I

~
~

t
-....
l::
~

...l:l
....
l:;'
Concepts
:::

and
..

~
~
~

~
'.
~
1

.......
~

~
Principles
~
~
"'i
....
:::-
~
of
~.

J,
.
...:
l:l
:::
...
~
Behavior
::
~

....:::
~
Analysis .,

, t::I:l
~
:::-.
l:l
....
...:
t ~
"'i
~
:::
l:l
~ Jack L. Michael
....'"
'"
ACKNOWLEDGMENTS

Chapters 1, 4, S, 6, 8, and 10 have not heen previously published. They were developed for use in my
courses taught in the Psychology Department at Western Michigan University·and in various public
presentations. .
Chapter 2: This paper is a modified version of Michael, J. (1991). Historical antecedents of behavior
analysis. The ABA Newsletter, Summer 7-12. Adapted by permission of the Association for Behavior
Analysis.
, .
Chapter 3: This paper is a modified version of Michael, J. (1985). Behavior analysis: A radical
perspective. In B. L. Hammonds (Ed.). Master Lecture Series, Volume 4: Psychology o/Learning.
Washington, D. C.: American Psychological Association. Copyright 1985 by the American Psychological
Association. Adapted by permission of the publisher. I am grateful to Ernest A. Vargas for his very helpful
comments on an earlier version of the paper.
Chapter 7: Sections of this paper were reprinted with the permission of Macmillan PUblishing Company
from Science and Human Behavior by B. F. Skinner. Copyright © 1953 by Macmillan Publishing
Company, renewed 1981 by B. F. Skinner.
Chapter 9: This paper is a modified version of Michael, J. (1983). Evocative and repertoire-altering effects
of an environmental event. The Analysis o/Verbal Behavior, 2, 19-21. Adapted by permission of Mark
Sundberg, Editor, The Analysis o/Verbal Behavior.
Chapter 11: This paper is a modified version of Michael, J. (1986). Repertoire-altering effects of remote
contingencies. The Analysis o/Verbal Behavior, 4, 10-18. Adapted by permission of Mark Sundberg,
Editor, The Analysis o/Verbal Behavior. Many of the notions presented here were developed in discussion
with Michael A. Minervini and from his presentation at the 1985 annual conference of the Association for
Behavior Analysis (Minervini, 1985). His doctoral dissertation (Minervini, 1989) contains a more detailed
treatment of the same general topic.
Chapter 12: This paper is a modified version of Michael, J. (1982). Skinner's verbal operants: Some new
categories. Verbal Behavior News, 1,1. Adapted by permission of Mark Sundberg, Editor, Verbal Behavior
News.
Chapter 13: This paper is a modified version of Michael, J. (1988). Establishing operations and the
mand. The Analysis 0/ Verbal Behavior, 6, 3-9. Adapted by permission of Mark Sundberg, Editor, The
Analysis o/Verbal Behavior.
Chapter 14: This paper is a modified version of Michael, J. (1985). Two kinds of verbal behavior plus a
possible third. The Analysis a/Verbal Behavior, 3,14. Adapted by permission of Mark Sundberg, Editor,
The Analysis a/Verbal Behavior.
Chapter IS: This paper is a modified version of Michael, J. (1991). A behavioral perspective on college
teaching. The Behavior Analysis, 14, 229-239. Adapted by permission of the Society for the Advancement
of Behavior Analysis.
 Contents

Acknowledgments, iii
1. Outline of the Science of Behavior, 1
Abbreviated Outline, 3
Expanded Outline, 4
I. Stimulus, 4
ll. Response, 5
Ill. Unlearned Behavior, 5
IV. Learning and Motivation, 6
Y. What a Specific Organism Learns, 17

2. Historical Antecedents of Behavior Analysis, 19

Behavior Analysis History Chart, 21 ., .
Principal Contributors, 22

3. Behavior Analysis: An Overview, 29

A Little Recent History, 29
Concepts and Principles, 30
Unlearned Functional Relations, 30
Respondent Functional.Relations, 31
Operant Functional Relations, 32
Methodology, 36
Rate of Response, 37
Within-subject Comparisons, 37
Statistical Inference, 38
Theory as a Basis for Research, 38
vi Contents

Extensions, .39
Behavior Modification or Applied Behavior Analysis, 39
Four Interpretive Analyses by Skinner, 40
Conclusion, 43

4. Contrast Between the Procedures

of Respondent and Operant Conditioning, 45
Respondent Conditioning, 45
Operant Conditioning, 46
The Procedures Exemplified, 46 .
Attempting to Operantly Condition Peripheral Vasoconstriction, 46
Attempting to Respondently Condition the Lever Press, 47
Procedures Versus Underlying Process, 47

5. Common
, . Errors in the Use of Behavioral Terms, 49
Main Terms, 49
Other Common Usage Errors, 50
Some Errors of Pronumciation, 52

6. Explanatory Fictions, 53
7. Establishing Operations, 57
Establishing Operation Defined in Terms of Two Features, 58
Unconditioned Establishing Operations, 60
Deprivation and Satiation, 60
Temperature Changes, 60
Variables Relevant to Sexual Reinforcement, 60
Painful Stimulation: Escape, 61
Unconditionedn Reinforcement, 62
Painful Stimulation: Aggression, 63
Other Emotional EOs, 63
EOs and Punishment, 64
A Respondent Analogy, 64
Conditioned Establishing Operations (CEOs), 64
Surrogate CEO: Correlating a Stimulus with a UEO, 66
Reflexive CEO: Correlating a Stimulus with Worsening or Improvement, 68
Transitive CEO: Conditional Conditioned Reinforcement and Punishment, 70
Gene~al Implications, 72

8. The Discriminative Stimulus, 73

Del}nition, 73
SD and SA (S delta), 73
Neutral Stimuli, 73
 Contents vii

Discrimination Based on Punishment, 73

The Common Definition, 74
Not a Procedural Definition, 74
Contrast with Cognitive Defmitions, 75
The Correlation of the SD versus that of the sr, 75
Motivative Variables,75
During Acquisition of Stimulus Control, 75 ,
When the SD Occurs After Acquisition, 76
The Strength of an SD, 76
Operant Distinguished from Respondent Stimulus Control, 76

9. lWocative Versus Function-altering F1feets ofThvironmentai Events, 77

Definition, 77
Operant Evocative and Function-altering Relations, 77
Related Terms, 78
Provenance, 78
Motivative Relations, 78
Respondent Evocative and Function-altering Relations, 78
Multiple Functions of Environmental Events, 79
Direct Versus Indirect Effects of Environmental Events, 81

10. Stimulus Change Decrement and Stimulus Generalization, 83

Defmition, 83
lllustration of this Principle, 83
With an Ambient Stimulus in an Operant Conditioning Procedure, 83
With a Discriminative Stimulus (SD), 83
With Extinction of an SD, 84
With a Conditioned Eliciting Stimulus (CE), 84
With an Ambient Stimulus in a Respondent Conditioning Procedure, .'84
With Respect to Extinction of a Conditioned Elicitor, 84
As an Explanation of the Partial Reinforcement Effect, 84
Decrement Compared with Stimulus Generalization, 85
Generalization of Extinction, 85

11. Behavioral Effects of Remote Contingencies, 87

Direct and Indirect Effects, 87
Other Clues that an Effect is Indirect, 90
How Should We Talk? 91
But Why Are We Successful? 92

,
"
viii Contents

12. Elementary Verbal Relations, 95

Mand,95
Tact, 95
Intraverbal Behavior, 96
Codic Behavior, 96
Duplic Behavior, 97

13. The Mand, 99

Refinement of the Definition, 99
Practical Implications, 101
Conclusion, 103

14. Two Kinds of Verbal Behavior Plus a Possible Third, 105

Tppography-based Verbal Behavior, 105
Selection-based Verbal Behavior, 105
Important Differences, 106
Manded Stimulus Selection, 108

15. A Behavioral Perspective on College Teaching, 109

An Important Type of College Teaching, 109
Student Motivation, 110
Intrinsic Interest in the Subject Matter, 111
Approval and disapproval from Significant Others in the Study Setting, .J 11
Social Reactions of Others to the Repertoires Acquired Through Study, 111
Short-Term Advantages to Oneself from the Newly Acquired Repertoire, 112
Long-Range Payoffs, ·112
The Course Grade, 112
University Grading Practices, 112
The Procrastination Scallop, 114
Competing Activities, 115
Course Grade Importance, 115
The Relation Between Exam Grade and Course Grade, 115
The Relation Between Studying and the Exam Grade, 115
Task arid TIme Monitoring, 117
Conclusions, 118

16. References, 121

 1
OUTLINE OF THE SCIENCE OF BEHAVIOR
This paper consists of a one-page outline
and an expanded treatment of the topics
shown in the outlin~, and it serves several
putposes. The research and applied work
of those who call themselves behavior
analysts (e.g., the curr~nt memlx:rs of the
Association for BehavIOr AnalysIs) are
relatively narrow in focus. This .. .
limitation in current interests and acnvlttes
sometimes leads outsiders and behavior
analysts to conceptualize the field
erroneously in similarly narrow terms,
and to neglect some topics as . subcategory that has been the ~ri~ary
unimportant, to assign them to som~
other discipline, or even to deny therr .
existence. In this paper, I try to descnbe
and indicate relations arnong all possible
aspects of the science of behavior,
whether or not they are a current focus of
behavior analysis. One can approach the
topic by asking what do we need to know
in order to predict, control, or interpret
the behavior of an organism.
First, we need to know how the
organism is affected by the environment,
and especially what stimuli can affect that
organism. Next we have to know how
that organism can affect the environment;
that is, what are its response
mechanisms. Stimulus and response are
thus the first two major categories of the
outline. It is also imponant to know what
environment-behavior relations make up
the organism's unlearned or built-in
repertoire (the third major category)..
Behavior analysts have not been heavily
involved in the discovery of knowledge
in these three broad areas, but they do not .
ignore them in any comprehensive
presentation.
Next, one must understand how the
organism is modified by its interaction
with the environment, how new
environment-behavior relations are
learned and unlearned. This is the founh
major category, and it can be broken
down into tlrree subcategories:
habituation, respondent functional
relations, and operant functional
relations. It is, of course, this third
area of basic research and apphcanon for
those who consider themselves behavior
analysts.
Finally, to understand any particular
organism one must know not only those
functional relations that are common to all
members of that species, its unlearned
repenoires and how it learns, QVt also
what it has learned, or more generally,
what that type of organism typically
learns during its lifetime. This fifth
category has not been a major focus of
interest for behavior analysts, although
earlier and recent effons to behavioralize
the areas of child development, social
psychology, personality, lang?age and
communication, and other tOpiCS of
traditional psychology are clearly in this
category.
Although our field, as exemplified
by the activities and interests of those
who consider themselves behavior
analysts, has consisted largely of research
and application related to operant
functional relations, all of the rest of the
 2 Outline of the Science ofBehavior
science of behavior is important and must
be a part of any comprehensive treatment.
Some topics have not been a part of
behavior analysis simply because of
historical factors related to the
development of the current version of the
field. Some, especially those in the fIfth
major category, may have been excluded
because of the nonbehavioral, mentalistic,
cognitive, or other orientation of those
who currently work in those areas. This
neglect, however, should be considered
only temporary. As our effectiveness in
dealing with those basic and applied
topics we now specialize in continues to
improve, we will undoubtedly begin to
consider the subject maner of topics that
are now dealt with from other theoretical
perspectiveS:' It would be a serious
mistake to reject this subject maner as a
part of our rejection of the current
theoretical approaches.
Thus, one major purpose of the
outline is to display the topics of the
science of behavior so that none are
erroneously ignored, devalued, or
assigned to nonbehavioral fields.
Another is to locate various topics in an
organized comprehensive system, so that
their relations to other topics can be bener
understood and more effectively used,
both theoretically and practically. The
outline is also helpful for locating some
traditional topics, like motivation and
memory, which are usually dealt with as
separate psychological processes, but
from a behavioral perspective are best
considered as parts of several different
behavioral topics.
Still a third purpose of the outline is
to counteract an oversimplified
.;
characterization of the area in which we
are most involved, operant functional
relations. In our·efforts to bring our
knowledge to as many potential users as
possible, we have often emphasized only
the most basic and simplest concepts and
principles. Many of us will happily bring
this knowledge to an audience of
whatever scholarly level and present our
subject matter for whatever duration is
available (a I-hour lecture, a 3-hour
workshop, a part of an academic course,
the entire course, etc.). As a result of
some of the shorter formats, a
misconception issometimesgenerated to
the effect that there really isn't much to
leam in this area A display of the details
in outline form is one way to avoid this
possible misconception.
In the expanded portion of the
outline, I have also elaborated on some
aspects of my approach to teaching
behavior analysis that may not be dealt
with in other assignments. In this way I
have a wri nen version of various lecture
topics that I can assign when appropriate.
These topics include the definitions of
stimulus and response, the concept of
stimulus change, evocative versus
function-altering relations, a rather
detailed analysis of the operant dependent
variable, and establishing operations.
The material in the expanded outline
is not meant to function without lecture or
some other supplementation. I usually
include it as text material in any
behavioral course I am teaching, provide
a brief lecture on the major categories,
and then assign various parts of the
material as the topics arise.
 ABBREVIATED OUTLINE OF THE SCIENCE OF BEHAVIOR

I. Stimulus , b. Uncond establishing ops (UEOs)

A. Defined c. Cond establishing ops (CEOs)
1. Human receptors i. Surrogate CEO
2. Others ii. Reflexive CEO
B. Stimulus change iii. Transitive CEO
d. Emotional EOs f
II. Response 7. Stimulus control ,
A. Defined a. S~hange decrement re ambient S
1. Human effectors b. Discriminative stimulus
2. Others c. Stimulus control procedures
d. Operant discrimination based on pmt
III. Unlearned Behavior e. Relevance of establishing op
, A. Specific S-R relations . f. Oeneralization and S-<:hange dec re SD
B. Behavior of the whole orgaDlsm g. S class membership and equivalence
1. Kineses h. Attentionlblocking re compound S
2. Taxes i. Conditional discrimination
3. Fixed action patterns 8. Conditioned rfmt and pmt
4. Reaction chains a. Procedure
C. More complex relations b. Generalized conditioned rfmt or pmt
IV. Learning and Motivation c. Relevance of establishing operation
d. Chaining
A. Habituation 9. Operant weakening operations
B. Respondent functional relations a. Extinction
1. Type of effector b. Two ways to weaken an SD
2. Experimental preparations
3. Conditioning c. Two ways to weaken an ST or SP
4. Higher order conditioning d. Punishment:
i. Uncond and cond pmt
5. Dependent variables
6. Temporal relations ii. Other variables
iii. Recovery from punishment
7. Other variables
8. Extinction e. Conditioned suppression
9. Stimulus control 10. Escape and avoidance
a. S-<:hange decrement re ambient S 11. Schedules of reinforcement
b. Generaliz and S-<:hange dec re the CS a. types
c. Stimulus discrimination b. why schedules are important
d. Stimulus class membership 12. Miscellaneous
e. Attention/blocking with compound S a. Another type of memory
b. Adjunctive behavior '1
10. Miscellaneous
a. Emotion c. Skinner's operant perceptual response
13. Evocative vs function-altering relations
b. Conditioned perceptual rsps
c. Drug addiction and overdosing V. What a specific organism learns
II. Evocative vs. function-altering 1. Basic condo reinforcers and punishers
relations 2. Basic sensory-motor repertoires
C. Operant functional relations 3. Elementary social repertoires
1. Type of effector 4. Elementary verbal repertoire
2. Experimental settings 5. Control by private stimuli
3. Rfmt and pmt: definitions & basic 6. Intellectual repertoires
function-altering relation 7. More complex sensory-motor repertoires
4. Other independent variables 8. Social and interpersonal skills
5. Dependent variables 9. Interacting repertoires in the individual
6. Establishing operations (BOs) 10. Repertoires related to group practices
a. Two defining effects 11. Repertoires related to cultural practices
i. rfer-establishing effect
ii. evocative effect
4 Outline of lhe Science ofBehavior

EXPANDED OUTLINE

I. Stimulus
A. Energy change that affects an organism through a receptor.
Defining stimulus in terms of an organism's receptor systems is a form of caution based on
the fact that all of what we know about the effects of stimuli on behavior has come from
laboratory study of energy changes that can be easily related to specific receptor systems
(tones, lights, touches, etc.). It seems safer to use the term stimulus analogy for events
ordinarily referred to as mental images, dreams, hallucinations, etc. that seem to function as
though they are stimuli, even though no specific receptor systems are being activated. The
energy changes in central nervous system structures produced by receptor input may very
well function like stimuli, even when produced in some other way (e.g., by the activity of
other central nervous system structures). However, there may be important differences
between the behavioral effects of such energy changes and the effects of those that activate
receptors. Because almost nothing is known at present about the scientific status of such ,~
"
events, it would seem appropriate to identify them as analogies, and reserve stimulus for
energy changes whose behavioral effects are better understood.
1. Human receptors
Vision, hearing, smell, taste, cutaneous sense (surface touch, surface temperature,
surface pain), kinesthesis (muscle sense), vestibular sense (balance), organic sense
(deep touch, deep temperature, deep pain).
2. Others
Some organisms have other receptor systems, such as the voltage receptor ofthe
shark and the infrared receptor of some reptiles.
B. Stimulus change versus a static stimulus condition
Stimulation is one of the main ways the environment affects an organism, but there is a
complication arising from the fact that stimulus conditions whose importance for the
organism is of ontogenic provenance-the result of a learning history-are primarily
important as stimulus changes, not as static conditions. Another way to express this notion
is to refer to the occurrence of a stimulus, or to refer to a stimulus event; both of these
terms are dynamic rather than static in implication. This means that when the stimulus
appears in the context of behavior analysis, it usually means a stimulus change. For
example, in respondent conditioning the conditioned stimulus may be referred to as a tone.
However, the relevant stimulation actually consists in a change from the absence of tone to
the tone sounding, and although this is usually understood without having to be mentioned,
it is easily overlooked in the analysis of more complex phenomena. In the case of operant
functional relations, discriminative stimuli, conditioned reinforcers, conditioned puniShers,
and conditioned establishing operations are also usually important as stimulus changes, not
as static conditions.
This complicates our analysis in that the behavioral characteristics of a stimulus change
must be described in terms of at least two conditions: the prechange condition and the
postchange condition. This is not a problem if all we are concerned with is a description of
the two st;imulus conditions themselves (e.g., tone off and tone on). But for many .
purposes our analysis requires consideration of other behavioral aspects of the stimulus
conditions. Thus, it is not possible to refer to a static stimulus condition as a conditioned
reinforcer, because the actual reinforcing effectiveness depends as much on the reinforcing
features of the prechange condition as of the postchange condition. For example, to
describe a particular stimulus condition (e.g., a tone being souncied) as a conditioned
reinforcer because in the presence of the tone food (for a food-deprived organism) is
delivered at a particular rate, would be quite incomplete without a description of the rate of
 Outline of the Science ofBehavior 5

food delivery in the absence of the tone. Even with such a seemingly simple phenomenon
as punishment by painful stimulation, it is not possible to quantify the punishment effect
simply by stating the intensity of the stimulation that occurs after the response (e.g., 50
volts of shock stimulation). It is true that a change from 0 volts to 50 volts will function as
punishment, but a change from 100 volts to 50 volts is reinforcement, not punishment.
The analysis of stimuli as stimulus changes is important for understanding several
phenomena that are sometimes cited as evidence for the inadequacy of a behavioral
approach and for the necessity of cognitive concepts, such as the weakening effect of
unconditioned-stimulus-alone trials in respondent conditioning, as discussed below (see
IVB8 under respondent extinction). It is also relevant to the explanation of some
phenomena that are cited as evidence for the necessity of a molar as opposed to a molecular
approach (e.g., avoidance behavior).

II. Response
A. The neurally innervated action of an organism's effector
Defining response in terms of effector action is based on the same argument given above
for defining stimulus in terms of receptor action. All of our knowledge regarding
responding is based on laboratory study of the kind of behavior that is easily seen to
consist of effector action. It seems safer to use the term response analogy for response-like
events such as thinking and perceptual behavior. The same is true for mental images,
dreams, and hallucinations, mentioned above as stimulus analogies, which in some
respects may be more like responses than like stimuli. Although it may well be
theoretically and practically useful to conceptualize such events as stimuli and responses, it
is nevertheless important to identify such speculative extension as such with the term
analogy.
1. Human effectors
Muscles (striped and smooth) and glands (exocrine or duct and endocrine or
ductless).
2. Others
Some organisms have other effectors, such as the luminescent organ of some insects
and deep-sea organisms, the voltage-generating organ of the electric eel, and the
color-change effectors of the chameleon and the octopus.

III. Unlearned Behavior

Most of our knowledge of unleamed behavior, with the exception of the first cate~ory
below (reflexes) comes from the study of nonhuman organisms. Humans have a'number
of unconditioned reflexes, but the actual status of proposed human examples of any of the
more complex kinds of unlearned behavior is still somewhat problematic. This is because
of the heavy involvement of learning in all human behavior and because the most powerful
means of distinguishing learned from unlearned behavior-raising the organism in
conditions in which learning cannot occur-is not ethically possible with humans. (The
functional relations described below under IV are also inherited, in the sense that organisms
are innately equipped to be altered by the environment in the manner described.)

A. Specific stimulus-response relations: Reflexes

Vasoconstriction to coldness, activation syndrome to painful or other intense
stimulation, salivation to food in the mouth, etc. Humans have plenty of these.

B. Behavior of the whole organism

1. Kineses
The wood louse and moisture regulation.
6 Outline ofthe Science ofBehavior

2. Taxes
Positive phototropism in the moth.
3. Fixed action patterns
Egg retrieval by the goose; some spider's nest weaving.
4. Reaction chains
Hermit crab shell selection.
C. More complex relations.
The first several categories of unlearned behavior listed above seem to involve relatively
specific stimulus-response relations, but as the behavior becomes more complex (mating,
nest building, aggressive behavior, etc.) it seems to consist of a mixture of relatively
specific establishing operations (see IVC6 below), which then lead to operant conditioning
of whatever behavior achieves the relevant event that is made effective as reinforcement by
that establishing operation. Skinner refers to this as an intermingling of the contingencies
of survival (unlearned or innate determination) and the contingencies of reinforcement
(learning). .

;. IV. Learning and Motivation.

A. Habituation
Repeated presentation of an unconditioned stimulus for startle or orientation leads to a
lasting but not irreversible decline in response magnitude to that stimulus; this is the
simplest form of learning. Because the effect of such repeated presentation persists for a
time even in the absence of further presentations, this is probably the simplest form of
behavior change to invoke the term memory, which of course doesn't constitute an
explanation but only another narne for the facts of the phenomenon.

B. Respondent functional relations

1. Type of effector
Smooth muscles and glands, mostly, but there are some striped muscle reflexes.
2. Experimental preparations
Dog salivation, human eye blink, human galvanic skin response, human heart rate,
rabbit nictitating membrane, rabbit eye blink, pigeon autoshaping, rat conditioned
suppression (note that here the actual measured response is lever pressing established
and maintained by operant reinforcement), rat taste aversion. (A single occurrence of
a taste stimulus prior to an organism's becoming nauseated becomes a conditioned
aversive stimulus, even if there is a long delay between the taste and the nausea It
only seems to work with taste as the conditioned stimulus and nausea as the
unconditioned stimulus/response. As with conditioned suppression the supposed
conditioned response is only indirectly observed.)
3. Conditioning
If one pairs a neutral stimulus with an unconditioned stimulus (US), then the neutral
stimulus becomes a conditioned stimulus (CS) in that it will elicit a type of response
similar to what the US elicited. From a molar perspective conditioning is said to
result from correlating CS presence/absence with US presence/absence (however, see
below with respect to respondent extinction-IVB8-and the weakening effect of US-
alone trials).
Because the effects of the respondent conditioning procedure last over time-a
dog conditioned to salivate to a tone on one day will salivate to the sarne tone the next
day-this is a phenomenon to which the term memory is sometimes applied. Thus
the dog is said to remember the relation between CS and US. However this is just
another way of naming the results of the conditioning procedure and is not in any
 Outline of the Science ofBehavior 7

sense an explanation of the fact that the es elicits the response after a period of time.
The only evidence for the remembering is the responding, which makes the
remembering an example of an explanatory fiction.
As described below (see 1VB1l), we are here concerned with the function-
altering effect of a US, and it is the essence of function-altering effects that they .
produce lasting (but not irreversible) changes in the organism's capacity to be affected
in various ways by environmental variables. In general, the concept of memory is
quite unnecessary in dealing with function-altering effects, which by definitidn last
over time. Some day it will be possible to identify neurochemical changes in me
central nervous system that will explain the results of the various functioncaltering
effects. Until then, the procedural history is the only adequate explanation for the
later effect
4. Higher order conditioning
If one pairs a neutral stimulus with an already effective es, then the neutral stimulus
will become a es and will,elicit a response similar to what the already effective es
elicited.
5. Dependent variables
With smooth muscle and gland responses, the usual dependent variable is the
magnitude of the response produced by the es or the latency of the response to the
es. For example, Pavlov used the amount of saliva (number of drops) and the time
between the es presentation and the first drop of saliva. Proportion of successive
blocks of conditioning trials on which the conditioned response occurred is the most
common dependent variable used with striped muscle reflexes such as the eye blink.
With conditioned suppression, the dependent variable consists of a change in the rate
of lever pressing (maintained by food or water reinforcement) which is used to infer
the presence of an emotional conditioned response.
6. Temporal relations
Short delay, long delay, trace, simultaneous (ineffective), backward (ineffective),
temporal conditioning (no es, just regularly timed presentations of the US).
7. Other variables
In addition to the temporal relation between es and US, other variables relevant to
the effectiveness of the conditioning procedure include es and US intensity, es and
US duration, and the sense mode of the es with respect to the particular US being
used.
8. Extinction "
Presenting the es without the US is the ordinary extinction procedure, but presenting
the US without the es also decreases the elicitative strength of the es. From a molar
perspective, the correlation bet:ween es presence/absence and US presence/absence is
the essential feature of respondent conditioning, and both eS-alone trials and
US-alone trials have the effect of decreasing this correlation. A molecular perspective
implicates temporal contiguity (pairing) of es and US as the essential operation, and
eS-alone trials certainly constitute unpairing. To explain the weakening effect of
US-alone trials, however, requires an analysis of es presentation as a stimulus
change (see Ie above) consisting of both a pre- and a postchange stimulus condition
The US-alone trials represent pairing of the US with the prechange condition, thus
decreasing the degree to which es presentation is a change with respect to US
presence.
As with conditioning, extinction causes lasting (but not irreversible) changes in
the organism's capacity to be affected in a certain way by a stimulus, and thus
invokes the concept of memory; in other words, extinction is a function-altering
operation. But again, memory is no explanation of the phenomenon, and if used as
such is an explanatory fiction.
8 OUlline of lhe Science ofBehavior

Sponmneous recovery, disinhibition, and conditioned inhibition are additional

phenomena related to extinction.
9. Stimulus control
a. Stimulus-change decrement re ambient stimuli
Mter conditioning, any change in any of the ambient stimuli (room illumination,
ambient noise level, kinesthetic stimuli resulting from body posture, tactile stimuli
caused by contact with the restraining apparatus, etc.) will result in the es having
a smaller elicitative effect than if the stimuli had remained exactly as they were
during conditioning.
b. Generalization and stimulus-change decrement regarding the CS
Any change in the es from what it was during conditioning will result in its
having a smaller elicitative effect than if the change had not occurred. The decrease
resulting from this change is called stimulus-change decrement. The fact that the
changed stimulus still has some elicitative effect, which it would not have had
before conditioning with the original es, is referred to as stimulus generalization.
In a quantitative sense, stimulus generalization and stimulus-change decrement are
the complements of one another.
c. Stimulus discrimination
ByTepeatedly pairing a es with the US and repeatedly presenting, without the
US, a stimulus that through generalization at first produced the same response as
the es, the generalized elicitative effect can be reduced while the effect of the
original es is maintained. This type of training and the resulting kind of stimulus
control is referred to as discrimination. Discrimination is a name for the resulting
differential stimulus control, not an explanation for the control. The proper
explanation for the differential stimulus control is the history of differential pairing
of the different stimuli with the US. Thus it is quite mentalistic to say that the
organism responds differentially because it "discriminates between the two
stimuli." In other words, discrimination is a name for the training procedure or for
the behavioral result of this procedure, not for an underlying mental process.
Thus, it does not constitute an explanation for the behavioral result.
d. Stimulus class membership
Two stimuli are said to be members of the same stimulus class when it is possible
to change the behavioral function of one and find that the other has already been at
least partially changed in the same way. Stimuli that physically resemble one
another are usually members of the same class in this sense, without any leaming
history being required. However, after various kinds of leaming histories (such as
repeated insmnces of temporal contiguity or pairing), stimuli that have no physical
resemblance may become members of the same stimulus class.
e. Attention and blocking with compound stimuU
With compound stimulation (for example simultaneous presentation of a tone and a
light) as the es, the extent to which anyone component of the compound acquires
elicitative control over the response is not simply a matter of the number of
pairings of c<1IDpound es and US. For example, if one of the components has
., . previously been developed as a es by itself for the same response, it may block
, , ' W,e acquisition of elicitative control by the other component, irrespective of how
"x','mMytimes
. ,., ..1._, ?
.~,.,c.j".,
the compound es and the US are paired. The issue is qnite complex,
,., . ·Sj.IJ,1ilar relations are found with compound discriminative stimuli (SD s) in the
;of?perant stimulus control. Both respondent and operant phenomena are
:~~discussed in terms of attention; for example, it may be said that the
1ii'\lgesnot attend equally to all components of the compound stimulus.
~is is not an explanation. It is simply another name for the facts, but it
\".lJe<:;ome an explanatory fiction.
" :',
 Outline of the Science ofBehavior 9

10. Miscellaneous
a. Emotion
Although much of the topic of emotion is concerned with the operant relations
involving establishing operations (see IVC6 below) to some extent emotion refers
to smooth muscle and gland responses, and thus emotional responses are
respondently conditionable. The Staats and Staats experiment (1958) is an
interesting example of this conceptualization, especially in that it does not involve
any actual measurement of smooth muscle and gland responses.
. b. Conditioned perceptual responses
Skinner (1953, pp. 266-270) makes considerable imerpretive use of a' type of
response analogy that he calls a perceptual response. He refers to seeing (hearing,
etc.) something as a result of light waves from that thing affecting the visual
receptors as an unconditioned perceptual response. The light waves are the
unconditioned stimulus for that unconditioned response. Thus seeing the breaking
surf as a result of light waves from such surf is unconditioned seeing. Then,
using the respondent conditioning paradigm, he suggests that any stimillus that is
paired with such an unconditioned stimulus (for the present example, the sound of
,the breaking surf) will come to elicit a similar perceptual response, a conditioned
perceptual response. Although indirect, there is considerable evidence for such an
interpretation.
c. Drug addiction and overdosing
Some drugs (e.g., amphetamines, opiates and cocaine), in addition to producing
the effects for which they are known, seem to function as unconditioned stimuli
for bodily effects that counteract or are in opposition to the effect for which the
drug is being taken. Neutral stimuli, such as the visual characteristics of the place
where the drug is typically taken, by being paired with this unconditioned
stimulus, become conditioned stimuli for a similar counteracting effect. This
conditioning may then result in the necessity for increased dosage in that place
because the previously neutral stimuli come to elicit an increasingly strong fonn of
counteraction. When this increased dosage is then taken in a novel setting, a
setting in which the conditioned stimuli for the counteracting effect are not as
strong, it may constitute a serious fonn of overdosage.
11. Evocative versus function-altering relations
The US functions to elicit the UR (the evocative effect of the US) and also to alter
the future function of a neutral stimulus (the function-altering effect of the US); that
is, it conditions the organism so that the neutral stimulus becomes a CS. It is
convenient to have separate symbols for each of these functions; thus, UE stands for
unconditioned elicitor (the evocative effect) and UC for unconditioned conditioner
(the function-altering effect). Similarly, in higher order conditioning, the CS
functions as a conditioned elicitor (CE) in evoking a response somewhat similar to
that evoked by the US with which it was correlated, and as a conditioned conditioner
(CC) in altering the future function of a neutral stimulus.
C. Operant functional relations
1. Type of effector '
Operant conditioning is generally successful with striped muscles. Whether smooth
muscle or gland responses can be operantly conditioned (Le., can be affected by
consequences), is still unclear. Studies have shown such conditioning, but ruling out
operant conditioning of striped muscle responses that in turn elicit the smooth muscle
or gland responses has been difficult.
 10 Outline of the Science ofBehavior

2. Experimental s e t t i n g s . .
Lever pressing, wheel turning, chain pulling; and so on for food or water
reinforcement in the rat, monkey, and many other species; key pecking in the pigeon;
many other kinds of operanda with a wide variety of species.
3. Reinforcement and punishment definitions and basic function-altering
relation
(a) A stimulus change occurs (b) immediatelyl after a response, and (c) in the future
the frequency of that type of response is increased (d) in similar stimulus conditions,
(e) and when the relevant establishing operation (EO) is in effect, (f) but not if for
other reasons such as elicitation. (positive reinforcement =presentation; negative .
reinforcement = removal or tennination.)
For the definition of punishment, change increased to decreased and change
elicitation to fatigue, extinction, habituation, etc. The possible relevance of EOs to
the punishment effect is more complex than with reinforcement Painful stimulation
seems not to require an EO. For the type of punishment thatconsists of the removal
of a reinforcer, such removal will not function as punishment unless the EO relevant
to that reinforcer is in effect. And it is to be expected, although there is no laboratory
data on·this topic, that the future suppressive effect of punishment consisting of
reinforcer removal will be seen only when the EO relevant to that reinforcer is in
effect.
The fact that the change in the organism that is brought about by reinforcement or
punishment in operant conditioning lasts over time results in the concept of memory
being invoked as a fictional explanation for its persistence over time. The rat is said
to be pressing the lever today because it remembers that lever pressing produced food
yesterday when the rat was first operantly conditioned. As with respondent
conditioning and extinction, the essence of a function-altering effect is its persistence
over time, and an instance of such persistence requires no explanation beyond
mention of the the relevant history and citation of the descriptive principle of operant
conditioning. Again, as with respondent conditioning, the neurochemical changes
that are produced by the conditioning procedure will someday be better understood
and will constitute another type of explanation for the persistence.
4. Other independent variables
The effectiveness of reinforcement is dependent upon the establishing operation as
described below in more detail (see NC6), but it is also directly related to the
magnitude and quality of reinforcement and indirectly to the delay of reinforcement
and the response effort needed to obtain the reinforcement
S. Dependent variable
When behavior can be repeated during a short period of time, as with a rat pressing a
lever on some schedule of intermittent reinforcement, frequency of responding refers
to the number of responses per unit time. A water-deprived rat with a good history of
intermittent water reinforcement for lever pressing might be expected to press several
times per minute. However, if each response changes the situation so that further
responding is no longer appropriate (as with continuous reinforcement, where

1Earlyw()l:!< on delay of reinforcement seemed to imply that consequences delayed by more than 10 or 20'
seconds would generally be ineffective. More recent research suggests that considerably longer delays may
.still result in behavior maintenance and even acquisition. The variables relevant to the effectiveness of long
~elayed c~msequen':e5 are only beginning to be determined. It can probably be safely said that the less that
IS happenIng ~th m term~ of stimuli and responses) during the delay interval, the more likely a delayed
com;~ue~ce 18 to be effective. It may ~ be that some specific kinds of responses can be affected by some
~dic kin~s of consequences afte~ qUite long delays, somewhat analogous to the taste aversion phenomena
m the tyspondellt case. I? ~eneral, It does seem clear that the longer the delay the less effective the
consequence, but upper limits cannot at present be stated with confidence.
 Outline of the Science ofBehavior 11

reinforcer consumption lasts longer than the typical response duration, or in a

discrete-trials procedure) responses per unit "time is not as easily obtained as a
dependent variable. Much human behavior is of this sort in that 'it can only occur
once on each appropriate occasion, because the occurrence of the response changes
the situation so that it is no longer appropriate. For example, when a person asks
someone what time it is and is told the time, it is not appropriate to ask again, because
the person asking is no longer reinforceable by information about the time. The
person who asked is now said to know what time it is and therefore has no further
need to know. " ,
Frequency can still be measured as a string of latencies as follows: Start a timer
timing as soon as the appropriate condition begins, and stop it when the response
occurs. This measures the time between the onset of a condition in which a response
wonld be appropriate and the occurrence of the response. The accumulation of these
measures over many appropriate conditions will result in a total time and a total
number of responses, and the ordinary frequency measure-responses per unit
time-can be calculated. (A problem arises for those occasions on which no
response occurs. Whatis the time measure for those occasions? Typically one
arbitrarily agrees to an upper limit for the duration of the condition and uses this value
for those conditions, assuming that a response would have occurred after that upper
time limit)
Another commonly used dependent variable in such situations is the frequency of
response occurrences relative to the frequency of occurrence of the conditions
appropriate for such a response. In other words, this is the proportion or percentage
of appropriate conditions in which a response occurred, sometimes referred to as the
relative frequency of responding. (Responses per unit time can then be called
absolute frequency.) For such conditions, the dependent variable is sometimes
referred to as the probability or likelihood of the response. However, the term
probability and its various synonyms is a source of possible confusion. If all that is
meant by probability or likelihood is the observed proportion or percentage of those
circumstances appropriate for a response when a response actually occurred (in other
words, the observed relative frequency), then it is better to use relative frequency
measured as percentage or proportion, because this term refers to an actual dependent
variable, something that can be, at least in principle, directly observed.
Probability more often appears in the role of a dispositional concept. It refers to
what might be expected to occur upon certain conditions. It is like the term brittle
when applied to some material like glass, It does not refer to any feature of the
material that conld be observed at the time of the description, but rather to what would
happen if the material were struck or distorted in some way-it would shatter into
many pieces. A description of brittleness is typically provided, not from close
observation of the material, but from knowledge of a history of other examples of the
same material shattering on being struck or distorted. The same is true when, for a
particular organism, a comment is made about the probability of a certain type of
behavior. It is a statement about what would be expected to happen if other
conditions were realized, and it is based on a knowledge of the organism's history
and the effects of similar histories on other similar organisms. Ryle (1949) analyzes
many such dispositional concepts in psychology, both in terms of their usefulness
and the ease with which they can be misinteI]lreted.
Such terms lend themselves very easily to inteI]Jretation as some kind of
underlying characteristic of the organism that is responsible for the absolute or
relative frequency of responding. Both absolute and relative frequency are sometimes
spoken of as a basis for inferring the response probability, or as the manifestation of
the underlying response probability. Such verbal practices encourage explanatory
fictions and the invention of cognitive or mentalistic entities, and must be engaged in
12 Outline ofthe Science ofBehavior

very carefully, or avoided altogether. All that is usually necessary for a behavioral
analysis is to describe the functional relations among environmental histories, current
conditions, and absolute or relative frequencies of responding. .
The term response strength like probability, may refer to the observed proportion
or percentage of those circumstances appropriate for a response when a response
actually occurred, in which case it seems bener to use the term relative frequency ,
because this term refers to an actual dependent variable. Response strength,like
probability, is often used as a dispositional concept Strength is sometimes criticized
as being just another name for some kind of inferred internal entity, in which case it is
common to explain that nothing more is meant than probability of response
occurrence. A strong response is simply one that has a high probability of occurrence
under a given set of circumstances. Because the value of our verbal practices
regarding probability is questionable, it seems to be the better part of caution to avoid
strength as well as probability.
6. Establishing operations (EOs)
a. Two defining effects
i. Reinforcer-establishing effect
An EO momentarily alters the reinforcing effectiveness of some event or stimulus
'(food deprivation momentarily increases the effectiveness of f~ as
reinforcement; food satiation momentarily decreases the effectiveness of food as
reinforcement).
ii. Evocative effect
An EO momentarily alters the frequency of any behavior that has been reinforced
by the event or stimulus whose reinforcing effectiveness is altered by the same
establishing operation (food deprivation momentarily increases the frequency of
any behavior that has been reinforced by food; food satiation momentarily
decreases the frequency of any behavior that has been reinforced by food).
Some consider it preferable to state this effect as momentarily altering the
evocative effect of relevant SDs .
b. Unconditioned establishing operations (UEOs)
The main ones are deprivation and satiation with respect to food, water, activity,
sleep, oxygen; temperature too high or too low; variables relevant to sexual
reinforcement (deprivation but others as well); and painful stimulation. These
UEOs each produce a corresponding form of unconditioned reinforcement, but the
UEO should not be confused with the unconditioned reinforcement. For example,
food is unconditioned reinforcement given food deprivation as a UEO, but food
deprivation is not a form of unconditioned reinforcement nor is food a UEO.
The notion that some organisms are motivated l7y curiosity, or that they have
an exploratory drive, may be equivalent to proposing that stimulus change
functions as a form of unconditioned reinforcement, with the establishing
operation possibly consisting of stimulus-change deprivation. Similarly, the
notion that humans are motivated to be competent may imply that stimulus changes
that are sigris of competence (whatever they might consist of) function as
unconditioned reinforcers, possibly not related to any unique EO, but under the
control of any EO (in other words, under any condition in which something else
functions as effective reinforcement). An alternative interpretation of the
phenomena supposed to be evidence for this latter type of motivation is in terms of
generalized conditioned reinforcement, described in IVC8b below.
c. Conditioned establishing operations (CEOs)
i. Surrogate CEO
A stimulus that has been correlated with another EO may become a CEO itself in
evoking the behavior that is typically evoked by that other EO and in altering the
 Outline of the Science ofBehavior 13

reinforcing effectiveness of the events whose reinforcing effectiveness is

typically altered by that other EO. .
ii. Reflexive CEO
A conditioned aversive stimulus, in that it establishes its own removal as a fonn
of reinforcement and evokes the behavior that has accomplished such removal, is
functioning as a CEO. Probably a similar effect, but in the opposite direction,
can be seen with any stimulus that has been systematically correlated with some
fonn of reinforcement.
iii. Transitive CEO
The stimulus upon which conditional conditioned reinforcement (or punishment)
depends increases the reinforcing effectiveness of that type of conditioned
reinforcement and evokes the behavior that has produced such·conditioned
reinforcement, and is thus functioning as a CEO. In a common human situation,
a stimulus functioning as SD evokes a particular fonn of behavior, but the
behavior cannot be executed-is blocked-because some other object or
stimulus condition is absent. The occurrence of the SD for the blocked behavior
results in an increase in the reinforcing effectiveness of the missing object or
stimulus condition, evokes the behavior that has obtained the object or stimulus
condition, and in this capacity functions as CEO. .,
d. Emotional EOs or Emotional Predispositions
B. F. Skinner's concept of the emotional predisposition identifies an operant
aspect of emotion, very milch like the basic motivational functional relation, as a
fonn of establishing operation (although he did not use this tenn. Consider the
emotion of anger. The causal environmental eve!lt is some kind of worsening in
the presence ofanother organism-painful stimulation, interference with ongoing
reinforced behavior, removal of reinforcers that are present, etc. For some
organisms, this event seems to function as a 000 in causing signs of damage or
discomfort on the part of the other organism to function as reinforcement, and in
evoking any behavior that has been reinfQrced by such signs in the past. Whether
such effects are related to UEOs in humans, or primarily to different kinds of
CEOs, or to OOo-cEO mixtures is presently unclear. 2
7. Stimulus control
a. Stimulus change decrement with respect to ambient stimuli
Mter an operant function-altering effect, the changed function is seen at its
maximum value when the stimulus conditions are exactly the same as during the
function-altering operation. This holds for operant conditioning, operant
extinction, and the operant weakening effect of punishment.
b. Discriminative stimulus
SD and S6: A discriminative stimulus is a stimulus condition that (a) alters the
momentary frequency of a type of behavior (b) because of a correlation between
the presence/absence of that stimulus and the relation between that type of behavior
and an effective behavioral consequence.
c. Stimulus control procedures
Multiple schedule, matching to sample, and many others with no special names.
d. Operant discrimination based on punishment
The defmition of the SD and S6 given above, in referring to an effective behavioral
consequence, includes the development of stimulus control by punishment

2The essential similarity of emotional and motivational functional relations was well developed in his 1938
book, The Behavior ofOrganisms, but the term emotional predisposition as well as a more extensive
analysis appears in Scieru:e and Human Behavior, pp. 162--170.
14 Outline of the Science ofBehavior

procedures, but because such stimulus control is not usually mentioned in this
context an elaboration will be provided here. If behavior is ordinariIy reinforced in
the presence of a stimulus condition that serves as an SO for this behavior, but is
punished when another stimulus is present along with this So, then this second
stimulus functions to decrease the frequency of the behavior.. There is no
conventional term for this kind of stimulus control, but it could reasonably be
called a warning or threat So. .
e. Relevance of establishing operations
Control by a discriminative stimulus is seen only when the relevant EO is in effect.
f. Generalization and stimulus-change decrement re the SD
Any change in the SO from what it was during its development will result in its
having a smaller evocative effect than if the change had not occurred. The decrease
resulting from this change is called stimulus-change decrement. The fact that the
changed stimulus still has some evocative effect, which it would not have had
before development of the original So, is referred to as stimulus generalization. In
a quantitative sense, stimulus generalization and stimulus-change decrement are the
.complements
,. of one another.
g. Stimulus class membership and stimulus equivalence
Two stimuli are said to be members of the same s.timulus class when it is possible
to change the behavioral function of one and find that the other has lj.!ready been at
least piutially changed in the same way. Physically similar stimuli are usually
members of the same class in this sense, without any learning history being
required. However, after various kinds ofleaming histories (such as repeated
instances of temporal contiguity or pairing), stimuli that have no physical
resemblance may become members of the same stimulus class. Recently a
procedure referred to as equivalence training [Sidman, M. (1971). Reading and
auditory-visual equivalences. Journal of Speech and Hearing Research, 14, 5-13]
has been found to be especially effective in developing stimulus class membership.
Two physiCally dissimilar stimuli are each, in separate training situations, made the
correct comparison stimulus to pick in the presence of the same sample stimulus.
If one is then made either the comparison or the sample in further training
involving different sets of stimuli, the other will to some extent function similarly.
This procedure results in a conditional discrimination, as described below.
h. Attention and blocking with respect to compound stimuli
With compound discriminative stinmIi (e.g., simultaneous presentation of a tone
and a light) as the So condition, the extent to which anyone component of the
compound acquires evocative control over the response is not simply a matter of
the extensiveness of discrimination.training with the compound stimulus. For
example, if one of the components has previously been developed as an So by
itself for the same response, it may block the acquisition of evocative control by
the other component, irrespective of discrimination training with the compound
stimulus. Similar relations are found with compound CSs in the case of
respondent stimulus control (see IVBge above). Both respondent and operant
phenomena are sometimes discussed in terms of attention; for example, it is said
., that the organism does not attend equally to all components of the compound
stimu}us. This is not an explanation but rather is simply another name for the
facts, but can easily become an explanatory fiction.
i. Conditional discrimination
When the nature or extent of operant control by a stimulus condition depends upon
some other stimulus condition, the relation is'feferred to as a conditional
discrimination. The most common form of conditional discrimination is one in
 Outline of the Science ofBehavior 15

which the correct stimulus to select (by pecking it in the case of the pigeon as
subject, or touching or pressing it in the case of rat, monkey, or human) depends
on the characteristics of another stimulus display. For example, when the sample
stimulus displayed is a triangle, then pecking the red comparison stimulus is
correct, but when the sample stimulus is a circle, pecking the green comparison is .
correct Thus, the control of the pecking response by the color of the comparison
stimulus is conditional upon the shape of the sample stimulus.
8. Conditioned reinforcement and punishment
a. Procedure
A neutral stimulus is paired with reinforcement or punishment: Stimulus
presence/absence is correlated with reinforcement or punishment presence/absence.
But stimulus should be taken to mean stimulus change rather than to refer to a
static stimulus condition (see ill above).
b. Generalized conditioned reinforcement or punishment
A neutral stimulus is paired with several different kinds of reinforcement or
punishment: Stimulus presence/absence is correlated with several different kinds
of reinforcement or punishment presence/absence.
c. Relevance of establishing operation
These stimuli function as conditioned reinforcement or conditioned punishment
only when the EO relevant to the reinforcement or punishment with which they
were correlated is in effect.
d. Chaining
Rl produces S, in the presence of which R2 receives some form of effective
reinforcement. Under these conditions S functions as sr (conditioned
reinforcement) for Rl and as SD for R2
9. Operant weakening operations
a. Extinction
This term refers to the procedure whereby a response occurs without its
reinforcement, or to the weakening effect of this procedure. It is not a general term
for any procedure that weakens or for any weakening effect. In particular, it is
quite inappropriate to refer to "extinguishing a response by punishing it."
b. Two ways to weaken the evocative strength of an SD
Extinguish the response in the presence of the stimulus or reinforce the response in
the absence of the stimulus. .,
c. Two ways to weaken the effectiveness of a conditioned reinforcer or
punisher
Let the stimulus occur but without SR or sP or provide SR or sP in the absence of
the stimulus.
d. Punishment
(See the definition above with the definition of reinforcement-IVC3)
i. Unconditioned and conditioned punishment
Unconditioned punishment may consist of painful stimulation (by which is
meant an increase in painful stimulation-see ill above), some tastes, some
odors, muscular effort, possibly any very intense form of stimulation, loss of
bodily support, physical restraint.
Common conditioned punishment consists of any stimulus change that has
been systematically related to any form of worsening with respect to
reinforcement and punishment, such as social disapproval, removal of
privileges, fines, and so on.
16 Outline of lhe Science ofBehavior

ii. Other variables

The suppressive effectiveness of punishment is directly related to the intensity of
the punishing stimulus, and is indirectly related to the delay of punishment
iii. Recovery from punishment
Somewhat analogously to operant extinction with respect to reinforcement, the
suppressive effect of punishment decreases when response occurrences are not
followed by punishment. The situation is complicated by the fact that the
response that was punished must either be currently reinforced or must have .
been reinforced prior to the punishment, otherwise there would be no behavior to
punish; the effects of discontinuing the punishment will then depend on the past
and current reinforcement contingency.
e. Conditioned suppression
The suppressive effect on food- or water-reinforced behavior by a stimulus that
has been paired with a painful stimulus.
10. Escape and avoidance .
Although escape behavior and avoidance behavior do not involve any principles
other than those already considered under the headings of reinforcement, establishing
operations (especially the reflexive CEO), and conditioned reinforcement, these topics
are'often considered separately under the heading of aversive control. Escape refers
to behavior that is reinforced by the tennination of an unconditioned aversive
stimulus, usually painful stimulation. Avoidance is behavior that is reinforced by the
termination of a stimulus that has been systematically related to the onset of an
unconditioned aversive stimulus (the so-called warning stimulus), thus by
condilioned reinforcement. Escape is evoked by the onset of the unconditioned
aversive stimulus functioning as a UEO; avoidance is evoked by the onset of the
warning stimulus functioning as a reflexive CEO. Avoidance without a warning
stimulus (nondiscriminative or Sidman avoidance) requires a more complex
explanation and raises the issue of whether molar or molecular explanations are more
appropriate.
11. Schedules of reinforcement
a. Types
Time based, response based, combinations, stimulus control procedures, and
others.
b. Reasons why schedules of reinforcement are important
(a) as a standard research tool, (b) as a research area itself, (c) as an addition to the
technology of behavioral control, (d) as an important theoretical enrichment of the
independent variable, (e) as a clarification of the topic of motivation.
12. Miscellaneous
a. Another type of memory
The persistence over time of behavior changes produced by function-altering
effects requires no special consideration, as might be implied by referring to such
persistence as a form of memory. Some special analysis may be required,
however, when a stimulus chaJ:)ge having only evocative effects is presented and is
then removed in such a way that the evocative effect cannot occur until after a '
,period of time, and then does occur appropriately. The most common example of
'.such a procedure is delayed matching to sample, where the sample is presented,
. then removed, and after a time period (e.g.,IO seconds) the comparison stimuli are
. presented and the one selected is appropriate to the no-longer-present sample
stimulus. Most explanations consist of inferring that some form of subtle
function-altering effect takes place at the time the evocative stimulus is presented
 Outline of the Science ofBehavior 17

and then removed, even though the stimulus is not ordinarily thought of as having
any function-altering effects. . .
b. Adjunctive behavior
Sometimes reinforcing or in some other way increasing the frequency of one type
of behavior leads to an increase in some other behavior that has not been directly .
reinforced. Thus intermittent presentation offood to a food-deprived organism
will lead to corresponding intermittent food consumption, and for some organisms
if water is available there will be a brief period of drinking inunediately after each
occurrence of eating. The eating is evoked by the food presentations, and me
drinking is called adjunctive behavior in that it is attached to the other behavior in a
subordinate manner. Much adjunctive behavior seems closely related to the
phylogenic organization of behavioral sequences.
c. Skinner's operant perceptual response
As with respondently conditioned perceptual responses (see IVBI0b above),
Skinner (1953, pp. 27--275; 1974, pp. 91-95) proposes that some perceptual
responses are in a sense analogous to operantly conditioned behavior, in that they
occur when the relevant EO comes into effect. Thus, food deprivation leads to an
increase in the frequency of all the behavior that has been reinforced with food,
and because under such conditions the sight of food functions a reinforcement,
seeing food can occur even in the absence of food.
13. Evocative versus function-altering relations
SD, VEO, and CEO refer to evocative effects; SR, SP, sr, and SP refer to
function-altering effects; the development of conditioned reinforcers, conditioned
punishers, and CEOs by correlating stimuli is a type of function-altering effect that is
respondent in procedure (pairing or correlating stimuli), but is operant in terms of
what is developed.
V. What a Specific Organism Learns
With humans, it is clear that there are important differences in behavior from one human to
another. These differences are the essential subject matter of much of traditional
psychology. The principles oflearning and motivation described above are characteristic of
all humans, and although quantitative differences in these principles may account for some
of the differences between the behavior of different persons, much of this individual
variability is undoubtedly due to differences in what each person learns. An understanding
of the principles of learning and motivation, no matter how complete, will tell us· little about
the significant personal differences that are of such practical importance. We are all
susceptible to respondent and operant conditioning, but to understand individual
differences it is further necessary to know for each of us what stimuli have become
conditioned stimuli, what behaviors have been increased in frequency by what kinds of
reinforcers, and so on.
Those behavioral psychologists who refer to their field as behavior analysis have not
been much concerned with this aspect of the science of behavior, but rather have aimed
primarily at discovering principles that are general to all humans or to all organisms. This
should not by any means imply that such differences are beyond a behavioral analysis, or
belong to some other field. The listing below is far from complete, but should suggest the
direction that is being taken by a behavioral analysis of personality, individual differences,
and so forth. These topics should more properly be arranged as a smaller number of
categories, each containing several subcategories,but a simple listing will probably suffice a
this time. [Probably more than any other behaviorist A. W. Staats (1963, 1968, 1971,
1975) has been at the forefront of empirical and theoretical work in this area.]
1. Basic conditioned reinforcers and punishers
2. Basic sensory-motor repertoires
18 Outline ofthe Science ofBehavior

3. Elementary social repertoires

4. The elementary verbal repertoire
Examples: rnands, tacts, echoic behavior, textual behavior and inttaverbal behavior.
5. Verbal behavior under the control of private stimuli
6. Intellectual repertoires
Examples: reading, writing, math, etc.
7. More complex sensory-motor repertoires
Examples: musical behaviOr, athletic behavior, artistic behavior, etc.
8. Social and interpersonal skills
9. Interacting re.pertoires within the individual
Examples: self-control, problem solving, a type of memory related to problem
solving [Skinner's (1957, pp. 142-143) reacting to the past rather than simply
profiting from the past and (pp. 357-367) the capacity to be directly affected by
function-altering rules or instruction.
10. Repertoires related to group practices
Here behavior analysis blends with social psychology.
11. R!!pertoires related to cultural practices
Here behavior analysis blends with cultural anthropology and sociology.
 •
2
HISTORICAL ANTECEDENTS OF B EHAVIORANALYSIS
Purpose and Limitations
Students of behavior analysis who know
little of its history will be less than
optimally effective in acquiring new
knowledge. They will also be unaware
of relations among various parts of their
professional and scientific repertoires. In
short, it is important to know where we
came from. The purpose of the present
document is to provide an easily learned
introduction to the history of behavior
analysis. However, there are several
limitations to my approach to this topic.
In the first place, the infonnation
below provides historical perspective
only on the unique aspects of the area
called behavior analysis. The
professional repertoires of many behavior
analysts also have more general historical
antecedents, such as those of
experimental psychology, American
psychology, or fields other than
psychology.
A second limitation is the obvious
one, that any version of history as
abbreviated as this is bound to be grossly
oversimplified. It can be argued,
however, that an oversimplified version,
so long as it is not clearly misleading, is
much better than no historical knowledge
at all, which seems to be the most
common alternative.
Finally, it is quite clear that others
who have thought much about the history
of our field might well organize historical
infonnation differently, placing greater
emphasis on some contributions than I
have, and de-emphasizing some that have
c.;
seemed to me especially important. I like
to think, however, that there would be
little serious disagreement among us, and
that the present document will be useful
even though most users will have to make
some modifications. .
An Unusual History
As behavior analysts, our historical
antecedents are somewhat unusual in that
contributions to the science of behavior
occurring priorto 1938, the date of
publication ofB. F. Skinner's The
Behavior ofOrganisms, have affected us
mainly through their influence on
Skinner. A reasonable representation has
a number of historical factors relevant to
Skinner's intellectual repertoire (sort of
like the top half of an hourglass),
Skinner's contributions then play the
major role in the subsequent development
of experimental and applied beh.avior
analysis (like the bottom half of the
hourglass). Of course many scholars and
scientists in addition to those considered
below played important roles in the
distribution, interpretation and refinement
. of Skinner's contributions, and also made
original (Le., non-Skinnerian)
contributions to the field of behavior
analysis. The field is not solely B.F.
Skinner, but his intellectual repertoire
played a very important role in
subsequent developments. An
abbreviated version of the history of
behavior analysis can be portrayed as an
effort to understand the origins of
Skinner's behavior with respect to the
science of behavior, and then to trace the
20 Historical Antecedents ofBehavior Analysis
effect of his achievements, in
combination with other factors, on the
development of the field.
Because of its considerable influence
on the development of behavior analysis,
Principles ofPsychology, the
introductory text by Keller and
Schoenfeld (as well as other aspects of
the program.at Columbia College),
should be shown as another major focal
point. Its antecedents include the
;.
i·
methods and results of Skinner's
Behavior ofOrganisms (1938) and other
behaVioral aspects of experimental
psychology; it also influenced many of
the subsequent contributions and events.
Antecedents, other than Skinner's
Behavior of Organisms, and the
consequents of this remarkable book are
not indicated on the chart, but only
because the diagram is already too
complex to be easily remembered.
 Behavior Analysis History Chart

The philosophical works of Francis Bacon (1561-1626) . 'Darwin (1859)

Sechenov (1863)
. The Origin ofSpecies
Reflexes Of\Brain
1870 ~ \
1880 Mach (1883) .
1890 cience ofMechanics
Thorndike (1898) Pavlov
1900 Cat in puzzle box
A,series of
.ssertation, Law of Effect
The published
Organis lectures on
1910
asa Watson (1913) - - - conditioned
Whole "Psychology as 'the reflexes
(1916) behaviorist views it" (1908-1928)
1920
Conditioned
Reflexes
Russell 1927) (1927)
Philoso hy
1930
Crozier (around 19 \
Dept of Physiology

~Sk:inner(193 )
Behavior of Organisms
1940

Keller and Schoenfeld (1950)

1950 . Principles of Psychalogy

I Skinner (1953)
Science tmd Human Behavior
.,
Skinner (1957) Ferster and Skinner (1957)
1 Verbal Behavior
________________
Journal ofthe Experimental Analysis Schedules ofReinforcement
E[§~~~~j!§~L!~~~ _
J 1960

Many important books and articles were published during this period-5idman's Tactics of
Scientific Research (1960), Ullman and Krasner's Case Studies in Behavior Modification, (1965).
There were also a number of important organizational events (e.g., Division 25 of APA was
formed, behavioral academic programs got started at University of Washington (Bijou), Arizona
State University (Staats), Southern Illinois University (Sulzer, Azrin,); Kansas University (Baer,
Wolf, Risley etc.) and others.
Journal ofApplied Behavior Analysis (JABA) 1968
1970
Behaviorism 1972
Association for Behavior
Journal ofOrganizational Behavior Analysis (ABA) 1975
Management (JOBM) 1977
The Behavior Analyst 1978
The Analvsi.' ofVerbal Behavior 1982
 22 Historical Antecedents ofBehavior Analysis

Principal Contributors
Francis Bacon l (1561-1626)
In Europe during the Middle Ages, science (philosophy) and technology (crafts) had almost
no points of contact.
It fell within the province of the philosopher to explain the phenomena of nature; their use for
practical ends was left to the craftsman. The philosopher was much concerned with books and opinions,
and but little with things; he displayed admirable intellectual ingenuity in framing his explanations of the
natural world in general terms while lai'gely neglecting their application in detail. The craftsman, on the
other hand, knew little or nothing beyoiKl trade methods and processes which he followed because they had
been handed down to him and because they brought the results he sought; he was altogether innocent of
theories to explain his actions. Only in the seventeenth century...was it realized--and even then by few-
that science and the crnfts were alike concerned with natural phenomena and could aid each other. (Singer,
Holmyard, Hall & Williams, 1957, p. v)
Francis Bacon, writing at the beginning of the 17th century, was one of those few.
He was a philosopher, but strongly opposed the separation of science and technology, and
seized every opportunity to exalt technological innovation, not only for the power of its concrete
products to relieve human suffering but, more important, as a new model of the knowing process. Bacon's
epistemology elevated the role of the artisan above that of the theoretician, and he repeatedly contrasted the
steady piecemeal progress exhibited by the technical arts (what we would ca11technology) with what he saw
as the stagnant and fruitless speculations of the Aristotelian tradition.
The kernel of Bacon's new conception of knowledge is that we know best what we can ourselves
make....To understand nature...is to be able to bend the ways of nature to the production of novel
effects. (Smith, 1992, pp. 216-217)
Bacon's view of science consisted of two related notions that were in sharp contrast
with the views of most of his philosophical contemporaries. He argued that the most
effective scientific methodology was experimentation, an active manipulative intervention in
natural processes (as contrasted with passive contemplation), and that the main criterion of
scientific truth was the ability to control nature. His emphasis on science as the power to
produce effects was also relevant to his "reforrirlst" view, that "a proper understanding of
the value of the technical arts and the acceptance of the manual arts as a new model of
knowing would bring about unprecedented changes in society, leading steadily to the
satisfaction of all human wants and the emergence of a harmonious society (Smith, 192, p.
219). His Utopian fable, The New Atlantis, is an account of the government and manners
of the inhabitants of a fictional island "where productive knowledge is given full respect;
nature is experimented with and remade with an eye toward human benefit. There are
synthetic dyes, ...submarines, methods of flight, telephones,...methods of transplanting
organs, and artificial means of extending the life span" (Smith, 1992, p. 219).
?
There is clearly a close relation between Baconian thought and the unique
methodological-epistemological view ofB. F. Skinner, as well as Bacon's reformist
approach. By Skinner's time, of course, science was no longer primarily contemplative,
but within psychology Skinner's approach is still exceptional for its advocacy of an
interventionist methodology and for its emphasis on control as the primary criterion of
scientific validity.. A clear statement of this latter point occurs early in Verbal Behavior:
'.The "understanding" of verbal behavior is something more than the use of a consistent vocabulary
with which sPecific instances may be described. It is not to be confused with the confirmation of any set of
theoreti&! principles. The criteria are more demanding than that. The extent to which we understand verbal
behavior in a "causal" analysis is to be assessed from the extent to which we can predict the occurrence of

1Lawrence D. Smith, historian of science, has Ween a special interest in the origins of B. F. Skinner's
distinctive approach to scientific methodology and epistemology. In particular he has shown a close
relation between Skinner's views and the works of Francis Bacon (Smith, 1992), and also Ernst Mach's The
Science ofMechanics (Smith, 1986). My treatment of Bacon's influence and of Mach's is taken almost
entirely from Smith.

I
 Historical Antecedents ofBehavior Analysis 23

specific instances and, eventually, from the extent to which we can produce or control such behavior by
altering the conditions under which it occurs. (Skinner, 1957, p. 3)
While in the eighth grade, Skinner read biographies of Bacon, summaries of his
philosophical position, and several of his works, including Essays, New Organon,
Advancement ofLearning, and the utopian novel, The New Atlantis (Smith, 1992, p. 217).
In his autobiography Skinner remarks that "this was stretching my abilities pretty far, and I
doubt whether I got much out of it at the time, but Francis Bacon was to serve me in more
serious pursuits later on". (Skinner, 1974, p. 129).
There were undoubtedly many other influences on Skinner's approach to sciente,
especially the views of Ernst Mach as will be discussed later. Still his early contact with
the works of Bacon may well have detennined his receptivity to later, more sophisticated
versions of some of those early Baconian themes.
With respect to the design of a culture on the basis of the science of behavior, Skinner
credits Bacon as follows:
The New Atlantis was the first utopia I real!. A better world was possible, but it would not come
about by accident. It must be planned and built with the help of science. Salomon's House inThe New
Atlantis was the model of the Royal Society, and the American Philosophical Society, which Benjamin
Franklin founded on the model·of the Royal Society, was dedicated to "the promotion of useful knowledge."
It was the theme of the Enlightenment and, very early, of my own intellectual life.
By its very nature an experimental analysis of behavior spawns a technology because it points to
conditions which can be changed to change behavior. I said as much in my own New Atlantis, Walden
Two. (1983, p. 412) .
Walden Two (1948) was a novel, but the later Beyond Freedom and Dignity (1971)
was a detailed argument for replacing outmoded and essentially harmful cultural practices
with practices based on a science of behavior, a direct and forceful example of the Baconian
refonnist theme.
Charles Darwin (1809-1882) The Origin of Species (1859)
Darwin indirectly influenced Skinner by establishing the continuity of the species. In a
sense he dethroned our species from the center of God's creation, just as Copernicus
dethroned the earth from the center of the universe. The theory of evolution by natural
selection made human behavior an appropriate subject matter for the natural sciences. It led
to the development of the field called comparative psychology and inspired the early
students of comparative psychology (Loeb, Jennings, etc.). With continuity of the species,
behavior ofnonhumans became more relevant to human behavior, and vice versa. This
continuity led to much effort to frod evidence of human consciousness and mental
processes in the behavior of nonhumans.. This effort was not the intellectual movement,
however, that influenced Skinner, but rather the somewhat later efforts to explain
nonhuman behavior without such concepts (by Loeb, for example).
Darwin's concept of natural selection had a more direct influence on Skinner's more
recent work, as can be seen from the following quotation2:
Human behavior is the joint product of (i) contingencies of survival responsible for natoral selection,
and (ii) contingencies of reinforcement responsible for the repertoires of individuals, including (iii) the
special contingencies maintained by an evolved social environment. Selection by consequences is a causal
mode found only in living things, or in machines made by living things. I~ was first recognized in natoral
selection: Reproduction, a first consequence, led to the evolution of cells, organs, and organisms
reproducing themselves under increasingly diverse conditions. The behavior functioned well, however, only
under conditions similar to those under which it was selected.
Reproduction under a wider range of consequences became possible with the evolution of processes
through which organisms acquired behavior appropriate to novel environments. One of these, operant

2The quote is the abstract for the article "Selection by Consequences" as written to accompany a reprinting
in The Behavioral and Brain Sciences. (1984) 7,477-481. The original was Skinner, 1981.
 24 Historical Antecedents ofBehavior Analysis

conditioning, is the second kind of selection by consequences: New responses could be strengthened by
events which followed them. When the selecting consequences are the same, operant conditioning and
natural selection work together redundantly. But because a species which quickly acquires behavior
appropriate to an environment has less need for an innate repertoire, operant conditioning could replace as
well as supplement the natural selection of behavior.
Social behavior is within easy range of natural selection, because other members are one of the most
stable features of the environment of a species. The human species presumablY)lecame more social when
its vocal musculature came WIder operant control. Verbal behavior greatly increased the importance of a
third kind of selection by consequences, the evolution of social environments or cultures. The effect on the
group, and not the reinforcing consequences for individual members, is reSPDnsible for the evolution of
culture (Skinner, 1984). .
Selection has here become the basis for a comprehensive integration of the biological,
behavioral, and social sciences.
Ivan M. Sechenov (1829-1905) Reflexes of the Brain (1863)
At a time when physiologists were generally quite dualistic, especially with respect to
human thought and consciousness (believing these to be mental rather than physical
events), Sechenov proposed that all aspects of consciousness, states of mind, etc., in
humans a~ well as nonhumans, were reflexes. This was a carefully worked out attempt to
explain complex behavior by deriving it from an analysis of simple reflexes and their
combinations. No mental processes were made use of. This completely behavioral
approach-although having to rely heavily on inference in some of the essential steps of the
argument-influenced many Russian physiologists, especially Pavlov (discussed below).
Ernst Mach (1836-1916) The Science of Mechanics (1883)
Mach had a direct influence on Skinner's general approach to scientific methodology,
epistemology, and philosophy of science. (Skinner read The Science ofMechanics as a
graduate smdent at Harvard) There was also a strong indirect influence, in that Loeb and
his student Crozier (discussed below) were influenced by Mach's orientation to science,
methodology, and epistemology, and Skinner worked in Crozier's laboratory as a student
at Harvard., Important features of Mach's approach that can be seen in much of Skinner's
own orientation to science are as follows: 3
1. The nature and origin of science: "An outgrowth of the practical concerns of
everyday life"; "the evolution of animal behavior and the history of physics [and any
science] are but two parts of a single historical line of epistemological development."
2. Biological economy in science: Science is just behavior, helping organisms to be
more effective. To do so it must show "efficiency of investigation, immediacy of
observation, economy of description aJ.1d communication....".
3. Cause as nothing more than a functional relation between independent and
dependent variables, as opposed to the notion of a cause as pushing or pulling to produce
its effects. Explanation is nothing more than a description of a functional relation.
4. Hypotheses and theories are unnecessary and often harmful.
5. The only effective epistemology must be empirical (and with Mach, this often meant'
behavioral, although behaviorism had riot been identified as a general approach at the time
Mach wrote The Science ofMechanics).
EdwarU L. Thorndike (1874-1949) Animal Intelligence (1898)
Loeb and others had argued that it was not necessary to infer consciousness and mental
processes to explain some of the behavior of "lower" species (invertebrates such as insects,
worms, snails). Their behavior could be understood as the result of some simple
mechanistic principles-as in tropisms, for example. Thorndike's puzzle box experiments

3The quotes are taken from Smith 1986, pp. 264,275.

 Historical Antecedents·ofBehavior Analysis 25

were meant to show that the problem-solving behavior of higher organisms (mariunals such
as cats), which might seem to require consciousness and reasoning processes, could be
understood as the inevitable result of a simple principle like the law of-effect. Skinner cites
Thorndike in this respect, but most probably got his mechanistic orientation from Crozier,
and from Loeb's work, rather than from Thorndike. It is quite clear that Thorndike's law
of effect, even though it sounds very much like the principle of operant conditioning, did
not give rise to an understanding of operant as opposed to respondent functional relations.
Watson later made no use of Thorndike's law of effect in his behavioristic approach, but
rather relied heavily on Pavlov's work. It is not clear how Skinner was influenced by
Thorndike's work; thus, no line was drawn from Thorndike to Skinner.
Ivan P. Pavlov (1849-1936) Conditioned Reflexes (1927)
Pavlov discoveredmost of the facts and principles that constitute our cwTent understanding
of respondent functional relations. This was the first really convincing experimental
analysis of behavior. The unconditioned reflex could not function as the basis for more
complex behavior because there would be no new functiona1 relations, no learning. But the
fact that new functional relations between stimuli and responses could be developed and
eliminated in the laboratory constituted a major step toward a completely mechanistic
behavioral account. Pavlov gave a series of public lectures from 1903 to 1928 that were
published (in Russian and also translated into German) and these influenced a number of
American psychologists, but it was the English translation of 1927 that Skinner read.
Skinner stated that he was much influenced by the rigor of Pavlov's experimental control.
He was also influenced in that he attempted to conceptualize the behavior of the whole
organism (after Crozier and Loeb) in the same tenns that Pavlov had developed for the
conditioned reflex. The basic tenns and concepts-conditioning, extinction,
discrimination, generalization, unconditioned stimulus, conditioned stimulus-were all
simply adopted from Pavlov.
.lohnB. Walson (1878-1958) ''Psychology as the beltavioristviews it"(1913)
This was the most effective call for an outspokenly behavioral position. In this paper,
p.ublished in The PsycJwlogical Review, the journal of the American Psychological
Association, Watson denied that consciousness and mental processes had any explanatory
value for the behavior of nonhumans or humans. This paper started the movement that
called itself beluiviorism. Skinner made first contact with Watson's work when he read
favorable comments about Watson's Behaviorism (a nontechnical book published in 1924
for popular consumption) in a book review by Bertrand Russell; as a result he read the
Watson book and also Russell's P h i l o s o p h y . ,
Bertrand Russell (1872-1970) Philosophy (1927)
According to Smith (1986, pp. 262), Russell's Philosophy was critical for Skinner's
appreciation of a strong link between epistemology and behaviorism. The book was "a
lengthy, detailed, and direct application ofWatsonian behaviorism to the traditional
problems of epistemology. Russell argued thatj,n most respects behaviorism provided an
adequate and fruitful account of both ordinary and scientific knowledge." ..
Jacques Loeb (1859-1924) The Organism as a Whole (1916)
Using invertebrates, Loeb studied the behavior of the whole organism, what would today
be called kineses and taxes, and attempted to explain such behavior in strictly mechanistic
terms. He strongly opposed inferences of mental functions as explanatory concepts for
such organisms. Loeb influenced Skinner indirectly because he was Crozier's teacher (see
below) and directly in that Skinnerread and was quite impressed by two of Loeb's books·
prior to entering graduate school. (Watson was influenced by Loeb's earlier writings and
through personal contact as well, because Watson took a course with Loeb at the
University of Chicago.)

•
 ~ ~ -~-------------------

26 Historical Antecedents ofBehavior Ana}ysis

W. J. Crozier, Head of the Department of General Physiology at Harvard

University during the period from 1925 to 1935
Skinner worked in Crozier's laboratory when he entered graduate school at Harvard and
was clearly more influenced by Crozier's general approach to science (a modification of
Loeb's and Mach's) than by that of the psychologists at Harvard at that time. Crozier's
unwillingness to infer mental or neural explanations, his interest in developing
mathematical functional relations between environmental variables and behavior, and his
interest in the behavior of the whole organism are very similar to important features of the
descriptive behavioral approach developed and refined by Skinner. It was also in Crozier's
Department of Physiology that Skinner made contact with the work of Magnus on postural
and locomotor reflexes and the work of Sherrington on spinal reflexes, both of which
topics were important to his early research.
B. F. Skinner (1904-1990) The Behavior of Organisms (1938)
This was an attempt to extend Pavlov's analysis of the reflex to the behavior of the whole
organism. In this book Skinner laid out almost all of the basic concepts that are currently
the substance of the area: called behavior analysis: a sharp distinction between respondent
and operant procedures and concepts, rate of response as the main operant dependent
variable,'tbe cumulative record as a way of studying rate of response, the methods and
results of operant conditioning and operant extinction, unconditioned and conditioned
reinforcers, the development of operant stimulus control (the discriminative stimulus and its
clear difference from the conditioned elicitor of respondent functional relations), stimulus
generalization, an analysis ofintertnittentreinforcement (schedules of reinforcement), and
the role of motivative and emotional variables.
B. F. Skinner (1904-1990) Walden Two (1948)
This was a utopian novel in which behavioral principles from The Behavior ofOrganisms
are used to design a culture. This effort to design a culture was considered by Skinner to
be one of his most valuable contributions. It is also the contribution that attracted a great
many people to behavior analysis who might not have come to this approach from a basic
science interest
F. S. Keller and W, N. Schoenfeld, Principles of Psychology (1950)
This introductory text, used at Columbia College for a number of years, was the first easily
understood version of the methods, concepts, and principles that Skinner had presented in
The Behavior ofOrganisms. In addition, Keller and Schoenfeld integrated Skinner's
approach with a number of important methods and results from experimental psychology in
general. It influenced many more people---undergraduates in the Columbia program,
graduate students who worked in the laboratory that went with the course, instructors and
students at other universities who adopted the text for their own courses-than did The
Behavior ofOrganisms. UK and S" clearly prepared the way for Skinner's own
introductory text, Science and Human Behavior.
B. F. Skinner (.1904-:-1990) Science and Human Behavior (1953)
This was a text written for one of Harvard's general studies courses in the natural
sciences-there were no prereqnisites, and students were not necessarily psychology
majors, Science and Human Behavior starts with basic concepts and principles as [lIst
presented in The Behavior ofOrganisms, then goes on to apply these principles to the ~
behavior of the individual as a whole, the behavior of people in groups (social behavior),
the behavioral functions of controlling agencies or institutions, and ends with the design of
a culture. The book is full of very effective extrapolations to all kinds of human situations.
It is a very persuasive presentation of the relevance of the behavioral approach to
understanding and improving human behavior by altering the environment in which the
behavior occurs. Like Keller and Schoenfeld, this book got many people started as
 Historical Antecedents ofBehavior Analysis 27

dedicated behavior analysts. In addition to its very effective introduction to the sCience of
behavior, the book contains many sophisticated analyses that are of interest to the more
advanced behavioral scholar.
B. F. Skinner (1904-1990) Verbal Behavior (1957)
Verbal Behavior extends the basic operant and respondent concepts and principles to all
aspects of human language. The emphasis is on the variables controlling the behavior of
the individual speaker, rather than on the listener or on the practices of the verbal
community, which are the topics of most interest to the traditional language scholar:
Ferster and Skinner, Schedules ofReinforcement (1957)
This book reports the results of a number of years of extensive study of the behavior of
pigeons on various schedules of reinforcement Most of the subsequent research in JEAR
(see below) began with methods described in Ferster and Skinner.
•
Journal of the Experimental Analysis of Behavior (JEAB) (1958)
Skinner, Keller and their students and associates were having trouble getting their operant
research published in the journals that would have been appropriate-the Journal of
Experimental Psychology and the Journal o/Comparative and Physiological Psychology.
This was because the research methodology differed considerably from that of mainstream
research in leaming and motivation. In particular, it was usually based on data from only a
small number of subjects (although there was usually a great deal of data from those few
subjects), and the effects of the different values of the independent variable were typically
assessed by visual comparison of cumulative records rather than with statistical significance .
tests. The operant research also seemed peculiar to other experimental psychologists in that
it was descriptive rather than a test of a theory. Actually, quite a few operant papers were
published in traditional journals, but in addition to the mild to moderate difficulty of getting
papers accepted was the fact that they were scattered over several different journals and
constituted only a small portion of anyone journal. As has happened in other areas, and
for somewhat similar reasons, the operant researchers eventually decided to publish their
own journal, in which the methodological standards would be more appropriate (and even
exclusive of much traditional group statistical research) and which would be full of papers
of interest to those researchers. JEAR has continued to be the main outlet for basic
research in experimental analysis, although such research is increasingly appearing in
several other experimental journals as behavior analysis seems to be merging somewhat
with other basic research areas (e.g., ethology and brain physiology). .
1958-1968
Many important books and articles were published during this period, and there were also a
number of important organizational events. these include Holland and Skinner, The
Analysis ofBehavior (a programmed textbook), 1961; Sidman, Tactics ofScientific
Research, 1960; Bijou and Baer, Child Development I, 1961; Division 25 of the American
Psychological Association formed in 1964; Staats and Staats, Complex Human Behavior,
1964; Ullman and Krasner (Eds.), Case Studies in Behavior Modification, 1965; Ulrich,
Stachnik and Mabry (Eds.), Control 0/ Human Behavior, 1966; Honig (Ed.), Operant.
Behavior: Areas 0/ Research and Application, 1966. Major uniyersitytraining centers
were started during this period at the University of Washington (Bijou), Arizona State
University (Staats, Goldiamond, Michael, and others), Southern Illinois University
(Sulzer, Azrin), the University of Kansas (Baer, Wolf, and Risley), and Western Michigan
University (Ulrich, Kent, Malott, and others).
Journal of Applied Behavior Analysis (lABA) (1968)
By the late 1960s so many applied (as contrasted with basic) research papers were being
submitted to JEAB that it became necessary to start a new journal devoted to applied
28 Historical Antecedents ofBehavior Analysis

behavioral analysis. JABA is mainly for the original publication of reports of experimental
research involving applications of the experimental analysis of behavior to problems of
social importance. Prior to JABA, research in what was then called behavior modification
was essentially an application of the methods used in basic research (as reported in JEAB)
to various practical problems involving the mentally ill, the mentally retarded, children,
juvenile delinquents, and others. JABA editors and contributors have developed and
refined methOdological strategies more appropriate to research in applied settings. JABA
articles have become model demonstrations of how to conduct and interpret applied
behavioral research, and JABA has become one of the most prestigious and widely cited
journals in any area of applied psychology.
Behaviorism (1972)
By this time many articles that were neither reports of basic nor applied research were being
submitted to JEAB, JABA, and other journals, and especially to philosophy journals.
Willard F. Day started Behaviorism which had this general goal: "to serve as a forum for
the critical discussion of issues pertaining to the contemporary practice of behaviorism."
The kinds of issues covered include conceptual issues related to the practice of
behaviorism, methodological innovations, ethical issues, philosophical issues related to
behavioriijl;Il, and others. Many of the articles have dealt with Skinner's analysis of
language in Verbal Behavior.
Association for Behavior Analysis (ABA) formed in 1974
Partly as a result of dissatisfaction with the programming of the annual convention of the
Midwest Psychological Association (a regional organization of the APA), a group of
behavioral psychologists from various universities in the Midwest decided to start their
own organization with the primary purpose of sponsoring an annual convention. It was at
rmt called the Midwest Association for Behavior Analysis (MABA) but when it attracted
behavioral psychologists from all over the country, and from other countries, the name was
changed to the Association for Behavior Analysis. It is at present the main organization of
professionals with a behavior-analytic orientation. Division 25 of the APA is largely
restricted to psychologists and heavily represented by academicians, whereas ABA is
composed of professionals (an M.A. or higher degree is generally required) in all aspects
of the human services, business and industry, education, as well as colleges and
universities.
1975 to the present (1990)
Several new journals have begun publication: Journal o/Organizational Behavior
Management (started in 1977 by Behavior Systems, Inc., under editorship of Aubrey C.
Daniels); The Behavior Analyst (started in 1978 by W. Scott Wood as a publication of
ABA); The Analysis o/Verbal Behavior (started in 1982 by Mark Sundberg as the
newsletter of the Verbal Behavior Special Interest Group of ABA; it became The Analysis
ofVerbal Behavior in 1985); and others. In addition hundreds of books devoted to
behavior analysis have been published during this period. ABA now has a number of very
active regional associations that have their own annual conventions. And there are a
number of other developments, consisting of the spread of behavior analysis into new areas
of applicatio!1, cooperative relationships with other professional groups, and increasing
availabiljty of behavioral training programs.
 3 f
BEHAVIOR ANALYSIS: AN OVERVIEW
A Little Recent History
When he published The Behavior of
Organisms in 1938, Skinner added the
subtitle "An Experimental Analysis."
Experimental had its ordinary meaning,
but analysis was apparently an
affIrmation of his analytic approach in
spite of the widely known criticism of
analysis by Gestalt psychologists
(Skinner, 1979, p. 203). By 1946 there
was a small group of experimental
psychologists working with Skinner at
the University of Indiana and with Fred
Keller and W. N. Schoenfeld at
Columbia University. Because of their
unconventional methodology (discussed
in detail later), these psychologists were
having trouble getting their work
published in experimental journals, and at
meetings their papers were presented in
sessions that were not related to this
group's interests.
As a temporary solution, Skinner,
Keller, and Schoenfeld organized their
own conference at Indiana University on
the experimental analysis of behavior, a
term taken from the subtitle of The
Behavior ofOrganisms. Before long,
members of this group were using this
term to refer to their work. Experimental
analysis now meant research on operant
functional relations employing the rather
unique methodology that characterized
Skinner's approach. Behavior was an
affIrmation of their behavioristic
orientation in considering behavior to be a
proper subject matter in its own right,
rather than simply a manifestation of
underlying inner causes.
Others have usually considered this
work to be a subclass oflearning theory
research, but learning in the sense of the
acquisition of new behavior is too
narrow, especially because a large
proportion of the studies were concerned
with factors influencing the maintenance
of behavior after it had been learned-the
effects of schedules of intermittent
reinforcement. And the fact that the
research was rather specifically. not the
testing of theories and not an application
of the hypothetico-deductive approach to
science made theory quite inappropriate
as well. The field is often referred to as
operant psychology, which is all right if
reference is being made to the general
research interests of this group, but this
label is not good if it carries the
implication that its adherents deny the
existence or importance of unconditioned
behavior or of respondent relations and
processes, which they clearly do not.
By 1958, the research output from
this orientation had .become large enough
that the group could establish their own
journal, the Journal ofthe Experimental
Analysis ofBehavior (JEAB), and in
1964 they formed a division of the
American Psychological Association, the
Division of the Experimental Analysis of
Behavior. During the late 1950s and
early 1960s the number of practical
applications·of operant behavioral
procedures in the areas of mental illness,
mental retardation, education, child
 30 Behavior Analysis: An Overview
development, rehabilitation, and other
areas increased rapidly, justifying the
establishment of the lourfUll ofApplied
Behavior Analysis (lABA ) in 1968.
The term experimental was now
being contrasted with applied, although
basic is actually a better contrast with
applied because applied researchers also
rely on experimentation. A considerable
body of literature was also developing
that stemmed from Skinner's more
theoretical and philosophical writings,
and that was closely related to basic and
applied operant research. So the
enterprise of behavior analysis was no
longer solely experimental as opposed to
applied, and not even experimental as
opposed to theory and philosophy.
Therefote, in 1975, when a group of
people with these diverse interests formed
the Midwest Association for Behavior
Analysis, soon to become the Association
for Behavior Analysis, experimental was
dropped. In 1978 this association
established another journal, and called it
The Behavior AfUllyst. So behavior
analysis is how first- and second- .
generation (and now third, fourth, etc.)
Skinnerians refer to their substantive and
methodological activities and interests.
What this consists of in more specific
terms is the subject of the sections that
follow.
Concepts and Principles
Unlearned Functional Relations
Behaviorism is widely, but quite
erroneously, believed to be committed to
Ii'
the proposition that all human behavior is'
learned. Watson did not hold this view
Ii,I and neither does Skinner, but the truth is
, they follow (operant conditioning). In
somewhat complex and it is easy to see
how the rnisconci::ption arises.
Some important behavioral
differences between species are due to
geneti1: factors and, ultimately, to the
action ofthe environment in natural
selection. In some cases it is even
possible to identify the chromosomal
region that carnes a behavioral trait.
Most of the available information relates
to the behavior of invertebrates
(mollusks, insects, worms, etc.) or that
of nonmammal vertebrates (fish, birds,
amphibians, reptiles), but there is also
some such information regarding
mammals. (For an introduction to this
general area see Fantino & Logan, 1979,
chapter 9.)
In terms ofjust what is inherited in
any particular species, it is convenient to
consider three kinds of behavioral
relations. (a) All organisms inherit some
fairly specific stimulus-response relations
called unconditioned reflexes. The reflex
whereby a decrease in skin temperature
leads to peripheral vasoconstriction is an
example. (b) Organisms also inherit'
more complex environmenhbehavior
relations, previously referred to as
instinctive behavior and now more likely
to be called released behavior. In many
species mating behavior, nest building,
and some aspects ofthe care ofthe young
illustrate this type of inherited functional
relation. It is an empirical matter in each
instance whether these kinds of behaviors
are best described as built-in response
patterns triggered by specific though
complex stimuli, or as the effect of,
special kinds of reinforcement that
develop and maintain the specific form of
behavior that is followed by them (in
which case they belong in the next
category), or as some combination of
both processes. (c) Organisms also
inherit certain capacities to be changed
behaviorally by the environment; thus,
the capacity for reflex responses to be
conditioned to novel stimuli (respondent
conditioning) and the capacity for some
environmental occurrences to increase the
future frequency of the type of behavior
fact, all principles of behavior are
descriptions of built-in capacities in this
sense.
Between-species differences are
often explained in terms ofphylogenic
provenance; that is, behavior analysts
readily attribute to innate endowment
many behavioral characteristics that are
shared by all members of a species and
that can distinguish one species from
another. So why is it so often claimed

that behaviorism denies the importance of
innate endowment? There are several
reasons. For one thing, there is very little
research being done at present on
unlearned behavior by psychologists or
zoologists operating from a behavior-
analytic perspective. Also, most of the
current efforts at theoretical integration in
this area by ethologists make extensive
use of mentalistic or cognitive
explanatory concepts, which makes such
theory (but not necessarily the data)
unattractive to behavioral psychologists.
But the main problem is that laymen
and professionals in various fields are not
primarily interested in the kind of innate
endowment relevant to differences
between species. The important innate
endowment for most people is the one
that might help to explain behavioral
differences among different people. The
behavioral position regarding this type of
inheritance does not radically differ from
the general view, but typically does not
go far enough to suit most
nonbehaviorists. Certainly people must
inherit physiological characteristics that
make them behave differently than other
people. It is completely unreasonable to
assume that we are all exactly alike in all
behavioral characteristics (e.g., in rate of
learning new stimulus-response
relations). This admission is easily
obtained from most behavior analysts,
but they are likely to become reluctant to
attribute any specific difference to
inheritance.
There seem to be two reaSons for
this reluctance. Often such attribution is
simply based on ignorance of any other
relevant variables, and in this sense
inheritance is nothing more than an
explanatory fiction. But in addition, the
basic issue is really a matter of degree.
The behaviorist is very likely to argue that
there is already strong evidence for large
environmental effects, which must be
well understood before one attributes
much to innate determination.
Take athletic ability, for example. It
would be quite unreasonable to assume
that we are all born exactly equal with
respect to the ability to run a fast mile.
Behavior Analysis: An Overview 31
There must be many relevant inherited
characteristics, such as height, body
proportions, lung capacity, and more
complex neuromuscular characteristics.
However, it would certainly be a mistake
to attribute the large differences between
mile times of members of a randomly
sarnpled group of humans to innate
endowment It seems quite reasonable to
expect that the range of such differences
would be greatly reduced if all members
of the sarnple had been.trained from birth
to be mile runners. Even after such
training there would, of course, still be
differences, but they might not be very
interesting. And in terms of practical .
matters, a training program would clearly
gnarantee large effects in a relatively short
time as compared with a selective
breeding program. The same is true for
so-called personality characteristics--
before we accept a predominantly genetic
explanation of such :individual
differences, we must have a better
understanding of the obviously.relevant
environmental factors.
So, behavior analySts readily
acknowledge innate determination, but
when it comes to the only kind that most
people are really interested in, they
become increasingly resistant and are then .
quite justifiably credited with a preference
for environmental explanations.
Respondent Functional Relations
In The Behavior ofOrganisms (19'38),
Skinner began with a description of
functional relations involving reflexes
(unlearned behavior) largely based on the
work of Sherrington (1906). He next
considered conditioned reflexes
(respondent or classical conditioning)
based on Pavlov's work (1927), and
finally considered functional relations
involving behavioral consequences
(operant conditioning). In Science and
Human Behavior (1953) and About
Behaviorism (1974), the systematic
treatment of behavior also began with
reflexes and respondent conditioning.
Skinner never questioned the existence or
importance of respondent behavioral
relations. Specific reflex responses (e.g:,
 32 Behavior Analysis: An Overview
salivation to taste stimuli or peripheral
vasoconstriction to a decrease in skin
temperature) are not of great importance
in human affairs, but the functional
relations revealed by studying the
conditioning of such responses are
important for understanding stimulus
control in general, and for understanding
emotional variables and effects. It is also
very important to understand the
respondent type of stimulus control so as
not to confuse it with operant stimulus
control.
Briefly stated, Pavlov studied the
way in which stimuli that did not elicit
reflex responses could be brought to do
so by being paired with the unconditioned
stimuli for such responses. The former
stimuli are then called conditioned
stimuli, and Skinner systematically used
the term elicitation to refer to the effects
of both unconditioned and conditioned
stimuli. Presentation of the. conditioned
elicitor by itself leads to a decrease in its
effectiveness as an elicitor and this
procedure is called respondent extinction.
Respondent relations most
commonly involve smooth muscles and
glands as effectors, as opposed to the
striped muscle effectors most commonly
subject to operant conditioning, as
described later. The expression
respondent behavior and its contrasting
operant behavior might imply that one
could classify a response as respondent
or operant solely on the basis of a careful
inspection of the behavior itself.
I
However, there are a number of
exceptions to this rough correlation
between type of effector and type of
11
! ]1
effective conditioning procedure, which
is why the term respondentfunctional
relation is preferred in the present paper.
It is the controlling relation between
independent variable and behavior that
can be identified as respondent or
operant, not the behavior alone.
Most of The Behavior ofOrganisms
is concerned with the investigation of
operant functional relations. It was the
effect ofbehavioral consequences that
had been previously neglected or
confused with respondent or classical
conditioning and that needed extensive
exploration, and it is the human operant
repertoire that is of most significance for
other humans. Likewise, the researchers
working in the experimental analysis of
behavior who followed up Skinner's
work have been primarily involved in
extending our knowledge of operant
relations. Also, when the field of
behavior modification or applied behavior
analysis was started in the late 1950s, the
manipulation of behavioral consequences
quickly becarne the standard technique of
that movement. As a result of this
emphasis on the operant relation,
Skinner's approach is often called an
operant conditioning approach, or
operant psychology, or operant
behaviorism, especially by those who
wish to contrast it with some supposedly
broader orientation. This is clearly a
mistake. Irrespective of a justifiable
emphasis on operant relations, behavior
analysts acknowledge the importance of
. unlearned behavior and of respondent
conditioning as a form of learning in
addition to operant conditioning. In fact,
recent work in the area of respondent
conditioning (Rescorla & Wagner, 1972)
has revealed some general features of
compound stimulus control-.blocking
and overshadowing-that are the subject
of much current research in the
experimental analysis of operant stimulus
control.
Operant Functional Relations
Operant conditioning. Operant
conditioning is typically explained in
terms of the response and the reinforcing
consequence without reference to prior
stimuli. Stimulus control is then
introduced later in connection with
generalization and discrimination. Nor
are motivative (same as motivational but
possibly less likely to be used for an
intemal mental condition or a hypothetical
construct) relations typically brought in at
the beginning. However, these
postponements can cause conceptual
trouble, so this treatment will be a little
more complex at the beginning but
conceptually clearer in the long run.

The basic operant functional relation
can be stated in tenns of four kinds of
events, as follows: In a particular (a)
stimulus situation, some (but not all)
kinds of (b) behavior, when followed by
certain (c) environmental consequences
(reinforcement), show an increased future
frequency of occurrence in the same or in
a similar stimulus situation. For the
environmental consequences to have
functioned as reinforcement, some (d)
motivative variable must have been in
effect, and the increased future frequency
of response occurrence is only seen when
that motivative variable is again in effect.
This is often referred to as the three-tenn
contingency of (a) stimulus, (b)
response, and (c) reinforcement; but the
(d) motivative variable responsible for an
environmental event functioning as
reinforcement must also be specified. A
food-deprived rat reinforced with food
for pressing a lever in a particular type of
experimental chamber will have an
increased tendency to press the lever
when it is next in that chamber or one
resembling it, but only if it is food
deprived at that time.
Motivation in behavior analysis
usually refers to one or both of two
effects produced by variables like food
deprivation, water deprivation, painful
stimulation, etc. These motivative
variables (a) alter the effectiveness of the
relevant event (food, water, pain
reduction) as a fonn of reinforcement;
and (b) change the relative strengths of
different parts of the organism's
repenoire. The food-deprived organism
is (a) reinforceable by food, in that any
behavior that immediately preceded food
delivery would be increased 1 in future
1Motivative variables actually have effects in
both directions-food deprivation increases and
food satiation decreases the effectiveness of food
as a fonn of reinforcemenI, and similarly with
the effect on the momentary strength of the
relevant part of the organism's repertoire.
Further directional complexity results from the
fact that some punishers are also altered in
effectiveness by motivative variables and also by
the fact that discriminative stimuli may be
suppressive as well as facilitative. For more
Behavior Analysis: An Overview 33
frequency, and (b) any behavior that had
previously been reinforced by food is
momentarily increased in frequency at the
expense of other behavior. Because most
operant relations involve discriminative
stimuli (see below), it is more precise to
say that the motivative variable alters the
effectiveness with which those
discriminative stimuli increase the'
frequency of the relevant behavior.
Another way of looking at the basic
operant relation is to say that
reinforcement produces in the organism's
repenoire a lasting increase in the strength
of a unit involving (a) a particular
motivative variable (the variable
responsible for the effectiveness of that
particular type of reinforcement), (b) a
particular stimulus situation, and (c) a
particular type of behavior.
Stimulus control is actually
somewhat more complex than implied by
reference to the basic three-tenn relation.
It is true that a single instance of
reinforcement will result in some control
of the response by the stimulus situation
present during reinforcement, but more
commonly the organism also has an
opponunity to make the same response
but not be reinforced in a stimulus
situation that differs in some way from
the original one. Responding without·
reinforcement is called operant extinction,
and operant stimulus control generally
involves repeated reinforcement in the
presence of a particular stimulus
condition but extinction in the absence of
that stimulus condition. Skinner
introduced the tenn discriminative
stimulus (1938, p. 177) for the stimulus
condition in which the response is
reinforced, abbreviated it as SD, and
referred to the condition in which
reinforcement is not provided as SII. (S
delta). An SD, then, is a stimulus
condition in the presence of which a
response has occurred and been
reinforced, and in the absence of which
the response has occurred and has not
been reinforced, and for this reason, in
details on motivation see the later paper in this
collection titled "Establishing Operations."
34 Behavior Analysis: An Overview
the presence of that stin:mlus condition the
momentary frequency of that type of
response increases .
It is essential to emphasize the
importance of the immediacy of
reinforcement Events that are delayed
more than a few seconds after the
response do not directly increase its
future frequency. When human behavior
is apparently affected by long-delayed
consequences, the change is
accomplished by virtue of the human's
complex social and verbal history, and
should not be thought of as an instance of
the simple strengthening of behavior by
reinforcement. When, for example,
industrial work behavior is increased by
public posting of daily productivity, this
effect could not possibly be the direct
result of reinforcement, because the
posting occurs hours (sometimes days)
after the relevant behavior, and the
worker is not working but is instead
reading (the notice) at the instant the
reinforcement occurs. This is not to deny
that such procedures actually alter the
relevant behavior, but only to insist that
they do it in a more complex way, which
is only just beginning to be understood.
This point will become important when
rule-governed behavior is considered
below in connection with extensions of
these basic concepts to more complex
human behavior. . generally be made between an offer of
Some events function as
reinforcement, given the proper
motivative operation, irrespective of any
learning history on the part of the
organism: we do not have to learn to be
reinforced by food when food deprived,
or by water when water deprived, for
example. Such events are referred to as
unconditioned reinforcers. Some
stimulus changes acquire their capacity to
reinforce behavior by being
systematically related to other reinforcing
events, and these stimulus changes then
function as conditioned reinforcers. In a
common laboratory demonstration, a
food-deprived rat is trained to press a
lever in the presence of an overhead light
through the use of food reinforcement.
Typically the light-on condition is
alternated with a light-off condition in
which pressing can occur but is not
reinforced. When good stimulus control
is developed---the rat presses when the
light is on but not when it is off-it is
possible to show that the light onset will
now function as reinforcement for any
response that produces it, such as pulling
a chain that is suspended from the ceiling.
In addition to being classified as
unconditioned or conditioned (primary
and secondary are sometiI;lles used, but
these terms too easily imply more than
just the relevance of a learning history),
reinforcers are often called positive, if
they strengthen by being immediately
presented, and negative, if they
strengthen by being immediately removed
following the response. The term
negative reinforcement has caused
considerable confusion, however,
because it is so easily equated with
punishment (described later). The
distinction between positive and negative
reinforcement may be unnecessary and
seems to be made less often now than a
few years ago, but in any case, a possible
way to avoid the confusion is to note that
negative reinforcement is highly
desirable. If offered negative
reinforcement you should always accept
the offer. It may also help to reflect on
the fact that an intelligent choice cannot
positive and negative reinforcement
without knowing what specific
environmental changes are being
considered.
In The Behavior ofOrganisms,
Skinner described the detailed results of
what he called periodic reconditioning,
later referred to as a fixed-interval
schedule of reinforcement. This is an
arrangement in which the first response
following afixed time period from the
last reinforced response is reinforced, and
also initiates the next fixed time period,
and so on. It is one of the large variety of
intermittent reinforcement schedules
falling between the extremes of
reinforcement for every response and
reinforcement for no responses. Work
on intermittent reinforcement has

constituted a large proportion of the
research in the experimental analysis of
behavior, starting with the extensive
results reported in Ferster and Skinner's
Schedules ofReinforcement in 1957 and
continuing to occupy much of the content
of the early volumes of JEAB. This
work has been important for several
reasons. First, some schedules of
reinforcement have become standard
research tools that are used to study a
wide variety of other phenomena in the
science of behavior. Second,
understanding how a particular schedule
produces its characteristic pattern of
performance has been an intriguing
research area in its own right, and has led
to much new knowledge about behavior.
Also, the knowledge gained from
studying various schedules of
reinforcement has greatly improved our
technology of behavioral control. But
probably the most important effect has
been the enrichment of the behavioral
independent variable.
Prior to Skinner's discovery of
interval and ratio schedules, the basic
behavioral independent variables were the
variables ofrespondent conditioning and
operant reinforcement (also punishment)
either given or withheld. Yet behavior
seemed to be a very complex dependent
variable, a situation that encouraged the
invention of inner determiners to account
for unexplained behavioral variation. The
study of the effects of intermittent
reinforcement has shown that behavior is
sensitive not only to whether it is
reinforced or not but also to the
a"angement of reinforcement in
combination with requirements or
contingencies involving time, number of
responses, and the presence of stimuli.
Adding time and number of responses to
the ordinary three-term continge!lcy
(really four) gives us an independent
variable more than sufficiently complex to
be considered as the prime determiner of
variations in the dependent variable,
behavior.
Finally, the study of intermittent
reinforcement has also revealed the role
of consequences in maintaining behavior
Behavior Analysis: An Overview 35.
after it has been acquired. Historically,
differences or fluctuations in response
frequency, after learning, have been
assigned to the topic of motivation.
Removing from this topic differences that .
are due to different schedules of
intermittent reinforcement leaves for
motivation only the differences due to
variables (like deprivation) that alter,
response frequency while at the· same
time altering the effectiveness of various
events as reinforcement .Reinforcement
is thus seen to be a much more important
variable than was previously realized, and
the topic of motivation has become
somewhat more restricted. .
Operant weakening by
punishment. Some (but not all) kinds
of behavior, when followed by certain
environmental changes (punishment),
show a decreased future frequency of
occurrence in a similar stimulus situation.
As with reinforcement, punishment must
be immediate to have much of an effect,
and as with reinforcement, punishing
events can be classified as unconditioned
and conditioned, dependent upon the
necessity of a learning history.
It now seems reasonable to describe
this basic operant weakening relation as
the opposite of the operant strengthening
relation, but the issue has had a long and
complex history, and many behavior
analysts, especially in the applied field,.
will probably contest the correcmess of
this approach. Punishment has lx;en
more difficult to study than . .
reinforcement, largely because it cannot
be studied easily by itself. One must have
some behavior to punish, which means
behavior that has been or is still being
reinforced. Thus, punishment has
always been superimposed on a recent or
a continuing schedule of reinforcement
for the same behavior.
Early experiments by Skinner (1938)
and by Estes (1944) seemed to show that
punishment did not actually weaken
behavior directly, but rather only
constituted an arrangement that favored
the development of behavior that was
incompatible with the punished behavior.
On the basis of these experiments and
 36 Behavior Analysis: An Overview
possibly other less fo~ kinds of .
evidence, Skinner proVIded an extensIve
treatment of the topic in Science and
Human Behavior (1953), which for a
time was the generally accepted view. In
a series of experiments begun in the late
1950s, Azrin and his colleagues corrected
flaws in previous experiments and
provided a comprehensive picture of the
effects of electric shock punishment on
the operant behavior of pigeons. A
detailed treatment of the topic of
punishment, including the results of the
previous 10 years of research, appeared
as a chapter (Azrin & Holz, 1966) in the
first handbook of operant behavior
(Honig, 1966). This research is the basis
for the currently accepted interpretation
described at the beginning of this section.
Another common form of operant
conditioning consists of the reinforcement
of behavior by the termination of
environmental events that when presented
would function as punishment. Such
stimuli have been called aversive, and
their termination is the negative
reinforcement referred to earlier. When
the response terminates an unconditioned
aversive stimulus, the procedure is called
escape; when it terminates a warning
stimulus that has become a conditioned
aversive stimulus due to its relation to the
unconditioned aversive stimulus, the
procedure is called avoidance. The
interpretation of avoidance behavior is
currently somewhat controversial,
especially with respect to
nondiscriminative avoidance, that is,
avoidance without a warning signal.
Rats, monkeys, and humans readily learn
to press levers that do not actually
I
II terminate any form of exteroceptive
,Ii stimulus but insteild only postpone the
iii onset of the next (typically brief) aversive
stimulus. In the molar interpretation the
I
reinforcement for the lever pressing
'I behavrt>i'is simply the decreased
i was typically compared: (a) the main
frequency of aversive stimuli that results,
I,'
even if there is no immediate
I
reinforcement (Hermstein & Hineline,
1966; Sidman, 1962). In the molecular
interpretation, the stimuli resulting from
the lever-pressing behavior itself
(kinesthetic, tactile, etc.) are the
reinforcement, in that they are never
followed immediately by shock, as
contrasted with the stimuli they replace
(Dinsmoor,1977).
The molar-molecular controversy is
not restricted to the interpretation of
avoidance, but is in fact a major current
topic of research and theorizing. The
basic issue is whether reinforcing or
punishing consequences aiter behayior
primarily by being temporally conttguous
with it-the molecular view (which was
Skinner's interpretation when he first
focused on the importance of behavioral
consequences}-or by being dependent
upon behavior even if temporally remote
from it-.the molar view (for exarnple,
Baum, 1973; Hermstein, 1969; Hineline,
1977). At present temporal contiguity is
. seen as quite important, but whether it is
strictly necessary or strictly sufficient is
still controversial (as are many issues
within the field of behavior analysis).
Although the general outline seems
quite clear, the field should not be
thought of as being committed to any
existing principle. Behavioral knowledge
is constantly being refilled, and in some
cases drastically revised. In spite of the
unfinished nature of the field, however, I
think the general view can be safely
described as highly deterministic.
Although the details are by no means
completely known, it is assumed that
human behavior will be completely
accounted for by specifying the relevant
unlearned and learned functional
relations.
Methodology
Skinner's research in the 1930s and
1940s and the work of those who
followed him differed in the following
ways from other research in the
psychology of learning, with which it
dependent variable was rate of response,
typically portrayed as the cumulative
curve of responding produced by a single
animal; (b) each subject was exposed to
all values of the independent variable, and
only a few subjects were studied in any
"
one experiment; (c) no use was made of
statistical inference (significance tests);
and (d) the research was not construed as
a test of a theory.
Because large numbers of subjects,
compl~x statistical treatment (typically the
ana!ysls ofv~ance~, and explicit theory
testmg were mcreasmgly the basis for
evaluati.on of researc~ on learning during
that penod, the expenmental analysis of
~havior seemed very different and its
differences were often interpreted as
deficiencies. Because these four
methodological idiosyncrasies still
characterize much of the behavior
analysis research and are seen as tts
distinguishing features by many, they
\Yill be described in some detail.
Rate of Response
For Skinner, rate of response was more
!han just a conve.nient dependent variable;
It was the only direct reflection of
response probability, which was the basic
concern of a science of behavior. "A
natural datum in a science of behavior is
the probability that a given bit of behavior
will occur at a given time. An
experimental analysis deals with that
probability in tenns of frequency or rate
of responding" (1966, p. 213). Other
measures commonly used in experimental
psychology-latency, reaction time, time
to complete a task (to traverse a runway,
for example), number of errors made,
number of trials to a criterion--were all
criticized as "not related in any simple
way to probability of response" (p. 214).
And again, "Rate of responding is a basic
dimension, not simply because responses
can be accurately counted, but because
rate is relevant to the central concern of a
science of behavior" (p. 214).
The appropriate use or meaning of
the ~ncept of J?robability is presently the
subJc::ct of ~onslderablecontroversy in the
phYSical sCiences and in the philosophy
of science (Fine, 1973). Skinner
acknowledged the difficulty with the
concept (1953, pp.62-63; 1957, pp. 22,
28) but did not deal with it, and any
justification of rate in terms of its relation
to probability remains problematic.
Behavior Analysis: An Overview 37
Response rate is much easier to justify
empirically. Its study has led to the
accumulation of increasingly precise
information about the principles of
behavior. Response rate is also easy to
justify in the applied areas in which
frequency of occurrence (too high or too
low) is the essential feature of most
behavioral problems. f
Within-subject Comparisons
Behavior analysis is often said to favor
single-subject rather than the more
con,nnon group-statistical experimental
deslgus. The essence of this distinction
concerns the way the different values of
th~ independent v.ariable are compared
With each other: m tenns of their effects
o~ ~e sam~ subject at 9ilferent times (a
within-subJect companson) or in terms of
their effects on different subjects (a:
between-subjects comparison). When the
within-subject comparison is made, the
reliability of the differences between .the
effects of the different values of the
independent variable is assessed by
observing the performance of a single
subject at each of these values, often
repeatedly introduced and removed.
Several experimental subjects (seldom
more than three or four) are used to
assess the generality of the findings. If
the performances of the small number of
subjects differ very much from one '
another, one solution is to increase the
number of subjects and average their
results to reveal the general functional
relation; another is to try to isolate the
sources of uncontrolled variation and then
perform the experiment again under better
conditions of experimental control. It is
this latter approach that has typified
behavior-analytic research.
Skinner's early research on
conditioning and extinction using rate of
response in a free responding (as
contrasted with discrete trial) situation
required prolonged study of the same
organism, and the similarity of the
cumulative curves across subjects made it
unnecessary to study more than three or
four animals in any given experiment.
The later research on intermittent
 38 Behavior Analysis: An Overview
reinforcement resulted in interesting
behavioral changes even after many hours
of exposure to the independentvariable,
which made the careful study of a few
organisms the only practical option. Most
of the research published in lEAR also
involves prolonged exposure to
independent variables and an emphasis on
stable performances prior to changing
conditions, factors that work against the
use of the between-subjects comparisons.
Although it is the within-subject
feature of this research that is
methodologically somewhat unique (not
the actual number of subjects used),
between-subjects comparisons typically
require quite a few more subjects. When
the different values of the independent
variabl6'are applied to different subjects,
individual differences are confounded
with those values, and typically can only
be unconfounded by averaging the results
of several subjects (several usually means
five or more). Thus within-subject
experiments typically use three of four
subjects; between-subjects experiments
are likely to involve a group of five or
more subjects for each value of the
independent variable, although with
extremely stable and well-understood
individual performances a between-
subjects comparison can involve only a
few subjects.
Statistical Inference
Skiuner and subsequent ba~.lc and applied
researchers have not been reluctant to use
ordinary descriptive statistics such as
frequency distributions; means, medians,
percentages, scatter diagrams, and
correlation coefficients when they
facilitate effective reaction to data. They
have made little use, however, of the
significance tests of statistical inference.
~e reasons are probably quite complex.
Highly effective experimental control .
makes such techniques superfluous.
When repeated application of the different
values of the independent variable leads
to ~bvious changes in the dependent
vanable, ~e assessment of the probability
?f su~h diff~rences occurring by chance
IS unmteresnng. It is also true that the
langnage of statistical hypothesis testing
seems most appropriate when the
experiment is itself formulated within a
hypothetico-deductive or theory-testing
approach to research, and much of the
early behavior-analytic work explicitly
disavowed this general orientation. It is
also not entirely irrelevant that there were
no easily applicable significance tests
(and still aren't) for the type of within-
subject comp:irisons most commonly
made in the analysis of h:havU.
Theory as a Basis for Research
During the late 1930s and throughout the
1940s much research in leaming was an
explicit application of the hypothetico-
deductive approach. This was especially
true of research generated by Clark Hull's
postulate system (Hull, 1943), aimed at
either refining the postulates and
theorems or refuting them. Skinner's
orientation was by contrast much more
descriptive. He had discovered a
sensitive dependent variable and had
developed a methodology for studying its
relation to obvious and important
independent variables. With the further
discovery of the area of intermittent
reinforcement, research questions were
easily formulated and easily investigated
in the sense that one study led to another,
and there was plenty of work for anyone
who was interested. In the context of the
leaming psychologists' love qffair with
theory testing, however, the work of .
Skinner and his followers was often
criticized as a sort of purposeless data
gathering. Skiuner reacted by writing the
widely quoted paper, "Are Theories of
Leaming Necessary?" (1950), which was
an emphatic negative answer to the
question posed, and eamed him the anti-
theory reputation that he now holds. It
was only certain types of theories that he
opposed, namely "any explanation of an
observed fact which appeals to events
taking place somewhere else, at some
other level of observation, described in
different terms, and measured, if at all, in
different dimensions" (1950, p. 193).
He specifically criticized the explanatory
use of inferred physiological entities or
relations, mental events, and the

hypothetical constructs of the Hullian and
similar theory builders (pp. 193-194).
Skinner's orientation still characterizes
the behavior analysis approach, with
inferred cognitive processes being the
current opposition party. However, a
good deal of the research recently
published in JEAB is considerably more
theory oriented than would have been
considered appropriate 20 years ago.
This is especially true in connection with .
efforts at quantification (see, e.g.,
Commons & Nevin, 1981).
These last three methodological
features, within-subject versus. between-
subjects comparisons, visual inspection
versus. significance tests, and description
versus a theory-testing basis for research,
are logically quite independent of each
other. Currently, all possible
combinations are found in JEAB and
JABA, although methodologically many
studies still resemble the earlier work in
the experimental analysis of behavior.
It is in connection with theory that
Skinner's orientation has sometimes been
called antiphysiological. His opposition,
however, is only to inferred physiological
processes as explanations of behavior.
He has never expressed any opposition to
actual physiological independent of . . almost any human problem with some
dependent variables, and has been quite
enthusiastic about the possible
cooperative roles of the experimental
analysis of behavior and the physiological
analysis. It is true that he sees the
direction of usefulness, at least in the near
future,jrom the experimental analysis of
behavior to physiology. Valid behavioral
relations cannot be refuted by any
physiological discoveries, but
physiologists must know principles of
behavior in order to know what to look
for (1966, pp. 282-284).
It is important to point out that only
by comparison with other work in the
experimental psychology of learning did
Skinner's methodology seem unusual. It
was not at all unlike the methodology
practiced in physiological psychology,
psychophysics, and, for that matter, in
much of biology, chemistry, and physics.
Sidman's Tactics ofScientific Research
Behavior Analysis: An Overview 39
describing and justifying this
methodology was published in 1960 and
quickly became the standard reference for
researchers in behavior analysis. More
recently several other books describing
various aspects of this approach have
become available (Bailey & Bostow,
1979; 1981; Hersen& Barlow, 1977;
Johnston & Pennypacker), and the f
approach appears tobe increasingly
accepted within experimental psychology
as a whole.
Extensions
Behavior Modification or Applied
Behavior Analysis
There are undoubtedly many antecedents
to the applied branch of behavior
analysis, but Skinner's Science and
Human Behavior (1953) is probably the
most relevant. There, for the first time,
respondent and operant functional
relations were used to interpret many
aspects of human individual and social
behavior. Contingencies of
reinforcement, ubiquitous but often quite
subtle, played the leading role in this -1
sensitive and highly sophisticated look at
the human condition. Mastery of this text
prepared the behaviorist to approach
chance of success, and further, to have a
behavioral interest in almost any aspect of
human behavior.
By the late 1950s and early 1960s
descriptions of specific research projects
in the areas of mental illness, education,
and mental retardation began to appear in
the literature, and in 1968 the Journal of
Applied Behavior Analysis (JABA) was
founded. In early applications, behavior
problems were interpreted in terms of the
concepts and principles of the operant
(and to some extent, also respondent)
conditioning laboratory, and treatment
procedures followed from these
interpretations. More recently many
applications have used the methodology
of behavior analysis, but with an outcome
orientation that does not emphasize
interpretation in terms of conditioning
concepts and principles.
 ----_ ..
40 Behavior Analysis: An Overview
At first, behavior modification was
applied mainly with institutionalized
clients and with children, but it has now
expanded to include all phases of human
endeavor. This can be nicely appreciated
by considering the chapter headings of a
typical recent text in behavioral
applications, Behavior Change by
Lutzker and Martin (1981).
Behavior Change: History, Principles,
and Processes
Assessment and Evaluation
The Community and the Environment
(littering, recycling, etc.)
The Working World: Business,
Industry,'and Government
Health Care: Adults (cardiovascular
disorders, neurological disorders,
rehabilitation, etc.) .
Health Care:,children (constipation,
seizures, compliance with medical
procedures, fear of medical
procedures, etc.)
Children (environments for young
children, toilet training, bed wetting,
delinquency, etc.)
Classrooms (preschool, teacher behavior,
higher education, etc.)
Severe Problems: Institution and
Community Intervention (mental
illness, mental retardation) .
Personal Problems: Obesity, Smoking,
Chemical Dependency
Personal Problems: Anxiety and
Depression
Relationships and Sexual Behavior
Training, Ethics, and the Future
As can be seen from this list, applied
behavior analysis includes all the
traditional areas of applied psychology
(clinical, industrial, educational) as well
as more recent developments such as
behavioral medicine, rehabilitation, and
environmental psychology. Several
additional areas actually being dealt with
by behavioral psychologists are not
covered in theLutzker and Martin text,
such as problems of aging (note Skinl\er
_-------------------
& Vaughan's Enjoy Old Age, 1983),
legal jurisprudence and criminal justice,
the analysis and design of instructional
materials (as with programmed
instruction), and sports psychology.
Four Interpretive Analyses by
Skinner
Even with the preceding description of
basic concepts and principles, distinctive
methodology, and the applied branch of
the field, the character of behavior
analysis is not accuratelyportrayed
without mention of some of Skinner's
unique theoretical contributions. They
represent interpretive analyses of complex
human behavior, and four of the most
important are presented below.
Verbal behavior. Even before
The Behavior ofOrganisms was
published in 1938, Skinner began the
analysis of human language that was
ultimately published as the book Verbal
Behavior in 1957. Although Skinner
considered this to be the contribution that
will ultimately be seen as his most
important (personal communication,
1981), it has only recently begun to have
a noticeable impact on the field.
However, this book represents a
completely behavioral treatment of human
language that is not only unique, but in
many ways is completely incompatible
with both common sense and
professional (linguistic, psycholinguistic)
treatments. Not only that, but it is also
very complex, as it must be to do justice
to the topic and the basic notions
presented in the first part of the boOk
must be mastered for the later sections to
make sense. The journal Behaviorism
began publication in 1972, and has had a
steady stream of contributions related to
Verbal Behavior. The Association for
Behavior Analysis began annual
conventions in 197.4, and the convention
has had a verbal behavior specialty area
since 1978. Perhaps it simply took 20
years or so even for behaviorists to
appreciate this very revolutionary way of
thinking about language.
Skinner begins by defining verbal
behavior as behavior that achieves its

effect on the world through the mediation
of someone else's behavior. Its
reinforcement is thus indirect, as
compared with the reinforcement of
nonverbal behavior. The nonverbal
behavior of reaching for an object is
reinforced by tactile contact with the
object; the verbal behavior of asking for
the object is only reinforced with the
object if there is someone available who
responds to the request. The implications
of the indirectness of the reinforcement of
verbal behavior are far reaching; in fact,
this indirectness is responsible for many
of the unique and powerful features of
human language (1957, pp. 203-206).
Next, Skinner identifies several
elementary verbal units (mand, echoic,
textual, intraverbal, and tact)
distinguished from each other by the
relation between the form of the response
(what is said or written) and the nature of
the controlling variable. He then
considers the extension of these
elementary verbal operants to novel
conditions on the basis of stimulus
generalization (generic extension,
metaphor, metonymy, etc.); the effects of
multiple controlling variables; and finally
the role of ongoing verbal behavior as a
controlling variable for further verbal
responses (autoclitic verbal behavior).
This basic approach is enriched with
hundreds of examples illustrating the
interpretive power of the small number of
completely behavioral concepts.
Unfortnnately, but unavoidably, the book
assumes a good deal of familiarity with
topics considered in traditional treatments
of language: this adds to its general
difficulty but also contributes to the
persuasiveness of the argument. Skinner
essentially demonstrated that there.are no
aspects of human language that cannot be
"understood," at least in this interpretive
sense, in completely behavioral terms.
As mentioned above, it seems that
increasing use is being made of Skinner's
analysis in philosophical and theoretical
activity and also in research. The effect
of the book has not been seen in some
areas in which it might have been
expected (e.g., the recent efforts to
Behavior Analysis: An Overview 41
develop human language in apes,
UJifortunately for that enterprise, show no
detectable influence by Verbal Behavior),
but it has influenced many in the field of
behavior analysis and is being taught
about and written about now more than
ever before. (For a recent review of this
area see Michael, 1984.)
Private events. In "The f
Operational Analysis of PsychOlogical
Terms" in the Psychological Review in
1945, Skinner first presented his analysis·
of verbal behavior under the control of
private stimuli. Rather than attempt to
provide operational defmitions for terms
like ache, itch, hunger pangs, anger, and
so forth, he asked how the verbal
community could develop in the speaker a
verbal repertoire under the control of such
private stimuli. He concluded that it is
accomplished indirectly, in one of two
ways. First, the community can react to
the learner by providing reinforcement,
extinction, punishment, etc., on the basis
of common public accompaniments of the
private stimuli (including the public
stimuli resulting from behavior already
controlled by the private stimuli, such as
scratching, crying, etc.). For example,
an adult can use the language of pain in .
comforting a child who has been cut by a
sharp object. The child is affected
primarily by the private stimuli generated
by the injury, which are completely
unavailable to the adult. The adult can,
nevertheless, react to the public visual .
features of the damaged tissue and the
public features of the child's crying.
Control by private stimuli also comes
about as the result of stimulus
generalization, whereby a response
learned with respect to public stimuli .
occurs simply because a private stimulus
has something in common with the public
one. For example, "sudden" may be
acquired underthe control ofpublic
visual or auditory events, and by stimulus
generalization be at some strength when a
private stimulus such as a stomach pain
shows the same temporal property.
This brief treatment cannot do justice
to the thoroughness of Skinner's
derivation of this basically simple notipn,
 42 Behavior Analysis: An Overview
nor Js there space for considering its
significance for understanding many
longstariding psychological and .
philosophical problems. One implication,
however, is that the language offeelings
can never be precise, and another is that
introspection is not a way of making
contact with basic causal variables, but is
only verbal behavior controll~aD.d not
well-by stimuli. Skimier's analysis of
the role of private stimuli and responses
as just more stimuli and responses to be
accounted for, rather than essential causal
variables underlying all behavior, is in
fact the defining feature of "radical
behaviorism" and has been the subject of
much discussion and debate. For
examp~e, every volume of Behaviorism
contains several articles directly related to
Skinner's analysis of private events.
Chapter 17 ('The Role of Private
Events in a Natural Science") of Science
and Human Behavior (1953) is an
expanded version of the 1945 article; the
same topic is considered in the chapter on
the tact in Verbal Behavior and is the
basis for Skinner's 1963 Science aniele,
"Behaviorism at Fifty." Most recently,
the role of private events was
systematically considered in chapters 1
and 2 of About Behaviorism (1974).
Although his analysis of private events is
considered by many to be one of
Skinner's most valuable contributions, it
would be a mistake to assume that for
Skinner, private stimuli are of great
importance in human life. For Skinner,
the main reason for dealing with private
stimuli was to put them in proper
perspective, which represents a
considerable reduction in their
significance from the essential causal role
they are thought to play in traditional and
common-sense accounts.
Rule-governed behavior. In
chapter 5 ("Operant Behavior") of
Contingencies ofReinforcement (1969),
Skinner introduces the distinction
between rule-governed and contingency-
shaped behavior (pp. 121-125). In
Chapter 6 ("An Operant Analysis of
Problem Solving"), the distinction is
made further use of (pp. 146-152), aJ:ld
in the notes to that chapter the distinction
is dealt with specifically and in
considerable detail (pp. 157~171).
Contingency-shaped behavior refers to a
repertoire developed as a result of
exposure to reinforcement contingencies
(involving stimuli, responses, motivative
variables, and effective consequences);
rule-governed behavior refers to a
repertoire developed as a result of
exposure to a verbal description of
contingencies (involving stimuli, etc.) but
not to the contingencies themselves.
Speaking one's native language is an
example of the former; the early phases
ofleaming a second language in school-
composing sentences by using a grammar
book and a bilingual dictionary-is an
example of the latter.
Skinner originally emphasized the
distinction in order to offset the
increasingly common practice (among
coguitive psychologists) of interpreting
contingency-shaped behavior as the
following of rules. This point is
especially clear in the chapter "Causes
and Reasons" in About Behaviorism.
"We do not need to describe
contingencies of reinforcement in order to
be affected by them. Lower organisms
presumably do not do so, nor did the
human species before it acquired verbal
behavior" (1974, p. 141).
Within the last ten years, it has
become increasingly clear that much
human behavior is rule governed rather
than contingency shaped. Whenever our
behavior is affected by consequences that
occur more than a few seconds after the
behavior, and where bridging stimuli are
not present, the effect cannot generally be
interpreted as the direct result of the
consequence, but ~s probably related to
our ability to generate and to be affected
by descriptions of contingencies. It thus
becomes incumbent upon us to deal more
specifically with such behavior and to
understand how it works in terms of
. elementary behavioral principles.
Skinner's earlier analysis of instruction in
Verbal Behavior (1957, pp. 357-367)
was a good beginning, and his treatment
of rule-governed behavior in

Contingencies ofReinforcement
acknowledged more explicitly the
importance of this type of behavioral
control and provided a technical tenn for
it. Others have begun to deal with the
issue (e.g., Hayes, Zettle, & Rosenfarb,
1982) and there is now an active area of
investigation, referred to as human
operant research, that is also much
concerned with the effects of rules
(provided by others or by the behaver) on
nonverbal behavior (e.g., Lowe, 1979).
In many respects this area, initiated or at
least emphasized by Skinner's
distinction, is the behavior-analytic
version of the cognitive behaviorist's
concern for such topics as modeling and
observational learning. It can also be
seen as an inevitable extension of basic
behavioral concepts and principles to
increasingly complex human behavior.
Design of a culture. Skinner's
anempts at cultural redesign are probably
bener known than most of his other
contributions. Walden Two, his utopian
novel, was published in 1948, and
Beyond Freedom and Dignity, a
nonfiction argument for the design of a
culture along behavioral lines, was
published in 1971. Both received
extensive popular press coverage. Less
well known are the last three chapters in
Science and Human Behavior (1953) and
numerous separate articles on the same
general issue, namely, that it is time to
abandon the traditional view of human
nature and deliberately design cultural
practices on the basis of the science of
behavior.
Although these writings are full of
intriguing suggestions for new and more
effective cultural practices, they are really
most important as a critique of existing
practices, and especially of the traditional
view of human nature that is used to
justify these practices. A scientific
approach that views human behavior as
the completely determined product of
innate endowment plus environmental
events is contrasted with a traditional
view of man as a self-determining
autonomous agent. Freedom of choice is
seen as an illusion based on failure to
Behavior Analysis: .!\nDverview 43
appreciate the nature of control by .
positive reinforcement-we tend to..
recognize control only when aversive
events. are the responsible variables.
Most importantly, our continuing failure
to accept the scientific view is responsible
for the majority of our current problems,
and unless we begin to design the culture
in order to achieve long-range ,
effectiveness or survival, it may soon
perish. These points are, of course, not
just asserted, but rather are argued in
meticulous and persuasive behavioral
detail.
In the process of making his general
point about the necessity of deliberate
cultural design unimpeded by the
traditional notion of freedom, Skinner
attacks the longstanding philosophical
practice of distinguishing sharply
between facts and values. Contrary to the
view that science is concerned with facts,
but that value judgments are the province
of some other kind of knowledge,
Skinner identifies the value judgment as a
particular type of behavior, and thus a
proper topic within the science of
behavior. This particular analysis is
clearly made in Science and Human
Behavior (pp. 428-436) and in Beyond
Freedom and Dignity (chapters 6, 7, and
8). Like Skinner's analysis of language,
his behavioral value theory is not easy to
understand on first contact, because it is
so different from the consideration of
values that is implied in everyday
language and that is a part of our religious
and legal heritage. But the argument is
logically tight, and the implication for
survival of the human species is quite
clear.
Conclusion
The approach to the prediction, control,
and understanding of human behavior
that has been described above should not
be thought of as static. Behavior analysis
is constantly changing in linle ways, and
every once in a while a big change---a
breakthrough-occurs. It is a
deterministic view that sees human
behavior as the inevitable product of
innate endowment and environmental
44 Behavior Analysis: An Overview
events taking place during the person's
lifetime. In many respects, it is scientific
method applied to all aspects of behavior.
This view is not concerned only with
operant conditioning. It does not exclude
private stimuli and covert behavior from
scientific consideration. It does not insist
that behavior can change only as a result
of direct exposure to contingencies, but .
readily acknowledges behavior change by
instruction, and by the description of
contingencies. It is not antiphysiological,
antigenetic, or antitheoretical (except with
respect to inferred inner explanations). It
is the science and technology of behavior.

4 !
CONTRAST BETWEEN THE PROCEDURES OF
RESPONDENT AND OPERANT CONDITIONING
It is typical for students to learn about
respondent conditioning in the context of
Pavlov's work with the dog's salivary
response, followed by a few other
examples involving smooth muscles and
glands, possibly the eye blink striped
muscle response, and possibly the
conditioned emotional response.
Similarly, operant conditioning is usually
introduced with reference to the rat's
lever pressing and the pigeon's key
pecking. I have found that the imponant
procedural features of the two different
kinds of conditioning are often so closely
linked to the specific learning contexts
that they cannot be described or
recognized outside of those contexts.
This paper is an attempt to disentangle
procedure from response mechanism by
deliberately attempting to apply each
procedure in a context more appropriate
to the other procedure.
Respondent Conditioning
Peripheral vasoconstriction (typically
measured with a plethysmograph) occurs
as a reflex response to a decrease in skin
temperature sufficient to stimulate cold
receptors. Many mammals, including
humans, have this unconditioned reflex.
To condition vasoconstriction to a tone,
we turn on the tone for a few seconds,
then direct a strearn of cold air against one
of the person's hands, turn off the tone
and the strearn of cold air, wait for a
while, then repeat the process, and so on.
.,il
The unconditioned stimulus (US) is the
cold air on the hand, because it elicits
vasoconstriction without the necessity of
a learning history. The unconditioned
response (UR) is vasoconstriction to the
cold air stimulus. The conditioned .
stimulus (CS) is the tone, because tones
do not ordinarily elicit peripheral
vasoconstriction, but do so only after a
history of pairing with the cold
stimulus-that is, they have to be
conditioned to have the effect. The
conditioned response (CR) is
vasoconstriction to the tone l .
Note that no aspect of the procedure
depends on the organism's behavior.
The procedure consists only of the
manipulation of stimuli, which ~e
presented according to a prearranged
temporal plan. The experimenter ddes
not have to monitor the organism's
behavior in order to know whether or not
1A more complete description of conditioning is
that US frequency in the the tone (US
occurrences per unit time) must be greater than in
the absence of the tone. This is sometimes
referred to as a contingency between US and CS.
In terms of pairing, the US must be paired with
the tone-on condition more than it is paired with
the tone-off condition, otherwise the change from
tone off to tone on will not constitute a change
with respect to US history. In other words, if
the absence of the tone elicits the CR as much as
the presence of the tone, then elicitation cannot
be detected (or will not occur) when the change
from tone off to tone on occurs.
 46 Respondenl-Dperant Contrast
to present any of the stimuli. There is no
response contingency in the procedure.
Neither the occurrence of the CS nor of
the US is dependent upon any particular
behavior by the organism. It is, of
course, necessary to measure the
response in some way to see if
conditioning is taking place, but this
doesn't constitute any aspect of the
conditioning procedure. .
Operant Conditioning
By contrast, now, consider typical
operant procedures, such as conditioning
a pigeon to peck a disk on the wall of its
chamber when a tone sounds, or
conditioning a rat to press a lever when a
tone souyds. In the former case, we
provide reinforcement by making the
grain hopper available for a few seconds;
in the latter, by operating a pellet feeder
that delivers a food pellet into the
organism's food cup. Note that for a
description of operant conditioning
procedures it is essential that the
reinforcement presentation depend upon
the organism's response. It is the
occurrence ofthe organism's response
that causes the reinforcement to be
delivered. The experimenter (or the
electronic apparatus) must monitor the
organism's behavior in order to present
reinforcement. In respondent
conditioning procedures, the conditioned
and unconditioned stimuli are presented
without regard to the animal's behavior.
In operant conditioning, we must detect
the response in order to know when to
deliver the reinforcement, not just to see
whether conditioning is taking place.
The Procedures Exemplified
It is important thal these procedures be
understood in their abstract form, and not
solely in the context of the particular .
typesef behavior being conditioned. For
this purpose it is useful to consider how
one IIiight try to use operant conditioning
to get a child to show 'a vasoconstriction
response (a brief period of
vasoconstriction) when a tone sounds;
and similarly how one IIiight try to
respondently condition a rat in the usual
operant conditioning chamber to press a
lever when a tone sounds. Of course,
given what is already known about
respondent and operant conditioning, we
would certainly not proceed in this way if
our goal was simply to bring the
respective behavioral changes under some
form of stimulus control. It would be
much more reasonable to use respondent
conditioning with vasoconstriction and
operant conditioning with the lever press.
The purpose of this exercise, however, is
illustrate the essential features of each
procedure, free from any particular
context.
Attempting to Operantly Condition
Peripheral Vasoconstriction
In the former case, one would have to
have available some form of
reinforcement-let us use a small piece of
food known to function as reinforcement
for a particular child when several hours
have passed since the last meal. Let us
assume that the condition of the
peripheral blood vessels shows some
variability from moment to moment even
when no experimental operations are
being performed. We should start with
the tone on and deliver the food (place it
in the child's mouth) whenever we detect
vasoconstriction. If this type of behavior
is subject to operant conditioning,
vasoconstriction should begin occurring
more often than it did before we started
reinforcing this response. After a few.
instances of vasoconstriction followed by
food reinforcement in the presence of the
tone, we should turn the tone off and
allow several instances of
vasoconstriction to occur without
reinforcement; then tum the tone on and
reinforce vasoconstriction; then tum it off
and extinguish vasoconstriction, and so
on. As a result of this training, (a) when
the child is food-deprived and (b) the tone
comes on, (c) vasoconstriction responses
should occur more often than when the
tone is off. In other words, if the child is
food deprived, the tone onset should
evoke vasoconstriction. We have
essentially developed the tone as a
discriIIiinative stimulus (an SD) for
vasoconstriction.

It may well be that peripheral
vasoconstriction is not subject to operant
conditioning-that this type of behavior
is not affected by its consequences, in
which case our effort will be
unsuccessful. Because vasoconstriction
consists of the behavior of smooth
muscles, and because most efforts to
operantly condition smooth muscle
systems have not been very successful, it
would be reasonable to expect failure.
Nevertheless, the operant conditioning
procedure can be applied to any response
system; it is importiUlt that the procedure
be understood for what it is, irrespective
of the response system it is being applied
to and irrespective of past or anti<;ipated
successes or failures in such application.
Note that it would be completely incorrect
to suggest using a cold stimulus, even
though we know that this would produce
vasoconstriction. In the context of the
present paper, causing vasoconstriction
with an unconditioned stimulus for the
vasoconstriction reflex has nothing to do
with operant conditioning.
Attempting to Respondently
Condition the Lever Press
Consider now the second task, trying to
condition a rat in an ordinary operant
chamber to press a lever when a tone
sounds, using respondent conditioning.
We would have to have available an
unconditioned stimulus that produced this
response as a part of a reflex. It is quite
likely that there is no such unconditioned
stimulus for lever pressing-that there is
no lever-pressing reflex in the rat's
innatelydetermined repertoire of
behavior. Still, our attempt can be
described as though there were. That is,
if we can fmd an unconditioned stimulus
for lever pressing, we have only to pair
this stimulus with the tone several times,
as with any respondent conditioning
procedure, and soon the tone should elicit
the lever pressing by itself. Note that it
would be completely inappropriate to
suggest providing any form of
reinforcement after a lever-pressing
response. Reinforcement in the sense of
Respondem-Operant Contrast 47
a .behavioral consequence has nothing to
do with respondent conditioning.
(Unconditioned stimuli are sometimes
referred to as "reinforcers" for the
respondent relatiqn, which was Pavlov's
usage, and which is quite compatible with
the dictionary definition of "reinforcer."
In technical usage, however, the term
now refers most often to any f
consequence of a response that increases
the frequency of that type of behavior as
an instance of operant oonditioning.
Using the same term for the
unconditioned stimulus of respondent
conditioning leads to considerable
confusion because the two procedures are
so different; thus such usage should be
avoided.)
Procedures Versus
Underlying Processes
This treatment of the two most basic
behavioral procedures is provided with
full knowledge that within the field of
behavior analysis, the validity of the
operant-respondent distinction is
questioned on theoretical grounds, and
that for those most concerned with inner
processes (e.g., cognitive psychologists)
the distinction may not have been of
major importance in the first place. Still,
as history has shown, any attempt to deal
with most human behavior as though it
were subject to principles of respondent
conditioning-any attempt to ignore the
role of behavioral consequences+-is
highly unsatisfactory, both theoretically
and practically. And although it is a less
serious error, the attempt to deal with
respondent relationships as though they
were a function of behavioral
consequences is equally unsuccessful.
Those who blur the distinction are usually
not primarily concerned with procedures,
because the procedures are easy to
distinguish. As to so-called underlying
processes, which may be the same for
both types of conditioning, they would
seem to be the subject matter of
physiology.
5
COMMON ERRORS IN THE USE OF BEHAVIORAL TERMS
Main Terms
stimulus. This is a singular form.
The plural is stimuli (last syllable either
rhymes with see or with pie, depending
on personal preference). Thus, "A
stimulus preceded the response."
"Several stimuli preceded the response."
elicit. To bring out, draw forth,
evoke. This term should not be confused
with illicit (illegal, unlawful), nor should
it be spelled with more than one 1. But
most important, in behavior analysis there
are two strict usage requirements: (a) The
only acceptable subject for this verb is the
singular or pluralform of stimulus or a
term referring to some particular type of
stimulus (light, tone, touch, etc,). It is
incorrect for an organism to be the subject
of elicit. Thus "The dog elicited
salivation" would only be correct in the
unlikely situation where the dog was a
stimulus for another organism, thus "The
dog drinking from the stream elicited
salivation OIfthe part of the watching
alligator." (I don't think alligators
salivate, but no matter.) (b) Elicitation is
relevant only to the respondent relation,
not the operant relation. Thus the only
kinds of stimuli that can elicit a response
are unconditioned or conditioned stimuli
of a respondent relation (a US or CS).
An SD is not said to elicit a response
(except by those unfamiliar with the
operant-respondent distinction).
emit. To release or send forth.
This term should not be confused with
omit (to leave out, fail to include), and it
should be spelled with only one m. But
most important, in behavior analysis there
are two strict usage requirements: (a) The
only acceptable subject for this verb is an
organism or a type of organism. A
common error would be to say that the
sound of the feeder emitted it salivary
response. Only an organism can emit a
response. (b) In addition, emit is used
only when referring to an operant
relation. Thus, 'The rat emitted a lever-
press response that operated the feeder;
the sound of the feeder elicited a salivary
response on the part of the rat" (Lever
pressing is the type of behavior that is
affected by its consequences, and the
rat's lever pressing is a part of an operant
relation involving such a consequence.
Salivation is part of a respondent relation,
and the sound of the feeder is a
conditioned stimulus based on some
unconditioned stimulus with which it was
paired.)
evoke. To call forth. In behavioral
usage, only a stimulus or type of stimulus
can be the subject of this verb, but the
term can be used for both respondent and
operant stimulus control. Thus behavior
can be evoked by a US, a CS, an SD, as
well as by unconditioned and conditioned
motivative variables. However, as with
elicit, organisms do not evoke anything in
technical behavioral use.
Note that all three of these verbs,
elicit, emit, and evoke, have response as
their object. The only things that are
elicited, emitted, or evoked in behavior
analysis are responses. A fl!irly common
 error consists in having a response elicit
or evoke a consequence of some kind.
This is not correct when all that is meant
is that the response resulted in or
produced a consequence. Onl~ stimuli.
elicit or evoke, and only orgamsrns ennt;
responses are the only things elicited,
evoked, or emitted.
Other Common Usage Errors
There are no circumstances under which
it is reasonable to refer to some procedure
as "reinforcing a stimulus."
Responses are the only things that are
reinforced.
Also, the expression behavioral
response sometimes appears, either in
contrast with physiological response,
or in implicit contrast wi~ mental . responses are always muscle action or
response. There are no CIrcumstances m
behavioral usage where either behavioral
response or physiological response are
appropriate terms, and these terms should
be avoided. Physiological response
sometimes seems to refer to the action of
smooth muscles or glands, possibly
because such ~sponses are often
measured with the methods of the
physiology laboratory. Behavioral
response, then, seems to refer to the
action of striped muscles, often measured
by psychologists in terms of their effects
on the environment But this is not a
reasonable contrast. Striped muscle
action is just as physiological as any other
bodily process and is studied by
physiologists. Neither is physiological
versus behavioral an adequate way to '
characterize the distinction between
respondent and operant conditioning
procedures. As to the contrast with
mental responses, because of the
vagueness of this .latter term, whatev~r
contrast is attempted can be accomplished
much more effectively in some other
way;
T6refer to behavior as being or
becoming "reinforcing" is inappropriate,
Only stimuli can function as forms of
reinforcement When behavior is saidto
, be reinforcing it may be that response-
produced stimuli are being referred to,
but if so, it is less confusing to refer to
II
i
"\i
those stimuli as the reinforcement, not the
responses that produced them.
It is sometimes said that "one
person's response is another person's
stimulus," but this is not correct. One
person's response consists of the action
of that person's muscles or glands, which
mayor may not result in stimuli that
affect another person. An ~ffective
repertoire in behavior analysis is unlikely
to develop if stimuli and responses are
not sharply distinguished from each
another.
Another source of possible
confusion is to use stimulus adjectives to
identify responses, as in the expressions
visual responses or auditory responses.
From a behavioral 'perspective, human
glandular secretion and are.not .
appropriately referred to WIth stunulus
terms. These expressions may be an
attempt to identify the stimuli leading to
the responses, in which case "responses
to visual stimuli" would be much clearer.
In a physiological context, visual
response may refer to some activity of the
visual receptor system or of the neural
connections to that system, but in a
psychological context the term is
arnbiguous.
Another error: There is some
tendency to refer to respondent functional
relations as innate, unlearned, phylogenic
in origin, and operant functional relations
as learned or ontogenic in origin. This is
quite incorrect, because the relation
between the previously neutral stimulus
and the response that it elicits after
respondent conditioning has occurred is
clearly a learned functional relation. This
usage probably arises from the fact that in
respondent conditioning, the response
itself was previously part of an unlearned
relation, and in this sense is an unlearned
response. But respondent conditioning
develops a new functional relation, a
learned relation between a previously
neutral stimulus and a type of response
that had previously been elicited by an
unconditioned stimulus.

Here is a summary of common
incorrect expressions. If you discover
any of them occurring in your repertoire
try to delete them. You will be a much
better person for having done so.
An organism elicited a response.
An SD elicited a response.
A stimulus emitted a response.
An organism emitted a smooth muscle or
glandular respo~e.
An organism evoked a response.
The response elicited or evoked a
reilfforcer.
The response elicited or evoked or
emitted a stimulus.
The organism elicited or evoked or
emitted a stimulus.
Training consisted in reinforcing some
stimuli and extinguishing others.
Behavioral responses are qffected by
consequences and physiological
responses are elicited by stimuli.
As a result ofthe training, the responses
became more reinforcing.
The organism's visual responses were
reinforced.
extinct. Inactive, as a fire or
volcano; no longer existing as a living
form; having died out (extinct animals,
such as dinosaurs). This term is not a
behavioral term at all. In particular, it is
not the past participle of the behavioral
verb, extinguish. It is not correct to say
that a response became extinct as a result
of nonreinfon:ement It is correct to say
that the response was extinguished as a
result of nonreinforcement
aversive and adversive. These
terms are a source of difficulty because
technical use in behavior analysis for
aversive has departed somewhat from
everyday use. Aversive is a behavioral
adjective (although it is sometimes used
as a noun) modifying stimulus. An
aversive stimulus is one whose
presentation evokes the behavior that
removes it (or whose onset weakens the
behavior that preceded the onset, or
whose offset strengthens the behavior
Common Errors 51
that preceded the offset). This use is n~t
listed in ordinary dictionaries, and the
adjective averse (usually with to) is not
equivalent to the technical aversive.
Adverse (with ad) has no technical use
in behavior analysis. It is a sort of
- synonym for hostile or opposed, and for
ulffavorable, thus somewhat related in
meaning to aversive, but the safest f
strategy in the context of behavior
analysis is to completely avoid its use.
Use only the term with nn d.
imply and infer. These two words
are quite valuable and many of you use
them fairly often. You may have leamed
them from context rather than by
deliberate addition to your vocabulary,
and from some contexts it is possible to
get their meanings confused. Infer-
almost always has as the subject of the
verb a person or group of persons. Only
persons can infer, that is, only persons
can draw conclusions or make
deductions. ImPly can also have a person
as the subject of the verb, who by their
verbal or nonverbal behavior states
indirectly, hints, or intimates something.
("She implied, but did not actually say,
that his behavior was inappropriate.")
But the subject of imply can also be a
person's behavior or any kind of event.
("Your hesitancy in answering my
question implies that you may not know
the answer." "This [situation] implies
that you probably have some higher
priorities than obtaining a good grade in
this course.") The most common itli.stake
is to use infer instead of imply in
sentences like the two preceding ones.
Response is never spelled with a c;
causal is not equivalent to casual;
anatomic, automatic, and autonomic
have very different meanings; withhold
and withdraw are not eqiJivaient
terms-to withhold something is to
refrain from providing ino an organism
that does not already have it, whereas to
withdraw something is to take it away
from an organism that does already have
it To withhold reinforcement when a
response occurs is the extinction
procedure, whereas to withdraw
reinforcement is a punishment procedure
52 Common Errors
(e.g., a child can be punished for
dangerous play with a particular toy by
withdrawing that toy-taking it out of the
child's hand). You might also work over
your use of affect/effect,
accept/except, sit/set, lie/lay.
(Correct use of these last two verbs is so
difficult that you are probably better off
just dropping them from your vocabulary
and using synonyms such as recline and
place.) Drop irregardless completely if
you haven't already done so.
Some Errors of Pronunciation
These are not behavioral or even
psychological terms, but psychology
students seem to use them often. The
Latin expression et cetera, usually
abbreviated etc., is often pronounced as
though it were spelled excetera which, of
course, is quite incorrect even though
common. The second phoneme is a t
sound, not an x or k sound. This error is
especially conspicuous when a person
really likes the expression and uses it
several times in succession for emphasis,
as "I'll bring the necessary tools-
hammer, pliers, excetera, excetera,
excetera." Another common error is in
.;
the pronunciation of the term asterisk,
usually evoked by the symbol *
indicating a footnote. You don't see the
word spelled out very much, but notice
carefully the last three letters. Many
people seem to believe that it is spelled
asterick. Another common error,
especially important to those of you who
aspire to success in the business
community, is to pronounce realtor (a
person who deals with real estate) as
though it was spelled re la tor. The first
two syllables have the same
pronunciation as the word real, namely
ree-ukl, thus ree-uhl-tor. Still another
common mistake is to pronounce
nuclear as though it were spelled 000
kyoo lur or nyoo kyoo lur instead of 000
klee uhr or nyoo klee uhr which are the
correct pronunciations (not
pron.Q]Jnciations). Finally, in this list of
local favorites, the word deteriorate
has five syllables when pronounced
correctly, thus dee-lir-ee-o-rae. It is not
pronounced dee-tir-ee-ate, except, of
course by the uninformed. (Note that
uninformed is not the same as
uniformed-isn't it a shame that every
little letter counts.)
6
EXP.LANATORY FICTIONS
Simple explanations usually consist of
describing a cause that is proposed to
account for some effect. Thus, when
one's car engine makes a loud clanking
noise, it may be proposed that the noise is
due to a worn or damaged main bearing.
The loud noise is the effect and the
damaged main bearing (or some aspect of
its action) is the cause. We will use these
terms, cause and effect without implying
anything more th31l a functional relation
between an independent variable (cause)
and a dependent variable (effect).
A primary goal of science is to
discover and describe the functional or
causal relations that permit the prediction,
control, and interpretation of dependent
variables in terms of independent
variables. However, in order to predict
an effect from a cause, or to control the
effect by manipulating the cause, or to
provide a valid interpretation of the effect
in terms of the cause, it must be possible
to observe the cause independently of the
effect. When the cause is just another
name for the effect the explanation is said
to be circular, and the supposed cause can
be called an explanatoryfiction. The
essential feature of a circular explanation
is that the cause is inferred from the same
information that constitutes the effect. Or
said another way, when a cause cannot be
observed independently of the effect for
which it is the explanation, such a cause
is an explanatory fiction.
Circular explanations and their
explanatory fictions are quite common in
dealing with behavior. This probably
comes abOut because it is quite useful to
identify and label relatively stable
characteristics of a person's behavior in
certain circumstances. The receiver of
such information benefits by being able to
anticipate and prepare for the relevant
behavior. For example, for a teacher to
be told that a child coruing into her class
has a hostile attitude toward authority,
that another is not very intelligent, and
that a third has a lot of musical talent will
permit the teacher to behave more
appropriately toward each child,
assuruing that the information is correct.
As a name for a consistent pattern of
behavior in reaction to certain
environmental circumstances "hostile
attitude toward authority" may be quite
reasonable. In such usage, attitude
simply refers to the fact that certain events
typically generate certain kinds of
behavior on the part of the child.. ;rhe
person giving the information to the
teacher may well think of attitude as an
internal quality that is causally
responsible for the behavior on the basis
of which the judgment about the attitude
was made, but no contact was ever made
with such an internal quality. The
judgment was based on the statements of
others or on observations of behavior
under certain environmental conditions.
Likewise, the teacher receiving the
information may think of the attitude as
an internal causal quality, but again, the
usefulness of the information is solely in .
terms of its relevance to environment-
behavior interactions.
 54 Explanatory Fictions
A problem with such terms arises,
however, when an explanation for
specific instances of environment-
behavior interactions is required. If it is
asked why the child is so uncooperative
when the teacher makes a simple and
reasonable request, and the answer is
"because of his hostile attitude," the
.attitude is no longer just a name for a
collection of environment--behavior
relations, but has become the explanation
for the relations. Attitude has become an
internal or mental event that supposedly
explains the environment-behavior
relation, but because the only evidence
for the hostile attitude is the very same
behavior that the attitude supposedly
explains, it is essentially an explanatory
fiction in'this usage.
In the case of athletic and artistic
performances, some individuals are
systematically better than average.
Ability or talent often appears as the
name for such unusual effectiveness, and
information couched in such terms may
be quite helpfnl to the recipient. In the
attempt to explain such unusual
effectiveness, it may be possible to
identify relevant environmental events. If
as a child the person had ample
opportuuity to participate in such
activities under highly effective training
conditions--parents, siblings, and
i'i friends were all competent musicians and
j'
I were highly supportive of the child's
! effons---the current effectiveness may
I seem quite understandable. However, if
II obvious environmental causes cannot be
found it is quite common for talent to be
proposed as the explanation, and thus to
function as an explanatory fiction. Why
does she play the violin so well? Why
does she learn new musical skills so
readily? Because of her outstanding
musical talent. As usual, the only
evidem;e fOf the talent is the unusual
musica1 performance, which is what
makes talent an explanatory fiction in
such usage.
'I
i!i The development of an effective way
r~ of talking about behavior has been made
more difficult by the fact that terms like
I
I
attitude and talent (and the thousands of
I
I
;:
I'"
I,
I'
Ii
'I
other similar terms) have both a useful
and a harmful application. As names for
relatively consistent environment-
behavior relations for particular persons,
they are useful; as explanations for the
environment--behavior relations, they are
almost always explanatory fictions. At
the present time, such fictions have
additional plausibility if they seem to refer
to coguitive or neural processes, or if
they have something to do with
computers.
Note that explanatory fiction is not a
general term for an inadequate
explanation, many of which are not
explanatory fictions at all. Most of us
would not consider a person's
astrological sign to be an adequate
explanation for her effective artistic
behavior, but the date of one's birth can
be observed quite independently of one's
artistic behavior. Also, historical
environmental causes, although often
quite useful, may be faulty (a) because
the events didn't actually occur or (b)
because of insufficient evidence for the
relation between those events and the
relevant type of behavior. Some aspects
of a person's behavior may be attributed
to the fact that his mother really didn't
want another child before he was born.
Information about what the mother
wanted may be inaccurate, however, and
even if accurate, the explanation suffers
because of the existence of many persons
with this same behavior whose mothers
wanted them and the existence of many
unwanted children without the behavior.
In respondent conditioning, it is very
common, but quite undesirable, to add
explanatory fictions to the basic
functional relations. After a dog has been
conditioned to salivate to a tone by
, pairing the tone with food, it is often said
that he salivates to the tone because of his
expectation that the food will follow,
,because of his knowledge of the tone-
food relation, or because he has formed
an association between the tone and the
food. The only evidence for the
expectation, knowledge, or association,
however, is the actual salivation when the
tone sounds, which is the effect we are

trying to explain. (Sometimes the
explanatory fiction is in verb fonn: the
dog expects, knows, or associates; but it
is still an explanatory fiction.) It is quite
sufficient to state that he salivates to the
tone because of the relevant history,
namely the past correlation of tone with
food, with further reference to the laws of
respondent conditioning, of which the
present case is a specific instance.
Expectations, knowledge, and
associations are simply inventions that
add nothing to our understanding of the
rellitions between environmental
procedures and the resulting changes in
behavior. Such explanatory fictions are
even worse than useless, however,
because they suggest to the listener (and
also to the user) that some additional
knowledge is available. Focusing on
inferred inner causes also serves to
distract us from a careful study of
environment-behavior relations.
Of course, the conditioning
procedure achieves its effect on current
behavior by producing physical changes
in the organism that persist over time, and
it is these physical changes that are
directly responsible for the changed
behavior. Unfortunately, there is at
present no physiological knowledge that
supplements the knowledge that has been
and is still being obtained by direct study
of environment-behavior relations.. And,
even when more is known about the
physiological changes underlying
respondent conditioning, the concepts
and principles of conditioning, which
constitute an increasingly extensive and
precise body of knowledge, will not be
contradicted or in some sense rendered
useless. It is, in fact, these concepts and
principles that can guide the search for
physiological understanding-they
constitute what it is that one is trying to
understand in physiological tenns.
Explanatory fictions are also popular
in explaining the facts of operant
conditioning, although here they are more
complicated because both motivation and
leaming must be dealt with. After
reinforcing a food-deprived rat's lever
pressing with food, students often
Explanatory Fictions 55
explain the animal's behavior as a joint
function of its desire for food and its
knowledge that lever pressing will result
in food. delivery. The desire is inferred
from the deprivation regimen, and the
knowledge is inferred from the lever
pressing. Without explanatory fictions,
the facts can be adequately dealt with as
follows: When some particular behavior
on the part of a food-deprived organism
is reinforced with food, that type of
behavior becomes a part.of the repertoire
that occurs more frequently under food
deprivation. This is, of course, a
statement of the principle of operant
conditioning. So why does the animal
press the lever if not because of its
desires and knowledge? Because (a) we
previously reinforced lever pressing with
food when the animal was food deprived,
and (b) the animal is currently food
deprived. It will also be necessary to
make reference to the laws of operant
conditioning, of which the present case is
a specific instance.
The explanatory fiction knowledge is
sometimes varied by substituting
understanding, expectancy, association,
cognition, belief, and more recently a
cognitive structure is said to be relevant
Need is often mentioned instead of -
desire, and it might seem to have'
biological support; the rat needs food and
presses the lever because if it doesn't it-
will die of starvation. This use of need is
an example of another type of inadequate
explanation: explaining some current
event in tenns of a future event, which is
called a teleological explanation. A future
event that has not occurred cannot be
considered to be an explanation of
anything. The rat's starvation, should it
occur, is clearly in the future of our
current lever presser, and cannot playa
role in determining the current pressing.
Of course, the fact that food deprivation
causes an appropriate change in the
animal's repertoire is relevant to the
evolution of such an animal. Those
whose behavior is fine-tuned to biological
needs must certain!y be more likely to
survive and pass on the relevant genes to
their offspring. The behavior of the
56 Explanatory Fictions
present lever presser, however, must be
explained in terms of the functional
relations descriptive of its behavior, not
the possible future results of such
behavior.
Operant behavior, by defmition
affected by its consequences, is especially
susceptible to teleological explanation.
The syntax usually involve.s the tenn to or
in order to, as when the rat presses the
lever to get (or in order to get) the water
or food that is being used as
reinforcement. Of course, the proper role
of consequences in explaining an instance
of behavior is in tenns of the past
consequences of that behavior, not the
consequence that may occur after that
instance of behavior has taken place. The
inadequacy of teleological explanations is
widely recognized, however, and when a
careless instance is questioned, the user
will quickly come up with an explanatory
fiction; which seems to solve the
problem. A student who says that a rat is
pressing the lever in order to get
reinforcement, when asked why the
pressing occurs before the reinforcement
has been obtained, or during extinction,
can easily reply that it is the rat's cu"ent
beliefthatit will be reinforced that is the
actual cause. This is an explanatory
fiction, but at least it is not teleological.
Current beliefis no real improvement, but
because its inadequacy is more complex,
it may well go unchallenged.
As in the respondentcase, the
procedure of operant conditioning must
surely alter the organism in some
relatively lasting way, and it is this
physical alteration that is directly
responsible for the fact that current
behavior differs from what would have
been seen if the operant conditioning had
not taken place. We look forward
enthusiastically to advances in the
neurosciences that will permit us to
observe and even measure these relevant
physical events within the organism. As
with respondent conditioning, however,
no useful physiological supplement to the
knowledge obtained from the direct study
of environment-behavior relations is
currently available.
Explanatory fictions continue to crop
up as we deal with more complex
relations between the environment and
behavior (such as generalization,
discrimination, abstraction) and serve the
same nonpurposes. Probably the safest
practice in dealing with the behavior of
. any nonhuman species is to avoid all
terms that refer to inner mental qualities .
or activities. Certainly one should avoid
those identified above and their various
synonyms. But what about human
behavior? The behavioral position is that
such tenns are also explanatory fictions in
relation to human behavior, but the
argument becomes complex because of
the introspective reality of such terms to
the individual whose behavior they
supposedly explain.. This is a topic that
requires an analysis of the type of verbal
behavior that is under the control of .
private stimuli.
7 f
ESTABLISHING OPERATIONS
In common-sense psychology, what a
person does is generally thought to be a
function of two broad factors, knowledge
and motivation. For any particular
behavior to occur (except for
"involuntary" acts such as reflexes), the
behaver must know how and must also
want to do it. A· good deal of traditional
psychological theory concerns the
different kinds of wants and the way they
interact with other mental functions; much
of applied psychology is concerned with
getting people to do things that they know
how to do but don't want to do.
Motivation seems to be an important
topic, yet the basic notion plays only a
small role in the approach currently
referred to as behavior analysis.
In applied behavior analysis or
behavior modification, the concept of
reinforcement seems to have taken over
much of the subject matter that was once
considered a part of the topic of .
motivation. To some extent this is a
reasonable replacement. With the
discovery of the role of reinforcement in
the. maintenance of behaviOt-schedules
of intermittent reinforcement-many
examples of insufficient motivation could
be better interpreted as examples of
insufficient ongoing reinforcement. The
replacement was also attractive because
the more common motivational tenns--
wants, needs, drives, motives-usually
referred to inner entities whose existence
and essential features were inferred from
the very behavior that they were
~upposed to explain.
ReiIiforcement history is. not,
however, a complete replacement for
motivative functional relations. Skinner
(1938, chapters 9 and 10; 1953, chapter
9) clearly distinguishes deprivation and
satiation from other kinds of
environmental variables and relates these
operations to the traditional concept of
drive, as did Keller and Schoenfeld
(1950, chapter 9).. Skinner's treatment of
aversive stimulation (1953, chapter 11) is
very similar to his treatment of
deprivation, and Keller and Schoenfeld
classify aversive stimulation as one of the
drives (1950, chapter 9). Later, in his
treatment of verba1behavior (1957, pp.
28-33; 212-214) Skinner again identifies
deprivation and aversive stimulation as
independent variables that are quite· .
different in function from reinforcement
and stimulus contro1. l
Subsequent behavioral texts at first
continued to provide a separatechapter on
deprivation (Holland & Skinner, 1960;
Lundin, 1961, 1969; Millenson, 1967,
Millenson & Leslie, 1979), but more
recent texts have almost dropped the topic
(e.g., Catania, 1979, 1984; Fantino &
Logan, 1979; Mazur, 1986, 1990;
Powers & Osborne, 1976). The
handbook by Honig (1966) and the later
1The topic of emotion is closely related to
motivation in these treatments, either as an
adjacent chapter (chapter 11 in Skinner, 1938;
chapter 10 in Keller and Schoenfeld, 1950;
chapter 10 in Skinner, 1953) or as part of a
group of closely related independent variables (as
in Skinner, 1957).
II'I
I'
" 58 Establishing Operations
:1
one by Honig and Staddon (1977) each
contain a chapter on motivation by
Teitlebaum (1966, 1977), which differ
from the earlier treatments in being
concerned largely with relations between
physiological variables and behavior. The .
Honig and Staddon handbook also
contains a chapter by Collier, Hirsch, and
Kanarek (1977), in which feeding
behavior is analyzed in the 'context of its
ecological significance. Like that of
Teitlebaum,this is a very different
approach from the earlier ones, and one
that is to some extent critical of some of
the assumptions about motivation in
Skinner's earlier treatments. Neither the
physiological nor the ecological type of
analysis seems to have been incorporated
in the mote recent "nontreatments" of the
topic of motivation.
The present state of affairs, with
motivative variables being dealt with as
reinforcement history, deprivation and
satiation, or aversive stimulation, is not
entirely satisfactory, however. 2
Variables with behavioral effects like
those of deprivation and aversive
stimulation, but that cannot be easily
classified as either, are likely to be
ignored or misclassified (usually as
discriminative stimuli). What follows is
an attempt to provide a more thorough
and systematic approach than usually
appears, and one that corrects this latter
difficulty.
,, Establishing Operation
Defined in Terms of Two
:1
Ii Features
'i An establishing operation-the tenn was
first used by Keller and Schoenfeld
(1950) and later by Millenson (1967)---is
an environmental event, operation, or
stimulus condition that affects an
organism by momentarily altering (a) the
reinforcing effectiveness of other events,
2Kantor's selling/actor (1959, p. 14) includes
motivative variables but until recently this
concept has not been much used within the
behavioral community (but see Morris, Higgins,
& Bickel, 1982, especially pp. 161 and 167 for a
contradictory view).
and (b) the frequency3 of occurrence of
the type of behavior that had been
consequated by those bther-events.
The first effect can be called
reinforcer establishing and the second
evocative. Thus food deprivation is an
establishing operation (EO, read as ee oh)
that momentarily increases the
effectiveness of food as a fonn of
reinforcement But food deprivation not
ouly establishes food as an effective fonn
of reinforcement, if the organism should
encounter food; it also momentarily
increases the frequency of the types of
behavior that have been previously
reinforced with food. In other words, it
evokes any behavior that has been
followed by food reinforcement. This
evocative effect is probably best thought
of as (1) the result of a direct effect of the
EO on such behavior, (2) an increase in
the evocative effectiveness of all SDs for
behavior that has been followed by food
reinforcement, and (3) an increase in the
frequency of behavior that has been
followed by conditioned reinforcers
whose effectiveness depends on food
deprivation. Food satiation-consuming
food-is an EO working in the opposite
direction, and it is actually more accurate
to think of motivative variables as
establishing or abolishing operations, and
to think of their evocative effect as either
an increase or a decrease in the
momentary or current frequency4 of the
relevant kind of behavior.
3In this contextJrequency should be taken to
mean both number of responses per unit time and
relativejrequency, the proportion of response
opportunities in which a response occutred. This
usage makes it possible to avoid such
controversial terms as response strength and
response probability.
4A change in the momentary or current frequency
of all behavior that has been followed by a
particular type of reinforcement is to be
contrasted with a change in thefuture frequency
of the particular type of behavior that preceded a
particular instance of reinforcement Changes in
future frequency define function-altering relations
(reinforcement, punishment, extinction), whereas
changes in current frequency define evocative
relations (the effects of a discriminative stimulus

An effort is sometimes made to
interpret the evocative effect of an EO as
the result of stimuli produced by the
relevant deprivation (hunger pangs, dry
mouth and throat) functioning as
discriminative stimuli (SOs) for the
relevant behavior (e.g. Staats and Staats,
1963, pp. 111-112). At this point it will
be useful tQ provide a somewhat more
restrictive definition of the discriminative
relation than usually appears in basic texts
and to contrast this relation with that of
the establishing operation.
An So is a stimulus condition that
has a history of correlation with the
differential availability of an effective
form of reinforcement given a particular
type of behavior. Differential availability
implies that the relevant consequence has
been in some way more available in the
presence than in the absence of the
stimulus condition or So. Lesser
availability or unavailability in the
absence of the stimulus condition further
implies that the unavailable event would
have been effective as reinforcement if it
had been obtained. It is with respect to
this latter requirement that most
motivative variables fail to qualify as
discriminative variables. The ordinary
procedure for developing an So, as when
a rat's lever-pressing behavioris brought
under the control of an auditory stimulus
in a laboratory setting, consists in
allowing reinforcement for lever pressing
in the presence of the stimulus, but
extinguishing the response in the absence
of the stimulus-letting the response
occur but not following it with
reinforcement. Although this point is
seldom made, the extinction condition
always occurs at a time when the
unavailable consequence would be
effective as reinforcement if it were
obtained. The unavailability of an
ineffective consequence is a behaviorally
neutral event It doesn't constitute an
and of the various kinds of establishing
operations). For a detailed treatment of this
distinction, see the paper in this collection titled
"Evocative versus function-altering effects of
enviromnental events."
Establishing Operations 59
extinction condition, nor does it
contribute to a correlation with differential
availability. While ids true that food
reinforcement is in a sense unavailable in
the absence of the stimuli that we call
hunger pangs, this is not the kind of
unavailability that contributes to a
correlation. This issue is also critical for
an understanding of the role of pairtful
stimulation as an EO, as discussed later.
To summarize, a useful contrast can
usually be made as follows:
Discriminative variables are related to the
differential availability of an effective
form of reinforcement given a particular
type of behavior; motivative variables are
related to the differential reinforcing
effectiveness of environmental events.
It seems that most events,
operations, or stimuli that alter the
reinforcing effectiveness of other events
also alter the momentary frequency of .
occurrence of any behavior that has been
consequated by those other events. Why
this should be the case is probably not the
type of question that can be answered by
staying within the science of behavior,
but can eventually be understood in
neurophysiological terms or in terms of
evolutionary biology. Although the
reinforcer-establishing effect is important
for identifying th.e various EOs and.
distinguishing them from other types of
variables, it is the evocative effect of the
EO that is most consistent with common-
sense concepts of motivation and,with
motivational concepts appealed to in other
psychological theories. To "want"
something can be most easily
behavioralized in terms of the momentary
frequency of the behavior that has
typically obtained whatever is wanted-
the evocative EO effect. That the thing
wanted will function as more effective
reinforcement when and if it is
obtained-the reinforcer-establishing EO
effect-is not a convenient behavioral
interpretation of "want," because it refers
to an event that is in the future with
respect to the time the "want" is
observed.
60 Establishing Operations
Unconditioned Establishing
Operations (UEOs)
For all organisms there are events,
operations, and stimulus conditions
whose rei1!forcer-establishing effects are
unlearned. They depend upon the
evolutionary history of the species, and
vary from species to species. Note that it
is the unlearned aspect ofthe
reinforcer-establishing effect that results
in an EO being classified as
"unconditioned." The behavior evoked by
the EO is usually learned. Said another
way, we are probably born with the
capacity for our behavior to be more
reinforceable by food as a result of food
deprivation and more reinforceable by
pain cesSation as a result of pain onset,
but we have to learn most of the behavior
that produces food and terminates pain
and is thus typically evoked by these
EOs.
Deprivation and Satiation
Humans and related mammals are
equipped with a number of unconditioned
establishing operations (UEOs, read as
yew ee ohs) consisting of some form of
deprivation. Food, water, activity, and
sleep are things that these animals can be
deprived of with resultant reinforcer-
establishing and evocative effects. Food
and water consumption, engaging in
activity, and sleeping are operations that
work in the opposite direction. Salt
ingestion, perspiration, and blood loss
are operations that have much the sarne
effect on behavior as water deprivation-
water becomes more effective as a form
of reinforcement, and behavior that has
been reinforced with water becomes
momentarily m~ frequent.
Temperature Changes
A decrease in skin temperature or in
genemI body temperature below normal
values increases the reinforcing
effec~veness of a change in the opposite
direction, and more remotely of warm
objects, places, and clothing; and also
increases the momentary frequency of the
behavior that has accomplished such
changes. An increase above normal
values is an EO that has
reinforcer-establishing effects in the
opposite direction and evokes the
behavior that has resulted in a
body-cooling effect. These EOs could
also be dealt with under the category of
aversive stimulation, which would
include painful stimulation, but it may be
clearer to consider them separately,
especially considering the problems of
aversive stimulus as an omnibus term, as
discussed later.
Variables Relevant to Sexual
Reinforcement
For many nonhuman mammals,
hormonal changes in the female are
triggered by time passage, arnbient light
conditiO/IS, daily average temperature, or
other features of the environment related
phylogenically to successful
reproduction. These environmental
features, or the hormonal changes
themselves, can be considered UEOs in
making sexual contact with the male an
effective form of reinforcement for the
female, and simultaneously causing the
production of chemical (olfactory)
attractants that function as EOs for the
male. The various hormonal changes may
also elicit (or sensitize for easier
elicitation by other stimuli) certain
behaviors (e.g., the assumption of
sexually receptive postures) that then
function as UEOs or as respondent
unconditioned elicitors (UEs)5 for sexual
behavior by members of the opposite sex.
Superimposed on this collection ofUEOs
and unconditioned elicitors (UEs) is a
deprivation effect that may also function
as a UEO.
In the human, the situation is quite
complex and not well understood. The
behavioral role of hormonal changes in
the female is not clear, nor is the role of
chemical attractants in changing the
male's behavior. There do seem to be
5For the distinction between unconditioned
elicitor (UE), unconditioned conditioner (UC),
conditioned elicitor (CE) and conditioned
conditioner (CC), see the paper in this collection
titled "Evocative versus function-altering effects
of environmental events."

deprivation effects for both sexes. In
.
addition , tactile stimulation of erogenous
regions. of the body seems.to.functIon as
a UEO in making further s~ar
stimulation even more effective as
reinforcement and in evoking the
behavior that has in the past achieved
such further stimulation. But the possible
role of learning in these vanous relations
cannot be ruled out
Painful Stimulation: Escape
The onset of painful stimulation
establishes the reduction or offset of this
stimulation as an effective form of
reinforcement and evokes the behavior
that has achieved such reduction or
offset That painful stimulation is a UEO
rather than an SD is not well appreciated.
The issue can be most easily analyzed by
reference to a typical laboratory
shockcescape procedure. The response
that turns off the shock (often a lever
press) is clearly evoked by the shock
onset, and it is also clearly a part of an
operant rather than a respondent relation
because it was developed through the use
of shock offset as reinforcement. If the
only known operant evocative relation
were the SD, then shock onset would
seem to be an SD; but like the stimuli
called hunger pangs, it fails to qu~fy as
a discriminative variable because Its
absence has not been a condition where
an effective form of reinforcement was
unavailable for a particular type of
behavior.
To repeat the argument, an SD is a
stimulus condition that has been
correlated with the availability of a type of
consequence given a type of behavior. A
correlation with availability has two
components: An effective consequence
, (one whose EO was in effect) must have
,:
followed the response in the presence of
the stimulus' and the response must have
occurred wiihout the consequence (which
would have been effective as ' (unconditioned conditioner) such as food,
reinforcement if it had been obtained) in
the absence of the stimulus. The
correlation between painful stimulation
and consequence availability fails in the
second component In the absence of
Establishing Operations 61
shock, there is no effective consequence
that could have failed to follow the
response in an analog to the extinction '
responding that occurs in the absence of
an SD. In the absence of shock, the
relevant establishing operation is absent.
The fact that the lever press does not turn
off the "nonpresent" shock is in ~o sense
extinction responding, but rather IS I
behaviorally neutral, like the '
unavailability of food reinforcement when
one is food satiated. Coutrast this
situation with extinction when an SD is
absent-the food-deprived rat presses the
lever but fails to receive food, which
would at that moment be effective as
reinforcement if it were obtained. The
absence of the shock is more like the
absence of food deprivation than like the
absence of the SD. With no shock
present, shock termination is not an
effective fonn of reinforcement.
Note that in addition to its EO
evocative effect, the onset of a painful
stimulus also functions as a respondent
UE (unconditioned elicitor) in eliciting a
number of smooth muscle and gland
responses such as increased hero: rate,
pupillary dilation, adrenal secretion, etc.;
as a respondent UC (unconditioned
conditioner) in conditioning these
responses to other stimuli present at the
time; as an SP (it is convenient to use the
symbol SR for unconditioned
reinforcement and sr for conditioned
reinforcement; similarly SP for '(
unconditioned punishment and SP for
conditioned punishment) in decreasing
the future frequency of any type of
behavior that precedes it; possibly as an
SD for any response that has been
correlated with the availability of some
form of reinforcement other than pain
reduction (e.g., lever pressing for food);
and possibly as a respondent CE ,
(conditioned elicitor), if pain has been ,
paired with some other UC
in which case pain might elicit salivation
(this would probably work only with
mild pain because intense pain as a UE
would probably elicit autonomic activity
that is incompatible with salivation).
 62 Establishing Operations
It is possible that in general there is a
close correlation among several of these
functions. Maybe any stimulus, the onset
of which can function as a VEO in
evoking its own termination, will also
function as a VE and a UC with respect
to certain·smooth muscle and gland
responses (the activation syndrome, for
example), and will also function as an SP
in weakening any response that precedes
its onset This seems to be the implication
of much current use of the term "aversive
stimulus," where the specific behavioral
function is not identified. It is not dear at
present just how close the correlation
among these functions is, nor is it clear
that the advantages of an omnibus term of
this sort outweigh the disadvantage of its
lack cif specificity. It is clear that some
use of this term is simply a behavioral
translation of common-sense expressions
for "feelings," "states of mind," and so
on, an undesirable usage that is fostered
by the term's lack of specificity, The
same problem arises when
"reinforcement" is used without the
implication of a strengthening effect on
preceding behavior, but rather as a
synonym for "pleasant" or "desired" as in
the too often heard "That's very
reinforcing! "
To keep the term reinforcement
specific to its strengthening function and
still have an omnibus term for positive
events one might use appetitive as the
opposite of aversive. In this usage, an
appetitive stimulus or condition would be
one that elicits certain smooth muscle and
gland responses, conditions neutral
stimuli so that they elicit similar
responses, increases the future frequency
of the type of behavior preceding its onset
(SR), suppresses the behavior that
removes it (VEO), and decreases the
future frequency of the behavior that
pre9edes its termination (SP). This use of
appetitive is often presented in basic
textbooks, but seems not to be used much
in the area of behavior analysis as
compared with aversive, possibly
J>e<:ause it still seems to refer primarily to
eatlng.
Maybe the omnibus usage is
essential for effective behavioral
discourse, or maybe it can be and should
be avoided in favor of more specific
terms. In any case, proper use of aversive
stimulus is often problematic, and
appetitive doesn't seem to be catching on,
but this is not the place to do other than
identify the difficulty.
Unconditioned Reinforcement
It is common in introductory behavioral
treatments to identify environmental
changes that function as reinforcers in the
absence of any particular learning history
as unconditioned reinforcers (SRs). They
are contrasted with conditioned
reinforcers (srs), whose reinforcing
effectiveness depends upon some
association with unconditioned
reinforcers. Unconditioned establishing
operations are the environmental events,
operations, or stimulus conditions that
regulate the momentary effectiveness of
unconditioned reinforcers (and also the
momentary effectiveness of the
conditioned reinforcers based on those
unconditioned reinforcers). The VEO and
the SR are obviously closely related, but it
would be a mistake to consider the terms
synonymous. Food deprivation is the
VEO that regulates the effectiveness of
food as an SR. The list ofVEOs above
implies a corresponding list of
unconditioned reinforcers.
It is possible that there are forms of
unconditioned reinforcement that are not
related to any particular VEO. For
example, infants' behavior seems to be
reinforceable by a variety of.mild
stimulus changes 6 (Finkelstein and
Ramey, 1977; Kalnins and Bruner, 1973;
McKirdy and Rovee, 1978; Rovee and
Fagen, 1976; Watson, 1967; Watson and
Ramey, 1972) not easily related to any
obvious EO. The concept of intrinsic
motivation, for example, as proposed by
Deci and Ryan (1985, chapters 1 and 2)
seems to imply that "signs of
6 I am indebted to Henry Schlinger for reminding
me of this point and for directing my attention to
the references shown.

competence" and "signs of
self-detemJination" always function as
effective fonns of unconditioned
reinforcement, although they can be
temporarily weakened by the presence of
other UEOs and other momentarily strong
forms of reinforcement. What such signs
might consist of for the untrained
organism is not clear, and current
evidence for such reinforcement is
equally well interpreted in tenns of
Skinner's concept of generalized
conditioned reinforcement (1953, pp. 77-
81). However the issue is essentially an
empirical one, and either interpretation is
compatible with a behavioral approach to
motivation.?
Painful Stimulation: Aggression
In addition to establishing its own
reduction as a fonn of reinforcement and
evoking the behavior that has produced
such reduction, painful stimulation in the
presence of anothe): organism evokes
aggressive behavior toward that
organism. Skinner, in Science and
Human Behavior (1953), discussed this
effect in attempting to explain the
responses that vary together in emotion.
He identified two bases for such
covariation.
Responses which vary together in
an emotion do so in part because of a
common consequence. The responses
which grow strong in anger inflict
damage upon persons or objects. This
process is often biologically useful
when an organism competes with other
.organisms or struggles with the
inanimate world. The grouping of
,- responses which defme anger thus in
part depends upon conditioning.
Behavior which inflicts damage is
reinforced in anger and is subsequently
controlled by the conditions which
control anger. (163)
This is covariation based on
ontogenic factors or on a history of
reinforcement. Phylogenic factors may
also be involved.
?For a detailed behavioral analysis of the topic of
intrinsic motivation see Dickinson (1989).
Establishing Operations 63
Some of the behavior involved in
. an emotion is apparently unconditioned,
however, and in that case the grouping
must be explained in terms of
evolutionary consequences. For
example, in some species biting,
striking, and clawing appear to be
strengthened during anger before
conditioning can have taken place.
(163--164) f
About 10 years after Science and
Human Behavior was published, Ulrich
and Azrin (1962) first reported their
discovery of the phenomenon that came
to be known as elicited aggI'!:ssion: the
unleamed occurrence of biting, striking,
etc. as a result of painful stimulation,
foun~ to occur in a wide variety of
species. It thus appears that one of the
effects of painful stimulation is simply to
elicit aggressive behavior, as an
unconditioned response. But the issue is
still not entirely clear, because some or all
of the effect may be the result of pain as a
~O (rather than as a UE), where pain
mcreases the effectiveness of some form
of reinforcement specific to that EO, such
as signs of damage to another organism ..
or the feel of one's teeth being pressed
against something. The aggressive
behavior. would thus be an example of
what Skinner later referred to as·an ..
intermingling of phylogenic and
ontogenic contingencies (1974, pp. 45-
50), where the EO has its reinforcer-
establishing effect, which then leads to a
fonn of rapid operant shaping o( the .
appropriate behavior. These various
mechanisms needn't be mutually .
exclusive, and there is considerable
evidence for multiple provenances for
these kinds of behavior. .
Other Emotional EOs
The above analysis of painful stimulation
as an EO with respect to aggressive
behavior can probably be extended to the
other operations that produce so-called
"emotional behavior" or "emotions." In
Sc~ence and Human Behavior (1953),
Ski~er describes the operant aspect of
emollon as a predisposition, as follows: .
The "angry" man shows an
increased probability of striking,
 64 Establishing Operations
insulting, or otherwise inflictiog injury
and a lowered probability of aiding,
favoring, comforting, or making love.
The man "in love" shows an increased
tendency to aid, favor, be with, and
caress, and a lowered tendency to injure
in any way. "In fear" a man tends to
reduce or avoid contact with specific
stimuli-as by running away, hiding,
covering his eyes and ears; a\ the same
time he is less likely to advance toward
such stimuli or into unfamiliar territory.
These are useful facts, and something
like the layman's mode of classification
has a place in a scientific analysis. (p.
162)
He doesn't specifically identify the
reinforcer-establishing effect of the EO
here, but in the previous passage this
implicatiotl: is quite clear. Of course, as he
points out in discussing "the total
emotion" (p. 166), one must add the
respondent UE and DC effects of
emotional operations, and we should also
add the possible effects of such
operations as unconditioned
reinforcement or punishment for the
behavior that precedes the occurrence of
the Operation, for a complete picture of
the "total emotion" (Skinner, 1953, p.
166).
EOs and Punishment
IfEOs must be in effect for events to
function as reinforcement, it is reasonable
to consider their function with respect to
punishment. Other things being equal, an
increase in painful stimulation seems to
function as punishment without any
.establishing operation, but those fonns of
punishment consisting of the withdrawal
or removal of reinforcing events, would
only.be expected to function as
punishment-.that is, would only
decrease the future' frequency of the kind
of behavior that they followed-if those
events were effective as reinforcement at
the timd" they were withdrawn. It is not
punishment to take food away from a
food-satiated organism.
The evocative effect is more
complex. Suppose food removal from a
food-deprived animal was used as a form
of punishment for some particular type of
behavior (which was being reinforced in
some other way-for example, by access
to sexual stimulation). 'We must assume
that the current weakening effect of this
history of punishment would be seen
only when the organism was food
deprived. TIlis relation has not received
much theoretical or experimental
attention, but seems to follow naturally
from existing knowledge of
reinforcement, punishment, and
establishing operations. (Note that this
relation makes the detection of a
punishment effect doubly complex: The
effect of punishment will not be seen
unless the EOs relevant to the .
reinforcement maintaining the behavior
and to the punishment weakening it are
both in effect; in the example above,
unless sexual stimulation and food are
both effective as reinforcement)
A Respondent Analogy
As described above the EO is smctly an
operant evocative relation, but there may
well be a respondent analogy. Food
deprivation probably increases the
evocative and function-altering
effectiveness offood as a stimulus (that
I
is, as UE and DC) as well as of stimuli
that have been correlated with food (as
CEs and CCs). Sitnilarrelations may
exist for many of the unconditioned
reflexes, in which case respondent, like
operant evocation, should be
conceptualized as jointly controlled by EO
and UE (or CE).8
Conditioned Establishing
Operations (CEOs)
Ordinary fonns of conditioned
reinforcement do not require a special EO
for their effectiveness; the UEO
appropriate to the relevant unconditioned
reinforcement is sufficient. In other
words, many learned fonns of
reinforcement do not require learned EOs.
But there are variables that alter the
reinforcing effectiveness of other events
81 am grateful to Michael Commons for pointing
out the possible relevance of the EO concept to
respondent relations.

but only as a result of the individual
organism's history. These are learned or
conditioned establishing operations
(CEOs)9. As with the UEOs, they also
alter the momentary frequency of the type
of behavior that has been reinforced (or
punished) by those other events. There
are at least three kinds of CEO. They are
all stimuli that were motivationally neutral
prior to their relation to another EO or to a
form of reinforcement or punishment.
They differ in terms of the nature of their
relation to the behaviorally significant
event or condition. The simplest relation
is a correlation in time; the neutral event is
paired with or systematically precedes a
UEO (or another CEO). As a result of
this pairing the neutral event may acquire
the motivational characteristics of the
UEO that it is paired with. I refer to this
as a surrogate lO CEO.
A more complex relation is where a
stimulus systematically precedes some
form of worseningll , and if the stimulus
is terminated prior to the occurrence of
this worsening, the worsening does not
9In Michael (1982), I suggested establishing
stimulus (SE) for a learned motivative relation,
with establishing operation (EO) referring to the
unlearned relation. It now seems that conditioned
establishing operation (CEO) works better
because of the easier contrast with unconditioned
establishing operation (UEO). This approach
also leaves establishing operation (EO) as a
useful term for the general motivative relation,
without specifying provenance.
l~e term surrogate was suggested by Michael
Urbach.
11I use the term worsening to refer to any
stimulus change that wouldfunction as
punishment for the type of behavior that preceded
it I don't want to use punishment since in the
context of describing this CEO I am not referring
to the decrease in future frequency of any
behavior. Similarly I use improvement for a
change that wouldfunction as reinforcement for
the type of behavior that preceded it, but am not
referring to an increase in the futore frequency of
any behavior in this context The term aversive
stimulus would be appropriate, except for my
general uneasiness about omnibus terms, as
expressed earlier. Worsening should be
considered a term from everyday usage, not a
technical term.
Establishing Operations 65
occur. This relation is exemplified by the
warning stimulus in an avoidance
procedure, and·such·a stimulus acquires
the capacity to establish its own
termination as an effective form of
conditioned reinforcement, andto evoke
any behavior that has accomplished this
termination. In the opposite direction, a
stimulus that systematically precedes
some form of improvement, and whose
termination prevents the occurrence of the
improvement, will acquire the capacity to
establish its own termination as a form of
conditioned punishment, and to suppress
any behavior that has accomplished this
termination. In an earlier paper (Michael,
1988) I referred to these as a threat CEO
and a promise CEO. It now seems more
reasonable to refer to a CEO that
establishes its own termination as a form
of reinforcement or punishment as a
reflexive CEO, a term which is more
indicative of the effect of this CEO in
altering its own function. (Reflexive here
is meant in the grammatical sense, not as .
referring to a reflex. This usage is thus
somewhat similar but not identical to the
mathematical and logical use as occurs in
the context of equivalence relations.)
An even more complex relation
consists in the correlation of a stimulus
with the correlation between another
stimulus and a form of unconditioned .
reinforcement The term conditional
conditioned reinforcer refers to just such
a relation. The stimulus upon which the
effectiveness of the conditioned reinforCer
depends is a CEO in that it establishes the
effectiveness of another event as
reinforcement, and evokes any behavior
that has produced this other event. This
type of CEO can be called transitive as
contrasted with the prior reflexive type of
CEO. (Again, this is the grammatical
usage, as with a transitive verb which
takes a direct object.) As with the
reflexive CEO, one must consider both
the positive and the negative case. With a
conditional conditioned punisher, the
stimulus upon which the effectiveness of
the conditioned punisher depends is a
CEO in that it establishes the
effectiveness of another event as

66 Establishing Operations
punishment, and suppresses any behavior
that has produced this other event. In an
earlier paper (Michael, 1988) I suggested
blocked-response CEO for this relation,
because many human examples were
characterized by a stimulus change
functioning as an SD for a response that
could not take place until some object was
available, and thus functioning as a CEO
in establishing the object as a conditioned
reinforcer and evoking the behavior that
had obtained such an object. The slotted
screw example described later has this'
pattern, but some CEOs of this type are
simply a stimulus upon which the
reinforcing effectiveness of another
stimulus depends, but with no response
blocked, ~ike the nonhuman example of
lever pres'sing. The three CEO types will
now be considered in detail.
Surrogate CEO: Correlating a
Stimulus with a UEO
The development of the CE (conditioned
elicitor),Sr(conditioned reinforcer), and
i
SP (conditioned punisher) each involves
I pairing or Correlating a neutral event with
I
~I
a beha.viorally effective one as a way of
giving the neutral event some of the
I behavioral properties of the effective one.
Ii It is not unreasonable to suppose that EO
"I properties could be developed in the same
way. The question is, would a stimulus
that had been correlated with a UEO
become capable of the same
reinforcer-establishing and evocative
effects as that VEO? The terms learned
drive or acquired drive appeared quite
often in the early learning literature. A
chapter by Miller with the title "Leamable
drives and rewards" was included in the
1951 Stevens Handbook ofexperimental
psychology. Muc4 speculation regarding
human behavior has taken the form of
postulating various learned motives, but
as Millet pointed out, "the experimental .
work. db. learned drives and rewards is
limited almost exclusively to (1) fear as a
learnable drive and fear reduction as a
reward, or (2) .learned rewards and drives
based on hunger imd food" (1951, pp.
435-436). Although couched in the
language of hypothesized internal drive
states, the work on fear as a learnable .
,
drive is primarily concerned with the
motivative characteristics of the warning
stimulus in an avoidance situation, the
reflexive CEO described in detail below.
Miller's section on leamed rewards and
drives based on hunger and food (1951,
pp. 454-462) dealt extensively with the
development and use of learned rewards
(conditioned reinforcers), but only briefly
with learned drives. Subsequent to that
publication there was some research on
the possibility of developing a learned
appetitive drive in a laboratory situation.
The question was whether stimuli
correlated with high levels offood
deprivation would produce, by
themselves, a momentary increase in the
frequency of the behavior that had been
reinforced by food. Also, would they
increase, by themselves, the effectiveness
of food, water, etc. as forms of
reinforcement? (Of course these questions
could be asked just asreasonably about
other VEOs, such as water, sleep,
activity, or sex deprivation, but most of
the research involved food deprivation.)
In the first experiment of this type,
Calvin, Bicknell, and Sperling (1953),
placed rats in a distinctively striped box
for 30 minutes a day for 24 days. During
this training one group was placed in the
box while food deprived for 22 hours,
and the other group while deprived for
only one hour. After training, both
groups were allowed to eat in the striJ!!:d
box following 1l.5 hours of food
deprivation, and the rats with the history
of 22-hour deprivation ate significantly
more than the group with the history of
one hour of deprivation. There were
several attempts to replicate these results
during the next several years, some
successful, but most failing to produce
similar results. A comprehensive review
ofthis line of research by Cravens and
Renner in 1970, identified several major
methodological problems with most of
the research, and concluded that the
results were essentially uninterpretable.
Mineka (1975) suggested that
gustatory and olfactory stimuli would be
more appropriate as conditioned elicitors
for a hunger drive than the visual stimuli

that had been used in most of the
previous studies. She conducted a series
of experiments comparing visual and
gustatory stimuli, with favorable results
when the latter were used, but then failed
to replicate those results in a subsequent
experiment, and ultimately concluded that
the phenomenon may not exist. Mineka
also made an interesting point about the
possible biological uselessness of such
learned appetitive drives, in that eating
more than is appropriate for a given
deprivation level simply because one has
been hungry in that particular stimulus
condition before would not be to the
organism's long term advantage or
survival.
There has not been much research of
this type since Mineka's 1975 report, but
it would be premature to exclude the
possibility of this CEO on the basis of the
unclear empirical evidence or
hypothesized negative survival value.
Deprivation-satiation UEOs typically
build up slowly, and it is not easy for a
stimulus to become correlated with the
extreme values of such a build-up. UEOs
with more rapid onset, however, are
often paired with relatively unique stimuli
that might be expected to develop CEO
properties. Would stimuli that were
correlated with decreases in temperature,
for example, have CEO effects similar to
the effects of those temperature decreases
themselves? In the presence of such
stimuli, would warmth be more
reinforcing than would be appropriate for
the actual temperature, and would
behavior that has produced such warmth
be more frequent than it ought to be for
the actual temperature?
Note that this is not a question about
the possibility of conditioned elicitation or
conditioned reinforcement or punishment.
It is well known that a neutral stimulus
(e.g., a tone) paired with cold (hand
dipped in ice water) will come to elicit
appropriate smooth muscle responses
(peripheral vasoconstriction) when it is
presented alone. It is also quite clear that
if the onset of the cold stimulus functions
as punishment (sI) for any response that
precedes it, then so too will any stimulus
Establishing Operations 67
that is correlated with such onset (SP).
Neither of these functions (CE, SP) are
synonymous, however, with the operant
evocative effect of an EO, although their
occurrence might always be a good basis
for predicting the CEO effect, which
might well be based on the same
physiological processes. I know of no
research bearing directly on the existence
of a CEO evocative effect based on
pairing with temperature changes, but the
possibility seems worth Gonsidering.
With sexual motivation, EOs for
aggressive behavior, and the other
emotional EOs, the issue has not been
addressed in terms specific to the CEO,
because its distinction from CE, sr, and
SP has not been previously emphasized.
There is evidence that a stimulus
correlated with painful stimulation will
increase the frequency of aggressive
behavior when presented alone (Farris,
Gideon, & Ulrich, 1970), but it is not
clear whether it is functioning as CEO,
CE, or both. The basic experimental
design is simple enough: Correlate a
neutral stimulus condition with a UEO,
and then see if by itself it increases the
reinforcing effectiveness of the
consequence relevant to the UEO and
increases the momentary frequency of the
behavior that has been developed through
reinforcement by that consequence. ,
Reinforcing effectiveness is not easy to
quantify, but the evocative effect should
be easy to measure, and its presenpe
should be evidence enough for the CEO
effect. One must, of course, use behavior
that is clearly of leamed operant origin to
prevent confusion of CEO with CEo
Thus, with painful stimulation and
aggressive behavior, if the potential CEO
evokes some arbitrary response such as
lever pressing, which has been developed
by reinforcement with access to another
organism to attack, then it is functioning
as CEO rather than CE, because there is
no UE for such behavior. The issue
would not be clear if the behavior studied
was striking, biting, etc. because these
may be elicited by painful stimulation as a
UE. Similarly, with sexual motivation, if
the previously neutral stimulus evoked an
 68 Establishing Operations
arbitrary response such as lever pressing,
which had been reinforced with access to
sexual stimulation, it would be
functioning as CEO rather than CE, but
the issue would be unclear if the behavior
studied was pelvic thrusting, which might
have been elicited by a UE.
The possibility of developing a
surrogate CEO based on painful . This process, like the ones described
stimulation as a UEO for escape behavior
must be carefully distinguished from the
next type of CEO to be discussed below.
The issue is whether correlating a neutral
stimulus with painful stimulation will
increase the effectiveness of pain
reduction as a form of remforcement and
evoke the behavior that has been
reinforcep with pain reduction. It is not
clear what it means to increase the
effectiveness of pain reduction when no
pain is present, but such a stimulus in the
presence of mild pain might cause the
mild pain reduction (along with the
reduction of the CEO) to be more like the .
reduction of more severe pain. Much less
difficult to measure would be the extent to
which such a stimulus evoked the pain-
escape response in the absence of pain. In
the typical shock-escape experiment, all
that is necessary is to precede the onset of
shock with a warning stimulus and see if
the shock-escape response is increased in
fr~uency by the ons~t ?f the warning
stlmulus. Note that thIS IS not the typical
escape-avoidance procedure. As
described below, there is no question that
a stimulus that systematically precedes
pain will evoke the behavior that
terminates that stimulus and thus avoids
the onset of the pain. Here the question is
whether the warning stimulus will evoke
the response that terminates the pain,
even though the pain is not present, and
even though such a response has not
prevented ):he onset of pain.
1'he situation can be clarified by
reference to an unusual type of avoidance
exp~riment, one with escape and
aVOIdance responses of quite different
.topographies. Imagine a rat in a
procedure where a lever press terminates
the. shock but a wheel tum terminates the
warning stimulus and avoids the shoc~.
The present CEO would be demonstrated
if the warning stimulus evoked the lever
press; the CEO discussed below if the
warning stimulus evoked the wheel tum.
The occurrence of such behavior could
have other interpretations, but it will
probably be possible to exclude these
with appropriate experimental designs.
above, seems intuitively quite plausible,
butresearch directed precisely at the CEO
issue has not yet been conducted.
Reflexive CEO: Correlating a
Stimulus with Worsening or
Improvement .
In the traditional discriminatedl2
avoidance procedure, an intertrial interval
is followed by the onset of an initially
neutral warning stimulus, which is in tum
followed by the onset of painful
stimulation-usually electric shock.
Some arbitrary response (i.e., one that is
not part of the animal's phylogenie pain-
e~ape ~pertoire) termin~tes the painful
stlmulanon and starts the lDtertrial
interval. The sarne response, if it occurs
during the warning stimulus, terminates
that stimulus and starts the intertrial
interval, thus avoiding the shock. As a
result of exposure to this procedure,
many organisms acquire a repertoire that
consists of making the relevant response
during most of the warning stimulus
occurrences.
12The tenn discriminated arose so that this type
of procedure could be distinguished from an
avoidance procedure with no progrnmmed
exteroceptive stimulus except for the shock
itself. It also implies that the warning stimulus
is a discriminative stimulus for the avoidance
response, but the main point of the present
section contradicts this practice-thus we should
develop a new name for this type of procedure.
Sometimes this procedure is called avoidance
without a warning stimulus and is then
contrasted with avoidance with a warning
stimulus. This may be preferable to
discriminated, but it implies the effect of the
stimulus on the organism-it warns the
organism---;md it would be preferable if the terms
for procedures did not presuppose their behavioral
functions.

Recall the analysis of the role of the
shOCk as an EO for the escape response,
the reinforcement for which is the shock
termination. The warning stimulus has a
similar function, except that its capacity to
establish its own termination as an
effective form of reinforcement is of
ontogenic provenance-due to the
individual's own history involving the
correlation of the presence of the warning
stimulus with the onset of the painful
stimulation. In other words, the warning
stimulus as a CEO evokes the so-called
avoidance response, just as the painful
stimulation as a UEO evokes the escape
response. In neither case is the relevant
stimulus correlated with the availability of
the response consequence, but rather with
its reinforcing effectiveness.
In more general terms, any stimulus
that is positively correlated with the onset
of painful stimulation becomes a CEO, in
that its own offset will function as
reinforcement, and it will evoke any
behavior that has been followed by this
reinforcement But this set offunctional
relations is not limited to painful
stimulation as a form of worsening (or
even to worsening, as will be seen later).
It is well known that organisms can learn
to avoid forms of stimulus change other
than the onset of pain. Stimuli that warn
of a lowered frequency of food
presentation, increased effort, a higher
response ratio requirement, longer delays
to food, etc. will all evoke the behavior
that terminates such stimuli. Such events
have in common a form of worsening,
and stimuli positively correlated with
such events are often called conditioned
aversive stimuli, without specifying any
particular behavioral function. It is
possible that such stimuli will generally
function as CEOs in evoking the behavior
that terminates themselves, as conditioned
punishment (SP) for any behavior that
precedes their onset, and as conditioned
elicitors (CEs) for smooth muscle and
gland responses of the sarne type that are
produced by painful stimuli. And, of
course, the offset of such stimuli will
function as reinforcement (Sf) for any
behavior that precedes that offset. As
Establishing Operations 69
mentioned in connection with the term
aversive stimulus above, there are
advantages and disadvantages to such
omnibus terms, and in any case their
availability doesn't obviate the necessity
for more specific terms. With respect to
the CEO, the case can be most clearly
stated as follows: Any stimulus condition
whose presence-absence has been f
positively correlated with the presence-
absence of any form of worsening will
function as a CEO in establishing its own
termination as effective reinforcement and
in evoking any behavior that has been so
reinforced.
It is possibly useful to repeat the
argument against such stimuli being
considered discriminative stimuli. A
discriminative relation involves a
correlation with the availability of a type
of consequence given a type of behavior.
A correlation with availability has two .
components: An effective consequence
(one whose EO was in effect) must have
followed the response in the presence of
the stimulus; and the response must have
occurred without the consequence (which
would have been effective as
reinforcement if it had been obtained) in
the absence of the stimulus. The
correlation between the warning stimulus
and consequence availability fails in the
second component. In the absence of the
warning stimulus, there is no effective
consequence that could have failed to
follow the response in an analog to the
extinction responding that occurs in the
absence of an SD. The fact that the
avoidance response does not turn off the
nonpresent warning stimulus is in no
sense extinction responding, but rather is
behaviorally neutral, like the
unavailability of food reinforcement for a
food satiated organism.
Now consiqer a stimulus that is
positively correlated with some form of
improvement. Such a stimulus would
clearly function as conditioned
reinforcement (Sf) for any response that
preceded its occurrence, but that is not the
functional relation under consideration.
Its CEO effect consists in its establishing
its own offset as effective punishment

70 Establishing Operations
and suppressing (as an opposite to
"evoking") any behavior that has been so
punished. The relation is quite plausible,
although I know of no research that is
directly relevant Stimuli that are
negatively correlated with some form of
worsening-safety signals-have the
same status as those that are correlated
positively with improvement Their onset
would be expected to establish their
removal as a form of punishment and to
suppress any behavior that had been so
punished. Similarly, but in the opposite
direction, a stimulus that has been
negatively correlated with improvement
would be expected to establish its
removal as reinforcement and evoke
behavior that has been followed by such
reinforcefnent None of these last
relations have been verified directly by
research but seem to follow naturally
from existing knowledge.
There is an important additional
requirement involved in the correlation
histories discussed above. It is essential
for the stimulus not only to have preceded
the worsening or improvement but also
for removal of the stimulus to have
systematically prevented the worsening or
improvement Ifremoval of the stimulus
does not prevent the worsening or
improvement, there is no correlation
(D'Amato, Fazzaro, & Etkin, 1968; also
see Fantino & Logan, 1979, pp. 273-
275).
. the buzzer sound, and evokes the
Transitive CEO: Conditional
Conditioned Reinforcement and
Punishment
When a stimulus condition (SI) is
correlated with the correlation between
another stimulus (52) and some form of
improvement (or.worsening), the
presence of the 5 1 establishes the
reinforcing (or punishing) effectiveness
of S2,~andevokes (or suppresses) the
behaVIor that has been followed by that
reinforcement or punishment. Again, this
relation has not been directly researched,
but it follows easily from existing
principles and concepts. Many (probably
most) forms of conditioned reinforcement
or conditioned punishment are themselves
t :
..,
conditional upon other stimulus
conditions. This notion is sometimes
referred to by saying that conditioned
reinforcing effectiveness is dependent
upon a "context."
Imagirie a food-deprived animal in an
environment where it can always produce
a lO-second buzzer sound by pressing a
lever. Distinctive visual stimuli are related
to the relation of this auditory stimulus to
food. In the presence of a red overhead
light, the lO-second buzzer sound ends
with the delivery of food. In the absence
of the red light, the buzzer sound lasts for
10 seconds and then ends without any
food delivery. This is a situation where
the auditory stimulus functions as
conditioned reinforcement, but
conditional upon the color of the
overhead light. Thus the buzzer onset is
not effective as reinforcement until the red
overhead light comes on. When it does,
with a well-trained animal, the lever press
will be evoked. What is the reinforcement
for the lever press? Obviously the buzzer
onset. How does the red overhead light
evoke the lever press? Not as an SD,
because it is not correlated with
availability of the buzzer-the buzzer is
actually available irrespective of the light
condition, but it is not an effective form
of reinforcement in the absence of the red
light. It evokes the lever press as a CEO,
a stimulus change that alters the
reinforcing effectiveness-the value-of
behavior that produces it. Only in the red
light has the buzzer been paired or
correlated with food, so only in the red
light is it an effective form of conditioned
reinforcement. The basic relation is still
that of correlation, but of a more complex
type. The buzzer's correlation with food
is itself correlated with the light color.
There have been several attempts to
demonstrate this type of CEO with
nonhumans (Alling, 1991; McPherson
and Osborne, 1986, 1988). The
experiments have in common the
following arrangement. One stimulus
condition, S 1, can be produced both in
the presence and in the absence of another
stimulus condition, 52. Onsetof the flTst

stimulus systematically precedes food
reinforcement in the presence of S2, but
not in its absence. In other words, S1
onset should function as conditioned
reinforcement, but conditional upon the
presence of S2; it is thus a form of
conditional conditioned reinforcement.
The stimulus upon which its reinforcing
effectiveness is conditional is the
supposed CEO. Pigeons learn to stop
producing S1 except in the presence of
S2, but in the experiments cited it has not
been possible to exclude the possibility
that S2 is simply functioning as the first
discriminative stimulus in a two-response
chain.
The first element in a chain evoked
by an SD is often a CEO of this type.
Consider a rat in a chamber where an
auditory stimulus is related as an SD to
the availability offood for a lever press.
But the lever cannot be pressed until it is
located, so the auditory stimulus evokes
visual search behavior, which is
reinforced by seeing the lever. The
auditory stimulus is not related to the
availability of this reinforcement,
however, but rather to its value. (Once
the lever is seen, the other elements of the
chain-approaching, touching,
pressing-are controlled by a succession
of SDs, but not the first element.)
Similarly, in a avoidance situation the
warning stimulus evokes the avoidance
response as a reflexive CEO, but if this
requires locating an operandum, the
visual search behavior is evoked by the
warning stimulus as a transitive CEO,
which is correlated with the value of
seeing the lever, not the availability of its
sight.
This type of CEO is exemplified by
many human examples. A workman is
disassembling a piec~ of equipment. His
assistant hands him tools as he requests
them. In the process of disassembling he
encounters a slotted screw which must be
removed, and requests a screw driver.
The sight of the slotted screw "evoked"
the request, the reinforcement for which
is receiving the screw driver. To refer to
the slotted screw as an SD for the request,
however, raises the same difficulty as
Establishing Operations 71
before. This stimulus has not been
differentially correlated with successful
requests-- screw drivers are not more
available when slotted screws are around
than in their absence, but rather more
valuable. The slotted screw should be
considered a CEO for the request, not an
SD. Here the slotted screw is like the red
light. In its presence screw drivers !fave
been correlated with successful .
disassembly and are therefore valuable.
Another common human example is
a stimulus related to some form of
danger, in its evocation of protective
behavior. A night watchman patrolling an
area hears a suspicious sound and pushes
a button on his radio phone that causes
the other night watchman to answer the
phone and ask if help is needed. The
suspicious sound is not an SD in the
presence of which such help is more
available, but rather more valuable. Note
that this effect of the danger signal is not
to produce its own termination, but rather
to increase the value of some other event.
This type of analysis seems to be
required irrespective of the direction of
the first or second correlations, and
irrespective of whether the final event is
improvement or worsening.,'J'o consider
one more example, let the buzzer be a
stimulus that is negatively correlated with
a worsening of some sort, in other
words, let the buzzer be a safety signal;
but let this correlation on!y be in effect
when the overhead light is red. Under
other stimulus conditions, the buzzer is
uncorrelated with any form of worsening.
Now let the lever press be a response,
maintained by an unrelated form of
reinforcement, that has in the past also
terminated the buzzer, and thus in the red
light terminated the "safety," a form of
conditioned punishment for the lever
press. The lever.press, of course, also
terminates the buzzer when the red light is
not on, but in this case it is not
punishment, because in the absence of the
red light the buzzer is uncorrelated with
worsening. We would expect that when
the red light came on, and the buzzer was
on, any tendency to press the lever would
be "suppressed." The red light is
72 Establishing Operations
functioning as a CEO to cause buzzer
offset to function as effective punishment
and to suppress any behavior that has
been so punished. These higher order
relations may not playa major role in the
typical animal experiment because they
require such extensive histories.
However they would be expected in the
repertoires oflong-lived species in their
natural environments, and most certainly
in those of humans.
General Implications
Motivation as a topic within behavior
analysis can be reintroduced as
consideration of those variables-
establishing operations-that
momentaIjly alter the effectiveness of
other events as reinforcement (and
punishment), and simultaneously alter the
frequency of those types of behavior that
have been followed by that reinforcement
(or punishment). A clear distinction is
possible between motivative and
discriminative variables in terms of the
whether they are related to the availability
of an event or to its reinforcing
effectiveness. The application of this
distinction is especially critical for the
proper interpretation of some of the
effects of painful stimulation. When
applied to learned functional relations this
distinction pennits classification of a
number of seemingly discriminative
relations as conditioned establishing
operations, further enlarging the topic of
motivation within behavior analysis, and
facilitating a useful identification of the
various factors involved in the multiple
control of human behavior.
 8 f
THE DISCRIMINATIVE STIMULUS
Definition
SD and S~ (S delta)
A discriminative stimulus is a stimulus
condition that (a) alters the momentary
frequency <number of responses per unit
time or proportion of response
opportunities in which a response
occurred) of a type of behavior (b)
because in the presence of that stimulus,
the relation between that type of behavior
and an effective behavioral consequence
has been different from what it was in the
absence of that stimulus. If the term
evoke can mean either increase or
decrease in the momentary frequency of a
. type of behavior, then a discriminative
stimulus is a stimulus condition that (a)
evokes a type of behavior (b) because in
the presence of that stimulus, the relation
between that type of behavior and an
effective behavioral consequence has
been different from what it was in the
absence of that stimulus. When the
correlation is positive and the con-
sequence is a form of reinforcement, the
discriminative stimulus is called an SD or
S+ (S plus). When the correlation is
negative and the consequence is a form of
reinforcement, the discriminlltive stimulus
is called an S~ (S delta) or S- (S minus).
Neutral Stimuli
When stimuli are uncorrelated with the
relation between a type of behavior and
any form of effective consequence, they
are discriminatively neutral stimuli.
There is no special term for this type of
stimulus. In laboratory usage, stimulus
typically refers to some condition that is
superimposed on a background stimulus
condition, and a neutral stimulus is
simply a change that could affect the
organism's receptors, but in the presence
of which the relations between responses
and consequences are no different than in
the background condition. This type of
neutral stimulus has sometimes been
confused with the second type of
discriminative stimulus mentioned above,
the S~, which is by no means neutral.
The three different kinds of relations
between stimulus conditions behavior,
and reinforcement, can be portrayed as
two-by-two correlation scatter diagrams
as shown on the next page. The square
dots represent occurrences in an
organism's history. A high positive
correlation is shown for the SD and a ,
high negative correlation for the S~. The
correlations are not typically perfect, as
indicated by the occurrences in the top left
cell in the left figure and the top right cell
in the middle figure. In a well-controlled
experiment, such occurrences would
represent occasional apparatus failures or
human errors. For the neutral stimulus,
the correlation is zero.
Discrimination Based on
Punishment
Reinforcement is not the only type of
effective behavioral consequence.
 74 Discriminative Stimulus

··... ·... ·...

··• ... ·• ...
• • •••
yes--> ...
••
...
•
• ••• • ••

R-->SR?
no-->
·· ...
... · .
·• ...
··...
·• ...
... ·... ·• ...
·...
•
•••
••• ••
no yes no yes no yes
S Dpreseilt? S /'; present? neutral S present?

Correlations for discriminative and neutral stimuli.

The presence/absence of a stimulus can The Common Definition

also be correlated, either positively or
negatively '(or not at all), with the relation
between a type of behavior and a
punishing consequence. This kind of
stimulus'control is more complex than the
previous cases because it presupposes
and is superimposed upon other
controlling relations. Suppose that a
particular type of behavior is momentarily
strong because it has generally been
reinforced in a particular stimulus
situation, whichwe can call S~, an
"
The SD is usually described simply as a
stimulus condition in the presence of
which 'a response is reinforced and in the
absence of which it is not (This last
phrase is sometimes omitted, but it is
always implied.) This definition is
incomplete in that it omits mention of the
evocative effect of the SD on behavior
(thus describing the procedure only-see
discussion below), it neglects or obscures
the possible contrast with a neutral
stimulus, and it fails to include stimulus

i ordinary discriminative stimulus. Once in control based on punishment. It is

'il a while, however, another stimulus limited in a fourth sense in that operant
'I
I:!
II condition, S~ ,has been superimposed stimulus control can also be developed

"'
II
i~ upon S~, and although the response has because a stimulus condition is related to
quantity and quality of reinforcement,
I
, I
i: I
,'Ii I
received its reinforcement in the presence
of this new superimposed stimulus, it has
also been followed by some form of
effortfulness involved in obtaining
reinforcement, and so on. In general,
and in somewhat common-sense terms,
i punishment. This history of punishment
would result in the response being operant stimulus control can develop ,
I
weaker when the superimposed stimulus, when' a stimulus condition is related to
II:
,1 any change that constitutes some form of
I~!
1
! S~ ,is present. The response could be improvement or some form of worsening
,I ' said to be suppressed by such a stimulus. (unless, of course, the change is too
Iii
There is no well-established technical slight, or is distributed over too long a
term for this relationship, even though a time period to affect the organism).
good deal of human behavior is under Sometimes the term stimulus control is
this type of control. The common-sense distinguished from discrimination by
term threat or warning are possibly limiting discrimination and the term
appropriate to S~ , and discriminative discriminative stimulus (SD) to the case
stimulusfor punishment is sometimes of reinforcement versus no reinforcement
used (although this is just as potentially and using the term stimulus control for all
cognitive in its implication as other cases.
discriminative stimulusfor reinforcement
which is discussed below). Not a Procedural Definition
Some behavioral approaches do not
defme the SD in terms of behavior but

only in terms of its relation to a
contingency of reinforcement; that is, to
the availability of reinforcement for some
type of behavior. Such a definition is a
description of a particular type of
environment, irrespective of the
organism's exposure to that environment
With such an approach, it is then
customary to introduce another term,
effective discriminative stimulus, for
those stimulus conditions that actually
control the behavior of an organism
because of its history with them. By
including the actual control in the
defmition, this second step can be
obviated, with no apparent loss in
terminological effectiveness.
Contrast with Cognitive Definitions
The definition in terms of an evocative
effect on behavior is in sharp contrast
with cognitive language in which the
stimulus is said to predict or signal the
reinforcer, or to inform the organism
about the availability of the reinforcer,
with the organism not necessarily doing
. anything as a result of the presence of the
stimnlus. The behavioral definition has
the SD evoking some behavior. It is a
stimulus for a response, not a stimulus
for a reinforcer. This latter expression
has the environment providing
information to the organism, with the
appropriate action being left to the
organism's discretion or to the operation
of some other variables acting upon the
organism; this is an essentially cognitive
or mentalistic approach rather than a
behavioral one. (Emphasizing the
evocative effect of the SD does not, of
course, preclude simultaneous control by
S D and
yes
·...
• •••
R__>SR? • •••
• • •• •
no ••.
••
no yes
SD present?
Discriminative Stimulus 75
other variables, such as establishing
operations, as described below.)
The Correlation of the SD Versus
that of the sr
Related to the point made above, the SD
is sometimes described as a stimulus
condition that is correlated with
reinforcement. As it stands, this
statement omits reference to the response
and is thus an accurate description of an
sr, not an SD. The SD is it stimulus
condition that is correlated with the
relation between a type ofbehavior and
reinforcementfor that behavior. The two
types of correlations can be represented
graphically (see below). In both
diagrams, the square dots represent
occurrences in an organism's history.
The patterned lower left cell in the
diagram on the right represents the
indefinite number of occurrences when
neither SR nor sr were present. A high
positive correlation (and thus a strong sr
and a strong SD) is shown in the tWo
diagrams, with most of the occurrences
being in the bottom left and top right
cells. .
Motivative Variables
During Acquisition of Stimulus
Control
Reinforcement in the presence of the
SD and extinction in its absence (let us
use the simple defmition for now;,but
remember that it doesn't include enough)
both imply that some motivative variable
is in effect or else the event that is
supposed to function as reinforcement
sr correlations
yes ··......
••••
SR present?
no
no yes
Sf present.?

76 Discrimi1llJtive Stimulus
would not have that function. Food is
not an effective form of reinforcement
without food deprivation, nor would
responding and f~g to rec~ive.
reinforcement funcnon as exnncnon
unless what one failed to receive would
have been reinforcing had it been
received.
When the SD Occurs After
Acquisition
An SD will evoke (recall the bidirectional
implication of evoke) a type of behavior
only if the relevant motivative variable is
in effect at the time the SD occurs. For
example, if food was used as
reinforcement for lever pressing in the
presence of an overhead light (and the
food-depnved animal pressed the lever in
the absence of the light and received no
food), then the light will evoke lever
pressing as an SD, but only if the animal
is food deprived.
The Strength of an SD
It is probably reasonable to assume that
• the occurrence of an SD does not have an
all-or-norte effect on the relevant
response, but rather increases its
momentary frequency over a continuum
of values. When the SD has been well
conditioned and when the relevant
establishing operation (EO) is at a high
value, one would expect the SD to
.compete favorably with other SDs,
producing its response more quickly, or
.,
in the case of incompatible responses,
producing its response and thus
preventing the occurrence of. the behavior
controlled by weaker SDs or by those
related to other EOs. The streIigth of an
SD also seems to be a function of the
nature of the typical consequence, in that
an SD related to a larger or higher quality
consequence would have a stronger
evocative effect than one related to a
smaller or lower quality consequence.
This point has not been researched often,
but seems to be taken for granted in much
of our verbal behavior about behavior.
Operant Distinguished From
Respondent Stimulus Control
The difference between the SD and the
conditioned eliciting stimulus of the
respondent relation (the CE) is in the
different histories. Both evoke a type of
. behavior, but due to different kinds of
historical relations with the environment.
Elicit is used with the respondent relation,
and set the occasion or simply occasion is
used with the SD; this makes it possible
to infer the relevant histories without their
being specifically mentioned. This
terminological practice suggests that there
is some observable nonhistorical
difference between the two types of
evocation, but this is not clear at the
present time.
 9
EVOCATIVE VERSUS FUNCTION-ALTERING EFFECTS
OF ENVIRONMENTAL EVENTS
Definition
The controlling relations between
environmental events and behavior can be
classified into several main types, and a
single event (e.g., the onset of a light)
typically functions as more than one of
these types. (Skinner discusses this issue
in detail in chapter 14 of Science and
Human Behavior, 1953; it is also an
essential aspect of the material on multiple
variables in chapters 9, 10, and 11 of
VerbalBehavior,1957.) A behavioral
relation can be called evocative when a
change in the environment produces an
immediate change in the magnitude,
latency, or frequency of a type of
behavior, but a change that persists only
as long as the new condition of the
environment persists. A relation can be
called function altering when an
environmental event causes a relatively
permanent alteration in a functional
relation between environment and
behavior. 1 Said another way, a function-
altering effect occurs when an
environmental event alters the organism's
repenoire (without necessarily producing
any immediate change in behavior) so that
the organism will respond differently
, was in the absence of that stimulus.
1This distinction is difficult to express in brief
defining statements. These two definitions are of
value ouly after several examples and some
further elaboration have been provided. At this
point it is well to tolerate some ambiguity.
Please read on.
when the conditions are again similar to
those that prevailed when the function-
altering effect occurred.
The term evoke is useful because it
can include both operant and respondent
relations. In the respondent case it is
equivalent to elicit, and in the operant
case it is equivalent to occasion, or
increase the frequency of. It is somewhat
unsatisfactory, however, in suggesting
only an increase in momentary response
strength, because some of the relations
that will be considered evocative involve
decreases. Evocative or suppressive
would actually be more accurate, but also
more cumbersome, so in subsequent use
of this term consider it to have a '
bidirectional implication.
Operant Evocative a n d '
Function-Altering Relations
The term discriminative stimulus (SD)
refers to an operant evocative relation.
The SD is a stimulus that evokes (alters
the momentary frequency of) a type of
behavior because in the presence of that
stimulus the relation between that type of
behavior and some effective behavioral
consequence was different from what it
Reinforcement refers to an operant
function-altering effect, and the effects of
extinction and punishment are also
function-altering effects in the present
sense, although in the direction opposite
 78 Evocative vs. Function-Altering Effects
that of reinforcement. An SD evokes
behavior because of previous function-
altering effects. Reinforcement
conditions a new evocative relation, that
is, brings a new type of behavior under
the joint control of an establishing
operation (see below) and a
discriminative stimulus; extinction and
punishment weaken 9r eliminate old
evocative relations.
Related Terms
A similar distinction is sometimes made
by contrasting stimulus control with
reinforcement control and also by
contrasting antecedent with consequent
control. These contrasts, however, imply
the absence of stimulus control when
beltavior change is accomplished by
reinforcement, extinction, and
punishment, which is clearly incorrect.
These consequence operations affect the
orgaIlism by altering existing relations or
developing new relations between
stimuli, establishing operations, and
types of responses. These terms are also
somewhat unclear with respect to
respondent functional relations..
Provenance
It is useful to classify behavioral
functional relations further in terms of the
way the environmental change acquired .
its effectiveness. Most such changes
affect behavior either because of a
learning history of the individual
organism or because of the evolutionary
history of that species, and can thus be
considered to be of ontogenic or of
phylogenie provenance.
The SD clearly acquires its control
over behavior as a result of a learning
history. Events that have a reinforcing or
punishing function, however, must be
further classified, because some have this
~unction innately and some acquire it as a
result of their historical relation to other
reinforcing or punishing events. This
distinction is currently made by
identifying reinforcement and punishment
as either unconditioned or conditioned.
The symbols SR and sP stand for the
former; and sr and SP for the latter. (The
terms primary and secondary are also
appli¢ to these two kinds of
consequences, but seem less accurate
than the innate-learned terms, because of
other possible implications such as
importance to the organism.)
Motivative Relations
Establishing operations (described in
detail in an earlier paper in this volume)
are environmental events defined in terms
of two effects on behavior: (a) They alter
the effectiveness of other events as forms
of reinforcement, and (b) they evoke the
type of behavior that has been reinforced
by those other events in the past. Thu~,
food deprivation (a) makes food a more
effective form of reinforcement and (b)
evokes the behavior that has been
reinforced with food. Painful stimulation
(a) makes pain reduction possible as a
form of reinforceme.nt and (b) evokes the
behavior that has previously terminated
pain. The first effect of an establishing
operation (EO) can be called its
reinforcer-establishing effect. The
second defming feature is an evocative
effect with respect to the behavior that has
been reinforced by the relevant
consequences. Because of this second
effect, the EO can be added to the SD as a
second operant evocative functional
relation. EOs can be further classified as
phylogenie or ontogenic in provenance,
depending upon whether their reinforcer-
establishing effect is unlearned (as with
food deprivation and painful stimulation)
or learned. The terms unconditioned
establishing operation (UEO) and
conditioned establishing operation (CEO)
are convenient ways of identifying the
two kinds of motivative relations.
Respondent Evocative and
Function-Altering Relations
The respondent unconditioned stimulus
(US) has both an evocative and a
function-altering effect: (a) It elicits the
unconditioned response (UR) at the
moment that it is presented, and (b) it
alters the function of any other stimulus
affecting the organism immediately prior
to the occurrence of the US, causing that

other stimulus to function in the future as
a conditioned stimulus (CS) in eliciting a
conditioned response (CR). In a typical
Pavlovian conditioning experiment, for
example, the meat powder that was given
to the dog both evoked salivation at the
moment that it affected the taste and touch
receptors in the mouth, and altered
(conditioned) the dog so that future
presentations of the conditioned stimulus
(e.g., the sound of a bell) would also
elicit salivation. In other words, the US
functioned as an elicitor and as a
conditioner, two quite different functions.
That the term US refers to either or both
of these functions is sometimes an
inconvenience, which can be overcome
by the invention of two new terms,
unconditioned elicitor (UE) and
-unconditioned conditioner (UC).
In general, it seems communicatively
most convenient to have technical terms
that refer to only one behavioral function,
even though many environmental events
~ave more th~ a single function. Thus,
111 any analySIS of behavior, such a term
woul? unambiguously identify the
functton under consideration, and if more
than one function was involved, more
than one term would be used. In the case
of the stimulus change that occurs
between two responses in an operant
chain, ~l--->S--->R2, the term SD refers
unambIguously to the evocative effect of
S with respect to R2, and Sr refers
unambiguously to the function-altering
effect of S with respect to Rl. Even if
the two (or more) functions are invariably
linked, it is still important to be able to
refer to each separately. Thus the terms
~D and sr are w~ys of identifying _
Important behaVIoral functions, even if it
were true (which it probably isn't) that a
stimulus could not function as an sr
unless it also functioned as an SD, and
vi.ce versa. Similarly, most (but not all)
snmulus changes that function as
unconditioned elicitors also function as
unconditioned
. . . conditioners' but even so,
It IS convement to have a separate term for
each function. And if the existence of
higher order conditioning is granted, then
Evocative vs. Function-Altering Effects 79
the ,CS should also be partitioned into a
conditioned elicitor (CE) and a
conditioned conditioner (CC).
It may also be convenient to have
some terms that refer to collections of
functions, as US and CS do at present
In the operant case, aversive stimulus is a
term of this sort, as is the less often used
appetitive stimulus. It is not clear r
however, that the convenience of ~uch
omnibus terms is sufficient to offset their
built-in vagueness. -
Multiple Functions of
Environmental Events
It should now be clear that most
environmental events that are not
beha,?orally neutral actually have multiple
f~mc~ons. In the Pavlovian conditioning
situatton, the meat powder functioned as
UE and UC, the second of which Was
most interesting to Pavlov. But it also
functioned as an SR for any behavior that
happened to occur inImediately prior to
the meat powder being provided --
(assuming that the dog \yas to some
degree food deprived). We would not
expect to see much of such behavior
because even if strengthened by one'
presentation of the meat powder it would
probably extinguish before the next
presentation. Had he been looking for
such an effect, however, it seems likely
that Pavlov could have seen OCcasional
superstitious operants, such as head
limb, or torso movements that happ~ned
to occur just prior to the presentation of
the US and were thus strengthened
enough to occur for a while before
extinguishing.
To the SD and sr functions of the
stimulus in an operant chain should be
add~ a CE function, depending in
qualIty on the event that functioned as a
UC at the end of the chain. Thus if food
was d~livered to a food-deprived' .
organIsm after the terminal response in
the chain, the stimulus change (e.g., a
light onset) that as an SD evoked this
response and that as an sr increased or
maintained the frequency of the initial
response in the chain would also be
 80 Evocative vs. Function-Altering Effects
expected to function as a CE with respect
to salivation. A light onset would also
undoubtedly function as a UE for
pupillary constriction. [Interestingly, the
pupillary constriction reflex does not
seem to be conditionable. It has not been
possible (young, 1965) to use a light
increase as a UC to develop pupillary
constriction to a previousfy neutral
stimulus. This is an example of a UE that
does not also function as a U C.]
Painful stimulation must be
considered separately in terms of onset
and offset. Onset functions as sP
(unconditioned punishment) for the type
of behavior that preceded it, in that such
behavior is less likely to occur in the
future (when painful stimulation is again
absent); as UEO in evoking the behavior
that has been followed by pain reduction
(an effect that continues as long as the
painful stimulation is present); as UE for
a variety of emotional reflex responses;
and as UC with respect to conditioning
the organism so that previously neutral
stimuli present at the momentof pain
onset will in the future also elicit some of
BEHAVIORAL FUNCTIONS
Phylogenie
Evocative --------------- ------------ ------------
Ontogenic
Phylogenie
Function-altering
Ontogenic
Although not shown in the table, the
extinction effects of a nonchanging
environment are also function altering.
When a CE occurs without being
followed by the relevant UC, it loses
some ofits evocative (eliciting) control
over the relevant type of behavior.
those emotional responses. (The painful
stimulation may also cause any neutral
stimulus that it was paired with to
function as a CEO when it is next present
by itself. This is the surrogate CEO
described in the earlier article on
motivative relations.) Pain offset
functions as SR for the type of behavior
that preceded such offset, in that such
behavim: will be more likely to occur the
next time painful stimulation is present; as
a UEO (like satiation) in momentarily
weakening (son of un-evoking) the
behavior that has been reinforced by pain
reduction; and possibly as a form ofUE
and UC with respect to the emotional
reactions (relief) opposite to those
produced by pain onset.
Similar analyses can be made of
most behavioral situations, and to prevent
overlooking functions it is probably
useful to be able to run through the
various functions, classified as operant or
respondent, evocative or function
altering, and phylogenie or ontogenic.
The various relations can be represented
in a table, as shown below.
Respondent Operant
UE UEO
CE CEO SD
UC
CC
Similarly, when an SD evokes a response
that is not followed by reinforcement, the
SD and the relevant establishing operation
lose some of their evocative control over
the relevant type of behavior. Also,
when an SD suppresses a response that
has been weakened by punishment and

punishment does not occur, some of the
suppressive effect is lost.
In addition to the effects mentioned
thus far, all of the events listed in all cells
of the table seem capable of reproducing
themselves by the operation of pairing.
This effect for the UE and CE is shown
as the UC and CC entries immediately
below them. The best known of the other
such reproductive effects are related to the
development of conditioned
reinforcement and puIiishment When a
stimulus event that functions as
reinforcement or punishment (either
unconditioned or conditioned) has been
paired with some other stimulus event,
the latter becomes capable of functioning
as reinforcement or punishment in its
own right The pairing of stimuli is the
procedure of respondent conditioning,
but in the present case, the evocative
effects of the stimuli are irrelevant.
Instead, it is the capacity of the
previously neutral stimulus to function as
reinforcement or punishment that has
been conditioned. That a CEO may be
developed by the pairing of neutral
stimuli with a UEO is discussed in the
earlier paper on motivative relations.
Pairing a neutral stimulus with an SD is
not the usual way that SD control is
developed, but there is some evidence for
such an effect (Morse & Skinner, 1958).
Direct Versus Indirect Effects
of Environmental Events 2
All of the function-altering effects
discussed above-S R , SP, sr, SP,
operant extinction, UC, CC, and
respondent extinction-are changes in the
function of EOs or of stimuli brought
about by events that are very close in time
to the relevant response or stimulus.
Thus reinforcement refers to· an increase
in future frequency of a type of behavior,
given some particular EO and some
particular stimulus condition, because a
response of that type was followed by.
2A later article in this collection, "Behavioral
Effects of Remote Contingencies," is a more
detailed consideration of these same issues.
Evocative vs. Function-Altering Effects 81
some form of improvement relevant to
that EO, andfollowed means "occurred
within afew seconds ofthe response". It
is convenient to refer to such effects as
direct, and to contrast them with indirect
function-altering effects of environmental
events in which the environmental event
(usually reinforcement or punishment)
occurred minutes, hours, or days aftet the
relevant behavior or stimuli.
It often happens with humans that a
remote event has an importimt effect on
the type of behavior that produced this
event, even though the behavior and its
consequence are separated by a long time
period. For example, a student was
induced by a fellow student to try a novel
study procedure in preparation for the
next course exam. Possibly as a result,
the exam performance was much
improved over previous performances on
similar exams. It would be quite
reasonable to assume that success on the
exam would result in the student's
continued use of the new study
procedure, and one might hear the
success referred to as reinforcement for
the use of the new study procedure.
However, studying for the exam must
have occurred hours or days before I
leaming of the exam success, and should i
certainly not be referred to as a direct
effect of reinforcement
Whether or not the use of the new
study procedure is increased in future
frequency by the favorable exam .;
performance is not the issue, but rather
how such an increase is to be understood
in terms of the principles of behavior
analysis. In general, effective scientific
terminology employs different terms for
different processes, operations, events,
etc. To attribute the increase in the use of
the new study procedure as a result of the
exam success to reinforcement is to
identify the episode as an instance of
operant conditioning. At present, this
identification seems quite unjustified, and
therefore, it is important to distinguish
direct from indirect effects clearly.
An adequate analysis of indirect
function-altering effects is yet to be made,
although several steps in that direction
82 Evocative vs. Function-Altering Effects
although several steps in that direction
have recently been taken. It appears that
certain verbal stimuli, called by Skinner
rules or contingency-specifying stimuli,
can produce, qfter an appropriate verbal
and social history, approximately the
same conditioning effects that are
produced by repeated exposure to
ordinary conditioning environmental
changes. This possibility is fIrst
mentioned in Verbal Behavior (1957,pp.
357-367). Later, in a quite different
context (Contingencies ofReinforcement,
1969, pp. 146-171), the concept of rule-
governed behavior is introduced, at fIrst
to form a contrast with contingency-
shaped behavior, then later analyzed in
terms of possible provenance and
function. Two recent papers dealing with
this issue are those of Schlinger and
Blakely (1987) and Blakely and Schlinger
(1987). This highly verbal way of
altering a repertoire without actual
exposure to ordinary repertoire-altering
environmental events is just beginning to
be studied from a behavioral perspective.
We can expect a better understanding of
this type of behavior change in the near
future.
The distinction between direct and
indirect also applies to the evocative
effects of discriminative stimuli. It is not
uncommon to hear some event referred to
as an SO for behavior that occurs hours
or days after the supposed SO. As in the
case of function-altering events, the facts
are not being contested. There is no
question that some behavior by humans
may actually be attributable in some way
to a stimulus event that occurred long
before the relevant behavior. For
example, taking an-umbrella when one
leaves for work is undoubtedly
sometimes attributable to having heard the
prediction of rain made on the morning
television weather report, possibly an
hour earlier. To refer to the heard
prediction as an SO for taking the
umbrella, however, implies that the
functional relation is the same
discriminative relation studied in the
laboratory where the behavior occurred in
the presence of the SO, or immediately
after its occurrence, but it is quite likely
that the relation is actually much more
complex. The same argument is relevant
to the other evocative relations, especially
the CEOs that are currently mistaken for
SOs.
As with the function-altering
relations, an adequate analysis of such
long-delayed effects is not readily
available. Some of them are currently
researched and theorized about by
cognitively oriented psychologists under
the topic of memory, but behavior
analysts have only just begun to deal with
such issues (e.g., see Palmer, 1991).
For now, however, it is important to
recognize the problem and not to confuse
the more complex indirect effects with the
simpler and better understood direct ones.
~.
 10
STIMULUS CHANGE DECREMENT AND STIMULUS
GENERALIZATION
Definition
As discussed below, stimulus change
decrement and stimulus generalization are
the complements of one another, and
because generalization is more familiar
than decrement, it might be better to start
with generalization. However, in some
ways, decrement is an easier concept to
apply and is more likely to be overlooked
than generalization. The principle can be
stated as follows: Whenever behavior is
strengthened by respondent or operant
conditioning or weakened by respondent
or operant extinction or by punishment,
-the future effect of the behavior-change
procedure will be at its maximum value
when the stimulus conditions are exactly
like they were when the behavior-change
procedure occurred. A smaller future
effect will be seen if at that time the
stimulus condition differs in any way
from what it was at the time of the
original behavior-change procedure. In
other words, there will be decrement in
the effect if there is any stimulus change.
Furthermore, the greater the stimulus
change, the greater the decrement in
effect Note that effect here means either
an increase in behavior due to
. conditioning or a decrease due to .
extinction or punishment A behavlOr-
change procedure is one that has a
function-altering effect Another way to
describe the stimulus change decrement
principle is to state that function-altering
effects are somewhat specific to the
stimulus conditions in which they take
place.
Illustration of this Principle
With an Ambient Stimulus in an
Operant Conditioning Procedure
Train a rat to press a lever in a chamber
(use food reinforcement contingent on
lever pressing) with a dim overhead light.
Next observe a decrease in pressing rate
when the overhead light is made bright.
Note that no discrimination training was
given. The rat had not been exposed to
the bright overhead light prior to our
demonstration of stimulus change
decrement. Also note that it is only at the
very beginning of the exposure to the
bright light that we are measuring
stimulus change decrement. If we .,
reinforce in the bright light, the decrement
will be quickly overcome as a result of
this reinforcement; if we extinguish in the
bright light, the behavior will become
even weaker due to the extinction.
With a Discriminative Stimulus
(SD)
Train a rat to press l\ lever when a dim
overhead light is on but extinguish lever
pressing when the light is off. The dim
overhead light can then be considered a
discriminative stimulus for lever pressing
(the S delta, or Sd, is the absence of the
dim light). When in a darkened chamber
the dim overhead light is turned on, the
rat begins pressing. Next, in a darkened

84 Stimulus Change Decrement
chamber, turn on the overhead light but at
a considerably brighter level, and observe '
that the lever pressing is weaker in some
way (slower, longer latency, etc.). Note
that there was no real SD in the first
situation above because there had been no
extinction training in the absence of the
dim light In the second, however, the
dim light is a real SD because there was
real S.1 training.
With Extinction of an SD
Train a pigeon to peck a disk on the wall
of the experimental chamber when the
disk is illuminated with r~ light, yellow
light, and green light; but extinguish
pecking when the disk is not illuminated.
That is, reinforce disk pecking (typically
on a'variable-interval schedule of
reinforcement) when the disk is one of
the three colors, but do not reinforce
pecks to the unillurninated disk. Each of
these three colors can then be considered
discriminative stimuli for disk pecking.
Next, extinguish pecking when the disk
is red, and observe that some pecking still
occurs when the disk is yellow, and even
more when it is green. Here the effect
that shows a decrement due to stimulus
change is the weakening effect of the
extinction training. (This same approach
can be used to demonstrate decrement
with respect to punishment of responding
in the presence of an SD.)
With a Conditioned Eliciting
Stimulus (CE)
Condition a dog to salivate to a tone of
SOD Hz (by repeatedly pairing this tone
with an unconditioned stimulus such as
meat powder squirted into the dog's
mouth), then present a tone of 1000 Hz
(but without.the meat powder) and
observe less salivation to this novel tone
than t<:,the tone of the original
c()nditioning.
With an Ambient Stimulus in a
Respondent Conditioning
Procedure
Condition a dog to salivate to a tone of
SOO Hz as above, in a room with dim
overhead lights. Then observe reduced
oq
salivation to the SOD-Hz tone when it is
presented in the room with the lights on
bright.
With Respect to Extinction of a
Conditioned Elicitor
Condition a dog to salivate to tones of.
300, SOO,and 700 Hz by pairing each
repeatedly with meat powder. Extinguish
salivation to the 700-Hz tone by
presenting it repeatedly without the meat
powder. Observe more salivation to the
SOO-Hz tone than to the 700, and even
more to the 300 than to the 700. With
respect to the extinction procedure, SOO
represents stimulus change and 300 even
more stimulus change; thus, when these
stimuli are presented one observes a
decrement in the effect of extinction with
respect to the 700-Hz tone.
As an Explanation of the
Partial Reinforcement Effect.
Extinction after intermittent (or partial)
reinforcement takes much longer, results
in many more responses, than extinction
after continuous or regular reinforcement
This has historically been referred to as
the partial reinforcement effect which can
be explained in terms of stimulus change
decrement with respect to ambient
,stimuli. Let us use a rat pressing a lever
and receiving food reinforcement as our
example. During continuous
reinforcement there are a number of
stimulus events related to the
reinforcement as follows: almost
constant gustatory and olfactory
stimulation from the food pellets;
rhythmic visual, tactile, and kinesthetic
stimulation resulting from moving from
lever to food tray and back; and frequent '
auditory stimulation from the operation of
the food-delivery mechanism. In terms
of ambient stimulus events, the situation
could be called a very busy one. When
extinction begins, there is a sudden and
dramatic change in the ambient stimulus
situation. The gustatory, olfactory, and
auditory stimuli related to reinforcement
are absent, and the stimuli related to
movement from lever to food tray and
back become quite irregular. It is as

though reinforcement had typically taken .
place in the light and it was now dark.
Thus, superimposed on the weakening of
behavior due to its occurrence without
reinforcement is a large stimulus-change-
decrement effect. With intermittent
reinforcement, however, the stimulus
situation when extinction begins is not
very different from that present during
periods of nonreinforced responding on
the intermittent schedule of reinforce-
ment The extinction-produced decrease
in responding does not have the large
stimulus change decrement superimposed
on it, and responding continues for much
longer than after continuous reinforce-
ment These facts are often unnecessarily
mentalized in terms of the animal's not
being able to tell the difference between
extinction and intermittent reinforcement,
or continuing to expect to receive food for
responding after intermittent
reinforcement. "However, it is quite
sufficient to explain the continued
responding in terms of the reduced
stimulus change decrement due to the
stimulus similarity of the reinforcement
and the extinction conditions, without
reference to mental states.
Decrement Compared with
Stimulus Generalization
In a sense, stimulus change decrement is
the complement of stimulus
generalization. Assume that some
behavior has been brought under the
control of a particular stimulus, say a red
light. If we change the light so that it is
now orange and observe that in the
presence of this changed stimulus,
responding is in some way weaker (lower
rate, longer latency, smaller magnitude,
etc.) than it was in the presence of the
original stimulus, we are observing
stimulus change decrement. The
decrement could be quantified by
subtracting the measure of response
strength in the orange light from that in
the presence of the red light. In other
words, stimulus change decrement is
Stimulus ehange Decrement 85
seen when a novel or changed stimulus
produces less responding than the
original stimulus.
Stimulus generalization refers to the
increase in responding to a novel stimulus
as a result of training with a different
stimulus. Thus, to use the same ~
example, assume that prior to traini¥g
with the red light, there was little or no
responding in the presence of the orange
light. After training with the red light,
there is some responding'in the presence
of the orange light. The difference
between responding in the orange light
after training with red and responding in
orange before training with r¢ is a
measure of stimulus generalization.
Stimulus change decrement is the loss
due to the change from the original
stimulus; stimulus generalization is the
responding that remains in spite of the
change. Said another way, function-
altering effects (which includes the effects
of extinction and punishment) are
somewhat specific to the stimulus
conditions in which they take place
(stimulus change decrement), but they are
not completely specific to those
conditions (stimulus generalization).
Generalization of Extinction
When extinction with respect to one
stimulus condition results in decreased
responding in a different (but usually
somewhat similar) stimulus condition, the
decrease is referred to as the .,
generalization of extinction. To the extent
that the responding still occurs in the
changed or novel condition (as in the
extinction examples above), that is, to the
extent that generalization of extinction is
incomplete, we are seeing stimulus
change decrement with respect to an
extinction effect. This concept may seem
like wheels within wheels, but it is
relevant to a number of everyday
occurrences, and failure to understand it
is likely to lead to the invention of
inferred mental entities (such as
expectation).

11
BEHAVIORAL EFFECTS OF REMOTE CONTINGENCIES
Direct and Indirect Effects
Operant conditioning in~olves t~mporal
relations between behavIor and Its
consequences that are on the order of
seconds. It is convenient to refer to such
effects as direct, and contrast them with
the indirect effects of more remote
relations between environmental events
and behavior (Malott, 1984).
A grant-writing example.
Consider the case of a research grant
proposal, and the notification several
months later that the grant was approved.
It is reasonable to suppose that such a
favorable outcome would have a positive
effect on future grant-writing activity. It
is also reasonable to suppose that many
who are familiar with operant
conditioning might refer to the favorable
outcome as reinforcement for grant-
writing behavior. To do ~o, of cou~~, is
to imply that the increase m grant wntmg
is to be understood as an instance of the
kind of operant conditioning that has ~n
studied in the nonhuman laboratory. It IS
quite clear, however, that if it were not
for an extensive verbal repertoire and
social history involving such events, the
grant approval could ha:,~ no effect
whatsoever on grant wnung:
Assume instead of grant approval
that some other highly favorable event
occurred several months after the grant
request had been submitted, but let it be
an event that is in no way related to grant
writing-a very close friend who had
been dying of an illness makes a sudden
"ii
I,
iIiil
'I
!i
,
"
, i:
'i:I
'Ii
i'i
HI
"
and complete recovery, or one leams that
he is to receive an unexpected and large
inheritance. Could such events function
as reinforcement for the grant-writing
behavior that occurred several months
earlier? The layman and the cognitive
psychologist would both say that.tl,1ey I
liii
would have no effect on grant wnung "
'I'
because the recipient knows that such
events are not related to grant writing.
Although intuitively plausible, such an
answer should trouble the behaviorist,
who has not (one would hope) previously
considered the organism's knowledge of
causal relations to be an essential
requirement for an event to fu~ction as
reinforcement. In fact, behaVIOr changes
resulting from accidental reinforcement
are often cited as evidence for the
automaticity of operant conditioning.
Because of their importance to the
recipient (their relation to other forifts of
reinforcement), it is quite reasonable to
believe that the good news about the
grant, about the sick frien~, or about the
inheritance would all funcuon as
(accidental) reinforcement for whatever
behavior was occurring immediately prior
to receiving the good news. Thus we
might expect a future increase in the
frequency of letter-opening behavior or
the tendency to go to the place where mail
is received. The good news would also
function evocatively as a discriminative
stimulus (SD) for a variety of responses
related to reactions from others (telling
them about the good news) and also as a
88 Behavioral Effects ofRemote Contingencies
conditioned elicitor (CE) for some
emotional respondents.
The direct conditioning effects as
well as the various evocative effects are
all interpretable in tenns of our technical
behavioral concepts and principles, but
not the future increased frequency of
grant writing (seen when it is time to
write another grant) that might result from
the present request having been awarded.
That indirect effect must be analyzed in
terms of existing verbal repertoires,
history with respect to similar events,
rule-governed behavior (Skinner, 1969),
verbal stimulus equivalences (Hayes &
Hayes, 1989; Sidman & Tailby, 1982),
and no doubt other behavioral processes
and functions. Idon't mean to imply that
a behavioral analysis cannot be made, but
only that research with nonhumans and
with humans in applied settings and in the
laboratory has not yet resulted in a clear
and generally agreed upon understanding
of the indirect effect of long delayed
consequences on human behavior.
The grant-writing example should
not be thought of as an exception to the
rule of appropriate verbal practice. It may
well be that most of our examples
showing the effects of reinforcement and
punishment with normal humans are
oversimplified in the same sense. Many
important consequences for humans are
complex events, or relations among
events, that affect the relevant person in a
number of ways and over a long period
of time. The humanly important effects
of such consequences are seldom the
changes in frequency of the behavior
immediately preceding the stimulus
correlated with (the announcement of)
such a consequence, but rather the
indirect effects of such announcements on
the type of behavior that occurred much
earlier but was causally related to the
consequence. The essence of human
intellectual superiority over nonhuman
organisms is in our control by remote
contingencies, which at another level is a
major advantage of the educated over the
uneducated person. But it is a drastic
oversimplification to interpret these
indirect effects as nothing other than the
well-known and well-researched direct
effects of behavioral consequences.
More indirect effects. Some
areas in which indirect effects are often
treated as though they were direct are
behavioral contracting, self-management,
community applications (the control of
littering, energy use, etc.), interpretations
of broad cultural phenomena (the effects
of third-party payments on clinical
practice or the effects of insurance
policies on safety behavior), and
especially the rapidly growing area
referred to as organizational behavior
management (OBM). In this latter field,
most incentive programs in which
consequences are specifically related to
job performance involve consequences·
provided long after the relevant behavior
occurs; similarly, most efforts to control
tardiness, absenteeism, safety in the
workplace, and company theft involve
remote consequences. The consequences
used are monetary bonuses, time off,
favorable work schedules, opportunity
for advanced training, mention in a
newsletter or on a wall poster, and so
forth, all of which are presented to the
recipient long after the relevant behavior
occurred (or as is discussed below, even
before the behavior has a chance to
occur).
The only directstrengthening effects
of such events would be on the trivial
behaviors of opening pay envelopes,
approaching wall charts, and so on. Note
that the delay of the monetary bonus
(e.g., in next month's pay envelope) is
not a problem if a stimulus associated
with this event (some kind of
announcement) is available immediately
after the relevant behavior, but such
stimuli are themselves usually provided
long after the relevant behavior. One
form of reinforcement often
recommended by OBM consultants is
praise by supervisors, which may have
useful direct effects, but see the section
below on two other clues to the
involvement of indirect complexities.
A lottery example. It may be
helpful to consider an OBM example in
some detail. Suppose a manager of a

smaIl factory approaches an OBM
consultant because his workers typically .
arrive late for work. The consultant
might be expected to inquire about
existing consequences for being on time
or for being late. The manager admits
that there is clearly no positive
consequence for being on time and not
much of a negative one for being late,
even by as much as 30 minutes or so. Of
course repeated occurrences of extreme
tardiness lead to warnings and possibly
ultimate discharge, but this seldom
happens. Work is supposed to begin at 8
a.m. but things don't really get going
until around 8:45.
It would not be atypical at this point
for the consultant to suggest that because
behavior is a function of its
consequences, it would be reasonable to
consider providing something positive for
being on time, something negative for
being late, or possibly both. The positive
approach is favored by most consultants,
so assuming that the manager agrees, the
consultant might institute a lonery system
for coming to work on time. Either
through some time-clock punch-in or
through the efforts of someone appointed
to the task, workers arriving before 8
a.m. are provided with a lottery ticket as
they come in the factory entrance. They
put their names on the tickets and at the
end of the week a drawing is held and
several winners are given dinner-for-two
at a local restaurant or some other prize.
The consultant believes, and perhaps
quite correctly, that this procedure will
lead to a considerable decrease in late
arrivals, and he justifies this procedure in
terms of the law of effect, as a well-
verified principle of behavior. It is, of
course, the relation between the lottery
tickets and the prizes that makes the
tickets valuable. This is a type of token
system, which though somewhat
complex, is not too difficult to understand
in behavioral terms. Let us assume, then,
that for many of the workers the lottery
tickets are, themselves, effective forms of
conditioned reinforcement How might
they indupe an individual who is ofte?
late to work to be more regularly on ume?
Behaviorpl Effects ofRemote Contingencies 89
. As a direct effect of reinforcement,
we might well expect an increase in rapid
locomotion through the factory entrance,
if such had been followed by the receipt
of a lottery ticket. The trouble is that
workers are not late to work because they
don't walk rapidly through the front gate.
There is an old saying to the effect that "It
does no good to hurry. You have t6 start
on time," and any effect of the lottery
ticket on starting on time is most likely
the result of verbal and other compl~x
processes occurring in the evening after
the worker comes home (setting the alarm
clock to an earlier wake-up time) or in the
morning prior to leaving for work
(scanning instead of careful reading of the
morning paper).
At the point where the consultant
mentioned the law of effect and the role
of behavioral consequences, the manager
should have brought up the topic of the .
devastating effects of even short delays of
reinforcement, as found in laboratory
investigations of operant conditioning
(e.g., Williams, 1976). He should
possibly have asked that the consultant
either provide a derivation of his expected
effect from known principles of behavior
that overcomes the problem of the
consequence of delays of minutes or
hours, or provide some other rationale.
The consultant would not have been able
to provide such an analysis, or at best it
would have been highly speculative.
About all that he could do, givel).,our
current understanding of such
complexities, would be to point out that
this procedure was quite successful at a
similar factory (if true), appeal to
common sense (Wouldn't you be more
likely to come to work on time if you
could get a lottery ticket?), or suggest that
the procedure be tried and if it doesn't
work they would know soon enough and
could then try something else. This more
modest approach, though possibly not
permitting as high a fee-there is often
some monetary advantage to technical
jargon-is safer in the long run, and is all
that is really possible at present.
 90 Behavioral Effects ofRemote Contingencies
Other Clues that an Effect is
Indirect
A preconsequence increase. In
addition to a delay between response and
consequence of more than 30 seconds,
there are two other clues that a behavior
change is probably an indirect elfect of
the relevant contingency. One is that the
behavior shows some increase in
frequency prior to occurrence of the
consequence, and the other is that a single
occurrence of a consequence produces a
large change in behavior. When a
neighborhood youth is offered $5 for
mowing a lawn, and then does so, it is
common to refer to the $5 as the
reinforcement for mowing the lawn, with
the implication that this behavior is to be
understood as an instance~of operant
conditioning, with the money as the
relevant reinforcement But he mowed
the lawn before he got the money, so that
instance of mowing cannot possibly be
attributed to that $5. We might speak of
that instance of mowing being related to
past promises that were fulfilled,
similarities between those circumstances
and the present one, and so on, but we
are speculating, and doing so after the
fact. Even ifwe confine ourselves to
mowings that occur after receipt of the $5
it is still risky to suppose that we are now
dealing with simple operant conditioning,
because the variables that controlled the
first mowing are possibly still relevant
and could continue to playa role. It
could, of course, be pointed out that
youths won't mow very often if promises
to pay are not fuifI.11ed. True, but our
belief in this common wisdom is not
based on our understanding of the
laboratory phenomel)on of operant
extinction, nor does it presently help our
understanding of this situation in any
simple way to' have studied the science of
behaviodn its present form. In a
practicalsense, it may help to know about
other studies of mowing or similar tasks
and their relation to pay, but it helps in
the way that knowledge of the experience
of others helps and not in the way that a
science helps.
A large change as a result of a
single consequence. The second
additional clue to the involvement of
indirect effects is a large behavioral
change resulting from a single occurrence
(or nonoccurrence, in the case of
extinction) of a reinforcing or punishing
consequence. Consider the common use
of descriptive praise, providing some
general sign of social approval (a smile
plus some comment such as "Good
work!") and in addition. a brief
description of the bebavior that is
responsible for the approval ("I like the
way you're ... I"~). When such praise is
provided to a normally verbal person over
5 or 6 years of age, it probably functions
as a form of instruction or as a rule, much
as if the praiser had said, "If you want
my continued approval you have to...."
For example, a factory supervisor walks
up to an employee who is cleaning up a
oil spill on the factory floor, smiles .
broadly and says "George, I really like
the way you're cleaning up that spill
before anyone steps in it. That's very
considerate of you." Now suppose that
George cleans up spills from that time
forward-a rather large change in
behavior considering that it was followed
by only a single instance of
reinforcement. We might suspect that the
praisefunctioned not simply as
reinforcement but rather as a form of rule
or instruction, and that George, for .
various reasons, provided himself with
similar instruction every time another spill
occurred.
Behavioral analysis of such rule-
governed behavior is just beginning to be
made, but it certainly shouldn't be
interpreted as a simple instance of operant
conditioning. Even when the
consequence is not accompanied by a
description of the relevant contingency,
the occurrence of the consequence may
evoke a description by the person
receiving it, and this self-produced
description may have important effects,
although at present we can only speculate
about such a process. Nor is the process
limited to consequences arising from the
behavior of another person. In

interacting with the physical environment,
as in working with some type of
mechanical tool, if a type of behavior is
followed by any kind of favorable effect
the person involved may describe the
relationship and be affected by the
description, much as though someone
else had stated a general rule.
The nonoccurrence of a customary
consequence may also lead to a sudden
and large decrease in the tendency to
engage in the relevant behavior, which in
some cases is probably due to some form
of self-rule statement. This is quite
common in our interaction with
mechanical or electrical equipment that
sometimes breaks and then is inoperative
until repaired. Our extensive experience
with such equipment often permits us to
react to the single failure by identifying it
as an ins!aDce of relatively permanent
nonfunction, describing it as such to ,
ourselves, and reacting to this self-
description as a form of instruction or
rule. To refer to the large and sudden
decrease as an example of extinction is
usu~y a drastic oversimplification.
How Should We Talk?
Modesty is appropriate. The
implication of the arguments given above
is that we should refrain from using our
technical behavioral terms for what may
seem like the functional relations of
operant conditioning when talking about
indirect effects. Incorrectly used
technical language is worse than
common-sense language because'it
suggests that the situation is well
understood, and it may displace serious
attempts at further analysis. Until we are
able to provide an accurate analysis of the
various processes relevant to indirect
effects, we are better off using ordiniuy
descriptive language. Thus, say "The
successful grant application is likely to
encourage future efforts in tile same
direction," but don't say it as though you
had the science of behavior behind you.
Stop referring to successful settlements of
a labor dispute as reinforcement for
striking, and successful election of a
political candidate as reinforcement for
Behavioral Effects ofRemote Contingencies 91
political activity. Similarly, when
someone does something that you like,
don't say "That's very reinforcing,"
unless you wish to direct the listener's
attention to the (probably trivial) direct
effects on the immediately preceding
behavior. It might be correct to say
"That's very eliciting," in reference to the
emotional respondents produced by ~eir
act, but it would be better just to say
"Thank you." Don't talk about good
grades as reinforcement fpr effective
study behavior, although they are no
doubt responsible for maintaining it in
some cases. Just say that they're
responsible for maintaining it. Restraint
of this sort will deprive some of us of an
opportunity to (incorrectly) display our
technical knowledge, but so much the
better.
Other proposed solutions. One
simple kind of indirect effect is fairly well
understood. When the delay of
reinforcement is on the order of seconds
or even minutes, the relation of a
particular type of behavior to its ultimate
effective consequence may be an instance
of simple chaining. In a nonhuman
experimental setting, one response may
produce the appropriate So for a different
response, which may in turn produce the '
So for still another response, and so on
until the final response in the chainis '
followed immediately by some form of
unconditioned reinforcement. Although
one might carelessly attribute the '
maintained occurrence of the first
response in the sequence to the ultimate
unconditioned reinforcement provided at '
the end of the sequence, it is quite clear'
that the main reinforcement for each
response is the immediate stimulus
change that it produces, that functions as
the So for the next response in the
sequence and also as the sr (conditioned
reinforcement) for the response that
produced it. This type of chaining cannot
be taken for granted, however. In work
with nonhumans, with young children, or
with low-functioning older humans, a
special kind of training called backward
chaining must be explicitly provided in
order to build up the effectiveness of each
,/
92 Behavioral Effects ofRemote Contingencies
stimulus change in the chain as an sr. To
interpret the effects of consequences that
occur hours or days after the relevant
behavior as instances of this relatively
simple type of indirect effect is quite
unjustified, however, unless an
uninterrupted chain of responses and
response-produced stimuli can be
identified (and this is almost never
possible), because any interriJption
simply locates the problem of delay at a
different point in the behavioral sequence
but does not solve it.
It is sometimes suggested that our
large brain permits us to react to
delayed consequences as though they
were immediate ones, and thus, we can
si~ply ignore the delay. But our large
bram would" also have to avoid linking the
consequence with all the other behaviors
that occurred between the relevant
response and the consequence, that just
gets us back to knowing what is related to
w~at, and this appears to depend on the
eXIstence of a verbal repertoire, rule-
governed behavior, and so on.
In simple operational terms, it "
could be argued that reinforcement is
usually defined as "an environmental
change that increases the future frequency
of any behavior that precedes it."
Therefore, if the grant-writing behavior
increased in frequency, the grant approval
must be considered an instance of
reinforcement. It is true that most formal
definitions of reinforcement do not
include a temporal qualification in their
defmition (similarly with punishment),
but the qualification is there by
implication when the results of research
with nonhuman species are cited as the
basis for the science of behavior being
extended to the human condition. To
~gnore the implied temporal requirement
IS to use the same term for effects that are
probablybased on quite different
beh~vioral proce~ses, a verbal practice
that IS surely detnmental to effective
technical communication.
Another approach is to cite the molar
orientations currently under study in the
nonhuman experinlentalliterature. These
-
deemphasize strict temporal contiguity as
the defining feature of the reinforcement
(or punishment) contingency in favor of
the effect of responding on overall
reinforcement frequency. The
correlation-based law of effect of Baum
(1973) is an example of such an
approach. However, it should be
remembered that proponents of such
molar approaches are dealing with much
shorter time spans than those involved in
most indirect effects of the type described
above, and even for those time spans
(typically a response may alter the rate of
reinforcement during the next several
lninutes although it has no immediate
consequence), the point is quite
controversial (e.g., see Vaughan, 1984).
The notion of a correlation-based law as
an explanation of changes in behavior due
to effects that are hours or days removed
from the relevant behavior has not, to my
knowledge, been seriously proposed by
those most involved in such research.
But Why Are We Successful?
As a fmal point, it is appropriate to
consider the seeming paradox that
although our treatment of indirect effects
as if they were the same as direct effects
is quite unjustified, we have nevertheless
been moderately successful from a
practical perspective. Reconsider the
rapidlr growing area of organizational
behaVIOr management, an area that is
providing employment for an increasing
number of behaviorally oriented
psychologists at both the M.A. and the
Ph.D. levels. It is quite clear from the
many published reports and conference
presentations that these behaviorally
oriented psychologists are often able to
effect considerable improvements in
various aspects of organizational
behavior. It is also clear that these
improvements are usually related to the
manipulation of behavioral consequences,
that are temporally quite remote from the
relevant behavior. It is a somewhat
unusual situation, in which weakness in
theory is accompanied by considerable
success in practice. "

I think there are several reasons for
this success. First, some of the problems '
dealt with in this area do involve the
direct effects of contingencies. Some use
of supervisor praise is of this sort. Some
incentive programs involve the
installation of counters or other devices
that provide stimulus changes
immediately contingent on important
features of the work activity.
Interventions of this sort are not likely be
thought of by consultants with a
nonbehavioral perspective, especially
when the relevant stimulus changes are
not the type that evoke comment by the
recipient.
Second, many of the improvements
have been made in settings in which
behavior has traditionally been
"understood" in terms of a constantly
changing variety of mentalistic concepts
and principles. Many of these
explanatory fictions direct attention away
from environmental variables that from a
common-sense point of view are -
important sources of organizational
behavior. The behaviorist is relatively
immune to such inner directedness and is,
Behavioral Effects ofRemote Contingencies 93
in fact, ever on the lookout for
manipulable environmental variables.
But probably the most important
reason for success is the empirical and
scientific orientation coupled with a
powerful research methodology. This
methodology, in contrast with what was
in effect prior to behaviorists' appearance
on the scene, emphasizes direct '
observation involving an easily ,
understood form of quantitative
measurement (usually frequency of
occurrence of something), high standards
of measurement reliability, and within-
subject experimental designs that permit
assessment of an effect under conditions
inappropriate for traditional group
comparisons. With this methodology the
practitioner can be successful,
irrespective of the incompleteness of
available theory and irrespective of verbal
practices that are possibly inappropriate.
But even though one may be able to do
good works without talking about it '
correctly, I can't help but believe that
even better works are possible when
verbal practices are not seriously flawed.
, ,
j
 12
ELEMENTARY VERBAL RELAnONS
In Verbal Behavior (1957), Skinner
identified and named six types of
functional relations between controlling
variables and verbal responses: mand,
tact, intraverbal, textual, echoic, and
audience relations. In the section on
transcription (pp. 69-71) he almost
named two more, which can be usefully
referred to as copying a text and taking
dictation (see paragraphs 2 and 3 of
p. 70). Skinner's general analysis of
verbal behavior has greatly facilitated our
ability to talk effectively about human
behavior, and these elementary verbal
units are an essential part of this analysis.
In teaching from Verbal Behavior, I
have found it. convenient to add two more
special terms to the list of elementary
relations. This addition does not identify
new or previously overlooked relations,
but rather provides names for implied
categories, and thus a place for several
forms of verbal behavior that were not
previously classifiable. The suggested
change also makes the basic categories
more nearly collectively exhaustive. The
new terms are codic and duplic, which
like echoic, textual, and intraverbal
function as adjectives preceding behavior
or relation, and like these others can
occur alone when behavior is understood.
The basic arrangement is described below
in a form that is convenient for
instructional purposes.
Mand
The response form (topography) is
controlled by a current unlearned or
!
learned motivational variable,an
unconditioned or conditioned establishing
operation. Said another way, the
response form is most closely related
historically to what has previously
functioned as reinforcement for such
responses. The response can consist of
speaking, writing, signing (as with the
sign language of the deaf), fmger
spelling, sending Morse code, etc. l
Skinner classifies mands as requests,
commands, or entreaties (on the basis of
how the listener is reinforced) and as
mands for nonverbal action versus mands
for verbal action (which latter are called
questions), plus some other associated
types (pp. 38-41). Another way of
classifying mands is to say that one can "
mand objects, actions, attention, and .;
more complex events, as when one asks
for information or instruction, or says
"Thank you" because an increase in the
li,stener's future favorable behavior woulil
be effective as a form of reinforcement:
In common-sense terms, the response
form of the mand is determined by what
the speaker wants .
Tact
The response form is controlled
primarily by an immediately prior
lSkinner's verbal relations were all described in
terms of topography-based verbal behavior.
Similar relations exislfor selection-based
behavior, for which see the later article in this
collection, "Two Kinds of Verbal Behavior and a
Possible Third:'
 96 Elementary Verbal Relations
nonv¢rbal stimulus (an Q1>Je,qt;;l!c;tjon,
relati(ln,prpperty, etc.). As with all of
the elementary verbal relations except the
mand, ilie~.ff¢':t ()f the establishing
oIJeflitidn on the response forn1 is
Illininili.ed by the fact that the
reinforcement for the.tact is usually
gen~rafu:edconditioned reinforcement
(Skinner, 1957, pp. 52-55) The response
can consist of speaking, writing, signing,
fmger spelling, sending Morse code, etc.
It might seem reasonable to substitute
some temi such as naming or describing
for the tact relation, but as Skinner makes
quite clear (1957, p. 82), there are good·
reasons lor avoiding such a substitute. A
useful contrast between mand and tact is
that "". ,: the Inand permits the listener to
infer s'omer.!ilrig about the speaker
regardless ,9.f the external circumstances,
while the tiictpermits him to infer
something about the circumstances
regar4!¥ss;of the condition of the
speaker"(Skinner, 1957, p. 83). In
terms 9f gI:oup coordination, the mand
pennitS th~spi:aker to alter the
environment through someone else's
behavior, and the tact permits the listener
to reactJothe behavior of others "...
rather than directly to things and events"
(Skinner, 1957, p. 432).
The audience relation is a special and
very important type of tact Skinner
(1957) devotes a chapter to this relation,
but the audience differs from the tact only
in the size'ofthe repertoire controlled,
, and in the facfthat the nonverbal stimuli
usually consist hf the collection of stimuli
arising from the listener.
Intraverbal Behavior
The response form is controlled by (a) a
verbal stimulus (the product of
someone's verbal behavior-but this is
not a simple concept, because the same
behavior Il;IlIy have verbal and nonverbal
products) 'Yith which (b) the response
does not have point-to-point
correspondence. Point-to-point
correspondence between stimulus and
response (or between stimulus and
response product) is in effect when
subdivisions or parts of the stimulus
control subdivisions or parts of the
response (or response product). In
intraverbal behavior the parts of the
stimulus are not related in any special
way to parts of the response. An
example of intraverbal behavior is a
tendency to say swamp as a result of
hearing someone say alligator. The
response can be speaking, writing,
signing, etc. and the verbal stimulus can
be the result of someone's vocal, writing,
or signing, etc. behavior. Note that for
American Sign Language, also called
Ameslan or just Sign (which is not
equivalent to fmger spelling), vocal or
written responses to signs or signing.
responses to vocal or written words are
intraverbal behavior. There is generally
no point-to-point correspondence
between signs and words (although the
situation is somewhat complicated by the
existence of initialized signs-signs that
incorporate some aspect of finger
spelling). The sign for cat, for example,
consists of stroking imaginary facial
vibrissae. This clearly has no point-to-
point correspondence with either the
spoken or the written cat. The finger-
spelled cat, of course, has point-to-point
correspondence with both spoken and
written cat, but not with the sign for cat
Codic Behavior
The response form is controlled by (a) a
verbal stimulus, with which it (b) has
point-to-point correspondence, but (c)
there is no formal similarity between
stimulus and response product;· Formal
similarity is Skinner's term for the case in
which the controlling stimulus and the
response product are (a) in the same
sense mode (both are visual, or both are
auditory, or both are tactile, etc.) and (b)
resemble each other in. the physical sense
of resemblance. Note that codic is meant
to suggest the kind of relation seen in a
formal code, where one stimulus is said
to stand for another stimulus that it does
not resemble in any physical way.
Textual behavior and taking dictation are
special types of cOOie behavior. In the
textual relation, the stimulus is visual
(written or printed words), and the

response consists of speaking. In
common-sense tenns, textual behavioris
reading out loud, without the implication
that the reader understands or can react in
any other way to what is being read. In
taking dictation, the stimulus is auditory
(the result of someone's vocal behavior),
and the response consists of writing what
is heard. There is at present no
commonly used fonn of codic behavior
involving signs, although such a system
was developed some time ago and is
available in dictionary fonn (Stokoe,
Casterline, & Croneberg, 1965). The
relation between finger spelling and vocal
or written stimuli may be codic. In the
same way, reading Braille out loud is
codic behavior as is writing in Braille
what one hears spoken.
Duplic Behavior
The response fonn is controlled by (a) a
verbal stimUlUS, and (b) the response
product has fonnal similarity with the
controlling stimulus. (Sometimes the
necessity for point-to-point
correspondence between stimulus and
response is cited as a third requirement,
but fonnal similarity between stimulus
and response product always implies
point-to-point correspondence between
stimulus and response, so this third
requirement needn't be listed.) The
response can be speaking, writing,
signing, etc. Duplic implies duplicates or
copies. Echoic behavior and copying a
text are special types of duplic behavior.
In the echoic relation, the stimulus is
auditory and the response is speaking
(echoing what one hears). In copying a
Elementary Verbal Relations 97
text, the stimulus is visual and the
response is writing (copying what one
sees in written fonn), Imitating
someone's signs is also duplic behavior,
as is fmger spelling what one sees
someone else finger spell.
The general importance of the basic
distinction between codic and duplic
behavior is dealt with in Verbal Behavior
in several places, but especially pp. 67-
68. With the five basic category names it
now becomes possible·to identify all of
the common fonns of verbal behavior in
terms of important defining propemes, as
well as to,classify immediately any new
fonn that develops. The two new terms
also make it unnecessary to extend
existing categories to novel conditions
because no technical tenn is available, as
when one refers to Braille; reading as
textual behavior. Braille reading is
clearly a fonn of codic behavior and
would be expected to share functional"
properties with other members of the .' ,
same category, but to call it and all other
forms of codic behavior involving a vocal
response textual is potentially confusing.
A similar undesirable extension occurs
when sign imitation is called'echoic
behavior, and is avoided by identifying it
as one of the several types of duplic "
behavior. The suggested terminology is
an instance of the general effort to
eliminate ambiguity from t~chriical and
scientific language, an effoh that is often
initiated and possibly most keenly
appreciated by those whinperid most of
their time teaching others to use that'
language.
 13
THEMAND
Refinement of the Definition
In Verbal Behavior,Skinner defines the
mand as "a verbal operant in which the
response is reinforced by a characteristic
consequence and is therefore under the
functional control of relevant conditions
of deprivation or aversive stimulation...
and in contrast with other types of verbal
operants... the response has no
specified relation to a prior stimulus"
(1957, pp. 35-36). In other words, with
the mand, what is said, written, signed,
etc. is primarily detennined by the
motivative variable (establishing
operation or EO) currently in effect. With
the more extended treatment of motivative
variables that appears in the earlier paper
("Establishing Operations") in this
collection, it becomes possible to
supplement Skinner's mand definition in
several directions.
But first, it might seem attractive to
avoid dealing with establishing operations
entirely and simply relate the mand to its
history of reinforcement: "a verbal
operant in which the response is
reinforced by a characteristic
consequence" (1957, p. 35). But history
of reinforcement explains only the origin
of the functional unit involving SD, EO,
and R, and not why the response occurs
on a particular occasion. In other words,
if a child says "water" as a mand, the
explanation of that instance of behavior
cannot consist simply of the statement
that such responses were reinforced with
water in the past, without adding that an
EO related to water (e.g., water
deprivation) was in effect at the time the
response occurred. The layman would
say that the child knew how to ask for
water when thirsty (had the relevant
history of reinforcement) and did so at
that time because of being thirsty at that
time (the relevant EO was in effect).
Skinner expresses this general point
when he says that the mand is under the
functional control of relevant-conditions
of deprivation and aversive stimulation
(1957, p. 35), which brings us to the
next point.
It is clear that the terms deprivation
and aversive stimulation are not broad
enough to cover all of the variables that
control the mand. Deprivation seems to
be lUi operation that is primarily relevant
to unconditioned establishing operations
or UEOs, and although aversive
stimulation can include any operation that
can be considered a form of worsening,
the transitive CEO is not covered by
either term (nor are UEOs such as salt
ingestion, temperature changes, and some
others).
The transitive CEO, as illustrated
with the following example, is especially
relevant to the mand. Two people are
walking toget1:Ier and one sees something
that must be written down so that it will
not be forgotten--a store name or an
address, for example. The would-be
writer, however, does not have a writing
tool, so requests one from the other
person, who readily provides it. It might
seem reasonable to consider the stimulus
responsible for the request to have been
 100 TheMand
{' an SD for that request, but this is not
J correct. The immediate reinforcement for
j requests of this type has clearly been
i receipt of the thing requested, in this case
i
I
i a pencil; but the stimulus that evoked the
request did not do so because it was an
especially favorable situation for
obtaining pencils-the companion would
have provided the pencil whenever
requested-but rather because it resulted
in the increased reinforcing effectiveness
of pencils. That is, it did not evoke the
request as an SD because of a correlation
with the availability of pencils, but rather
as a CEO because of a correlation with
the reinforcing effectiveness of pencils.
This type of CEO often seems to be a
stimulus event that functions as an SD for
a type of behavior that cannot occur
successfully until some other object or
event becomes available. The stimulus
event then also functions as a transitive
CEO with respect to the behavior that has
been reinforced by obtaining this other
object or event, which for the human is
often a mand. In the pencil example
above, the store name functioned as an
SD for writing, but writing was not
possible without a writing tool, so the
store name also functioned as a transitive
CEO for pencils, increasing their
momentary reinforcing effectiveness, and
more importantly, evoking all behavior
that had been reinforced by receiving a
pencil, such as asking for one. (One
might try to consider the transitive CEO
to be a form of deprivation in the sense
that something is absent, but again such
absence is not the precipitating cause of
the relevant behavior. In the previous
example, it was not the absence of pencils
that evoked the request, or there would
have been requests for all the other things
that were absent)
The mand, then, can be defined as a
verbal 6perant in which the response is
reinforced by a characteristic consequence
and is therefore under the functional
control of the establishing operation
relevant to that type ofconsequence. The
.next part of Skinner's definition, that "the
response has no specified relation to a
prior stimulus" should also be modified .
slightly. He is saying that with the mand,
unlike all of the other elementary verbal
relations, what it is that is agtually said,
written, signed (as in the sign language of
the deaf), is not determined by a prior
discriminative stimulus. This does not
mean (a) that the frequency or occurrence
of a mand is unrelated to prior
discriminative stimuli, nor does it mean
(b) that the form of the response is not
determined by prior stimuli functioning in
some other way than as discriminative
stimuli-namely as establishing
operations.
. With regard.to the first point, as
Skinner points out (1957, p. 52), prior
stimuli are not irrelevant to the actual
occurrence of the mand response form.
Consider the mand "water," where the
response form is determined by water
deprivation. Saying "water" in the
absence of an appropriate audience or
under circumstances in which water has
never been available has typically
undergone extinction, and thus even
under water deprivation the response will
not ordinarily occur until appropriate
circumstances are in effect. The audience
or the circumstances are clearly
functioning as SDs, but not in the sense
of determining the form of the response.
The EO contributes to an increase in the
momentary frequency of "water" as a
response form, but SDs related to past
reinforcement of such a response form
also contribute. In common-sense terms,
the deprivation produces some tendency
to ask for water, but such asking will not
occur under circumstances in which it has
been systematically unsuccessful in the
past. On the other hand, even in
circumstances in which the mand "water"
has always been reinforced, the response
form would not occur uuless an EO
related to water reinforcement was in
effect. This is what Skinner means by
the statement that with the mand "the
response has no specified relation to a
prior stimulus," but it seems somewhat
more precise to say that with the mand,
the form of the response is not
determined by a prior discriminative
stimulus.

Now with respect to the second
point, prior stimuli functioning as EOs
may well determine the fonn of the mand
response. UEOs such as painful
stimulation and temperature changes are
certainly prior stimuli, and, even more
common, both reflexive CEOs and
transitive CEOs are prior stimuli and are
clearly the detenninants of response fonn
in the mand relation. A revised
description of~mand and its contrast
with other verbal operants, then, is as
follows: Themand is a type of verbal
operant in which a particular response
fonn is reinforced by a characteristic
consequence and is therefore under the
functional control of the establishing
operation relevant to that consequence.
And in contrast with other types of verbal
operants, the response form has no
specified relation to a prior discriminative
stimulus. The other elementary verbal
operants (tact, echoic, etc.) consist of
response forms that are reinforced by
generalized conditioned reinforcement
(Skinner, 1957, pp. 53-54) but in the
presence of characteristic discriminative
stimuli, and are therefore under the
functional control of those discriminative
stimuli. Ideally, these other verbal
operants have no specified relation to any
establishing operation (but see chapter 6
of Skinner, 1957).
Practical Implications
Considering only mands controlled by
UEOs, one could easily underestimate the
ubiquity of the mand. When mands
related to the reflexive CEOs and to the
!fR'lsitive CEOs are added in, however, it
IS reasonable to assume that about half of
the adult's ordinary daily verbal
interaction consists of mands. In addition
to the mands for objects and actions,
there are the mands for SDs and CEOs
.(i.e., for infonnation) that constitute such
a large share of what we say to others.
Much of the verbal behavior controlled by
other ongoing verbal behavior in the same
speaker (autoclitic verbal behavior) is also
a type of mand, but due to the complexity
of this relation it cannot be dealt with in a
paper of the present scope. (For more on
TheMand 101
the autoclitic mand see Skinner, 1957
pp. 311-367, especially pp. 321-330;
also Peterson, 1978,' pp. 177-180.)
Underestimation of the mand's
importance in our speculative analysis of
everyday language is of little practical
significance. Normal children and adults
do not .need much professional support
for therr mands, because the mand is the
type of verbal behavior that directly
benefits the speaker. If anything, a more
common concem is to induce those who
mand too much to be more considerate of
the needs of others.
Of much more practical significance
is the relative neglect of the mand in
language training programs for the
developmentally disabled. Such
programs devote very little time to the
mand, in favor of training the tact relation
and what is referred to as receptive
language. There are several reasons for
this neglect, in addition to general
ignorance of Skinner's analysis of verbal
behavior. First, acquiring a verbal
repertoire is seen by many in the speech
and language area as learning the
meanings ofwords. It is assumed that
when such meanings have been acquired,
the words can then be used in various
ways with no further training. From this
perspective, receptive language training is
clearly one. of the easiest ways to teach
such mearnngs, and tact training is .
probably next Based on experience with
nonnal children and ad.ults, onc,y a person
has l~amed W~lat an.object is called (by
leammg to pomt to It when given its
name, or to say the name when the object
is shown), it is reasonable to assume that
when the ?bject becomes important the
leamer will be able to ask for it without
further training.
It is clear that this does not happen
with low-functioning individuals, many
of whom have had a good deal of
receptive language and tact training but
are sai~ to lac~ a ~unctionallanguage
repertorre, which IS then explained in
terms of their general intellectual deficit.
They can often point to several kinds of
object~ when the name is spoken and can
sometimes even say the name when the
 102 TheMand
object is shown, but they have no
tendency to request the object when it is
clear from other evidence that it would be
an effective fOIm of reinforcement for
. them. This point was dpunatically made
in the study by(Jjr"ati~gfSundberg,
~Ji3,8i7i)~ Two subjects were taught to
peifoIm a sequence of activities (without
responding verbally in any way) that
culminated in the production of
something that was known to be effective
as reinforcement. For example, both
subjects were taught to make a cup of
soup using a pack;ige of instant soup, a
bowl, hot water, arid a spoon, with the
last step being consuming the soup. In a
different setting, the same subjects were
taught to tact all of the objects used in the
other setting, but not in the process of
producing and consuming the reinforcer.
Later the subjects were returned to the
setting where they had produced the
reinforcer, but this time the sequence
could not be completed because the .
experimenter had removed a critical object
(the hot water in the case of making
soup). Under these conditions the
subjects had no tendency to mand the
missing object, although they could tact
the object in the other setting. When the
same subjects were next taught to mand
the missing objects by either echoic or
tact prompting, they readily leamed to do
so, and fairly soon acquired a more
general tendency to mand other things
when they became effective as
reinforcement even though they had only
been foImally taught to tact those things.
This suggests that a little bit of mand
training might have drainatic effects with
respect to the development offunctional
language.
Another reasqn for the neglect of
mand training, even by those who might
well appreciate its significance, is that the
trainer must contrive appropriate EOs or
take ad~antage of those that develop
naturally. Contriving a variety of
effective EOs for the learner seems at first
glance much more difficult than providing
a ,:ariety of objects (usually pictures of
obJt:Cts) to be named or pointed at. And
relymg on naturally occurring EOs in a
language-training setting will not usually
result in sufficient variety, although the
variety can be increased by providing
language training under other
circumstances not instituted for that
purpose. The pre>cec:lure called iifGlllW1itdl~'
teaching (Hatt;&RisfeYl,'197S) makes
some use of this latter approach, in that
verbal prompts for mands are provided
whenever the learner needs help in
obtaining some kind of reinforcement
during any training or care-giving
activities.
Although it might seem difficult to
contrive EOs in the artificial setting of a
language-training program, an .
understanding of the transitive CEO
should make it easier. In general, when
some known fOIm of effective
reinforcement cannot be obtained without
some additional object or action, that
object or action becomes the basis for a
reinforceable mand. This strategy is well
illust:ated in the Hall and Sundberg study
mentioned above, and could be a major
part of any language-training program.
Finally, mand training is neglected
because it is not well appreciated that it is
the only type of verbal behavior that
directly benefits the learner. When an EO
is strong, an appropriate response may
produce consequences that are specific to
that EO, thus providing a stronger form
of reinforcement than the generalized
conditioned reinforcement available for·
other types of verbal behavior. When
these two types of consequences are
contrasted, specific reinforcement
characteristic of the mand produces
stronger behavior (Stafford, Sundberg &
Braam, 1988).
Receptive and tact repertoires permit
the learner to follow directions given by
others, and to provide information to "
others. Of course such directions and
such information may well be to the long-
range advantage of the learner, but long-
range advantage is seldom effective as
reinforcement. There is some evidence
(Caroll & Hesse, 1987; Stafford et aI.,
1988) that mand training also makes other
aspects of language training more
effective. The EO and specific

consequences can be used in combination
with other variables (nonverbal stimuli,
verbal prompts, etc.) to evoke verbal
, responses and can then be faded out once
the response is strong. Also, a client
who can successfully mand for and
receive specific reinforcers is often much
more willing to participate in training
sessions. Receptive training, tact, and
intraverbal trials can then be interspersed
with mand trials, and language training
becomes much more like typical verbal
interaction rather than the standard
situation in which the trainer does all the
manding (e.g., "What's that?" "Touch
red") and the student simply complies
(Sundberg, 1987).
A final point concerning the
importance of mand training with the
low-functioning developmentally disabled
client is that it will often lead to a
considerable reduction in the frequency of
various kinds of inappropriate behavior
(crying, aggressive behavior, loud
unintelligible vocal responses, etc.). As
Sundberg (1987) points out much of this
behavior is actually under the control of
some strong EO, but the behavior is
either insufficiently specific for the trainer
or caregiver to comply with it, or
compliance does occur, which then
functions as reinforcement for the
inappropriate behavior. An appropriate
mand response, if generally successful,
will displace the inappropriate behavior,
and the client can then function more
normally.
TheMand 103
. The neglect of mand training is
naturally paralleled by a neglect of the
mand during language assessments. This
would be expected to result often in
assessments that credit the client with
better language skills than are actually
available. The relative lack of functional
language would then be the basis for
underestimating the client's actual'
potential for language performance. A
detailed treatment of language assessment
from the perspective of Skinner's
analysis of verbal behavior is available in
Sundberg (1987).
Conclusion
Skinner's analysis of language is a major
behavioral breakthrough, with many
theoretical and practical implications. Its
advantage over traditional language
theory is especially clear in its
identification of elementary verbal
relations and its implications for teaching
and learning these relations. Because of
the mand's introduction in terms of
unlearned motivative variables and
because its relation to prior controlling
events is quite complex, its general
significance has probably been
considerably underestimated. A more
extensive treatment of establishing
operations and a description of the mand
relation in such terms will, I hope,
prevent such a mistake, as well as lead to
a more appropriate emphasis on the mand
in language training prograrns for
subjects who do not develop nofmal
language.
 I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
 14
Two KINDS OF VERBAL BEHAVIOR PLUS A POSSIBLE
THIRD
Topography-Based Verbal
Behavior
'Verbal behavior consists of relations
between controlling. variables (verbal
stimuli, nonverbal stimuli, motivative
variables or establishing operations, and
consequences) and behavior. In terms of
a speaker, writer, or user of sign
language, the topography of the response
is an important aspect of the verbal
relation, being one of the factors that
distinguishes one verbal operant from
another. Saying "cat" and saying "dog"
differ from one another in response
topography (in the direction, force,
duration, etc. of the action of the various
components of the vocal musculature) as
well as in terms of the controlling
variable. The unit of verbal behavior can
be described as a specified frequency or
relative frequency 1 of a distinguishable
fOpOgraphy given some particular
controlling viriiable(s). In the case of the
1In most examples of verbal behavior, it is the
frequency of occurrence of a particular· type of
response relative to the frequency of the
circumstances in which that response would be
appropriate that is the actualor conceptual
dependent variable of interest. This is referred to
as the relalivefrequency and can be conlIaSted
.with absolute frequency, the number of responses
per unit of time. For a more extensive treatment
of this issue, see the earlier paper in this
collection titled "An Outline of the Science of
Behavior," the section on the operant dependent
variable (section IVC5).
,
tact, in the presence of a dog the speaker.:
shows an increased relative frequency of
saying "dog." In the case of the mand, as
a result of an increase in the strength of
some motivative variable or establishing
operation (Michael, 1982) which makes a
particular dog or dogs in general more
effective as reinforcement, the speaker
has an increased relative frequency of
saying "dog"; likewise for the other
verbal operants. It will be convenient to
refer to verbal relations of this sort as
topography based.
Selection-Based Verbal
Behavior
There is another type of verbal behavior,
however, that does not involve
distinguishable topographies and yet
provides equally distinguishable stimuli
to a listener. Under appropriate
conditions a person may behave verbally
by simply pointing at, touching, or in
some sense selecting a stimulus that then
plays the same role for the listener or
viewer as the response-produced stimulus
of a topography-based verbal response.
Although the effects on a listener of such
stimulus-selection-based verbal behavior
seem quite similar to those of
topography-based verbal behavior, from
the behaver's perspective they are quite
different The basic verbal relation is
between a controlling variable (verbal
stimulus, nonverbal stimulus,
establishing operation) and the control
106 Twa Kinds a/Verbal Behavior
that another specific stimulus has over the
pointing, touching, or other such
stimulus-identifying behavior. It is a
conditional discrimination in which a
stimulus (or an establishing operation)
alters the controlling strength of another
stimulus over a nondistinctive response
such as pointing or touching. The
topography of pointing is pretty much the
same irrespective of the thing pointed at,
especially if the stimuli to be selected are
not in constant positions relative to the
pointer's body. The unit a/verbal
behavior can be described as an increased
control over the pointing response by a
particular stimulus as a result of the
presence of a different stimulus (or the
strength of a particular establishing
operation). Pointing at a written word or
at a picture to alter the behavior of a
viewer exemplifies selection-based verbal (
behavior. Communication boards used
with the physically handicapped, and the
lexigram selection system used by
Rumbaugh (1977) and Savage-
Rumbaugh (1982) in their work with c .
chimpanzees, are probably the best
known such examples.
Important Differences
The difference between topography-based
and selection-based verbal behavior is
likely to be overlooked or considered
unimponant by those who are primarily
interested in the effects of verbal stimuli
on listeners. This is especially true in the
case of writing and reading, because
reacting to a written word would not
seem to depend on whether it was written
or simply selected by the verbal behaver.
The difference is also likely to be ignored
by cognitivists who see the selection of
words from memory as the most
significant aspect' of topography-based
verbal behavior, with the particular means
of display for the listener (speaking,
writing, pointing at words or at symbols
or. at pictures, etc.) being relatively
unimponant. This is somewhat
equivalent to inferring an internal
stimulus selector when external behavior
doesn't seem to involve this process, and
is common in coguitive interpretations in
which internal choice is used as an
explanation of external differential
responding.
But even behaviorally oriented
students oflanguage often seem to favor
selection-based over topography-based
conceptualizations, especially when
developing verbal behavior in nonverbal
organisms. As an exercise in my
graduate course on verbal behavior,
which uses Skinner's Verbal Behavior
(1957) as a text, I ask the students to
explain how they would develop a color-
naming repenoire in a pigeon that was as
much like the analogous human repenoire
as possible. Their pigeon color-naruing
behavior almost invariably has the bird
pecking a key with a word or symbol on
it appropriate to the particular color
displayed. Thus, in the presence of three
response keys, each "(ith a different
shape projected on it (circle, triangle,
square), the pigeon is reinforced for
pecking the circle when another display is
red; the triangle when the other display is
green; and the square when the other
display is blue. This, of course, is not
ordinary human color-naruing behavior,
yet almost never does the student have the
pigeon eruit different topographies in the
presence of the different color displays,
for example tum in a circle when the
display is red, peck its foot when the
display is green, stretch its neck upward
when the display is blue. It is possible
that the somewhat famous experiment by
Epstein, Lanza, and Skinner (1980) has
become such a pan of the behavioral
culture that the students produce a similar
experiment even though they claim not to
be familiar with the experiments. It is
also possible that the selection-based
analogy is favored because of the ease of
automating this type of experiment, but
all the student is actually asked to create is
a thought experiment. One student, even
though this exercise occurred near the end
of the semester, said that for her,
language still seemed to be essentially
related to words, and a pigeon's pecking
disks with words (or symbols) written on
them seemed more like real language than
a pigeon's turning in a circle when the

display was red, pecking its foot when
the display was green, and so on. I
suspect that her view (she was an
excellent student with an extensive
behavioral background) is not
uncommon. I'm reasonably confident
that laymen, linguists, philosophers, and
others with a strong predisposition for
mentalistic explanations of human
behavior would agree with her sentiment,
except they would probably fmd the
notion of pigeon verbal behavior
preposterous to begin with.
From a behavioral perspective, the
differences between the two types of
verbal behavior seem to be potentially
quite important A conditional
discrimination involves two primary
controlling variables, whereas an
unconditional discrimination involves
only one. Of course, additional variables
such as establishing operations, audience
characteristics, and others make
unconditional discriminations actually
conditional, but this means that selection-
based verbal behavior has a further
degree of significant conditionality. Our
verbal behavior about behavior has not
dealt much with discriminations of the
sort that do not involve distinguishable
responses. When such discriminations
do occur, we typically retain the
descriptive language appropriate to
differential topographies. For example,
we speak of the dependent variable in a
simultaneous color-discrimination
procedure as the response of pecking the
red key or pecking the green key, but
these are not different responses, if
response refers to topography, because
the pecking may well be the same
topography irrespective of the color that
evokes the pecking behavior. It is hard
not to believe that this further
conditionality is relevant to such factors
as ease of acquisition of a verbal
repertoire, effectiveness of control by
motivative variables, ease of interference
by similar functional relations, and so on.
Another difference (first suggested
by Paul Whitley) between these two types
of verbal systems is that topography-
based verbal behavior always involves
Two Kinds o/Verbal Behavior 107
point-to-point correspondence between
the response form and the response
product, whereas there is no such relation
in selection-based behavior. When one
speaks, there is correspondence between
the details of the vocal muscle action and
the relevant details of the auditory
stimulus that results; likewise with
writing and the use of signs and their
respective visual response products.
When one points at a word, picture, or
symbol, however, the muscle action of
the pointing response has no
correspondence with the important
features of the selected stimulus. Again,
this difference does not seem to be
irrelevant to such factors as ease of
acquisition, precision of control,
susceptibility "to interference, and so
forth.
Still another difference is the
necessity of an effective stimulus-
scanning repertoire in the case'of
selection behavior. Typically the various
visual stimuli from which the selection
must occur cannot all be viewed at the
same time. In the case of a selection-
based tact, for example, some nonverbal
stimulus affects the organism by
increasing the control of one of the verbal
stimuli over the pointing response.
However, if the set of verbal stimuli is
reasonably large and the scanning
repertoire is not systematic, the
appropriate verbal stimulus may be
overlooked. Also, if the scanning takes
much time, the effectiveness of(the
nonverbal stimulus may be lost by the
time the appropriate verbal stimulus is
encountered. A good scanning repertoire
is so well developed in the normal adult
that one might overlook its existence, but
when it is ineffective or absent, as with
young children or with some severely
retarded individuals, selection-based
verbal behavior is not possible.
Topography-based behavior requires no
such scanning (although of course
cognitivists typically invent an internal
scanning process, as mentioned above)
and thus seems behaviorally simpler.
The necessity of an effective scanning
repertoire for selection-based behavior
 108 Two Kinds ofVerbal Behavwr
also constitutes an additional basis for the
disruption of such behavior.
Manded Stimulus Selection
Closely related to the selection-based tact
is what is ordinarily called receptive
language, better referred to as manded
stimulus selection. In typical receptive
language training, an individual is
presented with a set of stimuli (objects or
pictures) and is aske4 to point to or touch
a particular item in the set. For example,
a teacher may present pictures of a cup, a
doll, a spoon, and an apple, and say
"point to the apple." The teacher mands
behavior on the part of the leamer with
respect to a particular stimulus. The
leamercan respond correctly only if the
pointiri'g response is jointly controlled by
the auditory verbal stimulus provided by
the teacher ("apple") and the nonverbal
visual stimulus provided by the object
(the apple). The reinforcement for such
,behavior is typically praise, an edible, a
trinket, etc., depending on the nature of
the learner.
The repertoire developed by such
training is, in a sense, the opposite of a
selection-based tact repertoire. In the
latter, an array of verbal stimuli is
presented along with a nonverbal
stimulus; which momentarily strengthens
the control by one of the verbal stimuli
over a pointing'response. With the
present example, the array would consist
of the printed words cup, doll, spoon and
apple, and the learner would have an
increased tendency to point to the word
,I apple when shown an apple and asked by
I I
the teacher "What is this?" In manded
stimulus selection, the array consists of
[' the several nonverbal stimuli (objects or
pictures), and the teacher provides the
verbal stimulus ("show me the apple ").
I Both of these relations involve joint
il control by a nonverbal and a verbal
Iii I
II!
;11
I!
IIII"
II1I
Ili
i~i!:
"
stimulus, thus both are clearly conditional
discriminations. Both also require an
effective scanning repertoire.
This type of instruction is quite
popular with the developmentally
disabled, even to the neglect of other
verbal relations, such as the mand and the
intraverbal, which seem to be more
directly valuable to the learner. Its
popularity is probably related to the
general belief that language learning
consists in leaming the meanings of
words, which can then be used for
various purposes; this way of leaming the
meanings of words doesn't require the
shaping of vocal or other topographies
nor the arrangement of motivative
variables (as in teaching the mand
relation). It may also be relevant that
manded stimulus selection enhances the
control of the leamer by the staff. These
comments should not be taken to suggest
that manded stimulus selection is
unimportant as a part of one's repertoire,
but only that it should not be considered
equivalent to other equally important and
behaviorally quite different kinds of
functional relations.
Topography-based verbal behavior,
selection-based verbal behavior, and
manded stimulus selection are often
considered to be equivalent forms of the
same underlying language processes. It
is true that the highly verbal adult has
well-developed repertoires of all three '
sorts, and behavioral relations acquired as
one type readily occur in the other types
without further training. From a
behavioral perspective, however, there
are differences between these types of
verbal relations; these differences would
be expected to be of special significance
when verbal behavior is being developed
in those whose verbal repertoires are
seriously deficient, and it is important not
to overlook them.
15
A BEHAVIORAL PERSPECTIVE ON COLLEGE
TEACHING
There are a number of college students
who learn very little from their courses.
Most of these students spend almost no
time studying their textbooks, and if they
even attend lectures they typically do not
spend much time outside of class
studying lecture notes. It is customary to
locate the problem within the student as a
lack of intellectual ability, a lack of
motivation, or possibly both. An
alternative approach would inspect the
environmental contingencies that are
supposed to maintain class attendance and
effective out-of-class study. What
follows is an analysis of these
contingencies and some suggestions for
improving them.
An Important Type ofCollege
Teaching
This analysis is not directed at laboratory
courses, seminars, skill courses such as
in fine arts or in spons training, or
courses that are "experiential," or aimed
primarily at such goals as personal
growth or effective interpersonal skills;
nor is it aimed at courses with very small
enrollments (15 or less) irrespective of
content or goal. I arn primarily concerned
here with courses having fairly clear
instructional goals related to a specific '
content or subject matter area, and with a
class enrollment of 40 or more.
At universities with 10,000 or more
students, a large proportion of most
students' formal education takes place in
such content courses. In the sciences and
humanities, most of the introductory and
quite a few of the upper level courses are
of this type. This kind of instruction is
also quite common in business and
education courses, less so in engineering
and fine ans.
These content courses are often.
required as a pan of the major or minor,
or they are semirequired in that they are
pan of a small set from which the student
must select one or more. Text material
covering most of the content is usually
available, and the instructional method
consists largely in assigning such text
material and supplementing it with
lectures. Assessment is with an in-class
exaruination, usually consisting largely of
multiple-choice questions, typically given
more often than twice a semester but
seldom as often as once a week.
Students have to learn from the text
by abstracting out main points, .
organizing and outlining, practicing
solving problems, classifying examples, .
making up original examples, making up
and answering test questions, and soon.
They must also learn during lecture and
from studying' notes taken during lecture
(which they reorganize, outline,
memorize, etc.), just as with the text
material.
I know that some professors and
some students consider such courses to
be necessary evils, at best, brought on by
mass education, but I don't share that
 110 A Behavioral Perspective on College Teaching
view. This type of course, when it
generates effective study, is responsible
for a great deal of learning. It is in these
courses that students learn much of what
is already known in a field. They acquire
extensive and often fluent repertoires that
pennit them to begin talking and thinkiJig
about the subject matter in the sarne way
that professional experts talk and think.
.These courses are sometimes criticized
for not teaching creativity, or how to
obtain new knowledge, but only how to
parrot back old knowledge. However,
for most scientific and professional areas,
creativity requires an extensive familiarity
with what is already known-an
extensive knowledge base with which
one can be creative. Creative and
innovative use of a subject matter can be
more easily addressed in later courses
with smaller enrollments, or in graduate
school.
DigreSsion on differences
between cOllege and high school.
Ahe;lvy reqance on texts and lectures as
source material is in sharp contrast with
teaching practices in many high school
courses, where c;:lassdiscussion,
individual and group projects,
home;work, and other activities constitute
a sigrtificant basis for the learning that
takes place. Related to the dependence on
texts and lectures is the use of exarn
scores as the main basis for the course
grade in college. High school course
grades are often based on a variety of
aetivities,including attendance and class
participation, with exarn scores
sometimes counting less than half of the
course grade. Some entering college
students who earned rather high grades in
high school may be quite ineffective at
studying texts and lecture notes and at
taking exams over such,material. It is
important for college counselors to be
sensitive to this cause of poor
perfor,rnarice in seemingly well-educated
enterihg freshmen. Another important
difference between high school courses
and the type of college content course
dealt with in this paper is the requirement
for many hours per week of outside class
study, as described below. In college,
homework is not just an occasional
inconvenience, it is where most of the
learning takes place. (It is my impression
that these three important differences
between college and high school are
seldoin properly emphasized in freshman
orientation programs.)
Student Motivation
At many colleges and universities
students are advised that to obtain a
passing grade-a grade of C-for typical
lecture courses they will be expected to
spend about two hours outside of class
for every hour in class. Thus, for a
three-credit-hour lecture course they .
would be expected to spend nine hours
each week, three in class and six in
outside study. I don't think our own
undergraduate catalogue makes any such
statement, but I consider it a very .
reasonable expectation and have always
taught my courses accordingly. This
outside study requirement is sometimes
referred to as the "two-for-one" rule, and
is based on a full semester load consisting
of 12 to 15 credit hours, which makes the
student's work week 36 to 45 hours
long, for what could be called mediocre
academic accomplishment For grades of
B or A, the work week may well become
50 hours or more, assuming that all of
one's courses are content lecture courses.
In actuality, most students do not take 12
to 15 hours of these courses each
semester, and some students with
favorable educational histories are able to
achieve at a high level in less time than
described above. However, it is not
unreasonable for a student who wants
high course grades to either put in more
time or take fewer courses. This is
clearly not the nine-to-five work week
considered reasonable in some kinds of
errtployment, but being a student should
be considered more like having one's
own business than working for someone
else. Unfortunately not all students see it
this way.
Now let us ask what might motivate
students to spend a large portion of their
waking hours attending classes and
studying academic materials outside of

class. The main problem is that there are
many other activities that compete for the
student's time. A good case can be made
(to the student, at least) that the college
setting is an appropriate place to acquire
repertoires other than academic ones--
leadership, organizational and
interpersonal skills, for example. For
many students it is also a rich source of
entertainment in terms of concerts, plays,
movies, social gatherings, and athletic
events as participant or spectator. And of
I course television programs may be as
readily available in college as they were
when the student was in high school.
Social relations involving
companionship, romance, sex, even
marriage compete with studying for a
course. And for those students who are
already married the demands and
pleasures of family life require a moderate
amount of maintenance time. What
variables are available to motivate the
study necessary to master the contents of
three to five college courses during a
semester instead of engaging in other
attractive or demanding activities? The
motivational variables usually mentioned
can be grouped into several classes, as
follows.
Intrinsic Interest in the Subject
Matter
Because of a favorable social and
intellectual background or unusually
effective lecture or text material, some
students find contact with a particular
topic sufficiently rewarding in itself that
they are happy to have the opportunity to
spend the necessary time studying.
However, I seriously doubt that this
factor can play much of a motivational
role. There are two main difficulties.
Very few students will find many of their
content courses of sufficient intrinsic
interest to maintain the necessary amount
of study time. Many topics are somewhat
interesting to most students, but this
interest is usually satisfied with far less
contact than is required by the instructor
for even barely adequate accomplishment.
In addition, this factor is unfortunately
susceptible to postponement as contrasted
with the competing activities, for which
A Behavioral Perspective on College Teaching 111
"seize the moment" is the relevant
attitude. The spontaneous social
gathering, the unexpected dating
opportunity, scheduled athletic events, all
must be taken advantage of at the time
they become available, but the aspects of
the text that make it intrinsically
interesting will be available and just as
interesting at a later time. f
Approval and Disapproval from
Significant Others in the Study
Setting .
The reactions of parents playa significant
role in motivating the study behavior of
some high school students, but this
influence is greatly weakened when the
student moves away from home. Even
for those still living at home, school work
becomes less subject to effective scrutiny
and evaluation by parents as a result of
the new topics being studied and the
parents' inability to understand the .
contingencies affecting the student at
college. Roommates might provide
appropriate approval and disapproval, but
are not likely to react in such a way as to
sul:!iect themselves to any form of .
aversive countercontrol. Thus, !lot !TIuch
can be expected from this factor, ailllih
any case it is not one that the instruCtor
has any control oveL .
Social Reactions of Others to the,
Repertoires Acquired Through
Study . '.,
Praise and admiration for displaYing .'
one's newly acquired knowledge may be
available from the instructor in class,
from other students in the same class, or
from other people in the academic or
nonacademic environment. The .
instructor has some control over this
variable within the class situation, but not
much when there are 40 or more students
in the class. l:he approval of others, in or
out of class, is not under instructor
control at all, and for most content
courses such approval is pretty skimpy
and not tightly related to mastery of the
assigned material. It is hard to imagine a
student giving up an attractive social
activity or an interesting entertainment for
 112 A Behavioral Perspective on College Teaching
fear of having to forego temporarily this
type of approvaL
Short·Term Advantages to Oneself
from the Newly Acquired
Repertoire
Some things learned in college courses
have relatively immediate practical value,
as when an automotive engineering
course facilitates repaU: of one's own car.
Or the things learned, as in an economics
course, may pennit a better understanding
of a puzzling political or economic
situation. Many of the content courses,
however important for their contribution
to further learning in the same field,
provide relatively little of this kind of
shon-term payoff. Even when they do, it
is not likely to suppon much sacrifice of
attractive competing activities, and in any
case it is not closely related to the details
of the study assignment or under much
control by the instructor.
Long~Range Payoffs
Getting a degree and a good job, going
to graduate school, contributing to the
improvement of the human condition, and
so on are events that are sometimes cited
as possible sourCes of motivation for
studying. Future possibilities of this son
may well have played a role in students'
and parents' decisions regarding college
attendance, and the general importance of
such events seems to contribute to their
strength as possible motivators.
However, their temporal remoteness
works in the opposite direction, and even
more critical is the impossibility of
relating such events to the details of a
weekly study assignment: It is easy to
believe that one can contribute to the
human condition even if one d6esn't
really understand a particular graph in the
textbOok. Also'these long-range payoffs
suffer from the sarne susceptibility to
postponement as intrinsic interest: One
can probably get almost as good a job on
graduating if instead of studying tonight
one takes advantage of an unusual, social
opponunity and studies tomorrow night.
Instructors should cenainly direct
students' attention to such future
possibilities when appropriate, but my
"
experience doesn't suggest that this type
of motivator plays any appreciable role in
maintaining daily and weekly study.
The Course Grade
Now we come to the one motivational
factor over which the instructor has
considerable control and which is easily
related to the details of the study
assignment. It is also a factor of
considerable strength, as evidenced by
the intensity of study activity occU¢ng
immediately before a major exam, Some
instructors will assen that grades do not
actually serve as effective motivators for
many of the current crop of college
students, and this issue will be addressed
later. But first a few comments about
university grading practices.
University Grading Practices
A university is a place in which students
acquire repertoires that are of value to
them and to society. It is generally
considered to be the responsibility of the
university to assess the extent and quality
of the student's accomplishment and
provide a record of this accomplishment,
the academic transcript. This record
becomes important when the student
attempts to qualify for a further
educational opponunity or for
employment. In some cases, the agency
or institution offering the further
opponunity will perform some
assessment-give a specially prepared '
examination or conduct an interview-but
most of the assessment is considered best
done by those who taught the student'! It
is certainly possible for a person to
acquire a professional intellectual or
technical repertoire without the aid of an
educational institution, but most graduate
schools and employers do not readily
accept the task of assessing the extent or
quality of such self-made repertoires.
1This involves some conflict of interest, of
course, because the educator's assessment that
most of the students failed to accomplish
anything would certainly raise questions about
the adequacy of the instructional technology.

Course grades are assigned
according to two quite different
principles, referred to as norm-referenced
and criterion-referenced grading. A grade
that reports the student's standing in the
course relative to the accomplishments of
the other students is norm referenced. An
instructor uses norm-referenced grading
when giving the grade orA to those
students who received the top 10% of the
exam scores. Criterion-referenced
grading involves grade assignment
according to an absolute standard,
irrespective of the performance of other
students. Giving the grade of A to those
students who obtain 90% of the available
exam points is criterion-referenced
grading. In practice, these two grading
principles are seldom seen in a pure form.
An instructor using a norm-referenced
approach may still refuse to give the top
10% an A if the overall class performance
seems below the usual level experienced
by that instructor, or may exceed the 10%
A assignment if the class seems unusually
effective. Likewise, an instructor using a
criterion-referenced approach may adjust
the absolute criterion downward if it
seems that an exam was inappropriately
difficult when other evidence implies
good effort and accomplishment by the
students.
At this point it will be useful to .
consider the often-discussed notion that
an emphasis on grades fosters an
undesirable form of competition among
students. It is important first to
distinguish between two kinds of
competition, which can be called vicious
competition and friendly competition.
Competition is vicious when an increase
l in one student's course grade will result
in a decrease in another'S; it is friendly
I when anyone who achieves at a certain
accomplishment level will receive the
grade appropriate for that level. It is quite
A Behavioral Perspective on College Teaching 113
c~ear that norm-referenced grading--
grading on a curve--is a vicious
competition contingency, because if only
the top ten percent of the class receives an
A, students improving their performance
and moving into that category bump other
students out Criterion-referenced
grading is a friendly competition
contingency, because each grade i~
assigned to anyone who achieves at that
level (gets a certain number of points
related to exams, papers, etc.)
irrespective of how many other students
achieve at that level. With criterion-
referenced grading there is, of course, the
competition related to having the highest
point score, or having a higher score than
some other particular student, but this is
not the type of competition that makes
students rejoice in the misfortune of other
students, or refuse to help others for fear
of lowering their own course grade. For
this reason as well as others, I believe
that criterion-referenced grading is the
only appropriate form of grading for'
college academic performance.' .
To return to grading practices, ,
various scales are used, with 'the most
common being the five levelA, B, C,D;
and E system, or an: eight level scale of
A, BA, B, CB, etc.(sometimes recOrded
as A, B+, B, C+, etc.) In general the'
larger the number of steps in the scale the '
more precise the information being,
reported, up to the point where the '
precision becomes unreasonable given the '
relative crudeness of the measUfement '
system I suspect that the eight~point
scale is about as precise as is appropriate'
for most classroom grading practices.
Near the end of the term, eight steps is
better than five, because the value of
further study vanishes when it cannot
lead to an advance to the next letter-"grade,
and this occurs earlier with the larger'
steps.
114 A Behavioral Perspective on Col/ege Teaching
SbJdying for an Exam
all
finished
safe
) no time left
start
;. time left to complete the task
The Procrastination Scallop
Studying for a college course exam is like
many tasks.for which a certain amount of
work must be completed before a
relatively inijexible deadline,and for
which the quantity and quality of the
work completed determines the value of
some subsequent event-in this case, the
grade earned on the exarn. As with many
such tasks, the work could be done at any
time during the interval between the
announcement of the task and the
deadline, but typically is not done until
near the end of the interval and then with
increasing frequency. This pattern of
behavior is sometime$ erroneously
referred to by behavioral psychologists as
aflXed-intervai scallop, implying the
same functional relations between
behavior, time, and reinforcement as seen
with fixed-interval schedules of
reinforcement. The pattern is similar, but
the controlling relations are quite
different.
For the students in my courses, the
task €onsists of studying a text and
lecture notes in conjunction with study
objectives, preparing further study
materials (flash cards, notes of various
kinds, etc.), and rehearsing or working
with these materials until they are well
very very
good good
~
"e
~
"0
§
~
very
brl L very
brl
learned. 2 The figure above portrays this
situation as a task-completion ftime-
passage space. At the beginning of the
study interval-immediately after
finishing the last exam-the student's
situation is represented by the bottom left
corner. Working toward c.ompleting the
task moves up; time passage moves to the
right. At the end of the interval (at the
time ofactually writing the exam), the
vertical location (how much of the task
has been completed) determines the value
of the outcome (the exam score). The
value of the exam score, in turn, is
largely a function of its relation to the
21 have always used essay exams in my courses.
The questions are typically the kind that require
at most a paragraph, usually a sentence or two,
sometimes just a single word as an answer. 1
recruit sbJdents who do well in the course one
semester as grading assistants for the next
semester, for which they receive some form of
course credit or just do it for the increased
learning that comes from such an activity. They
grade with a strigency bias, and 1review those
answers that were considered incorrect in any
way. My courses typically have from 40 to 80
students, and if1had to' teach a course with over
a hundred students, 1 would use multiple-<:hoice
questions. 1currently supervise a system
involving over 700 students in sections of 120
taught by graduate sbJdents. Assessment is
presently by 40-item multiple-choice exams
given every week.
 A Behavioral Perspective on College Teaching 115
overall course grade given at the end of
the semester.
Task completion and time passage
together determine the aversiveness of the
situation at any point in the interval
relevant to the task. To have completed
most of the task and to have a good deal
of time left before the end of the interval
is safe or nonaversive. But to have
completed very little of the task and to
have very little time left is a condition that
is dangerous or aversive, and that
generates escape behavior. Usually the
only relevant form of escape under these
circumstances is to work on the task, that
is, to study for the exam. The usual
reinforcement for studying thus consists
in the fact that studying changes one's
status with regard to the task-time relation
to a less dangerous condition. A number
of details are relevant to the dynamics of
this environment-behavior relation; these
will now be considered.
Competing Activities
As mentioned above, we can usually
assume that a number of opportunities or
demands regarding activities other than
studying will be superimposed on this
situation. I will also assume that if
studying has any intrinsic reinforcement
value, it does not compete effectively
with the reinforcing value of the other
behaviors, either because of its relative
weakness or because the other
opportunities and demands are much less
postponable. These assumptions mean
that studying the subject matter of the
college course will not ordinarily occur to
a sufficient degree to develop the mastery
required by the instructor unless the
studying is relevant to course credit and
to the course grade.
Course Grade Importance
Secondly, if the course grade is of little
importance to the student, then the
possibility of receiving a low exam score
will certainly not function as a basis for
aversive control, and studying as a form
of escape behavior will certainly not
compete with behavior related to other
sources of reinforcement.
The Relation Between Exam Grade
and Course Grade
Even if the co~e giade is important to
the student, anyone exam grade does not
determine the course grade by itself, but
does so only in combination with the
grades on the other exams and/or other
assigned course activities. This is the
implication of the diverging an:owion the
figure above leading from the exam
outcome to the course outcome. The
exam outcome influences the course
outcome but doesn't completely
determine it; this factor is responsible for
an unavoidable reduction in the
effectiveness of a low exam gi"ade as a
basis for aversive control. The student
can believe, and not incorrectly, that a
low score on the current exam can
possibly be compensated for by a high
score on a later exam. This source of
vagueness is enhanced when a large
portion of the course grade depends on
some end-of-semester activity SJlch as a
comprehensive final exam or a term .
paper. However, in many college .
courses there is a good deal of additional
vagueness in the relation betW!ien.. exam .
outcome and course outcortIeJhat results
from the instructor's general failure.to
specify precisely the basis for the c0w.:se
grade. Another source of vagueness IS.
the use of a norm-referenced or .. .
relativistic grading system under which ...
the exam and course grades are notonly ..
based on the student's own pelt}jrrtIance; .
but also depend to some extent on the
performance of other students inthe .
course.
The Relation Between Studying
and the Exam Grade
For the potential exam grade to motivate .
task completion before the time runs out,
the relation of task completion to exam
outcome musrbe a strong one. If failure
to complete a sizable portion of the task
does not result in a bad grade, or task
completion does not result in a good
grade, then studying as escape behavior
will not be strong, nor will task progress
function as a form of reward for the
various aspects of study behavior. The
 116 A Behavioral Perspective on College Teaching
most common difficulty here is the
instructor's failure to be sufficiently clear
about the relation of text and lecture
material to the exam content-usually out
of fear that with a more specific
assignment, the student would learn what
was specified but would not learn all the
other things that would have to be studied
if the exam content could not be
anticipated. Such specification may also
be withheld on the grounds that part of
the scholar's repertoire consists of
bringing order out of chaos, and having
to deal with the instructor's vague
assignments is a good opportunity to
acquire this skill. Students are often
heard to say that in some cases they
studied extensively and received a low
grade ll,J1d in others they studied hardly at
all and received a high grade. They will
say that they have no idea where the exam
questions came from, or that topics that
were barely touched in lecture and hardly
mentioned in the text constituted a .
significaritponion of the exam. Of
course, some of this complaint is
rationalization for poor performance due
simply to insufficient or ineffective study,
but as 'an ex-student I fmd myself quite
sympathetic to such complaints. In some
cases the only thing that saved me was
the instructor's strict adherence to a
norm-referenced grading system under
which no one learned what appeared on
the exams, but those who did least worst
received high grades.
In my courses, and in those of some
of my colleagues, students are provided
with weekly study objectives-statements
identifying essential parts of text and
lecture material. These function to define
the study task, making it possible for the
student to know at any point what
remains to be studied and when to quit
studying because the task has been
accomplished. Some study objectives
identify relatively explicit defmitions,
facts, principles, and so forth. but quite a
few are related to more complex kinds of
knowledge, such as being able to apply
some concept or principle to a novel
situation, identify or classify examples,
generate original examples, explain the
point being made by a text example, etc.
Digression on lecture content
and class attendance. For large-
enrollment content courses, lecture
attendance is closely related to how
essential the lecture material is for doing
well on the exam, not, as is widely
believed, how interesting or inspirational
the lectures are. Of course, other things
being equal, more interesting lecturers
command better attendance than those
who are less interesting, but it is easy to
overemphasize this factor. Because there
are usually pretty good textbooks for
such courses, the basic problem for many
instructors is how the lecture material
should be related to that presented in the
text. One popular, but I believe
inappropriate, approach, based on the
belief that many of the students will not
study the text, or if they do will not
understand it, is to provide in lecture
essentially the same information that is in
the assigned text readings. This renders
the lectures unnecessary for students who
can understand the text and the text
unnecessary for those who attend and
make good use of the lectures, and it
greatly reduces the amount of material
that can be adequately covered in the
course. Much better is to provide detailed
study objectives that guide the students'
contact with the text-indicating what is
important from the instructor's
perspective---and that contain instructor-
prepared supplementary material that
clarifies and expands on inadequate or
especially difficult sections in the text.
With such an approach the lectures can
then provide more recent developments in
subject matter, elaborate funher on
especially difficult text material, add
material that the instructor believes is
important to a thorough understanding of
the subject matter but that is not covered
at all in the text, and so on. 3 It should, of
3However, it is not reasonable for a very large
portion of the course content to be available only
as lecture material. Adequate text material can be
found for most specific-content courses at the
freshman or sophomore level, and too-heavy
 A Behavioral Perspective on. College Teaching 117
course, be clear by announcement and by
practice that much that is in the text will
not be covered in lecture but will
nevertheless be covered on exams. (The
world of scholarship depends on the
written word. College students must be
effective readers and studiers when they
come to college or they must quickly
.acquire these skills, or they should not be
able to obtain a college education. If they
can, the college education doesn't amount
to much.) It should also be clear, again
by practice as well as by announcement,
that any material presented in lecture that
is not in the text will be on the
examinations. Lecture attendance under
such conditions is generally quite
satisfactory, and not for artificial reasons.
Task And Time Monitoring
The student's task monitoring and time
monitoring, of course, also determine the
aversiveness of the situation, and thus the
frequency of the relevant escape
behavior. If the task is vaguely described
by the instructor or vaguely understood
by the student, the current situation will
not function as an accurate source of
aversive control, with the usual error
being that the student underestimates the
amount of work left to be completed.
Similarly if the student's time assessment
is inaccurate-the usual error being to
overestimate available study time-the
situation at any point in the assign-study-
test cycle will not be as aversive as it
should be. Even with accurate
monitoring, many of us have a significant
tendency in the direction of unrealistic
optimism, resulting in our being slightly
to moderately late in completing most
tasks. Weakness in these skills is of
importance, however, only at the
beginning of exposure to a well-managed
motivational system: The system itself
will automatically develop adequate task
and time monitoring after exposure to
several assign-study-test cycles.
reliance on lecture results in a reduced content and
is too dependent on students' note-taking skills.
If essential material is not available in text form,
then the instructor should write it.
: To summarize! the threat ofr~eiviilg.
a lower grade than IS acceptable willc,.",'.'\"
usually generate appropriate exam study .
behavior, but not if any of the following
cO';lditions prevail: (a) the coursegradeis
unllllportant to the student; (b) the relation.
of exam grade to course grade is unclear:
and (c) the relation of studying for the '
exam to the exam grade eamed is unclear.
Now back to the claim, often heard,
that many students are simply not
motivated by grades. Stich a claim is
undoubtedly based on instructors'
experiences, but before taking this claim
too seriously, we should first eliminate
evidence that is confounded with the
second and third deficiencies described
above. My experience has been that very
few students are willing to take a low
grade when the contingencies are
properly arranged.
Students do poorly in college for two
main reasons: inadequate intellectual
repertoires and ineffective motivational
contingencies. The present analysis deals
only with motivation. Even when the '.
detrimental conditions listed above arenot
present, students with grossly defective
rntellectual skills will not be easily .
successful. However, an effective
motivational system can go a long way
toward supporting the automatic , .
development of better intellectual 'skills.
The same is true with respect to self-
management-control by long-range
contingencies as opposed to ewessive .
influence by the pleasures and pains of
the moment. Self-management skills
develop as a result of personal experience
and instruction by others, but this .
development depends to some extent on a
relatively stable environment. Most .
students already have some tendency to
study rather than to play when it is
appropriate, and if doing so generally has
good results, and if failing to do so
generally has bad results, self-
management skills automatically improve.
Repeated exposure to an effective and
reasonable assign-study-test cycle can
play an important role in generating and
maintaining effective student behavior,
 118 A Behavioral Perspective on College Teaching
and seems especially imponant at the
freshman and sophomore levels.
Before drawing some general
conclusions, it is necessary to comment
on exam coverage and exam frequency.
For motivational purposes4 any s~dent
who masters only a small proportIon of
the assigned material s~ould ..
systematically and ineVItably fail the . will only be present during the last week
exam and students who master most of It
should always receive high exam grad~s.
Exam coverage insufficient to accomplish
these goals greatly weakens the
effectiveness of the exam as a
motivational basis for studying rather
than doing other things. If stud~nts
correctly believe that they can skip
studying any of the assigned materi~ and
still get'an acceptable grade, many will do
so.
. With respect to exam frequency, my
goal is t9 generate six or more hours of
effective out-of-class study every week.
WhelH give an exam every week, and
when the exam covers most of the
assignment, such study does take place.
I have been teaching this way for about
20 years, and there is no question in my
mind that it prOduces the necessary study
by most of the students. What if I gave
an exam every three weeks? I use one-
hour essay exams, and it is quite clear
that if I gave such an exam pver three
times as much material, it would have to
be a much smaller sample of the material
assigned. This would result in a less
accurate measure of accomplishment and
would constitute an instigation to omit
some part of the assignment because of
the distinct possibility that it w~uld not
appear on the exam. But more Imponant,
if it takes six or more hours to master a
4It may be.possible to show thai an ex~ that
coven> only a moderate sample of the assIgned
material will nevenheless function as an adequate
assessment device, but this is irrelevant to the
motivational issue. For motivational purposes,
the student must correctly believe that failure to
complete even a small portion of the assigned
study task will inevitably lead to an appropriate
lowering of the exam grade and ultimately the
course grade.
week's assignment, then it should
Genainly take 18 or more hours of out-of-
class study to master a· three-week
assignment.. The nature of ~e
procrastinallon scallop descnbed above
strongly suggests, and experience
. ..
verifies, that sufficient fear to generate
study in competition with other aCllVIttes
of a three-week periOd, and it is absurd to
suppose that students.will systeJ?atically
put in 18 hours studYIng for a smgle
course during anyone week. What.
happens is that instructors lower therr
standards so as not to fail too large a
proportion of the students-they do not
expect 18 hours of outside study for an
exam over a three-week assignment.
Even less do they expect 30 or more
hours of study for an exam given after
five weeks of instruction with no
previous exams. Under such conditions .
there develops a strong tendency to favor
norm-referenced grading and give high
grades to the best students, even though
they didn't learn very much.
Because exams are seen as the most
aversive aspect of college leaming, it has
been suggested that a system involving
weekly extensive-coverage exams would
be highly unpopular with students, but I
have not found this to be the case. The
idea is unpopular, but on course .
evaluations most students strongly praise
the system, correctly identifying the
weekly exams as being responsible for
generating extensive and well-paced
study. Students who can be identifi~ as
highly effective-those who are eammg
grades of A and BA-seem espe~ally
satisfied with such a system, saymg that
they very much appreciate knowing
exactly what they have to study (as a
result of having the detailed weekly study
objectives), being able to thoroughly
master small units of work, and being
induced to spend much more tiI?e
studying than with many of their other
courses.
Conclusions
1. It is unrealistic to consider
sources of motivation other than exam
 A BehavioralPerspective on College Teaching 119
and course grades to be of much
imI!Ortance in generating study in the
typiCal content course.
2. Exam grades, course grades, and
what they represent can function to
motiv~te a more than adequate
commitment to out-of-class study, if three
important conditions are met: (a) The
course grade is important to the student.
(b) Exam grades are closely related to
course grades. (c)Studying is closely
related to exam grades.(The instructor has
little or no control over the first, but is
completely responsible for the other two.)
3. To meet these conditions:(a)
Exams must be extensive and must occur
quite often.(b) Assignments must be quite
specific as to the relation between text and
lecture material and exam content.(c)
Weekly exams must contribute in a large
way to the course grade. Said another
way, fmal exams and term papers cannot
contribute too much to the course grade.
4. There are three popular notions
about college teaching that I believe are
quite incorrect for the type of course I
have been considering, and when taken
seriously impede the development of
effective instructional systems. The first
is that if you teach properly, the students
will fmd learning both fun and easy. Not
true, for two reasons: intensity and
assessment. Leaming new things is often
enjoyable, but seldom so under
conditions in which a great deal must be
learned in a relatively short time. The
college student taking several content
courses will have to spend over 40 hours
a week attending classes and studying
outside of class. But as described above,
such a student is also physically, socially,
:md economically prepared to participate
m a great many other available activities
of a kind that require little or no effort,
and are either intrinsically interesting or
achieve ample extrinsic reward. Students
will often find themselves saying
something like "I can't do it now because
I have to study." Turning down many
attractive opportunities in favor of, at
best, moderately interesting textbook and
lecture study is essentially aversive. In
addition to this source of aversiveness,
college learning is the kind of activity that
is not taken for granted, but rather has to
be formally assessed. Spending time and
effort trying to accomplish something
with the distinct possibility that when it is
assessed the result will be disappointing
is also an intrinsically aversive situation.
The aversiveness is, of course, tolerated
because of the long-range value oftile
repertoires being acquired, but the value
of the remote end doesn't completely
eliminate the aversiveness of the more
immediate means to its achievement.
The second false notion is that
grades should not be emphasized Not
true! Grades are the primary motivative
variable, which means that current
standing in the course and progress
toward a final grade should always be
clear and frequently brought to the
student's attention. With the current
ready availability of microcomputers and
spread sheet programs, an instructor can
easily provide students with a weekly
printout showing their past and present
performance along with that of the other
students, providing a projection to their
final course grade, and indicating how
they must perform on remaining course
activities to achieve any particular course
grade. I provide this type of weekly
information, and it is always well
appreciated. On course evaluations many
students write that they wish all their
instructors gave them such information.
The third false notion is thllt good
teaching consists primarily of good
lecturing. Of course it is important for
lecture material to be interesting and
valuable for learning the course subject
matter, but lecture presentation is only a
small part of the total activity. It is much
more important that the instructor have an
effective instructional system involving
clear study objectives over both text and
lectures, frequent well-designed exams,
an exam envrronment that renders any
form of cheating relatively useless,
appropriate remedial opportunities for
students who experience occasional poor
performances, and so on.
5. Effective instruction, as implied :
above, is labor intensive. The notion that
120 A Behavioral Perspective on College Teaching
a content-course system that generates
effective intellectual repertoires in 40 or
more students can be managed and
continuously updated by spending little
more than the time required to present
three one-hour lectures a week is way off
tjIe mark.
,
6. Perhaps in a more general sense it
would be useful to conclude as follows:
The world runs on feat. College learning
is largely under aversive control, and it is
our task to make such control effective, in
which case it becomes a form of gentle
persuasion.
16 !

REFERENCES
Alling, K. (1991). The effects of a conditioned establishing operation on performance of
a two-component chain. Unpublished master's thesis, Western Michigan University,
Kalamazoo.
Azrin, N. H., & Ho1z, W. C. (1966). Punishment. In W. K. Honig (Ed.), Operant
behavior (pp. 380-447). New York: Appleton-Century-Crofts.
Bailey, J. S., & Bostow, D. E. (1979). Research methods in applied behavior analysis.
Tallahassee, FL: Copygrafix.
Baum, W. M. (1973). The correlation-based law of effect. Journal ofthe Experimental
Analysis ofBehavior, 20, 137-153.
Bernstein, D. J. (1990). Of carrots and sticks: A review of Deci and Ryan's Intrinsic
Motivation and Self-Determination in Human Behavior. Journal of the Experimental
Analysis ofBehavior, 54,323-332.
Bijou, S. W., & Baer, D. M. (1961). Child development Vall: A systematic and
empirical theory. New York: Appleton-Century-Crofts.
Blakely, E. & Schlinger, H (1987). Rules: Function-altering contingency-specifying
stimuli. The Behavior Analyst, 10,183-187.
Calvin, J. S., Bicknell, E. A., & Sperling, D. S. (1953). Establishment of a conditioned
drive based on the hunger drive. Journal ofComparative and Physiological,
Psychology, 46, 173-175. . .
Caroll, R. J. & Hesse, B. E. (1987). The effects of alternating mand and tact training on
the acquisition oftacts. The Analysis ofVerbal Behavior, 5, 55-65.
Catania, A. C. (1979). Learning. Englewood Cliffs, NJ: Prentice-Hall.
Catania, A. C. (1984). Learning (2nd ed.). Englewood Cliffs, N.J: Prentice-Hall.
Collier, G., Hirsch, E., & Kanarek, R. (1977). The operant revisited. In W. K. Honig & J.
E. R. Staddon (Eds.), Handbook ofoperant behavior (pp. 28-52). Englewood Cliffs,
NJ: Prentice-Hall. .
Commons, M. L., & Nevin, J. A. (Eds.). (1981). Quantitative analyses ofbehavior: Vol.
1. Discriminative properties of reinforcement schedules. Cambridge, MA: Ballinger.
Cravens, R., & Renner, K. E. (1970). Conditioned appetitive drive states: Empirical
evidence and theoretical status. Psychological Bulletin, 73, 212-220.
 122 References

D'Arnato, M. R, Fazzaro, J., & Etkin, M. (1968). Anticipatory responding and avoidance
discrimination as factors in avoidance conditioning. Journal ofExperimental
Psychology, 77, 41--47.
Darwin, C. (1859) On the origin ofspecies by means of natural selection. London:
Murray.
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human
behavior. New York: Plenum.
Dickinson, A. M. (1989). The detrimental effects of extrinsic reinforcement on intrinsic
motivation. The Behavior Analyst, 12,1-15.
Dinsmoor, J. A. (1977). Escape, avoidance, and punishment: Where do we stand?
Journal ofthe Experimental Analysis ofBehavior, 28,83-95.
Epstein, R, Lanza, R P., & Skinner, B. F. (1980). Symbolic communication between
two pigeons (Columba livia domestica). Science, 207, 543-545.
Estes, W. K. (1944). An experimental study of punishment. Psychological Monographs,
57, (Whole number 263).
Fantino/E., & Logan, C. A. (1979). The experimental analysis ofbehavior. San
Francisco: Freeman.
Farris, H. E., Gideon, B. E., & Ulrich, R. E. (1970). Classical conditioning of aggression:
a developmental study. The Psychological Record, 20,63-67.
Ferster, C.B.; & Skinner, B. F. (1957). Schedules of reinforcement. New York:
Appleton-Century-Crofts.
Fine, T. L. (1973). Theory ofprobability. New York: Wiley.
Hall, G. & Sundberg, M. L. (1987). Teaching mands by manipulating conditioned
establishing operations. The Analysis ofVerbal Behavior, 5, 41-53.
Hart, B. M. & Risley, T. R. (1975). Incidental teaching of language in the preschool.
Journal ofApplied Behavior Analysis, 8, 411-420.
Hayes, S. C., & Hayes, L. J. (1989). The verbal action of the listener. In S. C. Hayes
(Ed.), Rule-governed behavior (p. 185). New York: Plenum Press.
Hayes, S. C., Zettle, R. D., & Rosenfarb, I. (1982, May). An empirical taxonomy ofrule-
governed behavior. Paper presented at the meeting of the Association for Behavior
Analysis, Milwaukee.
Herrnstein, R. J. (1969). Method and theory in the study of avoidance. Psychological
Review, 76, 49-69.
Herrnstein, R J., & Hineline, P. N. (1966). Negative reinforcement as shock-frequency
reduction. Journal ofthe Experimental Analysis ofBehavior, 9, 421-430.
Hersen, M., & Barlow, D. (1977). Single case experimental designs. New York:
Pergamon Press.
Hinelin.e, P.' N. (1977). Negative reinforcement and avoidance. In W. K. Honig & J. E.
R. Staddon (Eds.), Handbook ofoperant behavior (pp. 364--414). Englewood Cliffs,
• NJ: Prentice-Hall.
Holland, J. G., & Skinner, B. F. (1961). The analysis ofbehavior. New York: McGraw-
Hill.
Honig, W. K. (1966). Operant behavior: Areas of research and application. New York:
Appleton-Century-Crofts.
 References 123

Honig, W. K., & Staddon, J. E. R. (Eds.). (1977). Handbook ofoperant behavior.

Englewood Cliffs, NJ: Prentice-Hall. '
Hull, C. L. (1943). Principles ofbehavior. New York: Appleton-Century-Crofts.
Johnston, J. M., & Pennypacker, H. S. (1981). Strategies and tactics of human
behavioral research. Hillsdale, NJ: Erlbaum.
Kantor, 1. R. (1959). Interbehavioral psychology. Granville, OH: Principia Press.
Keller, F. S., & Schoenfeld, W. N. (1950). Principles ofpsychology. New York: f
Appleton-Century-Crofts.
Loeb, J. (1916) The organism as a whole,from a physicochemical viewpoint. New York:
Pumam.. .
Lowe, F. (1979). Determinants of human operant behavior. In M. D. Zeiler & P.
Harzem (Eds.), Reinforcement and the organization ofbehavior (pp. 159-192). New
York: Wiley.
Lundin, R. W. (1961). Personality, an experimental analysis. London: MacMillan.
Lundin, R. W. (1969). Personality, a behavioral analysis. London: MacMillan.
Lutzker, J. R, & Martin, J. A. (1981). Behavior change. Monterey, CA: Brooks/Cole.
Mach, E. (1883) The Science ofMechanics. Chicago: Open Court.
Malott, R,'(1984). Rule-govemed behavior, self-management, and the developmentally
disabled. Analysis and Intervention in Developmental Disabilities, 4, 199-209. .
Mazur, J. E. (1986). Learning and behavior. Englewood Cliffs, NJ: Prentice-Hall.
Mazur, J. E. (1990). Learning and behavior (2nd 00.). Englewood Cliffs, NJ: Preritice-
Hall.
McPherson, A., & Osborne, J. G. (1986). The emergence of establishing stiniuhis'"
control. Psychological Record, 36, 375-386.
McPherson, A., & Osborne, J. G. (1988). Control of behavior by an establishing'
stimulus. Journal ofthe Experimental Analysis ofBehavior, 49,213-227.
Michael, J. (1982a). Distinguishing between discriminative and motivational functions of
stimuli. Journal ofthe Experimental Analysis ofBehavior, 37,149-155.'
.,
I . Michael (1982b). Skinner's verbal operants: Some new categories. VB News, I, 1
Michael, J. (1983). EvOCative and repertoire-altering effects of an environmental event.
The Analysis ofVerbal Behavior, 2, 19-21. '
Michael, J. (1984). Verbal behavior. Journal of the Experimental Analysis ofBehavior,
42,363-376. .
Michael, J. (1985a). Behavior analysis: A radical perspective. In B. L. Hammonds
(Ed.), Master lecture series: Vol.e 4. Psychology of learning (pp. 95-121).
Washington, DC: American Psychological Association. .
Michael (1985b). Two kinds of verbal behavior plus a possible third. The Analysis of
Verbal Behavior, 3, 1-4.
Michael, J. (1986). Repertoire-altering effects of remote contingencies. The Analysis of
Verbal Behavior, 4,10-18.
Michael,1. (1988). Establishing operations and the mand. The Analysis ofVerbal
Behavior, 6, 3-9.
124 References

Michael (1991a). A behavioral perspective on college teaching. The Behavior Analyst,

14, 229-239.Michael, J. (1991b) Historical antecedents of behavior analysis. The
ABA Newsletter, Summer, 7-12.
Millenson, J. R. (1967). Principles ofbehavioral analysis. New York: MacMillan.
Millenson, J. R., & Leslie, J. C. (1979). Principles ofbehavioral analysis (2nd. ed.). New
York: Macmillan.
Miller, N. E. (1951). Leamable drives and rewards. In S. S. Stevens (Ed.), Handbook of
experimental psychology, (pp. 435-472). New York: John Wiley & Sons.
Mineka, S. (1975). Some new perspectives on conditioned hunger. Journal of
Experimental Psychology: Animal Behavior Processes, 104, 143-148.
Minervini, M. (1985). Metaphorical extension in the analysis ofbehavior. Paper
presented at the meeting of the Association for Behavior Analysis, Columbus, OH.
Minervini, M. (1989). An Analogy ofScience. Unpublished doctoral dissertation,
Western Michigan University, Kalamazoo, MI.
Morris, E. K., Higgins, S. T., & Bickel, W. K. (1982). The influence of Kantor's
interbehavioral psychology on behavior analysis. The Behavior Analyst, 5, 159-173.
Morse, W. H. & Skinner, B. F. (1958). Some factors involved in the stimulus control of
operant behavior. Journal ofthe Experimental Analysis ofBehavior, 1 103-107.
Palmer, D; C. (1991). A behavioral interpretation of memory. In L. 1. Hayes & P. N.
Chase (Eds.), Dialogues on verbal behavior (pp.261-279). Reno, NY; Context
Press.
Pavlov, I. P. (1927). Conditioned reflexes (G. V. Anrep, Trans.). London: Oxford
University Press.
Peterson, N. (1978). An introduction to verbal behavior. Grand Rapids: Behavior
Associates.
Powers, R. B., &Osborne, J. G. (1976). Fundamentals ofbehavior. St. Paul, MN: West.
Rescorla, R. A., & Wagner, A. R. (1972). A theory ofPavldvian conditioning:
Variations in the effectiveness of reinforcement and nonreinforcement. In A. H.
Black & W. F. Prokasy (Eds.) , Classical conditioning II: Current research and
theory (pp. 64-99). New York: Appleton-Century-Crofts.
Ryle, G. (1949). The concept ofmind. New York: Harper & Row.
Rumbaugh, D. M. (Ed.). (1977). Language learning by a chimpanzee: The Lana project.
New York: Academic Press.
Russell, B. (1927) Philosophy. New York: Norton.
Savage-Rumbaugp, E. S. (1982). Verbal behavior at a procedural level in the
chimpanzee. Journal of the Experimental Analysis ofBehavior, 41, 223-250.
Schlinger, H. & Blakely, E. (1987). Function-altering effects of contingency-specifying
stilllUli. The Behavior Analyst, la, 41-45.
Sechenov, I. M. (1863) Reflexes of the Brain. St. Petersburg: Sushchinskii.
Sherrington, C. (1906). The integrative action ofthe nervous system. New York:
Scribner's.
Sidman, M. (1960). Tactics ofscientific research. New York: Basic Books.
 References 125

Sidman, M. (1962). Reduction of shock frequency as reinforcement for avoidance

behavior. Journal ofthe Experimental.Analysis ofBehavior, 5, 247-257.
Sidman, M. & Tai1by, W. (1982). Conditional discrimination vs. matching to sample:
An expansion of the testing paradigm. Journal ofthe Experimental Analysis of
Behavior, 37, 5-22.
Skinner, B. F. (1938). The behavior oforganisms. New York: Appleton-Century-Crofts.
Skinner, B. F. (1945). The operational analysis of psychological terms. Psychological
Review, 52,270-277. .
Skinner, B. F. (1948). Walden two. New York: Macmillan.
Skinner, B. F. (1950). Are theories ofleaming necessary? Psychological Review, 57,
193-216.
. Skinner, B. F. (1953). Science and human behavior. New York: Macmillan.
Skinner, B. F. (1957). Verbal behavior. New York: Appleton-Century-CroftS.
Skinner, B. F. (1966). What is the experimental analysis of behavior? Journal ofthe
Experimental Analysis ofBehavior, 9, 213-218.
Skinner, B. F. (1969). Contingencies ofreinforcement. New York: Appleton-Century-
Crofts.
Skinner, B. F. (1971). Beyondfreedom and dignity. New York: Bantam/Vintage.
Skinner, B. F. (1979). The shaping ofa behaviorist. New York: Knopf.
Skinner, 1981. Selection by consequences. Science,213, 501-504.
Skinner, B. F. (1974). About behaviorism. New York: Vintage.
Skinner, B. F.,& Vaughan, M. E. Enjoy old age. New York: Norton.
Smith, L. D. (1986) Behaviorism and logical positivism. Stanford, CA, Stanford
University Press.
Smith, L. D. (1992). On prediction and control: B. F. Skinner and the technological ideal
of science. American Psychologist, 47,216-235. .
Stafford, M., Sundberg, M. L., & Braam, S. J. (1988). A preliminary investigation of
the consequences that define the mand and the tact. The Analysis ofVerbal ,Behavior,
6,61-71.. .
Staats, A. W. (with contributions by C. K. Staats) (1963). Complex human behavior.
New York: Holt, Rinehart and Winston.
Staats, A. W. (1968). Learning. language and cognition. New York: Holt, Rinehart and
Winston.
Staats, A. W. (1971). Child learning, intelligence and personality. New York: Harper
and Row.
Staats, A. W. (1975). Social behaviorism. Homewood, IL: Dorsey Press.]
Staats, A. W. and StaatS, C. K. (1958). Attitudes established by classical conditioning.
Journal ofAbnormal and Social Psychology, 57,37-40.
Stokoe, W. C., Casterline, D. C. & Croneberg, C. G. (1965). A dictionary ofAmerican
Sign Language on lingUistic principles. Washington, D. C.: Gallaudet College Press.
Sundberg, M. L. (1987). Teaching Language to the Developmentally Disabled: A Course
Manual. Prince George, B.C.: College of New Caledonia Press.
126 References

Teitlebaum, P. (1966). The use of operant methods in the assessment and control of
motivational states. In W. K. Honig (Ed.), Operant behavior: Areas ofresearch and
application (pp. 56~08). New York: Appleton-Century-Crofts.
Teitlebaum, P. (1977). Levels of integration of the operant. In W. K. Honig & J. E. R.
Staddon (Eds.), Handbook ofoperant behavior (pp. 7-27). Englewood Cliffs, NJ:
Prentice-Hall.
Thorndike, E. L. (1898). AnimalinteIIigence: an experimental study of the associative
processes in animals. Psychological Review Monograph Supplement, 2(8).
illlman, L. P., & Krasner, L. (Eds.). (1965). Case Studies in Behavior Modification.
New York: Holt, Rinehart & Winston.
Ulrich, R. E., & Azrin, N. H. (1962). Reflexive fighting in response to aversive
stimulation. Journal ofthe Experimental Analysis ofBehavior, 5,511-520.
Ulrich, R., Stachnik, T., & Mabry, J. (Eds.). (1966) Control ofHuman Behavior, (Vol. 1).
Glenwood, IL: Scott, Foresman.
Vaughan, W. (1984). Optimization versus response-strength accounts of behavior.
Journal ofthe Experimental Analysis ofBehavior, 42, 337-348.
Watson, J. B. (1913) Psychology as the behaviorist views it. Psychological Review, 20,
158-177.
Williams, B. (1976). The effects of unsignaled delayed reinforcement. Journal of the
Experimental Analysis ofBehavior, 26, 441-449.
Young, F. A. (1965). Classical conditioning of autonomic functions. In W. F. Prokasy
(Ed.), Classical conditioning (pp. 358-377). New York: Appleton-Century-Crofts.