Download as pdf or txt
Download as pdf or txt
You are on page 1of 39

Alternative Sets as a Framework for Behavioral Formulations and Research

Author(s): Israel Goldiamond


Source: Behaviorism, Vol. 3, No. 1 (Spring, 1975), pp. 49-86
Published by: Cambridge Center for Behavioral Studies (CCBS)
Stable URL: http://www.jstor.org/stable/27758831 .
Accessed: 22/06/2014 00:53

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Cambridge Center for Behavioral Studies (CCBS) is collaborating with JSTOR to digitize, preserve and extend
access to Behaviorism.

http://www.jstor.org

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as a Framework for
Behavioral Formulations and Research

Israel Goldiamond1
The University of Chicago2

The relation between talk about events, and what we otherwise do about
how we
them, is, of course, of interest not only to students of philosophy but also to students of
behavior analysis. The terms we use tend to classify our observations. They may thereby
dictate what it is thatwe admit as data, and govern our data collecting procedures. In a
scientific discipline, terminological problems can lead to the formulation of scientific
puzzles and to the pursuit of experimental problems which might otherwise not have been

pursued. The present discussion concerns such a case study.


In this issue, Michael and Mixon independently address themselves to the terms,
negative and positive reinforcement, and punishment. Both note the confusion generated
as well.
by present usage not only among laymen, but among professional psychologists
Both suggest the replacement of the three terms by, simply, reinforcement and punish
ment, although they arrive at this conclusion by different routes. They differ inwhat they
would retain. Mixon is content to live with negative and positive reinforcers, but Michael
would excise both as well as those compound terms containing the modifier, aversive.
He would replace them all by reinforcement and punishment. The two terms would refer
tomaintenance by consequential means. He could live with
and decrement of behavior
-ment
simple derivatives such as reinforcers and punishers. The differences between the
and -er words would be governed by standard English usage as applied to other words
(e.g., government, governors, governed). Definitions would be provided by their
contexts.

Both authors address themselves has been noted by others as


to a problem which
well. The terminological confusion currently prevailing is indeed appalling. Moreover,
suggestions are being made which would encumber the language with even more terms,

1
Professor of Psychology in the Departments of Psychiatry and of Behavioral Sciences (Biopsychology),
and in the College. Mailing address: Department of Psychiatry (Box 411), The University of Chicago, 950
East 59th Street, Chicago, Illinois 60637.
article was written with support
2This from a grant from the State of Illinois, Department of Mental
Health, entitled: "Self-Control Procedures: Variables in Training Social Agents and Patients in Construc
tional Diagnosis and Intervention." The author wishes to express his appreciation toMarvin Segal, Research
Associate in the Department of Psychiatry, for helpful comments and criticisms of sections of the manus

cript. The views expressed are those of the author. The figures were made by Mrs. Judith E. D. MacPhail.

49

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

eg., that there be two punishments, negative and positive, to parallel the two reinforce
ments; that passive and active avoidance be substituted for punishment and avoidance.
An overhauling seems overdue. The use of confusing terms is not restricted to behavior
analysis, of course. Nor
is the precise specification of different procedures always
accompanied by equal in the descriptive terms used for them. This is not an
niceties
apology for slovenly, confusing, or out-dated terms. Other disciplines live with such
terms (possibly not happily) as long as their procedural and relational referents are not
confused thereby. Here, I wish to address a different problem, not considered by the
suggested revisions. These revisions implicitly accept an extant framework of discourse
which can present problems more serious than terminological confusion. It has also
produced confusion in social application. An alternative framework addressed to these
issues will be proposed. This framework also specifies more closely than the present one
certain of the behaviors of the investigator in socially applied situations, and in some
cases also does so in a laboratory situation.
The two frameworks
to be contrasted are ( 1) the present unilinear framework which
describes behavior
by the contingencies into which it enters,3 and (2) a proposed
framework of alternative sets, which requires description not only of behavior and its
relation to environmental events, but also of alternative sets of behavior-environment
relations. In a previous discussion (Goldiamond, 1974, pp. 54-63) some of the alterna
tives were sketched and their relevance to unilinear descriptions was noted in passing.
Here, that discussion will be expanded insofar as itbears on the present terminological
issue.

The present unilinear system has only one well-defined set, with the behavioral
element designated as the response (R). The alternatives are
("competing behavior")
defined solely by negation or exclusion from that set. For a
example, key-peck with a
specified range of force may be R, with everything else excluded from consideration.
Machine definition can readily enter. It becomes redundant to consider
competing
behaviors, since R. This exercise can be
they merely complement logically trivial.
Further, the explanatory status of competing behaviors ismade untenable by the fact that
they are poorly defined as the open expanse of anything outside the boundaries of the
well-defined R set.
However, triviality is not the case when there are at least two well-defined sets. Such
designs are sample by research in concurrent behavior, but are not limited to them.
Indeed, the matching relation (Herrnstein, 1961) involves such "choice" (cf. Baum,
1974). The present discussion will consider as basic the situation inwhich there are at
least two well-defined sets. This situation will be considered as theminimal case of those
which have a larger number of well-defined sets; the present formulation treats only one
well-defined set as the basic one.
The two well-defined behaviors which enter into the two sets of behavior
environment relations of the proposed system will be
designated as target behavior (TB)

as used here, refers to the discriminative


'^Contingency, repertoire of the investigator who defines (a) a
response, (b) a stimulus, and (c) the relationship between the two whereby he make /; contingent on a.
Contingency can be extended to include other events but this is not relevant to the distinction between
frameworks discussed.

50

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

and alternative behavior (AB). A third class which is defined by exclusion of both TB and
AB will be designated as neither behavior (NB). The NB category can be subdivided into
other well-defined behaviors, and the classification can be extended, but for purposes of
discussion, two well-defined behaviors will suffice.
To any student of stimulus control, such a classification should present no conceptual

problems. Two well-defined sets typify op?rant research in stimulus discrimination and
stimulus control. Indeed, consideration of only-one well-defined stimulus set as basic,
rather than minimally-two, has led to differing experimental results. These differences
have been used to support and generate theoretical differences which are not trivial in
science or application.
In the instrumental situation in which there is only one well-defined set, Sd is
topographically defined as a stimulus, say, a key which is trans-illuminated red, inwhose
presence a response is consequential, and SA is defined as anything else, that is, the
not-red inwhose presence the response is inconsequential. Stimulus discrimination is
defined as differential responding between these "two." Generalization has been loosely
defined as responding to some stimulus other than the training one (Sd). Since no two
stimuli are every completely identical, the question of the extent to which Caesar will
keep putting his foot into a riverwhich differs from the one of original immersion, has not
been considered a trivial one. The curve which describes the functional relation between
(y) different rates, strengths of behavior and (x) stimuli which differ in their distance
from the stimulus inwhose presence control was established is defined as the generaliza
tion gradient. The stimulus dynamism of classical behaviorism is derived from experi
ments inwhich a "tent-like" gradient was produced, that is, the further the stimulus was
from the supporting pole (the original stimulus in whose presence reinforcement was
provided) on either side, the smaller its effects upon behavior. Control was considered
quite specific to the original stimulus.
K?hler' s experiment is one of the crucial experiments forGestalt theory as opposed to
behaviorism. Chicks found food available in a medium gray box rather than a light gray
one. When the light gray box was replaced by a dark gray box, the chicks should have
gone to themedium box, response towhich had previously been reinforced, according to
the classical behaviorist theory of specific stimulus control, as exemplified by the
tent-like gradient. Instead, theywent to the (different) dark box. Behavior was governed
by the relations between stimuli rather than by their absolute properties. Behaviorism had
been dealt a mortal blow and a configurational (Gestalt) theory was upheld! In applica
tion, educators sought to teach children to read by the contextual properties and
memorized form of the whole word rather than by the specific letter-sounds of the
alphabet.
What was overlooked at the funeral was that classic formulations rested on data from
situations with only one well-defined set, and K?hler used two. Indeed, behaviorism is
not discomfited in the latter case, which ismore representative of op?rant discrimination
research. Here Sd may be a well-defined red key in whose presence responding is
consequential, and SA may be a well-defined greey key inwhose presence responding is
inconsequential. The fact that responding in the presence of anything else is also
inconsequential, namely, responding to anything which is not red (including the key

51

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

when it is not illuminated or a bolt on thefloor) does not, in the usual discussions, lead to
formulations of red versus everything else. Stated otherwise, the discussions center on
differential responding in the presence of red and in the presence of green. Here, the

generalization gradient which can be obtained is described as a "peak-shift." As the


name suggests, a novel stimulus can produce ^higher rate of responding than the original
stimulus which had hitherto been the peak of the curve. This can occur when it is further
removed from SA on the continuum described by Sd-SA than is the original Sd.
K?hler's results are replicated, and within a behaviorist framework, when his conditions
are met. The differences in results obtained by classic investigators and K?hler can be

interpreted in terms of differences between consideration of Sd as the one well-defined


stimulus, with SA defined as anything else, and consideration of Sd as one well-defined
stimulus, with SA being another equally well-defined stimulus.

A series of experiments by Schoenfeld and Farmer (1970) inwhich both TB and AB


were well-defined is relevant. TB was a well-defined key peck reinforced, on the
=
average, every twentieth (p .05) occurrence. Since AB would occur at a time other
than TB, the total duration of TB-related time was, in essence, subtracted from total
session duration.4 The alternative time was subdivided into equal intervals, the total
number of whichequalled the total number of TB during that session. Since TB
= .05 schedule differed for each bird, the frequency of the AB
frequency under the p
units also differed and, accordingly, unit size differed for each bird. AB is not simply an
inter-response interval, since theremight be three units in one TB-TB interval, two in
another, etc. (Each TB reset the AB-unit counter.) Over twenty post-stabilization
was then
sessions were required to establish the size of the AB unit for each bird. AB
=
reinforced on schedules ranging from p = .0125 to .20; TB remained on the p .05
schedule. AB was a function of its schedules. Having established thatAB, like TB, was
amenable to schedule control, a series of different experiments was then run, and in the
final one they noted that "when an independent variable (a DRL schedule) can produce

seemingly divergent findings" with respect to TB in the standard literature, "other


factors must be the causes. In the present case, the contribution [of AB] seems evident."
How these sets are "manipulated will help determine what is chosen as the sample R in
any experiment" (p. 244).

The Present Schema

The discussion will open with a brief consideration of the highlights of the present
unilinear definitional schema. Detailed coverage is not attempted.
? The major contingencies which underlie the present de
Major contingencies :
are are on the left,
scriptive system presented inFigure 1. The presentation contingencies
and the withdrawal contingencies are on the right.

The schematic diagrams depicted describe the experimenter s laboratory arrange

4
The procedure used was more complex, but the present description is close enough for our purposes.

52

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

merits. The uppermost line describes TB.5 The line below defines the consequences
which the experimenter arranges, contingent on the presence or absence of TB. The third

A. Presentation contingencies B. Withdrawal contingencies

- -
R? R
0 0
?
JT1
O?
o
CB
Z LT CBZ
LT U LT

Figure 1. Schematic diagrams of the basic defining contingencies of the

prevailing unilinear system.

line describes the undefined set of all other behaviors (competing responses). Time flows
from left to right. A perpendicular line through all three lines describes contemporaneous
events.

The diagrams depict a pattern of behavior, by a naive organism, subjected to various


consequential arrangements. Behavior is presented in two states: behavior present and
behavior absent; length of the horizontal indicates duration. Where the line ascends,
behavior occurs. Descent is its null state. Similarly, where the stimulus line forms a
plateau, the experimenter has introduced a stimulus. The stimulus is either "positive" or

"negative," depending on which the experimenter had introduced. Where this stimulus
is absent (withdrawal or zero state), the line is at the lower null position. Positive and
negative stimuli are not depicted opposites. Both are presentations of choice by the
investigator, and the presence of either is indicated by the same upper position.
In the presentation relations, positive reinforcement and punishment are both de
scribed by the same procedural schematic. They may be restated unilinearly in the

following manner:
[1.1] TB?? Event delivered.
If pTB then increases (or ismaintained), positive reinforcement is defined, and the
event delivered is assigned a reinforcing property. If pTB decreases (or ismaintained at a
lower probability), punishment is defined, and the event delivered is considered as being
aversi ve (or negatively reinforcing).6
In thewithdrawal relations, negative reinforcement and time-out are both described
by the same procedural schematic, restated unilinearly in the following manner:

5In the conventional notation system, TB is the response, designated R. It is so noted in the schematic

diagrams. For purposes of comparison in the text, it is referred to as TB, namely, the behavior which is the

target of our interest, whether or not an AB has been explicitly considered.


"The reader should note that for a variety of experimental and theoretical purposes, the probability of
on occasion,
behavior, pB, is not identical to the rate of behavior and can, produce conflicting expectations.
For the purposes of this article, this important distinction is being overlooked, and pB is being used
for probability of behavior, or its rate, or both.
generically

53

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

[1.2] TB^ Event removed.


If pTB then increases, negative reinforcement is defined, and the event then removed
is assigned an aversi ve (or negatively reinforcing) property. Indeed, both Keller and
Schoenfeld (1950), p. 61) and Skinner (1953), p. 73) agree on the definition of negative
reinforcers as those events which strengthen behavior or increase its probability when
they are removed. If pTB decreases, time-out from positive reinforcement, a form of
punishment, is defined, and the event removed is considered as being positively reinforc
ing. Time-out from positive reinforcement is discussed in different contexts elsewhere
(cf. Herrnstein, 1955; Ferster and Skinner, 1957).
For both presentation and withdrawal relations, for all four contingencies, extinction
is procedurally described by the following unilinear statement of a relation substituted for
procedures [1.1] and [1.2] some time after their introduction:
[1.3] TB ?> No contingent environmental changes.
If the effects of the contingencies described by [1.1] and [1.2] are undone overtime,
extinction is defined. If the effect of [1.3] is decrease in pTB, it can be assumed that the

contingencies in [ 1.1] and[ 1.2] were (positive or negative) reinforcement. If the effect of
[ 1.3] is increase inpTB, itcan be assumed that the prior contingencies were punishment.
Itwill be noted that description of the (behavior ?? consequence) procedures ( [1.1],
[1.2] ) differs from definition of the contingencies. These are defined functionally, that
is, by the effects the procedures have on behavior. The event delivered or removed is also
defined functionally.
The parsimony of the present definitional system is evident from
Figure 1, and may
have contributed to its staying power. Where TB is depicted as the same in both
presentation (left) and withdrawal (right) relations, the corresponding consequences
which are made contingent on TB will have identical temporal patterns, but are comple
ments of each other, in accord with differences between
presentation and withdrawal.
one of freedom is
Accordingly, only degree (df) logically necessary. For each contingen
cy, CB ismerely the complement of TB, since it is defined only by negation of TB. Only
one df holds here, as well. And reinforcement and
punishment are functionally defined
by opposite effects on behavior. Accordingly, attempting to explain an obtained effect on
TB by an opposite effect on CB (e.g., TB is maintained because CB is is
punished)
questionable on at least three grounds. First, since one df describes TB and CB, such
explanation ismerely a restatement and provides no logical gain. Second, ifan explana
tion in terms of TB is unsatisfactory, then by virtue of the
logical identity, an explanation
using an opposite function for the of TB must also be unsatisfactory. Third,
complement
whereas TB constitutes a well-defined CB is the
set, poorly-defined class of anything
else and can be regarded as an open set with questionable
explanatory value.
In summary, it should be noted that the present definitional
system is highly
parsimonious and logically elegant. It has served well to formalize much
laboratory
research. This elegant formulation has served to provide a framework for discussion of
laboratory findings which have suggested ever-expanding areas of investigation.
Discriminative and nondiscriminative avoidance: ? One other form of negative
reinforcement will be noted in addition to the escape contingency. Two avoidance forms
are outlined in
Figure 2. Discriminative avoidance is depicted on the left, and nondis

54

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

stages of establishment (acquisition) are depicted: prior to


criminative on the right. Three
establishment, during establishment, and avoidance behavior established.
In one form of discriminative avoidance (left), a preaversi ve (or conditioned
is presented, usually a tone or a signal light. Its termination is coupled with the
stimulus)
presentation of a noxious stimulus, such as shock, for a brief period. TB at any point in

A. Discriminative avoidance B. Nondiscriminative avoidance

r r
o o
-
iti
s s
o o?

cb ? cb ?
o
- U-LT 0
-

Figure 2. Discriminative and nondiscriminative avoidance, described by


the prevailing unilinear system.

the coupling terminates the presentation(s) at that point. During establishment, TB may
occur during the noxious component, thereby terminating it, as depicted by the shortened
consequence. This is the escape behavior of Figure 1. In the established state, TB
terminates or escapes from the preaversi ve stimulus, and the coupled noxious stimulus
is, accordingly, never delivered. It is avoided.
Since TB is defined by discrete occurrence, and nontarget behavior
a is anything else,
the conventional systems of description allow for only one df.
In the early stages of nondiscriminative avoidance (NDA), presented at the begin
a noxious stimulus such as
ning, a recycling timer periodically produces delivery of
shock. Absence of TB is accompanied by such cyclical presentation of shock. There is no
no warning is given. Each full cycle of the clock is indicated
preaversive stimulus, that is,
by an arrow. TB interrupts the clock cycle and resets it.As can be seen in the schematic

diagram, appropriately spaced TB is accompanied by the delivery of no consequences.


Further, there are no other stimulus changes associated with it. The term, avoidance, is
In discriminative avoidance (left), TB during the preaversive
clearly descriptive.
stimulus may be defined either as escape from the preaversive stimulus or avoidance of
the aversive stimulus (shock), depending upon the writer. Occurrence of TB in either
case defines negative reinforcement, and the effective stimulus is assumed to be nega
tive.

The scientific puzzle provided by NDA may be introduced by the following unilinear
statement of successful avoidance:
[2.1] TB?? Event not delivered (No environmental change).
Such behavior is characterized by itspersistency. If the experimenter disconnects the

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

aversi ve consequences (the negative reinforcer), the following unilinear statement is

descriptive:
[2.2] TB ~? Event not delivered (No environmental change).
This, of course, is extinction, cf. [1.3]. As far as an observer who is naive about the
actual contingencies (such naivete can also characterize the subject himself), Lines [2.2]
and [2.1] are identical, yet one is defined as a behavioral maintenance procedure, and the
other as an extinction procedure. Indeed, under the extinction procedure, well
established behavior can continue for an inordinate period of time.

Another puzzle is generated by the absence of stimulus change in [2.1]. Prior to the
establishment of successful NDA avoidance, there were no antecedent discriminative
stimuli associated with absence or delivery of consequences, and after its establishment,
there are no stimulus changes generated by themaintaining consequence itself! The issue
of explaining the persistence of behavior under such conditions has attracted considera
ble theoretical and experimental attention.

One resolution has been to reject reinforcement theory as inapplicable. Another has
been to use some term suggesting abnormality, such as compulsion. Yet another has been
to maintain that (negative) reinforcement is still operative, but it has become inter
nalized, hence its external (re)presentation is no longer necessary. It is here that the
two-factor theory (Mowrer, 1947) has been pressed into heroic service. This theory
opens with discriminative avoidance. According to the theory, the preaversi ve stimulus
becomes classically conditioned to the aversive event. This event is internalized as
anxiety. Accordingly, the preaversive stimulus produces anxiety (factor one). In ter
minating this stimulus, TB terminates (escapes) anxiety, and ismaintained as moperant
(factor two) by this consequence. In NDA, which lacks such stimulus changes, anxiety
builds up over time during the timer cycle. This is terminated by TB in accord with the
termination of the preaversive stimulus. As in psychoanalytic theory, the compulsive
behavior wards off anxiety. This time interval is considered as the conditioned aversive
temporal stimulus by Anger (1963), in a parallel theory which omits terms such as
anxiety. The "need to rationalize stimulus nonoccurrence as reinforcing" (Schoenfeld,
1969) has also been met by assigning the reinforcing function to other nonobservables
such as proprioceptive-kinesthetic stimuli, as Schoenfeld also observers. Increase inTB,
through attenuative punishment of CB, has also been pressed into service.

Sidman, on the other hand, has attributed the increase in response probability to

consequential (overall) reduction in delivery of shock (1962). And, in accord with this
interpretation, Hineline and Herrnstein (1970) note that a difficulty with "traditional
[avoidance] theories [is that they] . . .were conceived to explain responding to proce
dures that provide an immediate stimulus change after each response" (emphasis
mine?I.G.). Their experiment is one of the many generated by these interpretations,
and probably the most significant one to date. Indeed, the role
played by unilinear
description of the procedure (Line [2.1] ) in creating the puzzle is suggested by the
opening sentence of their study: "What is the reinforcer for avoidance responding?"
Their results support neither the overall shock reduction
interpretation nor the two-factor
theory, and question the ability of the latter "to define the substantive issues in avoid

56

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

ance" (p. 126). An experimental extension by Hineline (1970) is more definite in its
challenge to the two-factor theory. Space precludes further discussion.

Alternative Sets

Although the descriptions made will be set in the context of laboratory arrangements,
the formulations and outcomes so derived have often been extended to application and

practice, and some implications will be noted.


For a given arrangement, two types of situations and their outcomes will be com
are so arranged
pared. In the firstof these, the investigator so arranges the events, or they
in the natural environment, that only TB is operative. AB, so to speak, is idling. This
situation will always be depicted on the left-hand side of a schematic diagram. The
second situation is so arranged that both AB and TB are operative. This situation will
always be depicted on the right-hand side of a schematic diagram. For both first and
second situations, NB is depicted as inoperative.
The Law of Effect refers to effects on TB, as does the present system, and such
outcomes will be the focus of the discussion. They critically enter not only into the
functional definitionsof the present system, but into decision-making for applied

purposes. ABis ignored. The outcomes so obtained in the present schema should be
similar to the gross expectations derived from an analysis of alternative sets, when AB
idles and TB alone is effective (left schematic). To the extent that similarity in expecta
tions occurs, the alternative-sets formulation does not violate present understanding. In
the present schema, competing behavior is the residual outside the well-defined re

sponse, which is effective. NB is also a residual when TB and AB are each effective

(right schematic). Here, the outcomes presently obtained should be the gross expecta
tions for the combination (TB, AB). However, the TB outcome, the presently-defined
"outcome", should then not be generally comparable for unilinear and alternative sets.
To the extent that an effective AB enters into a situation described unilinearly, TB can
become unpredictable. The second situation then becomes a requisite for understanding
and application. This point has been made for the laboratory by Schoenfeld and Farmer
(1970). For application designed to change behavior, and for understanding (theory)
derived from the natural environment, the differences are impressive. Use of unilinear
formulations, as will be noted, is not confined to behavior analysis.
It should be reiterated that the expectations presented are gross. They are not intended
to serve as a theoretical explanation of present schedules, nor as solutions to the puzzles
they present. Such explanation and resolution require a fine-grain analysis under condi
tions which make itpossible, namely, the laboratory. Nor are the expectations intended
to serve explanatory functions for themore "socially relevant" phenomena of applica
tion. Analogous immersion in these phenomena is also necessary. The presentation is
intended only to recommend a framework which may be more useful for fine-grain
analysis and application than the present unilinear one, and to explore relations to the
terminological issues raised.
The settings inwhich the two situations and their outcomes are compared will include
"pr?sentation" and "withdrawal" contingencies. These contingencies will be consid
ered separately. The former will be characterized as a setting in which there is a

57

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

continuous zero state, interrupted occasionally by discrete positive or negative presenta


tions. The latter will be characterized as a situation in which there is a continuous
presentation state, positive or negative, interrupted occasionally by discrete zero states.
Continuous states can, of course, be described in terms of successive discrete events, and
discrete events as episodes from a continuous state, with a continuum constructed either
way. The first arrangement will bear an odd number, and the second an even one, so that
3-4, 5-6, refer to pairings whose major procedural differences are continuous zero and
continuous presentation.
The schematic diagrams will follow those which described the present definitional
will describe three
contingency, with exceptions to be noted. Each schematic diagram
pairs of events (three sets) occurring at the same time. The first set will describe TB on
one line and stimulus events on the line beneath. The second set directly beneath itwill

similarly pair AB and contemporaneous stimulus events. The third set will pair NB and
contemporaneous stimulus events. Since this pair is defined only by negation of (AB,
TB), itwill be presented in full only when necessary.

Continuous Zero State: Discrete Presentation Imposed

The alternatives described when the experimenterinterrupts a continuous zero state


by making a discrete presentation contingent on behavior are depicted in Figure 3.

A. TB alone effective B. TB, AB each effective

TB
O? n n_TLn
?
?
o

AB
b
0 n
?
?
0

?
NB
0
-
TUUriuru
?
0_

Figure 3. Continuous zero state interrupted by discrete presentations.


Alternative sets.

As
is evident from the schematic diagram for TB alone
operative (left), each TB is
of a discrete event. whether
accompanied by presentation AB, occurring instead of TB
(b), or at some other time (a), is a zero state of the environment, as in
accompanied by
occurrence of NB. Nonoccurrence of NB (the competing response explanation) is
accompanied by a discrete presentation only if nonoccurrence represents occurrence of
TB, and not when it represents occurrence of AB. Even when AB idles, NB is not the
complement ofTB.

58

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

Extension of the schematics over time suggests that if the alternative sets are
continually in force, and they are accompanied by increase in pTB, with reference toTB
alone the procedure would be conventionally defined as positive reinforcement, and the
stimulus event defined as a (positive) reinforcing stimulus. Where pTB decreases, the
stimulus is assumed to be aversive and the procedure is defined as punishment. These
expectations accord with present formulations.
As is evident from the schematic diagram forTB and AB each operative (right), TB is
accompanied by presentation of an event, and AB is also so accompanied whether
instead of TB (b) or at some other time (a). NB is accompanied by a zero state, but its
nonoccurrence is always accompanied by presentation of an event. Accordingly, its
nonoccurrence becomes
explanatory (but see earlier discussion) only when AB is
explicitly effective. Under such conditions, as will be seen, it can become difficult to
define effects on TB. These procedures and outcomes are related to no existing designa
tions.

Where are effective, presentations will occur (with respect to theTB


both TB and AB
set) also when TB does not occur, that is, when AB occurs. Similarly, with respect to
AB, presentations will occur in its absence. This also holds with respect toAB when only
TB is effective. This joint-environment-effect (j.e.e.) is described in the presentations
related to NB in both sides of Figure 3, and may be substituted for consequences
arranged by the experimenter, depicted for TB and AB in both sides. The j.e.e. iswhat
the subject may, of course, experience. Where the existence of AB is overlooked, in the
?
+ case, where TB and AB are effective, it is "free" reinforcement. Similarly, in the
case, it is "arbitrary" punishment.
The alternative (behavior-environment) sets can be described as follows:
a. TB alone operative b. TB, AB, each operative
TB?> Discrete presen. TB?> Discrete presen.
[3.1] AB^ Continuous zero AB?> Discrete presen.
NB?> Continuous zero NB^> Continuous zero
?
With regard to assigning + or status to the presentation, scanning [3. la] suggests
that if pTB increases, the presentation functions positively, and if it decreases, it
functions negatively. With regard to [3.1b], the same conclusions can be derived for
p(TB, AB). However, from the effects on pTB alone (the conventional basis for stimulus
definition), inference is hazardous. Decrease of pTB may be relatable to either a negative
presentation (TB?? ?) or to a positive presentation (TB?? + ). The latter decrease may
accompany an increase in pAB (AB?? +) when its repertoire is stronger and the

contingency thereby preempts TB. Inference is safe only when pTB increases (TB?? + ),
but prediction is then not possible.
If the joint environment effects are considered, the alternative sets from the subject's
vantage point are the following for [3.1a], TB alone operative:

TB absent^ Contin. zero TB present^ Discrete pres.


[3.1j] AB abs-pres^ Contin. zero AB abs-pres^ Discrete pres.
NB abs-pres^ Contin. zero NB abs-pres^ Discrete pres.

59

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

Comparison of left and right sides indicates thatTB is the critical behavior here. The
discrete presentations for AB, NB can be considered "free" presentations.
-
A stimulus having been assigned + or status through functional scanning of [3.1],
itmight now be applied in a contingency to change TB. The expected long-range
outcomes may be inferred grossly from inspection of limits:
a. TB alone operative b. TB, AB each operative
- -
Presentation + Presentation Presentation + Presentation
TB lim-? 1.00 TB lim-> .00 TB lim-> .00
TB]
[3.2] AB lim-* .00 AB? >lim-? 1.00 AB lim-? .00
NB lim-? .00 1.00 A?] NB lim-? 1.00
|lim-> NB lim-? .00
NB)

Where either (TB, AB) approach lim?? 1.00, prediction of outcome of one is, of
course, impossible, since either can rise as the other decreases, or both can rise in a
variety of ways. A more interesting inference emerges from comparing the effects of
presenting 4- in [3.2a] and [3.2b]. The p(TB) should increase in [3.2a], but not
necessarily in [3.2b]. Expectations derived from one situation may not hold for the other.
The sets described when TB and AB are each effective are by no means trivial. A
social system which desires increase of some specified TB for individuals under its
purview must engineer the situation extremely carefully. As long as AB is present and
produces the same positive consequences as TB, itmay replace it, especially if it is less
costly or more readily available in the response repertoire. If theAB substitute is costly to
the system, it may attempt to decrease AB through means described as punitive.
Examples of such patterns are cheating, stealing, lying, or any one of a variety of other
shortcuts. Indeed, laboratory animals have also been known to engage in such shortcuts,
for example, discriminating the sound of their own relays, rather than the "approved" Sd
and SA. The experimenter also must ever be on the alert. Many patterns of neurotic
behavior may also be considered as examples of an AB class under the benign positive
stimulus sets of [3. lb]. The question of why the patient (or system) then applies for
therapy has already been answered. He may be "hurting" from the negative conse
quences^/?? also apply toAB in an effort to decrease it,or AB may be preempting TB.
History will enter importantly into the availability of TB, AB repertoires. The therapeu
tic effort described as pathological will focus on AB and itshistory, at least initially, and
may regard it as a class to be eliminated. However, the therapeutic effort described as
constructional will seek to ascertain from AB itsmaintaining consequences and seek to
program a TB repertoire which produces the same consequences. The history sought will
be relevant to such tasks (Goldiamond, 1974). A patient I interviewed at a state mental
hospital clearly indicated the existence of such alternatives. To attain sustenance and
shelter when he had outworn his stays at all the homes of his friends and relatives, he
could either engage in criminal behavior and be sent to prison, or engage in crazy
behavior and be sent to themental hospital, or engage in neither behavior and starve and
die of exposure. Viewed unilinearly, engaging in behaviors whose consequence is
confinement in a ward in a state hospital does not make "sense," hence is "crazy." Nor
does repeated self-infliction of pain make "sense," hence is "masochistic." Indeed,

60

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

much of classic psychiatric description is unilinear, and the classic op?rant unilinear
system is not alone in this regard. They both may represent a Zeitgeist inwhich such
was taken for The now be ? witness the
unilinearity granted. Zeitgeist may changing
attacks against "mechanistic" approaches. Returning to the patient under discussion, his
"crazy" behavior did not represent "psychosis," nor would criminal behavior have

represented "criminality." Both were AB which were maintained by the same conse
quences, namely, sustenance and shelter, albeit in different packages. The socially
a
approved TB, working for salary, was in his repertoire, but its availability was not in
the repertoires of prospective employers who avoided hiring him or who fired him when
they found out about his hospital record. His "psychotic" behavior, considered in terms
of alternative sets, made eminent sense, as did the behaviors of his employers. These
were not necessarily governed by his "labelling" as a psychotic, as some sociological
theorists would currently hold in a unilinear statement (Labelling?> Nonhiring). Rather,
they could be related to consequences to employers made by insurance companies, often
based on actuarial tables.
It may also socially useful to encourage AB, as when
be inventions and other

labor-saving alternatives to current procedures produce the consequences presently


on TB. Indeed, it
may be argued that those whom we regard pathologically as
contingent
character or emotional disorders may be regarded constructionally as people who have
been responsive to shaping by nonconventional social programs which provide conse
quences we obtain more conventionally. Accordingly, they have often been credited with
ingenuity and creativity. However, the fact that thousands of patients engage in similarly

disturbing behaviors suggests that the social programs are not that rare, and that they are

subject to the same trite program-behavior relations that we are.


Where both TB and AB produce punishment, AB may also be substituted, if it is less
not
costly, as in plea-bargaining. That the offering of sacrifices to propitiate the gods is
confined to ancient history is indicated by the recent sacrificial offering of smaller fry to
avert the forced resignation of the target. Although this involved sacrifice of others, one
can figuratively or literally "give one's right arm."
An inference which is not evident from consideration of either [3.1a], TB alone
effective, or [3. lb], TB and AB effective, does appear when they are compared, or are
considered as alternatives. The issue is one of coercion. It can be argued that positive
reinforcement is used coercively when the controlling system makes a potent conse
quence contingent only on the TB it requires, and especially so when it also creates the
conditions which make the consequence potent. A system which creates some critical
state of need, such as famine (or doesn't create it, but takes advantage of it), and then
makes food delivery contingent only on the specified behavior repertoire it desires,
or acceding, can certainly be called
leaving the recipient with the choice of starving
coercive. The moral indignation aroused by suggestions of social application of op?rant
laboratory procedures derives not from misunderstanding, but from a literal application

of standard laboratory procedure, which involves a TB alone effective [3. la] paradigm.
However, to the extent that AB's are available [3. lb], for the same positive presenta
tions, the equation of TB^ discrete presentation with coercion may become inappro
The can escape the critical state of need by engaging inAB other than
priate. organism

61

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

the TB the system has established. It is not necessarily coerced intoTB. To get TB, the
or capitalizing on an extant history,
system might then facilitate it through programing
among other procedures.
Such design also relieves behavior analysts from the conceptual dilemma created
when a consequence iswithheld (zero state) unless TB occurs, and all other behaviors are
ill-defined. When is defined in terms of behavioral outcomes, that is,
punishment
functionally (Azrin and Holz, 1966), the results can be described either as punishment of
non-TB or reinforcement of TB. Since the defining procedures are identical, the terms
Nor does insistence thatwe
punishment and reinforcement appear to lose theirmeaning.
differentiate extinction (zero state) from punishment wash well given functional defini
tion under such circumstances. Conceptualization in terms of alternative sets helps
resolve the dilemma, and consideration of alternatives between sets of alternatives is
even more helpful.
Consideration of the joint-environment-effect when the subject overlooks AB has an
for
interesting implication for investigator reliance on subject's report of his experiences
The subject may ascribe his behavior to all kinds of other
explanatory purposes.
relations, and this can define superstitious experience. This can also be defined, with
respect to AB, when he overlooks TB when it alone is effective, or when both are. Of
course, if the investigator is similarly ignorant, he, too, may derive spurious relations.
On the other hand, both may be spurred to more careful investigation. In all events,
reliance on the subject's reported experience is hazardous, even when his report is a
"valid" indicator of his subjective experience, at least when the conceptual system is
described in terms of alternative sets.

Continuous Presentation: Discrete Cessation

The alternatives described when the experimenter interrupts a continuous presenta


tion by making a discrete zero state contingent on behavior are described in Figure 4.

A. TB alone effective B. TB, AB each effective

TB -
0 ?

-
O o
b - a b a a b
AB AB
0
? J~L O
?
ru7L_n_n_a.

-
0 O
?
NB
O
-
Untnun nu
O -

Figure 4. Continuous presentation interrupted by discrete zero periods.


Alternative sets.

62

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

As is evident
from the schematic diagram for TB alone effective (left), TB is
accompanied by a brief zero state. This is followed by a continuous presentation. AB,
whether occurring instead (b), or at some other time (a), is accompanied by a continuous
presentation. Occurrence of NB is also accompanied by a continuous presentation.
However, is accompanied by the substitution of a brief zero state only
its nonoccurrence
when TB occurs
instead of NB, and not when AB occurs instead of NB. Stated
otherwise, all occurrences of NB are accompanied by continuous presentation, but all
nonoccurrences of NB are not accompanied by zero states. NB is not the complement of
TB.

of the schematics suggests that if the alternative sets are continually in


Extension
force, and they are accompanied by increase in pTB, with reference to TB alone the
procedure would be conventionally described as negative reinforcement, and the
stimulus event defined (the non-zero state) as a negative reinforcer or aversi ve stimulus.
Where pTB decreases, the presentation is assumed to be (positively) reinforcing and the
procedure is defined as time-out from positive reinforcement. These expectations accord
with present formulations.
As
is evident from the schematic diagram forTB and AB each operative (right), TB is
accompanied by a discrete zero state, and AB is also so accompanied whether instead of
TB (b) or at some other time (a). The nonoccurrence of NB is equivalent to the
occurrence of either TB or AB and is relatable to the zero state. The procedures described
are related to no existing designations.

A joint-environment-effect analogous to thej.e.e. considered for Figure 3 also holds


here, of course, and will not be elaborated.
The sets can be described as follows:
a. TB alone operative b. TB, AB each operative
TB?> Discrete zero TB?? Discrete zero
[4.1] AB?> Continuous presen. AB?> Discrete zero
NB?? Continuous presen. NB?> Continuous presen.
-
With regard to assigning 4- or status to thepresentation, scanning [4. la] suggests
that if pTB increases, it functions negatively, and if pTB decreases, it functions
same conclusions can be derived for p(TB, AB).
positively. With regard to [4. lb], the
However, from the effects on pTB alone, inference is hazardous. Decrease of pTB may
- .
occur either when the prevailing presentation is + or In the + case, TB interrupts the
?
prevailing positive environmental state, and should decrease.
In the case, either TB or
AB interrupts the prevailing negative state, and ifAB> TB (in terms of repertoire,

history, happenstance occurrence at the time, etc.), pAB may increase, and pTB then
decrease. Inference
concerning the prevailing presentation is safe only when pTB
?
the presentation is then negative. Such inference about the presentation must
increases
be distinguished, of course, from prediction of pTB. The former is possible for a negative
state only if pTB increases. This depends on AB.

Consideration of the joint-environment-effect here is analogous to [3. lj] for [4. la],
TB alone effective.

63

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

TB absent?? Contin. pres. TB present?* Discrete zero


[3. lj]AB abs-pres.^ Discrete zeroes
AB abs-pres.^ Contin. pres. NB abs-pres.-? Discrete zeroes
NB Contin. pres.
abs-pres.-^

Comparison of leftand right sides indicates thatTB is critical; theAB, NB zeroes can
be considered as "free" discrete cessations.
-
The presentation having been assigned + or status through functional scanning of
[4.1], itmight now be applied in contingency sets to change TB. The long-term expected
outcomes may be inferred grossly from inspection of limits:

a. TB alone operative b. TB, AB each operative


? ?
Presentation + Presentation Presentation + Presentation
TB lim-? .00 TB lim -? 1.00 TB lim-? .00
TB]
[4.2]AB] AB lim-> .00 1.00
1.00 |lim->
AB lim-? .00 A?)
[lim-> NB \\m-> .00 NB lim-? 1.00 NB lim?> .00
NB]
Where either of two sets approaches lim?? 1.00, prediction of outcome of one is, of
course, impossible. A more interesting inference emerges from comparing the effects of
-
presenting in [4.2a] and [4.2b]. The pTB should increase in [4.2a], but not necessarily
in [4.2b]. Expectations derived from one situation may not hold for the other.
Comparison of discrete presentation during continuous zero (the '4presentation
contingencies," Figure 3) and discrete zero during continuous presentation (the "with
drawal contingencies," Figure 4) reveals an interesting paradox between functional
definition and prediction when (TB, AB) are effective.
Functional definition is possible only when "reinforcement" procedures (positive or
negative) are applied. While pTB may not increase after discrete presentation during
continuous zero (pAB may increase instead), if itdoes then increase, the presentation is
positive. While pTB may not increase after discrete zero during continuous presentation
(pAB may increase instead), if it does then increase, the presentation is negative.
Increase in pTB, while not inevitable, permits functional definition of positive and
negative reinforcement and the corresponding presentations as positive or negative,
respectively. In contrast, ifpTB decreases after discrete presentation during continuous
zero this may occur when a negative presentation accompanies TB, or when a positive

presentation for (TB, AB) effective increases pAB instead. Further, if pTB decreases
after being accompanied by discrete zero interruption of a continuous presentation, this
may occur because pAB increases instead when interruptions in a continuous negative
presentation are attached to each, or because TB accompanies interruption of a continu
ous positive presentation. Accordingly, functional definition of procedures and presenta
tions based on relations between procedures and changes inpTB is possible only when
positive or negative reinforcement procedures are applied either to TB alone (the
classical or (TB, AB) effective. Such definition is not possible when either
procedure)
punishment procedure is applied to (TB, AB), although it is possible when TB alone is
effective. The classical insistence on basing functional definition on pTB incremental
procedures, i.e., reinforcement, rather than d?cr?mentai punitive procedures turns out to

64

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

have beena happy choice when alternative sets are considered. Basing functional
definitions on
incremental procedures instead of d?cr?mentai procedures appears.as
simply arbitrary when a unilinear approach is used.
With regard to prediction of pTB, the same analysis which indicates that functional
definition is determinate only during reinforcement paradoxically indicates that predic
tion of pTB is determinate only when d?cr?mentai procedures ("positive" or "negative")
are applied. When (TB, AB) are operative, positive or negative reinforcement applied to
both may result in either an increase or decrease in pTB, with decrease occurring ifpAB
increases (TB, AB) are effective, punishment of each, either by applying
instead. When
a discrete presentation during continuous zero, or a discrete zero during continuous
presentation, should result in decrease of the effective behavior. When TB alone is
effective, either form of punishment should produce attenuation of pTB. Accordingly,
prediction of pTB is possible only when d?cr?mentai (punitive) procedures are used.
The statement that the effects of punishment are unpredictable appears to be a
generalization from limited conditions. The world outside the laboratory abounds inAB
sets, and such an analysis suggests that the continued social use of punishment may
reflect its predictive properties for TB, rather than reflecting ignorance of the lessons of
behavior analysis.
It is also instructive to examine negative reinforcement when (TB, AB) are operative.
Lewin (1935) places a child into a psychological box whose walls are impenetrable
barriers which radiate negative valences in all directions around him (e.g., Figures 13
and 14, page 135). Rather than engaging in TB, the child may "escape the field" by
lifting himself, so to speak, off the page along a different dimension through fantasy or
psychological withdrawal. Skinner notes that "if you make a student study to escape
or
punishment, then he will soon escape in other ways; he'll play hookey, be a truant,
become a dropout. Or he may counterattack. Vandalism against school property is easily
explained just by looking at the techniques schools use to control their students" (in
Evans, 1968, p. 34). Both stress the controls necessary to prevent AB, in contrast to the
ease of instituting positive reinforcement. Under negative reinforcement, AB "necessar

ily occurs unless especial measures are taken to prevent it" (Lewin, 1935, p. 125,
emphasis Lewin's). All of the foregoing comments about negative reinforcement are

unquestionably supported by social experience, but itdoes not necessarily follow that use
of positive reinforcement relieves the system of the necessity for taking "especial
measures ... to prevent" AB. Such socially-undesirable AB were noted in discussion of
(TB, AB) effective under continuous zero interrupted by discrete presentation. And it
does not follow that use of positive reinforcement is any the less coercive than use of
negative reinforcement. Noncoercion requires that "especial measures are taken." These
measures would not include tightening of barriers to prevent AB, but the opposite,
AB rests
namely, to allow forAB, so that the increase of system-defined TB rather than
upon considerations other than simple escape from the coercion of an applied positive or
reinforcement system. Other procedures have been discussed elsewhere (Gol
negative
diamond, 1974).
Prisons are, indeed, expensive both in construction and maintenance. Analyzed in
terms, what is about the may not involve their use
contingency expensive contingencies

65

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Gold?amond

as instruments of social punishment (decrease of "criminal" behavior) or social negative


reinforcement (avoidance of incarceration through "prosical" behavior). Rather the
expense may be related to the differences between the requirements for continuous zero
state (Figure 3) and continuous presentation (Figure 4). In the latter, the system may be
required to introduce and maintain such continuous presentation, whereas in the former,
the same presentation is discrete. Negative reinforcement can, accordingly, be more

expensive than discrete presentation of the same event. An applied behavior analyst who
seeks to "meliorate" in a socially-established
the conditions continuous (negative)
presentation institution by introducing some "positive reinforcement" in this situation,
at null states? within the confine of a
may best be introducing occasional continually
negative system. The time-out procedure can be cut of similar cloth. Indeed, removal of
freedom contingent on TB is relatable to the negative reinforcement situation just
described. In one children's home I visited, the time-out room could best be described as
a small jail on the premises of a therapeutic institution.Where the total environment is a
positive one, and the staff is pleasant throughout as a standard operating procedure, then
time-out may be inexpensive to introduce. Similarly, where the conditions are exceed
ingly harsh, for example, intense starvation and deprivation, as a standard situation
(recurrent famines), then negative reinforcement (occasional relief) may also be an
economical procedure to introduce. The change agent may readily find some people who
are "willing to sell their souls" for a slight respite. Citing the conditions under which the
procedures are inexpensive does not imply endorsement. It is simply intended to
demonstrate that comparison of the relative costs of different contingency systems
requires careful analysis.
Systems which institute procedures which disturb us can be viewed pathologically as
incompetent, malicious, "sick," and so on. Their institutional behavior may also be
examined constructionally. The maintaining consequences of such behavior may be the
optimization of certain cost-benefits. Other (TB) procedures may have been tried and
may not have produced such optimization. Other historical features may enter, as well as
present knowledge and availability of alternative procedures. The task of the social
analyst-programmer would seem evident, namely, to help construct and systematize
programs and demonstrate their effectiveness in producing the critical consequences
currently governing socially-disturbing procedures. All the ethical and co-contractual
issues applicable to individuals will enter. It would seem that we might divert to the
development of constructional theory and procedures some of the prodigious effort

already expended on pathological theory and eliminative procedures. In such an effort,


explicit consideration and formulation in terms of alternative sets would appear to be
critical.

Continuous Zero State: Cyclic Discrete Presentation Imposed

The alternatives described when a recycling timer interrupts a continuous zero state
by presenting a discrete event, with behavior interrupting the cycle, are depicted in
Figure 5.
As evident from the schematic for TB alone effective (left), TB interrupts the timer
cycle, and the timer resets for a full cycle. The presentation is delivered on completion.

66

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

A. TB alone effective B. TB, AB each effective

jxxiJri
?
?
o

Figure 5. Continuous zero state, cyclic presentations. Alternative sets.

When TB occurs at less-than-cycle (LTC) intervals, the timer is continually reset and
there are no presentations. AB, whether occurring instead of TB, or at some other time,
may be accompanied or followed by a presentation or zero, as in NB.
The joint-environment-effect noted earlier will analogously be present here, except
that its impact will indirectly affect presentation of an event, through itsdirect effect on
the timer. If both (TB, AB) occur at a rate sufficient to interrupt the timer cycle
continually, this will suffice to maintain continuous zero. However, the rate of neither
TB or AB alone may be so sufficient.
The alternative sets may be described as follows:
A. TB alone operative b. TB, AB each operative
TB-* Continuous zero TB?? Continuous zero
[5,1] AB?? Occasional presen. AB-> Continuous zero
NB-* Occasional
presen. NB-? Occasional presen.
+ ?
With regard to assigning or status to the presentation, scanning [5.1a] suggests
that ifpTB increases, the presentation functions negatively, and ifpTB decreases, the

presentation functions positively. These are derivable from current procedures. With
regard to [5.1b], the same conclusions can be derived from p(TB, AB). The procedures,
?
of course, are defined as nondiscriminative avoidance (presentation ), and DRO
(presentation + ). However, from the effects of pTB alone (the conventional basis for
stimulus definition), inference is hazardous. Either or both may increase or decrease for a

negative presentation. They should both decrease for a positive presentation. Inference
?
about the presentation is safe only for an increase inpTB (presentation ), but prediction
of pTB is safe only when d?cr?mentai procedures are applied.

67

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

Consideration of the joint-environment-effects for [5.1a] follows:


TB absent-? Occas. presen. TB pr?sent?? Continuous zero

[5.1j] AB abs-pres.-^ Occas. presen. AB abs-pres.-^ Continuous zero


NB abs-pres.^ Occas. presen. NB abs-pres.?? Continuous zero

The implications are evident.


The expected long-term outcomes may be inferred grossly from inspection of limits
for [5.1]:

a. TB alone operative b. TB, AB each operative


? ?
Presentation + Presentation Presentation + Presentation
TB lim-> .00 TB lim-> LTC pattern TB lim -> .00 TB?
[5.2]AB? AB] .00
LTCpat.
\ lim-> 1.00 1.00-LTC (TB) AB lim-? A?]
Ilim-?
|lim-?
NB) NB] NB lim-?1.00 NB-? 1.00-(TB,
AB)
The outcomes forTB alone operative (left) accord with those classically obtained for
DRO and NDA. For (TB, AB) operative (right), when either TB or AB occur at LTC
intervals whether in patterns such as TB-TB-TB-TB, or AB-TB-TB-AB, or AB-AB
AB-AB, there will be a continuous zero state, that is, TB or AB will never be accom
panied by a presentation. If neither TB or AB occur before the timer resets, the zero state
will be interrupted by a presentation.

The scientific puzzle created by nondiscriminative avoidance and described earlier


becomes glaring ifone's observation is limited to TB. Behavior seems to be maintained
in the absence of any related stimulus change. An explanation in terms of the punishment
of NB glides over the fact that not all such behaviors are punished, and also the fact that to
the extent that there are a variety of AB which supplant NB, the absence of NB is not
associated with successful maintenance of a zero state. For DRO, appropriately timed
occurrence of TB will effectively maintain a no win situation, and occasional TB will
result in longer than usual wait for the positive delivery. To be consistent with DRO, one
might refer to nondiscriminative avoidance as a DPO arrangement. However, to the
extent that a name suggests an explanation, DPO has been considered inadequate, and
the same objections would hold for the designation DRO. On the other hand, if the
negative presentation is designated as nondiscriminative avoidance (NDA), perhaps the
positive one might be designated as nondiscriminative reinforcement (NDR). Since a
two-factor theory has been proposed for NDA (DPO), wherein immediate escape from
an internalized state of anxiety is proposed as the maintaining consequence for target
behavior, one should expect two-factor theory to account for NDR (DRO). Here, the
cessation of TB might be related to some kind of cyclical changes in elation or eagerness
(anticipated elation). Conceivably the cessation of internalized elation which initially
accompanied TB is experienced internally as the occurrence of anxiety of internal
punishment which becomes conditioned toTB, hence its approach to .00 as the schedule
takes effect. The reader is invited to exercise his imagination on the form which a
two-factor theory of DRO would take.

The relation of NDA to themore conventional negative reinforcement-escape proce

68

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

dure (Figure 4) is worth noting. In escape negative reinforcement, each TB is accom

panied by a zero state. InNDA, each TB is also accompanied by a zero state. Further, it is
the only response which has this consistent relationship. Both AB and NB will occasion
ally be accompanied by a negative condition. True, the zero condition which accom
panies TB in negative reinforcement was immediately preceded by a negative condition,
and the zero condition which accompanies TB in Sidman avoidance is not immediately
preceded by a negative state. Rather, it is preceded by the same condition which is
present when the behavior occurs.
However, successful avoidance through TB is
generally preceded by ahistory of negative presentations in the absence of LTC patterns.
Instatement of "free shocks" when TB is established might then increase pTB not
because of related increase in anxiety, but by reinstatement of a past early history when
increase in pTB eliminated shock.
In escape, the change in conditions from negative to zero has been interpreted as the
stimulus in negative reinforcement, but exactly how a zero state qualifies as a stimulus
has puzzled many observers. It may be assumed to be the functional equivalent of a

reinforcing event and may acquire this functional quality through the fact that the
alternative state of the environment is negative. In Sidman avoidance, the negative
alternative is occasional, but is historically relevant to TB. Conceivably, appropriate
occurrence of TB can facilitate occurrence of all theworkaday behaviors which are not in
the investigator's experimental purview, and which are governed by the consequences
contingent upon them. In successful avoidance, these are disrupted neither by the

periodic presentation of a negative stimulus, nor by the preparatory behaviors relevant


to them. Morse and Kelleher (1970) report that an event such as shock can be program
med to assume the functional equivalence of a positive event, and thereby maintain
a zero event might also be programed to
complex schedule performance. Conceivably,
function in thismanner ("no news is good news") through careful attention to schedule
histories and alternatives.

Continuous Presentation: Cyclic Discrete Zero Imposed

The apposition of presentation contingencies and withdrawal contingencies have


reinforcement in the classical
provided the basic definitions of positive and negative
unilinear formulations. Accordingly, itmay be of interest to place in apposition to a
a withdrawal contingency. Such contingencies are consid
periodic presentation periodic
ered in Figure 6. In parallel manner with the periodic interruption of discrete zero states
discussed earlier, the prevailing conditions constituting the continuous presentation, are
either positive or negative.

As is evident from the schematic forTB alone operative (left) TB interrupts the timer
and the timer resets for a full cycle, with the continuous presentation then
cycle,
a zero. When TB occurs at LTC the timer is
interrupted by discrete intervals, continually
reset, and the continuous presentation is not interrupted. AB, whether occurring instead
of TB, or at some other time, may be accompanied or followed by a presentation or zero,
as is NB.

69

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

TB alone effective B. TB, AB each effective

TB -
0
?
JTLJTI_TLITL
-
0

AB -
n (-^
AglTLnJTl TX
ururui

NB

-
O

Figure 6. Continuous presentation, cyclic zero periods. Alternative sets.

The joint-environment-effect discussed earlier holds here, as well.


The alternative sets may be described as follows:
a. TB alone operative b. TB, AB each operative
TB?> Continuous presentation TB-* Continuous presentation
[6.1] AB-* Occasional zero AB?> Continuous presentation
NB-* Occasional zero NB?> Occasional zero.

Scanning [5.1a] suggests that ifpTB increases, the presentation functions positively,
and ifpTB decreases, the presentation functions negatively. With regard to [6. lb], the
same conclusions can be derived, but for p(TB, AB). However, for the effects on pTB,
inference is hazardous, with the same relations noted earlier obtaining, as they do for
prediction of pTB. The long-term expected outcomes may be inferred grossly from
inspection of limits:
a. TB alone operative b. tb, ab each operative
? -
Presentation + Presentation Presentation + Presentation
TB lim -* LTC pat. TB lim -> .00
tb]
[6.2] AB] AB]
1 lim-> ltc pat.
1.00-LTC 1.00 ab)
[lim-* [lim-*
NB] (TB) NB] nb lim-* 1.00-ltc-(tb, ab).

an intuitive basis, the gross expected outcomes seem self-evident.


On
No name is available for these relationships. In the laboratory, a procedure which
suggests itself readily involves the correlation of a discriminative stimulus with the

70

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

presentations, at least initially. The key lightmight go on or off when the timer is reset,
and then be faded out to establish nondiscriminative control. Space precludes further
discussion.

Continuous Zero State: Complex Discrete Presentations (1)

In the procedures to be discussed, a timer locks upon completion of a single cycle.


Behavior at any time recycles the timer, returning the timer to its origin for one more
cycle, either while it is cycling or is locked upon completion. The continuous zero state is
interrupted by a presentation only if behavior occurs when the timer has locked upon
completion of a cycle. There is no presentation when behavior occurs during a cycle, and
such behavior adds another full timer cycle period to the time elapsed. The schematic
diagrams are depicted in Figure 7.
As is evident from the schematic forTB alone operative (left), TB interrupts the timer
cycle, and institutes a new cycle. A continuous zero state accompanies these TB-timer
resets, and also the completion of cycle. It is only when TB occurs after completion of

A. TB alone effective B. TB, AB each effective

JTLJUJTLJTl^
tb
tb
0
? 0
?
JTL-TLJTUar^
? ?
h-h
-???v?

ad a a b a a b a a b a b
ab
oZTLTLTl_TL -TUTLn_n_n?n?

? ?
0 ?
>>>>>>>>>>> > >>->>
0? n_n_ji
_???>

nb
'grurLfiiriJJi^
- ? n
?
o
?
tin o n.

Figure 7. Continuous zero state, discrete presentations during indefinitely


locked timer. Alternative sets.

71

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

the cycle that a presentation interrupts the continuous zero state, and the timer is reset to
run its cycle. As is evident, the TB which are accompanied by presentations are those
spaced at greater than cycle (GTC) intervals. AB, whether occurring instead of TB (b), or
at some other time (a), is accompanied only by a zero state, as is NB.
Analogous joint-environment effects also hold.
Description of alternative sets requires three groupings, which are related to different
rates of TB:
a. TB alone operative
i. ii. iii.
TB zero Cont. zero TB rare-* Discr. pres. TB freq.^ Cont. zero
[7.1a] AB-> Cont. zero AB?? Cont. zero AB-> Cont. zero
NB-> Cont. zero NB-^ Cont. zero NB-^ Cont. zero
IfTB is absent, continuous zero ismaintained (i). IfTB occurs more frequently than
zero, itmay be accompanied by discrete presentations (ii), providing it is not so frequent
that a continuous zero is reinstated (iii). If rate of TB increases from a zero rate, then
increases further, then drops, hovering around a low rate defined by GTC intervals, the
presentation can be considered +. This procedure is classically designated as DRL
(differential reinforcement of low rates). If rate of TB is zero, or extremely high, this
suggests that the presentation is ?. The latter procedure might be designated as DPL
(differential punishment of low rates) and itmight be of interest to investigate the
conditions under which DPL results in absence of TB, or in an extremely high rate of TB.
Description of alternative sets when (TB, AB) are both operative also requires three
groupings:
b. (TB, AB) each operative
i. ii. in.

(TB, AB) zero^ Contin. (TB, AB) rare-* Discrete (TB, AB) freq. Contin
zero presens. zero

[7.1b]
NB-* Contin. NB-* Contin. NB Contin
zero zero zero

The same relations noted for TB alone operative [7.1a] also hold for combined (TB,
AB). However, except for (TB, AB) zero (i), their rates can vary independently of each
other.

Inspection in terms of limits indicates the complexities:


a. TB alone operative b. TB, AB each operative
- -
Presentation + Presentation Presentation + Presentation
Either Either
TB lim-> .00 TBi lim^> low TB lim-* .00

(AB, NB) 1.00 1 (GTC) AB lim-* .00

ABl NB lim-* 1.00

[7.2] TB lim-? low rate


(GTC)

72

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

Or Or
(AB, NB) lim-? high TB lim-? 1.00 TB?
rate (high or Ltc Him-^ 1.00
(AB, NB) lim-? .00 A?) or
(high
(low or 1.00-LTC) AB lim high LTC)
rate NB lim-? .00, low
limits approached are 1.00 or high rate (LTC, less than cycle intervals), low rate
The
(GTC, greater than cycle intervals), and .00. Itwill be noted that forTB, when (TB, AB)
are operative, a .00 rate, or a .00 or low rate (AB instead of TB lim?? LTC), or a high rate
(TB lim-?- LTC), can ensue. However, only an incremental outcome for TB can

functionally define the presentation. These definitional properties of the schedules are
similar to the others.
The situation when (a) no behavior produces no effects and (b) a high rate also has no
effects ("overeagerness"), but (c) a low rate is effective ("playing itcool") is, of course,
not confined to the laboratory.

Analogous Continuous Presentation: Complex Discrete Zero (1)

The "withdrawal" analogous to the DRL-DPL arrangements (Figure 7), are de


scribed in Figure 8. The behavior-timer-presentations are identical, except that the

presentation is continuous, but appropriate behavior is accompanied by a zero interrup


tion of the presentation.

A. TB alone effective
B. TB, AB each effective

-
tb ? tb
o ?
o

o ITU LTU
>??h
j"-^

ab a a b a b
ab
a a b

0
-^>>> ?> >>>>>>>>>>*>>>>>>>
o UU?Lf

nb nb
o^UTJTJJTJJL^^ o-1

?
U

Figure 8. Continuous presentation, discrete zero periods during indefinitely


locked timer. Alternative sets.

73

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

The schematic diagrams forTB alone operative are presented on the left, and for (TB,
AB) operative on the right. Analogous j.e.e. hold. For TB alone operative, the alterna
tive sets, given three different rates, are described as follows:
1. TB alone operative
i. ii. iii.
TB zero-*- Contin. TB low^> Discrete TB high?> Contin.
zero
pres. pres.

[8.1a] AB^ Contin. AB-> Contin. AB^ Contin.


pres. pres. pres.
NB-* Contin. NB^ Contin. NB^ Contin.
pres. pres. pres.

IfTB is absent (i) or at a high rate (iii), the presentation is continuous, and only if it is
low (ii), is itaccompanied by a temporary interruption. IfTB hovers around this low rate,
-
the presentation can be considered ,with a temporary "restraining injunction" (neither
overdone, nor neveruseu; in effect. IfTB is either zero or at a high rate, the
presentation
can be considered + ,with a low rate
accompanied by discrete losses.
Description of alternative sets when (TB, AB) are both operative follows:
b. (TB, AB) each operative,
i. ii. iii.
TB, AB) zero-^ Contin. (TB, AB) rare-^Discrete (TB, AB) freq.^ Contin.
zero
pres. pres.

[8.1b] NB-* Contin. NB^ Contin. NB-* Contin.


pres. pres. pres.

The same relations noted for TB alone operative [8.1a] also hold for combined (TB,
AB). However, except for (TB, AB) zero (i) their rates can vary independent of each
other. Inspection in terms of limits indicates the complexities:
a. TB alone operative b. TB, AB each operative
- -
[8.2] Presentation + Presentation Presentation + Presentation
(See 7.2- ) (see 7.2a+ ) (See 7.2b- ) (See 7.2b+ ).
As noted for [7.2], only an incremental outcome can functionally define the presenta
-
tion. In the "presentation" contingency , increment may define a presentation, while in
the "withdrawal" contingency, it
may define a -I-presentation. This is also the case in
positive (Figure 3) and negative (Figure 4) reinforcement. Reinforcement provides the
only possible basis for functional definition here as well as in the other situations
discussed.

Continuous Zero State: Complex Discrete Presentations (2)

One final set of schedules will be considered, simply to contrast itwith DRL. Here,
both a timer and a response counter are involved in the procedures to be discussed. Upon
the completion of its cycle the timer automatically resets the timer and the counter.
Nothing else is then programmed. A predetermined number of responses counted during
the cycle (that is, before the slate iswiped clean at completion), resets both the timer and

74

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

the counter but, in contrast to cycle completion, also interrupts the continuous zero state
by a discrete presentation. The alternatives are presented'in Figure 9. Analogous j.e.e.
can be derived.
As is evident for the schematic for TB alone operative (left), the timer recycles and
resets the counter when less than required (LTR) TB occurs during a cycle. If the required

A. TB alone effective B. TB, AB each effective

TB
Ml I lllll,1111
II 11 1111 Mill
Hill,

AS"1111 MM IM 1Mill INI INI 11inI mi mi


-
?
?
o

NB-
0 un mn
urn
TTTTTT Ngrnn
-
?
-
0 wuui_nn
-->

5:
VUW V

Figure 9. Continuous zero state maintained by recycling timer. Alternative sets.

number occurs during a cycle, the last response is accompanied by a presentation. Since
the timer is then reset, it is obvious that the duration of the TB-controlled reset is
governed by the time inwhich the requirement was met. A short time produces a short
cycle, at which time there is a presentation, and a longer time produces a longer cycle,
with the maximum being the setting of the timer, at which point there may be no
presentation, and the counter is reset. The alternative sets require two groupings:
a. TB alone operative
h.
TB zero, low Continuous zero TB high Occasional presen,
[9.1a] AB?? Continuous zero (every /?thTB)
NB^ Continuous zero AB?? Continuous zero
NB^ Continuous zero

75

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

Inspection suggests that if rate of TB increases, and ismaintained at a high rate, the
presentation is +. This, of course, is theDRH (differential reinforcement of high rates)
schedule. If rate of TB drops, and ismaintained at a low or .00 rate, the presentation is -.
This arrangement has no name, but might be called a DPH.
Description of alternative sets when (TB, AB) are both operative follows:
b. (TB, AB) each operative
i. ii.
[9.1b] (TB, AB) zero, low-* Continuous zero (TB, AB) high-* Continuous pres.
NB?> Continuous zero NB-* Continuous zero.

or
High rate of TB can represent either high TB (and low AB), high AB (and low TB),
both. Similarly, both or either can be low when (TB, AB) low, but each must be zero if
(TB, AB) zero.

Inspection of limits suggests straightforward long-term outcomes:


a. TB alone operative b. (TB, AB) each operative
- -
Presentation + Presentation Presentation + Presentation
TB lim-* 1.00 TB lim-> .00 TB lim-? .00
TB|
[9.2]AB lim-* .00 AB lim-* .00
NB lim-* .00 1.00AB] A?]
[lim->1.00
NB lim-* 1.00
[lim-* NB lim-* .00
NB)

The DRH schedule [9.2a+] has often been bracketed with DRL as arate procedure
which is opposite in effect. It is evident that such a bracketing is oversimplified. In
contrast to DRH, whose outcomes are straightforward, the DRL outcomes [8.2+] are

complex and wavering. Further, the procedures (Figures 7, 9) are not complementary (as
they are in the reinforcement and punishment of Figure 3), but are markedly disparate.

Analogous Continuous Presentation: Complex Discrete Zero (2)

The "withdrawal" analogies to Figure 9 are presented in Figure 10. The predeter
mined number of responses interrupts a continuous presentation and substitutes a discrete
zero. Analogous j.e.e. may be described.

The schematics may be summarized by the following sets of alternatives:


a. TB alone operative b. (TB, AB) both operative
TB high-* Occasional zero (TB, AB) high-^ Occasional zero
[10.1] AB?* Continuous presen.
NB^ Continuous presen. NB-* Continuous presen.

Inspection suggests that for TB (a), if rate of TB increases, the


alone operative
-
presentation is the presentation is + .For (TB, AB) both operative
, but if itdecreases,
-
(b), either TB or AB or both can be at high rates for presentations, but rate of both
should decrease when the presentations are +.

76

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

A. TB alone effective B. TB, AB each effective


- "
TB
11111111
uh TB
1111,1
\\\\ 0
111i 11
?111111 II.Hill I III!

o o
->
5 K

AS"ini iiiiiii mum mi mi ab-mii nulli m IUI Ill I


0
"1_^l-^r?

? -
NB
0
NB
0 11
0
->

5r 5:

0e \aAA7 uwwww
0:

Figure 10. Continuous presentation maintained by recycling timer. Alternative


sets.

Inspection of limits is equally straightforward:


a. TB alone operative B. (TB, AB) both operative
- ?
Presentation + Presentation Presentation + Presentation
TB lim-? .00 TB lim-? 1.00 TB lim-? .00
tb)
[10.2]AB) AB lim-? .00 AB lim-? .00 1.00
Jlim-?
1 lim-+ 1.00 NB lim-? .00 NB lim-? 1.00 ab)
NB) nb lim-? .00

Decision Theory

One major logical extension will be noted, namely, decision theory. The simplest
two well-defined responses. A thirdNR class must
paradigm for decision theory requires
also be considered, although this class is usually implied. Although it is assumed that
-
other well-defined behaviors may exist and, indeed, are explicitly analyzed (n alterna
tives), the (TB, AB) case is considered the simplest expository case.

77

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

Decision theory procedures differ from typical concurrent behavior paradigms in


terms of one major extension. Each of the two well-defined behaviors is coupled with at
least Aro consequences, which are typically relatable to two states of the environment.
The familiar 2x2 matrix is formed, given by the intersection of states of the environment
and behaviors, inwhich case the cell-entries are the consequences. Under one state of the
environment, TB is relatable to one consequence and AB to another (or the same), and
under the alternative state the consequences for each differ. It should be stressed that
although the two classes of behavior can be considered simply as alternative concurrents,
namely, as Bi and B2, their designation as TB and AB is often more appropriate, as
judged by the different designations given their cell entries, e.g., False Alarm and Hit in
the TB case, and Correct Rejection and Miss in the AB case, in the case of Signal
Detection Theory. The reader is familiar with False and True Positives and Negatives in
other test or diagnostic cases.
Expected outcomes are estimated only from consideration of all sets of all alternative

behavior-consequential relations, often in terms of net outcome for the entire matrix.
Decision theory can thus be considered an,extension of the cases which have been
presented, in which expectation of outcome also required such inspection.

Differential Reinforcement

In differential reinforcement, of course, one pattern of behavior is reinforced, while


all patterns excluded from that definition are not. They are to "undergo extinction."
Since the excluded class is defined only by reference to the other, this is a poorly-defined
class, and such usage of differential reinforcement is not to be confused with the discrete
presentation alternatives of Figure 3. Rather, it is described by Figure 1. Stated other
wise, the procedures and derivative definition are unilinear. It seems as if the subject has
a "choice" between alternatives, but as noted, alternative sets required well-definition in
both cases. In addition to this possibly illusory ascription of choice to the procedure, it is
often considered as the humane op?rant alternative to aversi ve control, inwhich patient

understanding and reinforcement are substituted for impatient punishment. Accordingly,


it finds widespread sponsorship in texts and other discussion of applied behavior
analysis. As will be noted briefly, its sponsorship and association with applied behavior
analysis can be disastrous in a variety of ways.
The actual situation has been obscured, to a considerable extent, by a unilinear
definition. Reinforcement, the substitute for punishment, is defined unilinearly in the
present system. In the system of alternatives proposed, it is defined by thewhole set of
alternatives, rather than any single class. In the present defining laboratory situation,
patience is, indeed, often required. In the laboratory, the experimenter (or, more
properly, his equipment surrogate) literally does nothing (1) until the response require
ment is met, that is, he does nothing (2) when it is not met.
Doing nothing (1) until the response requirement ismet also describes those prison
systems whose "behavior modification" procedures include deprivation of privileges, or
assigning the prisoner to "the hole" until he "shapes up." Such practices, of course, have
existed before behavioral research was even dreamed of, as have related practices in
other total and quasi-total institutions (Goldiamond, 1974; cf. also Holland, 1974). An

78

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

applied behavior analyst who steps into such total institutions under the impression that
by substituting differential reinforcement for aversi ve control, he is tempering the
system, is deluding himself. His delusion is supported by a unilinear definition of
reinforcement as a positive contingency and the ensuing conceptual neglect of the
alternative class which necessarily enters into the definition of reinforcement. And he has
been surprised by the vehemence of the reaction. The alternative class is not neglected by
others. It is noticed by the subjects. They react to the excluded and much larger set of
behaviors which are treated to a neglect which by no stretch of the imagination can be
called benign, and to the system-imposed deprivation necessary to make the positive
consequence potent. In the mental hospital, this procedure is usually tempered by the

hospital's curative mission. However, where the administration neglects to supervise its

professionals or patients, gross abuses can also occur inmental hospitals and children's
homes. "Benign" neglect also characterizes those humanist approaches which
a
rationalize doing nothing for mental patient until he asks for therapy.
The parents or other custodians of a child whose behavior disturbs them often can not
simply do nothing until the child does the right thing. Depending on the urgency created
by the ensuing behaviors, they or the institution may apply physical or chemical
constraints to avoid such behaviors, or they may apply punishment when they occur.
Parents may withhold reinforcement ? for a time? and then yield, but the response
pattern in the meantime may have escalated. Thereby behaviors further removed from
parental requirements are reinforced. Programed (or nonprogramed) disaster can accom
pany programed (or nonprogramed) differential reinforcement or the humanist ap
proaches noted.
The practical difficulty of doing nothing when the response requirement is not met
outside the laboratory has already been noted even where the aim is scientific (e.g.,
Azrin, Holz, Ulrich, andGoldiamond, 1961). It is often insurmountable where the aim is
applied. The term, differential reinforcement, as then used, may then be only metaphori
cal.

Where the requirements of differential reinforcement have actually been fulfilled in


such applied situations, that is, have been literally rather than metaphorically applied, I
suggest that (1) whatever success is attained may often be derived from procedures
defined by considerably more than simple differential reinforcement, and that (2) these
procedures can often be related to laboratory formulations. Further, (3) which laboratory

procedures have been tapped for application is often obscured by describing the proce
dures in the language of laboratory differential reinforcement. Explanation of success in
such simple termsmay underlie the unfortunate statement by Hilgard and Bower (1966)
that "the practical use of the (op?rant) system is based on the complementary principles
of control through presenting and withholding reward. The supplementary principles of
stimulus discrimination and response differentiation suffice to inaugurate shaping
through themethod of successive approximations. Beyond that, all that is needed is the
experimenter's ingenuity" (p. 144). Indeed, training-courses and manuals for parents
and others often do not provide even as much.
The "experimenter's ingenuity" consists of applying procedures other than those
confined by the oversimplified formulations used. Itmay involve contriving situations

79

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

which will increase the likelihood of occurrence of the response required. Itmay involve
the selection of response-requirements considerably removed from the outcomes de
sired, and which are farmore probable than such targets. Itmay involve the development
of curricula in applied areas (Goldiamond and Dyrud, 1967). The issue often is the
program developed, and there is all the difference in theworld between telling someone
to reinforce differentially in a step-by-step manner, and the actual development of a
program, e.g., as by Sidman and Sidman (1967). The impression created in the former
case is that anyone can do it.Experience, training, and supervision are not required. The
picture of false simplicity created, as reflected in the distorted image of Hilgard and
Bower, may be related to accurate reflection of much practice, as reported, rather than to
distortions in theirmirror.
No pretense is made that adoption of a framework of alternative sets provides
conceptual solutions to problems whose resolution must be experimental. The
framework is one inwhich certain issues can be seen more clearly. It was noted that,
when applied problems are examined within this framework, the understanding of
successful solution requires more than is implied by the present oversimplified formula
tions. Among the things required is the consideration of a curriculum, that is, a program
(as used in programed instruction). A program may be considered as an active and
manipulative establishment of a sequence of events to produce an explicitly stipulated
outcome. A history may be considered a retrospective attempt to reconstruct a sequence
of events which have culminated in a particular outcome. An appropriate history, may,
accordingly, be said to be the resurrection of the program involved. Morse and Kelleher
(1970) report research inwhich a variety of different op?rant responses by monkeys was
maintained by scheduled-presentation of shock which was not relatable to any consequ
ences intuitively regarded as positive. Stated otherwise, shock by itself served to
maintain behavior. They ascribe their results to the schedules and to the history of their
establishment: "It should be understood that a scheduled performance represents what
has happened to a subject and as part of that subject's experience will determine its

subsequent behavior" (p. 182). Stated otherwise, the program, or history, is critical, and
it should simply be added that history should be formulated for laboratory animals, as it
must for people outside, in terms of the alternatives, past and present, involved.

Discussion

Presentations have been kept to their simplest possible forms. Only two well-defined
sets of behavior have been considered; in actuality, many more sets are often found. The
- -
alternative to + or has always been 0, rather than such combinations as 4- and or
4- + and +. TB and AB have either been related to the same consequences or to 0 forAB.
Schedules considered have been variants of CRF or a few inwhich a fixed timer-period is
involved. The joint-environment-effects have not been pursued fully. And so on.
Nevertheless, even wtih these limitations, some interesting patterns emerge. Among
these is the happy decision, made some time ago, to define the stimulus functionally only
by incremental effects on TB, rather than also by d?cr?mentai outcomes. This holds even
when the definitional system is strained by extending the defining effects toAB, as well.
Accordingly, itwould appear that the terms positive and negative reinforcement are

80

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

useful laboratory terms. The nondifferentiated use of the itxmpunishment also appears to
be a useful usage, since differentiation (negative and positive, Type I and Type II, among
others) serves no purpose in functional definition. Some savings may ensue from
winnowing terms from the two reinforcements and punishment, but their nicety in
functional definition may thereby be blurred. The cost of restructuring might also be
considered. Stated otherwise, a decision formulation is appropriate forwhat is, after all,
a decision. Of course, there are costs attached to any restructuring, but the question is
raised, are they justified by the benefits? In a different context, Michael (1974) has
elegantly argued that the choice between using statistical tests of significance and using
experimental analysis be examined in terms of decision outcomes. Favoring retention of
the terms noted is the fact that the typical laboratory situation, which can be described in
terms of TB idling, is sufficiently close to the situation described by a unilinear notational
system that terms derived from that system may still serve parismoniously. As long as
this laboratory situation is dominant, either in practice or conceptually, revision of the
three terms which derive from them should be considered with deliberation.
However, this does not necessarily hold for the status of terms such as positive and
negative reinforcers, aversive stimuli, and the like. Substitution of positive (+) and

negative (?) serves present laboratory purposes and also brings them more into line
with notations and descriptions in other disciplines (e.g., positive and negative sanc
tions, gains and losses), including those formally applying decision theory. And be
havior analysis can make a singular contribution by defining these terms functionally,
instead of the presents priori basis. However, their functional definition would derive not
from their effects on TB alone, but by considering the alternatives involved. The Law of
Effect, as was noted, refers to TB alone in a unilinear system. Definition through its
alternatives as well removes TB from itspresent central position. Itmust be considered as
one among equals. However, it is not thereby necessarily removed from privileged
attention. Conditions abound in which we focus on specific patterns for increment,
decrement, or other purposes. Even then, we must consider as equals the AB our
procedures may also affect in order to understand or control TB. In such a paradigm, the
laboratory condition when experimental control is such that AB is not affected, is a

limiting case which might be best encompassed within the same framework, namely, AB
idling, as has been done here. We have, so to speak, deprived TB of its exclusive control
of the reins, but not its title; it is no longer the exclusive conceptual dispenser.
It is also suggested that terms related to "withdrawal" contingencies might also be
reconsidered, together with that term What
itself. the stimulus is in such a case is always
a problem. Instead, we might consider arrangements whereby the investigator sets up a
continuous presentation, positive or negative, and then, for some period of time does not
do so (discrete zero). These can be compared, as has been done, to those arrangements in
which, for contingency contrast, he does nothing for a continuous zero, and then
term
occasionally presents a positive or negative event. It will be noted that the
presentation is used consistently throughout all contingencies. The relations of this
redefinition to application have been noted in the body of the discussion.
If the two reinforcement terms and the punishment term have a certain elegance in the
laboratory as long as the present one-defined-response laboratory system is regnant (and

81

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

will therefore probably remain with us for some time) can one make predictions, or at
least recommendations concerning their staying power if the one-response-system be
comes no longer standard? Certainly, such a restatement is by this discussion,
suggested
and implied by research such as concurrent schedules, choice and matching, theHerrn
stein and Hineline experiments noted, and the differences between tent gradients and
peak-shifts when analogous reformulation was introduced into discrimination research.
The discussion will be centered on the conditioning laboratory itself, since the terms
noted derive from its requirements. How well does thismodel account formore recent
data and formulations?
The use of the laboratory term reinforcement in application, has produced at least as
much confusion, ifnot more, than itsmodifiers, positive and negative. Reinforcement
has prior association with Pavlovian conditioning and, in the popular press, at least, has
led to confounding with op?rant procedures. Behavior analysts, whether experimental or
applied, have always had to take pains to explain the differences between op?rant and
respondent which, after all these years, are still not evident. This is true not simply for
popular audiences, but also for more sophisticated members of our literate culture.
Indeed, Kling (1971) notes the existence of confusion even among "many biological
scientists that these are merely two slightly different forms of the same procedure and that
they probably involve the same neuronal process" (page 566). He suggests that clarity is
facilitated when "we reserve the term 'conditioning' for themethods of Pavlov, and say
'instrumental' or'op?rant' training or learning when referring to the latter procedures"
{idem., emphases in original). If there is one termwe might dispense with readily, it is
conditioning. Its relationship to op?rant procedures has been implicitly tenuous for some
time. Indeed, Holland and Skinner ( 1961 ) open their still classic program (derived from a
section of Skinner's 1953 text) with respondent procedures. These procedures are
thereafter almost completely ignored. One gets the impression gained from a scientific
treatise of a few hundred years ago which opens with statements of fealty to theCrown
and the true precepts of the true Church, and having done so, forthwith presents an
independent coherent account; or from a Soviet textbook on experimental psychology
which opened with the indication that the contents to follow were in accord with and
inspired by Marxist-Lenninist doctrine, followed by a straightforward textwhich could
have easily been accommodated within the American range of offerings. Reconsidera
tion of conditioning might simply make explicit what is currently implicit. It
might
relieve introductory authors of the necessity for their Pavlovian introductions. If the
relation of op?rant procedures to conditioning is to be questioned, we might also
reconsider derivatives such as conditioned (conditional) reinforcers and unconditioned
stimuli.
Itmay be worthwhile to consider the very term reinforcement. Michael would
accept
this term and also punishment, and use the two and their derivatives (reinforcers,
punishers) as substitutes for negative-positive reinforcements, negative-positive rein
forcers, punishment, aversives, and the like.We run into some problems even with this
reformulation when we examine the situations for TB, AB each effective. Itwas noted
that under appropriate conditions, pTB can be decreased in the benign condition when a
consequence functionally defined as positive is also attached to AB. The resultant

82

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

attenuation of TB fits the definition of punishment he approves, namely, the attenuation


of behavior by consequential means. In this case, the consequences are attached to a class
which includes both TB and AB. The face of a mother would make an interesting study as
it is explained to her thatwhen she effectively uses this commendable procedure she is
actually being punitive, or punishing TB (since it is going down)! Indeed, popular
lectures aside, the term punishment is being stretched. Itmoves toward meaninglessness
as we depart from a unilinear formulation of consequential relations.
Most properly, when we consider the rate changes from which the terms at issue are
derived, what we are discussing, in science or application, is increment or decrement and
subsequent maintenance at some stated level. We can, using consequential procedures
and a well-defined set of alternatives, produce or study such effects by attachingpos/f/v^,
negative, orzer? consequences defined functionally by consideration of the relations of
procedures to effects for all operative sets of alternatives. Where all the operative
alternatives had not been spelled out (even when we thought we had), we might infer
such alternatives from conflict between expected and obtained outcomes, and then test
our inferences.

If the field is to move toward


simplification and rationalization of terms, it is
recommended that we move in these directions. The following specific recommenda
tions are made, in order of the assumed ease of their adoption.
1. That the environmental conditions
provide which
the present consequential
settings be described in terms such as continuous
zero state, continuous presentation
(among others), which are interrupted by discrete presentations or discrete zero states,
cyclic or aperiodic (among others). These describe more closely what goes on than do
current presentation-withdrawal-extinction formulations.
2. That procedures be primarily defined by how the presentation-zero arrangements
are scheduled, rather than by whether the presentation is + or ?. Such definition would,
for example, consider DRO and NDA as the same arrangement. Their present stimulus
bound classification (reinforcing, aversive stimuli) has placed them inwidely disparate
categories (reinforcement, avoidance).
These two recommendations are independent of alternative sets. Although the next
recommendation is stated in terms of alternatives, it need not be. It is compatible with
present formulations and terminology.
3. That positive, negative, and zero substitute for present stimulus terms in a set of
well-defined alternatives, the interplay of whose procedures and outcomes provides
functional definition.
The following recommendation, of course, has been made throughout the discus
sion.

4. That alternative sets replace present unilinear formulations.

Conceivably, for some time, unilinear and alternative-sets formulations might


coexist, depending on the type of research, the explanatory adequacy of the formula
tions, and the comparative conceptual clarity they provide.
Finally, and not without after-thoughts, the following three recommendations are
made.

5. That increment, decrement, and maintenance substitute for reinforcement,

83

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

punishment, and various maintenances (of high rates, attenuated rates, zero or extinc
tion).
6. That along with (or independently of) the preceding recommendation, nouns such
as analysis and behavior (and the like, depending on context) replace the term condition

ing as the substantive which op?rant modifies. Wherever possible, we might shed
derivative conditioning terms such as unconditioned stimulus, conditioned reinforcer.
7. That we begin reconsidering terms related to behavior analysis, modification, and
similar compounds.
Although we do analyze behavior at all times, what we analyze is much more
extensive. And behavior modification is simply a misnomer. Sometimes in this area,

changes in behavior are programed. However, the behavioral change in establishment of


reading (eye movements) is trivial; what is initially changed is the stimulus control the
environment exerts over unchanged verbal behavior, so that textual stimuli now govern
the same words that verbal stimuli governed previously. Thereby, behavior is exposed to
thewritten repertoire of the human race. Changes in consequences can be programed ?
and the changes can be dramatic. Changes in schedule can be programed, and so on.
What is involved in experimental or applied "behavior analysis" is a fine-grain analysis
of "the an organism and its environment"
interaction between (Skinner, 1969, p. 7).
This includes(1) the occasion, (2) the response, (3) the consequences, and (4) the
interrelations among them, that is, the "contingencies" (p. 7). Skinner (1974) lists
twenty misconceptions about behaviorism (pp. 4-5). Many of these derive from an
accurate reading of "methodological behaviorism." Skinner must then dissociate "radi
cal behaviorism" from it.What empirically supports his approach derives mainly from
differences in the subject matter analyzed and relations inferred (contingencies rather
than stimulus-response relations) and the experimental procedures used (the experimen
tal analysis, as opposed to the statistical analysis, of behavioral experiments).7
Much of this paper is about schedules. They are not simply an interesting field of

specialization which happens to characterize this area. Schedules define the precise way
inwhich consequences are related to behavior, for the environmental settings. (The point
has repeatedly been made elsewhere, e.g., Ferster and Skinner, 1957; Skinner, 1969).
Given identical availability of identical consequences, identical occasions, identity in
anything else necessary to define environments, and genetically identical organisms with
identical responses, different patterns of behavior will emerge if the precise ways in
which the consequences are related to behavior differ. Further, given the same sets of
schedules, whether behavior is sensitive to the new schedule or insensitive, that is, a
behavioral requirement for a previous one persists, will be a function of the serial order of
these schedules. These relations tend to be glossed over by statistical studies. They may
be subsumed under interaction variance, but this does not suffice. To these requirements
for analysis, the present discussion suggests the consideration of alternative contingen
cies. Behavior can increase or decrease for the same schedule, depending on the

7Skinner opens the book with the following definition of behaviorism:kk Behaviorism is not the science of
human behavior; it is the philosophy ofthat science." (p. 3). A behaviorist would then be one who expounds
that philosophy or uses it to guide his practice, and not necessarily one who uses certain experimental

procedures or draws inferences from them.

84

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Alternative Sets as Framework for Behavioral Formulations and Research

alternatives. Where possible schedule differences and alternatives have not been consid
ered, it is difficult to justify inferences of hereditary differences from behavioral
differences, when consanguinity variance and environmental variance represent 'the
major independent variables, or to justify a variety of other inferences being made.
It would be poor programing to recommend that these reformulations replace the
present ones m toto and immediately. Good programing suggests thatwe proceed slowly
and make those changes that our own requirements call for, examine the extent towhich
we gain or lose thereby, and so on. The basic procedural heritage we have been

bequeathed has been a goodly one, and can be used as long as is necessary and be referred
to as the original referent framework.
The limited presentation made has generally abided by the recommendations pro
posed, and the reader is invited to assess the effectiveness with which they describe
procedures, relations, and outcomes. The currently prevailing formulations are encom
passed by the framework proposed. Prevailing definitions and predictions are not
violated. Within the same proposed framework, other predictions, paradigms, and
schedules flow systematically, and are neither derived nor expected from the currently
prevailing system. Whether the apparent implications of the reformulation for the
laboratory are fulfilled requires experimental validation. However, their implications for
application are glaring. That the applied implications have been overlooked by experi
mental and applied behavior analysts singularly illustrates the powerful hold of an
intellectual formulation which, within itsown limited area, is elegant. The illustration is
all themore interesting in view of the fact that these applied implications have been noted
not only by those vehemently opposed to the system, but also by neutrals and friends?
and widely publicized. The intellectual formulation referred to is, of course, the present
unilinear framework. Critics have attributed its limitations and implications to experi
mental and applied behavior analysis. Unfortunately, behavior analysts have contributed
to this confusion by couching their presentations in unilinear terms. We can do ? and
deserve ? It that some of the critics were but for the
certainly better. may be said right,
wrong reasons and against the wrong target.

REFERENCES
Anger, D. The role of temporal discriminations in the reinforcement of Sidman avoidance behavior. Journal

of the Experimental Analysis of Behavior, 1963, 6, 477-506.


Azrin, N.H. & Holz, W.C. Punishment. InW.K. Honig (Ed.), Op?rant behavior :Areas of research and
application. New York: Appleton-Century-Crofts: 1966.
Azrin, N.H., Holz, W.C., Ulrich, R., & Goldiamond, I. The control of the content of conversation through
reinforcement. Journal of the Experimental Analysis of Behavior, 1961, 4, 25-30.
Baum, W. On two types of deviation from the matching law: Bias and undermatching. Journal of the
Experimental Analysis of Behavior, 1974,22, 231-242.
Evans, R.I. Skinner: The man and his ideas. New York: E.P. Dutton,
B E. 1968.
Ferster, C.B. &Skinner, B.F. Schedules of reinforcement. New York: Appleton-Century-Crofts, 1957.
Goldiamond, I. Toward a constructional approach to social problems: Ethical and constitutional issues raised

by applied behavior analysis. Behaviorism, 1974,2 (No. 1), 1-84.


Goldiamond, I. Si Dyrud, J.E. Some applications and implications of behavioral analysis for psychotherapy.
In: J. Schlien (Ed). Research in psychotherapy, Volume III. Washington, D.C.: American Psychologi
cal Association, 1967, pp. 54-89.

85

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions
Israel Goldiamond

Herrnstein, R.J. Behavioral consequences of the removal of a discriminative stimulus associated with
variable-interval reinforcement. Unpublished doctoral dissertation, Harvard University, 1955.
Herrnstein, R.J. Relative and absolute strength of response as a function of frequency of reinforcement.
Journal of the Experimental Analysis of Behavior, 1961, 4, 267-272.
Hilgard, E.R. & Bower, G.H. Theories of learning (3d Ed). New York: Appleton-Century-Crofts, 1966.
Hineline, P.N. Negative reinforcement without shock reduction. Journal of the Experimental Analysis of
Behavior, 1970, 14, 259-268.
Hineline, P.N. & Herrnstein, R.J. Timing in free-operant and discrete-trial avoidance. Journal of the
Experimental Analysis of Behavior, 1970, 13, 1 13-126.
Holland, J.G. Behavior modification for prisoners, patients, and other people as a prescription for the

planned society. Paper delivered at annual meeting of the Eastern Psychological Association, Philadel
phia, April, 1974.
Holland, J.G. & Skinner, B.F. The
of behavior. New York: McGraw-Hill,
analysis 1961.
Keller, F.S. & Schoenfeld, W.N. of psychology.
Principles New York: Appleton-Century-Crofts, 1950.

Kling, J.W. Learning: Introductory survey. In J.W. Kling & L.A. Riggs (Eds.), Woodworth and Schlos

berg's Experimental Psychology (3d Ed). New York: Holt, Rineheart and Winston, 1971.
Lewin, K. A dynamic theory of personality: Selected papers. New York: McGraw-Hill, 1935.
Michael, J. Statistical inference for individual organism research: Mixed blessing or curse? Journal of
Applied Behavior Analysis, 1974, 7, 647-653.
Morse, W.H. & Kelleher, R.T. Schedules as fundamental determinants of behavior. InW.N. Schoenfeld.
The theory of reinforcement schedules. New York: Appleton-Century-Crofts, 1970, pp. 139-185.
Mowrer, O. On the dual nature of learning: A reinterpretation of "conditioning" and "problem-solving."
Harvard Educational Review, 1947, 17, 102-148.
Schoenfeld, W.N. "Avoidance" in behavior theory. Journal of the Experimental Analysis of Behavior,
1969, 12, 669-674.
Schoenfeld, W.N. & Farmer, J. Reinforcement schedules and the "behavior stream.11 InW.N. Schoenfeld

(Ed), The
theory of reinforcement schedules. New York: Appleton-Century-Crofts, 1970.
Sidman, M. Reduction of shock frequency as reinforcement for avoidance behavior. Journal of the

Experimental Analysis of Behavior, 1962, 5, 247-257.


Sidman, R. & Sidman, M. Neuroanatomy, a programmed text. Boston: Little, Brown and Co., 1967.
Skinner, B.F. Science and human behavior. New York: Macmillan, 1953.
Skinner, B.F. Contingencies of reinforcement: A theoretical analysis. New York: Appleton-Century
Crofts, 1969.
Skinner, B.F. About behaviorism. New York: Knopf, 1974.

86

This content downloaded from 91.229.229.96 on Sun, 22 Jun 2014 00:53:22 AM


All use subject to JSTOR Terms and Conditions

You might also like