one series (called A, for attended) and to ignore the other (U,
for unattended). They were also to report both series in a specified order, with either the attended or the unattended series first (i.e., the
orders were UA or AU). The striking result was the difference in the
shape of the serial position curves for the A and U groups. As may be
seen in Figure 8-1, the serial position curve -is Hat for the A group, but
there is a pronounced recency effect with U items. Significantly, this
result occurs regardless of the order of report.
Another interesting feattlre of Figure 8-1 is that the order of report
(AU or VA) has less effect on the U-message than on the A-message.
The A-message is more susceptible to output interference, the disruption
of memory which is caused by the activity of recall. This finding is consistent with the idea that the U -message is stored as an acoustic rather
140
220
210
200
190
0
...qN
180
c 170
E
....0 160
Q)
''-
0
'Q)
...0
E
:::>
150
140
130
120
110
A-U order
0---0 U-A order
100
OT
Attended
Unattended
FIGURE8l
Recall as a function of attention and order of report (Bryden, 1971, with permission).
than as an articulatory code. Perhaps subjects "attend" to a message by a

form of enactiveencoding, or rehearsal (Murray & Hitchco~k, 1969;
Neisser, 1967) which alters the nature of the memory trace, and also
strengthens it.
Bryden's study is especially noteworthy because of the novel use of
attention as an instruction variable, but its conclusions concerning the
nature of storage have been challenged. The observation of a pronounced recency effect is consistent with the .idea that the unattended
message is stored in a precategorical store, but it provides no conclusive
proof. A similar recency effect can be obtained with visually presented
words, which are probably "read" at the time of exposure (Corballis &
Luthe, 1971). Massaro (1972) has argued that the subjects in Bryden's
experiments perceived all items as soon as they were shown, and that
the difference between the A and the U serial position curves was due
entirely to rehearsal.
In summary, the split-span experiment has shed less light on theoretical issues than was originally hoped. Certainly, subjects prefer to
group their responses in a manner reflecting the perceived organization
of the input, and they perform best when allowed to do so. However,
141
there is no evidence that subjects must respond in this manner, nor that
they cannot perceive simultaneous items in parallel.
F AlLURES
OF DIVIDED ATTENTION
A striking failure of parallel processing was reported by Colavita

(1971). His subjects were instructed to press one key to a light flash and
another key to a tone. The subjects expected a single event on each trial,
but on some trials both the light and the tone were presented. These occurrences were attributed to equipment malfunction. On 49 of 50 conHict trials (five each for ten subjects), subjects pressed the "light" key
alone. On the single trial when the "tone" key was pressed, the subject
apologized for his mistake! The subjects were unaware that the tone had
been presented on 17 trials. For man, a visual stimulus is clearly dominant over a concurrent auditory stimulus, and it captures both awareness and response.Colavita observed the opposite pattern in nocturnal
animals.
Most studies of divided attention have used verbal material. For
example, Mowbray (1953), found that subjects could not listen to one
story while reading another. On a subsequent test of comprehension of
the two messages, the poorer score was consistently near chance level.
Despite the intention to divide attention, sllbjects were apparently focusingon only one of the messages. In a subsequent study, Mowbray (1954)
simultaneously presented an auditory and a visual message, which were
to be used in a complex task. His subjects were unable to use the simultaneous messages and usually denied noticing the simultaneity. "Many
were vehement in their denials and expressed outright surprise, and in
some instances, sheer disbelief that such had really been the case [p.
90]." In a simulation of the air-traffic cqntroller's task, Webster and
Thompson (1954) also observed very poor handling of simultaneous
auditory messages, except when these were highly redundant.
Subjects have also been instructed to shadow a continuous auditory message in one. ear and to note for later recall an isolated word presented either to the other ear (Mowbray, 1964) or visually (Mowbray,
1962). Shadowing was usually disrupted by the presentation of the
critical word, and the disruption was more severe when that word was
spoken than when it was shown visually. Treisman and Geffen (1968)
also observed interference with the shadowing of a message to one ear
when the subject detected a target word in a concurrent message to the
other ear. The word that coincided with a target item was frequently
missed in shadowing.
A study by Moray and O'Brien (1967) provides a crucial observa-
142
tion concerning the limitations of divided attention. Moray and O'Brien

presented dichotic messages consisting of 90 percent digits and 10 percent letters at a rate of 100 items/minute in each ear; the subject pressed
a key with the appropriate hand whenever a letter was heard on either
ear. A striking result occurred when two letters were simultaneously presented to the two ears: subjects pressed at least one key on 99 percent of
such occasions, but they pressed both keys on only 17 percent of occasions. In a subsequent series of experiments, Moray (1970a, b) obtained
essentially the same results in the detection of transient increments of
loudness in tone series. Again, when two simultaneous targets were presented, the listener was very likely to respond to one but unlikely to respond to both. These results can be stated in correlational terms: there is
a strong negative correlation between responses to concurrent targets on
the two ears.
Shaffer and Hardwick (1969b) have reported a monitoring experiment rather similar to Moray's. Their subjects reported the successive
repetition of the same word in either of two dichotic messages. Shaffer
and Hardwick concluded that some processing of the two messages was
possible because subjects detected about 60 percent of these targets.
However, there were systematic sequential dependencies between successive items: if a subject had detected a target on one ear, he was likely
to detect an additional target on that ear, and he was also likely to miss
a target on the other ear. The positive correlation between detections on
one ear and the negative correlation between ears suggest t)lat subjects
were not dividing their attention equally between the two ears. At any
time in the task they were listening more to one ear than to the other.
Moray (1969b) was led by his observations to a modified version of
filter theory. He postulated a filter that can alternate very rapidly between channels, as long as no stimuli of special significance are detected.
When an important stimulus is recognized, the filter remains locked on
the channel where that stimulus arrived until its processing is terminated. These hypotheses explain why a single target is easily detected in
divided attention, while an either-or pattern of detection is) approximated
with simultaneous targets. However, it has been pointed out that performance with such targets is actually too good to be explained by a
strict application of Moray's time-sharing theory (Treisman, 1972). Another difficulty for Moray's theory arises because of the extremely important role that it assigns to the timing of stimuli: if two targets are
presented in slight asynchrony, the one presented first should always be
detected. In fact, the precise synchrony of inputs does not appear to be
particularly important in divided attention (Treisman & Davies, 1972).
When two auditory targets are presented in near synchrony, the factor
of ear dominance is the primary determinant of which is detected, rather
143
than the precise temporal relations between the onsets of the targets
(Avner, 1972).
EXPERIMENTAL TESTS OF THE DEUTSCH-NORMAN THEORY
A central issue among theories of attention concerns the effects of

inattention on perceptual analysis. Several theories assume that the
analysis of unattended items is impaired or precluded (e.g., Broadbent,
1958, 1971; Moray, 1969b; Neisser, 1967; Treisman, 1960, 1964d, 1969).
The analysis of attention in this book also adopts this position. On the
other hand, other theorists have assumed that attention does not. affect
the contact of sensory information with recognition units (Deutsch &
Deutsch, 1963; Keele, 1973; Norman, 1968). In that view, attention
merely determines which of the currently activated recognition units will
be allowed to control awareness and response.
The (~wo critical experiments in this area are due to Treisman. In
the initial study, Treisman and Geffen (1967) attempted to determine
whether selective attention operates on perception or response. They
reasoned that filter theory and t4e Deutsch and Deutsch theory imply
different predictions for the following experimental situation: a subject
is required to shadow all items arriving on a designated ear and to perform an additional response (tapping) to some items, regardless of their
ear of origin. According to Deutsch and Deutsch, a critical item should
elicit a response regardless of the ear on which it is heard, because the
corresponding recognition unit is pres~t for it. According to filter theory,
on the other hand, the requirement to shadow the message presented to
one ear prevents the allocation of attention to the processing of items on
the other ear, and therefore precludes the recognition of target items on
that ear. In other words, filter theory asserts that a listener cannot comply with the instruction to divide attention between the two channels,
while the Deutsch and Deutsch theory asserts that he can.
The results confirmed the prediction derived from filter theory.
Treisman and Geffen (1967) reported that subjects detected 87 percent
of the target. words in the attended message, but only 8 percent in the
unattended message. When instructed to tap to a word (e.g., rite) but
not to its homophone (e.g., write), subjects were able to use contextual
cues to avoid tapping to incorrect homophones only in the attended message. Context had virtually no effect on the detection of target words in
the rejected ear.
Deutsch and Deutsch (1967) did not accept this experiment as
critical evidence against their' theory. They pointed out that a subject
was required to perform two tasks, repeating and tapping, .in response
144
to target items on one ear and only one task in response to critical items
on the other ear. This difference could be reflected in the importance attached to critical items on the two ears, with a consequent bias in favor
of the slladowed message.
To overcome this criticism, Treismanand Riley (1969) required
subjects to shadow one of two dichotic lists of computer-synchronized
digits and to detect occasional letters on either ear. The ingenious feature of the design. was the response that the subject w.as to make to a
target item on either ear: he was to immediately stop shadowing. Thus,
the response to a critical item was the same, regardless of the ear on
which is was heard. The subjects detected "76 percent of letters on the attended ear, and 33 percent on the other. The large bias in favor of the
shadowed message' provides strong evidence against the Deutsch and
Deutsch theory (1963, 1967). However, it should be noted that selectivity
was less impressive than in the Treisman alld Geffen study. This result
would be expected by a theory which emphasizes grouping effects. The
use of computer-synchronized digits eliminates most factors essential to
the formation of effective perceptual groups, and thereby hampers
selection.
In other conditions of the same experiment, Treisman and Riley
(1969) observed that critical letters which differ in voice from the background digits are always detected on the unattended channel. This effect
of voice quality relates' to Lawson's (1966)' finding that listeners easily detect a tone on the rejected ear. 'As was noted in the preceding chapter,
such results are consistent with filter theory, which assumes that all stimuli are tested on obvious physical characteristics.
Treisman's experiments demonstrate that attention cannot be divided between .concurrent stimuli if the listener is' biased toward one
channel by the instruction to shadow one of the 'messages. The bias
favoring the shadowed message overcomes the effects of the relevance of
designated targets, such as the letters in the Treisman.. Riley study. Another demonstration of attentional bias was offered by Weg (1'971) in
my laboratory. Subjects listened to dichotic messages, each ten words
long, presented at a rate of two pairs/ second. The recognition of rightear words was tested, as in the focused attention experiments described
in the preceding chapter. In addition, subjects were required to note the
occurrence in the left-ear message of one or two target items, which they
were asked to recall immediately after the presentation of the message.
Three types of target items were used: (1) words belonging to a content
category, such as animal names; (2) digits; (3) isolated digits in a male
voice inserted in a message spoken by a female. 'Subjects were paid a
substantial bonus for successful recall of critical left-ear words and were
paid at a lower rate for recognizing right-ear items. They were penalized
for mistaking left-ear words as familiar in the recognition test.
145
The subjects were highly successful in the recall test, remembering

over 90 percent of the critical left-ear items in all three conditions. However, their ability to selectively recognize words presented on the right
ear was severely impaired.' In conditions (1) and (2), monitoring for content and for digits, subjects were unable to prevent massive intrusions of
left-ear items in the recognition test. Among the words that they identified as having been heard on the right ear, only 40 percent were in fact
on that ear, 44 percent were left-ear words, and 16 percent were unpresented distractors. Performance was better in the condition (3),where
critical items were identified by voice quality. In that condition, 53 percent of the recognized items had been heard on the right ear, 34 percent
were left-ear items, 'and 13 percent were'distractors.
The high rate of intrusions in the content conditions ".is not easily
accommodated by the theories of Deutsch and Deutsch or Norman:
the prediction from this theory is that a listener should be able to attach high pertinence values both to right-ear items (when they occur),
and to left-ear animal names (in advance), thereby prev~nting intrusions of irrelevant items from the left ear. The experiment demonstrates
that subjects cannot perform in this manner. When expecting an important item on the left ear, they attend to all words on that ear, and they
are evidently unable to prevent attended irrelevant words from being
stored in memory. When the targetis expected to be labeled by aphysical characteristic (e.g., a distinctive voice quality), the subjects presumably have more confidence in their ability to detect it, and they pay less
attention to the irrelevant message in which it is embedded.
SUCCESSFUL DIVISION OF ' ATTENTION
In contrast to the experiments described in the preceding section,

other studies conclusively demonstrate that parallel processing of concurrent verbal' stimuli is possible. In an important experiment, Treisman
(1970; Treisman & Fearnley, 1971) presented subjects either with single
items or with pairs of precisely synchronized auditory items, consisting
either of two nonsense syllables or of a nonsense syllable and a digit.
The subject was to press a key if one of the items was a digit, and another key if neither item was a digit; his reaction time was measured.
On some trials the subject was given advance information about which
digit, if any, would be presented. This precuing is known to reduce RT.
The experiment was intended to test the hypothesis that concurrent
items are processed serially. This hypothesis entails the following sequence of events in response to the dichotic stimuli of the experiment:
(1)
Determine if the item on one of the ears is a digit.
146
If it is not(2)
(3)
Switch to the item on the other ear.

Test that item.
When a single stimulus is presented, the process is completed after

phase 1. Now consider the effects of precuing. With a single stimulus,
precuing abbreviates phase 1. When two stimuli are presented, both
phases 1 and 3 will be shortened by precuing. Consequently, the hypothesis of serial processing entails a larger reduction of RT by precuing with pairs than with single stimuli. However, the striking result
of this experiment was that precuing reduced RT by 115 milliseconds
for both single items and pairs. Evidently, the decision that neither of
two simultaneous items is a digit could be made in parallel. for the two
items.
Although the analysis of precuing effects indicates that processing
was parallel, it is important to note that the efficiency of parallel processing was less than the efficiency of processing a single item: RT to pairs
was longer by about 80 milliseconds than RT to single stimuli.
Similar results were obtained in a study by Niniopand Kahneman
(1973). Subjects were exposed to brief dichotic word lists, and they
pressed a key whenever they heard an animal name. Reaction times
were measured in two conditions: attention divided between the two
ears or focused on one message. Despite, the fast presentation rate (two
pairs/ sec) the subjects detected 77 percent of the targets in the divided
attention condition, demonstrating some ability to deal with both messages. A more precise test of filter theory was obtained in that study by
comparing the RT distributions in focused and divided attention. According to Broadbent's initial statement, RT in divided attention depends
on the setting of the filter at the instant of target presentation: if the filter happens to be selecting the appropriate channel, RT should be short.
On the other hand, RT should be long if the filter was set to the wrong
channel at the critical instant. Thus, filter theory entails much greater
variability of RT in divided attention than in focused attention: the
fastest RT's should be identical in both conditions, and there should be
very slow RT's in divided attention. Contrary to this prediction, there
was little difference in the variability of the two RT-distributions, and
the difference between the three fastest reactions in the two conditions
(110 msec) was not much less than the difference between the means
(140 msec). This pattern of results supports the hypothesis that the processing of simultaneous words is parallel.
Lindsay (1970) has reviewed some studies of psychophysical tasks
in which attention was divided among different stimuli and among dif-
147
ferent relevant aspects of the same stimulus. Subjects in these experiments made absolute judgments of various attributes of simultaneous
visual and auditory stimuli. Their performance was evaluated by a measure of information transmission, which reflected their consistency, in
assigning distinct labels to different stimuli. Subjects transmitted almost
as much information on each dimension when they judged both stimuli
together as they did when the judgments were made one at a time.
Manipulations of stimulus duration had no effect, suggesting that the two
discriminations were indeed performed in parallel. Thus, attention was
effectively divided between the two tasks (Lindsay, Cuddy & Tulving,
1965; Tulving & Lindsay, 1967), at least when the stimuli to be judged
were easily discriminable from one another. Very different results were
obtained when discriminability was reduced, and these additional findings will be discussed in the next section.
Evidence of parallel processing was oqtained by the present author
and his students (Levy, 1971) in studies of recognition memory following
dichotic presentation of word lists. The experimental situation has already been mentioned in' the discussion of focused attention. In the divided attention condition, the subject~ were 'exposed to dichotic lists of
31 words each, and they subsequently attempted to distinguish words
that had been presented from unpresented distractors. The recognition
list always, included several pairs of items that had been presented
simultaneously to the two ears. Filter theory entails a strongly negative
relation between the recognition of simultaneous words: if an item on
one ear is recognized, then attention must have been directed to that ear
at the time of presentation, and therefore away from the other ear. Thus,
the probability of recognizing two simultaneous, items should be substantially lower than the product of their separate probabilities of recognition.
Several experiments, conducted at various presentation 'rates on a
total of 260 subjects, failed to confirm the, prediction from filter theory.
The interaction between simultaneous items predicted by this theory
simply did not occur. A typical subject in these experiments recognized
about 51 percent of right-ear items and 48 percent of left-ear items
(equivalent to about 30 percent recognition after correction for chance
success), and these values were unaffected' by the recognition of the
corresponding item on the other ear. In contrast to this independence between simultaneous items, there was' a slight but highly consistent negative relation between successive items.
The results indicated that man can listen to both ears at once and
store some part of what he hears, although recognition performance is
far poorer than when he listens to only one ear. The subjects faced with
the overwhelming task of listening to two messages at once quickly real-
148
ized the futility of any active strategy and usually reported adopting a
passive, receptive attitude. With this attitude, there were no indications
of interference between simultaneous items, but performance on all items
was quite poor.
EFFORT AND THE ALLOCATION OF ATTENTION
The evidence of the preceding section is consistent neither with

filter theory nor with the Deutsch-Norman theory. Filt~r theory is invalid
because it cannot accommodate the reaction-time evidence which suggests strictly parallel processing. The Deutsch-Norman theory is also incorrect, because it cannot accommodate the effect of channel bias on the
detection of a highly significant target item (Treisman & Riley, 1969).
Thus, there appears to be more division of attention than filter theory can
allow and less than the Deutsch-Norman theory would suggest.
Parallel performance has been evaluated by two main measures:
(1) comparisons of the effectiveness of joint performance to effectiveness
in a single task; and (2) computations of the correlations between concurrent activities. By both measures, different experimental tasks yield
strikingly different results:
(1) Parallel processing of simultaneous inputs sometimes occurs
with little interference (e.g., Lindsay, 1970; Lindsay & Norman, 1969).
In other situations processing is parallel, but its effectiveness is impaired
(Ninio & Kahneman, 1973; Treisman, 1970; Treisman & Fearnley, 1971).
Finally, there are cases where parallel processing fails altogether.
(2) The achievements in two concurrent activi~es are sometimes
uncorrelated, although both are impaired (e.g., my own'studies ofrecognition after dichotic listening). In other cases there is a marked negative
correlation between the processing of concurrent stimuli, so that successful processing of one makes the response to the other very unlikely (e.g.,
Moray, 1969b, 1970a, b; Moray & O'Brien, 1967; Shaffer & Hardwick,
1969).
The concept of limited capacity is helpful in ordering these chaotic
results. If the effort that man can exert at any time is limited, then any
two tasks whose joint demands exceed that limit must be mutually interfering.,Thus, the main prediction from an effort theory is that the ability
to respond to simultaneous inputs should depend primarily on the demands of the activities among which attention is to be divided.
There is some evidence to confirm this prediction. As was already
mentioned, effective parallel processing of stimuli was observed by Lindsay and his associates (Lindsay, 1970). They found that the information
conveyed by absolute judgments on a dimension did not decrease mark-
149
edly when the subject made several judgments on concurrent 'stimuli,

but this pattern held only when the stimuli were highlydiscrimin;able. A
different result was obtained when the chosen stimuli were less discriminable (Lindsay, Taylor & Forbes, 1968). The total information conveyed
in multiple judgments was then rather less than in the single-judgment
condition. Thus, "... it is not the amount of information in a stimulus,
as defined by measures of stimulus uncertainty, but rather the difficulty
in discriminating the signal in a given channel that causes performance
to break down under multi-channel conditions [Lindsay, 1970, p. 154]."
A series of careful control experiments established that the interference between the simultaneous judgment tasks occurred at the time
of stimulus presentation. When a subject was told after the presentation that he was not to report one of the stimuli, his judgments of the
other stimuli did not improve. Thus, the interference had -already occurred when the irrelevant dimension was indicated. On the other hand,
precuinghad a positive effect, because it permitted attention to be
focused on the ,appropriate stimulus. The cqnclusionof this work is that
man~s capacity to perform concurrent perceptual. tasks is limited. Easy
tasks can be performed together with little interference, but more difficult tasks cannot.
The capacity model may also explain why the correlation between
concurrent activities is nil in recognition, and strongly negative in monitoring, but the interpretation of these results requires further elaboration
of the model. The new hypothesis is that an even distribution of attention among concurrent activities is possible only at a low level of total
effort. When total effort is high, one of the activities typically draws most
of the attention, leaving little for the others. This suggestion is related to
the Easterbrook hypothesis discussed in Chapter. 3. It implies that attention is divisible at low levels and more nearly unitary at high levels of
effort. (See also the discussion of spare capacity on p. 16.)
In the recognition experiments conducted in my laboratory, subjects appear to adopt a passive. attitude during the presentation of long
(31 wordsj ear) dichotic lists of words. The phenomenology of the situation is suggestive: subjects report that they deliberately refrain from
paying particular attention to any word, because they realize that doing
so involves "missing" several other words. Thus, the listener allocates
some effort to the processing of both words in a simultaneous pair; but
he refrains from checking whether either of the words has been effectively stored in his memory. Under these conditions, recognition performance is poor, but the two items in a pair are prevented from
competition oi mutual interference.
The situation is quite different in Moray's monitoring experiments
(Moray, 1970a, b; Moray & O'Brien, 1967). In the monitoring situation,
150
the activation of a recognition unit corresponding to a target causes a

series of changes in the subject's state of attention, as' indicated in
Figure 8-2. The surge of effort initiated by a tentative recognition is
associated with a reduction of the capacity allocated to other signals.
The outcome is often failure to perceive another target, rather than a
mere delay of response. In discussing a similar monitoring experiment,
Shaffer and Hardwick (1969b, p. 403) noted: "It seems that the commitment to translate a signal into a response imposes a further limitation
upon, or interferes with, speech recognition." Thus, the even distribution
of attention in multiple monitoring represents an unstable equilibrium.
An initial tendency to allocate capacity to a target on one channel disrupts that equilibrium, and callses attention to focus on that target.
Whether two targets can be processed at once depends on the
amount of effort demanded by the processing. Moray's monitoring task
requires an immediate overt response, while our word-recognition task
does not. One of my students attempted to find an intermediate level of
effort between recognition and monitoring (Avner, 1972). His subjects
performed a monitoring task, but they made a covert decision rather
than an overt response whenever they detected a critical item. All sub-
AROUSAL
/.--
..........
/ increased "
/
arousal
\
(
reduces
\
\
spare
J
',capacity /
1----.
rlrvv-L
I
:AVAILABLE:
: CAPACITY i
/----....,
-T/
II \
.....-.--.'/ /
\:---
"-
D-rr:~ ~==D . .---'----.. ".,.-/

POSSIBLE ACTIVITIES
TENTATIVE
RECOGNITION
OF TARGET
..-----,
/'"
/
causes
\
'I increase ,
~----....I of demands .....- - - - - - - - - - - \\ on capac I ty /
"
'---...".,.-/
-....-.,
/ / surge of "
arousa I to "
/........---,
\
meet
/
,
\
increased ;1
may cause \ , demand /
I interruption \ ,____ ""
\
of other
J
\
activities I
""""'-_/
/ / directs '
I attention to \
\further activities,
\
related
/
" to target / /
Mt SC ELlANEOUS
MANIFESTATIONS
OF AROUSAL
FIGURE 8-2
Effects of the tentative recognition of a target stimulus.
151
jects listened to 48 five-second dichotic messages (ten unrelated words to

each ear) and filled a, recognition form where items presented to
both ears were to be marked. One group performed only this task. Subjects in two other groups were instructed, in addition, to notice target
words which they recalled in writing before the recognition test. In the
content condition targets were defined by a relevant category (e.g., animal names, capital cities) which was stated before each message. In the
voice condition, the targets were defined by voice quality (words in a
male voice). The nllmber of targets varied from one to three; 16 messages included a pair of simultaneous target items.
There was a marked .negative correlation between the recall of
such simultaneous targets. Given that one member of the pair had not
been recalled, the probability that the other member would be recalled
was 88 percent in the voice group and 96 percent in the content group.
Given that one member, of the pair had been recalled, the probability
that the other would also be recalled was only 58 percent, and 55 percent, respectively for the two groups. The negative correlation between
simultaneous targets was p,ronounced, but it was distinctly lower than
in the experiment by Moray and O'Brien (1967), which required an immediate overt response to each target. Thus, Avner appears to have succeeded in producing a situation in which less effort was required than in
Moray's tasks, and in which the negative correlation between the. responses to simultaneous stimuli was correspondingly less marked.
Avner's subjects detected target items on either ear very effectively.
When confronted with simultaneous targets, however, they showed pronounced ear dominance, usually of the right ear..As in the previously
described experiment by Colavita (1971) a dominance rule is applied to
"break the tie" when two stimuli require a simultaneous response. The
dominance effect in Avner's experiment was more prevalent in the voice
group ,than in the content group. In a subsequent study using the same
recordings, Gopher (1971) observed that the pattern of ear dominance
can be altered by requiring subjects to fixate 20 degrees to the right or to
the left of the frontal plane.
Two additional observations of Avner's study are relevant to the interpretation suggested in Figure 8-2. The first concerns the correlation
between simultaneous items in recognition. The preceding discussion attributed the independence- between simultaneous items in .our recognition studies to the subjects' passive attitude. The messages in Avner's
study were shorter (ten words per ear, compared to 31 in the other
studies), and the subjects appeared to adopt a more active attitude. They
also showed a con'sistent negative correlation between simultaneous
items on the recognition test. The results of the experimental and control
groups did not differ (the control group performed only the recognition
152
task). The probability (corrected for chance success) that a word would
be recognized increased from 40 percent to 50 percent, depending on
whether the word with which it was paired was or was not recognized.
The negative correlation in recognition, while consistent, was much less
pronounced than in recall. This finding is consonant with the view that
the negative correlation between concurrent activities is a function of
effort.
Another observation concerned the recognition of a non-target
word that had been presented simultaneously with a target. The recognition of those words was very severely impaired in the voice group, but
there was little or no impairment in the content group. A similar result
was obtained by Weg (1971). As may be recalled, Weg instructed subjects to listen to all words on the right ear for a recognition test, and to
detect and later recall critical items occasionally presented on the left
ear (see p. 143). The recognition of the right-ear word presented simultaneously with a left-ear target was more severely depressed when the
target was a digit in a male voice (the rest of the message was in a
female voice), than when the target was a digit or an animal name
spoken in the same voice as the rest of the message.
The difference between the interfering effects of voice and content
targets can be interpreted by noting that a voice target may be "recognized" as a target at an earlier stage of analysis than a content target.
Consequently, the narrowing of attention indicated in Figure 8-2 will be
more' detrimental to the item paired with a voice target,because attention will be withdrawn from it sooner. .
In' summary, the findings reviewed thus far indicate that parallel
processing of simultaneous inputs is pOSSible, contrary to 'filter theory.
There seems to be no single-channel bottleneck in the perceptual system,
. but attention tends to be more nearly unitary at high levels of effort than
when little effort is exerted.
MODALITY EFFECTS IN DIVIDED ATTENTION
Treisman (1969) has offered a theory of divided attention which

departs from filter theory in important respects. She proposed that
different analyzers can operate in parallel-without interference, but that
processing within anyone analyzer is necessarily serial. This hypothesis
explains why responding to the same dime.nsion of various objects is
hard, while responding to various features of a single stimulus is easy (see
Chap. 6). Another inference from this view is that dividing attention
among stimuli in the same modality will be more di.fficult than when
the stimuli are in different, modalities: stimuli in anyone modality are
Attention Divided
Amo~g
Inputs
153
more likely to reach overlapping analyzers. Indeed, Mowbray (1964) has

shown that shadowing a message on one ear is more severely disrupted
by the presentation of a single significant word on the other ear than by
the visual presentation of a word.
Treisman's argument implied that each analyzer functions as a
single channel in which processing is necessarily serial. It also implied
that there is only one analyzer for each stimulus feature. This is clearly
incorrect: for example, there certainly is spatial duplication of analyzers
in the visual system, since we are able to see more than one color and
more than one shape at a time. This position was moderated in a subsequent paper by Treisman and Davies (1972): "There may well be some
common pool of capacity, perhaps that involved in control processes, but
there may also be some more specific limits within the relatively independent perceptual analyzers."
Treisman's theory can best be tes~ed in a monitoring task, where the
subject responds to critical targets on several channels. The theory predicts that subjects should be able to monitor two messages as easily as
one provided that the messages do not reach the same analyzers. In
other words,dividing attention between analyzers should be as effective
as focusing attention on one analyzer. On the other hand, dividing attention between' messages reaching the same analyzer should be extremely difficult.
Treisman and Davies (1972) tested these predictions. They found
that monitoring two auditory or two visual messages was much harder
than monitoring messages on different modalities. However, performance
in even the easiest conditioll of divided attention was substantially worse
than in focused attention. These results are incompatible with the idea
that processing in separate analyzer systems occurs in parallel and without interference. Nevertheless, they indicate an interaction between tasks
which departs from a mere summation of effort: tasks involving the
same modality become disproportionately more difficult when performed
together. Thus, Treisman and Davies have shown that it is particularly
difficult to divide attention within a modality, but they did not show
that it is easy to divide attention between modalities.
Allport (1971) also reported a study of attention divided among
analyzers. He presented three colored outline shapes in which a numeral
was sometimes inscribed. The subject was instructed to report the values
of the three items on one or two dimensions (color, shape, numeral).
Interference occurred when the subject was to report shapes and numerals, but not when he reported color with another dimension. Allport
explained this result on the grounds that reading numerals and naming
shapes require overlapping analyzers. However, the fact that interference
was limited to the second dimension reported suggests that the inter-
154
ATTENTION AND EFFOIlT
ference may occur in encoding or in memory rather than in perception,

as Allport assumed.
Division of attention within a modality often involves a conflict of
orientation tendencies (Kahneman, 1970). In the visual case, for example,
it'is highly unnatural to "think" about either of two displays without
looking at them. - Orientation tendencies also arise in audition. Thus,
Gopher (1971) has found that the eyes continuously make small sideways excursions during dichotic listening. This elicitation of conflicting
orientations may have a detrimental effect on auditory perception. Treisman (1971) observed that shadowing a message which is rapidly alternated between the two ears is more -difficult than shadowing a monaural
message. A series of studies (Axelrod & Guzy, 1968; Axelrod, Guzy &
Diamond, 1968) have shown that the apparent rate of a series of clicks
is lower when the clicks are alternated between the two ears than when
they are presented to one ear or binaurally. Axelrod and Powazek (1972)
showed that the apparent click-rate increases as the spatial separation
between the sources is reduced, and they pOinted out that this result
could be due to the elicitation of motor response tendencies.
When the stimuli to be attended are in different modalities, however, orienting appropriately is not difficult. Imagine trying to listen to a
message to your right ear while looking to your left. You will probably
notice a tendency to cock your head toward your right shoulder.
Dividing attention within a modality is also difficult whenever a
task involves storage. Kroll et ale (1970) have found that a single target
word presented during shadowing is retained better if it is visual than if
it is included within the auditory message. The visual item appears to be
relatively immune- to. retroactive interference from the shadowing task.
This observation suggests that visual and auditory storage mechanisms
are at least partly independent. Consequently, interference- is more likely
to arise between items presented to the same modality than between items
on different modalities (Parkinson, 1972). Related results were reported
by Treisman and Davies (1972, exp. 1) in a bisensory split-span experiment.
The usual memory loss for the second series was avoided if that series waS
presented on a different modality, and if different modes of response were
employed for the series. Response mode plays an important role in this
context, and it is discussed in greater detail in Chapter 10.
REVIEW
This chapter began with the question of wheth~r simultaneous inputs are 'processed in parallel or in sequence. The conclusion is that
both modes of processing occur, depending on the task and on the cir-
155
cumstances. Several experimental designs have been used to study this

question.
Split-span. Subjects tend to group their responses in accordance
with the perceived grouping of stimuli. In general, this strategy yields
grouping by channel in the dichotic case, or by modality in the bisensory
case. However, if grouping by ear yields an uncomfortable response
sequence (e.g., a medley of digits and letters in each group), grouping by
type can be adopted.
Monitoring several channels is possible, but less effective than
monitoring a single channel. Reaction-time evidence suggests that analysis of concurrent signals occurs in parallel, but at a slower rate than in
focused attention. It appears to be especially difficult to monitor spatially
separated messages in the same.modality, .perhaps becau~e~onHicting
orielltation tendencies are elicited. Special attention to one channel
makes a significant target on another channel difficult to detect, contrary
to the Deutsch-Norman theory.
Responses to simultaneous items frequently fail. The subject almost
always responds to one of the stimuli but often remains unaware' of the
other, producing a negative correlation between the occurrence of the
two responses. "Tie-breakin,g" rules are .consistently applied in such-cases,
including oominance of a light over a tone, or of a word presented to
the right ear over a word to the left ear. The negative correlation between responses to simultaneous stimuli is reduced when the responses
are made less demanding. This result supports an effort theory of divided
attention. In addition, there is a modality effect, such that simultaneous
targets on the same modality are more difficult to detect and to process
than simultaneous targets on different modalities.
9
Speeded Responses
to Simultaneous
and to
Immediately Successive Signals
Measures of reaction time have been used extensively in attempts to
study man's ability to divide his attention between two response tasks
which overlap in time. The' first section of this chapter discusses the
issue of response integrality:. when are two physically distinct responses
properly viewed as components o~ a single molar response? Subsequent
sections deal with results obtained with quickly successive signals. Additional studies in which reaction time is used to measure the division of
attention are described in Chapter 10.
MULTIPLE RESPONSES AND MULTIPLE TASKS
How does man produce multiple responses to mul~iple simultaneous or immediately successive signals? A vast number of studies have
been devoted to this question, in an attempt to clarify the interactions
between concurrent processes of perception, decision, and response. A
preliminary question that must be answered, however, concerns the very
definition of multiple signals and multiple responses. The discussion of
integrality in Chapter 6, and of grouping processes in Chapters 5 and 7,
156
Speeded Responses to Simultaneous and to Immediately Successive Signals
157
made it clear that the question of what constitutes a single stimulus is

an empirical one. The mere fact that the experimenter can independently
manipulate some physical dimensions of objects does not guarantee that
these dimensions function as separate stimuli in the control of the subject's behavior. Similarly, the fact that the components of a complex
response can be measured separately does not guarantee that it is useful
to view the performance of each of these components as a separate
activity. It must be admitted, however, that our understanding of integrality and grouping of responses lags far behind our understanding
of integrality and grouping of percepts.
Consider, for example, an experiment in which the subject presses
a key with his right index finger when he hears a tone and another key
with his left index finger when a light is flashed. The unexpected presentation of a light and a tone at the same time leads to an either-or
choice, which typically favors the response to the light (Colavita, 19'71).
With proper instructions, however, a subject will easily learn to treat the
joint occurrence of a light and a tone as a single event, to which he will
respond by pressing both keys. The physically separate responses of the
two fingers, which constitute alternative acts in Colavita's experiment,
can readily be combined into a molar response unit.
The nature of the effective response unit is not always so easy to
discern. As an example of this problem, we shall consider in detail a
careful study by Schvaneveldt (1969, exp. 2). The experiment w.as conducted as follows: the subject was seated facing four display units and
four response buttons. On a trial, a sin,gle numeral was shown in one
of the units. The subject responded by pressing the corresponding button (the manual task) and by saying a letter corresponding to the numeral: A for 1, B for 2, C for 3, and D for 4 (the verbal task). The
information associated with each of the tasks was varied in a series of
experimental conditions. In the simplest condition (0 bits of information
for both the manual and the verbal tasks), the numeral "1" was shown
on every trial, always in the same display unit. Whenever the stimulus
was shown, the subject pressed a button with the same finger and said
"A," as fast as he could. In the most complex condition (2 bits of information for both the manual and the verbal tasks), anyone of the four
numerals could appear in anyone of the four units. The information of
both tasks was varied in a complete factorial deSign, with three levels of
uncertainty (0, 1, 2 bits) represented for each task. The subjects also
performed in Single-task conditions, where only one response (verbal or
manual) was required. Unfortunately, the single-task conditions were
always tested early in practice and cannot be compared unequivocally
with the double-task conditions.
Schvaneveldt compared the observed latencies of manual and ver-
158
bal responses to two theoretical models: (1) An independence model,

according to which the latency of the verbal and the' manual responses
should depend only on the number of choices available for each of these
responses. In this model, the reaction time for a response should be the
same, regardless of whether that response is performed singly orin conjunction with another response. (2) A single-channel or successiveness
model, according to which the two responses can only be performed in
strict succession. In this model, the latency of the slower of the two
responses in each double-task condition should equal the sum of the
latencies of both responses in the corresponding Single-task conditions.
Both models failed. The latency of each response was found to
depend on the complexity of the other, contrary to the independence
model, and the latency of the slower response was far shorter than the
sum of single latencies, contrary to the successiveness model. Part of the
data were displayed in the manner of Figure 9-1. This figure, indicates
an interaction: the verbal latency increases with the information of the
verbal task, but the rate of increase depends on the complexity of the
manual task. Specifically, the number of choices in the verbal 'response
has relatively little effect on RT when the manual response also involves
a choice. Schvaneveldt (1969, p. 296) concluded: "The interesting result
. . . is the suggestion of increasing overlap between two tasks as the
uncertainty in the tasks is increased." It seems natural to infer from
these results that the decision processes involved in the two tasks can
occur in parallel with little or no interference (Keele, 1973).
A different approach to the same results is possible, if one rejects
the basic assumption that the two responses correspond to distinct tasks
that the subject performs in parallel. Let us consider the alternative
possibility that the verbal and manual responses are components of a
molar response unit, much as the distinct movements of hand and foot
are components 'of the act of switching gears while driving and automobile. In this view, the subject makes a single decision on each trial
"press-this-button-and-say-that-letter." Note that this hypothesis does not
entail that the responses should occur in precise synchrony, since they
involve different muscle groups. If each molar response,' is controlled by
a single decision process, however, the latencies of both components
should vary in unison with the latency of the decision. Thus, the interval between the two responses (IRI) should remain about the same
under all conditions. In contrast, an independence model in which the
latencies of the component responses ar~ the same in single-task and
dual-task conditions, entails large variations of IRI across conditions:
from +328 milliseconds (where the verbal response involves a 2-bit
decision and the manual response is simple) to -163 milliseconds (where
the'verbal response is simple' and the manual is complex).
159
fv4tcnual information:
0 bits
1 bit
2 bits
700
600
u
eu
en
E
eu
E
.... 500
c
.2
....
u
eu
L-
'0
...0
400
L-
eu
>
0
300
200 '-----'-
__'__
...a..-.
2
Verbal information (Bits)
FIGURE 9-1
Verbal RT as a function of information in verbal response, for three
levels of information in manual response (from Schvaneveldt, 1969,
with permission).
Another implication of the t:esponse-grouping hypothesis is tllat the

latency of the compound decision should depend on the overall complexity of that decision, i.e., on the total number of different response
patterns among which the subject is required to choose. For example, the
latencies of both the verbal and the manual responses, should be about
the same in the three experimental conditions that require a choice
among four alternative compound responses, namely the conditions in
which: (1) the verbal and the manual responses each involve a choice
between two alternatives; (2) the verbal response is a four-alternative
choice and the manual response is simple; (3) the verbal response is
simple and the manual response is a four-alternative choice.
160
The data relevallt to these predictions are shown in Figure 9-2,

which presents the average RT of the manual and verbal responses in
the nine conditions of Schvaneveldt's experiment. The grouping hypothesis fares rather well. The verbal response is consistently slower than
the manual, regardless of the complexity of the choice associated with
each of these responses. Furthermore, the latencies of the responses are
clearly dependent on the overall complexity of the compound choice.
Note that the verbal latencies recorded in Figure 9-2 are the same data
that were presented in Figure 9-1. What appeared to be an interaction
in the former display now appears as a nonlinear trend relating verbal
RT to total information.
While the major predictions of the grouping hypothesis are con-
700
0
0
600
,.......
Cl)
en
Cl)
....
500
c:
2
....
~
400
300
...
cCl)
...
...
o
Total information (Bi ts)
FIGURE 9-2
Manual RT (filled symbols). and verbal RT (unfilled symbols) as a function of
total information. Symbols identify the information in the manual respon,se: 0
bits (triangles); 1 bit (squares); 2 bits (circles).
Speeded Responses to Simultaneous and to Immediately Successive SigMls
161
firmed, there is a trend in Figure 9-2 that this hypothesis does not predict: the IRI between the two responses tends to increase with the
complexity of the task. It appears that the organization of the compound
response becomes looser at a high level of complexity. Why this occurs is
not clear. Certainly, however, the implication of this result is that there
is less overlap between the processes leading to the two responses when
the situation is complex than when it is simple. This conclusion is diametrically opposite to the conclusion that Schvaneveldt drew from consideration of the data of Figure 9-1.
The preceding discussion shows that the question of whether two
distinct responses constitute two tasks or one, cannot be dismissed as a
matter of definition. It is an empirical question. Integrality of responses
has observable consequences, and response units must be discovered,
not defined. In Schvaneveldt's study, there were two indications that
response . grouping or integration occurred. First, the latency of the
manual response, which was always the first to occur, depended equally
on the information conveyed by both the position and the identity of the
numeral. Second, the IRI between the two responses was relatively short
and varied only within narrow limits. However, the fact that IRI did
vary systematically with the overall complexity of the task suggests that
response integration may be a matter of degree.
Phenomena of grouping and organization are as important in, the
context of response as they are in the context of perception. Re~ponse
grouping and integration extend over both space and time: complex
coordinated acts such as shifting gears in an automobile involve different
limbs and a relatively prolonged sequence of subordinate activities. As is
also true of perceptual organization, response organization is hierarchical, and response units are integrated in groups of increasing size. It is
often. easy to discover the size of the dominant unit of organization.
Speak aloud, for example, and try to obey the instruction "say everything twice." What did you discover? What was the size of the units
that you chose to repeat? Almost invariably the repeated unit consists
of more than one word, though the words are clearly present as distinct
subordinate units. The effect is not restricted to verbal response. Set
yourself to make free-form movements with both hands. Now try to "do
everything twice." The analogy of the motor experience to the verbal
will be clearly evident.
The isolation of valid response units is an essential prerequisite
to the study of divided attention in motor performance. It is only meaningful to speak of attention as divided among isolable processes, but
these isolable processes must first be discovered (Posner, Lewis & Conrad, 1972). The discussion of perceptual attention in earlier chapters led
162
us to reject the concept of cnanneI and to 'De highly sKeptical of the

psychological validity of arbitrarily defined dimensions. The analysis of
Schvaneveldt's experiment suggests that we must be equally skeptical of
arbitrarily defined response units.
The discussion of perceptual units in an earlier chapter suggested
that the rules of grouping are relatively impervious to learning. Response
units, on the other hand, are often fashioned in prolonged experience.
The acquisition of complex skills consists in large part of the formation
of extended units. An impressive example of this process was described
by Seibel (1963). He employed a display,often bulbs, any of which was
equally likely to be illuminated or left dark on each trial. The subjects
responded to. the pattern of lights by pressing corresponding keys with
the fingers of both hands. After extremely prolonged practi~e (75,000
trials I), RT no longer depended on the number of keys that were
pressed, or on the "size of the ensemble of possible patterns. Apparently,
the pattern of lights was perceived as a unit and elicited a unitary, integrated response.
In many situations, response integration does not take place. When
the stimuli are not expected to occur togetl)er, the responses to them
will tend to be successive (Dimond, 1971), or only one response will
occur (Avner, 1972; Colavita, 19"71; Moray, 1970a, b; Moray & O'Brien,
1967). Finally, response grouping can be prevented when the two stimuli
are separated in time. Subjects" can obey the instruction to respond to
the first stimulus without waiting for the second, even if the interval between the two stimuli. is as brief as 50-100 milliseconds (Sanders &
Keuss, 1969). In that case, the response to the second stimulus is often
delayed. The next sections review some of the research conducted in this
paradigm.
INTER- RESPONSE I NrnRvAL
.AND TIlE
PSYCHOLOGICAL REFRACfORY PERIOD
A vast amount of research has been devoted to the question of how

attention is allocated when two distinct stimuli demand r~sponses in very
rapid succession. Crai.k (19'47, 1948) is usually credited with the main
discoveries and the first theoretical formulation in this area. He studied
a tracking task in which the target followed a course that jumped from
one level to another at variable intervals (Vince, 1948). The basic finding'
was that, whenever two signals followed one another within 0.5 seconds,
the reaction to the second signal was markedly delayed. The interval
163
between the two signals was the only effective variable; the magnitude
of the second signal did not seenl to matter. This result led C'raik to
describe the intermittence of corrective processes by the term "psychological refractory period" (PRP) which had been introduced earlier
(Telford, 1931). Craik suggested that man behaves as a,rl intermittent
servomechanism; the main characteristic of such a mechanism is that
the corrections it makes when performing a continuous action are discrete, and limited in rate. Information that arrives during the refractory
period which follows each correction is acted upon only at the next
instant of sampling.
Subsequent investigations of refractoriness have largely abandoned
the tracking task in favor of the simpler situation in which the subject
reacts to two rapidly, successive stimuli, Sl and S2. The sequence of
events in a typical trial is shown in Figure 9-3. The question that is
raised in such experiments is whether the subject can prepare the response (R2) ,to the second stimulus (S2) while engaged in preparing or
executing the response (R1 ) to the first stimulus (Sl).
The data of a reaction-time experiment in the refractoriness paradigm are usually plotted as in Figure 9-4, in which RT2 is pl9tted as a
function of the interval (lSI) between Sl and S2. Figure 9-4 presents
theoretical functions for the dependence of RT 2 on lSI, which are derived from the single-channel hypothesis, as formul~ted by Welford
(1952, 1959', 1967) and by D'avis (1957). This hypothesis is an application
of Craik's original view to the reaction-time situation. The mainassumption of single-channel theory is that the response-selection stage of information-processing is a bottleneck, or single channel, which can select
responses only one at a time. The one exception that Welford admitted
RT 2
(,----------')...'--------.....1
RT 1
r".-------A"------I----.....'
S1
S2
Jl~
\.
R1
R2
n------n-----_--rL
l'-
-y---- _ _.........)
I RI
'S I
T i m e - - -.....
)
FIGURE 9-3
The sequence of events in a typical trial of a refractoriness study: 8 1 and 8 2 are
successive stimuli; R1 and R2 are the corr~spondingresponses.
164
,
""
""
""
"
"
( B)
",
""
""
""
""-
"""-
""---------.----a.------.
" . . .- ------
Control RT 2
RT 1
---'A""---------.-.\_ ~
I,--
Time--..
)
FIGURE
Feedback
delay
Inter-stimulus interval ( 151 )
9-4
Reaction time of second response (RT 2) plotted as a function of interstimulus interval (lSI). Two predictions from single-channel theory are
shown: (A) prediction on the assumption that the processing of 8 2 begins
with the execution of Rt ; (B) prediction on the assumption of a further
delay for processing feedback.
was the occurrence of response grouping when8 t and 8 2 are nearly

simultaneous.
The .simplest version of the single-channel hypothesis asserts that
the decision-mechanism becomes available to prepare the second response only after the first response is made. This prediction is labeled A
in Figure 9-4. When 82 is shown before R t occurs, it is held ~n storage,
and processing of 82 begins only with the occurrence of Rt In this
view, a stimulus 82 that preceded R1 is treated as if it had in fact been
simultaneous withR t Accordingly, RT2 should be normal for all 82 that
occur after Rt , and it should be delayed for all 8 2 that precede Rt . The
165
function shown in Figure 9-4 is idealized; in actual data, the variability

of RT1 would be expected to cause a smooth transition between the
two arms of the RT function (Bertelson, 1967; Welford, 1968).
In many experiments RT 2 is longer than normal even when S2 is
presented after the occurrence of R1. Davis (1957) attempted to explain
these additional delays by postulating an additional central refractory
state which lasts about 100 milliseconds. Welford (1952, 1959, 1967)
proposed that the system may be occupied for some time by feedback
from the first response, and he also advanced the interesting suggestion
that waiting for feedback from R1 is optional: it is most likely to occur
when the' execution of R1 demands a high degree of precision, or in
early stages of practice. The prediction from single-channel theory which
incorporates the additional- assumption of post-response delay is labeled
B in Figure 9-4.
Single-channel theory asserts that division of attention between response processes is impossible. This is a surprising contention, in view of
the vast amount of ~vidence indicating that attention is often divisible.
Indeed, there are very few experimental reports in which the d-ata fit the
theoretical predictions that were illustrated in Figure 9-4; in most studies
the discrepancies betwe~n observations and predictions are large and
systematic. Nevertheless, single-channel theory has often been viewed
as the dominant theory in this area (Bertelson, 1966; Smith, 1967b).
The survival of single-channel theory in the face of massive contradictory evidence can be traced, at least in part, to the tradition of plotting experimental results in the manner of Figure 9-4, whereRT2 is
shown as a function of lSI. It is equally reasonable, however, to formulate the experimental question as follows: How does the interval between the two responses R1 and R2 (IRI) vary with the interval between
the two.stimuli, S1 and S2?
This formulation suggests that experimental data should be displayed as in Figure 9-5, which again presents two alternative predictions
from single-channel theory. According to that theory, IRI should be
constant up to a value of lSI which is equal to R1 (version A) or larger
(version B). Beyond that point, IRI should rise directly with lSI. Note
that a constant IRI is predicted by two very different hypotheses: response grouping and single-channel operation. However, the singlechannel hypothesis also entails that IRI should be relatively long, and
that the latency of R1 should be independent of the complexity of the
subsequent response, R2. As was shown in the preceding section, the
grouping hypothesis entails that RT 1 and RT ~ should vary in unison
whenever the complexity of either component of the task is altered.
As an illustration of the two modes of analysis of refractoriness
data, consider an experiment by Smith (1969), in which the subject was
166
( B)
o
c:
c
Q)
Feedback delay
Q)
(A)
o
a.
U)
Q)
'I
'-
....c
Q)
Control RT 2
Q)
RT 1
t.=
r
Time
Feedback
delay
Inter-stimulus interval
FIGURE 9-5
Inter-response interval as a function of inter-stimulus interval. 'J;he
predictions from single-channel theory correspond to the functions
labeled (A) and (B) in Figure 9-4.
required to make two choice-responses in quick succession. The second

response (R2 ) was always a two-alternative choice. The experimental conditions varied in the complexity of the first response (R 1 ). The three conditions studied were 2-2, 4-2, and 8-2, where the two numbers in
each pair refer to the number of alternatives for R1 and R2 respectively.
Figure 9-6 presents the main results of this experiment, analyzed
and plotted in terms of RT2 (panel A) or IRI (panel B). The two panels
respectively correspond to the graphical representations introduced in
Figures 9-4 and 9-5. The data of panel A appear at first glance to correspond quite well to the predictions from. single-channel theory, and
they were interpreted as supporting that theory (Smith, 1969).
To draw the results in panel B, IRI was computed for each data
point separately, because RT1 varied slightly as a function of lSI and
of the complexity of R:!. The equation for the .computation is:
IRI == RT 2
+ lSI -
RT 1 (see Fig. 9-3).
( A)
3 bits
2 bits
1 bit
167
700
400
--0-- Control RT
2
--0-- Control RT
400
B)
o
u
Q)
300
200
50
150
300
500
Inter-stimulus interval (msec)
FIGURE 9-6
Data from Smith (Acta Psychologica, 30, 1969). RT 2 and IRI are shown
as a function of lSI. The parameter is the information of R1
As plotted in Figure 9-6, panel B, the data are seen to violate drastically
the predi~tions of single-channel theory, since the horizontal segment predicted by that theory is missing in all cases. The discrepancy between
the impressions that are gained from observing the two panels of Figure
9-6 is due to a simple fact of sensory discrimination: we are much
168
more sensitive to deviations of a line from the horizontal than to deviations from a slope of minus one!
The results of Figure 9-6 are incompatible with single-channel
theory for two reasons: (1) because they indicate that IRI can be shorter
than the control value of RT 2, so that some processing must be parallel;
(2) because the slope of the functions that relate IRI to lSI is always
positive, again indicating parallel processing. These. deviations from
single-channel predictions are much too large to be explained by random
fluctuations of RT1
The slope of the function that relates IRI to lSI in Figure 9-6, panel
B, is a meaningful parameter: it represents the amount by which IRI may
be shortened (in msec) if the presentation of S2 is advanced by one millisecond. Both single-channel theory and a grouping hypothesis entail a
slope of zero for" the range of short lSI's. On the other hand, the hypothesis that the processes leading to the two responses are completely independent entails that the slope of the function should be unity. The result
shown in Figure 9-6B is typical: the slope of the IRI function is positive
throughout, and the function is pOSitively accelerated. This result is incompatible with the three hypotheses that have been introduced in this
discussion: single-channel theory, and the grouping and independence
models. The results imply that some attention is devoted to the processing of S2-R2 as soon as S2 is presented. Furthermore, the amount of
attention devoted to S2 increases steadily during the latency of R 1
These results are typical of a large number of studies of refractpriness
(e.g., Bertelson, 1967; Broadbent & Gregory, 1967, exp. 1; Nickerson,
1967; Sanders & Keuss, 1969).
It may be noted in Figure 9-6 that the slope of the IRI function
varies inversely with the complexity of R1 :IRI increases more slowly
with lSI when R 1 is complex than when it is simple. Since the slope of
the IRI function reflects the rate at which S2 is processed, this finding
appears to support an effort theory, which entails a reduced sharing of
capacity when one of the two competing activities is highly demanding.
However, a more fundamental observation is that the shortest IRI is
almost the same at the three levels of complexity. This result suggests a
modified concept of refractoriness, i.e., that there is a minimal interval
that separates successive responses when these responses are not grouped.
If such a minimal IRI is a basic feature of the system, the divergence of
the curves follows necessarily, as the following argument shows. At a low
value of lSI, both RT1 and RT 2 are affected by a change in the complexity of R 1 , but IRI is the same for different levels of complexity. At a high
value of lSI, on the other hand, only RT 1 is affected by the complexity of
R 1 and IRI is consequently longer when R 1 is Simple then when it is
169
complex. Between the two values of lSI, therefore, the slope of the IRI
function must be generally steeper for the simpler R 1 Since this result
follows necessarily from ,the assumption of a common minimal value of
IRI, the temptation to interpret the differences between the slopes of the
curves must be firmly resisted.
There is additional evidence for the notion of a minimal IRI between ungrouped responses. Karlin and Kestenbaum (1968) carried out
an experiment very similar to that of Smith (1969). They studied five
different combinations of RT tasks. In the notation introduced earlier,
the tasks were: i-2; 2-2; 5-2; 1-1; 2-1. The minimal values of IRI
were almost the same for' all conditions: they varied only from 220
milliseconds (for the 1-2 condition) to 244 milliseconds (for the 2-1
condition). The data were generally very similar to those shown in Figure 9-6: the slope of the IRI function was positive in all conditions and
at all values of lSI, and the curves diverged systematically as a function
of R 1 complexity. The complexity of R2 , on the other hand, had very little
effect on the IRI functions.
Keele (1973) has emphasized the importance of these observations
by Karlin and Kestenbaum, and he made them the cornerstone of a general view of attention. Although he did not analyze the data in terms of
IRI, it is probably easiest to present his approach in such terms. 'In his
view, the finding that the minimal IRI does not vary greatly with the
complexity of responses indicates that the processes leading to the two
responses interact only at the stage of response initiation, while earlier
operations occur in parallel and without interference. Thus, Keele separated the stages of information-processing into two sets: (1) perceptual
analysis and memory retrieval (including response selection); (2) initiation and execution of responses. He suggested that the earlier operations
occur in parallel and without interference because they require no attention. Only response-related operations, such as rehearsal or the initiation of overt responses, demand attention and are mutually interfering.
The constancy of the minimal IRI with variations of response complexity
is COllsistent with this hypothesiS of a conflict at the stage of response
initiation.
Keele's position that the processes of perception and retrieval do
not depend on attention is similar to the views of Deutsch and Deutsch
(1963) and Norman (196'8), which were found inadequate in preceding
chapters. However, the finding which Keele emphasizes, i.e., the near
constancy of minimal IRI over experimental conditions, does appear to
be of fundamental importance. Perhaps this was the kernel of truth in
the original hypothesis of psychological refractoriness. If the minimal IRI
is independent of response complexity, the single channel cannot be 10-
170
cated at the stage of response selection, as classical single-channel theory

would have it. It must be a feature of response organization. Unfortunately, however, there is not enough information concerning the
generality of this effect. There appear to be conditions where responses
to independent stimuli, apparently ungrouped, nevertheless occur in very
close succession (Posner, personal communication). There is a basic, unsolved problem here.
While this section ends ona note of doubt, it may be useful to review its more positive conclusions. It was suggested that a more incisive
analysis of the refractoriness paradigm is possible when IRI, rather than
RT2 , is adopted as the basic dependent variable. Two parameters of the
IRI function were isolated: the minimum value> of IRJ (typically observed when lSI is very short) and the general slope of the function. The
slope is positive in most studies of refractoriness, but it is almost always
less than one. This finding is incompatible with the hypotheses of strict
successiveness (single-channel), independence and grouping. Thus, while
there must be substantial temporal overlap between the processes
elicited by S1 and S:!, these processes do interact. Another finding is that
the shortest IRI is sometimes approximately constant in different experimental conditions. This result is compatible with the existence of a state
of motor refractoriness following R1. Keele has inferred the more radical
conclusion that competition occurs only at the stage of .response initiation, but the next section will present some evide.nce against this hypothesis.
REFRACCORINESS AND EFFORT
The main conclusion of the preceding section was that subjects in

the r~fractoriness paradigm usually allocate some capacity toS 2 as soon
as it.is shown, well before R1 is completed. This is contrary to any singlechannel theory. In isolated cases, however, the predictions of singlechannel theory are quite strictly upheld. It is therefore of interest to
isolate the conditions under which this finding is obtained.
Figure 9-7 is redrawn from a study by Broadbent an-d Gregory
(1967). In that experiment, the subject was first shown one of two lights
on one side, to which he responded by depressing one of two keys. He
was then shown one of two lights on his other side, to which he responded with the other hand. The instrllctionsand the knowledge of
results given after each trial defined the response to the first of the two
signals as the primary task. Two different conditions are shown in the
figure. In one (broken line), the responses to the lights were compatible
Speeded Responses to Simultaneofls and to Immediately Successive Signals
171
700
600
~
Q)
500
E
..........
c
c:
~
c
400
<J)
V)
c
o
a.
V)
300
'-
....c
Q)
200
Controt
100
100
200
300
400
500
Inter-stimulus interval {msec}
FIGURE 9-7
Data from Broadbent and Gregory (1967), with permission of The Royal Society.
with the stimuli: the subject pressed the key located under each light
that was Hashed. In the other condition (solid line), the S-R relations
were incompatible in both R1 and R2 because the subject was required to
press the key under the light that was not Hashed in each pair.
The results in the two conditions are markedly different. When
the two reactions are compatible, there is considerable parallel processing, as indicated by low values of IRI and by the fact that the slope of
the IRI function is consistently positive. When the reactions are incompatible, in contrast, there is no evidence of any sharing of capacity during the first 250 milliseconds of exposure. In addition, the processing of
the second reaction is relatively ineffective even after R1 is completed,
as indicated by the fact that the IRI function is of less than unit slope.
The shortest IRI in that condition is substantially longer than the con-
172
trol value of RT2, indicating a prolonged disruption of th~ second reaction by the occurrence of the first.
These results are not consistent with the theory proposed by Keele
(1973). If the processes that precede response initiation require no attention and occur in parallel, there is no special reason for a variation of
stimulus-resp'onse compatibility to affect the IRI function. The locus of
difficulty in an incompatible situation is in the stages of retrieval and
selection of the appropriate response, which Keele assumed to be
non-attentive. At the stages of initiation and execution, compatibility
should have no further effects. Thus, Keele's theory lacks a mechanism
that would explain the effect of compatibility on the minimal value of
IRI and on the slope of the'function.
The qualitative difference between the two IRI functions of Figure
9-7 suggests a far-reaching conclusion: perhaps no model can be correct
which assumes that the processing of 8 1 and 52 is necessarily parallel,
nor can a model be correct which assumes that processing of such stimuli is necessarily serial. Models that assume a consistent mode of operation under all conditions may be termed "hardware" models. They
attempt to explain the results in terms of the structure of the machine.
However, the machine seems to be able to organize its operation in different ways. Thus the device with which we are concerned is capable of
purely serial processing on some occasions, and of parallel processing on
others. The behavior of such a device in any situation is, perhaps better
explained by reference to the program which governs its opera~on than
by assuming that its function necessarily mirrors its st:fucture. This is
not to deny that structural limitations exist, but merely to state the obvious point that the observation that a system behaves in a certain mannerdoes not' imply 'that the system must behave in that manner. When
the operation of the system is shown to be qualitatively different in different conditions, its behavior in anyone condition is best explained in
terms of software, program, or strategy. The use of the concept of allocation policy in the present work is intended to suggest such an approach to attention. In Chapters 7 and 8, it was shown that man can both
focus, and divide his attention, within certain limitations that depend on
the task and on the circumstances. It should not be surprising to observe
a similar flexibility in the allocation of capacity in the context of successive speeded responses.
A study in the refractoriness paradigm carried out in my laboratory (Kafry, 1971) led to the fortuitous discovery of another case in
which a slight modification of experimental conditions causes a qualitative alteration in the allocation of capacity. Kafry investigated refractoriness in the R5I case, i.e., the experimental situation in which 52 is always
presented some time after the occurrence of R1 (Rabbitt, 1969; Triggs,
173
1968). She was looking for possible refractoriness effects following a response (Davis, 1957; Welford, 1959).
In all the experiments in her study, R2 was a compatible threechoice response to one of three lights, executed with the right hand. In
the two conditions shown in Figure 9-8, the subject's first task (R1 ) was
to stop a digital millisecond counter as close as possible to a specified
value (600 or 1200 msec in these data). The counter started at zero on
each trial, and the subject stopped it by depressing a key with his
left hand. The key-press caused 8 2 to appear, either immediately
(RSI = 0) or after a variable delay. Figure 9-8 includes data for two
groups of subjects who differ markedly, on the average, in psychomotor
.
skills: 20 untIergraduates and 20 Hight cadets.
The results for both groups were very similar: the "600" condition
caused total refractoriness for about 200 milliseconds following the
First task- 11600 11
900
-...
... ...
Q)
0---0
First task- n1200"
E 800
......,
o
c:
CI>
en
c:
8.en
700
~
I
'CI>
....
.E
600
500
100
200 .
300
400
Response-stimulus interval (msec)
FIGURE 9-8
Data from Kafry (1971), with permission. Filled circles: data of 20 students.
Unfilled circles: data of 20 flight cadets.
174
response, as if a stimulus that was shown at R8I == 0 did not become

effective until 200 milliseconds later. In contrast, there was no indication
of absolute refractoriness in the "1200" condition, where the effectiveness
of processing, measured by the slope of the line,/ was about 75 percent
of normal for several hundred milliseconds after the occurren'Ce of Rl .
The subjects' reports give a clue to the nature of the qualitative difference between the two conditions. In the "1200" condition, the subject
has time enough to prepare for both tasks. Indeed, he may throw anticipatory glances at the lig~t display while the counter runs. In the "600"
condition, on the other hand, attention is riveted continuously on the
counter, since the subject must "decide" -to press the key when he sees
the counter reach about 400, in order to execute the movement at the appropriate time. The inability to prepare for the next task is reflected in
the prolonged refractoriness of this condition.
The two experiments that have been discussed in this section suggest that attention is focused exclusively on the first stimulus-response
task only when that task is exceptionally difficult. When the first task is
easier, some attention is diverted to the execution of the second task or
to preparations for it, and the typical rising IRI function is observed.
OTHER FINDINGS AND THEORIES
The interpretation of the refractoriness paradigm as a special case

of divided attention is similar in some respects to the response-conflict
theory of the psychological refractory period which was originally presented by Reynolds (1964, 1966) and vigorously supported by Herman
and Kantowitz (1970). This theory proposes that 8 1 and 8 2 elicit response tendencies that are likely to conflict. The responses to both stimuli
will be retarded when such a conflict occurs, but it is assumed that the
prepotent response suffers the smaller delay. A response is prepotent
either by instruction or because. it was already in preparation when conflicting tendencies were aroused. The latter factor,' of course, always
favors R l over R2 , and it explains why RT2 is relatively slower than
RTI in the double-stimulation paradigm. Response-conflict theory leads
to the prediction that the. interaction will be most detrimental if 51 and
52 are 3:ssociated with incompatible or ~ntagonistic responses. This prediction has been confirmed (Herman & Kantowitz, 1970).
The present interpretation of refractoriness and response-conflict
theory shares the assumption that 8 1 and 8 2 can be processed in parallel.
It is not obvious, however, how a response-conflict theory could account
for the effects of task demands that were illustrated in the, preceding
section. In addition, response-conflict theory cannot readily explain the
175
finding of major delays ofR2 in the RSI design (Kafry, 1971; Rabbitt,
1969), since there are no conflicting tendencies when S2 is presented
after the completion of R1. The simplest explanation of these delays is
that the preparation for a subsequent stimulus and a subsequent response demands effort. Under some conditions (see, e.g., Fig. 9-8), this
preparation is precluded during the processing of another response.
Response-conflict theory and the limited capacity hypothesis both
suggest that R1 should be somewhat slower in the double-task situation than when it performed alone. Results. confirm this expectation.
Many studies have reported the consistent finding that the reaction to
the first stimulus is slower in the double-task paradigm than when a
single stimulus is presented (Bertelson, 1967; Broadbent & Gregory, 1967;
Gottsdanker, 1969; Gottsdanker, Broadbent & Van Sant, 1963; Herman &
Kantowitz, 1970; Nickerson, 1967; Smith, 1967c; Triggs, 1968). The delay
is usually quite small (around 30 msec). The delay of R1 has been foun'd
to vary inversely with lSI in some experiments: when S2 followed 8 1 in
quick succession, RT1 was slow (Herman & McCauley, 1969). The delay
of R1 also increases with the complexity of the processing that S2 and R2
require (Karlin & Kestenbaum, 1968). The competition between the
processes leading to the two responses is further connrmed by the observation that the speed of R1 and R2 can be manipulated by instructions: as one of these responses is made faster, the other correspondingly
slows (Triggs, 1968). Herman and Kantowitz (1970) have reviewed these
effects in detail.
An important observation that must be considered in explaining refractoriness is that a stimulus which does not require a response can
nevertheless delay the resp(onse to another stimulus. Thus, a large number of studies have s'hown that the.interpolation of an irrelevant stimulus
82 after S1 causes H1 to be delayed (Davis, 1959', 1962; Elithorn, 100'1;
Fraisse, 1957; Kay & Weiss, 1961; Nickerson, 1967; Rubinstein & Rutschman, 1967; Sinith, 1967a). The delay is small (usually 40-60 msec), and
its interpretation is controversial (Bertelson & Tisseyre, 1969; Davis, 1959;
Herman, 196~). A larger delay has been observed where a stimulus 8 1
inhibited a response. When S2 was presented shortly after such an inhibitory stimulus, RT2 was longer than normal (Sanders & Keuss, 1969). These
results are consistent with a theory of limited and shared capacity, but
they are. also easy to explain within a response-conflict theory.
Bernstein (1970; Bernstein, Clark & Edelstein, 1969a, b) has reported the ihterestingfinding that visual .RT can be facilitated by
presenting a loud auditory stimulus some time after the relevant vishal
stimullls. A plausible explanation of this effect is that the tone increases arousal and therefore 'facilitates ongoing processes. When the second
176
stimulus is associated with competing response tendencies, RT is delayed

(Herman, 1969).
A complete account of the interactions between stimuli and response in the double-stimulation paradigm must also include the effects
of expectancy and preparation. In general, a subject's RT is shorter if
the signal to respond arrives precisely at the instant it is expected. This
is the expectancy effect. In addition, there is an effect of the foreperiod
that is available for preparation: following a warning signal, it takes
about half a second for a subject to be at his best. In the double-stimulation paradigm, the occurrence of Sl provides a warning that S2 will SOon
occur. The readiness to respond .to S2 will therefore increase gradually,
reaching a maximum no sooner than 500 milliseconds afterS1. Furthermore, if the average value of lSI is longer, the gradient of maximal
, readiness for S2 will shift toward the value of lSI at which S2 is most
likely to occur.
There has been a major attempt to describe the so-called refractoriness effect in terms of expectancy and preparation (Adams, 1962;
Elithorn & Lawrence, 1955), but comprehensive reviews of the evidence
have concluded that this attempt was unsuccessful (Bertelson, 196'6;
Nickerson, 1967; Smith, 1967b). In accordance with the predictions of
expectancy theory, the average lSI affects RT2 in the double~stimulation
paradigm (Adams, 1962), suggesting that Sl functions as a warning signal
which causes the readiness for S2 to increase, as in the foreperiod effect.
However, the foreperiod effect in the single-stimulus case is smaller
than the refractoriness effect, and therefore insufficient to account for it
(Shaffer, 1968). Furthermore, refractoriness occurs in the absence of
temporal expectancy effects, e.g., when the interval between the two signals is constant (Borger, 1963; Creamer, 1963). Thus, expectancy can be
ruled out as a general explanation of refractoriness effects. Nevertheless,
the idea that preparedness for a stimulus and for a response vary in
time cannot be neglected, particularly in the explanation of refractoriness
in the RSI paradigm, i.e., where the occurrence of S2 follows the execution of R1 (Kafry, 1971).
REVIEW
This chapter was concerned with the organization of performance

in tasks that require two speeded responses. The suggestion was advanced that an analysis of the interval between the two responses (IRI)
is often more illuminating than separate analyses of their latencies. The
occurrence of response-grouping is indicated by an approximate constancy of IRI over conditions, and by a relatively low value of IRI: Evi-
177
dence for response-grouping was found in a reanalysis of a study by

Schvaneveldt (1969). More generally, it was proposed that the separability of the processes that lead to physically distinct responses must be
demonst~ated empirically, not assumed.
The application of an IRI-analysis to data in the refractoriness
paradigm suggested that some results which have been interpreted as
supporting single-channel theory actually provide conclusive evidence
against that theory. Typical results in the double-reaction paradigm indicate that processing of S2-R2 typically begins as soon as S2 is presented,
and continues at an accelerated rate throughout the' latency of R t .
An unexpected result observed in two studies (Karlin & Kestenbaum,
1968; Smith, 1969) is that the minimal IRI between ungrouped responses
does not seem to depend on the complexity of the interacting responses,
at least within the range of complexity included in these studies. The
minimal IRI may correspond to a state of motor refractoriness.
Keele (1973) has argued from these observations that mental operations of perception, memory retrieval, and response selection require
no attention and can be performed in parallel. His theory cannot account
for the isolated conditions in which the predictions of single-channel
theory are quite strictly upheld (Broadbent & Gregory, 1967, exp. 2;
Kafry, 1971). A strategy of strictly serial processing was adopted in these
experiments when the first response task was exceptionally difficult. No
structural theory which assumes that processing is always serial, or
always parallel, can account for these results, which tend to support the
concept of a flexible policy of attention allocation.
Addition,al compleXities of the refractoriness paradigm were briefly
discussed in the last section. The concepts of preparation, expectancy,
and response-conflict must be included in a comprehensive account of
results in this paradigm. The refractoriness paradigm appears to be too
complex to provide definite tests of theoretical positions concerning the
division and the focusing of attention.
10
Attention and
Task Interference
We ofteJ?- find it exceedingly difficult to execute. two activities together,

although each. alone is easy. This mutual interference between concurrent tasks is som.etimesexplained in structural terms, on _.the. assumption
that the competing tasks simultaneously elicit inco-mpatible responses, or
impose simultaneous demands on sp-ecific perceptual or motor mechanisms. An effort theory seeks to explain interf~rence in terms of a competition for a general limited capacity. This chapter reviews evidence
which shows that concepts of capacity and of structure are both needed
to explain the phenomena of interference. The results of sqme studies
of dual-task performance are interpreted in terms of the effort theory introduced in Chapter 2, and the reader may find it useful to quickly scan
the illustrations of that chapter before reading the present one.
CAPACITY INTERFERENCE
A theory which identifies attention with effort and with a limited

capacity entails two predictions concerning interference between concurrent activities: (1) interference will arise even when the two activities
178
Attention and Task Interference
179
do not share any mechanisms of either perception or response; (2) the

extent of interference will depend in part on the load which each of the
activities imposes, i.e., on the demands of the competing activities for
effort or attention.
The support for both propositions is overwhelming. The activities
of walking and mental arithmetic, for example, are as distinct as can be.
Nevertheless, the following experiment usually succeeds: while walking
casually with a friend, ask him to perform a complex .operation of mental
arithmetic; he is very likely to stop in his tracks. Even the highly automated act of walking apparently demands some central capacity.o Another: example that was introduced earlier is the combination of driving
and conversing. The conversation is interrupted when the demands of
the driving activity become critical.
There is much' experimental documentation for task interference
that arises from capacity overload. Thus, Posner and Rossman (1965)
asked th.eir subjects to retain three letters for a brief interval, during
which they engaged i~ mental tasks of varied complexity. The amount
of retention decreased regularly with increasing difficulty of the interpolatedtask. Similar results have been obtained by many other investigators, with different combinations of memory task and interpolated
activity (Baddeley, Scott, Drynan & Smith, 1969; Broadbent & Heron, 1962;
Dillon & Reid, 1969; Murdock, 1965; Peterson, 1969). This finding is
most naturally iriterpreted by assuming that rehearsal demands a considerable amount of effort or attention. When attention is preempted by
the interpolated task, rehearsal is disrupted and retention suffers. This
interpretation is not affected by the finding (Reitman, 1971) that a nonverbal interpolated task of signal detection may effectively prevent rehearsal without destroying the 'menlory of the stored material. This
finding implies that rehearsal is not always necessary to retard forgetting.
When required rehearsal is precluded by concurrent activities, retention
suffers.
Johnston, Greenberg, Fis'her, and Martin (1970)' employed tracking
as a subsidiary task in several studies of memory. The subjects in one
of these studies were shown lists that they were to recall after a retention interval. Tracking scores during the retention interval were inversely
related to memory load, which presumably 'controls rehearsal activity.
Tracking performance was also related to the complexity of an opera:0 It seems fair to raise the question of why I have been pacing the corridor
while formulating the preceding sentences. The hypothesis I find most attractive is
that walking is often used to pace oneself down, thus slowing the rate of thought-and internal speech so as to minimize confusion. Deliberately slowing down is not
advantageous in activities that impose a high load on short-term memory, such as
mental arithmetic. Accordingly, one tends to stay still while performing such activities.
J80
tion that the subject was asked to perform on material stored in memory (e.g., mentally arrange five words in alphabetical order). Finally,
the activity of verbal recall of stored material caused severe interference
with tracking. This reslllt is consistent with physiological measures of
effort, as well as with other studies of divided attention (Kahneman &
Peavler, 1969; Kahneman & Wright, 1971; Trumbo & Noble, 1970).
In a related study, Shulman and Greenberg (1971) observed that
the probability that a subject would recognize an item in a tachistoscopic
exposure was inversely related to the length of a list,that he was silently
rehearsing at the time. However, the relation between perc~ptual deficit
and memory load appeared to break down when the amount of material
exceeded memory span. This interesting, result confirms the suggestion
that effort no longer increases when a task becomes impossibly difficult.
The same authors also found that reaction time in deciding which of two
lines' is longer is delayed by concurrent rehearsal (Shulman & Greenberg,
1971; Shulman, Greenberg & Martin,1971).
The interaction of learning activity with other tasks may follow
different rules in motor le.arning, which does not involve rehearsal.
Eysenck and Thompson. (1966) reached the surprising conclusion that
concurrent activity disrupts the performanc~, but not the learning, of a
motor skill. Subjects pressed a foot pedal in response to auditory signals
while learning to track on the pursuit rotor. The rate of foot responses
imposed by the auditory signals was varied. The tracking performance
deteriorated as the rate of this interfering response was increased, but
the difference between groups exposed to different levels of distraction
vanished as soon as the distraction was removed. Fo~lowing a rest period,
all groups showed a large reminiscence effect and' precisely identical
tracking ability. Eysenck and Thompson (1966) co,ncluded that attention
is not involved in the acquisition of skill during massed practice. This
provocative conclusion. demands further study.
A study by Keele (1967) provides strong evidence for a hypothesis
of limited capacity. Keele instructed his subjects to turn off a series of
lights; he controlled the difficulty of that task by the compatibility of
the stimulus-response arrangements. In addition, the subjects were asked
to count backward, by one, three, or seven. Measurements of the speed
of both responses indicated some gain from performing the tasks together, when both were easy. The total time required to perform a certain number of responses of both kinds was less when two easy tasks
were combined than when they were performed successively. When the
tasks were both difficult, on the other hand, the attempt to combine or
interweave them resulted in a marked loss of efficiency. As predicted by a
capacity model, the quality of performance on each task decreased regularly with the difficulty of the other.
Attention and Task .Interference
181
182
tasks and the variability which they induced in tapping performance.

Baddeley (1966) measured load by requiring subjects to produce a random series of digits. He observed that redundancy tends to' increase under high load.
The results of these studies are generally encouraging. Nevertheless, Brown (1966, 1968) has listed several limitations in the use of subsidiary tasks to measure load. In particular, he noted that comparisons
of different tasks require great caution, since the amount of disruption
of the subsidiary task depends on the structure of the primary task, as
well as on its difficulty. The subsidiary-task method does not provide a
pure measure of capacity interference, because any particular combination of primary and subsidiary tasks is likely to involve some structural
, interference. Brown noted, for example, that Michon's interval-production task (Michon, 1966) is most severely disrupted by primary tasks in
which the subject responds at a high rate, whereas Baddeley's (1966)
random generation task appears to be sensitive to information la'ad,
rather than to response rate. Tasks that impose a high motor load and
tasks that impose a high perceptual or conceptual load are therefore
likely to have different effects on the subsidiary tasks that Michon and
Baddeley introduced (Brown, Simmons & Tickner, 1967). The conclusion,
of course, is that capacity interference is best measured by means of a
battery of subSidiary tasks, rather than by a single task.
DECISION BO'TTLENECK
OR COMPETITION FOR EFFORT
The evidence of the preceding section is consistent with_ the notion

of a general limit on capacity, but it can also be interpreted in other
terms. Welford (1968) has proposed a single-channel theory, according
to which interference arises in the dual-task paradigm when the two
tasks compete for the control of the response-selection stage. In Welford's theory, this stage is.a bottleneck, which can only deal with one
response process at a time. This theory is formally similar to Broadbent's
filter theory, except for the location of the bottleneck, which Broadbent
placed at the P-system, and Welford placed at a stage which translates
percepts into acts. In both,theories, the effect of complexity on interference is explained in terms of time: the single channel is. occupied for a
longer period by a complex operation than by a simple one, and the
severity of the interference increases with the duration of the delay.
The assumptions of single-channel theory are much more precise
and restrictive than those of a limited capacity model which permits
Attention and Task Interference 183
parallel processing. In particular, 'single~channel theory yields precise

predictions for the refractoriness paradigm that was "discussed in the
preceding chapter. Th~sepredictions, however, have generally failed
to be confirmed.
In the dual-task paradigm, single-channel theory entails that a task
which does not require response selection should neither interfere with
any other task, nOr be subject to .interference. In contrast, the limited capacity hypothesis entails that any two tasks should be mutually interfering to some extent, and that the extent of interference should vary
with effort, r~ther than with requirements of response selection.
Results that support Single-channel theory were reported by
Trumbo, Noble, and Swink (1967), who combined a tracking task with
several other; aGtivities. They found that tracking performance was disrupted equally by tasks' of different difficulty. The following two activities, for example, interfered equally with tracking: a complex learning
task, in which the subject serially anticipated e~ch member in a series
of stochastically dependent .numbers; and an apparently much simpler
task, in which the subject eIllitted a series of freely selected numbers.
The general conclusion of the study was that the "a priori difficulty of the
secondary task was not predictive of the. amount of interference, nor was
the extent of interference a function of primary task difficulty [Trumbo,
Noble & Swink; 1967,p. 239]." The authors concluded that the initiation
of responses was the main source of interference between concurrent
tasks.
The results of this experiment cannot be accepted without reservation, because similar studies in which tracking was the primary task have
reported a'substantial effect of secondary task difficulty, even when the
response elements of that task are kept constant (Johnston, Greenberg,
Fisher & Martin, 1971; Naylor, Briggs & Reed, 1968)~ However, an additional stlldy by Trunlbo and Noble (1970) provides more compelling evidence for the conclusion of their original study. Trumbo and Noble
adopted a theoretical framework suggested by Smith (1968), in which
the stimulus-response chain is divided into {ollr stages: (1) stimulus
preprocessing; (2) stimlllus classification; (3) respollse selection; and (4)
response execlltion. They compared the effects of a series of secondary
tasks, which were designed to impose different demands on each of
these stages. The primary task was always the learning of a list of nonsense syllables, presented at a three-second rate. The following secondary
task conditions were studied:
(a)
(b)
Control. No task.
Free response. Pressing one of five buttons, freely chosen, once
184
(c)
(d)
(e)
every three seconds. This task involves only stages 3 and 4, i.e., the
selection and execution of a response.
Learning the stochastic rules governing a sequence of lights, shown
at the rate of one every three seconds. This task involves only
stages 1 and 2.
"Shadowing" the series of lights, without learning instructions.
Shadowing was done. by pressing the button spatially corresponding
to each light that came on. This task was assumed to involve stages
1,2, and 4.
Anticipating each of the lights by pressing the appropriate button.
This task was assumed to involve all stages.
Condition (e) severely retarded verbal learning, and condition (b)

was also disruptive. Conditions (c) and (d) did not differ significantly
from the control condition. It is easily seen that this result implicates
stage 3, response selection, as the locus of interference. This conclusion
appears to support single-channel theory, since task (c) causes less interference than task (b) although it is more complex.
On closer examination, the results are consistent with the approach
to effort which was introduced in earlier chapters. Indeed, one may predict with some confidence that phYSiological measures of effort, such as
the dilation of the pupil, would' reproduce the ordering of conditions obtained by Trumbo and Noble. Free choice of a response, as in condition
(b), is known to elicit substantial pupillary changes (Simpson & Hale,
1969), whereas the dilations that accompany silent associative learning
are small (Kahneman & Peavler, 1969). Thus, the finding that a "Simple"
task of free response causes greater interference than a complex learning
activity is quite consistent with physiological studies of effort. Physiological studies also indicate that considerable effort is involved in overt
tests of recall, which are present in condition (e) of the Trumbo-Noble
experiment. Thus, the ordering of conditions by effort and by i~ter
ference is probably the same, and this ordering violates intuitive notions
of difficulty and complexity. How.ever, a discrepancy remains, since
Trumbo and Noble reported a dichotomy between some tasks which
cause interference and others, including associative, learning, which do
not. Pupillary studies, on the other hand, suggest that the activity involved in associative learning does require effort. The pupillary dilations
that accompany such learning are extremely consistent, although they are
only 15-20 percent as large as the dilations that occur during tests of
recall. The absence of statistically significant interference in some conditions of the Trumbo-Noble study could well be due to the difficulty of
obtaining reliable and sensitive interference measures in a relatively
small number of trials (Kahneman, 1970).
185
The reinterpretation of the experiment of Trumbo and Noble relies on speculations about what pupillary measurements would have
shown, if collected. This type of reasoning is hardly conclusive. However,
there exist experimental results that directly confirm the continuous covariation of a measure of interference with physiological indications of
effort and arousal (Kahneman, 1970; Kahneman, Beatty & Pollack, 1967).
These results were discussed in an earlier chapter (see Fig. 2-3 on page
21). The subjects in a. series of experiments performed a demanding
digit transformation as their primary task, and as a subsidiary task they
monitored a visual display for a significant signal. Two of the curves in
Figure 2-3 illustrate the time-course of the perceptual deficit that occurred during the transformation task, while a third curve traces concurrent changes of pupil size. Control experiments in which an artificial
pupil was used showed that the dilations of the pupil were not the cause
of the visual deficit. The observation of a close correspondence between
behavioral and physiological measures provides strong support for an
effort theory. Another important observation in Figure 2-3 is that the
perceptual deficit was severe during the pause between the presentation
of the .digits and the subject's response. Thus, the interference with perception was due neither to the presence of concurrent stimuli nor to the
occurrence of concurrent responses.
The present argument suggests a reformulation of single-channel
theory. This theory assumed that the stage of response selection is a
bottleneck, which can only deal with one response at a time. Instead, it
appears plausible to assume that the selection of a response is often highly
demanding of attention and effort. As ;a result, activities that demand
response selection will tend to interfere with other activities. Response
selection, however, is neither a necessary condition for the occurrence
of interference, nor a sufficient condition for the total refractoriness postulated by single-channel theory.
PROBE MEASURES OF SPARE CAPACITY
The observation of a perceptual deficit that accompanies the transformation of a series of digits illustrates the use of a probe signal to
measure- variations of spare capacity during the performance of a primary task. To obtain such a measure, the probe must be introduced at an
llnpredictable time. According to the theory of effort outlined in Chapter
2, the accuracy and the speed of the response to an unpredictable probe
reflect the spare capacity that is allocated to perceptual monitoring at
the instant of presentation. The theory assumes that spare capacity de-
186
creases regularly with increasing investment of effort in the primary

task (see Fig~ 2-1 on p. 15).
Two measures of the response to the probe-have been used in recent
studies: perceptual deficit and delayed RT. Anat Ninio, at the Hebrew
University, has investigated variations of perceptual de~cit as ~ function
of task load during the performance of a reaction- task. She showed the
subject a very large numeral, projected on a screen. The subject was required to read the numeral aloud (Add-O) or to transform it (Add-I). His
primary task was to perform this operation as fast as he could, and he
was rewarded for consistent maintenance of a fast RT. Some time after
the presentation of the numeral, an acuity target was briefly shown,
preceded and followed by a masking field to prevent visual persistence.
Acuity was found to vary sharply during the reaction time to the numeral in both task conditions. At "about 150-300 "milliseconds after the
presentation of- the numeral, acuity was significantly lower when the subject was enga,ged in the Add-l task than when he was engaged in the
Add-O task. Earlier and later, there were no significant differences.
Ninio's study was undertaken in the hope of clearing upa thorny
problem in the theory of effort: the confounding- of effort' with response
time. In general, there isa high correlation -between the time required
to produce a response and the physiological arousal that accompanies
that response. Because all autonomic measures of effort involve some lag
and temporal integration, these measures cannot be used to prove that
the rate at which effort is exerted is higher with a slow and difficult
respon~e than with a faster and easier one. The results
Ninio's experiment suggest that a more complex response task involves both a longer
latency and a greater investment of effort during at least some se'gments
of this latency.
A study by Blake and Fox .(1969) yielded discrepant results. These
authors presented an acuity target at various intervals during the reaction time to an auditory tone, and observed no decrement of visual
recognition. This unexpected failure to obtain interference could be due
to a combination of two factors: a very simple and fast manual task (RT
was 150-200 msec) and a mode of target presentation which permitted
prolonged visual persistence. It is at least possible that the subjects in
this study "read" the acuity target from an iconic image. The persistence
of this image would permit the subject to deal with the two tasks in sequence.This strategy is precluded when the probe stimulus is immediately masked.
Posner and Keele (1968, 1970) have used simple RTas a subsidiary
task. At various times during the execution of a visually guided movement, they introduced an auditory signal to which the subject was to
respond. The RT to the probe was longer if the probe coincided with
of
187
the initiation of the movement or with its terminal phase than if it

occurred during the intermediate period. In a further refinement, Ells
(1969) showed that the RT to a probe inserted just before the initiation
of a choice-response reflects the complexity of the choice (i.e., the number of alternative, responses). On the other hand, the RT to probes inserted during the movement reflects the accuracy demands of the task
(i.e., the size of a target toward which the movement is aimed).
Probe RT was used in a subsequent study (Posner & Boies, 1971)
to investigate a letter-matching task. The sequence of events on each
trial was as follows: there was a warning signal; some time later, a letter
was shown; then another letter was shown and the subject pressed one
of two keys with his right hand, depending on whether or not the
second letter was the same as the first. A tone was presented on half
the trials, in one of eight temporal positions. The subject was instructed
to press a key with his left hand whenever he heard the tone, but the
instr~ctions and the knowledge of results that the subjects were given
both emphasized the letter task. The RT to the auditory probe was
interpreted as a measure of the demands of the letter-matching task.
Posner and Boies (1971) observed that the presentation of the first
letter in the sequence did not cause an immediate rise in probe RT.
During the first 300 milliseconds after the presentation of the initial letter the subject is presumably involved in an operation of encoding,
which prepares him to judge whether the second letter is the same as the
first. Probe RT was not significantly delayed by this encoding activity.
However, RT started to rise about 500. milliseconds before the presentation of the second letter. When the interval between the first and the
second letter was prolonged, the rise in probe RT was correspondingly
delayed (Posner & Klein,1972).
Posner interpreted probe RT as a measure of competition for a
limited capacity, but it is not entirely clear that the delay of RT which
is observed in the letter-matching studies provides a pure measure of
capacity interference, since there seems to be little for the subject to do
during the 500 milliseconds that precede the presentation of the second
letter. A conflict between the anticipation of a response with the right
hand and the execution of a response with the left hand could contribute
to the delay. This interpretation is supported by a comparison of the
magnitude of the delays observed by Posner and Boies to those obtained
by Shulman and Greenberg (1971; Shulman, Greenberg & Martin, 1971),
cited earlier. Although involvement in rehearsal delayed RT very conSistently in these studies, the effect was much smaller than in Posner's
paradigm, where the subjects are instructed to make two speeded reSponses. The similarity of the primary and subsidiary tasks probably increases conflict and interference.
188
A subsequent study (Posner & Klein, 1972) provided additional evidence forthe validity of probeRT as a measure of task load. Enormous
delays were observed when the subject was instructed to apply a transfo~mation to the first letter and to match the second letter to the output
of the transformation. The subject was to make a positive response if the
second letter occurred in the alphabet three positions after the first (e.g.,
the response was positive if the first letter was M and the second was P).
This task certainly keeps the subjects very busy during the brief interval
between the first and the second letter. Accordingly, they tend to delay
responding to the probe until the completion of the matching task.
This brief discussion of the perceptual-deficit and probe-RT
methods echoes the conclusions reached earlier in the discussion of
measures of continuous load. The object of all these methods is to measure the attentional demands . of primary tasks, but the results of any
single method must be interpreted with caution, because of the everpresent possibility that the observed interference is due to structural
factors rather than to limitations of capacity. The methodological moral
is clear: effort or load should always be measured by at least two independent methods, so chosen that they are unlikely to cause structural
interference in the same way. For example, a perceptual subsidiary task
minimizes overt responses, but it usually involves some load on shortterm memory; aprobe-RT task causes response conflict, but imposes no
load on memory. The two methods appear to be complementary. Alternatively, either of these methods could be used in conjunction with
physiological measures of effort and arousal (see Chap. 2). The time-lags
involved in autonomic responses, however, make them inadequate for
the study of the microstructure of effort demands. For that purpose, the
only alternative to convergent behavioral measures may be a combination of a behavioral method with measurements of evoked cortical responses (e.g., Posner, Klein, Summers & Buggie, 1973; Posner & Warren,
1972).
PERCEPTION AND EFFORT
An important outcome of Posner's work (Posner & Boies, 1971;

Posner & Klein, 1972) was the conclusion that the process of encoding
does not require the limited-capacity mechanism: probe RTremained
unchanged or even decreased during the first 200 milliseconds after
the presentation of the initial stimulus in the matching task. Since the
first signal must be encoded at about that time, the absence of interference with probe RT suggested that the process of encoding is effortless.
Keele (1972) has used a reaction-time measure in another attempt
189
to demonstrate that certain mental activities are effortless. His subjects

were required to make a choice-response to the color of a visual stimulus,
which was sometimes a nonsense shape, sometimes an irrelevant word,
and sometimes a color word (e.g., the word Green printed in red, with
"red" the correct response). There was no difference in RT between the
responses to nonsense shapes and to irrelevant words. Nevertheless,
Keele could prove that words were read, because the presentation of
color words caused Significant interference. He concluded from this
finding that reading a word is effortless and demands no attention. As
was mentioned in the preceding chapter, Keele (1973) takes the position
that all mental operations prior to the initiation of, responses require no
attention, and therefore do not interfere with other activities.
The view of perception introduced in Chapter 5 suggests a different
interpretation of these results. It assumes that effort is invested in perception. The allocation of effort or attention to a particular perceptual object
is manifested in figural emphaSis. The effect of this allocation is to enhance the quality of the information which eventually reaches the recognition units. The number of activated recognition units and their degree
of activation are affected by the amount of 'attention that was paid to the
stimulus object. However, the activation of recognition units and the
achievement of perceptual interpretations do not require more attention
than was already allocated at the stage of figural emphasis. Thus, it takes
no more effort to look attentively at a familiar English word than at a
nonsense form. Whether such an attentive look results in "reading" the
word depends entirely on the availability of a recognition unit for the
pattern.
The occurrence of perceptual deficit during mental activity provides the most direct evidence for the relation between perception and
effort. 'If an activity can be carried out without effort, it should no more
be subject to capacity interference than be the source of such interference. Indeed, the most sensitive test of whether an activity demands effort is whether it can be disrupted by intense involvement in another
activity. An act that demands little effort may be vulnerable to interference, while having negligible effects on other acts.
. This methodological criticism of the Posner-Keele argument suggests that perceptual emphaSiS could demand attention after all. But
, a more significant aspect of this debate is conceptual: what is meant by
saying that an activity requires or demands effort? These verbs have two
distinct meanings: one, that we may label demand1 , merely states a necessary condition for some end to be achieved. The other meaning, demand2 , implies that some action is taken to ensure that the demand will
be met. Thus, it is proper to say that a particular Hower demands 1 a
great deal of water for normal growth, while a child loudly demands 2
mote marbles from his partner.
190
It will now be apparent that the terms "demand" and "require"

have been used in the preceding discussion in the two meanings of demand t and demand 2 Thus, it was said that many mental activities demand t effort, because they cannot be completed without attention. In
addition, some stimuli which are favored by a selective set demand2 attention, i.e., they attract more attention than do other stimuli. Finally,
the model of attention introduced in Chapter 2 assumed a feedback loop
by which an evaluation of current performance controls arousal, and
thereby the supply of effort for the successful continuation of that performance. It is through this feedback loop that a continuous mental
activity demands 2 attention and effort. Thus, a complex task such as
serial digit transformation cannot be carried out without attention (demand t ) and it also causes attention to be mobilized (demand2 ). The
elicitation of the orienting response was explained in similar terms: the
processing of a novel and significant stimulus requires (demands t ) a
relatively large amount of effort; a significant violation of the neuronal
model causes (demands 2 ) a subsequent surge of arousal and effort, which
is directed to a more detailed analysis of the stimulus.
Most stimuli, of course, do not elicit an orienting response, and it
is a reasonable assumption that most perceptllal activity rarely demands 2
any effort, although it depends on the continuous allocation of some
capacity (demand t ). If this idea is correct, minor changes in the structure
and complexity of perceptual acts will have no effect on the performance
of concurrent activities. The absence of interference between simultaneous dichotic items in our recognition studies was explained in similar
terms (see p. 149).
Another result that requires an explicit distinction between demand t anddemand2 was obtained in studies of monitoring for targets
identified by voice or by content, which were described in Chapter
8. Monitoring for a target defined by the sex of the speaker is certainly
not more difficult than monitoring for a semantically defined category:
it demands t no more effort. Neverthel~ss, the recognition of a word
presented concurrently with a target was more severely disrupted when
that target was identified by voice than by content. This finding was
explained on the assumption that a physically distinct target demands 2
attention very early in perceptual analysis, w.hile a content target must
activate the recognition system before it demands 2 attention (see p. 152).
The concurrent word presented to the other ear can be processed normally until attention is withdrawn to deal with the target. In this manner, a relatively easy monitoring task causes greater interference than a
more difficult task, precisely because it involves a rapid redirection of
attention.
The distinction between demand t and demand 2 provides the ra-
Attention and Task Interference' 191
tionale for the use of visually masked stimuli as probes in the measurement of spare capacity. Studies of the duration and locus of fixations
indicate that attention can be quickly directed to a potentially significant stimulus that is not immediately identified. The fixation on a
significant stimulus can also be extended-a" decision that is certainly
made within 150-200 milliseconds of the initial fixation. If the potential
target was first viewed in the visual periphery, a tentative detection can
control the choice of the next fixation (Gould & Schaffer, 1965). In these
examples, an activity of perceptual analysis demands 2 attention. However, a delayed allocation of attention cannot affect perception if .the
stimulus is immediately removed and its trace destroyed by a subsequent mask. In this manner, the use of masked stimuli provides a pure
measure of the attention that was allocated to visual perception at the
instant of presentation.
SET AND OTHER DETERMINANTS OF EFFORT DEMANDS
While the preceding section concluded that perceptual activity

demands effort, it also implied that these demands are slight, when compared to those of other activities. Choices, decisions, rehearsal, and the
mental manipulation of stored symbols, all appear more demanding than
routine perceptual analysis. These activities are particularly demanding
when executed under "pressure of time. Thus, the rate at which mental
activity ,is performed is a primary determinant of effort. In many activities, "taking it easy" simply means to slow down. There are activities,
however, which impose their own rate. This is especially true of any
mental act that depends heavily on short-term memory, since the rate of
rehearsal must compensate for the rate of decay of stored information.
In such tasks, one simply cannot "take it easy."
A concept of rate" becomes meaningful only when the units of activity are specified..However, the unit of activity is an elusive concept,
because of the hierarchical character of action. What is the unit, for
example, when one recites the alphabet? Is it the individual phoneme,
the individual letter, or perhaps such familiar groups as ABeD . . EFG
.. HIJK .. LMNOP? If the analogy of perceptual grouping is accepted, the answer to such a question is not arbitrary. A certain level of
organization may be dominant. Intuitively, it seems that performance is
monitored at the completion of units at that level, and that decisions
and choices are formulated in terms of these units.
In his classic paper on the serial organization of behavior, Lashley
(1951) introduced a vivid example. Imagine a piano with a defective key
that cannot be depressed. Any piano player will stop playing when he
192
unexpectedly encounters such a key. However, the expert player will

normally play several additional notes before he stops. Evidently, the
checkpoints at which behavior is monitored and controlled do not OCcur
~fter each note. Miller, Galanter, and Pribram (1960) expressed the same
idea in their notion of the TOTE. They analyzed behavior as a sequence of operations, with an objective defined for each such operation.
When the operation is completed, a test is carried out to confirm the
attainment of the objective. Only then is the control of action passed
on to the next objective. Thus, a continuous activity can be analyzed in
terms of units of Test-Operate-Test-Exit. The rate of activity is best
viewed as the number of TOTE's required per unit time. Th~s may be
the reason why Peterson (1969) found that such activities as rapid counting or speeded recitation of the alphabet did not cause a total disruption
of concurrent mental activities. With such highly overlearned sequ"ences,
a large number of distinct muscular activities are packed into each
TOTE.
The achievement of the most effective and economical organization
of action depends in large measure on the degree to which the task
allows anticipation of future stimuli and responses (Adams, 1966; Poulton~ 1952; Shaffer & Hardwick, 1969a, 1970; Shaffer, 1971). Activities such
as driving an automobile, reading, or shadowing an auditory message
usually permit the performer to anticipate each response before he
actually executes it.~ In reading aloud, for example, the anticipation is
provided by the eye-voice span: the subject's eye is usually several words
ahead of the word that he utters at anyone time. The eye-voice span
is easily measured by turning off the light by which the subject reads;
he will almost invariably continue to "read" a few words after the light
is off. In shadowing an a~ditory message, subjects typically adopt an
average lag of 1-1.5 seconds, which allows them continuous advance information about the phrase that they will utter in the immediate future.
The possibility of anticipation is essential to adequate performance. In
typing, for example, "response may lag the fixated letter by six or seven
letters, on the average, and . . . if lag is prevented by eliminating preview of text, then typing is about five times slower [Shaffer & Hardwick,
1970, p. 425]." Anticipation facilitates performance in several ways: it
permits response integration, and thereby effectively reduces the number of discrete choices and decisions that must be made. It also permits
a smooth adjustment of effort to the difficulty of each choice and each
response.
Anticipation is but one of the adjustments of which man is capable,
which reduce the effort required for adequate performance, or ensure
that the supply of effort will meet the demands. These adjustments are
193
often grouped under the collective label of set. .The present treatment
has distinguished several classes of preparatory adjustments.
A state of perceptual readiness for a particular perceptual interpretation'increases the likelihood that this interpretation will be adopted,
both when sensory informatio'n is appropriate to. it, and when the match
between the features of this information and the critical features of the
relevant recognition unit is less than perfect. Perceptual readiness is
mediated by a criterion bias favoring some interpretations over others.
A state of readiness for a particular interpretation implies that the
achievement of this interpretation demands 1 less information input, and
less attention, than does the achievement of other interpretations. Thus,
a stimulus for which one is ready is likely to be identified even when it is
presented on an unattended channel, or at a low level of intensity or
clarity.
A state of response readiness similarly lowers the criterion for the
elicitation of a particular response, or class of responses. It is reasonable
to assume that a response for which one is ready demandst less effort
than does a response for which one is not prepared.
Perceptual and response readiness may be viewed as altered states
of the specific units which are activated in the processes of perceptual
interpretation and response selection. In contrast, selective set is a characteristic of the allocation policy that controls figural emphasis and other
manifestations of selective attention. Here, a selected stimulus demands 2
attention: more attention or effort is allocated to it than to the processing
of other stimuli. Two variants of selective set have been distinguished,
of which one is mediated by the immediate allocation of attention to
stimuli isolated at an early stage of analysis, while the other involves
'
recognition units and a .recursive path of attention control.
The primary mechanism of selective attentionma)7 be identified
with Broadbent's filter. Perceptual emphasis is allocated to stimuli that
possess a particular attribute, e.g., sounds that originate in a particular
place or words printed in a particular color. A search set could affect
processing by the same mechanism, and it is conceivable that a target
for which one is set' can attract attention prior to the activation of the
recognition system, if the target is identified by obvious physical characteristics. A selected stimulus attracts more attention than do other
stimuli. Thus, a stimulus for which one is prepared will "jump" from
the background (e.g., Eriksen & Collills, 1969a; Neisser, 1967). An attended stimulus will also have prior entry, i.e., it will appear to have
occurred sooner than a physically simultaneous unattended stimulus
(Sternberg, Knoll & Gates, 1971). The reaction to a stimulus that matches
expectation is speeded (Egeth & Blecker, 1971). Indeed, some compo-
194
nents of the evoked cortical response occur sooner when the stimulus
matches expectations than when it does not (Posner, Klein, Summers &
Buggie, 1973). The effects of selective attention on the sensitivity parameter of signal detection can be mediated by this type of selective set.
Secondary selective attention is controlled either by a tentative
recognition of a significant stimulus, or by a failure to obtain an adequate
perceptual interpretation for an event which violates the neuronal model
of expectations. Such stimuli demand2 attention, which is allocated to
them via the recursive path of attention control. This mechanism is involved in some search tasks (e.g., monitoring a list for names of animals).
The tentative detection of the selected stimulus probably causes a surge
of eff~rt, as well as a redirection of attention to the detected target.
The various mechanisms of set are not mutually exclusive, and
more than, one mechanism may be engaged in any task. Thus, a set to
search for animal names may increase the perceptual readiness for these
names; it may also sensitize the process of secondary selective attention,
so that a tentative recognition of a target item will cause especially detailed analysis of that item. Preparatory adjustments appear to be
highly flexible.
Other aspects of preparatory set are the elicitation of anticipatory
arousal, and of a specific posture of orientation. The warning signals
commonly used in 'studies of reaction time and of the perception of brief
stimuli, serve both these functions :of orientation and arousal. To be fully
effective, such a warning signal must be delivered about 500 milliseconds before the relevant stimulus. Achieving a state of optimal readiness .takes time. Studies of the foreperiod effect also indicate that o,ptimal
readiness cannot be maintained very.long. Responses to. stimuli that follow the warning signal bya second or more tend to be slower than .when
the foreperiod is half a second. This failure to maintain readiness is consistent with the hypothesis that arousal is largely controlled by the feedback of ongoing activity. In the absence of such feedback, arousal diminishes.
The alerting function of warning signals has been studied in detail
by Posner (Posner & Boies, 1971; Posner, Klein, Summers & Buggie,
1973). ffe concluded that the presentation of the initial letter in the
letter-matching task can facilitate performance both. by increasing alertness and .by increasing the specific readiness for the repetition of that
letter. The two facilitative effects summate without interacting. This
finding suggested the hypothesis that the encoding process whichmediates the specific readiness for a letter is equally effective at various levels
of arousal. An additional discovery concerned the nature of the foreperiod effect: Posner was able to show that the U-shaped function which
relates RT to the duration of the foreperiod is associated with a
195
rt-shaped function for errors in a spatial choice-reaction. The high level

of alertness at the "optimal" foreperiod is accompanied by a relatively
high rate of errors.
There is other evidence which confirms the conclusion that high
arousal tends to be associated with a lowered response criterion, and
consequently with faster and less accurate responses (Broadbent, 1971).
Posner's interpretation of these results is .novel: he argues that alertness
does not affect the quality of the information which is available to the
decision mechanism, but merely the speed at which the decision is
reached. Because the decision is reached faster when alertness is high,
it is based on a reduced sample of evidence, and is consequently more
subject to error than when alertness is low.
It is very unlikely that the adequacy of perceptual analysis was the
.limiting factor in these experiments. Indeed, different results are' obtained when the stimuli for a task of simultaneous discrimination are
brief and faint: with such stimuli, an anticipatory warning signal reduces both the latency of responses and the probability of errors (Posner,
Klein, Summers & Buggie, 1973). Posner's interpretation is that a slow
response (associated with low alertness) does not yield the advantage of
a more protracted analysis when the stimuli are brief. An alternative
interpretation is that anticipatory alertness facilitates the immediate
perceptual analysis of stimuli, and also tends to alter the response criterion. When the stimuli are prolonged and easily perceptible, the only
measurable effect of the warning signal will be an altered value on the
speed-accuracy function. These are conditions where erroneous responses
do not reHectperceptual errors. The advantage of anticipatory allocation of attention only becomes evident when errors of perception begin
to limit performance. In this view, anticipatory arousal improves perceptual analysis, but does not facilitate the operation of the other mechanisms that determine the choice of a response in a discrimination task.
The preceding discussion of anticipatory adjustments indicates that
these adjustments affect both the amount of attention required for the
execution of an activity and the likelihood that attention will be effectively allocated to that activity in preference to others. These considerations introduce severe complexities in any analysis of performance in
dual tasks, since .the tasks interact at the level of preparatory set as well
as during the performance of demanding activities. This interaction is
sometimes favorable: the.anticipatory mobilization of effort for a primary
task :occasionally facilitates the response to a probe signal (Posner &
Boies, 1971). More often, the interaction is detrimental. There is much
evidence that a "divided set" hampers performance. In the refractoriness paradigm, for example, the reaction to the first stimulus is .generally
slowed by the anticipation of another response (Smith, 1967c; Triggs,
196
1968). Similarly, Broadbent (1956) found that subjects often fail in a

coding task when merely waiting for a buzzer to sound, and Malmo
(1966) reported that subjects who expect to shift from one mode of tracking to another track less efficiently than under unified set. Webster and
Solomon (1955) also observed that the comprehension of a single complex message is impaired if the subject had expected the presentation of
two simultaneous messages. Two plausible interpretations of these
findings are: (1) the divided set requires the maintenance of an orientation pattern which is both more strained and less effective than in unitary
set; (2) the organization of divided set draws directly on the capacity of
the organism.
STRUCTURAL IN'I'ERFERENCE
The introduction to this chapter distinguished two types of interference between tasks: capacity interference, which arises as a function
of the attentional demands of competing activities; and structural interference, which occurs because the activities occupy the same mechanisms of perception or response. Structural interference in perception
was illustrated in Chapter 8, where it was shown that concurrent monitoring tasks in one modality tend to be more difficult than concurrent
monitoring in different modalities (Treisman & Davies, 1972). This study
illustrates the gen'eral method by which structural interactions can be
demonstrated. Tasks A and B are equated by difficulty or by a physiological measure of effort, when performed singly. If the combination of
task A with a new task C is more demanding or difficult than the combination of tasks B andC, this result prOvides evidence for interference between A and C beyond what can be explained in terms of attention or
capacity. The alternative interpretation, that tasks Band C are mutually
facilitating, also assumes a structural interaction.
Structural interference appears to have been a confounding factor
in several of the studies that attempted to measure capacity interference.
Thus,Brown (1966) noted that the subsidiary tasks of interval production
and, random-number generation are affected differently by primary activities that involve a high rate of overt responses or a high rate of mental
activity. Similarly, there are indications that probe~RT measures are
especially sensitive to the motor component of the primary activity. The
general rule appears to be that similar activities' tend to be mutually
interfering, unless they 'can be integrated.
Structural interference can also arise within a single task, through
an interaction between the modality of the response and the modality of
the input that controls the response. Brooks (1968) has: offered an elegant
197
demonstration of this effect. In one of his experiments, he briefly presented a line diagram (e.g., Fig. lO-lA), and later required subjects to
begin at the star and categorize successive corners by saying "yes" if the
corner is on the extreme top or bottom and "no" otherwise. The correct
sequence of answers in this example is "yes,yes,yes,no,no,no,no,no,no,yes."
Three modes of response were compared: calling out the words "yes"
, or "no" for each corner; pointing to the appropriate word in columns of
"yes" and "no" (Fig. 10-lB); tapping with the left hand for "yes," and
with the right hand for "no." The first response was purely vocal, while
the second required visual monitoring. Subjects had much more difficulty
with pointing than with the other modes of. report. In another condition,
the subjects heard a sentence (e.g., "A bird in the hand is not in the
busll") and were asked to recall the sentence 'and to categorize each
word as a noun ("yes") or any other part of speech ("no"). The same
thre~ modes of response were used, but now the vocal response was by
far the most difficult. Brooks (1968, p. 354) remarked: "The subjects
reported that they 'could say the sentence to themselves' while tapping
or pointing, but not while saying 'yes' or 'no.' The diagrams could be
'pictured' while the subjects were tapping or saying 'yes' or 'no,' but not
while they were trying to point."
Brooks (1967, 1970) also showed that reading and visualization are
mutually interfering. Subjects were given a verbal description of a spatial
arrangement, and were asked to imagine and describe a rotation of that
198
arrangement. They were able to do so faster if they merely listened to

the original description than if they also read it. Structural interference
occurs between visual operations within a single task.
Others have reported related findings. Lowe and Merikle (1970)
found that spoken recall causes more output interference with the retention of auditory material than does written recall. Greenwald (1970a)
found that 'people are better able to resist auditory distraction when
they write than when they speak. Greenwald (1970c) reviewed James'
ideomotor theory of action, which explains such interactions by the idea
that images are involved in the control of action. A subject who prepares
to utter a word produces anticipatory acoustic imagery, and this imagery
may be disrupted if he hears a spoken word at the critical time.
These results extend the conclusion that simultaneous inputs on a
single modality are likely to be mutually interfering. Interference is also
likely when one modality is simultaneously involved in the control of
response and in the discrimination of inputs. Thus, concurrent tasks
that involve the same modality or response system are likely to suffer
from structural interference.
The suggestion that all interference between tasks may be structural was advanced by Allp9rt,' Antonis, and.Reynolds (1972). They proposed that an appropriate model of man may not be :a single general
purpose computer, but rather "a numb.er of special purpose computers (processors and stores) operating in parallel and, at least in some cases,
capable of accepting only one message or 'chunk' of information for
processing at one time [p. 233]." As evidence, the authors showed that
shadowing an auditory message impairs retention of a concurrent list
more severely if the list is auditory than if it is visual, and more ;severely
if the visual material consists of words than of pictures. In addition,
they showed that experienced piano players could sight-read and. shadow
an auditory message at th~ same time with little evidence of interference.
The authors justly emphasize the observation that subjects who shadow
an auditory message can play the piano, but cannot effectively listen to
another verbal message./
In contrast to these results is the finding of Peterson (1969) that
complex covert problem-solving, including the solution of. anagrams,
can be carried out while the subject is engaged in continuous high-speed
counting or recitation of the alphabet. Evidently, the involvement of.
verbal mechanisms in both tasks does not entirely preclude parallel performance. Interference was primarily determined by task complexity in
Peterson's study. These results present a difficulty for Allport's multichannel theory.
Structural interference between related tasks suggests the image
of antagonistic interactions among neural structures, such that a high
degree of activation of one structure tends to reduce the level of activity
199
in others. This mode of organization is .prevalent in the nervous system,

where it appears both in sensory analysis and in the control of motor
output. An enhanced input is required to keep any unit in such a system
at a specified level of response when another unit is simultaneously
activated. Thus, the simultaneous operation of two antagonistic units
demands! a greater input than the sum of the inputs that are required
, for separate operations. The strength of the inhibitory connections usually depends on the funQtional separation between the interacting units.
Neighboring units tend to interact more strongly than distant units. It
is readily seen that this feature of neural organization is quite compatible
with the suggestion by Allport (1971; Allport et al., 1972) and by Treisman (1969; Treisman& D'avies, 1972) that similarity between interacting
activities is the primary determinant of interference.
For an effort theory, the occurrence of interactions between tasks
is a complication, because the attractive notion that effort demands of
concurrent tasks are additive must be abandoned whenever such interactions 'occur. It is obviously impossible to predict the amount of interference between two tasks solely on -the basis of their separate demands for
effort. Overlap, similarity, and mutual compatibility must also be considered. However, it appears equally ~mpossible to account for the phenomena of interference without reference to the role of .task difficulty.
Thus, it is useful to retain the. term of structural interference for situations of strong interaction between similar tasks, and to apply the label
of capacity interference to situations where difficulty is the main determinant of results.
INTERFERENCE AND EFFORT THEORY
Let us now recapitulate the major assumptions that appear to be

required to explain the phenomena of task interference. First, we must
assume the existence of performance units, roughly equivalent to the
perceptual units that were discussed in Chapters 5 and 7. Attention, or
effort, is allocated to such units. We assume further that each such unit
is characterized by a certain level of demands, i.e., of need for attention
or effort. Performance falters if the amount of attention allocated to a
performance unit is less than the amount demanded. A further assumption is that the amount of .attention or effort supplied to- a unit rises with
demand, but not sufficiently (see Fig. 2~1 on p. 15). When a task is made
more complex, performance slows down and errors increase in spite of
augmented effort.
Consider now the case in which two distinct performance units are
simultaneously selected. We assume that these units are non-redundant,
so that there is no possibility of integrating them into a superordinate
200
structure. The perceptual equivalent would be the presentation of two

different words to both ears at the same time, where both must be identified. When the units are non-redundant, it is reasonable to assume the
following inequality:
Demand of Joint Performance ~ Sum of Separate Demands.
The difference between the left-hand and the right-hand sides of this inequality is a measure of structural interference. If the two performance
units are incompatible or otherwise mutually antagonistic, the effort required to perform both together will be greate~i than the sum of the
effort required to perform them separately. In addition, the total effort
required to perform two acts together can be greaterthan the sum of separate demands, if the organization of joint performance itself demands
attention (Lindsay, Taylor & Forbes, 19'68; Moray, 1967; Taylor, Lindsay
& Forbes, 1967).
The assumptions stated so far entail the prediction of some interference for all cases in which non-redundant tasks are performed together,even in the absence of structural interference. The basic assumption of the model is that the supply of effort is a negatively accelerated
function of demand. Since the joint demands of two performance units
are greater than the demands of either, the total deficit must be larger
in joint performance than when the tasks are executed in isolation. Thus,
Total Deficit ~ Sum of Separate Deficits.
According to the assumption that supply is an increasingly insufficient

response to demand, the total deficit increases with the total demand.
Consequently, there will be little interference when both tasks are easy,
and interference will increase with the difficulty of either task.
In this conception, interference is explained by the shape of the
function that relates the supply of effort to the demand. This assumption
is proposed instead of tIle commonly stated notion that a general limit
on capacity explains task interference. The idea of a constant limit on
capacity is inadequate, since it is easy to show interference occurring
even in situations where the actor does not exert the maximal effort of
which he is capable.
The preceding considerations indicate that interference must occur
whenever two distinct tasks are performed together. However, the actor
has considerable freedom to determine which task will suffer interference. Subjects are capable of protecting one task, so that it is performed in conjunction with another nearly as well as in isolation, and the
entire interference effect is then found in the performance of the subsidiary task (Kahneman, 1970).
201
The treatment so far has assumed that the competing units of

action are performed in parallel. This assumption was made because of
the well-documented failure of various single-channel models. However,
the maintenance of parallel organization of processing can sometimes
lead to a total failure of one or both acts, and a sequential strategy must
, be adopted to prevent such overload. When the two tasks both consist
of serial units, of performance, the units of both tasks are often interleaved. Indeed, a' basic rule of the policy that allocates attention appears
to be that jamming of the system is not permitted to occur. When the
demands of two tasks cannot be adequately satisfied, one is typically
_
selected and the other is delayed or abandoned.
A similar conclusion was reached earlier'in the discussion of dual
monitoring." When two targets are presented at once, the ,typical outcome
is for one to be perceived and for the other to be ignored entirely. If the
subject is expecting the simultaneous occurrence of the two' targets,
processing is sometimes parallel and sometimes strictly sequential. The
choice' of processing mode depends at least in part on the load imposed
by the competing activities.
The results in studies of divided attention are generally compatible
with a view of attention, or effort, as an input to central structures which
enables or facilitates their operation. The main attributes of attention
are the following:
(1)
(2)
(3)
(4)
Attention is limited, but the limit is variable from moment to

moment. Physiological indices of arousal provide a measure that
is correlated to the momentary limit.
The amount of attention or effort exerted at any time depends
primarily on the demands of current activities. While the investment .of attention increases with demands, the increase is typically
insufficient to fully compensate for the effects of increased task
complexity.
Attention is divisible. The allocation of attention is a matter of
degree. At high levels of task load, however, attention becomes
more nearly unitary.
Attention is selective, or controllable. It can be allocated to facilitate the processing of selected perceptual units or the execution of
selected units of performance. The poliCY of allocation reflects per- .
manent dispositions and temporary intentions.
REVIEW
This final chapter applied the theory of effort introduced in Chapter

2 to the interpretation of task interference. There is strong experimental
support for the main conclusion from this theory, that interference
202
between concurrent tasks depends on the demands that these tasks .

separately impose on the limited capacity system. The effort demands of
tasks do not always correspond to intuitive notions of task difficulty. For
example, subvocal rehearsal, the choice and execution of free responses,
and tests of recall on familiar material appear to require considerable
effort, although they would be judged simple.
The spare capacity which is available at any instant during the
performance of a primary task can' be measured by the accuracy and
speed with which unexpected probe signals are handled.
A distinction was drawn between two meanings of the termattention demands. Demand t denotes that an activity cannot be carried out
without a sufficient allocation of attention. Demand2 denotes that a prior
selectiv~ set or an evaluation of the quality of performance of an activity
controls the amount and allocation of attention. Perceptual analysis normally does not demand2 attention, although it demands t attention. These
terms were applied to an analysis of several variants of preparatory set,'
of which some reduce the attentional requirements of tasks, while others
ensure that these requirements will be met.
Some evidence for structural interference was reviewed....There
appear to be many situations in which concurrent tasks interact so that
the demands of dual performance greatly exceed what would be expected on the hypothesis that effort is additive. Structural interference
is typically observed when the interacting tasks require the operation of
similar mechanisms of perception or resp~nse.
The final section reviewed the interpretation of interference within
an effort theory. The concept that interference occurs only when a
limited capacity is exceeded was rejected, because capacity appears to
be variable, and because interference arises even among fairly undemanding tasks. Interference was explained on the alternative assumption
that the supply of attention generally fails to meet increasing demands.
This assumption is needed to explain why increased effort fails to compensate fully for increased difficulty, in both the single-task and dual-task
situations.
References
ADAMS, J. A. (1962) Test of the hypothesis of psychological refractory period.

Journal of Experimental Psychology, 64, 280-287.
---.~ (1966) Some mechanisms of motor responding: An examination of attention. In Acquisition of skills, ed. E. A. Bilodeau. New York: Academic
rress. Pp. 169--200.
AGNEW, N., and AGNEW, M. (1963) Drive level effects on tasks of narrow and
broad attention. Quarterly Journal of Experimental Psychology, 15, 5862.
ALLPORT, D. A. (1971) Parallel encoding within and between elementary
st~.:mulus dimensions. Perception and Psychophysics, 10, 104-108.
- - - , ANTONIS, B., and REYNOLDS, P. (1972) On the division of attention:
A disproof of the single channel hypothesis. Quarterly Journal of Experimental Psychology, 24, 225-235.
ALPERN, M. (1971) Effector mechanisms in vision. In Experimental Psychology,
eds. J. W. Kling and L. A. Riggs. New York: Holt, Rinehart & Winston.
AMADEO, M., and GOMEZ, E. (1966) Eye movements, attention and dreaming
ill. subjects with life-long blindness. Canadian Psychiatric Association
Journal, 11, 501-507.
ANTROBUS,
J.
S., ANTROBUS,
J.
S., and SINGER,
J.
L. (1964) Eye movements

203
204
accompanying daydreaming, visual imagery, and thought suppression.

Journal of Abnormal and Social Psychology, 69, 244-252.
ARCHER, E. J. (1962) Concept identification as a function of obviousness of
relevant and irrelevant information. Journal of Experimental Psychology,
6~, 616-620.
ARGYLE, M., and DEAN, J. (1965) Eye-contact, distance and affiliation. Sociometry, 28, 289-304.
ATKINSON, R. C., and SHIFFRIN, R. M. (1968) Human memory: a proposed
system and its control processes. In The PsychoZogy of Learning and
Motivation: Advances in Research and Theory, Vol. II., New York: Academic Press., Pp. 89-195.
ATTNEAVE, F. (1959) Applications of information theory to psychology: A
summary of basic concepts, methods and results. New York: Holt, Rinehart & Winston.
AVERBACH, E., and CORIELL, A. S. (1961) Short-term memory in vision. Bell
System Technical J~urnal, 40, 309-328.
- - - , and SPERLING, G. (1961) Short-term storage of information in vision.
In Information theory-Fourth London sym.posium, ed. E. C. Cherry.
London: Butterworths. Pp. 196-211.
AVNER, U. (19'72) Interdependence between right and left ear in processing
simultaneous items as function of attentional intensity. Master's thesis,
Hebrew University, Jerusalem. (Hebrew)
AXELROD, S., and GUZY, L. T. (1968) Underestimation of dichotic click rates:
Results using methods of absolute estimation and constant stimuli. Psychonomic Science, 12, 133-134.
- - - , GUZY, L.T., and DIAMOND, I. T. (1968) Perceived rate of monotic
and dichotically alternating clicks. Journal of the Acoustical Society of
America, 43, 51-55.
- - - , and POWAZEK, M. (1972) Dependence of apparent rate of alternating
clicks on azimuthal separation between sources. Psychonomic' Science, 26,
217-218.
BADDELEY, A. D. (1966) The capacity for generating information by randomization. Quarterly Journal of Experimental Psychology, 18, 119-129.
- - - , HATTER, J. E., SCOTT, D., and SNASHALL, A. (1970) Memory and the
time of day. Quarterly Journal of Experimental Psychology, 22, 605-609.
- - - , SCOTT, D., DRYNAN, R., and SMITH, J. C. (1969) Short-term memory
and the' limited capacity hypothesis. British Journal of Psychology, 60,
51-55.
BADIA, P., and DEFRAN, R. H. (1970) Orienting responses and GSR conditioning: A dilemma. Psychological Review, 77, 171-181.
BAHRICK, H. P.,FIT~, P. M., and RANKIN, R. E. (1952) Effect of incentives
upon reactions to peripheral stimuli. Journal of Experimental Psychology,
44, 400-406.
References
205
- - - , and SHELLY, C. (1958) Time-sharing as an index of automatization.

BAKAN, P. (1971) The eyes have it. Psychology Today, 4, (11), 64-67, 96.
- - - , BELTON, J. A., and TOTH, J. C. (1963) Extraversion-introversion and
decrement in an auditory vigilance task. In Vigilance: A symposium, eds.
D. N. Buckner and J. J. McGrath. New York: McGraw-Hill. Pp. 22-28.
- - , and LECKART, B. T. (1966) Attention, extraversion, and stimuluspersonality congruence. Perception and Psychophysics, 1, 355-357.
- - - , and SHOTLAND, R. L. (1969) Lateral eye movement, reading speed,
and visual attention. Psychono'm.ic Science, 15, 93-94.
BARTZ, D. H., SATZ, P., and FE~NELL, E. (1967) Grouping strategies in
dichotic listening: The effects of instructions, rate, and ear asymmetry.
BEATTY, J., and KAHNEMAN, D. (1966) Pupillary changes in two memory
tasks. Psychon.omic Science, 5, 371-372.
BECK, J. (1966) Effect of orientation and of shape similarity on perceptual
grouping. Perception and Psychophysics, 1, 300-302.
- - - . (1967) Perceptual grouping produced by line figures. Perception and
Psychophysics, 2, 491-495.
- - - . (1972) Similarity grouping and peripheral discriminability under uncertainty. American Journal of Psychology, 85, 1-19.
- - - , and AMBLER, B. (1972) Discriminability of differences in line-slope
and in line-arrangement as a function of mask delay. Perception and Psychophysics, 12, 33-38.
BERKSON, G. (1965) Eye fixation responses to one or two stimuli. Journal of
Com,parative and PhYSiological Psychology, 59, 422~425.
BERLYNE, D. E. (1950) Stimulus intensity and attention in relation to learning
theo~y. Quarterly Journal of Experimental Psychology, 2, 71-75.
(19-51) Attention to change. British Journal of Psychology, 42, 269278.
(1957) Conflict and information-theory variables as determinants of
human perceptual curiosity. Journal of Experimental Psychology, 53,
399-404.
- - - . (1958) The influence of complexity and novelty in visual figures on
orienting responses. Journal of Experimental Psychology, 55, 289-296.
(1960) Conflict, arousal and curiosity. New York: McGraw-Hill.
(1965) Structure and direction in thinking. New York: John Wiley.
(1966) Curiosity and exploration. Science, 153, 25-33.
~--. (1967) Arousal and reinforcement. In Nebraska symposium on motivation, ed. D. Levine. Lincoln: University of Nebraska Press. Pp. 1-110.
(1970) Attention as a problem in behavior theory. In Attention: Conte1nporary theory and analysis, ed. D. E. Mostofsky. New York: Appleton-Century-Crofts. Pp. 25-49-.
206
- - - , and BORSA, D. M. (1968) Uncertainty and the orientat~on reaction.

Perception and Psychophysics, 3, 77-79.
- - - , BORSA, D. M., CRAW, M. A., GELMAN, R. S., and MANDELL, E. E.
(1965) Effects of stimulus complexity and induced arousal on pairedassociate learning. Journal of Verbal Learning and Verbal Behavior, 4,
291-299.
- - - , BORSA, D. M., HAMACHER, J. H., and KOENIG, I. D. (1966) Pairedassociate learning and the timing of arousal. Journal of 'Experimental
Psychology, 72, ~-6.
- - - , and CARE~, S. T. (1968) Incidentalleaming and the timing of arousal.
Psychonomic Science, 13, 103-104.
- - - , and LAWRENCE, G. H. (1964) Effects of complexity and incongruity
variables on GSR, investigatory behavior and verbally expressed preference. Journal of General Psychology, 71, 21-45.
- - - , and McDoNNELL, P. (1965) Effects of stimulus complexity and incongruity on duration of EEG desynchronization. EEG and Clinical N europhysiology, 18, 156-161.
BERNSTEIN, A. S. (1969) To what does the orienting response respond? PsychophYSiology, 6, 338-350.
- - - , TAYLOR, K., AUSTEN, B. G., NATHANSON, M., and SCARPELLI, A.
(1971) Orienting response and apparent movement toward or away from
the observer. Journal of Experimental Psychology, 87, 37-45.
BERNSTEIN, I. H. (1970) Can we see and hear at the same time? Some recent
studies of intersensory facilitation of reaction time. Acta Psychologica, 33,
Attention and performance III, ed. A. F. Sanders. Amsterdam: NorthHolland Publishing Co. Pp. 21-35.
- - - , CLARK, M. H., and EDELSTEIN, B. A. (1969a) Effects of an auditory
signal on visual reaction time. Journal of Experimental Psycholo:gy, 80,
567-569.
- - - , CLARK, M. H., and EDELSTEIN, B. A. (1969b) Intermodal effects in
choice reaction time. Journal of Experimental Psychology, 81, 405-407.
BERTELSON, P. (1966) Central intermittency twenty years later. Quarterly
- - - . (1967) The refractory period of choice reactions with regular and irregular interstimuli intervals. ActaPsychologica, 27,45-56.
- - - , and TISSEYRE, F. (1969) The time-course of .preparation: Confirmatory
results with visual and auditory warning signals. Acta Psychologica, 30,
Attention and performance II, ed. W. G. Koster, 145-154.
BIEDERMAN, I. (1972) Human performance in contingent information-processing tasks. Journal of Experimental Psychology, 93, 219-238.
- - - , and CHECKOSKY, S. F. (1970) Processing redundant information. Journal of Experimental Psychology, 83, 486-490.
BLAKE, M. J. F. (1967) Time of day effects on performance in a range of tasks.
Psychonomic Science, 9, 349--350.
References
207
BLAKE, R. R., and Fox, R. (1969) Visual form recognition threshold and the
psychological refractory period. Perception and Psychophysics, 5, 46-48.
BOGGs, D. H., and SIMON, J. R. (1968) Differential effect of noise on tasks of
varying complexity. Journal of Applied Psychology, 52, 148-153.
BORGER, R. (1963) The refractory period and serial choice-reactions. Quarterly
BORING, E. G. (1950) A history 'of experimental psychology (2nd ed.). New
York: Appleton-Century-Crofts.
BRADSHAW, J. L. (1968a) Load and pupillary changes in continuous processing
tasks. British Journal of Psychology, 59, 265-271.
- - . (1968b) Pupil size and problem solving. Quarterly Journal of Experimental Psychology, 20,116-122.
BRAND, J. (1971) Classification without identification in visual search. Quarterly
BROADBENT, D. E. (1952) Listening to one of two synchronous messages. Journal of Experimental Psychology, 44, 51-55.
- - - . (1954a) The role of auditory localization in attention and memory
span. Journal of Experimental Psychology, 47, 191-196.
- - - . (1954b) Som~ effects ,of noise on ,visual performance.. Quarterly Journal
of Experimental Psychology, 6, 1-5.
- - - . (1956) Listening between and during practiced auditory distraction.
British Journal of Psychology, 47, 51-60.
- - - . (1957a) A mechanical model for human attention and immediate memory. Psychological Review, 64j 205-215.
- - - . (1957b) Effects of noise on 'behavior. In Handbook of noise control,
ed. C. M. Harris. New York: McGraw-HilI. Pp. 1-33.
- - - . (1958) Perception and communication. London: Pergamon Press Ltd.
(1963) Possibilities and difficulties in the concept of arousal. In Vigilance: A ~ymposium, eds. D. N. Buckner and J. J. McGrath. New York:
McGraw-Hill. Pp. 184-198.
(1967) Word-frequency effect and response bias. Psychological Review, 74, 1-15.
(1970), Stimulus set and response set: Two kinds of selective attention. In Attention: Contemporary theories and analysis, ed. D. I. Mostofsky. New York: Appleton-Century-Crofts. Pp. 51-60.
. (1971) Decision and stress. London: Academic Press.
, and GREGORY, M. (1961) On the recall of stimuli presented alternately
to two sense organs. Quarterly Journal of Experimental Psychology, 13,
103-109.
, and GREGORY,M. (1963) Division of attention and the decision theory
of signal detection. Proceedings of the Royal Society, B, 158, 222-231.
-----, and GREGORY, M. (1964) Stimulus set and response set: The alterna. tionof attention. Quarterly Journal of Experimental Psychology, 16, 309317.
208
- - - , and GREGORY, M. (1965) Some confirmatory results on age differences

in memory for simultaneous stimulation. British Journal of Psychology,
56, 77-80.
- - - , and GREGORY, M. (1967) Psychological refractory period and the
length of time required to make a decision. Proceedings of the Royal
Society, B, 168, 181-193.
- - - , and HERON, A. (1962) Effects of a subsidiary task on performance
involving immediate memory by younger and older men. British Journal
of Psychology, 53, 189-198.
BROADHURST, P. L. (1957) Emotionality and the Yerkes-Dodson law. Journal of
Experimental Psychology, 54, 345-352.
- - - . (19-59) The interaction of task difficulty and motivation: The YerkesDodson law revived. Acta Psychologica, 16, 321-338.
BROOKS, L. R. (1967) The suppression of visualization by reading. Quarterly
Journal-of Experimental Psychology, 19, 289~299.
- - - . (1968) Spatial and verbal components of the act of recall. Canadian
Journal of Psychology/Review of Cariadian- Psychology, 22, 349-368.
- - - . (1970) An extension of the conflict between visualization and reading.
Quarterly Journal of Experimental Psychology, 22, 91-96.
BROWN, C. R., and RUBENSTEIN, H. (1961) Test of response bias explanation of
word-frequency. Science, 138, 280-281.
BROWN, I. D. (1964.) The measurement of perceptual load and reserve capacity.
Transactions of the Association of Industrial Medical Officers, 14, 44-49.
- - - . (1966) Measuring the difficulty of perceptual-motor tasks: Some limitations to dual-task techniques. Paper presented to the Ergonomics Research Society, Annual Conference.
- - - . (1968) Criticisms of time-sharing techniques for the measurement of
perceptual-motor difficulty. XVI International Congress of Applied Psychology. Amsterdam: Swets & Zeitlinger. Pp. 322-328.
- - - , SIMMONS, D. C., and TICKNER, A. H. (1967) Measurement of control
skills, vigilance, and performance on a subsidiary task during 12 hours of
car driving. Ergonomics, 10, 665-673.
BROWN, J. (1970) Recognition and the direction of attention. Acta Psychologica,
33, 149--157.
BROWN, L. T., and FARHA, W. (1966) Some physical determinants of viewing
time under three instructional sets. Perception' and Psychophysics, 1,
2-4.
- - - , and GREGORY, L. P. (1968) Attentional response of humans and squirrel
monkeys to visual patterns: Final studies and resume. Perceptual and
Motor Skills, 27, 787-814.
BRUNER, J. S. (1957) On perceptual readiness. Psychological Review, 64, 123152.
References
209
BRYDEN, M. P. (1961) The role of post-exposuraI eye movements in tachistoscopic perception. Canadian Journal of Psychology, 15, 220~225.
- - - . (1962) Order of report in dichotic listening. Canadian Journal of Psychology, 16, 291-299.
- - - . (1964) The manipulation of strategies of report in dichotic listening.
Canadian Journal of Psychology, 18, 126-138~
- - - . (1971) Attentional strategies and short-term memory iIi dichotic listening. Cognitive Psychology, 2, 99-116.
BURSILL, A. E. (1958) The restriction of peripheral vision during exposure to
hot and humid conditions. Quarterly Journal of Experimental Psychology,
10, 113-129.
CALLAWAY, E. (1959) The influence of amobarbital (amylobarbitone) and
methamphetamine on the focus of attention. Journal of Mental Science,
105, 382-392.
- - - , and STONE, G. (1960) Re-evaluating the focus of attention. In Drugs
and behavior, eds. L. Uhr and J. G. Miller. New York: John Wiley. Pp.
393-398.
- - - , and THOMPSON, S. V. (1953) Sympathetic activity and perception:
An approach to the relationships between autonomic activity and personality. Psychosomatic Medicine, 15, 443-455.
CAMPOS, J. J., and JOHNSON, H. J. (1966) The effects of verbalization instructions and visual attention on heart rate and skin conductance. Psychophysiology, 2, 305-310.
- - - , and JOHNSON, H. J. (1967) Affect, verbalization, and directional fractionation of autonomic responses. Psychophysiology, 3, 285-290.
CARVER, R. P. (1971) Pupil dilation and its relationship to information processing during reading and listening. Journal of Applied Psychology, 55,
126-134.
CATLIN, J. (1969) On the word-frequency effect. Psychological Review, 76,
504-506.
CAVANAGH, J. P., and CHASE, W. G. (1971) The equivalence of target and nontarget processing in visual search. Perception and Psychophysics, 9, 493495.
CHASE, W. G., GRAHAM, F. K., and GRAHAM, D. T. (1968) Components of HR
response in anticipation of reaction time and exercise tasks. Journal of
Experimenta.l Psychology,. 76, 642-648.
CHERRY, E. C. (1953) Some experiments on the recognition of speech, with
one and two ears. Journal of the Acoustical Society of America, 25, 975979.
. (1957) On human communication: A review, a survey, and a criticism.
Cambridge: Technology Press, MIT; New York: John Wiley.
, and TAYLOR, W. K. (1954) Some further experiments upon the recognition of speech with one and two ears. Journal of the Acoustical Society
of America, 26, 554-559.
210
COHEN, J. (1969) Very slow brain potentials relating to expectancy: The CNV.
In Average evoked potentials: Methods, results, and evaluations, eds. E.
Donchin and D. B. Lindsley. NASA, SP-191, Washington, D.C. Pp.
143-198.
.
COHEN, M. J., and JOHNSON, H. J. (1971) Relationship between heart rate and
muscular activity within a classical conditioning paradigm. Journal of
COLAVITA, F. B. (1971) Interspecies differences in sensory dominance. Paper
presented at the Twelfth Annual Meeting of the Psychonomic Society,
St. Louis, Missouri, November.
COLMAN, F. D., and PAIVIO, A. (1969) Pupillary response and galvanic skin
response during an imagery task. Psychonomic Science, 16, 296-297.
- - - , and PAIVIO, A.(1970) Pupillary dilation and mediation process during
paired-associate learning. Canadian Journal of Psychology, 24, 261-270.
CONNOR, W. H., and LANG, P. J. (1969) Cortical slow-wave and cardiac-rate
responses in stimulus orientation and RT conditions. Journal of Exp~ri
mental Psychology, 82, 310-320.
COOMBS, C. H., DAWES, R. M., and TVERSKY, A. (1970) Mathematical psychology: An elementary introduction. Englewood Cliffs, N.J.: PrenticeHall.
COQUERY, J.M., and LACEY, J. I. (1966) The effects of foreperiod duration on
the components of the cardiac response during the foreperiod of a
reaction-time experiment. Paper presented at the Meeting of the Society
for Psychophysiological Research, Denver, October.
CORBALLIS, M. C., and; LUTHE, L. (1971) Two-channel visual memory. Perception and Psychophysics, 9, 361-367.
CORCORAN, D. W. J. (1965) Personality and the inverted-U relation. British
Journal of Psychology, 56, 267-273.
CORNSWEET, D. J. (1969) Use of 'Cues in the visual periphery under conditions
of arousal. Journal of Experimental Psychology, 80, 14-18.
CORTEEN, R. S. (1969) Skin conductance changes and word recall. British
- - - , and WOOD, B. (1972) Autonomic responses to shock-associated words
in an unattended channel. Journal of Experimental Psychology, 94,
308-313.
CRAFT, J. L., and SIMON, J. R. (1970) Processing symbolic information from a
visual display: Interference from an irrelevant directional cue. Journal of
CRAIK, K. J. W. (1947) Theory of the human operator in control system~ I.
The operator as an engineering system. British Journal of Psychology, 38,
56-61.
------. (1948) Theory of the human operator in control systems II. Man as an
element ina control system. British Journal of Psychology, 38, 142-148.
CREAMER, L. R. (1963) Event uncertainty, psychological refractory period, and
human data processing. Journal of Experimental Psychology, 66, 187194.
References
211
CROVITZ, H. F., and DAVES, W. (1962) Tendencies to eye movement and perceptual accuracy. Journal of Experimental Psychology, 63, 495-498.
CROWDER, R.G. (1967) Prefix effects in immediate memory. Canadian Journal
- - - . (1969) Phonic interference and the prefix effect. Journal of Verbal
Learning and Verbal Behavior, 8, 302-321.
- - - . (1971) Waiting for the stimulus suffix: Decay, delay, rhythm, and
readout in immediate memory. Quarterly Journal of Exp'erimental Psychology, 23, 324-340.
- - - , and MORTON, J. (1969) Precategorical acoustic storage (PAS). Perception and Psychophysics, 5, 365-373.
DALLETT, K. M. (1964) Effects of a supplementary prefix on recall. Journal
of Experimental Psychology, 67, 296--298.
DAVIES, D. R. (1968) Physiological and psychological effects of exposure to
high intensity noise. Applied Acoustics, 1, 215-233.
- - - , and HOCKEY, G. R. J. (1966) The effects of ,~oise and doubling the
'signal frequency on individual differences, in visual vigilance performance. British Journal of Psychology, ,57, 381-389.
- - - , and TAYLOR, A. (1969) Varied auditory stimulation, temperament
differences and vigilance performance. British Journal of Psychology, 60,
453-457.
DAVIS, R. (1957) The human operator as a single channel information system.
Quarterly Journal of Experimental Psychology, 9, 119-129. '
- - - . (1959) The role of "attention" in the psychological refractory period.
Quarterly Journal of Experimental Psychology, 11, 211-220.
- - - . (1962) Choice reaction times and the theory of intermitt~ncyin human
performance. Quarterly Journal of Experimental' Psychology, 14, 157166.
DAVIS, R. C. (1957) Response patterns. Transactions of the New York Academy
of Sciences, series 2, 19, 731-737.
DAY, H. (1965) Exploratory behavior as a function of individual differences
and level o~ arousal. Ph.D. dissertation, University of Toronto.
- - - . (1966) Looking time as a function of stimulus variables and individual
differences. Perceptual and Motor Skills, 22, 423-428.
DAY, M.E. (1964) An eye movement phenomenon relating to attention,
thought and anxiety. Perceptual and Motor Skills, 19, 443-446.
- - - . (1967a) An eye-movement indicator of type and level of anxiety:
Some clin'ical observations. Journal of Clinical Psy'chology, 23, 438-441.
- - - . (1967b) An eye movement indicator of individual differences in the
phYSiological organization of attentional processes and anxiety. Journal
of Psychology, 66, 51-62.'
DEANE~ G. E. (1961) Human heart rate responses during experimentally induced anxiety. Journal of Experimental Psychology, 61, 489-493.
DE GROOT, A. D. (1966) Perception and memory versus thought: Some old
ideas and recent findings. In Problem solVing: Research, method, and
theory, ed. B. Kleinmuntz. New York: John Wiley.
212
DEUTSCH, J. A., and DEUTSCH, D. (1963) Attention: Some theoretical considerations. Psychological Review, 70, 80-90.
- - - , and DEUTSCH, D. (1967) Comments on "Selective attention: Perception or response?" Quarterly Journal of Experimental Psychology, 19,
362-363.
DILLON, R. F., and REID, L. S. (1969) Short-term memory as a function of information processing during the retention interval. Journal of Experimental Psychology, 81, 261-269.
DIMOND, S. (1970) Reaction times and response competition between the
right and left hands. Quarterly Journal of Experimental Psychology, 22,
513-520.
DUFFY, E. (1957) The psychological significance of the concept of "arousal"
or "activation." Psychological Review, :64, 265-275.
DURHAM, R. L., NUNNALLY,J. C., and LEMOND, L. C. (1971) The effects of
levels of information conflict on visual selection. Perception and Psychophysics, 10, 93-96.
EASON, R. G. (1963) Relation between effort, tension level, skill, and performance efficiency in a perceptual-motor task. Perceptual and Motor
Skills, 16, 297-317.
EASTERBROOK, J. A. (1959) The effect of emotion on cue utilization and the
organization of behavior. Psychological Review, 66, 183-201.
ECETH, H. E. (1967) Selective attention. Psychological Bulletin, 67, 41-57.
- - - , BLECKER, D. L., and KAMLET, A. S. (1969) Verbal interference in a
perceptual comparison task. Perception and Psychophysics, 6, 355-356.
- - - , and PACHELLA, R. (1969) Multidimensional stimulus identification.
EISLER, H., and KNOPPEL, J. K. (1970) Relative attention in judgments of
heterogeneous similarity. Perception and Psychophysics, 8, 420-426.
ELITHORN, A. (1961) Central intermittency: Some further observations. Quarterly Journal of Experimental Psychology, 13, 240-247.
- - - , and LAWRENCE, C. (1955) Central inhibition: Some refractory observations. Quarterly Journal of Experimental Psychology, 7, 116-127.
ELLIOTT, R. (1969) Tonic he~rt rate: Experiments on the effects of collative
variables lead to a hypothesis about its motivational significance. Journal
of Personality and Social Psychology, 12, 211-228.
- - - , BANKART, B., and LIGHT, R. (1970) Differences in the motivational
significance of heart rate and palmar conductance: Two tests of a hypoth.esis. Journal of Personality and Social Psychology, 14,166-172.
ELLIS, S. H. ~nd CHASE, W. G. (1971) Parallel processing in item recognition.
ELLS, J. (1969) Attention requirements of movement control. Unpublished
Ph. D. dissertation, University of Oregon.
EpSTEIN, W., and ROCK, I. (1960) Perceptual set as an artifact of recency.
American Journal of Psychology, 73, 214-228.
References 213
ERIKSEN, C. W., and COLLINS, J. F. (1969a) Visual perception rate under two
conditions of search. Journal of Experimental Psychology, 80, 489-492.
- - - , and COLLINS, J. F. (1969b) Temporal course of selective attention.
- - - , and HOFFMAN, J. E. (1972) Some characteristics of selective attention
in visual perception determined by vocal reaction time. Perception and
Psychophysics, 11, 169-171.
- - - , and LAPPIN, J. S. (1967) Selective attention and very short-term recognitionmemory for nonsense forms. Journal of Experimental Psychology,
73, 358-364.
- - - , and ROHRBAUGH, J. W. (1970) Some factors determining efficiency of
selective attention. American Journal of Psychology, 83, 330-342.
EXLINE, R. V. (1963) Exploration in the process of person perception: Visual
interaction in response to competition, sex and need for affiliation. Journal
of Personality, 31, 1-20.
- - - . (1971) Visual interaction: The glances of power and preference. In
Nebraska symposium on motivation, ed. J. K. Cole. Lincoln: University
of Nebraska Press. Pp. 163-206.
E.YSENCK, H. J. (1967) The biological basis of personality. Sp.ringfield, Ill.:
Charles C. Thomas.
- - - , and THOMPSON, W. (1966) The effects of distraction on pursuit rotor
learning, performance and reminiscence. British Journal of Psychology,
57, 99-106.
FANTZ,R. L. (1958) Pattern vision in young infants. Psychological Record,
8,43-47.
- - - . (1965a) Visual perception from birth as shown by pattern selectivity.
Annals of the New York' Academy of Sciences, 1'18, 793-814.
- - - . (1965b) The ontogeny of perception. In Behavior of non-human
primates, eds. A. M. Schrier, H. F. Harlow, and F. Stolhitz. New York:
Academic Press. Pp. 365-463.
- - - . (1967) Visual perception and experience in early infancy: A look at
the hidden side of behavior development. In Early behavior: Comparative and developmental approaches, eds. H. W. Stevenson, E. H. Hess,
and H. L. Rheingold. New York: John Wiley.
FEHRER, E., and RAAB, D. (1962) Reaction time to stimuli masked by metacontrast. Journal of Experimental Psychology, 63, 143-147.
FELFOLDY, G. L., and GARNER, W. R. (1971) The effects on speeded classification of implicit and explicit instructions regarding redundant dimensions. Perception and Psychophysics, 9, 289-292.
FELLOWS,. B. J. (19'68) The discrimination process and development. Oxford:
Pergamon Press.
FITTS, P. M., and BIEDERMAN, I. (1965) S-R compatibility and information reduction. Journal of Experimental Psychology, 69, 408-412.
- - - , and POSNER, M. I.' (1967) Human performance. Belmont, Calif.:
Brooks Cole Publishing Co.
.214
FODOR, J. A., and BEVER, T. G. (1965) The psychological reality of .linguistic

segments. Journal of Verbal Learning and Verbal Behavior, 4, 414-420.
Fox, L.A., SHOR, R. E., and STEINMAN, R. J. (1971) Semantic gradients and
interference in naming color, spatial direction, and numerosity. Journal
.
FRAISSE, P. (1957) La periode refractaire psychologique. Annee Psychologique,
57, 315-328.
- - - , EHRLIcH,S.,and VURPILLOT, E. (19-56) Etudes de lacentration perceptive par la methode tachistoscopique. Archives de Psychologie
Geneve,35,193-214.
FREUD, S. (1900) The interpretation of dreams. (Original German ed., New
York: Basic Books, 1955).
FUREDY, J. J., and SCULL, J. (1971) Orienting-reaction theory and an increase
in the human GSR following stimulus change which is unpredictable but
not contrary to predictio,n. Journal of Experimental Psychology, 88, 292294.
FURST, C. J. (1971) Automatizing of visual attention. Perception and Psychophy~cs,10, 65-70.
'
GARNER, W. R. (1962) Uncertainty and struct'!Jre as psychological concepts.
New York: John Wiley.
- - - . (1970) The stimulus in information processing. American Psychologist,
25, 350--358.
---,a~d FELFOLDY, G. L. (1970) Integrality of stimulus dimensions in
various types of information processing. Cognitive Psychology, 1, 225241.
GARVEY, W. D., and TAYLOR, F. V. (1959) Interactions among operator variables, system dynamics, and task-induced stress. Journal of Applied
Psychology, 43, 79-85.
GERMANA, J. (1968) Response characteristics and the orienting reflex. Journal
GESELL, A., and ILG, F. L. (1949) Child development: An introduction tt;J
study of human 'growth. New York: Harper & Row.
GIBSON, E. J. (1965) Learning to read. S'cience,148, 1066-1072.
- - - . (1969) Principles of perceptual learning and development. New York:
Appleton-Century-Crofts.
GIBSON, J. J., and PICK, A. D. (1963) Perception of another person's looking
behavior. American Journal of Psychology, 76, 386-394.
GOLDWATER, B. C. (1972) Psychological significance of pupillary movements.
Psychological B~lletin, 77, 340-355.
GOODWIN, W. R., and LAWRENCE, D. H. (1955) The functional independence
of two discrimination habits associated with a constant stimulus situation.
Journal of Comparative and Physiological Psychology, 48, 437-443.
GOPHER, D. (1971) Patterns of eye movement in auditory tasks of selective
References
215
attention. Unpublished Ph.D. thesis, Hebrew University, Jerusalem.

(Hebrew)
- - - , and KAHNEMAN, D. (1971) Individual differences in attention and the
prediction of flight criteria. Perceptual and Motor Skills, 33, 1335-1342.
GORDON, I. E. (1969) Eye movements during search through printed lists. Perceptual and Motor Skills, 29, 683-686.
- - - , DULEWICZ, V., and WINWOOD, M. (1971) Irrelevant item variety and
visual search. Journal of Experimental Psychology, 88, 295-296.
GOTTSDANKER, R. (1969) Interacting responses to crowded signals. Acta Psychologica, 30~ 168-176.
- - - , BROADBENT, L., and VAN SANT, C. (1963) Reaction time to single and
to first signals. Journal of Experimental Psychology, 66, 163-167.
GOTTWALD, R. L., and GARNER, W. R. (1972) Effects of focusing strategy on
speeded classification with grouping, filtering, and condensation tasks.
GOULD, J. D. (1967) Pattern recognition and eye-movement parameters. Perception and PsychophysiCS, 2, 399-407.
- - - , and DILL, A. B. (1969) Eye-movement parameters and pattern discrimination. Perception and Psychophysics, 6, 311-320.
- - - , and PEEPLES, D. R. (1970) Eye movements during visual search and
discrimination of meaningless, symbol, and object patterns. Journal of Experimental Psychology, 85, 51-55.
- - - , and SCHAFFER, A. (1965) Eye-movement patterns in scanning numeric
displays. Perceptual Motor Skills, 20, 521-535.
GRAHAM, F.K., and CLIFTON, R. K. (1966) Heart-rate change as a component
of the orienting response. Psychological Bulletin, 65, 305-320.
GRAY, J. A., and WEDDERBURN, A. A. I. (1960) Grouping strategies with simnl..
taneous stimuli. Quarterly Journal of Experimental Psychology, 12, 180184.
GREEN, D. M., and SWETS, J. A. (1966) Signal detection theory and psychophysics. New York: John Wiley.
GREENWALD, A. G. (1970a) A double stim.ulation test of ideomotor theory with
implications for selective attention. Journal of Experimental Psychology,
84, 392-398.
- - - . (1970b) Selective attention as a function of signal rate. Journal of
Experimental Psychology, 86,48-52.
- - - . (1970c) Sensory feedback mechanisms in performance control: With
special referen~e to the ideo-motor mechanism. Psychological Review,
77,73-100.
GRINDLEY, G. C., and TOWNSEND, V. (1968) Voluntary attention in peripheral
vision and its eff.ects on acuity and differential thresholds. Quarterly JourfUll of Experimental Psychology, 20, 11-19.
HABER, R. N. (1964a) A replication of selective attention and coding in visual
perception. Journal of Experimental Psychology, 67, 402-404.
216
- - - . (1964b) Effects of coding strategy on perceptual memory. Journal of

Experimenta~ Psychology, 68, 357-362.
- - - . (1966) Nature of the effect of set on perception. Psychological Review,
73, 335-351.
HAKEREM, G., and SUTTON, S. (1966) Pupillary response at visual threshold.
Nature, 212, 485-486.
HARRIS, C. S., and HABER, R. N. (1963) Selective attention and coding in visual
perception. Journal of Experimental Psychology, 65, 328-333.
HASTINGS, S. E., and OBRIST, P. A. (19'67) Heart rate during conditioning in
humans: Effect of varying the interstimulus (CS-UCS) interval. Journal
HAWKINS, H. L. (1969) Parallel processing in complex visual discrimination.
HEIDER, F., and SIMMEL, M. (1944) An experimental study of apparent behavior. American Journal of Psychology, 57, 243-259.
HELD, R. (1968). Dissociation of visual functions by deprivation and rearrangement. Psychologische Forschung, 31, 338-348.
HENIK, A. (1972) Intrusions in dichotic listening. Unpublished M. A. thesis,
Hebrew University, Jerusalem. (Hebrew)
HERMAN, L. M. (1969) Effects of second signals on response time to first signals
under certainty and uncertainty. Journal of Experin~ental Psychology, 80,
106-112.
-_.- , and KANTOWITZ, B. H. (1970) The psychological refractory period
effect: Only half the double-stimulation story? Psychological Bulletin,
73,74-88.
- - - , and MCCAULEY, M. E. (1969) Delay in responding to the first stimulus in the "psychological refractory period" experiment: Comparisons
with delay produced by a second stimulus not requiring a response.
HERNANDEZ-PEON, R. (1964) Psychiatric implications of neurophysiological
research. Bulletin of the Menninger Clinic, 28, 165-185.
HERSHENSON,M., MUr--rSINGER, H., and KESSEN, W. (1965) Preference for
shapes of intermediate variability in the newborn human. Science, 147,
630-631.
HESS, E. H. (1965) Attitude and pupil size. Scientific American, 212, 46-54.
- - - , and POLT, J. M. (1964) Pupil size in relation to mental activity during
simple problem-solving. Science, 143, 1190-1192.
HILLYARD, S. A. (1969) Relationships between the contingent negative variation (CNV) and reaction time. Physiology and Behavior, 4, 351-357.
HINTZMAN, D. L., CARRE, F. A., ESKRIDGE, V. L., OWENS, A.M., SHAFF, S. S.,
and SPARKS, M. E. (1972) "Stroop" effect: Input or output phenomenon?
HOCHBERG, J. E. (1970) Attention, organization, and consciousness. In Atten-
References
217
tion: Contemporary theory and analysis, ed. D. I. Mostofsky. New York:

Appleton-Century-Crofts. Pp. 99-124.
HOCKEY, G. R. J. (1969) Noise and efficiency: The visual task. New Scientist,
42, 244-246.
- - - a (1970a) Effect of loud noise on attentional selectivity. Quarterly Journal of Experimental Psychology, 22, 28-36.
- - - a (1970b) Signal probability and spatial location as possible bases for
increased selectivity in noise. Quarterly Journal of Experimental Psychology, 22, 37-42.
- - - a (1970c) Changes in attention allocation in a multicomponent task
under loss of sleep. British Journal of Psychology, -61, 473-480.
HOUSTON, B. K. (1968) Inhibition and the facilitating effect of noise on interference tasks. Perceptual and Motor Skills, 27, 947-950.
- - - a (19-69) Noise, task difficulty, and Stroop color-word performance.
- - - , and JONES, T. M. (1967) Distraction and Stroop color-word performance. Journal of Experimental Psy-chology, 74, 54-56.
HOWES, D. H., and SOLOMON, R. L. (1951) Visual duration threshold as a function of word-probability. Journal of Experimental Psychology, 41, 401410.
HYMAN, R., and WELL, A. (1968) Perceptual separability and spatial models.
Perception and PsychophYSiCS, 3, 161-165.
IMAI, S., and GARNER, W. R. (1965) Discriminability and preference for attributes in free and constrained classification. Journal of Exp'erimental
INGLE, D. (1967) Two visual mechanisms underlying the behavior of fish.
Psychologische Forschung, 31, 44-51.
ITTELSON, W. H. (1952) The Ames demonstrations in perception: A guide to
their construction and use. Princeton, N.J.: Princeton University Press.
- - - , and KILPATRICK, R. (1951) Experiments in perception. Scientific
American, 185, 50-55.
JAMES, W. (1890) The principles of psychology. New York: Holt.
JENKS, R. S., and DEANE,G. E. (1963) Human heart rate responses during experimentally induced anxiety: A follow-up. Journal of Experimental
JENSEN, A. R. (1965) Scoring the Stroop test. Acta Psychologica, 24, 398-408.
- - - , and ROHWER, W. D. (1966) The Stroop color-word test: A review.
Acta Psychologica, 2-5, 36-93.
JOHNSON, D. A. (1971) Pupillary responses during a short-term memory task:
Cognitive processing, arousal, or both? Journal of Experimental Psychology, 90, 311-318.
JOHNSON, H. J., and CAMPOS, J. J. (1967) The effect of cognitive tasks and
verbalization instructions on heart rate and skin conductance. Psychophysiology,4, 143-150.
218
JOHNSTON, W. A., GREENBERG, S. N., FISHER, R. P., and MARTIN, D. W. (1970)

Divided attention: A vehicle for monitoring memory processes. Journal of
JONES, L. V. (1954) Distinctiveness of color, form, and position cues for
pigeons. Journal of Comparative and Physiological Psychology, 47, 253257.
JONGMAN, R. W. (1968) Het oog van de meester, Amsterdam: van Gorcum.
KAFRY, D. (1971) Task complexity and refractoriness in the RSI design. Unpublished M.A. thesis, Hebrew University, Jerusalem. (Hebrew)
KAGAN, J. (1972) Do infants think? Scientific American, 226, 74-82.
---,and LEMKIN, J. (1961) Form, color and size in children's conceptual
behavior. Child Development, 32, 25-28.
- - - , and LEWIS, M. (1965) Studies of attention in the human infant.
Merrill-Palmer Quarterly, 11, 9'5-122.
KAHNEMAN, D. (1968) Method, findings, and theory in studies of visual masking. Psychological Bulletin, 70, 404-425.
- - - . (1970) Remarks on attention control. Acta Psychologica, 33, Attention
and performance III, ed. A. F. Sanders, 118-131.
- - - , and BEATTY, J..(19'66) Pupil diameter and load on memory. Science,
154, 1583-1585.
- - - , and BEATTY, J. (1967) Pupillary responses in a pitch-discrimination
task. Perception and Psychophysics, 2, 101-105.
- - - , BEATTY, J., and POLLACK, I. (1967) Perceptual deficit during a mental
task. Science, 157, 218-219.
- - - , BEN-ISHAI, R., and LOTAN, M. (1972) Relation of a test of attention to
road accidents. Journal of Applied Psychology, in press.
- - - , and LASS, N. (19'71) Eye position in tasks of association and memory.
Unpublished manuscript; Hebrew Uni,versity, Jerusalem.
- - - , ONUSKA, L., and WOLMAN, R. E. (1968) Effects of grouping on the
pupillary response in a short-term memory task. Quarterly Journal of
- - - , and PEAVLER, W. S. (1969) Incentive effects and pupillary changes in
association learning. Journal of Experimental Psychology, 79, 312-318.
- - - , PEAVLER, W. S., and ONUSKA, L. (1968) Effects of verbalization and
incentive on the pupil response to mental activity. Canadian Journal of
- - - , TURSKY, B., SHAPIRO, D., and CRIDER, A. (1969) Pupillary, heart rate
and skin resistance changes during a mental task. Journal of Experimental Psychology, 79, 164-167.
- - - , and WRIGHT, P. (1971) Changes of pupil size and rehearsal strategies
in a short-term memory task. Quarterly Journal of Experimental Psychology, 23, 187-196.
KALSBEEK, J. W. H., and ETTEMA, J. H. (1963) Continuous recording of heart
rate and the measurement of perceptual load. Ergonomics, 6, 306-307.
References
219
- - - , and ETTEMA, J. H. (1964) Physiological and psychological evaluation

of distraction stress-Communication at the Second International Ergonomic Conference, Dortmund. Proceedings of the Second International
Congress on Ergonomics, Supplement to Ergonomics, pp. -443-447.
KAPLAN, I. T., and SCHOENFELD, W. N. (1966) Oculomotor patterns during the
solution of visually displayed anagrams. Journal of Experimental Psychology, 72, 447-451.
KARLIN, L., and KESTENBAUM, R. (1968) Effects of number of alternatives on
the psychological refractory period. Quarterly Journal of Experimental
KAUFMAN, L., and RICHARDS, W. (1969) Spontaneous fixation tendencies for
visual forms. Perception and Psychophysics, 5, 85-88.
KAY, H., and WEISS, A. D. (1961) Relationship between simple and serial reaction time. Nature, 191, 790-791.
KEEFE, F. B., and JOHNSON, L. C. (1970) Cardiovascular responses to auditory
stimuli. Psychonomic Science, 19, 335-337.
KEELE, S. W. (1967) Compatibility and time-sharing in serial reaction time.
- - - . (1970) Effects of input and output modes on decision time. Journal of
- - - . (1972) Attention demands of memory retrieval. Journal of Experimental Psychology, 93, 245-248.
- - - . (1973) Attention and human performance. Pacific Palisades, Calif.:
Goodyear.
KEELEY, S. M. (1969) Visual detection as a function of attentional demand and
perceptual system error. Perception and PsychophysiCS, 6, 73-77.
KENDLER, H. H., and KENDLER, T. S. (1962) Vertical and horizontal processes
in problem solving. Psychological Review, 69, 1-16.
- - - , and KENDLER, T. S. (1970a) Developmental processes in discrimination learning. Human DevelolJment, 13, 65-89.
KENDLER, T. S., and KENDLER, H. H. (1970b) An ontogeny of optimal shift
behavior. Child Development, 41, 1-27.
KENDON, A. (1967) Some functions of gaze-direction in social interaction. Acta
Psychologica, 26, 22-63.
KINSBOURNE, M. (1970) The cerebral basis of lateral asymmetries in attention.
Acta Psychologica, 33, 193-201.
- - - . (1972) Eye and head turning inqicates cerebrallateralization. Science,
176, 539-541.
KLEIN, G. S. (1964) Semantic power measured through the interference of
words with color-naming. American Journal of Psychology, 77, 576-588.
KLEINSMITH, L. J., and KAPLAN, S. (1963) Paired-associate learning as a function of arousal and interpolated interval. Journal of Experimental Psychology,65, 190-193.
220
-----, and KAPLAN, S. (1964) Interaction of arousal and recall interval in

nonsense syllable paired-associate learning. \ Journal of Experimental
KLEITMAN, N. (1963) Sleep and wakefulness. Chicago: University of Chicago
Press.
KOFFKA, K. (1935) Principles of gestalt psychology. New York: Harcourt,
Brace.
KOHN, B., and BRYDEN, M. P. (1965) The effect of lysergic acid diethylamide
(LSD-25) on perception with stabilized images. Psychopharmacologica,
7,311-320.
KORCHIN, S. J. (1964) Anxiety and cognition. In Cognition: Theory, research,
promise, edt C. Scheerer. New York: Harper & Row. Pp. 58-78.
KORNBLUM, S., edt (1972) Attention and performance IV. New York: Academic Press, in press.
KRECHEVSKY, I. A. (1932) "Hypotheses" in rats. Psychologicf!l Review, 89,
516-532.
- - - I (1938) A study of the continuity of the problem-solving process.
Psychological Review, 45, 107-133.
KRISTOFFERSON, A. B. (1967) Successive discrimi~ation as a two-state, quantal
process. Science, 158, 1337-1339.
KROLL, N. E. A., PARKS, T., PARKINSON, S. R., BIEDER, S. L., and JOHNSON,
A. L. (1970) Short-term memory while shadowing: Recall or visually and
of aurally presented letters. Journal of Experimental Psychology, 85,
220-224.
KULPE, O. (1904) ,Versuche tiber Abstraktion. Berlin International Congress
of Experimental Psychology, 56-68.
LA BERGE, D., and WINOKUR, S. (1965) Short-term memory using a visual
shadowing procedure. Psychonomic Science, 8, 239.;...240.
LACEY, B. C., and L~CEY, J. I. (1964) Cardiac deceleration and simple visual
reaction time in a fixed foreperiod experiment. Paper presented at the
meeting of the Society for Psychophysiological Research, Washington,
D.C., October.
- - - , and LACEY, J. I. (1966) Change in cardiac response and reaction
time as a function of motivation. Paper presented at the meeting of the
Society for Psychophysiological Research, Denver, October.
LACEY, J. I. (1959) Psych.ophysiological approaches to the evaluation of the
psychotherapeutic process and outcome. In Research in psychotherapy,
eds. E. A. Rubenstein and M. B. Parloff. Washington, D.C.: American
Psychological Association.
- - - I (1967) Somatic response patterning and stress: Some revisions of
activation theory. In Psychological stress, eds. M. H. Appley and R.
Trumbull. New York: Appleton-Century-Crofts.
- - - , KAGAN, J., LACEY, B. C., and Moss, H. A. (1963) The visceral level:
Situational determinants and behavioral co-rrelates of autonomic response
References
221
patterns. In Expression of the emotions of man. New York: International

Universities Press. Pp. 161-196.
LADEFOGED, P., and BROADBENT, D. E. (1960) Perception of sequence in auditory events. Quarterly Journal of Experimental Psychology, 12, 162-170.
LAPPIN, J. S. (1967) Attention in the identification of stimuli in complex visual
displays. Journal of Experimental Psychology, 75, 321-328.
LASHLEY, K. S. (1942) An examination of the "continuity theory" as applied to
discrimination learning. Journal of General Psychology, 26, 241-265.
- - - . (1951) The problem of serial order in behavior. In Cerebral mechanisms in behavior, ed. L. A. Jeffries. New York: John Wiley. Pp. 112136.
- - - , and WADE, M. (1946) The Pavlovian theory of generalization. Psychological,Review, 53, 72-87.
LAWRENCE, D. H. (1949) Acquired distinctiveness of cues: I. Transfer between discrimination of the basis of familiarity with the stimulus. Journal
- - - . (1950) Acquired distinctiveness of cues: II. Selective association in
a constant stimulus situation. Journal of Experimental Psychology, 40,
175-188.
- - - . (1963) The nature of a stimulus: Some relationships between learning and perception. In Psychology: A study of a science, Vol. 5, ed. S.
-Koch. New York: McGraw-Hill. Pp. 179~212.
- - - . (1971) Two studies of visual search for word targets, with controlled
rates of presentation. Perception and Psychophysics, 10, 85-89.
- - - , and LABERGE, D. L. (1956) Relationship between recognition accuracy and order of repor~ing stimulus dimensions. Journal of Experimental Psychology, 51, 12-18.
LAWSON, E. A. (19-66) Decisions concerning the rejected channel. Qttarterly
LEVINE, M. (1962) Cue neutralization: The effects of random reinforcements
upon discrimination learning. Journal of Experimental Psychology, 63,
438-443.
LEVY, R. (1971) Recognition of dichotic word lists in focused and divided attention. Unpublished M.A. thesis, Hebrew University, Jerusalem. (Hebrew)
LEWIS, J. L. (1970) Semantic processing of unattended messages using dichotic
listening. Journal of Experimental Psychology, 85, 225-228.
LEWIS, M., KAGAN, J., CAMPBELL, H., and KALAFAT, J. (1966) The cardiac
response as a correlate of attention in infants. Child Development, 37,
63-71.
.
- - - , and SPAULDING, S. J. (1967) Differential cardiac response to visual and
auditory stimulation in the young child. Psychophysiology, 3, 229~237.
LIBBY, W. L., JR., LACEY, B. C., and LACEY, J. I. (1973) Pupillary and cardiac
activity during visual attention. Psychophysiology, in press.
222
LIBERMAN, A. M., COOPER, F. S., SHA~WEILER, D. P., and STUDDERTKENNEDY, M. (1967) Perception of the speech code. Psychological Review, 74, 431-461.
LINDSAY, P. H. (19'70) Multichannel processing in perception. In Attention:
Contemporary theory and analysis, ed. D. I. Mostofsky.New York:
Appleton-Century-Crofts. Pp. 149-171.
- - - , CUDDY, L. L., and TULVING, E. (1965) Absolute judgments of simultaneously presented visual and auditory stimuli. Psychonomic Science, 2,
211-212.
- - - , and NORMAN, D. A. (1969) Short-term retention during a simultaneous
detection task. Perception and Psychophysics, 5, 201-205.
- - - , TAYLOR, M. M., and FORBES,S. M. (1968) Attention and multidimensional discrimination. Perception and Psychophysics, 4, 113-117.
LOCKHEAD,G. R. (1966a) Effects of dimensional redundancy on visual discrimination. Journal of Experimental Psychology, 72, 95-104.
- - - . (1966b) Visual discrimination and methods of presenting redundant
stimuli. Proceedings of the Seventy-fourth Annual Convention of the
American Psychological Association, 1, 67-68.
LO~ENS, S. A., JR., and DARROW, C. W. (1962) Eye movements, EEG, GSR
and EKG during mental multiplications.EEG and Clinical Neurophysiology, 14, 739-746.
LOVEJOY, E. P. (1966) Analysis of the overlearning reversal effect. Psych'ological Review, 73, 87-103.
- - - . (19'68) Attention in discrimination learning: A point of view and a
theory. San Francisco: Holden-Day.
LOWE, D., and MERIKLE, P. M. (1970) On the disruption of short-term memory
by a response prefix. Canadian Journal of Psychology, 24, 169-177.
LUBORSKY, L., BLINDER, B., and SCHIMEK, J. (1965) Looking, recalling and
GSR as a function of defense. Journal of Abnormal Psychology, 70,
270-280.
LURIA, A. R. (1961) The role of speech in the regulation of normal and abnormal behavior, ed. J. Tizard. New York: Liveright.
LYNN, R. (1966) Attention, arousal and the orientation reaction. Oxford:
Pergamon Press.
MACKINTOSH, N. J. (1964) Overtraining and transfer within and between dimensions in the rat. Quarterly Journal of Experimental Psychology, 16,
250-256.
- - - . (1965) Selective attention in animal discrimination learning. Psychological Bulletin, 64, 124-150.
- - - . (1971) An analysis of overshadowing and blocking. Quarterly Journal
MACKWORTH, J. F. (1969) Vigilance and habituation: A neuropsychological
approach. Harmondsworth, Middlesex, England: Penguin Books.
References
223
- - - . (1970) Vigilance and attention: A signal detection approach. Harmondsworth, Middlesex, England: Penguin Books.
MACKWORTH, N. H. (1965) Visual noise causes tunnel vision. Psychonomic
Science, 3, 67-68.
- - - , and BRUNER, J. S. (1970) How adults and children search and recognize pictures. Human Development, 13, 149-177.
- - - , and MORANDI, A. J. (1967) The gaze selects informative details within
pictures. Perception and Psychophysics, 2, 547-552.
MADSEN, M. C., ROLLINS, H. A., and SENF, G. M. (1970) Variables affecting
immediate memory for bisensory stimuli: Eye-ear analogue studies of
dichotic listening. Journal of Experimental Psychology Monograph
Supplement, 83, 1-16.
MALMO, R. B. (19'59) Activation: A neuropsychological dimension. Psychological Review, '66, 367-386.
- - - . (19'65) Physiological gradients and behavior. Psychological Bulletin,
64, 225-234.
- - _ . (1966) Cognitive factors in impairment: A neuropsychological study
of divided set. Journal of Experimental Psychology, 71, 184-189.
- - - , and SURWILLO, W. W. (1960) Sleep deprivation: Changes in performance and physiological indicants of activation. Psychological Monog.raphs, 74, 502.
MALTZMAN, I., HARRIS, L., INGRAM, E., and WOLFF, C. (1971) A primary
effect in the orienting reflex to stimulus change. Journal of Experimental
- - - , and RASKIN, D. C. (1965) Effects of individual differences in the
orienting reflex on conditioning and complex processes. Journal of Experimental Research in Personality, 1, 1-16.
MARCEL, A. J. (1970) Some constraints on sequential and parallel processing,
and the limits of attention. Acta Psychologica, 33, 77-92.
MASSARO, D. W. (1972) PreperceptuaI images, processing'time, and perceptual
units in auditory perception. Psychological Review, 79, 124-145.
MCCORMACK, P. D., and HALTRECHT, E. J. (1966) Monitoring eye movements
under two conditions of presentation of paired-associate materials.
Canadian Journal of Psychology, 20, 154-159.
- - - , HALTRECHT, E. J., and HANNAH, T. E. (1966) Monitoring eye move..
ments in nonlearning situations. Psychonomic Science, 6, 371-372.
- - - , HALTRECHT, E. J., and HANNAH, T. E. (1967) Monitoring eye movements during the learning of successive paired-associate lists. Journal of
Verbal Learning and Verbal Behavior, 6, 950-953.
- - - , and HANNAH, T. E. (1967) Monitoring eye movements during the
learning of high and low meaningfulness paired-associate lists. Psych0nomic Science, 8, 443-444.
- - - , and MOORE, T. E. (1969) Monitoring eye movements during the
224
learning of low-high and high-low meaningfulness paired-associate lists.

McDoNNELL, N. (1967) Effects of intensity, contrast and novelty on selective
attention and free-choice reaction-time. Unpublished M.A. thesis, University of Toronto.
McDoNNELL, P. M. (1970) The role of albedo and contrast in a test of selective attention. Perception and Psychophysics, 8, 270-272.
MICHON, J. A. (1964) A note on the measurement of perceptual motor load.
Ergonomics, 7, 461-463.
- - - . (1966) Tapping regularity as a measure of perceptual motor load.
Ergono111ics, 9, 401-412.
MICHOTTE, A. ,(1963) The perception of causality. London: Methuen.
MILLER, G. A., GALANTER, E., and PRIBRAM, K. (1960) Plans and the structure
of behavior. New York: Holt, Rinehart & Winston.
MONTAGUE, W. E. (1965) Effect of irrelevant information on a complex auditory-discrimination task. Journal of Experimental Psychology, 69, 230236.
MORAY, N. (19'59) Attention in dichotic listening: Affective cues and the influence of instructions. Quarterly Journal of Experimental Psychology, 11,
56-60.
- - - . (1967) Where is capacity limited? A survey and a model. Acta
- - - . (1969a) Listening and attention. Baltimore: Penguin Books.
- - - . (1969b) Attention: Selective processes in vision and hearing. London:
Hutchinson Educational Ltd.
- - - . (1970a) Time sharing in auditory perception,: Effect of stimulus duration. Journal of the Acoustical ,Society of America, 47, 660-661.
- - - . (1970b) Intro'ductory experiments in auditory time sharing: Detection
of intensity and frequency increments. Journal of the Acoustical Society
of America, 47, 1071-1073.
- - - , and BARNETT, T. (1965) Stimulus presentation and methods of scoring
in short-term memory experiments. Acta Psychologica, 24, 253-263.
- - - , and JORDAN, A. (1966) Practice and compatibility in two-channel
short-term memory. Psychonomic Science, 4, 427-428.
- - - , and O'BRIEN, T. (1967) Signal-detection theory applied to selective
listening. Journal of the Acoustical Society of America, 42, 765-772.
MORIN, R. E., FORRIN, B., and ARCHER, W. (1961) Information-processing behavior: The role of irrelevant stimulus information. Journal of Experimental Psychology, 61, 89-96.
MORTON, J. (1964) The effects of context upon speed of reading eye movements, and eye-voice span. Quarterly Journal of Experimental Psychology, 16, 340-355.
- - - . (1968) A retest of the response bias explanation of the word-frequency
References
225
effect. British Journal of Mathematical and Statistical Psychology, 21,

21-33.
- - - . (1969a) Interaction of information in word recognition. Psychological
Review, 76, 165-178.
- - - . (1969b) Categories of interference: Verbal mediation and conflict in
card sorting. British Journal of Psychology, 60, 329-346.
- - - . (1969c) The use of correlated stimulus information in card sorting.
- - - . (1970a) A functional model for memory. In Models of memory, ed.
D. A. Norman. New York: Academic Press.Pp. 203-254.
- - - . (1970b) The role of memory in the attention process. Acta Psychologica, 33, 271-279.
- - - , CROWDER, R. G., and PRUSSIN, H. A. (1971) Experiments with the
stimulus suffix effect. Journal of Experimental Psychology Monograph,
91, 169-190.
- - - , and HOLLOWAY, C. M. (1970) Absence of a cross-model "suffix effect"
in short-term memory. Quarterly Journal of Experimental Psychology,
22, 167-176.
MOWBRAY, G. H. (1953) Simultaneous vision and audition: The comprehension
of prose passages with varying levels of difficulty. Journal of Experimental Psycholog!J, 46, 365-372.
- - - . (1954) The perception of short phrases presented simultaneously for
visual and auditory reception. Quarterly Journal of Experimental Psychology, 6, 86-92.
- - - . (1962) Some human perceptual limits. APL Technical Digest, Jan.Feb., 21-24.
- - - . (19'64) Perception and retention of verbal information presented duringauditory shadowing. Journal of the Acoustical Society, 36, 14591465.
MURDOCK, B. B., JR. (1960) The immediate retention of unrelated words.
-_.-. (1965) Effects of a subsidiary task on short-term memory. British
MURRAY, D. J., and HITCHCOCK, C. H. (1969) Attention and storage in dichotic
listening. Journal of Experimental Psychology, 81, 164-169.
NAYLOR, J. C., BRIGGS, G. E., and REED, W. G. (1968) Task coherence, training time, and retention interval effects on skill retention. Journal of Applied Psychology, 52, 386-393.
NEISSER, u. (1963) Decision time without reaction time: Experiments in visual
scanning. American Journ{ll of Psychology, 76, 376-385.
- - - . (19'67) Cognitive psychology. New York: Appleton-Century-Crofts.
- - - . (1969) Selective reading: A method for the study of visual attention.
Presented at the Nineteenth International Congress of Psychology, London.
226
- - - , HOENIG, Y. J., and GOLDSTEIN, E. (1969) Perceptual org~nization in

the prefix effect. Journal of Verbal Learning and Verbal Behavior, 8,
424-429..
- - - , NOVICK, R., and LAZAR, R. (1963) Searching for ten targets simultaneously. Perceptual and Motor Skills, 17, 955-961.
- - - , and STOPER, A. (1965) Redirecting the search process. British Journal
NICKERSON, R. S. (1967) Psych.ological refractory phase and the functional
significance of signals. Journal of Experimental Psychology, 73, 303-312.
NINIO, A., and KAHNEMAN, D. (1973) Reaction time in focused and in divided
attention. Journal of Experimental Psychology, in press.
NORMAN, D. A. (1968) Towards a theory of memory and attention. Psychological Review, 75, 522-536.
- - - . (1969a) Memory and attention: An introd'f.lction to human information
processing. New York: John Wiley.
- - - . (1969b) Memory while shadowing. Quarterly Journal of Experimental
- - - , and RUMMELHART, D. E. (1970) A system for perception and memory.
In Models of human memory, ed. D. A. Norman. New York: Academic
Press. Pp. 19~64.
NOTTERMAN, J. M., SCHOENFELD, W. N., and BERSH, P. J. (1952) Conditioned
heart rate response in human beings during experimental anxiety. Journal of Comparative and Physiological Psychology, 45, 1-8.
NUNNALLY, J. C., FAW, T. T., and BASHFORD, M. B. (1969) The effects of degreesof incongruity on visual fixations of children and adults. Journal of
- - - , KNOTT, P. D., DUCHNOWSKI, A., and PARKER, R. (1967) Pupillary response as a general measure of activation. Perception and Psychophysics,
2, 149-155.
OBRIST, P. A. (1968) Heart rate and somatic-motor coupling during classical
aversive conditioning in humans.. Journal of Experimental Psychology,
77, 180-193.
- - - , WEBB, D. M., and SUTTERER, J. R. (1969) Heart rate and somatic
changes during aversive conditioning and a simple reaction-time task.
Psychophysiology,5, 696-723.
- - - , WEBB, D. M., SUTTERER, J. R., and HOWARD, J. L. (1970a) The
cardiac-somatic relationship: Some reformulations. Psychophysiology, 6,
569-587.
- - - , WEBB, D. M., SUTTERER, J. R., and HOWARD, J. L. (1970b) Cardiac
deceleration and reaction time: An evaluation of two hypotheses. Psychophysiology, 6, 695-706.
- - - , WOOD, D. M., and PEREZ-REYES, M. (1965) Heart rate during conditioning in humans: Effects of DCS intensity, vagal blockade, and adre-
References 227
nergic block of vasomotor activity. Journal of Experimental Psychology,

70,32-42.
OLSON, R. K., and ATTNEAVE, F. (1970) What variables produce similarity
grouping? The American Journal of Psychology, 83, 1-21.
OLTMAN, P. K. (1964) Field dependence and arousal. Perceptual and Motor
Skills, 19, 441.
OSGOOD, G. E. (1953) Method and theory in experimental psychology. New
York: Oxford University Press.
PAIVIO, A., and SIMPSON, H. M. (1966) The effect of word abstractness and
pleasantness on pupil size during an imagery task. Psychonomic Science,
5,55-56.
- - - , and SIMPSON, H. M. (1968) Magnitude and latency of the pupillary
response during an imagery task as a function of stimulus abstractness
and imag~ry ability. Psychonomic Science, 12, 45-46.
PARKINSON, S. R. (1972) Short-term memory while shadowing: Multiple-item
recall of visually and of aurally presented letters. Journal of Experimental
PAVLOV, I. P. (1927) Conditioned Reflexes, trans. G. V. Anrep. New York:
Dover Publications. (Replication of original, Oxford University Press, ed.,
1960.)
PAYNE, D. T., PARRY, M. E., and HARASYMIN, S. J. (1968) Percentage of pupillary dilation as a measure of item difficulty. Perception and Psychophysics, 4, 139--143.
PEAVLER, W. S. (19'69) Attention, processing load, and pupil size. Unpublished
Ph. D. thesis, Unive'rsity of Delaware.
PETERSON, L. R. (1969) Concurrent verbal activity. Psychological Review, 76,
376-386.
PILLSBURY, W. B. (1908) 'Attention. New York: Macmillan.
POLLACK, I., RUBENSTEIN, H., and DECKER, L. (1960) Analysis of incorrect
responses to an unknown message set. Journal of the Acoustical Society
of America, 32, 454-457.
PORGES, S. W. (1972) Heart rate variability and deceleration as indexes of reaction time. Journal of Experimental Psychology, 92, 103-110.
POSNER, M. I. (1964) Information reduction in the analysis of sequential tasks.
Psycholog,ical Review, 71, 49'1-504.
- - - . (1969) Abstraction and the process of recognition. In Advances in
learning, ed. G. Bower. New York: Academic Press.
- - - . (1970) On the relationship between letter names and super-ordinate
categories. Quarterly Journal of Experim.ental Psychology, 22, 279-287.
- - - , and BOIES, S. J. (1971) Components of attention. Psychological Review, 78, 391-408.
- - - , and KEELE, S. W. (1968) Attention demands of movements. Paper
presented to the Sixteenth International Congress of Applied Psychology,
228
Symposium on Work and Fatigue, August. Published in the Proceedings,

Amsterdam, 1969. Swets & Zeitlinger.
- - - , and KEELE, S. W. (1970) Time and space as measures of mental
operations. Invited address, Division 3, American Psychological <Association, September.
- - - , and KLEIN, R. M. (1972) On the functions of consciousness. In Attention and performance IV, ed. S. Kornblum. New York: Academic Press,
in press.
- - - , KLEIN, R. M., SUMMERS, J., and BUGGIE, S. (1973) On the selection
of signals. Memory and Cognition, 1, 2-12.
- - - , LEWIS, J., and CONRAD, C. (1972) Component processes in reading:
A performance analysis. In Language by ear and by eye, eds. J. Kavanaugh and 1. Mattingly. Boston: MIT Press.
- - - , and MITCHELL, R. F. (1967) Chronometric analysis of classification.
Psychological Review, 74, 392-409.
- - - , and ROSSMAN, E. (1965) Effect of size and location of informational
transforms upon short-term retention. Journal of Experimental Psychology, 70, 496-505.
- - - , and WARREN, R. (1972) Traces, concepts and conscious construction.
In Coding theory in learning and memory, eds. A. W. Melton and E.
Martin. New York: Winston, Pp. 25-43.
POULTON, E. C. (1952) The basis of perceptual anticipation in tracking.
British Journal of Psychology, 43, 29'5-302.
- - - . (19'53) Two-channel listening. Journal of Experimental Psychology,
46, 91-96.
PRATT, R. W. (1970) Cognitive processing of uncertainty: Its effects on pupillary dilation and preference ratings. Perception and Psychophysics, 8,
193-198.
PRITCHATT, D. (1968) An investigation into some of the underlying associative
verbal processes of the Stroop color effect. Quarterly Journal of Experimental Psychology, 20, 351-359'.
QUARTON, G. C., and TALLAND, G. A. (1962) The effects of methamphetamine
and pentobarbital on two measures of attention. Psychopharmacologica,
3,66-71.
RABBITT, P. M. A. (1964) Ignoring irrelevant information. British Journal of
- - - . (1967) Learning to ignore irrelevant information. American Journal of
- - - . (1969) Psychological refractory delay and response-stimulus interval
duration in serial, choice-response tasks. Acta Psychologica, 30, Attention
and performance II, ed. W. G. Koster, 195-219.
- - - . (1971) Times for the analysis of stimuli and for the selection of responses. British Medical Bulletin, 27, 259~265.
RAPAPORT, D. (1967) The theory of 'attention cathexis: An economic and
structural attempt at the explanation of cognitive processes. In The
References
229
collected papers of David Rapaport, ed. M. M. Gill. New York: Basic

Books. Pp. 778-794.
RASKIN, D. C., KOTSES, H., and BEVER, J. (1969a) Autonomic indicators of
orienting and defensive reflexes. Journal of Experimental Psychology,
80, 423-433.
- - - , KOTSES, H., and BEVER, J. (1969b) Cephalic vasomotor and heart
rate measures of orienting and defensive reflexes. Psychophysiology, 6,
149-159.
RAZRAN, G. (1961) The 'observable unconscious and the inferable conscious in
current Soviet psychophysiology: Introceptive conditioning, semantic
conditioning and the orienting reflex. Psychological Review, 68, 81-147.
REBER, A. S., and ANDERSON, J. R. (1970) The perception of clicks in linguistic
and nonlinguistic messages. Perception and Psychop'hysics, 8, 81-89.
REICHER, G. M. (1969) Perceptual recognition as a function of meaningfulness
of stimulus material. Journal of Experimental Psychology, ,81, 275-280.
REITMAN, J. S., (1971) Mechanisms of forgetting in short-term memory.
Cognitive Psychology, 2, 131-157.
REYNOLDS, D. (1964) Effects of double stimulation: Temporary inhibition of
response. Psychological Bulletin, 62, 333-347.
- - - . (1966) Time and event uncertainty in unisensory reaction time. Journal of Experimental Psychology, 71, 286-293.
REYNOLDS, G. S. (1961) Attention in the pigeon. Journal of the Experimental
Analysis of Behavior, 4, 203-208.
RIGGS, L. A., RATLIFF, F., CORNSWEET, J. C., and CORNSWEET, T. W. (1953)
The disappearance of steadily fixated visual test objects. Journal of the
Optical Society of America, 43, 495-501.
RUBINSTEIN, L., and RUTSCHMANN, J. (1967) Reduction of the "psychological
refractory period." Perception and Psychophysics, 2, 499-502.
RUMMELHART, D. E. (1970) A multicomponent theory of the perception of
briefly exposed visual displays. Journal of Mathematical Psychology, 7,
191-218.
SALAPAT'EK, P., and KESSEN, W. (1966) Visual scanning of triangles in the
human newborn. Journal of Experimental Child Psychology, 3, 155-167.
SANDERS, A. F. (196'3) The selective process in the functional visual field.
Monograph from Institute for Perception, National Defense Research
Organization TNO. Soesterberg, The Netherlands.
- - - . (1970) Some aspects of the selective process in the functional visual
field. Ergonomics, 13, 101-117.
- - - , and KEUSS, P. J'. G. (1969) Grouping and refractoriness in multiple
selective responses. Acta Psychologica, 30, 177-194.
- - - , and SCHROOTS, J. J. F. (19'68) Cognitive categories and memory span:
I. Shifting between categories. Quarterly Journal of Experimental Psychology, 20, 370-372.
230
SAVIN, H. B. (1967) On the successive perception of simultaneous stimuli.

SCHAEFER, T., JR., FERGUSON, J. B., KLEIN, J. A. and RAWSON, E. B. (1968)
Pupillary responses during mental activities. Psychonomic Science, 12,
137-138.
SCHLOSBERG, H. (1954) Three dimensions of emotion. Psychological Review,
61,81-88.
SCHNEIDER,G. E. (1967) Contrasting visuomotor functions of tectum and cortex in the golden hamster. Psychologische Forschung, 31, 52-62.
SCHOUTEN, J. F.,KALSBEEK, J. W. H., and LEOPOLD, F. F. (19'62) On the
evaluation of perceptual and. mental load. Ergonomics, 5, 251-260.
SCHVANEVELDT, R. W. (1969) Effects of complex;!ty in simultaneous reaction
time tasks. Journal of Experimental Psychology, 81, 289-296.
SCHWARTZ, F., and SCHILLER, P. H. (196'7) Rapaport's theory of attention
cathexis. Bulletin of the Menninger Clinic, 31, 3-17.
- - - , and SCHILLER, P. H. (1970) A psychoanalytic model of attention and
learning. Psychological Issues, 6 (3), Monograph 23. New York: International Universities Press.
SEIBEL, R. (1963) Discrimination reaction time for a 1,023-alternative task.
Journal of Experimental Psycholo'gy, 66, 215-226.
SENDERS, J. W. (1965) An investigation of the visual sampling. behavior of human observers. Bolt, Beranek, and Newman, Report No. 1246.
SHAFFER, L. H. (1968) Refractoriness in information processing. Quarterly
- - - . (1971) Attention in transcription skill. Quarterly Journal of Experimental Psychology, 23, 107-112.
- - - , and HARDWICK, J. (1969a) Reading and typing. Quarterly Journal of
- - - , and HARDWICK, J. (1969b) Monitoring simultaneous auditory messages. Perception and Psychophysics, 6, 401-404.
- - - , and HARDWICK, J. (1970) The basis of transcription skill. Journal of
SHEPARD, R. N. (1964) Attention and the metric structure of the stimulus space.
Journal of Mathematical Psychology, 1, 54-87.
SHEPP, B. E., KEMLER, D. G., and ANDERSON, D. R. (1972) Selective attention
and the breadth of learning: An extension of the one-look model. Psy.
chological Review, 79, 317-328.
SHIFFRIN, R. M., and GARDNER, G. T. (1972) Visual processing capacity and
attentional control. Journal of Experimental Psychology, 93, 72-82.
SHULMAN, H. G., and GREENBERG, S. N. (1971) Perceptual deficit due to division of attention between memory and perception. Journal of Experimental Psychology, 88,171-176.
- - - , GREENBURG, S. N., and MARTIN, J. (1971) Intertask delay as a param-
References
231
eter of perceptual deficit in divided attention. Journal of Experimental

SlMON, H. A., and BARENFELD, M. (1969) Information-processing analysis of
perceptual processes in problem solving. Psychological Review, 7-6, 473483.
SIMON, J. R. (1969) Reactions toward the source of stimulation. Journal of
- - - , CRAFT, J. L., and SMALL, A. M., JR. (1970) Manipulating the strength
of a stereotype: Interference effects' in an auditory information-processing
task. Journal of Experimental Psychology, 86, 63-68.
SIMPSON, H. M., and HALE, S. M. (1969) Pupillary changes during a decisionmaking task. Perceptual and Motor Skills, 29, 495-498.
- - - , MOLLOY, F. M., HALE, S. M., and CLIMAN, M. H. (1968) Latency and
magnitude of the pupillary response during an imagery task. Psychonomic Science, 13, 293-294.
- - - , and PAIVIO, A. (1966) Changes in pupil size during an imagery task
without motor response involvement. Psychonomic Science, 5, 405-406.
- - - , and PAIVIO, A. (1968) Effects on pupil size of manual and verbal indicators of cognitive task fulfillment. Perception and Psychophysics, 3,
185-196.
SMITH, E. E. (1968) Choice reaction time: An analysis of the major theoretical
positions. Psychological Bulletin-, 69, 77-110.
- - - , and HAVILAND, S. E. (1972) Why words are perceived more accurately
than nonwords: Inference versus unitization. Journal of Experimental
SMITH, F. (1969) The use of featural dependencies across letters in the visual
identification of words. Journal of Verbal Learning and Verbal Behavior,
8, 215-218.
SMITH, M. C. (1967a) The psychological refractory period as a function of performance of a first response. Quarterly Journal of Experimental Psychology, 19, 350-352.
- - - . (1967b) Theories of the psychological refractory period. Psychological
Bulletin, 67, 202-213.
- - - . (1967c) Reaction time to a second stimulus as a function of intensity
of the first stimulus. Quarterly Journal of Experimental Psychology, 19,
125-132.
- - - . (1969) The effect of varying information on the psychological refractory period. Acta Psychologica, 30, Attention and performance II,
ed. W. G. Koster, 220-231.
SNYDER, C. R. R. (1972) Selection, inspection, and naming in visual search.
SOKOLOV, E. N. (1963) Perception and the conditioned reflex, trans. S. W.
Waydenfold. Oxford: Pergamon Press.
232
- - - . (1965) The orienting reflex, its structure and mechanisms. In Orienting

reflex and exploratory behavior~ eds. L. G. Voronin, A. N. Leontiev,
A. R. Luria, E. N. Sokolov, and o. S. Vinogradova. Russian Monographs
on Brain and Behavior, 3. American Institute of Biological Sciences,
Washington, D. C.
------. (1969) The modeling properties of the nervous system. In A handbook
of contemporary Soviet psychology, eds. M. Cole and I. Maltzman. New
York: Basic Books.
SPERLING, G. (1960) The information available in brief visual presentations.
Psychological Monographs, 74, (Whole No. 498).
- - - . (1963) A model for'visual memory tasks. Human Factors, 5, 19-31.
- - - . (1967) Successive approximations to a model for short-term memory.
Acta Psychologica, 27, 285-292.
- - - , BUDIANSKY, J., SPIVAK, J. G., and JOHNSON, M. C. (1971) Extremely
rapid visual search: The maximum rate of scanning letters for the presence of a numeral. Science, 174, 307-311.
SPIETH, W., CURTIS, J. F., and WEBSTER, J. C. (1954) Responding to one of
two simultaneous messages. Journal of the Acoustical Society of America,
26,390-396.
SPOEHR, K. T., and SMITH, E. E. (1972) The role of syllables in perceptual
processing. In press.
STENNETT, R. G. (1957) The relationship of performance level to level of
arousal. Journal of Experimental Psychology, 54, 54-61.
STERNBERG, S., and KNOLL, R. L. (1972) The perception of temporal order:
Fundamental issues and a general model. In Attention and performance
IV, ed. S. Kornblum. New York: Academic Press. In press.
- - - , KNOLL, R. L., and GATES, B. A. (1971) Prior entry reexamined: Effect
of. attentional bias on order perception. Paper presented at the meeting
of the Psychonomic Society, November, St. Louis, Mo.
STRONGMAN, K. (1970) Communicating with the eyes. Science Journal, -6, 4752.
STROOP, J. R. (1935) Studies of interference in serial verbal reactions. Journal
SUCHMAN, R. G., and TRABASSO, T. (1966a) Color and form preference in
young children. Journal of Experimental Child Psychology, 3, 177-187.
- - - , and TRABASSO, T. (1966b) Stimulus preference and cue function in
young children's concept attainment. Journal of Experimental Child
SUTHERLAND, N. S. (1959) Stimulus analyzing mechanisms. In Proceedings of a
symposium on the mechanization of thought processes, II. London: Her
Majesty's Stationery Office. Pp. 575-609.
- - - . (1964) The learning of discrimination by animals. Endeavour, 23, 148152.
References
233
- - - , and HOLGATE, V. (1966) Two-cue discrimination learning in rats.

Journal of Comparative and Physiological Psychology, 4, 198-207.
- - - , and MACKINTOSH, J. (1964) Discrimination learning: Non-additivity of
cues. Nature, 2,01, 528-530..
MACKINTOSH, N. J., and MACKINTOSH, J. (1965) Shape and size discrimination in octopus: The effects of pretraining along different dimensions. Journal of Genetic Psychology, 106, 1-10.
- - - , and MACKINTOSH, N. J. (1971) Mechanisms of animal discrimination
learning. New York: Academic Press.
TANNER, W. P., JR., and SWETS, J. A. (1954) A decision-making theory of
visual detection. Psychological Review, 61, 401-409.
TAYLOR, M. M., LINDSAY, P. M., and FORBES, S. M. (1967) Quantification of
shared capacity processing in auditory and visual discrimination. Acta
---~
TECCE, J. J. (1972) Contingent negative variation (CNV) and psychological

processes in man. Psychological Bulletin, 77, 73-108.
TELFORD, C.W. (1931) The refractory phase of- voluntary and associative
responses. Journal of Experimental Psychology, 14, 1-35.
THACKRAY, R. I. (1969) Patterns of physiological activity accompanying performance in a perceptual motor task. FAA Office of Aviation Medicine
Report. No. 69-8.
THOMAS, E. L. (1963) Eye movements and fixations during initial viewing of
Rorschach cards. Journal of Projective Techniques and Personality Assessment, 27, 345-353.
- - - , and LANSDOWN, E. L. (1963) Visual search patterns of radiologists
in training. Radiology, 81, 28&-29'2.
TIKHOMI;Rov, O. K., and POZNYANSKAYA, E. D. (1966) An investigation of
visual search as a means of analyzing heuristics. Soviet Psychology, 5
(2), 3-15. (Trans. from Voprosy Psikhologii, 12, 39-53.)
TINKER, M. A. (1947) Time relations for eye-movement measures in reality.
Journal of Educational Psychology, 38, 1-10.
- - - . (1958) Recent studies of eye movements in reading. Psychological
Bulletin, 55, 215-231.
TITCHENER, E. B. (1908) Lectures on the elementary psychology of feeling
and attention. New York: Macmillan.
TRABASSO, T. R. (1963) Stimulus emphasis and all-or-none learning in concept
identification. Journal of Experimental Psychology, -65, 398-406.
- - - , and BOWER, G. H. (1968) Attention in learning: Theory and research.
New York: John Wiley.
TR~!SMAN,
A. M. (1960) Contextual cues in selective listening. Quarterly

Journal of ExperimentalPsycholog.y, 12, 242-248.
- - - . (19'64a) Verbal cues, language and meaning in selective attention.

234
- - - . (1964b) Effect of irrelevant material on the efficiency of selective

listening. American Journal of Psychology, 77, 533-546.
- - - . (1964c) Monitoring and storage of irrelevant messages in selective
attention. Journal of V erbal Learning and Verbal Behavior, 3, 449~59.
- - - . (1964d) Selective attention in man. British Medical Bulletin, 20, 12-16.
- - - . (1969) Strategies and models of selective attention. Psychological
Review, 76, 282-299.
- - - . (1970) Perception and recall of simultaneous speech stimuli. Acta
Psychologica,33, 132-148.
- - - . (1971) Shifting attention between the ears. Quarterly Journal of Expe'rimental Psychology, 23, 157-167.
- - - , and DAVIES, A. (1972) Divided attention to ear and eye. In press.
- - - , and FEARNLEY, J. S. (1969) The Stroop test: Selective attention to
colors and words. Nature, 222, 437~39.
- - - , and FEARNLEY, J. S. (1971) Can simultaneous speech stimuli be classified in parallel? Perception and Psychophysics, 10, 1-7.
- - - , and GEFFEN, G. (1967) Selective attention and cerebral dominance in
perceiving and responding to' speech messages. Quarterly Journal of Experimental Psychology, 19, 1-17.
- - - , and GEFFEN, G. (1968) Selective attention and cerebral dominance
in perceiving and responding to speech messages. Quarterly Journal of
Experimental Psychology, 20, 139~151.
- - - , and RILEY, J. G. A. (1969) Is selective attention selective perception or
selective response? A furt-her test. Journal of Experimental Psychology,
79,27-34.
TREISMAN, M. (1971) On the word frequency effect: Comments on the
papers by J. Catlin and L. H. Nakatani. Psychological Review, 73, 420425.
- - - . (19'72) Detection of binaural tone stimuli: time sharing or criterion
change? Journal of the Acoustical Society of America, 51, 625-630.
TREVARTHEN, C. B. (1968) Two mechanisms of vision in primates. Psychologische
Forschung, 31, 299-337.
TRIGGS, T. J. (1968) Capacity sharing and speeded reactions to successive signals. Technical Report No. 9-., August, University of Michigan, Contract
No. AF 49(638)-1736, United States Department of Defense.
TRUMBO, D., and NOBLE, M. (1970) Secondary task effects on serial verbal
learning. Journal of Experimental Psychology, 85, 418~24.
- - - , NOBLE, M., and SWINK, J. (1967) Secondary task interference in the
performance of tracking tasks. Journal of Experimental Psychology, 73,
232-240.
TULVING, E., and GOLD, C. (1963) Stimulus information and contextual information as determinants of tachistoscopic recognition of words. Journal
References
235
- - - , and LINDSAY, P. H. (1967) Identification of simultaneously presented

simple visual and auditory stimuli. Acta Psychologica,-27, 101-109.
TURSKY, B., SCHWARTZ, G. E., and CRIDER, A. (1970) Differential patterns of
heart rate and skin resistance during a digit-transformation task. Journal
ULEMAN, J. S., and REEVES, J. (1971) A reversal of the Stroop interference

effect through scanning. Perception and Psychophysics, 9 (3A), 293295.
UNDERWOOD, B. J., RUNQUIST, W. N., and SCHULZ, R. W. (1959) Response
learning in paired-associate lists as a function of intralist similarity.
- - - , and SCHULZ, R. W. (1960) Meaningfulness and v-erballearning. Philadelphia: Lippincott.
UNDERWOOD, G., and MORAY, N. (1972) Shadowing and monitoring for selective attention. Quarterly Journal of Experimental Psychology, 23, 284296.
UNGER, S. M. (1964) Habituation of the vasoconstrictive orienting reaction.
UNO, T., and GRINGS, W. W. (1965) Autonomic components of orienting behavior. Psychophysiology, 1, 311-321.
VINCE, M. (1948) The intermittency of control movements and the psychological refractory period. British Journal of Psychology, 38, 149-157.
VON WRIGHT, J. M. (1968) Selection in visual immediate- memory. Quarterly
- - - . (1970) On selection in visual immediate memory. Acta Psychologica,
33, 280-292.
WACHTEL, P. L. (1967) Conceptions of broad and narrow attention. Psychological Bulletin, 68, 417-429.
WALKER, E. L., and TARTE, R. D. (1963) Memory storage as a function of time
with homogeneous and heterogeneous lists. Journal of Verbal Learning
and Verbal Behavior, 2, 113-119.
WALTER, W. G., COOPER, R., ALDRIDGE, V. J., MCCALLUM, W. C., and WINTER, A. L. (1964) Contingent negative variation: An electric signal sensorimotor association and expectancy in the human brain. Nature, 203,
380-384.
WEBB, R. A., and OBRIST, P. A. (1970) The physiological concomitants of
reaction time performance as a function of preparatory interval series.
Psychophysiology, 6, 389-403.
WEBSTER, J. C., and SOLOMON, L. N. (1955) Effects of response complexity
upon listening to competing messages. Journal of the Acoustical Society
of America, 27, 1194-1198.
- - - , and THOMPSON, P. O. (1954) Responding to both of two overlapping
messages. Journal of the Acoustical Society of America, 26, 396-40-2.
WEG E. (1971) Unpublished study, Hebrew University, Jerusalem. (Hebrew)
236
WEITZMAN, B. (1963) A threshold difference produced by a figure-ground

dichotomy. Journal of Experimental Psychology, ,66, 201-205.
WELFORD, A. T. (1952) The "psychological refractory period" and the timing
of high-speed performance: A review and a theory. British Journal of
_ _---.I. (1959) Evidence of a single-channel decision mechanism limiting
performance in a serial reaction task. Quarterly Journal of Experimental

- - - . (1967) Single-channel operation in the brain. Acta Psychologica, 27,
5-22.
- - - . (1968) Fundamentals of skill. London: Methuen.
WELL, A. (1971) The inHuence of irrelevant information. on speeded classification tasks. Perception and Psychop,hysics, 10, 79--84.
WHEELER, D. D. (1970) Processes in word recognition. Cognitive Psychology,
1,59~85.
.
WILCOCKS, R. W. (1925) An examination of Kiilpe's experiments on abstraction.
WILKINSON, R. T. (1961) Interaction of lack of sleep with knowledge of results, repeated testing and individual differences. Journal of Experimental
- - - . (1962) 'Muscle tension during mental work under sleep deprivation.
- - - . (1963) Interaction of noise with knowledge of results and sleep
deprivation. Journal of Experimental Psychology, ,66, 332-337.
- - - . (1965) Sleep deprivation. In The physiology of human survival, eds.
o. G. Edholm and A. L. Bacharach. London and New York: Academic
Press. Pp. 399-430.
WILLIAMS, L. G. (1966) The effect of target specification on objects fixated
dtlring visual search. Perception and Psychophysics, 1, 315-318.
WILSON, R. S. (1964) Autonomic changes produced by noxious and innocuous
stimulation. Journal of Comparative and Physiological Psychology, 58,
290-295.
WOLFORD, G., and BOWER, G. H. (1969) Continuity theory revisited: Rejected
for the wrong reasons? Psychological Review, 76, 515-518.
WOODWORTH, R. S. (1938) Experimental psychology. New York: Holt.
- - - . (1958) Dynamics of behavior. New York: Holt.
WRIGHT, P., and KAHNEMAN, D. (1971) Evidence for alternative strategies of
sentence retention. Quarterly Journal of Experimental Psychology, 23,
197-213.
YARBUS, A. L. (1967) Eye movements and vision, trans. B. Haigh. New York:
Plenum Press.
YERKES, R. M., and DODSON, J. D. (1908) The relation of strength of stimulus
to rapidity of habit-formation. Journal of Comparative Neurology of
References
237
YNTEMA, D. C., and TRASK, F. P. (1963) Recall as a search process. Journal

of Verbal Learning and Verbal Behavior, 2, 65-74.
ZEAMAN, D. and HousE, B. J. (1963) The role of attention in retardate discrimination learning. In Handbook of mental deficien,cy: Psychological
theory and research. New York: McGraw-Hill. Pp. 159-223.
ZELNIKER, T. (1971) Perceptual attenuation of an irrelevant auditory verbal
input as measured by an involuntary verbal response in a selectiveattention task. Journal of Experimental Psychology, 87, 52-56.
ZIMNY, G. H., and KIENSTRA, R. A. (1967) Orienting and defensive responses
to electric shock. Psychophysiology, 3, 351-362.
- - - , and MILLER, F. L. (1966) Orienting and adaptive cardiovascular
responses to heat and cold. Psychophysiology, 3, 81-92.
- - - , PAWLICK, G. F., and SAUR, D. P. (1969) Effects of stimulus order and
novelty on orienting responses. Psychophysiology, 6, 166-173.
Name Index
Adams, J. A., 176, 192
Agnew, M., 39
Agnew, N., 39
Aldridge, V. J., 32
Allport, D. A., 153, 154, 198, 199
Alpern, M., 50
Amadeo, M., 64
Ambler, B., 74, 127
Anderson, D. R., 99
Anderson, J. R., 137
Antonis, B., 198
Antrobus, J. S., 64
Archer, E. J., 101, 109
Argyle, M.,63
Atkinson, R. C., 122
Attneave, F., 5, 74
Avetbach, E., 83, 103
Avner, D., 143, 150, 151, 162
Axelrod, S., 154
Baddeley, A. D., 42, 179, 182
Badia, P., 43
Bahrick, fI. P., 37
Bakan, P., 35, 54, 64
Bankart, B., 32
Barenfeld, M., 62
Barnett, T., 138
Bartz, D. H., 139
Bashford, M. B., 54
Beatty, J., 16-17, 19-20, 25-26, 46, 85,
94, 125, 185
Beck, J., 74-75, 127-29
Belton, J. A., 35
Ben-Ishai, R., 119
Berkson, G., 52
Berlyne, D. E., 3, 41, 43, 52-54, 137
Bernstein, A. S., 46
Bernstein, I. H., 175
Bersh, P. J., 31
Bertelson, P., 168, 175, 176
Bever, J., 43
Bever, T. G., 137
Biederman, I., 106, 109
Biryukov, D. A., 45
Blake, M. J. F., 42
Blake, R. R., 186
Blecker, D. L., 110, 193
Blinder, B., 59
Boggs, D. H., 35
Boies, S. J., 16, 33, 41, 187, 188, 194,
195
Borger, R., 176
Boring, E. G., 137
Borsa, D. M., 41, 43
Bower, G. H., 98-99, 101.
Bradshaw, J. L., 19, 26
Brand, J., 87
Briggs, G. E., 183
Broadbent, D.E., 6, 8, 12, 34-36, 38-39,
41, 70-71, 76, 84, 87-88, 91,
93-94, 96, 103, 112, 114, 116-18,
120-23, 125, 130, 135-39, 143,
146, 168, 170, 171, 175, 177, 179,
238
Broadbent (cont.)
182, 193, 195, 196
Broadhurst, P. L., 39
Brooks, L. R., 196, 197
Brown, C. R., 94
Brown, I. D., 182, 196
Brown, J., 118
Brown, L. T., 53-54
Bruner, J. S., 56, 58, 88, 93
Bryden, M. P., 41, 61, 64, 138-39, 140
Buggie, S., 41, 188, 194, 195
Bursill, A. E., 37, 38
Callaway, E., 38-40
Campbell, H., 29
Campos, J. J., 33
Carey, S. T., 42
Carver, R. P., 25
Catlin, J., 94
Cavanagh, J. P., 83
Chase, W. G., 31, 48, 75, 83
Checkosky, S. F., 106
Cherry, E. C., 112-13, 117
Clark, M. H., 175
Climan, M. H., 20
Cohen, J., 32
Cohen, M. J., 32
Colavita, F. B., 141, 151, 157, 162
Collins, J. F., 80, 83, 193
Colman, F. D., 18-20
Connor, W. H., 31, 32, 48
Conrad, C., 86, 161
C'oombs, C. H., 88
Cooper, F. S., 125
C'ooper, R., 32
Coquery, J. M., 31
Corballis, M. C., 140
Corcoran, D. W. J., 35
Coriell, A. S., 83
Comsweet, D. J., 38
Comsweet, J. C., 58
Cornsweet, T. W., 58
Corteen, R. S., 41, 118, 130
Craft, J. L., 48
Craik, K. J. W., 162, 163
Craw, M. A., 41
Creamer, L. R., 176
Crider, A., 18, 31
Crovitz, H. F., 61
Crowder, R. G., 121, 132-33
Cuddy, L. L., 146
Curtis, J. F., 114, 118, 138
Dallett, K. M., 132
Darrow, C. W., 64
Daves, W., 61
Davies, A., 142, 153, 154, 196, 199
Davies, D. R., 35
Davis, R., 29, 175
Davis, R. C., 163, 173
Dawes, R. M., 88
Day, H., 53, 54
Day, M. E., 63-64
Dean, J., 63
Deane, G. E., 31
Decker, L., 94
Defran, R. H., 43
De Groot, A. D., 56, 62
Deutsch, D., 6, 123-25, 135, 136, 143,
144, 145, 148, 155, 169
Deutsch, J. A., 6, 123-25, 135, 136, 143,
144, 145, 148, 155, 169
Diamond, I. T., 154
Dill, A. B., 59
Dillon, R. F., 179
Dimond, S., 162
Drynan, R., 179
Duchknowsky, A., 22
Duffy, E., 34
Dulewicz, V., 81
Durham, R. L., 54
Eason, R. G., 181
Easterbrook, J. A., 16, 28, 37, 38, 40,
42, 49
Edelstein, B. A., 175
Egeth, H., 103, 105, 108-10, 193
Ehrlich, S., 59
Eisler, H., 108
Elithom, A., 175, 176
Elliott, R., 32-33
Ellis, S. H., 75
Ells, J., 187
Epstein, W., 45
Eriksen,C. W., 78-80, 83-84, 193
Ettema, J. H., 17
Exline, R. V., 63
Eysenck, H. J., 35, 180
Fantz, R. L., 52
Farha, W., 54
Faw, T. T., 54
Feamley, J. S., 123, 145, 148
Fehrer, E., 70
Felfoldy, G. L., 107
Fellows, B. J., 98, 102
Fennell, E., 139
Ferguson, J. B., 25
Fisher, R. P., 179, 183
Fitts, P. M., 5, 37, 109
Fodor, J. A., 137
Forbes, S. M., 149, 200
Forrin, B., 109
Fox, L. A., 110
Fox, R., 186
Fraisse, P., 59, 175
Freud, S., 128
Furedy, J. J., 45
Furst, C. J., 59
Galanter, E., 192
Gardner, G. T., 16
Gamer, W. R., 5, 101, 106, 107, 108, 109
Garvey, W. D., 181
Gates, B. A., 137, 193
Geffen, G., 88, 94, 96, 141, 143
Gelman, R. S., 41
Germana, J., 43
Gesell, A., 52
Gibson, E. J., 86, 87, 102
Gibson, J. J., 63
Gold, C., 122
Goldstein, E., 133
Goldwater, B. C., 19
Gomez, E., 64
Goodwin, W. R., 101
Gopher, D., 62, 63, 119, 151, 154
Gordon, I. E., 81
Gottsdanker, R., 175
Gottwald, R. L., 106
Gould, J. D., 57, 58, 59, 60, 191
Graham, D. T., 31, 48
Graham, F. K., 31, 70
Gray, J. A., 139
Green, D. M., 88
Greenberg, S. N., 16, 85, 130, 179, 183,
187
Greenwald, A. G., 118, 198
Gregory, L. P., 53
Gregory, M., 85, 91, 94, 96, 122, 125,
138, 139, 168, 170, 171, 175, 177
Grindley, G. C., 59
Grings, W. W., 43
Guzy, L. T., 154
Haber, R. N., 103, 104
Hakerem, G., 46
Hale, S. M., 19, 184
Haltrecht, E. J., 61
Hamacher, J. H., 41
Hannah, T. E., 61
Harasymin, S. J., 19
Hardwick, J., 142, 148, 150, 192
Harris, C. S., 103, 104
Harris, L., 45
Hastings, S. E., 31
Hatter, J. E., 42
Haviland, S.E., 86
Hawkins, H. L., 106
Heider, F., 76
Held, R., 127
JIelmholtz, H. von, 59
Henik, A., 115
Herman, L. M., 174, 175, 176
Hernandez-Peon, R., 40, 78
Hershenson, M., 52
Hess, E. H., 19
Hillyard, S. A., 32
Hintzman, D. L., 110
Hitchcock, C. H., 140
Hochberg, J. E., 125, 127, 128, 131
Hockey, G. R. J., 34, 35, 38
Hoenig, Y. J., 133
Hoffman, J. E., 79, 84
Holgate, V., 101
Holloway, C. M., 132
House, B. J., 99, 100
Houston, B. K., 35, 39
Howard, J. L., 31, 32
Howes, D. H., 93
Hyman, R., 108
Ilg, F. L., 52
Imai, S., 101, 109
Ingle, D., 127
Ingram, E., 45
Ittleson, W. H., 92
James, W., 137
Jenks, R. S., 31
Jensen, A. R., 109
Johnson, D. A., 20
Johnson, H. J., 32, 33, 50, 65
239
Johnson, L. C., 43
Johnston, W. A., 179, 183
Jones, L. V., 101
Jones, T. M., 39
Jongman, R. W., 56, 62
Jordan, A., 138
Kafry, D., 172, 173, 175, 176, 177
Kagan, J., 29, 102
Kahneman, D., 15, 16, 17, 18, 19,20,22,
24, 25, 26, 46, 60, 61, 70, 85, 94,
114, 115, 11~ 125, 14~ 14~ 154,
180, 184, 185, 200
Kalafat, J., 29
Kalsbeek, J. W. H., 17, 181
Kamlet, A. S., 110
Kantowitz, B. H., 174, 175
Kaplan, I. T., 62
Kaplan, S., 41
Karlin, L., 169, 175, 177
Kaufman, L., 51, 56, 60
Kay, H., 175
Keefe, F. B., 43
Keele, S. W., 8, 9, 12, 16, 27, 106, 110,
143, 158, 177, 180, 186, 189
Keeley, S. M., 84
Kendon, A., 63
Kernler, D. G., 99, 100
Kessen, W., 52, 56
Kestenbaum, R., 169, 175, 177
Keuss, P. J. G., 162, 168, 175
Kienstra, R. A., 43
Kilpatrick, R., 92, 93
Kinsbourne, M., 63
Klein, G. S., 110
Klein, J. A., 25
Klein, R., 41
Klein, R. M., 187, 188, 194, 195
Kleinsmith, L. J., 41
Kleitman, N., 42
Knoll,R. L., 137, 193
Knoppel, J. K., 108
Knott, P. D., 22
Koenig, I. D., 41
Koffka, K., 119
Kohn, B., 41, 64
Korchin, S. J., 41
Kornblum, S., 12
Kotses, H., 43
Krechevsky, I. A., 99, 102
Kristofferson, A. B., 137
Kroll,- N. E. A., 154
Kiilpe, 0., 103, 105, III
La Berge, D. L., 103, 104
Lacey, B. C., 19, 29, 31, 32
Lacey, J. I., 19, 29, 31, 32
Ladefoged, P., 137
Landsown, E. L., 59
Lang, P. J., 31, 32, 48
Lappin, J. S., 84, 104, 105, 123
Lashley, K. S., 98, 99, 102, 191
Lass, N., 60, 61
Lawrence, C., 176
Lawrence, D. H., 83, 99, 101, 103
Lawrence, G. H., 53, 54
Lawson, E. A., 114, 123, 144
Lazar, R., 80
. Leckart, B. T., 54
240
Lemkin, J., 102

Lemond, L. C., 54
Leopold, F. F., 181)
Levine, M., 101
Levy, R., 115, 147
Lewis, J., 86
Lewis, J. L., 118, 130, 161
Lewis, M., 29
Libby, W. L., 19, 29, 31
Liberman, A. M., 125
Light, R., 32
Lindsay, P. H., 137, 146, 148, 149, 200
Lockhead, G. R., 107
Lorens, S. A., Jr., 64
Lotan, M., 119
Lovejoy, E., 98, 99, 101
Lowe, D., 198
Luborsky, L., 54
Luria, A. R., 78
Luthe, L., 140
Lynn, R., 42, 126
r
McCallum, W. C., 32
McCauley, M. E., 175
McCormack, P. D., 61
McDonnell, P., 53
Mackintosh, N. J., 98, 99, 100, 101
Mackworth, N. H., 56, 58, 84
Madsen, M. C., 138, 139
Malmo, R. B., 34, 36, 196
Maltzman, I., 43, 45
Mandell, E. E., 41
Marcel, A. J., 107
Martin, D. W., 179, 180, 183, 187
Massaro, D. W., 122, 140
Merikle, P. M., 198
Michon, J. A., 181, 182
Michotte, A., 76
Miller, F. L., 43
Miller, G. A., 192
Mitchell, R. F., 86
Molloy, F. M., 20
Montague, W. E., 109
Moore, T. E., 61
Morandi, A. J., 56
Moray, N., 8, 12, 91, 94, 114, 115, 117,
122, 125, 137, 138, 141, 142, 143,
148, 149, 150, 151, 162, 200
Morin, R. E., 109
Morton, J., 8, 58, 85, 93, 110, 122, 132,
133
Moss, H. A., 29
Mowbray, G. H., 138, 141, 153
Munsinger, H., 52
Murdock, B. B., 25, 179
Murray, D. J., 140
Naylor, J. C., 183
Neisser, D., 7, 57, 71, 75, 80, 81, 117,
122, 125, 126, 127, 128, 129, 132,
133, 135, 137, 140, 143, 193
Nickerson, R. S., 168, 175, 176
Ninio, A., 115, 146, 148, 186
Noble, M., 183, 184, 185
Norman, D. A., 17,88,96,122, 124, 125,
136, 137, 143, 145, 148, 155, 169
Notterman, J. M., 31
Novick, R., 80
Nunnally, J. C., 22, 54
O'Brien, T., 91, 94, 114, 115, 125, 141,

142, 148, 149, 151, 162
Obrist, P. A., 31, 32
Olson, R. K., 74
Oltman, P. K., 39
Onuska, L., 15, 19, 24
Osgood, G. E., 2, 75
Pachella, R., 108
Paivio, A., 18, 19, 20, 25
Parker, R., 22
Parkinson, S. R., 154
Pavlov, I. P., 42, 46
Pawlick, G. F., 45
Payne, D. T., 19
Peavler, W. S., 15, 19, 24, 36, 38, 184
Peeples, D. R., 59
Perez-Reyes, M., 31
Perry, M. E., 19
Peterson, L. R., 179, 192, 198
Piaget, J., 40
Pick, A. D., 63
Pillsbury, W. B., 12
Pollack, I., 16, 85, 94, 125, 185
Polt, J. M., 19
Porges, S. W., 29
Posner, M. I., 5, 9, 16, 33, 41, 86, 87,
106, 131, 161, 170, 179, 186, 187,
188, 189, 194, 195
Poulton, E. C., 114, 138, 192
Powazek, M., 154
Poznyanskaya, E. D., 62
Pratt, R. W., 19
Pribra~, K., 192
Pritchatt, D., 110
Prussin, H. A., 132
Quarton, G. C., 39
Raab, D., 70
Rabbitt, P. M., 81, 87, 106, 107, 109,
172, 175
Rankin, R. E., 37
Rapaport, D., 128
Raskin, D. C., 43, 44
Ratliff, F., 58
Rawson, E. B., 25
Razran, G., 45
Reber, A. S., 137
Reed, W. G., 183
Reeves, J., 72
Reicher, G. M., 85
Reid, L. S., 179
Reitman, J., 179
Reynolds, D., 174
Reynolds, G. S., 101
Reynolds, P., 198
Richards, W., 51, 56, 60
Riggs, L. A., 58
Riley, J. G. A., 114,' 123, 144, 148
Rock, I., 45
Rohrbaugh, J. W., 78, 83
Rohwer, W. D., 109
Rollins, H. A., 138, 139
Rossman, E., 179
Rubenstein, H., 94
Rubinstein, L., 175
Rummelhart, D. E., 84, 137
Runquist, W. N., 61
Rutschmann, J., 175
Salapatek, P., 56
Sanders, A. F., 57, 60, 139, 162, 168, 175
Satz, P., 139
Saur, D. P., 45
Savin, H. B., 139
Schaefer, T., 25
Schaffer, A., 57, 58, 191
Schiller, P. H., 128
Schimek, J., 54
Schlosberg, H., 34
Schneider, G. E., 127
Schoenfeld, W. N., 31, 62
Schouten, J. F., 181
Schroots, J. J. F., 139
Schultz, R. W., 61
Schvaneveldt, R. W., 157-61, 162, 177
Schwartz, F., 31, 128
Scott, D., 42, 179
Scull, J., 45
Seibel, R., 162
Senders, J. W., 58, 60
Senf, G. M., 138, 139
Shaffer, L. H., 142, 148, 150, 176, 192
Shankweiler, D. P., 125
Shapiro, D., 18
Shepard, R. N., 101, 108
Shepp, B. E., 99
Shiffrin, R. M., 16, 122
Shor, R. E., 110
Shotland, R. L., 64
Shulman, H. G., 16, 85, 130, 180, 187
Simmel, M., 76
Simmons, D. C., 182
Simon, H. A., 62
Simon, J. R., 35, 48
Simpson" H. M., 19, 20, 25, 184
Singer, J. L., 64
Small, A. M., Jr., 48
Smith, E. E., 86, 106, 183
Smith, F., 85
Smith, J. C., 179
Smith, M. C., 166, 169, 175, 176, 177,
195
Snashall, A., 42
Snyder, C. R. R., 84
Sokolov, E. N., 42, 43, 44, 46, 81, 126
Solomon, L. N., 138, 196
Solomon, R. L., 93
Spaulding, S. J., 29
Sperling, G., 84, 103, 114, 120, 121
Spieth, W., 114, 118, 138
Spoehr, K. T., 86
Steinman, R. J., 110
Stennett, R. G., 34
Sternberg, S., 137, 138, 193
Stone, G., 38, 39, 40
Stoper, A., 57
Strongman, K., 63
Stroop, J. R., 32, 33, 39, 40, 64, 98, 109,
110
Studdert-Kennedy, M., 125
Suchman, R. G., 102
Summers, J., 41, 188, 194, 195
Surwillo, W., 36
Sutherland, N. S., 98, 99, 100, 101, 102
Sutterer, J. R., 31, 32
Sutton, S., 46
Swets, J. A., 88
Swink, J., 183
241
Talland, G. A., 39
Tanner, W. P., Jr., 88
Tarte, R. D., 41
Taylor, A., 35
Taylor, F. V., 181
Taylor, M. M., 149, 200
Taylor, W.K., 113
Tecce, J. J., 32
Telford, C. W., 163
Thackray, R. I., 17
Thomas, E. L., 59
Thompson, P.O., 141
Thompson, S. V., 38
Tickner, A. H., 182
Tikhomirov, O. K., 62
Tinker, M. A., 58
Tisseyre, F., 175
Titchener, E. B., 1, 78
Toth, J. C., 35
Townsend, V., 59
Trabasso, T. R., 98, 99, 101
Trask, F. P., 139
Treisman, A. M., 3, 8, 70, 75, 76, 88,96,
102, 105, 114, 116, 117, 118, 121,
122, 123, 125, 127; 128, 131, 135,
137, 139, 141, 142, 143, 144, 145,
148, 152, 153, 154, 196, 199
Treisman, M., 94
Trevarthen, C. B., 127
Triggs, T. J., 172, 175, 195
Trumbo, D,., 183, 184, 185
Tulving, E., 122, 146
Tursky, B., 18, 31
Tversky, A., 88
Uleman, J. S., 72
Underwood, B. J., 61, 115
Unger, S. M., 43, 44
Uno, T., 43
Van Sant, C., 175
Vince, M., 162

Von Wright, J. M., 71, 74, 114, 129
Vurpillot, E., 59
Wachtel, P. L., 40
Wade, M., 98, 99
Walker, E. L., 41
Walter, W. G., 30
Warren, R;~ 131, 188
Webb, R. A., 31, 32
Webster, J. C., 114, 118, 138, 141, 196
Wedderburn, A. A. I., 139
Weg, E., 144, 152
Weiss, A. D., 175
Weitzman, B., 78
Welford, A.T., 163, 173
Well, A., 108
Wheeler, D. D., 85
Wilcocks, R. W., 103
Wilkinson, R. T., 35, 36
Williams, L. G., 57, 60, 72, 74, 129
Wilson, R. S., 31
Winokur, S., 123
Winter, A. L., 32
Winwood, M., 81
Wolff, C., 45
Wolford, G., 101
Wolman, R. E., 19
Wood, B., 118, 130
Wood, D. M., 31
Woodworth, R. S., 12, 54, 113
Wright, P.,' 19, 24
Yarbus, A. L., 54, 56, 58, 60
Yerkes-Dodson, 28, 33, 34, 35,37, 40, 49
Yntema, D. C., 139
Zeaman, D., 99, 100
Zelnicker, T., 116, 130
Zimny, G. H., 43, 45
Subject Index
Alertness, 33, 194-95
Allocation policy, 10-11, 201
and demand1, demand2, 190
in divided attention, 149-52
effects of arousal, 37-42
and perceptual emphasis, 69, 84
and preparatory set, 195
and the PRP, 172
in spontaneous looking, 54
Analysis-by-synthesis \
and focal attention, 125-27
and the prefix effect, 133
Analyzers
and discrimination learning, 99, 102,
III
and parallel processing, 105, 123, 137,
152-53
Anxiety, 23, 24
Arousal
and the allocation policy, 37-42
242
Arousal (cont.)
and capacity, 10
and directional fraetionation, 29-33
and effort, 17-24
and intensive attention, 3-5
in the OR, 43, 48
and performance, 33-42
preparatory, 194-95
Attensity, 78
Attention
to attributes, 98-111
bottleneck models, 5-7
capacity theory, 7-11, 13-17
characteristics of, 201
divided (see Divided attention)
and figural emphasis, 79
focal, 126-27
focused (see Focused attention)
guidance of, 71-72, 83-84, 114
intensive aspects, 3-5
Attention (cont.)
lability, 40-42
narrowing of, 37-42
selective, 2-3
taxonomy of, 70-72
spontaneous, 52-54
theories of, 6-7
unitary versus divisible, 5, 136
voluntary versus involuntary, 3-4, 42
Attenuation (see Filter attenuation theory)
Attributes
report of, 103-5
selection of, 70-71
Autonomic changes
directional fractionation, 29-33
in foreperiod of RT, 31-32
in looking, 19, 29-31
in mental effort, 16-20
in OR, 42-43
in resisting distraction, 114
in Stroop test, 32
Awareness (see Conscious perception)
Beta (see Criterion)
Bi-sensory stimuli (see Modality effects)
Bottleneck models, 5-7, 182
Capacity (see Attention, Effort, Spare
capacity)
and arousal, 10
and demands, 13-17
model, 7-11
Card sorting, 105-8, 110
Cardiac deceleration, 29-33, 48
Categorizing, 103
Category state, 8, 87
Channel
capacity, 5
in filter theory, 120, 121
and grouping processes, 132-35, 138-39
City-block model, 108
Classification task, 105-9, III
Cocktail-party problem, 7, 112, 129
Coding, 102
dimension, 104
object, 104
Collative properties (see also Enduring
dispositions), 3, 52-55, 60, 64, 7879, 81
Compatibility, S-R, 170-72, 180
Complexity, of visual scene, 52-53
Condensation, 106
Conscious perception, 70, 118, 126-28,
130-31
Conservation, 40
Contingent negative variation (CNV), 32
Criterion (Beta)
in filter-attenuation theory, 122
in Norman's theory, 125
and perceptual readiness, /91-95
and response readiness, 94-96
in signal-detection theory, 89-91
Cue dominance, 101-2
d' (see Sensitivity)
Defensive reaction, 43
Delayed auditory feedback (DAF), 116
Demand t , Demand2 (see Effort)
Deutsch-Norman theory
and divided attention, 136-37
and focused attention, 123-25
and the suffix effect, 134
tests of, 143-45
Dichotic stimulation, 75, 113, 115-19,
122, 125, 138-52
Dictionary unit, 8, 122-23
Digit-transformation, 20, 23, 186
Directional fractionation
and looking, 29-31
and response inhibition, 32-33
and waiting, 31-32
Discontinuity theory, 99-100
Discrimination
and arousal, 34,. 38
learning, 98-103
learning and perception, 102-3
tasks, and effort, 19, 20, 26, 148-49
Dispositions, enduring (see also Collative properties), 42, 52, 60, 78
Distorted room, 91-93
Diurnal rhythm, 42
Divided attention (see also Processing,
parallel or serial)
and the Deutsch-Norman theory, 124,
136-37, 143-45
and effort, 148-52
eye-movements in, 63, 154
and filter theory, 136, 138, 143, 147,
148
modality effects in, 152-54
in monitoring, 141-42, 144-46, 14853, 155
and Neisser's theory, 137
and the PRP, 170, 172, 174
in psychophysical judgments, 146--49
in the recognition task, 144-45, 14752, 190
in the split-span design, 137-41
Treisman's theory of, 137
Drug effects, 32, 38, 39, 41, 64
Dual-tasks (see also Divided attention,
Interference, Subsidiary tasks)
and arousal, 37-38
method, 181-82
Easterbrook's hypothesis, 16, 37-42, 149
EEG, 32, 42-43, 64
Effort (see also Attention, Capacity)
and attention, 4, 9
and consciousness, 131
demands
demand t , demand2, 189-91, 202
and perceptual readiness, 193
and response organization, 191-92
and response readiness, 193
of response selection, 185
supply of, 13-17
dual-task measures, 179-82
and focused attention, 116, 130
and interference, 20-22, 177-91, 199201
and the OR, 46-47
in perceptual processing, 84-85, 169,
188-91,202
physiological measures, 17-24, 184-85,
188
and the PRP, 170-74
243
Effort (cont.)
and spare capacity, 15-16, 20-22,
185-88
and task difficulty, 24--26
and time pressure, 25-26
voluntary control of, 14-15
Electrodermal response (see Skin conductance)
Emphasis
figural (see Figural emphasis)
in report of attributes, 104
Encoding, 103-4, 106, Ill, 187
Epistemic motivation,. 54
Euclidian model, 108
Evaluation of demands (see also Effort
.
demands, Recursive effects), 11,
42
Expectancy wave, 32
Expectations
and consciousness, 130-31
in Hochberg's theory, 128, 130
and the OR, 45, 49
and organization of responses, 192
and perceptual readiness, 91-93, 193
and recency, 45
Extroversion, 35
Eye-blinks, 64
Eye movements (see also Fixation, Looking)
of chess players, 62
lateral, in problem-solving, 63-64
in learning tasks, 61-62
in listening tasks, 62-63, 64, 119, 151,
154
pursuit, 51
in search tasks, 56-60, 81, 191
in social interaction, 63
and tachistoscopic accuracy, 61
and thought, 60-62, 64-65
types of, 50-51
Eye-voice span, 192
Feedback (see Evaluation of demands,
Recursive effects)
Figural emphasis
determinants of, 76-80
and effort, 84-85, 189
in focused attention, 129-30, 134-35
and Kiilpe's task, 105
in search tasks, 80-84, 193
and selection of inputs, 70
stage of, 67-68, 97
Figure-ground (see Figural emphasis)
Filter-attenuation theory
and divided attention, 137
and focused attention, 122-23, 125, 135
Filtering
and figural emphasis, 130
and selection of inputs, 70-71, 112
in speeded classification, 106, 109
Filter theory, 6-7, 120-23, 135
and divided attention, 136, 138, 143,
147, 148
and focused attention, 116, 118, 134~
35
Moray's version of, 142
of the suffix effect, 132-33
Fixation (see a~o Looking)
duration of, 58-59, 81
and Haidinger brush, 51, 60-61
244
Focused attention
and analysis-by-synthesis, 125-26, 133
and arousal, 38-39
on channels or on units, 129, 132-35
and definition of relevant message, 114,
'143-45
Deutsch-Norman theory of, 123-25,
143-45
and effort, 116, 130
eye-movements in, 64-65
and figural emphasis, 129-30, 134-35
filter-attenuation theory of, 122
filter theory of, 116, 120-21
individual differences in, 119
measures of, 113
and memory, 115, 117, 122, 139-40
and message duration, 118-19
in monitoring tasks, 114-15
and processing of unattended stimuli,
116-18, 122, 124, 130-31, 143-45
and the production of expectations, 128
psychoanalytic view of, 128
in recognition task, 115
and resistance to distraction, 113
and RT, 115, 153
in shadowing task, 113-14, 115, 14344
and signal-detection theory, 125
and suffix effect, 132-35
in tachistoscopic task, 112-13, 114
Foreperiod
autonomic changes in, 31-33, 41
effect
and alertness, 194
and errors, 41, 194-95
and RT, 41, 176, 194
Gating, 106
Gestalt
figure-ground, 76-79
laws of grouping, 68, 139
theory, 1-2
Grouping (see also Unit formation)
in filter theory, 130
laws of, 72-74
and response to rejected channel, 131
of responses, 156-62, 164, 165, 191
of responses and anticipation, 192
and TOTE, 192
and search, 74
in speeded classification, 106
in split-span design, 138-41
and the suffix effect, 132-35
GSR (Galvanic Skin Response) (see Skin
conductance)
Habituation, 43, 47-48, 59
Haidinger brush, 51, 60
Hardware models, 172
Hypotheses, in discrimination learning,
99, 102
Imagery tasks, 19, 20, 25-26
Incentives, 14-15, 37
Incongruity, 53, 65
Information theory, 5
Inhibition, motor
and directional fractionation, 31-33
in the OR, 48
Integrality
of dimensions, 107-9, Ill, 156
of responses, 156-62
Interference
capacity, 178-82, 196
in concurrent monitoring, 141-42 153
196
'
,
by distracting stimuli 113
in divided attention, '141-45, 148
effort theory of, 16, 199-201
as measure of effort 179-82 185-88
in memory tasks, 179-80 198
in motor learning, 180 '
output, 139-40
and response conflict 109 III
in shadowing, 127, i41 i53
and similarity of tasks '153 199
in speeded classificati~n, 107-8
in the Stroop test, 109-11, 189
structural, 178, 196-99, 202
suffix effect, 132-35
and task difficulty, 200
Inter-re7~onse interval (IRI), 158, 176-
Orientation (cont.)
versus reorientation, 119
and RT, 48
tendencies, 154
Orienting response (see Orientation reaction)
Overshadowing, 101
Overtraining-Reversal Effect, 100
Ne?ronal model, 43-45, 53, 60, 81, 190

NOIse, effects of, 5, 34, 38, 39, 41
Novelty, 42-47, 52-55, 81
Perception and effort (see Effort Spare

capacity)
,
Perceptual deficit
in mental tasks, 16, 20-22, 180 185
186 189
'
,
Perceptual' interpretation (see Selection
of, Perceptual readiness)
Perceptual readiness, 69, 88
and filter-attenuation theory, 122
and preparatory set, 193, 194
and processing of rejected message,
130
and signal-detection theory, 91-97
Pertinence, 125, 134, 145
Pigeon-holing, 96
p,.,pattern, 29
Pre-attentive mechanisms, 7, 71, 74-75,
126, 127, 129, 133, 135
Prefix effect, 132-33
Prior entry, 78-79, 137-38, 193
Probe measures (see Spare capacity)
Processing, parallel or serial
in compound responses, 158
of dichotic stimuli, 145-52
of dimensions, 104, 106, 107
and filter theory, 121, 136
in focused attention, 116-18
hardware or software models, 172
6f inpu~, 104, 124, 141
in Keele's theory, 169
and orientation tendencies, 154
prior entry, 78-79,137-38, 193
in psychophysical judgments, 146-47,
148-49
and response-conflict theory, 174-75
Treisman's theory of, 105, 123, 137,
153
Psychological refractory period (PRP)
and the allocation policy, 172
and delay of Rt, 179, 195
and effort, 170-74
expectancy theory of, 176
and feedback delay, 164, 166
response-conflict model, 174-75
and the RSI paradigm, 172-74
single-channel theory of, 163-65, 167,
170
in tracking, 162
Psychophysical judgments, 146-47, 14849
P-system, 120, 136, 182
Pupil dilation
in directional fractionation, 29-30, 33
as measure of effort, 18-27, ,184
in OR, 42,46
Orientation
reaction (OR), 42-49, 126-27, 190
reaction to unattended message, 126,
131
Reaction time (RT) (see also Foreperiod)

to directional cues, 48
in divided attention, 145-46
facilitation, by loud noise, 175-76
minimal, for ungrouped responses,

168-70, 177
and the PRP, 164-69
and response grouping, 158-61, 165
Introversion, 35
Knowledge of results (KR), 35-36
Learning tasks (see also Short-term memory)
and arousal, 41-42, 46
eye-movements in, 61
in measurements of effort, 19, 24, 26,
180, 183-84
Logogen 8, 85
Looking (see also Eye movements, Fixation)
in animals, 52-53
autonomic concommitants of, 19, 2931
collative properties in, 52-55, 60
in ,infants, 52
and information acquisition, 56-60
spontaneous, 51-55, 65
Mental arithmetic, 19, 24, 26, 179, 181
Modality effects
in dual-task performance, 198-99
in performance of single task, 196-98
in split-span design, 138
Monitoring (see also Search)
and divided attention, 141-45, 14853, 155, 1 9 0 '
and focused attention, 114-15, 119,
125, 153
Muscular strain, 18, 22, 109, 114
245
Reaction-time (cont.)
in focused attention, 115, 118
to a probe signal, 180, 186-88, 196
and the PRP, 162-77
to simultaneous stimuli, 157-62
Recognition memory
in divided attention, 144-45, 147-48,
150-52
in focused attention, 115
intederence with, 152, 190
Recognition units, 67-68, 94, 97, 130,
189
acquisition of, 87
organization of, 85-87
parallel processing in, 86-87
Recursive effects
and demand2, 190, 193
in Norman's theory, 125
and OR, 44, 81
in perception, 69-70
in search, 81, 193-94
Redundancy gain, 107
Rehearsal, 19,.26, 140, 179, 180, 202
Response conflict
and collative properties, 53
and the Stroop test, 109-11
theory of the PRP, 174-75
Response readiness, 69
and attention to attributes, 105
and signal-detection theory, 94-97
in the Stroop test, 110
Response selection (see also Response
readiness)
and attention to attributes, III
effort demands of, 185
free, and effort, 19, 183-84
as single-channel, 182-84
stage, 67, 69-70
Response set, 70, 71, 114, 121, 124-25
Reversal shift, 100
Rod and frame test, 39, 40
Saccade, 50
Scanning, 40-41
Search
control of, 71-72, 74
and figural emphasis, 80-84, 193-94
and grouping, 74-75
learning to, 81, 87
recursive process in, 81, 194
visual, 57
Selection
of attributes, 71, 98-111
of inputs, 70-71, 79, 112, 119 (see
Focused attention, Stimulus set)
of outputs, 70-71 (see Response set)
of perceptual interpretations, 67-70,
85 (See Perceptual readiness)
in focused attention, 130
and signal-detection theory, 94
of response (see Response selection)
of targets, 70-71 (see Search)
Sensitivity (d')
and attention, 94-95, 125
and availability of recognition units,
94-95
in signal-detection theory, 88-91
Set (see also Stimulus set, Response set,
Expectations)
246
Set (cont.)
divided, 195-96
preparatory, 191-95
Shadowing
of alternated messages, 154
and focused attention, 113-14, 115,
143-44
intederence with, 127, 141, 153
latency of, 118, 192
and processiJ;lg of other messages,
113-14, 117, 122, 198
Short-term memory
and arousal, 42
and attention, 116, 117-18, 122, 13235, 139-40
and effort, 19-20, 25, 26, 179-81
types of, 121-22
Signal-detection theory
and attention, 94-96, 125
basic concepts of, 87-90
and distorted room, 91-93
and word-frequency effect, 93-94
Similarity judgments, 102-3, 108
Single-channel theory
and dual-tasks, 182-85
and the PRP, 163-65, 167, 170
Sinus arrhythmia, 17
Skin conductance, 18, 32-33, 42-43
Sleep deprivation, 35-36
Spare capacity, 16
and arousal, 38
and demandt, demand2, 190-91
and primary effort, 15-16
probe measures of, 16, 202
perceptual, 16, 20-22, 180, 185,
186, 196
RT, 16, 180, 186-88, 195-96
Speed-accuracy tradeoff, 41
Split-span design
attention and retention in, 139-41
content effects in, 139
and perceptual grouping, 138-40
S-system, 120, 136, 138
Stimulus set, 70, 71, 114, 120, 124-25
Stroop test, 32, 33, 39-40, 64, 109-11
Structural models, 8, 11
Subsidiary tasks
other variants, 179-84
perceptual measures, 20-22, 180, 18586
probe RT, 180, 186-88
Suffix effect, 132-35
Sympathetic activity (see Autonomic
changes)
Tachistoscopic perception
and arousal, 38-39
eye movements in, 61
and focused attention, 38-39, 112-13,
114
selection of targets in, 83-84
of words and letters, 85-86
Time pressure, 25-26, 191
TOTE, 192
Tracking, 162, 179-81, 183-84
Vigilance, 35, 41
Word-frequency effect, 93-94

Attention and Effort. (1973)

Uploaded by

Copyright:

Available Formats

Attention and Effort. (1973)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Attention and Effort. (1973)

Uploaded by

Copyright:

Available Formats

Attention

Englewood Cliffs, New Jersey

Library of Congress Cataloging in Publication Data

Attention and effort.

1973 by PRENTICE-HALL INC., Englewood Cliffs, New Jersey

All rights reserved. No part of this book may be

Printed in the United States of America

PRENTICE-HALL INTERNATIONAL, INC.,

PRENTICE-HALL OF AUSTRALIA, PTY. LTD.,

1 BASIC ISSUES IN THE STUDY OF ATTENTION

2 TOWARD A THEORY OF MENTAL EFFORT

3 AROUSAL AND ATTENTION

5 ATTENTIO'N AND PERCEPTION

Stages of Perceptual Analysis, 66; Taxonomy of Selective Attention,

Discrimination Learning, 98; Selective Report of Attributes, 103;

7 FOCUSED ATTENTION-FINDINGS AND THEORIES

8 ATTENTION DIVIDED AMONG INPUTS

Concurrent Messages'and the Split-Span Experiment, 137;

9 SPEEDED RESPONSES TO SIMULTANEOUS AND TO

10 ATTENTION AND TASK INTERFERENCE

Capacity Interference, 178;

The concept of attention has had an uneven career in the history of

ATTENTION AND EFFORT

SELECTIVE ASPECTS OF ATTENTION

The existence of mechanisms that control the significance of stimuli

Basic Issues in the Study of Attention

to control the choice of stimuli that will be allowed, in turn, to control

INTENSIVE ASPEGrS OF ATTENTION

ATTENTION AND EFFORT

tion has become one of the central topics of experimental psychology. In

Basic Issues in the Study of Attention 5

BOTTLENECK MODELS OF ATTENTION

One of the classic dilemmas of psychology concerns the division

ATTENTION AND EFFORT

Model A illllstrates son1e central aspects of the filter theory first

Basic Issues in the Study of Attention

A CAPACITY MODEL OF ATTENTION

A capacity theory of attention provides an alternative to theories

ATTENTION AND EFFORT

there is a general limit on man's capacity to perform mental work. It

Basic Issues in the Study of Attention

noted that momentary variations in the difficulty of what a subject is

to other activities. In addition, of course, an action can fail because the

Basic Issues in the Study of Attention

ATTENTION AND EFFORT

REVIEW AND PREVIEW

THE MOBILIZATION OF EFFORT

ATTENTION AND EFFORT

Toward a Theory of Mental Effort

asked to perform an easy and a relatively difficult task separately, under

ATTENTION AND EFFORT

A hypothesis concerning the human response to demands on effort

Toward a Theory of Mental Effort

THE MEASUREMENT OF EFFORT BY AROUSAL

According to the capacity !ll0del introduced in the first chapter, the

ATTENTION AND EFFORT

available capacity and

Toward a Theory of Mental Effort

ATTENTION AND EFFORT