ABSTRACT
VERBAL CONDITIONING BY AVOIDANCE
LEARNING: AN EXPERIMENTAL ANADOGUE
TO A PARTICULAR INTERPFRSONAL SITUATION
by James PgfiMathie

Many researchers have recently attempted to apply the
principles which have evolved in animal and human learning
studies to the broad fields of personality, interpersonal
interactions, psychotherapy and behavioral disorders. The
present study attempted to apply some of the principles
of avoidance learning and verbal conditioning to the inter”
personal situation where one person verbally attacks another
person. It was posited that the replies associated with
”anxiety reduction" would mimic a person's predominant or
characteristic response to such a situation, eXpecially
under stress or stimulus change. An artificial interper’
sonal situation was used to increase eXperimental control.

Eighty male undergraduate students enrolled in ele-
mentary psychology courses served as subjects. They were
subjected individually to a three phase eXperiment. For
all three phases they were seated at a teaching machine
and were presented with the eXperimental stimuli through
a window in the machine. The first phase was an evaluation
phase, the second an acquisition or learning phase and the
third another evaluation phase.

During the two evaluation phases, the stimuli consisted
of twenty verbal attack statements presented one at a time.

The attack statements were printed in a cartoon-like balloon

James P. Mathis

accompanying the line drawing of a human face. A second
line drawing of a face was accompanied by a blank balloon
representing the person responding to the attack. The fife
composed and wrote replies to each of the twenty attack
statements.

After the first evaluation phase, each of the responses
was classified into one of five classes of response each
representing a different way in which the responder ex*

pressed or failed to express anger in his reply. The
class of response which occurred most often during this
phase was labelled. thw modal response and was excluded
from the acquisition phase.

During the acquisition phase the four remaining classes
of response (three of the four on each presentation) were
printed in the balloons accompanying the face, representing
the responder. Each response was followed by a type of
reinforcement. For each g two of these classes of response
were rewarded by being correct, one was punished, and one
led to the avoidance of punishment. Half of the g's were
aware that giving one of the reaponses would lead to the
avoidance of punishment (anxiety reduction or eXperimental
group) the other half were not aware of this (no clear
reinforcement or control group).

The third phase was identical to the first evaluation
phase except half the gfs were stressed during the evalua-
tion, half were not, half had stimulus material similar to

acquisition and half had new stimulus material (four groups).

James P. Mathis

It was assumed that the effects of the reinforcements
would be approximately equal on g's with different modal
responses and approximately equal on the five classes of
response. Both these assumptions were found to be unten-
able. Individual differences in g's led to differential
responses to reinforcement, and the different classes of
response were differentially affected. Even so, responses
reinforced by being correct in general increased, and
those followed by punishment decreased.

As eXpected stress and novel stimuli as Opposed to
lack of stress and similar stimulus material led to an
increased use of the modal response. The responses rein-
forced by avoiding the punishment without the fife awareness
led to results which mimicked those for the modal response.
fife who were aware they could avoid.punishment used the
modal response more than the unaware gfe particularly under
stress and with novel stimuli. The results were interpreted
in terms of learning two different kinds of avoidance
responses depending on the gfs level of awareness that

avoidance had occurred.

VERBAL CONDITIONING BY AVOIDANCE
LEARNING; AN EXPERIMENTAL ANALOGUE
TO A PARTICULAR INTERPERSONAL SITUATION

BY

("(II

)-

QT
James p. Mathis

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PhILOSOPhY

Department of Psychology

1965

To

Barbara

11

Acknowledgments

I wish to acknowledge my indebtedness to the Psycho-
logy Department at Western Michigan University who pro-
vided facilities as well as subjects for the research. I
also wish to thank Kalamazoo College and.Michigan State
University for their cooperation in gathering the prelimi“
nary data.

I am also indebted to Dr. Stewart Armitage and the
other members of the Psychology Department at the Veteran's
Administration Hospital at Battle Creek, Michigan for their
support during the course of the research.

In addition, I wish to express my gratitude to the
members of my committee: Dre. Joseph Reyher, R. E. Schell,
A. A. Seagull, and particularly to the chairman of my com'
mittee Dr. M. Ray Denny.

iii

Dedication
Acknowledgments
Introduction
Method

Results
Discussion
Summary
References

Appendix

Table of Contents

iv

. 11

p.111

p.
p.
p.
D.
p.
p.
p.

1
11
25
62
73
77
79

List of Tables

Table 1. Examples of Five Kinds of Replies Repre’ p. 9
senting the Five Classes of Response

Table 2. Pattern for the Association of Class of p.18
Response with Type of Reinforcement for
the Twenty Acquisition Programs

Table 3. Mean Frequency for each Class of Response p.29
chosen as the First Response during First
Ten and Last Ten Trials of Acquisition

Table A. Summary of Analysis of Variance for Class p.32
of Response X Stress X Generalization X
Membership in Experimental or Control
Group

Table 5. Mean Change Pre to Post for Classes of p.33
Response X Type of Reinforcement for
EXperimentsl and Control Groups

Table 6. Percent of Times each Class of Response p.3h
was associated with each Type of Rein-
forcement across all EXperimental Con'
ditions and EXperimental and Control
Groups

Table 7. Percent of Times each Class of Response p.35
was associated with each Reinforcement
for Experimental and Control Groups and
for Stress and Nonstress S's

Table 8. Mean Frequency with which S's with s p.h0
Given Modal Response gave Phnished, Safe
or Correct Responses for First Ten Trials
and Last Ten Trials of Acquisition

Table 9. Mean Change Pre to Post for Punished, p.’-L2
Safe and Correct Responses according to
Model Responses of Sis

Table 10. F Test and Newman'Keuls Procedure for p.h5
Mean Change Pre to Post for Responses
associated with Three Types of Rein‘
forcement

Table 11. Mean Pre to Post Change in Frequency of p.50
Modal Responses X Membership in Experi‘
mentsl and Control Groups, Stress“

Nonstress Conditions and Generalization-
Nongenerslization Conditions

V

Table 12.

Table 13.

Table 1h.

Table 15.

Mean Pre to Post Changes for Punished,
Safe and Correct Responses X Post Test
Conditions for EXperimental and Control
Groups separately and combined

Mean Pre to Post Changes for Punished.
Safe and Correct Responses X Member'
ship, in EXperimental and Control Group
for Stress, Nonstress, Generalization
and Nongeneralization combined and
separately

Mean Pre to Post Changes in Frequency
of Responses representing the Five
Classes of Response X Posttesting Con-
ditions and Membership in EIperimental
and Control Groups

Mean Pre to Post Changes in Frequency
of Punished. Safe and Correct Responses
X Stress vs. Nonstress and X EXperi-
mental vs. Control for S's with either
Self’attack or Denial as their Modal
Response

vi

1053

p.5h

p.57

p.61

Figure 1.

Figure 2.

Figure 3.
Figures u.

List of Figures

Schematic Presentation of Method p.
and Procedure

Saple Item on Pre and Post Test p.
Programs

Sample Item for Acquisition Program p.

5. 6, 7 and 8. Acquisition Curves for p.

Frequency of Punished, Safe and Correct
Responses across Sixty Trials (blocks
of Ten Trials) broken down according to
Class of Response being Reinforced;
Control Group

Figures 9. 10, ll. 12, and 1}. Acquisition Curves p.

Figures 14,

Figures 18,

Figure 22.

for Frequency of Punishment, Safe and
Correct Responses across Sixty Trials
(blocks of Ten Trials) broken down
according to Class of Response being
reinforced; Experimental Group

15, 16 and 1?. Acquisition Curves for p.
Frequency of Punished, Safe and Correct
Responses across Sixty Trials (blocks

of Ten Trials) broken down according to
st Modal Response; Control 8'3.

19, 20 and 21. Acquisition Curves for p.
Frequency of Punished, Safe and Correct
Responses across Sixty Trials (blocks

of Ten Trials) broken down according to
st Modal Response; EXperimental Group

12

13

17
27

28

37

38

Acquisition Curves for Twenty‘eight p.h7

Matched S‘s across Sixty Trials
(blocks 6: Ten Trials) for Experi'
mental and Control Groups

vii

Appendix
Appendix

Appendix 3

Appendix

List of Appendices
List of Eighty Attack Statements

Six Faces Used on Acquisition
Programs.

A Note on Training the Judges
Raw Data for Pre to Post Changes

X Posttesting Conditions X EXperi'
mental and Control Groups

viii

p.
p.

Introduction

Many researchers have recently attempted to apply
the principles which have evolved in animal and human
learning studies to the broad fields of personality,
interpersonal interactions, psychotherapy and behavioral
disorders (Shoben, l9h9; Mowerer, 1950, 1961; Bollard and
Miller, 1950; wolpe, 1952, 195%, 1958; Eysenck, 1960.)
The present study attempts to apply some of the princi-
ples and findings of avoidance learning and verbal con’
ditioning to an interpersonal situation which appears to
resemble, at least superficially, the classical avoidance
learning paradigm. The interpersonal situation under
study is that in which.one person verbally attacks a
second person and the second person replies (or fails to
reply) to the attack.

A Review of Related Research

In an elaboration of a study done by Diven (1936),
Lacey and Smith (195%) and Lacey, Smith and Green (1955)
demonstrated that an emotional response (in terms of a
heart response measure) could be conditioned to a pre‘
viously neutral word. They showed further, as Diven had,
that this conditioned.emotional response generalized to
words that were semantically related to the original sti'
mulus word; e.g. to synonyms and homonyms. This group of
studies along with a great many other related studies
(see Osgood 1953 pp.701 to 712; Kimble 1961, pp. 35h’355;
Franks 1961, pp.h78 to RSI; Jones 1961, pp.h88 to £91 for

1

2
reviews of the related studies) have shown not only that
emotional reactions can be conditioned to verbal stimuli
and subsequently by generalization to related words but
also that the verbal behavior itself can be modified.

Of particular interest for the present study is the
investigation of Eriksen and Kuethe (1956) who found
that if they asked Sis to free associate to a stimulus
word and then followed some of their associations with
a shock, the S's tended to avoid these words on subse-
quent word association tasks, even when the threat of
shock was not present. This was true for S's whether
they were aware of the contingency for shock or not.
Eriksen and Kuethe considered this to be an analogue to
repression and supported their argument on the basis of
differences in the reaction times of the 'aware' (aware
of the relationship between the giving of a particular
associate and the receiving of shock) and nunaware'g‘s.
The "aware! st showed increased reaction times in for-
ming associations to the critical stimulus words and
reported deliberately withholding the association that
would lead to shock. The “unaware'I st, who also avoided
the associations which lead to shock, showed no increase
in reaction times and reported no eXperience of with"
holding associations. Even the ‘ aware! subjects grad'
ually decreased their reaction times to the critical sti‘
mulus words and reported having lost the feeling of with-

holding associations. The authors discussed their

3
findings in terms of avoidance conditioning but did not

investigate the avoidance response itself, i.e. the words
substituted for the words not spoken. The characteris-
tic paradigm for avoidance learning is the presentation
of a as (light or buzzer usually but in the case of Erik-
sen and Kuethe, the stimulus word) followed after a delay
by a UCS (shock) and under the condition that a particu‘
lar response occurring during the delay period will keep
the U08 from occurring (for Eriksen and Kuethe, any asso’
ciate other than those chosen to be shocked). The re-
sponse occurring during the delay period between the CS
and the UCS which prevents the occurrence of the UCS is
called the avoidance response. The various studies on
avoidance learning especially those following from 8010‘
mon and Wynne (1953) and Solomon, Kamin and Wynne (1953)
on traumatic avoidance learning would suggest that the
substituted words might become conditioned to the stimu-
lus words because they would be associated with the non-
occurrence of shock (this is traditionally stated; asso-
ciated with anxiety reduction).

The studies cited on the conditioning of emotional
reactions to words have confined themselves primarily
to rather limited and quite specific types of verbal
responses e.g. word association tasks. Another body of
research on the conditioning of verbal behavior, that
growing out of the work by Greenspoon (1951, l95h, 1955)

has gone beyond the conditioning of specific types of

a

verbal responses e.g. the occurrence of plural nouns in
an interview setting (Greenspoon, 1951) particularly in
the work of Hildum and Brown (1956): Nuthman (1957) and
Salzinger and.Pisoni (1960).

Hildum and Brown (1956) found that in telephone
interviews, the number of statements favorable or unfavor‘
able to the topic of general education could be increased
by differentially reinforceing these statements with the
single word *good'. This result was obtained when the
subjects were not able to verbalize the contingency for
reinforcement. In their study, the subjects who had an
increase in unfavorable statements toward general educa-
tion still rated the experimenter as having a favorable
attitude toward the topic.

Nuthman (1957) showed that the number of statements
reflecting "acceptance of self" on a questionnaire could
be increased when such statements were followed by the
word 'good' spoken by the eXperimenter. This increase
occurred even though the words in the various statements
differed and even though the §_could not verbalize the
relationship between the experimenter's behavior and his
own.

Salzinger and Pisoni (1960) showed that the number
of statements reflecting affect could be significantly
increased in an interview setting if the experimenter
followed these statements on the part of the E by words

and gestures indicating approval.

5

In this last group of studies the emphasis was on
the modification of the verbal behavior with less empha-
sis on the conditioning of emotional responses. These
studies do, however, point toward the general finding
that verbal behavior can be modified by conditioning
techniques even when general classes of verbal responses
are reinforced rather than specific words or specific
types of words. It is suggested therefore that it may
be possible to apply some of the principles from the
first group of studies on the conditioning of emotional
responses to words, to the conditioning of general
classes of response. This study is such an attempt to
modify general classes of verbal responses within the
paradigm of avoidance learning mentioned above.

The verbal behavior to be dealt with was taken from
a real life interpersonal situation where the paradigm
for avoidance learning seems to exist. This was done
with the expectation that further research in less well
controlled but more natural settings would proceed more
meaningfully if some of the important variables in this
interpersonal situation could be identified in the better
controlled laboratory setting. Hopefully this would
eventually lead to a better understanding of the real
life interpersonal situation.

The real life situation is that in which one person
verbally attacks another person. In the actual setting,

the second person's reply (or lack of reply) is often

6
followed by some form of a renewed verbal attack by the
first person, or a spanking perhaps, if the second person
is a child, or any number of other happenings which could
be considered unpleasant or punishing. With some fre-
quency, however, the reply of the second person will
avert (or at least not be followed by) the occurrence of
such punishment. This set of circumstances fits the
avoidance learning paradigm and thus the replies not
followed by punishment would be presumably associated
with “anxiety reduction' and conditions favorable to
avoidance learning would be fulfilled. l'Anxiety reduc-
tion" is used here as an Operational term. This usage
is in accordance with the avoidance learning literature
(Solomon and Wynne etc.) and does not refer to a feeling
state of the person as used in the clinical and person“
ality literature.

Animal studies suggest that responses associated
with '.anxiety reductionl tend to extinguish very slowly
and can be reinstated very rapidly when the original
stimulus situation is re‘presented. If the interpersonal
situation referred to above fits the avoidance learning
paradigm, then it seems possible that the response a.per'
son gives to a verbal attack represents a response
learned as a consequence of having been associated with
“anxiety reduction!I in the past. Since peOple tend to
give a variety of responses to verbal attacks, the focus

in this study was on the kind of response S gave most

7
frequently in the eXperimental setting (this will be

referred to as the S's modal response.). The present
study then investigates the possibility of modifying the
frequency of occurrence of some other (nonmodal) kind of
response by associating it with "anxiety reductionI in
the experimental setting. The kind of response associated
with “anxiety reduction. in the experimental setting
could then be compared with the kind of response 2 gave
most frequently when entering the experimental setting.
Tc argue by analogy for the likelihood that the modal
response was also acquired at some earlier time because
it had been associated with "anxiety reduction“, three
reinforcement conditions in addition to .anxiety reduc'
tion! were used. The effects of these four types of
reinforcements (“anxiety reduction. plus three others) on
five different kinds of responses were evaluated under
stress and nonstress conditions and in situations either
different from or similar to the reinforcement situations
(a test for transfer). The effects of the types of reinr
forcemsnts on the kinds of responses could then be com-
pared with the effects on the S's modal response. The
argument by analogy is that if the responses associated
with "anxiety reduction! in the experimental setting and
the S's modal response were similarly affected by stress
and the test for transfer while the responses associated
with the other reinforcements were differently affected,

then the argument that the modal response was originally

learned to "anxiety reduction” would be strengthened.
The four reinforcement conditions were; 1) 'anxiety
reduction“, 2) punishment, 3) reward, in terms of cor'
rectness, and u) no clear reinforcement. They were cho’
sen because they seem to represent four major reinforce‘
ment conditions favorable to changes in behavior (or
lack of change in the case of the no clear reinforcement
type). These four reinforcement conditions will be
referred to as the 'types of reinforcement" from here on.
Table 1 gives actual examples of the five kinds of
verbal responses (replies of the person attacked), these
being; 1) replies reflecting a clear direct verbal attack
on the person making the original attack statement; 2)
replies reflecting verbal attacks on the person himself
and/or replies reflecting open or tacit agreement with
the attack statement; 3) replies reflecting an attack
but where the object of the attack is either unclear or
definitely not the original attacker; also those replies
reflecting an attack but where the cource of the attack
is unclear or where the attack nature of the reply is
only thinly disguised; h) replies which do not seem to
reflect an attack by anyone on anyone else or where the
attack nature of the reply is very unclear; 5) replies
which are specific denials of some part or all of the
attack statement, including replies which are positive
statements that are directly Opposite to the original

attack statement. These five "classes of response“ will

be referred to respectively as 1.) Anger, 2.) Self'attack,
3.) Object’displacement, h.) Isolation and 5.) Denial.

Table 1. Examples of Five Kinds of Replies
Representing the Five Classes of Responses

Attack Statement: You have a very annoying laugh.
Possible Replies:
Anger: You certainly are a rude person.

Isolation: Laughter is the medicine of the
soul.

Self‘attackg I can't change it.

Object'displacement; With me it's only my
laugh that's annoying.

Denial: It doesn't bother me.

The five classes of verbal replies, as the reader
has probably already noticed, were drawn from the liters?
ture on defense mechanisms, specifically from Miller and
Swanson (1960, pp.19h to 221) and were adapted to the
present type of verbal material.

This classification system was chosen for two main
reasons. First, the research reported in Miller and
Swanson (1960) seems to indicate that this type of cate‘
gorization system would handle the classification of a
large proportion of the verbal responses that might be
given in a verbal attack setting. Thus it would be pos‘
sible to construct replies for the reinforcement situa'
tion where each category could be represented often but
where the repetition of the exact wording of the replies

would be minimized. This allows for the learning of

10

general classes of response rather than specific words.
Secondly, the same research has shown that these cate'
gories represent psychologically meaningful response
dimensions; i.e. they have been shown in several research
settings to be related to other aspects of an individual's
behavior or life circumstances. Whether or not these
classes of responses relate to underlying personality
mechanisms is considered irrelevant to this particular
study.

A test for transfer and the effects of stress were
included in the comparisons between the S's modal response
and the responses associated with the experimental rein'
forcements. The test for transfer was used to assess the
effects of the reinforcements on the S's generalized
behavior. A complete failure to transfer would tend to
indicate that any learning that took place was situation
bound. It was expected that the response associated with
some of the reinforcements might generalize while those
associated with others might not.

The effect of stress on the responses was included
because considerable research (Child, 195u) has shown
that stress, defined in many ways, often leads to the
disruption of behavior patterns and the production of
new responses as well as to the reuse of old responses.
Thus the effects of stress on the modal response and on
the responses associated with the experimental reinforce'

ments could be compared.

The Method

The experiment consisted of three phases; pretesting,
acquisition and posttesting. Figure 1. presents the
design of the experiment schematically. The pretesting
phase was an evaluation phase where the frequencies with
which a g gave each of the five classes of response (to
20 attack statements) were determined. The most fre‘
quently given class of response was designated as the
modal response and excluded from the acquisition phase.
During the acquisition phase the four classes of responses
which had occurred less frequently in the pretesting
phase were associated with the eXperimental reinforce“
ments, Ianxiety reduction, punishment, reward in terms
of being correct and no clear reinforcement. The post‘
testing phase was an evaluation phase as was the pre-
testing phase. Again the frequencies of the five classes
of response (the modal response could again be given)
were determined. For some st the posttesting phase was
completed under stress (stress condition) while for the
others this was not the case (nonstress condition). A
test for transfer was also included in the posttesting
lphase. Some subjects had stimulus material similar to
the acquisition phase (nongeneralization) while others
had stimulus material less like the acquisition phase
(generalization).

The analysis of the results could then focus, pre

to post, on the change in frequency of the modal response.

11

12

.mocmlmac paoaw mo>oe magma gma.o an

nemcppav ochmeppn coo meagre ago: I mznuv pcmsdoaowcaea mamao oz.
Annapmmfiana ionsov snasoco coapfimasoom :soz. poms acsm. m
some phosoecfim mooaw ccwm ppmfi mo pooaaoo ”flunSom.H
Ammoapmcocl mchmmp mpcmaooaOMCAma m.m Hospcoo
npuoc acaasm anon on Ificoapmufi can modem
ufiaaecomv anemone ceapfitfiocoa anm spasm ”AmunovecoapoSQos >psaxc<=. m soSmcm co
pcsaueeap macaw can; whoa "mzc caom commacsm m mamas muons
- Ammdppuv pomambo.ppm3mm. H Umd pmmoQEoo
mcapucppmoo madame cmoLu “soap mpcaadomoeCHsm m.m ancmEHastm m “emcoommm
nouaanaocoacocv Fnauoao cedpfinasoon I Ammoam compo so one mace
so an usam macaw was; puma . mmz Ca op amp m Hespfi>accfi cmv “mmzomc m xcman
Ammsmumcocv auapneppaoc mad .COPpDQ poasppea meaccoomosaoo panama cooaamn
-aoe anon on . seasons eoapaeaoooa ozone mason omen useeomo aaoon on was seen unmam
co ma weds meomh ccmh pmea .mzwz nccormesaoo asppmfi epoaz newcoamsm
.asow we ccc mflco CH cL :do w Hemp .mCOOHHmQ
Ifi>apca “mcoapaccoo unappmom amok .20: was em: .:¢: consumed Eon» scam whoa CH
.smOmc assume no names CH pmoap uoafioma Spas mcooaanp m wpcmsmpmpn
opoas can cmmoosoo m "emcoammm an peficmmsooom «one use; psmHm xomppm om
.pcmEmpapu xomppm Spas cooaamn mamcam
.xcwan cooHHmD an meacaqsooom some pawn whoa mcooaamn
econ phMam .cooaaen one; peoH m .aeoee m
c« Eesmosc uwmpomo mm cacao came moomw m ”Eon“ roam ”Emu“ gown
Ca mpcasepmpa xoeppm om seam wsmpa ow "Emawoam zoom
.mcooaanfi m .mooam a “rape nomm _l Emswomm
Emamoma H mpmm m zomm co mampa om
msopfi om ”msmsmomn m maesmosd om

m.m.0w Ham
sou Emamoam H

 

mwompm coapmnaHmLcC¢m
.I.ImbnhwsprIpbapmNHHMmehb

  

.(\\\I, ososmquAMMbol.l.l.l

 
 

 

  

weanaoepaon meaemeponm
mwmppm coaumuaflmpmuwwcoad

mmeapmcea coapmuaanmiccmcoml

  
  

coapawflsvo<
_

waspmoosm rca pompaz mo coapmpcmwmsm oapmsmzom .H mammaa

13

and the responses associated with the various reinforce’
ment conditions. The effects of the conditions of post’
testing (stress‘nonstress and generalization'nongeneral-
ization) could also be evaluated. Further, the rein-
forcement condition could be evaluated by reference to
the acquisition curves of the training session.

The Stimuli: In all three phases, the stimuli were

 

reproduced on fanfolded paper and presented by means of
sn.MTA 100 Scholar teaching machine. The stimuli were
line drawings of two faces accompanied by cartoonvlike
balloons. (see Figure 2.) The balloon for the face on
the left always contained a verbal attack statement (to
be described later). During pretesting and posttesting,
the balloon for the face on the right was blank. During
acquisition, three balloons accompanied the face on the
right. Each of these balloons contained a potential reply
to the verbal attack statement in the balloon on the left.
(The nature of these replies will be described below).

Figure 2. SMple Item on Pre and Post Test Programs.

 
    
 

  

:2 srfierc
‘3“5’3p ever be I; g
:1, \ ‘Z'x.

During the acquisition phase and only during the acquisi‘
tion phase, the verbal attack statements were also pre“

sented by means of a tape recorder. The auditory

14

presentation coincided with the appearance of the visual
presentation in the window of the teaching machine.
The Verbal Attack Type Statements (VATS); These

 

statements consisted of 80 one sentence statements which
were composed in such a fashion that each could be con-
sidered somewhat unjustified in almost any context in
which they were said, e.g. “You can't ever be trusted".
The statements were general enough to apply to almost
anyone and were always critical in nature.

The Responses: All responses were intended to repre‘

 

sent replies to the verbal attack statements. During the
pre and posttesting phases S's responses consisted of the
replies which he composed himself and wrote on the blank
answer paper presented in the far right window of the
machine. During the acquisition phase, st responses
consisted of his reading aloud one of the three replies
typed in the right hand balloons, writing a letter on the
answer paper to record which of the three possible replies
he had chosen (they were lettered ”A“, “B”, and 'C”) and
pressing the appropriate lettered button on the left hand
side of the machine. The three buttons were labeled "A“,
“B”, and ”C" respectively and each would advance the pro*
grsm paper a different distance. On each item, one button
advanced the program paper less than an inch, one advanr
ced the paper until the stimulus was almost out of the
window, (approximately three inches) and the third
advanced the program to the next item. The lettered

15

button which would move the program a particular distance
varied randomly from item to item. During the pre and
posttesting phases, the st pressed any one of the three
buttons to move the program to the next item.

The Alternative Replies: The alternative replies

 

were chosen from the replies given by an independent
group of 118 college students from Michigan State Univer'
sity and Kalamazoo College (males and frmales). Each of
these students was given a booklet with 18 of the origi’
nal ninety verbal attack type statements. They were
instructed to Iwrite in two different ways that you might
reply to the statement depending on who said it to you
and the mood you were in when it was said to you". From
the replies obtained in this manner the E selected five
replies for each of the ninety verbal attack statements
which seemed to best represent the five classes of
response (Anger, Isolation, Selfbattack, Object’displace'
ment and Denial). The replies were also chosen so that
there would be a minimum of wording overlap from item to
item within a single class of response. Two judges then
attempted independently to classify the five responses to
each attack statement into one of the five classes of
responses. The judges agreed on 92% of the #50 replies
classified. The ten verbal attack statements on which
the judges showed the greatest disagreement in classifying
the five replies were eliminated to arrive at the list of
eighty attack statements with five replies to each

16

statement used in the study itself. The final list of
eighty attack statements and the five alternative replies
to each are listed in Appendix 1.

The Presentation of the Stimuli: Twenty-three prO'

 

grams were compiled to present the stimuli in accordence
with the design of the study. These consisted of one
pretesting program, two posttesting programs and twenty
acquisition programs.

The Pretest and.Posttest Programs: These three pro‘

 

grams each consisted of twenty items. Each item was com“
posed of two line drawings of human faces, each with an
accompanying balloon (see Figure 2.). The balloon accom-
panying the left hand face contained one of the eighty
verbal attack statements. The balloon accompanying the
right hand face was blank. The items were arranged on
the program paper in such a way that they would be viewed
one at a time by the S. The same twenty verbal attack
statements, in the same order were used on all three of
these programs.

The Acquisition Programs: Each of these twenty pro-

 

grams consisted of sixty items. Each item was composed
of two line drawings of human faces. The left hand face
was accompanied by a single balloon containing one of the
remaining sixty verbal attack statements (20 of 80 were
used on pre and posttesting programs). The right hand
face was accompanied by three balloons each of which con-

tained an alternative reply to the attack statement

17

(these replies were described above). The three alter-
native replies were lettered “A“, 'B‘, and 'C' (see
Figure 3). The items on the acquisition programs were

presented one at a time.

Figure 3. Sample Item for Acquisition Program.

 

    
     

You do the
stupidest
things.

   

All twenty acquisition programs had the same sixty
verbal attack statements arranged in the same order. The
programs differed from each other on the basis of the
content (class of response) of the alternative replies
and the order in which the alternative replies appeared
on a particular item. Table 2 gives a schematic presenr
tation of the differences between the programs.

Four of the twenty programs eliminated the Anger
class of response during acquisition. These programs
are listed under the Anger column in Table 2. Within
these four programs the remaining four classes of _
responses were each associated with reward (being correct)
twice and with both punishment and "anxiety reduction!
once (for half of the subjects, those in the control
group, one of their replies was associated with no clear
reinforcement rather than uanxiety reduction'). The four

remaining sets of programs (Isolation, Self‘attack,

H sarcoma

awooww m zrwoomv

E:

\N

z>wooww

O—v
Cr-

18

Table 2. Pattern for the Association of

Class of Response with Type of Reinforce“

ment for the Twenty Acquisition Programs
Class of Response Absent From Program

\

Anger 'Isclation Self' Object‘ Denial
attack displacement

O'Correct O‘Correct O'Correct S’Correct S'Correct
I'Correct A’Correct AfCorrect A’Correct A'Correct
S‘Safe S'Safe I‘Safe I'Safe O'Safe

D‘Punished D‘Punished D’Punished DTPunished D’Punished

S’Correct S‘Correct I-Correct I'Correct I'Correct
D’Correct D’Correct D’Correct ’Correct O'Correct
I’Safe A'Safe AfSafe A'Safe A‘Safe
O'Punished O‘Punished O’Punished S'Punished S’Punished

S’Correct S‘Correct I'Correct I'Correct I'Correct
O‘Correct O'Correct O‘Correct S'Correct S'Correct
D’Safe D’Safe D'Safe D'Safe O'Safe‘

I‘Punished A'Punished AfPunished S‘Punished A'Punished

I’Correct A'Correct A'Correct A‘Correct A‘Correct
D'Correct D‘Correct D'Correct D“Correct O'Correct
O'Safe 0'Safe O'Safe S'Safe S‘Safe
S'Punished S'Punished I'Punished I'Punished I'Punished

l9

Object'displacement and Denial) followed the same pattern
as that for Anger and each was missing the class of
response used to label it (e.g. Isolation contained no
Isolation type replies).

Any one S received only one of the acquisition pro-
grams. As the result of the above scheme S could have
been given an acquisition program that had any one of
the five classes of responses eliminated and which had
any one of the four remaining classes of responses asso‘
ciated with any one of the reinforcements.

The two correct responses each occurred on half the
items (one to an item). The punished and safe ('safe' is
used to refer to “anxiety reduction” and/or no clear
reinforcement) responses occurred on all the items.
Since a correct item had to be chosen before the S could
move to the next item, the correct responses each had to
occur thirty times in the course of acquisition. On a
chance basis the punished and safe responses also would
occur thirty times during acquisition.

The Faces: The six line drawings. all front face

 

portraits of men, were drawn with the intention of making
them markedly different from each other in appearance
(Appendix 2. contains reproductions of the six faces.).
One and the same face was present on the right hand side
of all programs. This face represented the person giving
the reply in all cases. The five faces which appeared

on the left hand side and which represented the person

20

making the attack statements appeared according to the
following scheme. One face appeared on all twenty items
of the single pretest program and only there. Two other
faces appeared on all twenty acquisition programs and on
the nongeneralization posttest program (the program on
which stimuli were to be similar to acquisition). The
last two faces appeared on the generalization posttest
program (the program on which the stimuli were to be
different than during acquisition). On programs where
more than one face appeared on the left hand side, the
faces appeared in random order.

Two different taped voices presented the verbal
attack statements during acquisition. For any particular
S, however, the same voice accompanied the same face
throughout the sixty trials (i.e. on thirty trials).
Furthermore, the correct reply on each of these thirty
trials was always the same class of response. The varia‘
tions in the programs allowed for the association of one
voice with one face and one class of response on one
program and with the other face and another class of
response on another program.

The Reinforcements: During acquisition a reply

 

given by 2 could be followed by reward (in terms of cor-
rectness), punishment, nanxiety reduction“, or no clear
reinforcement. Reward consisted of the program paper
moving immediately to the next item. Punishment con’

sisted of the program paper moving to a point just short

21

of the next item with the item responded to still in view.
The distance the paper moved for a punished item was pur'
posely similar in length to the correct item. In addition,
a very loud horn noise was presented by means of a tape
recorder at the instant that the paper stopped moving. The
horn was located slightly behind and to the left of the S.
The sound lasted four seconds. “Anxiety reduction'I (Experi-
mental st only) consisted of the paper moving a very short
distance and a green bar appearing in the window. The _S_'s
were instructed that the green bar meant the horn was
turned off and would no longer sound on that item. The

"no clear reinforcement'I type of reinforcement (Control

st only) was identical to the "anxiety reductionn rein‘
forcement in that the paper moved a short distance and
punishment could no longer occur. However, no green bar
appeared and no instructions were given to indicate that
the horn could no longer sound on that item.

The Conditions of Posttesting: The stress condition

 

of posttesting consisted of the presentation of the loud
raucous sound while S responded to the posttest program.

The sound lasted four seconds and Has presented at irregu-
lar intervals. The length of the intervals between the
onset of the horn were in order of occurrence, lo, 15, 20,
30, 10, 90, and 10 seconds respectively. The nonstress con-
dition of posttesting consisted simply of the absence of
the horn while the S responded to the posttest program.

The generalization condition of posttesting consisted

22

of the presence of two new line drawings of faces on the
left hand side of the posttest program. In the nongener‘
alization condition, the same faces which had appeared
on the left hand side of the acquisition program appeared
also on the posttest program.

Subjects: The §'s were 80 male students enrolled
in the elementary psychology courses at Western Michigan
University. All S's volunteered to participate in the
research by signing sheets passed out in their classes.

Procedure

At the beginning of the pretesting phase, the g was
seated at the teaching machine and was given the follow-
ing instructions.

”When I turn on the machine, the drawings

of two faces will come into view in this window

(E pointed to large center window in the ma-

chine). One of the drawings will be accom'

panied by a statement, the other face will have

an empty balloon next to it. Your task is to

write on the answer paper a reply that might

be given by the second person. After you have

written in your reply you push any one of these

buttons and the machine will move you on to the

next set of drawings. Write in a reply on each
of the items.

When the g conpleted the twenty items he was asked
to move back from the machine and the pretest program
and accompanying answer paper were removed from the
machine. The E then classified each of the E's replies
into one of the five classes of responses, giving half
scores when two different classes of response were repre’

sented in a single reply. The class of response occurring

23

most often was then designated as the modal response and
an acquisition program with that class of reaponse elimi-
nated was placed in the machine.

Each g was then assigned either to the eXperimental
or control group in an alternating order. The instruC'
tions given to the experimental group were as follows
(The parts in quotes in parentheses represent the modi-
fications of the instructions as given to the control
group). The first item was already in the window when
the instructions were given.

As you can see, this time three possible
replies to the first person's statements are
already typed in the balloons next to the
drawing of the second person. In a moment
the statement typed in next to the first per‘
son will be presented to you by this tape
recorder. You are to listen to the state“
ment and then choose one of the replies “A".
'B', or "C" next to the second face. Indi-
cate your choice by writing the letter ”A".
"B". or “C" on the answer paper. Then you
must read the words of the reply you have
chosen aloud and push the lettered button
that corresponds to the choice you have
made. On each problem, only one of the
replies will move you on to the next prO’
blem immediately. One of the other replies
will ('sometimes') cause this loud horn to
sound (the horn was sounded briefly).

Another one of the replies will cause a

green bar to appear in this window ("will
move the machine ahead a short distance“).
This green bar tells you that the loud horn
is turned off until you get to the next
problem (this last statement was not given

to the control group). You must continue to
choose replies by writing in the letter corres’
ponding to the reply you have chosen, reading
the words of your choice aloud. and pushing
the lettered button which also corresponds to
the reply you have made until you choose the
reply that moves you on to the next problem.
For example; if after you hear the voice on

24

the tape recorder, you choose reply ”A” you
would write the letter ”A" on the answer paper,
then read the words of reply “A" aloud and then
push the ”A" button. This choice could move
you on to the next problem or it could cause
the loud horn to sound or it could make the
green bar appear in the window ("make the
machine move ahead a short distance") which
would tell you that the horn could not sound
until you get the next problem (this was not
said to the control group). If this reply did
not move you on to the next problem you would
then have to choose one of the other two repliea
write in that letter on the answer paper, read
the words of that reply aloud and push the but'
ton corresponding with that second choice etc..
There will be two different faces on the left
hand side and two different voices on the tape
recorder. The same voice will always accompany
the same face.

When the sixty items of the acquisition program were
completed, one of the two posttest programs was put in
the machine and the g was given the same instructions as
he had been given for the pretest program. For the E's
in the stress groups the following statement was added to
the instructions. "From time to time while you are work-
ing on these problems the horn will sound“.

After the 2 finished the twenty items of the posttest
program a short interview was held in which E was asked
questions about the eXperiment. The essentials of the
questions asked were as follows.

1. What was the eXperiment all about? Did you see
any pattern to what moved the machine on to the next item?
What were you trying to do in the middle part (acquisition)?

2. “The last twenty questions were just like the
first twenty questions. Did you answer them in any special

way?

25

3. There were two faces and two voices in the mid’
dle part. What if anything did this mean to you?

h. What was your reaction to the horn? What was
your reaction when the machine moved ahead a little bit
(control group Efs only)? What was your reaction to the
green bar?

Results

The study, as originally formulated, was mainly con-
cerned with the effect of various types of reinforcements
on the probability of occurrence of various classes of
response during both an acquisition phase and a post‘
treatment test. In post‘treatment testing, gfs could be
either stressed or non’stressed and presented with either
stimuli similar to training or different from training
(generalization test). In the present design, the fol‘
lowing assumptions were involved in testing for these
effects:

1) The five response classes are affected in much the

same was by all treatments (e.g. all classes will show a
reduction in response frequency when punished, if any one
of them shows this effect).

2) Reinforcement effects will apply to all subjects alike,
independent of the initial response tendency (modal
response class) that gfs have at the beginning of the
eXperiment. (e.g. If punishment reduces response fre-
quency for E's with one type of prepotent response (modal)

response class) it will reduce reSponse frequency for

26

all E' s) .

It was found that these assumptions held only in a
limited way and that the data had to be analyzed and
interpreted accordingly. Thus the data pertaining to
these assumptions are presented first so that the main
analysis can proceed meaningfully. Let us first examine
assumption #1 regarding reinforcement for the five dif‘
ferent classes of response.

The relevant data regarding acquisition for correct
responses as well as safe and punished responses are
presented in Figures A through 13 and summarized in
Table 3. Here frequency of responding is the mean num-
ber of times a response is used as the first response in
each block of ten trials (each correct response can occur
only five times in ten trials while safe and punished
responses can occur a maximum of 10 times). During the
sixty acquisition trials, frequency of responding increar
sed for all classes of response when they were reinforced
by being correct. The amount of increase differed how-
ever for the classes of response. Table 3 shows that the
mean increase for denial responses exceeds that for any
of the other classes of response when the eXperimental and
control groups are considered separately or combined.
This can be seen graphically by comparing the curves for
the correct responses in Figures 8 and 13 with Figures
4 through 7 and 9 through 12. The difference is most

evident between denial and self‘attack where the mean

27

Figures h,5.6,7 and 8. Acquisition Curves for Frequency
of Puni Safe and Correct see across Sixty

lIl

 

28

s for Frequency

8
V
P
u
C
n
O
.1
t
1
S
1
U
0..
C
A
3
1..
d
n
8
2
1
l
.l.
O
1
9
8
e
r
U
8
1
F

Sixt
CQ.d n

8
8
O
r
C
R
8
9
S
n
O
D
8
6

Safe and Correct R

of Punished.

T
3’

f

a.

n
w

 

29

d.H
m.m
:.m
:H

o.H
m.m
o.m
ma

m.a
m.m
m.m
on

m.-
H.m
~.m

o.o
~.m
~.m

a.-
~.N
H.m
ma

Hmacoa

0.7 m. m4- m4- a. m.m- mm- a. m.. 0.7 m. a. m... :83
a.a m.m m.m a.a o.m o. w.a H.m..a.m m.m m.m m.a :.a om-om
~.m a.m m.m o.m m.m m.m 3.: :.m m.m m.m m.m m.a m.a oa-a
a om m oa ma : m am a a ma ma as a

am- e. he. .24.. a. m... a..- o;
m.H m.m ~.H m.a m.m m. m. m.m m.a w.m H.H ~.H omuom
m.» ~.m :.m o.m m.a m. m.a m.m H.m H.m H.m ~.H oaua
m ma Ha m Ha w m NH w Ha mm a m a

M.I N. O.NI 0.0 .MMHQ

.:rm
Nmr-l

C‘J

firm] w. moHl moHl m. Nod.- .JoH- m. m. moi we N.l m...- .MMHQ
m.a m.m m.a m.H m.m m. H.m m.m w.m o.m o.m :.H m.a omlom
Tm m4... m.m 9m as a; 9m an mm mm mm as. a; 2L.
ma on ma ma mm NH Ha mm ma ma mm ma as c
a o m m o a a o m a o a m
pceEmomHQme
upoeqno momppmnuaom cospeHOmH. nomc<

coapdmascod no mamas» cue and; paw cue
. pagan madame omcodmo ma
on» ma somozo omcoommm no mmmao Some sou mosescoam can: .M wanwm

% H

30

increase for the denial responses is three times that of
the self'attack response (combined eXperimental and con“
trol groups). Both classes of responses were occurring
with almost equal frequency after ten trials (mean fre-
quency after 10 trials for Self'attack 2.1; for Denial
2.2). For the last ten trials of acquisition, the
Denial responses occurred a mean of 3.8 times (out of a
possible five times) while the Self‘attack responses
occurred a mean of 2.5 times.

The differential effect of the reinforcements on the
five classes of response is even more evident for punish-
ment and safety, particularly when the experimental and
control groups are considered separately. Figures h
through 1} illustrate that the direction of change as
well as the amount of change differed among the classes
of responses. For the experimental group, the safe
response drOpped in frequency if it was an Anger type
response, (mean decrease 2.0), increased if it was an
Isolation response (mean increase 1.2), and stayed about
the same if it was a denial response (mean change 0.0).
For the punished responses in the experimental group;
Anger, Isolation and Self-attack responses showed little
change over the sixty trials (mean changes respectively
“0.0, '.3, and '.5) while Object'displacement and Denial
decreased (mean changes 'l.h and ‘2.h respectively).

For the control group, Anger and Isolation responses

changed little over the sixty trials (mean changes -1.u

31

and ‘2.M respectively).

For the control group, Anger and Isolation responses
changed little over the sixty trials (mean change '1.1
for Anger and ‘.2 for Isolation). Self'attack on the
other hand decreased a mean of 2.8 responses and Object‘
displacement decreased 1.3 responses per ten trials.

The significance of the differences in the various
comparisons above is hard to assess because the groups
are small and the n's vary. Furthermore, the data do
not reflect differences due to particular combinations
of the classes of response (e.g. when Anger is correct,
does it increase more if Isolation was the accompanying
punished response as Opposed to Self'attack etc.?). The
comparisons for the data on acquisition do suggest however
that assumption #1 cannot be accepted without qualifi-
cation.

An analysis of the pre to post treatment changes
likewise challenge assumptionfl..A2x2x2x5 analysis of
variance was used to compare the changes pre to post for
the five classes of responses. The eXperimental and
control groups and the four post testing conditions were
included in the assessment of the possible differential
effects the experimental conditions had on the five
classes of response. This analysis is presented in
Table fl. Table 5 presents the mean changes in frequencies
pre to post for the five classes of response broken down

by type of reinforcement.

32

Table h. Summary of Analysis of Variance for
Class of Response X Stress X Generalization X
Membership in EXperimental or Control Group

Source 83 df MS F
Classes of Response 1571.66 3 523.89 12.63,,
Generalization X M5.71 3 15.2% 1

Class of Response

Stress X Class of 396.00 3 132.00 3.1%'
Response

Group (see) x Class 382.22 3 127.u1 3.07'

of Response

Class of Reaponse X
Subjects within Groups 8957.10 216 u1.u7

* - (LOB
are . (.01

The analysis of variance suggests that across both groups
and all posttesting conditions, the five classes of
responses were not similarly affected by the experimental
conditions (main effect of response class is significant
at the .01 level). Anger and Denial responses showed an
increase across both groups and all conditions while
Isolation, Self'attack, and Object'displacement responses
showed little change. Table 5.suggests that these dif-
ferences vary with the various reinforcements.

One possible eXplanation for these differences is
that Anger and Denial were differentially associated with
correct reinforcement during acquisition. Table 6. pre-
sents the percentage with which each class of response

was associated with each type of reinforcement during

33

oa.m ca. oa.a-ma.a om.m-om.m- aw.m sw. we. om.m oa.a om. Om.m om.- o.m

o m m o m m o m a o m m o m m
0H m ma m a

0 m m o m m o m m o m A o m m

cm a ma m m
nauseowflmmap
amazon upooqno sospps.easm soapsaomH somc<

eczema Hospcoo use Hmpcmsapodxw you pcosooaoucaom no ache x
uncommom no nemmwao you upon 0p cam owcmno can: .m vanes

3n

acquisition. although the percentages are not identical
across class of response, they are nearly so; and above
all, the deviations do not correspond to the direction of
the differences in the means in the analysis of variance.
The differences between the means are therefore not a
reflection of the various responses having been unequally
associated with the various reinforcements during acqui'
sition.

Table 6. Percent of Times Each Class of Response

was Associated with each Type of Reinforcement

Across all Experimental Conditions and EXperimen-

tal and Control Groups.

Anger Isolation Self‘attack Object" Denial

displacement
2 2b. 27 2h 25 25
G M9 50 50 50 50
s 27 23 26 25 25

Membership in the experimental and control group and
the stress conditions during posttesting also influenced
the differences found between the classes of responses
(Groups X Class of response is significant at the .05
level; Stress X class of response is also significant at
the .05 level). The meaning of these significant inter-
actions is not entirely clear because the different
classes of responses were associated with the different
reinforcements disprOportionatly. Table 7. presents the
percentage with which each class of response was associa’
ted with each of the reinforcements during acquisition for

the eXperimental and control groups for the stress conditions

35

mm mm Ha m: mm mm mm as an am as mm mm mm mm
a: ma mm mm mm am me an NH ms mm mm mm as om
a: 0m mm mm Hm mm mm ma am om om om mm mm an

mm om am as .mm mm as mm am mm mm mm mm om as

o m m o m m o m m o a m o m m
pcmsoomadmap
Hanson poofiao mosses.eaom eoapwaomH somc<

.m .m mmeapwcoz use mmoapm ham mucosa Hocficoo cam Hmpcosdsomfim no.“ pcoEmoaowcacm
some Spas pcpwfioomw< mm: mwcoomcm no mwwHo Some moaaa no pcvoaem .N oHQma

36

Ideally each class of response should be associated with
punishment and safety 25% of the time and with correct
50% of the time. The Object'displacement response is the
only response that comes close to this ideal in either group.

Even though the meaning of the above significant
interactions (Group X Class of response and Stress X
Class of response)are unclear, the weight of the evidence
during acquisition and pre to posttesting suggest very
strongly that assumption #1 is untenable.

The second assumption was that reinforcement effects
would apply to all S's alike, independent of the initial
response rendency (modal response class). Figures 1h
through 21 show the learning curves for the eXperimental
and control groups in terms of the S's modal response
(st giving Anger responses as their most frequent response
during pretesting make up the group used for the learning
curves under Anger,etc.). No curves have been drawn for
the Object’displacement group since only 3 st gave
this response as their most frequent response during
pretesting.

E's in the control group with Isolation or Self‘
attack as their modal response (Figures 15 and 16) show
a greater consistency in the relative frequency of the
punished and safe responses across the sixty trials. The
Anger and Denial S's in the control group (Figures 1h and
17 in particular show a greater tendency for punished

and safe responses to be very close together in frequency

37

Figures 1h,15,l6 and 1?. Acquisition Curves for Frequency
of Puni Safe and Correct ses across Sixty

 

36

Figures 18, 19, 20 and 21. Acquisition Curves for Frequency
f Punished, Safe and Correct Responses Across Sixty Trials

 

39

at one point in acquisition while very far apart at
another point.

In the eXperimental group, the curves for the punished
and safe responses also tend to be smoother in the Isola'
tion and Self‘attack groups than in the Anger and Denial
groups (Figures 19 and 20 vs. Figures 18 and 21). Here,
however, the Self'attack st (Figure 20) can be distin‘
guished from the other three groups because they show a
consistent difference in the frequency of the punished
and safe reaponses across the sixty trials. The safe
response is a minimum of .9 responses above the punished
reSponse at all times.

Even the correct responses which show a rise in all
groups show some differences across modal response groups.
In the experimental group, the S's who gave Anger as their
modal response (Figure 18) were still not giving the cor‘
rect response as their most frequent response after thirty
trials. In all other groups, by ten trials, the correct
response was the most frequent response. {S‘s giving
Denial as their modal response also evidence some dif'
ference in the frequency with which they gave correct
responses across the sixty trials. In the control group
in particular (Figure 17) the Denial S's showed a tendency
to decrease the frequency with which they gave correct
responses after the first thirty trials of acquisition.

Table 8. presents the mean frequency for punished,

safe and correct responses for the first ten and last ten

no

:4 H. mél H.m m... :él aim Nialméu min o.m| 3. ..CHQ

mmsoaw
m.m o.m s.a m.m ~.H a.a :.w a. m.a w.m o.a m.m omuom
m.m m.m m.m w.: a.m w.m o.m m.a :.m m.: o.m m.m oa-a
escapaoo
oa as m a e
m.H n.. m.. nos m.H H.H: ~.u w.H ~.u H.Hu o.m m.u m.au.eeaa
m.e w. m.a a.m m.m ~.H a.m H.m m.a o.m :.m m.a omuom
o.m m.a m.m mam m.m ~.m s m m.: w.m m.m 0.: m.m s.m oaua
oa ma m m :
oops»
mg m.. 0.7 o.m w... a4- om ... 3- w; 0.7 w... .33
m.m m.a w.a sacs a.m H.m m.a m.m m.a m.H m.m w.a m.m omuom
w.a N.m w.m :.: m.m m.m m.: :.m ~.m a.: m.m H.m oaua
o m a o m m o m m o m m o m a
ON 0 an as m :
pcoeoomaamap
amazon Iuooqno xomppmlwaom coapmaomH named

.:0apawa5co¢ mo mamaae use pmeq use mHmaaB cmh panda sou momCOQmmm poeaaoo so when
.pmnmacsm seam emcoemmm Heep: coaaw a new: m.m soap: and: menopause use: .w manna

{13°80

Ml

trials of acquisition, grouped according to the S's modal
response. Data are presented for the experimental and
control groups separately and combined. Again the data
on st giving Object’displacement as their modal response
are excluded because there were only three such subjects.

The greatest drop in punished responses from the
first ten to the last ten trials of learning was found
in the experimental group for st with nger as their
modal response. "Anger"l S's in the control group, on the
other hand, were the only S's who showed an increase in
punished responses from the first to the last ten trials.
The other three modal response groups (Isolation, Self'
attack and Denial) showed a greater drOp in freouency in
the control group. For the safe responses, the two Anger
groups again stand out. The Anger control S's showed the
greatest drOp in frequency of the safe responses among
the control groups, while the Anger eXperimental Sis
showed the least drop in frequency among the eXperimental
groups.

The above data on acquisition seem to suggest very
strongly that it can not be assumed that Sis with dif'
fering modal responses react similarly‘tathe reinforce"
ments. Assumption #2 is therefore very much in question.

The changes from pre to post testing further suggest
that assumption #2 is questionable. Table 9. presents
the mean changes from pre to post testing for the 3's

grouped according to class of modal response and type of

M2
reinforcement.

Table 9. Mean Change Pre to Post for Punished, Safe and
Correct Responses according to Nodal Responses of S.

Modal Response Group

Anger Isolation Self'attack Object“ Denial

displacement

n 9 1n 3h 20

P 2.89 ‘l.79 “.53 '1.65
5 S '.78 “.28 .uu ”1.15
C c 3.83 6.60 3.h7 3.12
n 5 9 15 10
P .140 -2.89 '2.07 -1.90
E s -1.00 -1.22 -.07 '3.00
c n.90 8.33 n.23 n.15

n 4 5 19 10
P 6.00 .20 .68 -1.u0

C s -.50 1.u0 .8k .70
C 2.50 3.50 2.87 2.10

For the combined experimental and control groups,
the frequency for punished responses in the Anger group
increased a mean of 2.89 responses from pre to post
testing. The frequencies for the other three groups
decreased (Isolation 'l.79; Self‘attack ".53; Denial
“1.65). Likewise when the experimental and control groups
are considered separately the mean increase in frequency
of the punished reSponse for the Anger group always ex-

ceeded that for the other modal response groups. Six of

”3

the nine S's who gave Anger as their modal response
showed an increase of three or more punished responses
from pre to post testing. Only twelve of the other 71
subjects showed such an increase. It seems reasonable to
conclude that subjects who gave Anger as their modal
response did not react to punishment in the same way as
‘S's who gave Isolation, Self‘attack or Denial responses
as their modal response.

'g's giving Isolation as their modal response showed
a greater increase in the frequency of correct responses
from pre to post than any of the other modal response
groups. This is true considering the eXperimental and
control groups separately or combined. The consistencies
in the above findings suggest that some real differences
are present between the Isolation and Denial groups. The
pre to post testing data also suggest then that the
second assumption cannot be held with any confidence.

In summary, both assumption #1 and assumption #2
appear to be seriously challenged by the data on acquisi-
tion as well as the pre to post testing data. It seems
therefore that in considering the effects of a reinforce"
ment, both the class of response being reinforced and the
type of subject (in terms of his class of modal response)
may make a difference. Thus comparisons between the
eXperimental and control groups and between S's under the
various conditions of post testing (e.g. stress vs. non-

stress) as well as for vauisitkm are interpretable only

nu

where all classes of responses have been associated with
a particular reinforcement for approximately an equal
number of times for all groups. Such conditions hold
only for the comparisons between the reinforcement con“
ditions across both groups (eXperimental and control) and
across all post testing conditions (stress’nonstress and
generalization'nongeneralization). In other words the
only unbiased comparison between reinforcements is
achieved when you sum across all eXperimental conditions.

Table 10 presents the means for the three types of
reinforcements across both eXperimental and control
groups and across all post testing conditions for the
total sample of 80 gfs. This table also presents the
F ratio for these means together with the Newman’Keuls
Procedure for testing the significance of the difference
between all ordered pairs of means.

The means and the Newman'Keuls Procedure indicate
that the significant F was due to an increase in the
frequency of the correct response from pre to post. The
punished and safe responses showed an almost identical
decrease in frequency. Since this analysis includes both
experimental and control groups and all conditions of post
testing, the only reasonable interpretation would seem to
be that associating a response with reinforcement by being
correct can increase its frequency from pre to post
testing.

To assess the effects of type of reinforcement,

M5

Table 10. F Test and Newman‘Keuls Procedure far
the Mean Changes ore to post for Responses Asso‘
ciated with Three Types of Peinforcements.

Type of Reinforcement Mean Change
n = $0
Punished “.66
Safe ’.h9
Correct . .79
Correct .62

’ Based on 2 different correct responses during acquisition

Source 88 df MS F
Reinforcements 1858.11 3 619.37 15.37(p<.01)
Reinforcements

x 8'8 within 8705.07 216 no.3o
GrEups

Newman-Keuls Procedure

Punished Safe Correct a Correct b

 

Ordered totals 987 1001 13M} lulo
Punished 1n 356 ’* ue3’*
Safe 3M2“. 1109"”
Correct a 67

2 3 u
q.99(r,216) . 3.6M n.12 u.uo

q.99(r,216)‘{nMS error 207 23% 250

Q. I: p<01

46

membership in the eXperimental or control group, and post
treatment conditions, a system of extracting matched
pairs from the total sample population was devised to
eliminate the biases discussed above. The data on acqui-
sition are dealt with first, then the data on changes
from pre to post testing.

From the forty subjects in the control group, twenty'
eight st could be found, each of whom matched a‘S in the
eXperimental group in terms of the class of response
chosen as the modal response and in terms of the treatment
program received during vauisition. This means that
during acquisition all four classes of responses were
associated with the same reinforcement. If Self’attpck
was punished for a‘g in the control group then Self-
attack was punished for the matching subject in the
eXperimental group etc..

Figure 22. presents the acquisition curves for the
various types of reinforcements broken down according to
membership in the eXperimental or control group. The
data are presented in terms of the number of times a
response associated with a particular reinforcement was
given in a block of ten trials (only the first response
given to a particular item was considered).

These results in Figure 22. show a definite rise in
the mean number of correct responses in both groups with
the mean of the control group above the experimental

group in all points in acquisition up to the last ten

R7

Figure 22. Acquisition Curves for Twenty'eight
Matched 8': across sixty Trials (Blocks of Ten)

 

#8

trials. A drOp in the mean number of punished and safe
responses for both groups is also indicated. The mean
number of safe responses is higher in the experimental
group than in the control group at all times during
acquisition, while the mean number of punished responses
is lower in the eXperimental group for all points in
acquisition beyond the first ten trials. Within the
eXperimental group, but not the control, the mean number
of safe responses is always considerably above the mean
for the punished responses. In the control group the
punished and safe responses alternate at being higher
across the sixty trials and for all sixty trials occurred
exactly the same number of times.

In general, these results point toward a relatively
clear differentiation between the responses associated
with being correct and those associated with the other
reinforcement, for both groups. Furthermore, there seems
to be a differentiation between the safe and punished
responses within the experimental group but no such dif‘
ferentiation within the control group. These results will
be dealt with further after the data on the changes from
pre to post testing for matched pairs have been presented.

For the pre to post testing comparisons the matched
pairs were drawn as follows (the stress factor is used
here as an example). A S was selected from the stress
post testing condition. Then from a group of S's com'

parable on all factors except stress. a S was selected

#9

whose modal response was the same as tie first a, and who
had the same treatment program (for these two S's, all
experimental conditions were identical except for the
presence of the stress during posttesting for one of
them). This Process of selection continued until each

S that could be matched, was matched. This set of
matched pairs then constituted the group used for the
assessment of the effects of stress on the types of
reinforcement. For each of the other comparisons a new
set of matched pairs was drawn (e.g. generalization vs.
nongeneralization). This procedure lead to some §'9
being used in several comparisons while other E's were
not used in any comparisons. Thus a single ananysis that
compared all groups on the various factors could rot be
made; and the hazards of doing multiple t tests has to be
considered in the final assessment of the results.

The data on the changes in the frequency of the
modal responses from pre to post testing, under the
various post testing conditions will be dealt with first.
Three sets of matched pairs were drawn from the sample
population. One set had st matched on all eXperimental
conditions and class of modal response, with the only
difference being membership in the experimental or control
group. Thirty'four S's making seventeen matched pairs
constituted this set. A second set of sixteen matched

pairs were matched as above on everything except having

the generalization program during post testing. A third

50

set (eighteen matched pairs) was matched on everything

except stress non'stress during post testing.

Table 11

presents the results of the comparisons for these groups.

Table 11. Mean Pre to Post Change in Frequency
of Modal Response X Membership in EXperimental
Stress‘Nonstress Conditions
and Generalization‘Nongeneralization Conditions.

and Control Groups,

n
E & C 18
Combined
Within E 9
Within 3 9

n
E e c 16
Combined
Within E 7
Within 0 9

n
Across
s-Ns & 17
G'NG
s 9
NS 8
G 10
NC 7

For all

8

'3-33

-o.33
”7.33
s
-u.25

-1.86
'6.11
E

“3.6%

7
“3.50

O‘\

’2.

-3.20
-2.86

1'1

18

\O

16

\l

O9.

10
7

NS

‘9.00

'7.23
-10.78

NG
“9.88

'lo.h3

'9.hh

'5.80
-8.57

t

2.14

1.92
1.28

.62

1.59

(3025

U1

(.0
(.2

O

(.025

(.025
{.20

.10

/\

(.10

groups and for almost all individual E's the

modal response drOpped in frequency of occurrence from

pre to post testing.

Relative to pretesting,

the stress

and generalization groups gave significantly more modal

responses in post testing than did the nonstress and

nongeneralization groups (stress ) nonstress 334025 for
combined Experimental and Control; generalization) non-
generalization for combined EXperimental and Control
p>(.025). Since all other responses besides the modal
responses showed some increase from pre to post testing
when all types of reinforcements are considered, it seems
quite clear that the addition of stress and or general‘
ization led to the return to the response the person used
most frequently when first entering the eXperiment. When
the experimental and control groups were considered
separately the differences were significant only within
the eXperimental group (stress)nonstress p(.05 and
generalization )nongeneralization p <. 05) . Furthermore,
the trend throughout all comparisons was for the experi‘
mental group to give more modal responses (relative to
pretesting) than the control group. Table 11. shows that
the modal response occurred more often in all the eXperi-
mental groups than in the comparable control groups. The
significance levels are low (.10 and .30) but the trend

is consistent and in line with the findings that only in
the experimental group were significant differences found
between the stress-nonstress groups and the generalization“
nongeneralization groups. Membership in the eXperimental
as opposed to the control group seems to lead to a greater
use of the modal response and to a greater differentiation
in the rate of its occurrence under the various post‘

testing conditions.

52

The three sets of matched Efs used to assess the
changes from pre to post for the modal response were also
used to assess the effects of the three reinforcements.
Table 12. Presents comparisons between the mean changes
in frequency pre to post for the stress and nonstress
groups and for the generalization and nongeneralization
groups. In each comparison, the eXperimental and control
E's are considered separately and combined. Table 13.
presents comparisons between the mean changes in frequency
pre to post for the eXperimental and control groups across
all conditions of post testing and stress,nonstress,
generalization and nongeneralization conditions separately.

In Table l}. the punished response is the only type of
response where any differences appear between the eXperi‘
mental and control group. Across all posttesting condi‘
tions and for each condition considered separately, the
control group gave more punished responses than did the
experimental group. This trend reaches a significant
difference only in the nonstress and nongeneralization
conditions. As presented in Table 12. the punished
response occurred more frequently under the nonstress and
nongeneralization conditions for eXperimental and control
g's alike. None of the differences were statistically
significant at the .05 level. If differences do exist,
however, they would seem to be present in the control
gfs (the t tests for the control g's yielded p values

that were at or close to the .10 level while for the

Table

12.

Mean Pre to Post Clanges for Punished,

Safe and

Correct Responses X Post Test Conditions for EXperimental

and Control Groups separately and combined.

E & C

Combined

Across G &

NG

Within
only
Across
& NG
(E)

Within
only
Across
& NC
(0)

E & C

C

G

Combined

Across
S & NS

Within
E only
Across
S & NS
(E)

Within
C only
Across
S & NS
(C)

mm

(D

(n m

C)

O

18
18

16
16

16

«a -q -u

\D

-2.57
-2.00
3.21

.67
.22
2.61

n
18
18
18

16
16
16

NS
-.22
-.7s
5.00

NS

-1.78

-1.67
5.39

1.111;

1.1.61
no
.75

’1.62

8.38
NG

~1.29

.99
.115

1.68

.06
.793

.69

sig.

(.10

(.10

(.OC5

<. 10

(.10

5h

Table 13. Mean Pre to Post Changes for Punished,
Safe and Correct Responses X Membership in EXperi'
mental and Control Group for Stress, honstress,
Generalization and Nongeneralization Combined and

Separately.
Mean Mean
n E n C t

Across P 17 “2.30 17 1.29 2.80 (.01
s-Ns & s 17 0.00 17 -1.00 .su
G’éG c 17 2.79 17 2.9a .15

E C
s P 9 -1.00 9 .56 1.09
Across S 9 ‘1.33 9 ‘.8 .23
G-Ns C 9 2.50 9 2.17 .20

E C
NS 2 s -3.75 s 2.12 2.76 (.01
Across S 8 1.50 8 “1.12 .92
G-Nc C 8 3.12 8 3.87 26

E c
c P 7 '.86 7 2.28 1.6M (.10
Across S 7 0.00 7 ”1.00 .BM
s-Ns C 7 2.1u 7 3 6h 82

E C
NG P lo -3.30 10 .60 2.32 (.025
Across S 10 0.00 10 ”1.00 .hl

55

eXperimental ifs they were at the .30 level or greater).

The effects of the posttesting conditions on the
frequency of the punished response were directly Opposite
to the effects of the posttesting conditions on the modal
response. While the modal responses were more frequent
under stress and generalization than under nonstress and
nongeneralization, the punished responses were more fre'
quent under nonstress and nongeneralization than under
stress and generalization.

From Table 12. it can be seen that the correct
response like the punished response and Opposite to the
modal response was given more frequently in the nonstress
and nongeneralization conditions for both experimental
and control st. The experimental and control S's show
approximately equal differences in favor of the nonstress
and nongeneralization conditions. For the combined experi'
mental and control groups these differences approach the
.05 level. Table 13. shows no hint of a difference
between the experimental and control groups for any of
the posttesting conditions.

The safe response (Table 12.) for the eXperimental
‘S's was more frequent in the nonstress and nongeneraliza‘
tion conditions. For the control S's on the other hand,
the safe response was more frequent under the stress and
generalization conditions (stress) nonstress p<.005;
generalization >nongeneralization p<.10).

Briefly, the modal response was given more frequently

56

(relative to pretesting) under the stress and generaliza-
tion conditions than under the nonstress and nongeneral‘
ization conditions for both eXperimental and control gfs.
The control E's gave the safe response more often under
stress and generalization than under nonstress and non-
generalization. The eXperimental‘g's gave the safe
response more often under nonstress and nongeneralization.
Both the experimental and control ifs gave the punished
and correct responses more often under nonstress and
nongeneralization.

In all of the above comparisons the E's were matched
in such a way that each class of response was associated
with the same type of reinforcement for each pair of
matched fife. This eliminated one bias. The conditions
of posttesting or membership in the experimental or con-
trol groups still bias the findings by differentially
affecting the classes of response. Evidence for these
biases are shown in Table 14. which presents the mean
changes from pre to post for the five classes of response
(stress vs. nonstress, generalization vs. nongeneraliza'
tion and eXperimental vs. control). Each of the signi‘
ficant differences in this table will be discussed in
terms of the bias it may have introduced with respect to
the reinforcement effects.

For the stress-nonstress comparison, the Object‘
displacement response occurred more frequencty (relative

to pretesting) under the nonstress condition (p<1005 level).

57

Table 13. Mean Pre to Post Changes in Frequency of
Responses representing the Five Classes of Response
X Posttesting Conditions and_Membership in EXperi'
mental and Control Groups.

n S n NS t
A 18 1.28 18 3.06 .82
E & C I 15 ’2.13 15 ‘.hO .85
Groups 8 8 2.12 8 2.58 .07
Combined O 18 "2.11 18 2.33 3.05 .<.005
D 13 6.85 13 3.69 .60
n G n NC t
A 15 2 67 15 h 27 77
E e C I la .07 in .57 .06
Groups 8 5 ‘3.20 5 '2.20 .h7
Combined o 16 .00 16 .06 .03

n in 2.71 it 6.50 1.55 (110

15 -2.13 15 0.00 1.61 (.10

OLD

17 -2.65 17 2.06 2.83 (.005

C1

12 7.50 12 2.83 2.03 (.05

58

In considering the effects Of the reinforcements above,
the safe response for the control fife was found to occur
significantly more often under the stress condition of
posttesting than under the nonstress condition. Since
this difference is opposite to the effect of stress on
the Object-displacement response it seems reasonable to
conclude that the difference for the safe response was
not due to the effect Of stress on the Object-displacement
response.

In Table 12. the control gfs gave the punished
response more often under the nonstress condition (.10
level). When the two gfs who had Object‘displacement as
their punished response were eliminated from the compari“
son, the mean change for the stress group was -.1u and
for the nonstress group .57. The difference between
these adjusted means was not significant (p).lo). Thus
the difference in the frequency between the punished
response under stress as Opposed to nonstress conditions
for the control group may be due to the effect of stress
on the Object‘displacement class Of response.

Also in Table 12. the correct response for the com“
bined eXperimental and control groups occurred more fre‘
quently under the nonstress than the stress condition.
When the eight gfs who had Object’displacement as one of
their correct responses were eliminated from the compari-
sons the adjusted mean increase in the number of correct

responses for the stress group was u.65 and for the

59

nonstress group n.35. The difference between these
adjusted means was not significant (p>.lO) and the direC'
tion Of the difference was Opposite to that which occurred
when the Object‘displacement E's were included. It
appears therefore that the difference between the frequency
of the correct response under the stress and nonstress
conditions may mtbe dueto the effect of the stress con-
dition on correct responses per se. This is further
supported when the frequency of the Object’displacement
response is compared under stress and nonstress condi'
tions for E's who had the Object‘displacement response
reinforced by punishment and safety. The mean change in
frequency pre to post for the combined punished and safe
responses (only Object‘displacement responses considered
here) under stress was ”2.90 and under nonstress n.80.
This difference is significant at the .005 level. No
differences were found when all five classes Of responses
were considered together (Table 12. Stress vs. Nonstress
for EXperimental and Control ifs combined).

For the generalization'nongeneralization comparison
in Table 1a. the Denial class Of response showed a ten-
dency to occur more often under nongeneralization (p<.10).
When the seven E's who had Denial as one of their correct
responses were excluded in this comparison the mean changes
for nongeneralization was 3.17 and for generalization 3.00.
It would seem that the tendency to give more correct

responses under nongeneralizatton may be a tendency for

60

Denial responses to show a greater frequency under non‘
generalization conditions.

Also in Table 1U. the mean change in frequency for
the five classes of responses is given according to mem-
bership in the experimental and control group. here the
Isolation, Object'displacement and Denial responses all
show some differences between the two groups. In Table
13. only the punished reSponse showed a tendency to occur
with different frequencies in the experimental and con-
trol groups. When all ifs who had Isolation, Object-
displacement or Denial as their punished response were
excluded in the comparison between the eXperimental and
control groups the punished response still occurred sig-
nificantly more Often in the control group (adjusted mean
decrease for experimental group ‘3.50, adjusted decrease
for the control group ‘1.00 p<.05). The tendency for a
punished response to occur more Often in the control group
would not seem to be simply an effect of membership in the
eXperimental or control group on a particular class of

response.

Finally, the conditions of posttesting or membership
in the eXperimental or control group could have had a dif-
ferential effect on E's with differing modal reSponses.
Since more than half of the i's chose either Self‘attack
or Denial as their modal response, the number of matched

S's for the other three classes of responses was too small

for any analysis. Table 15 presents the comparisons between

61

f‘ ’Jr‘r‘+‘ 1‘ «‘7‘ .~ {v
To :..:.~-i. P

..—
‘I ‘

'7- . ',: \..~' ~ . “me.
31:19 1/ . 3.0.3.11 PI‘P ". ::

A

.v- ' N."
.:)r {Mud

1 1 p

Nonstress fixgerimehtal V3. Contri. iar 5"s
with either Self-attack or Denial as their Modal
Peeponse.

Q er
«1" v x
(1‘8“ \

n Stress n Nonstress t
B P 5 .00 5 ‘2.u0 .99
¥ s 5 -1.20 5 -2.u0 M1
2 c 5 .30 5 5.h0 2.00
Stress Nonstress
g g P 10 -2.30 10 .50 1.u6
% i s 10 .30 10 .30 '
- g c 10 1.95 10 1.70 1.19
Experimental Control
3 P 5 -2.h0 5 .60 .97
g s 5 -2.60 5 “.U0 1-21
i C 5 1.30 5 .70 .2
EXperimental Control
2; P 9 ‘2.67 9 1.77 2.60
:5: s 9 .33 9 -1.3b. .66
_ E c 9 3.38 9 1.22 .35

Punished, Safe and Correct Responses X Ftress vs.

0?

sig.

(.10

62

the stress and nonstress conditions of posttesting and

the comparisons between experimental and control E's
broken down according to the E’s modal response. The data
here are very meager and most comparisons would indicate
that E's with differing modal responses were similarly
affected by the posttesting conditions and membership in
the experimental or control group. In the stress-nonstress
comparison however, the E's who gave Denial as their modal
response tended to give the punished response more under
stress than nonstress while the Self‘attack gfs gave more
punished responses under nonstress than stress.

In short, the procedure of matching fife eliminated
some of the biases in the analysis of the effects of the
reinforcements. This procedure did not eliminate any
differential effects of the posttesting conditions and
membership in the eXperimental or control group on the
classes of response or on g's who gave differing modal
modal responses. Table 1M and 15 have shown that these
differential effects should not be discounted in further
research.

Discussion
Two foci of this study were the responses that g gave
most frequently (modal response) during the pretesting
phase of the eXperiment and the responses associated with
the various types of reinforcement. Since two of the
assumptions underlying the experimental design were chal-

lenged by the results of the study, the implications of

63

these results will be discussed first.

Challenging the assumptions: It was tentatively

 

assumed that E's who differed from each other in the class
of response given most frequently during pretesting would
not differ greatly from each other in their response to
the various reinforcements (e.g. Punishment would have
approximately the same effect for all gfs regardless of
modal response). The results for both acquisition and pre
to post comparisons contradict this assumption. In par'
ticular. fife who gave Anger responses most frequently
during pretesting seemed to react differently to punish?
ment than did other fife. It seems that separating g's
according to the class of response given most frequently
during pretesting also separates them to some extent in
terms of their reactions to certain reinforcements. Another
possibility, although the effect is the same, is that ifs
with different modal responses reacted differently to the
total experimental situation, and this affected their
reactions to the individual reinforcements.

If Efs with different modal responses differ in their
reactions to the various reinforcements, then any attempt
to change the frequency with which E gives a particular
response must take into account this individual difference
variable. Attempts to modify the frequency of a response
on the basis of a learning paradigm which predicts in most
cases but not in each individual case would seem to be

relatively inefficient. If the response to punishment of

6h

the §fs giving Anger as their modal response is any indi’
cation, the results could be directly Opposite to the
general prediction.

Thus future research in this area could do well to
emphasize the role of individual differences. The classes
of response (Anger, Isolation, etc.) used in this study
seem to have merit as a means of identifying individual
differences that are relevant for predicting response to
various reinforcements. There are, however, probably many
other response measures which might be as useful, if not
more useful (GSR, diversity vs. stereotypy in responding
under stress etc.).

It was also assumed that reinforcing one class of
response would produce the same sort of effect as for any
other class (i.e. when Anger responses were punished they
would show a frequency change in the same direction and
of the same approximate magnitude as Isolation responses
etc.). It seems more likely that this assumption is incor-
rect than correct. In figures 6 through 16. it was shown
that responses associated with "anxiety reduction“ drOpped
in frequency over the sixty acquisition trials if they
were Anger responses, increased if they were Isolation
responses, and showed little change if they were Denial
responses. As was the case with differences in modal
response, attempts to alter the frequency of a particular
response simply on the basis of a learning paradigm which

predicts for most classes of response, seems to be inefficient.

65

Presumably, future research should focus on a detailed
description of the responses under study. After the fact,
it seems reasonable that describing a response simply as
a “verbal response" is an insufficient description in a
learning context. Replying with "Charity begins at home!
to the attack statement "You think only of yourself? is
evidently psychologically quite different from replying
with "Thinking of you gives me a headache.'. The classes
of response used in this study seem to show some merit
for establishing a meaningful response division, at least
for handling replies to attack statements.

In brief, if principles evolved from human and animal
learning are to be applied to the broad fields of person'
ality and psychotherapy, then a great deal of attention
will need to be directed to the description of individual
differences as well as to what constitutes a psychologi-
cally meaningful response unit in a.particular situation.
This study, however, was concerned primarily with the
effect of reinforcements on "interpersonal behavior. While
acknowledging that the results of the reinforcement treat-
ments will have to be accepted with some reservation, we
believe that certain statements concerning’the effects of
the reinforcements can be made.

The effects of thegposttesting conditions on the rein-
forced responses in relation to the modal responses: One
of the main questions posed at the outset of this study

was as follows: Which of the reinforcement treatments,

66

if any, can modify a nonmodal response so that the effects
of the posttesting conditions on the nonmodal response
mimic the modal response? Reinforcement by ”anxiety reduC‘
tionI was posited as the most likely reinforcement to lead
to this result. The no clear reinforcement type of rein-
forcement was designed to function as the major contrasting
condition of reinforcement. No clear reinforcement dif‘
fered from "anxiety reduction" reinforcement only in the
fact that E's getting no clear reinforcement (the control
E's) were not informed that punishment would no longer
occur on a particular item once the response in question
was made. On the other hand, the eXperimental gfs (those
with "anxiety reduction"I as the "safe" response) were
informed by the green bar on the program paper each time
the aversive noise (punishment) was avoided. It was
assumed that the experimental E's, by virtue of their
knowledge that they had avoided punishment, would eXper'
ience Ianxiety reductionn while the control E's would not.
The other reinforcement conditions, reward by correctness
and.punishment, were required for the definition of the
critical reinforcement conditions and provided base line
data for this kind of learning situation.

The reinforcement effects during acquisition; The
key data on the matched pairs (Figure 22) suggest that
there were at least three different types of reinforce-
ments. Reward by correctness, punishment, and "anxiety

reduction" (experimental gfs) all led to acquisition

67

curves that were discernibly different from each other.
No clear reinforcement (control Eds) led to acquisition
curves very similar to punishment. The data seem to sup-
port the contention that "anxiety reduction“ as used in
this study was a bona fide reinforcement which was dif‘
ferent from the other three.

The differential effects of the posttesting_conditions
on the modal regponse and the reinforced responses: The
effects of the posttesting conditions on the modal response
were clear and in the direction anticipated in the design
of the study. The addition of stress or stimulus change
led to the use of the older stronger habit, the modal
response (the frequency of the modal response was greater
under stress and generalization than under nonstress and
nongeneralization). The answer to the question of which,
if any, reinforced response mimicked the modal response
also seems clear. The responses followed by no clear rein-
forcement occurred more frequently under stress and gener-
alization conditions than under nonstress and nongeneral-
ization conditions. The correct, punished and "anxiety
reduction" responses all tended to occur more frequently
under the nonstress and nongeneralization conditions. The
effects of the posttesting conditions on the correct and
punished responses were as anticipated at the outset of
the study. The effects of the posttesting conditions on
the ”anxiety reduction” responses were directly Opposite

to those anticipated. Even though the acquisition curves

68

for no clear reinforcement were very similar to the curves
for punishment, the posttest data seem to indicate that

no clear reinforcement led to a response that was affected
by stress and generalization in a manner similar to the
modal response, or in a manner different from the other
reinforced response.

The return to the modal response: The eXperimental

 

gfs tended to give the modal response more frequently than
the control gfs under all posttesting conditions. It
appears that "safe” response substituted in part for the
modal response for the control 2's and mimicked the modal
response under stress and generalization conditions. The
eXperimental E's clearly gave the modal responses more
frequently under the stress and generalization conditions
than under the nonstress and nongeneralization conditions
(stress) nonstress p (.05; generalization > nongeneraliza"
tion.p<.025). The control gfs on the other hand did not
show the same clear differentiation in the frequency of
the modal response in the stress vs. nonstress and gener‘
alization vs. nongeneralization comparisons (p for both
comparisons (.20).

One tentative interpretation of these findings is
that only the no clear reinforcement condition correspon‘
ded to the paradigm for avoidance learning and provided
for the learning of a response which mimicked the modal
response. A second, somewhat more speculative interpre-

tation of the results is that the paradigm for avoidance

69

learning was actually fulfilled by both the no clear rein-
forcement condition and the "anxiety reduction" condition.
Both types of reinforcements involved avoiding punishment
though only the eXperimental gfs were aware of the situa‘
tion. The element of conscious awareness or the lack of
it may then have led to the conditioning of two different
avoidance responses. The avoidance response learned in
the case of the control g's was the giving of the specific
class of response reinforced by the no clear reinforcement
condition during acquisition. The avoidance response
learned in the case of the eXperimental gfs was much more
general or abstract; namely, the giving of a safe response
rather than the giving of a specific "safe" response. Thus
when the modal response was again available in the post‘
testing phase the eXperimental‘g's having learned a more
general avoidance response, used the already safe modal
response as a safe response.

The sgpport for the second interpretation; The sup-
port for the interpretation that two different avoidance
responses were learned comes from two sources; the data
on the control gfs in this study, and the report of the
Eriksen and Kuethe study (1956).

Although the control gfs gave the no clear reinforce-
ment response more frequently under stress and generaliza-
tion than under nonstress and nongeneralization, the mean
number of no clear reinforcement responses accounted for

less than thirty percent of the responses given by control

70

‘é's during posttesting under stress and generalization.
This implies that the control E's could have used the
modal response more frequently under stress and generali'
zation than they did. The fact that the control gfs did
not give the modal response more frequently under stress
and generalization suggests that the "anxiety reductionu
reinforcement facilitated the use of the modal response
among the experimental gfs.

The Eriksen and Kuethe study (1956) also seems to
lend support to the interpretation that two different
avoidance responses were learned as a function of aware-
ness. Some of Eriksen and Kuethe's gfs were aware of the
relationship between shock and the giving of particular
words in their word association tasks while other E's
were not aware of this contingency. The "aware" _8_'s repor-
ted withholding associations and had longer reaction times
than the "unaware" gfs who did not report withholding
associations. Yet both groups ("award'and “unaware*)
avoided the shocked words. It seems reasonable to posit
that these two groups of §fs may have been learning two
different kinds of avoidance response as a function of
awareness. Awareness that an avoidance response is being
made may alter the nature of the response given.

Some igplications of two different kinds of avoidancg
response: Mednick (1958) in attempting to interpret some
of the behavior patterns of schizophrenics in learning

theory terms has posited that the behavior was learned

71

because Of association with anxiety reduction. He made
no distinction between avoidance responses learned by
accidental association with anxiety reduction and those
learned under conditions where the g was apparently more
aware that his response led him away from an anxiety pro-
voking tOpic or situation.

However, this distinction between the type of avoid?
ance response learned may be an important distinction for
conditioning types of psychotherapies to make. Because
Of the remoteness in meaning from the anxiety provoking
sitmulus, avoidance responses learned by accidental asso-
ciation with anxiety reduction might be amenable to sympto-
matic treatment without symptom substitution. On the
other hand, avoidance responses “chosen" by a person to
reduce anxiety are likely tO be Openly related to the
anxiety provoking stimulus. In this case anxiety reduc-
tion is directly associated with the occurrence of this
particular avoidance response. Thus eliminating the avoidance
response could leadtb mton substitution. In addition, non-
specific avoidance responses like “thinking away“ from an
anxiety provoking tOpic might be untouched by psychotherapy
as the virtually infinite pOOl Of ways to ''think away“ gets
"diminished” during therapy.

A word about the artificial nature Of this study seems
in order before closing the discussion. This was a labor-
atory study and the data were not based on the real life

situation they purport to study. The data, however, would

72

seem to be at least at the same level Of relationship to
1‘real life" as are data Obtained from the TAT, story tell-
ing'(Miller and Swanson, 1960), and many other techniques
used to predict the situations outside the laboratory.
Within this limitation it seems that this study has pointed
to the following general findings.

1.) Individual differences between gfs play an impor-
tant role in predicting their response to various reinforce-
ments. The scheme Of categorizing fife according to the
frequency with which they use the five classes Of responses
(Anger, Isolation, Self-attack, Object-displacement and
Denial) would seem to have some merit in delineating rele‘
vant individual difference variables.

2.) The modification Of the frequency with which
different responses are given in a verbal attack setting
should take into account the fact that different kinds Of
responses are affected differently by various reinforce-
ments. The scheme Of categorizing responses into the five
classes of response used in this study seems to have some
relevance for predicting the effects Of various reinforce-
ments on different kinds Of responses.

3.) At least within the limits Of this eXperiment, the
frequency with which gfs give nonmodal responses can be
modified by associating the nonmodal responses with various
reinforcements.

4.) The nO clear reinforcement type Of reinforcement

appears to modify nonmodal responses in such a way that

73

the response occurs more Often under stress and general-
ization than under nonstress and nongeneralization. This
effect is similar to the effect Of these conditions on the
modal response and different from the effect Of these con-
ditions on the correct, punished and nanxiety reduction“l
responses.

5.) The presence Of the "anxiety reduction" reinforce’
ment as opposed to the no clear reinforcement during acqui‘
sition seems to lead to a greater return to the use of the
modal response during posttesting, and to a greater dif‘
ferentiation in the frequency Of the occurrence of the
modal response under the various posttesting conditions.

Summary

The present study attempted to apply some Of the
principles Of avoidance learning and verbal conditioning
to the interpersonal situation where one person verbally
attacks another person. An artificial interpersonal sit“
uation was used to increase experimental control. The
study was primarily designed to explore the question as
to whether anxiety reduction could function as a rein-
forcement in such a situation. Would following a verbal
reply with anxiety reduction lead to a habit that would
generalize to a stressful situation as well as generalize
to a situation which differed from the original learning
situation?

Eighty male undergraduate students enrolled in ele-

mentary psychology courses served as subjects. They were

7h

subjected individually to a three phase experiment. For
all three phases they were seated at a teaching machine

and were presented with the eXperimental stimuli through

a window in the machine. The first phase was an evaluation
phase, the second an acquisition or learning phase and the
third another evaluation phase.

During the two evaluation phases, the stimuli consisted
Of twenty verbal attack statements presented one at a time.
The attack statements were printed in a cartoon-like bal-
loon accompanying the line drawing Of a human face. A
second line draWing Of a face was accompanied by a blank
balloon representing the person responding to the attack.
The st composed and wrote replies to each Of the twenty
attack statements.

After the first evaluation phase, each Of the responses
was classified into one Of five classes Of response each
representing a different way in which the responder ex’
pressed or failed to express anger in his reply. The
class Of response which occurred most Often during this
phase was labeled as the modal response and was excluded
from the acquisition phase.

During the acquisition phase the four remaining classes
of response (three Of the four on each presentation) were
printed in the balloons accompanying the face, representing
the responder. Each response was followed by a type of
reinforcement. For each 2 two of these classes of response

were rewarded by being correct, one was punished, and one

75

led to the avoidance Of punishment. Half Of the file were
aware that giving one Of the reSponses would lead to the
avoidance of punishment (anxiety reduction or experimental
group) the other half were not aware Of this (no clear
reinforcement or control group).

The third phase was identical to the first evaluation
phase except half the st were stressed during the evalua'
tion, half were not, half had stimulus material similar to
acquisition and half had new stimulus material (four groups).

It was assumed that the effects Of the reinforcements
would be approximately equal on S's with different modal
responses and apprOximately equal on the five classes Of
response. Both these assumptions were found to be unten-
able. Individual differences in st led to differential
reaponses to reinforcement, and the different classes of
response were differentially affected. Even so, responses
reinforced by being correct, in general, increased and
those followed by punishment decreased.

As eXpected stress and novel stimuli as Opposed to
lack Of stress and similar stimulus material led to an
increased use Of the modal response. The responses rein-
forced by avoiding the punishment without the st awareness
led tO results which mimicked those for the modal response.
st who were aware they could avoid.punishment used the
modal response more than the unaware st particularly under
stress and with novel stimuli. The results were interpreted

in terms of learning two different kinds of avoidance

76

responses depending on the S's level Of awareness that

avoidance had occurred.

References

Child, I.L. Personality. In C.P. Stone and Q. McNemar (Eds.),
Annual Review Of Psychology. Stanford, Calif.: Annual
Reviews, 195h. Pp 1&9'170.

Diven, K. Certain determinants in the conditioning Of
anxiety reactions. J. Psychol., 1936.3»291'308.

Dollard, J. and Miller, N.E. Personality and Psychotherapy.
New'York; McGraw'Rill, 1950.

Eriksen, C.W. and Kuethe, J.L. Avoidance conditioning Of
verbal behavior without awareness: a paradigm Of
re ression. J. abnorm. soc. Psychol., 1956:53. 203‘
20 .

Eysenck, H.J. Behavior therapy and the neuroses. New‘York;
Pergamon Press, 1960.

Franks, C.M. Conditioning and abnormal behavior. In N.J.
Eysenck (Ed.), Handbook Of abnormal Psychology. New
Your; Basic Books, 1961. Pp #57'h87.

Greenspoon, J. The effect of verbal and nonverbal stimuli
on the frequency Of members Of two verbal response .
classes. Unpublished.Ph.D. Dissertation, Indiana Uni‘
.versity., 1951. Cited by J. Greenspoon, Verbal Condi-
tioning and Clinical Psychology. In ANJ. Bachrach (se.).
EXperimental Foundations Of Clinical Psychology. New
York: Basic Books, 1962. Pp510’553.

Greenspoon, J. The effect Of two nonverbal stimuli on the
frequency Of members Of Two verbal response Classes.
Paper presented at the American Psychological Associar
tion Convention, New‘York, N.Y. l95h. Cited by J
Greenspoon, Verbal Conditioning and Clinical Psycho-
logy. In A.J. Bachrach (Ed.), Experimental rounda-
tions Of Clinical Psychology. New'York: Basic Books,
1962. Pp 510'553.

Greenspoon, J. The reinforcing effect of two spoken sounds
on the frequency Of two responses. Amer. J.Psychol.,

1955, 68. h09‘h16.

Hildum, D.C. & Brown, R.W. Verbal reinforcement and Inter-
view bias. J. abnorm. soc. Psychol., l956.53,lOS“lll.

Jones H.G. Learning and abnormal behavior. In H.J. Eysenck

(Ed.), Handbook Of abnormal psychology. New York;
Basic Books, Inc., 1961. Pp #88’528.

77

78

Kimble. G.A. hilgard and Marquis' Conditioning and Learning.
New'York; Appleton'Century'CrOfts, 1961.

LaceY. J.I. & Smith, R.L. Conditioning and generalization
Of unconscious anxiety. Science, 1954, 120, 10h5'1052.

Lacey. J.I., Smith, R.L. & Green. A. Use of conditioned
autonomic responses in the study of anxiety. PsychO‘
som. Med., 1955, 17. 218-217.

Mednick. A.A. A learning theory approack to research in
schizOphrenia. Psychol. Bull., 1958, 55, 316-327.

Miller, D. & Swanson, G.E. Inner conflict and defense.
New'York; Holt, 1960.

Mowerer, O.H. Learning theory and personality dynamics.
New'York: Ronald.Press, 1950.

Mowerer, O.H. Learning theory and behavior, New York;
Wiley, 1960.

Nuthmann, A.M. Conditioning Of a response class on a
personality test. J. abnorm. soc. Psychol., 1957,
5’4; 19- 23 .

Osgood, C.E. Method and theory in experimental psychology.
New York; oxford University Press, 1953.

Salzinger, K & Pisoni, S. Reinforcement of verbal affect
responses Of normal subjects during the interview.
J. abnorm. soc. Psychol., 1960. 60, 127'130.

Shoben, E.J. Jr. Psychotherapy as a problem in learning
theory. Psychol. Bull., 19h9.fl6. 366‘392.

Solomon, R.L., Kamin, L.J. and.Wynne, L.C. Traumatic avoi'
dance learning; the outcomes Of several extinction
Procedures with dogs. J. abnorm. soc. Psychol., 1953,
us. 291-302.

Solomon, R.L. & Wynne, L.C. Traumatic avoidance learning;
Acquisition in normal dogs. Psychol. Monogr., 1953.
67. NO 1h (whole NO. 35h).

Wolpe, J. Experimental neuroses as learned behavior. Brit.
Jo P8UChOl., 1952) ’43: 2143-268.

Wolpe, J. Reciprocal inhibition as the main basis Of
psychotheraputic effects. A.M.A. Arch. Neurol. Psychiat.,
195,4) 72: 205-226. '

WOl e J. Ps chothera b reci r c 1 inhibition. Stanford:
p Stanforg.UniversE¥y Press,pl95g.

79

Appendix 1. List Of Eighty attack statements

\OOQNmUWEWNH
O

H h‘ F4
n: Id 0
O O O

H
\N
I

15.
16.
17.

18.
19.
20.

Twenty Pre and Post Test Attack Statements
You're Just a big mama's boy.
You lOOk like you sleep in your clothes.
You never consider the other persons' point Of view.
Why can't you do your own thinking for a change?
You're nothing but a pretty boy.
You should do something about your weight.
You're ninety percent hot air.
You're going to have to get a hold on yourself.

When are you going to learn tO work things out for
yourself?

When the going gets tough. you always fold up.
anl never take anything seriously.
You don't give a damn for anyone but yourself.

You have nO principles at all.

. You louse up everything you do.

Couldn't you try tO look Just a little bit neater?
When are you going tO grow up?

You always depend on others tO make your decisions
for you.

You're as scatterbrained as any woman.
You're never completely honest about anything.
You clothes never look half‘way decent.

Sixty acquisition attack statements with
alternative replies

When are you going to do something without someone
holding your hand?
(D) Nobody's holding my hand.

S) As soon as I can. -

80

( ) At least I do something.
( ) What do you mean?
( ) How about shutting your mouth.
ere were you when the brains were handed out?
) At the end of the line.

) Who knows?

) In front of you.

) I was there.

)

O
I
A
h
S
I
A
D
0 At least I have an excuse.

w
(
(
(
(
(

You do things in a feminine way.
(I) It takes both kinds.

(D) Never.

(A) Go to hell.

(0) I see that you watch for these kind of things.
(8) Just call me Violet.

You can't ever be trusted.

(A) You're a liar.

(8) Really?

(I) Honesty is the best policy.
(p) I think I can be.

(0) And you?

You'd cheat anyone you could.
(D) I would not.
(I) Crime doesn't pay.
(0) Speak for yourself.
(a) You're out of your mind.
(8) Even my mother?

Horse play is all you're capable of.
(0) And some peOple have no sense of humor at all.
(A) Take a flying leap.

Do I really present that impression?

No. actually, I'm very serious.

All work and no play makes Jack a dull boy.

AAA
Hum
vvv

never contribute anything to solving a problem.
I contribute all that I can.

I'm not the only one.

It depends on what the problem is.

I'm a little slow at times.

Neither do you.

AAAAAN
>mHOuo
vvvvv c

K

pmuOHo

don't do anything on your own.
Two heads are better than one.
Who says I don't?

I do too.

I've done some things on my own.
How can I with you on my back?

AAAAA
vvvvv C

10.

ll.

12.

13.

14.

15.

16.

81

u're a slob.

) I am not.

And you act the part.

l‘ttakes one to know one.

Anything else the matter with me?
That's the way it goes.

AAAAAK
HmObu O

p

O H H
vvv VVH vvvv

you do is gossip all the time.

I'm watching people's reactions.

And some people I know repeat it every chance
they get.

That is not gossip.

I learned it from you.

It's in my nature.

AAA AA
(9 > U

won't ever get anywhere in life.

I believe I will.

I'll go further than you ever thought of going.
Time will tell.

I'm hoping.

Maybe not, but I know where I'm going.

AAA/\AK
O (D H): U 0
vvvvv C

u're always looking for the easiest way out.
) I'm lazy.

) Only a Jerk wouldn't.

) I work Just as hard as anybody else.

) Life is difficult.

) What's it to you?

AAAAAK

do you let women push you around all the time?
Easy going I guess.

There's more than one way to skin a cat.

What gives you that stupid notion?

I never let that happen.

You're crazy.

>UOHmD’ >Huomo

AIKAAA:
vvvvvq

n't you ever act your age?

0) People who ask that are usually getting old.
A) I don't want to emphasize your immaturity.
D) Always.

S
I

) How old do you think I am?

you always mumble when you talk?
I) What do you mean always mumble?
S) I hope not. do you think so?
A) Wash out your ears.
0) Have you ever listened to yourself.
D) I don't mumble.

{an galk like the president of the Ladies Garden Club.
am.
(I) Some women are very good speakers.

17.

18.

19.

20.

21.

22.

23.

82

(A) I guess you know because you attend the meetings.

(D) No I don't.

(0) Strange, that only happens when I talk to you.

At least once in a while you ought to be able to do

a good Job.

(A) Try it yourself sometime.

(0) It's funny how the biggest goldbricks are the
most critical.

(D) I do good work.

(8) I was Just born that way.

(I) What constitutes a good Job?

Aren't you ever going to learn how to take respon-
sibility?

D) I feel that I accept responsibility very well.
(A) You're the most irresponsible person I know.
(1) Responsibility is a vague word.

8) When I grow up.

(0) How can I when I'm not given the chance.

You think only of what's in it for you.

Who doesn't?

I'll never top you at it.

I suppose it has happened.

Self‘preservation is the first law of existence.

Hmbo

AAA
vvvv

daydream too much.

Not too much.

Not as much as a lot of people.

I like it better than listening to you.
What does daydreaming too much mean?
Very true.

AAAAAK

CDH>OUO
vvvvvc

always have someone else solve your problems.
Outsiders can sometimes see a problem more
clearly.

Someone else than you that is.

It's easier that way.

You are mistaken.

And you're always trying to solve someone else's
problem.

AAA/s AK
3’ U a: O H O
vvvv \’ c

do the stupidest things.
That's not so.

Maybe it's the company I keep.
Everything is relative.
Sometimes I do.

Like listening to you.

AAAA’NK

>mHouo
vvvvv C

5

not face it, you Just can't take it.
I never said I could.
T eke what?

AA:
H m
vv

2M.

26.

27.

28.

29.

30.

53

(A) Shut up.
(D) I disagree.
(0) I guess you think you're a good Judge.

Can't you come up with an original idea once in a
while?

(0) When I become as knowledgeable as you, I will.

I do quite often.

All ideas are merely repetitions of previous ones.
Once in.a while, puts me ahead of you.

No.

m>HU

AAA"

have a very annoying laugh.

You certainly are a rude person.

Laughter is the medicine of the soul.

I can't change it.

With me it's only my laugh that's annoying.
It doesn't bother me.

AAAAAM

UOKDH>O
vvvvvfi

always pick on the little guy.

Are you commenting on my stature.

You say that to everybody, shrimp.

I doubt it.

He's the only one I can beat.

The little ones always have the most bothersome
behavior.

AAAAAK

OUIU>HO
vvvvvc

Everything is Just a big Joke with you.

(A) Not everything. Just you.

(0) That shows how well you know me.

(3) Yes it is.

(D) You're wrong.

(I) Laugh and the world laughs with you.

n't you ever make up your own mind?

) I really feel that I do.

) Yeh, go to hell.

) Deciding things is a difficult task.
) I'm slow at decision making.

) Does somebody say that I don't?

AAAhfis
O m H” U 0

§

en are you going to start acting like a man?
A) When I stOp following your example.

3) Probably never.

I) What do you mean. "not like a man"?

0) When people stop asking foolish questions.
D) I don't have to act like one, I am one.

AAA’NA

You act Just like a woman.

(0) That was a rotten thing to say.

(A) You're nuts.

(I) What do you mean by acting like a woman?
(a) Do I 2

8%

(D) I don't think so.

31. You never care how you look.
(A) You're crazy.
(0) We both know that's a lie.
(I) Some things can't be helped.

(8) Well, maybe I don't.
(D) Sure I do.
32. You hate to see the other man get ahead.
(I) Anyone with the qualifications should be recognized.
(8) You could be right.
(D) I like to see the other man get ahead.
(0) Only if he's a slob.
(A) You don't know what you're talking about.
33. Why can't you stand on your own two feet?
(8) I don't know.
(D) I always have.
(0) Because usually somebody's stepping on my toes.
(A) Why don't you shut up.
(I) No one can depend completely on his own Judgment.
3h. Can't you do anything right?
(I) Very few people ever ask me that question.
(8) That's what I keep asking myself.
(D) or course.
(A) More than you do.
(0) So who's perfect?
35. You're the most insincere person I know.
(0) I'm not going to baby anyone.
(8) I guess I am.
(D) I'm very sincere.
(A) You're the craziest guy I've ever known.
(I) What is sincerity?
36. Y ve got an ugly puss.

u!
I) Beauty is in the eyes of the beholder.
D) That's not so.

0) Take a look at yourself.

8) I can't help what I look like.

A) Not half as ugly as yours.

37. You couldn't get by without someone telling you what
to do.
(D) I could so.
(8) Maybe not, that's Just me a follower.
(A) Time for you to get lost.
(I) Some people are born leaders.
(0) That's your opinion.

38. You're Just a weakling.

39.

kc.

M1.

M2.

#3.

1m.

’45.

85

I) Everyone can't be Hercules.

D) That's not quite true.

8) I'll make a sincere effort to correct it.
0) Not compared to the people around here.
A) I'll take you on any day.

on always look as if you're half asleep.
0) PeOple make me tired.

I) Judge by actions not looks.

8) Half of the time I am.

A) What's it to you?

D) I don't think so.

0

't you ever think things through before you act?
No, I'm naturally this way.

The first duty of life is to live.

Mind your own business.

I always think things through.

At least I can think.

AAAl-s’so
O U > H m

hings don't go your way you have a fit.

I do not.

Some things you can have and others you can't.
You're a liar.

At least I act my age.

Well. I'm concerned.

vvvvv cf vvvvv :3

AAAAA:
mo>Ho

walk Just like a woman.

You walk like an ape.

At least I walk and don't crawl.

NOpe.

Do women have a special way of walking?
Perhaps.

AAAAAK
U! H U 0 > O
VVVVVC

're nothing but a tightwad.

I resent that.

I didn't realize it.

And you're my inspiration.
Better tight than poor.

I think you're pretty generous.

AAA/KAN
U H b m 0 O
vvvvv c:

lean on someone else every chance you get.

I try not to.

it takes intelligence to know when and where to
ean.

I guess I should try to be more like you huh?

I never did and never will.

If I leaned on you, you'd collapse.

AAA “AK

>UO Hmo
vvv vv ‘3

u think only of yourself.

) Thinking of you gives me a headache.
) We all can't be perfect.

) I always consider the other person.

86

(I) Charity begins at home.
(3) Do you really think so?

M6. Ca
(D
(I
(o
(s
(A
#7. Y
(D
(s
(o
(A
(I
48. Yo
(3
ED
0
(I
(A
49. Whe
(D
(s
(A
(I
(o
50. Don'
(A
(I
(o
(s
(

wvvv

vvvvv :3

)
)
)
)

't you ever let go of your mother's apron strings?

I did a long time ago.

I don't believe in things like apron strings.
She's better to me than some peOple.

Perhaps not.

Don't sweat it fella.

ou haven't got any backbone.

Yes I do.

I'm working on that problem.

I suppose you do?

I have more guts than you.

Scientific evidence shows everybody has a backbone.

act Just like a two year old.

I feel like a two year old at times.

I always act my age.

I won't say what you look like.

You're only as old as you.feel.

How'd you like a twenty year old punch in the mouth?

are you going to take some pride in your appearance?
I take Pride in my appearance.

I don't know.

You Just have lousy taste.

Appearance isn't everything.

Comparatively speaking, my appearance is good.

t come sniffling to me for help.

I'd never ask a bum like you for help.

What gives you that impression?

I only ask peOple who are capable of giving help.
Who else can I go to?

D) I can take care of myself.

51. You're only interested in kid stuff.
(a) I am a kid.

(I)
(A)
(D)
(0)

52.

You

MAR;

D
S
(I

SVVV:

It depends on what you call kid stuff.

Get out of here, I'm tired of the sight of you.
No. I'm usually quite serious.

I'm not interested in you.

might as well wear a skirt the way you act.

You ought to try it when you get out of diapers.
There's nothing wrong with the way I act.

Am I really that bad?

A skirt would be cooler in the summer.

You're a fine one to talk.

53. Can't you shut up once in a While?
(D) I don't talk too much.

87

(A) I can shut you up.

(0) I. at least. have something to say.

(8) All right. I won't say another word.

(I) Do you feel that I talk too much?
5%. You never give anyone a break.
(A) You wouldn't recognize one if you got it.
(S) I am sorry.
(D) Oh yes I do.
(I) People make their own breaks.
(0) They don't deserve it.

55. n't you ever take a stand on anything?

) Try talking about something important.

Go Jump in the lake.

I always stick by my opinions.

Everything can't be set down in black and white.

I have a hard time making up my mind.

AAAA/‘Q
(DI—0c) p 0 p

vvvv

56. 't you ever settle down and get something done?

I get lots done.

You're worse than I am.

Planning is always a part of getting a Job done.
I'm a slow starter.

Who are you to talk about that?

“AAA/‘0
O m H :p U g
vvvvv

57. seem to be involved in something underhanded all
time.

A crook like you has no kick coming.

I'm certainly not the worst one around here.
That is not at all true.

I guess I'm no good.

Seeming doesn't make it so.

AAAAAdH:
H m U 0 > {3'0
VVVV'»; (b c

58. You're nothing but a sissy.
(D) No I'm not.
(0) Are you something better?
(S) I feel like one when you criticize continuously.
(A) You're looking for a black eye from me.
(I) The world is no longer only for the physically strong.
59. cu have a strange walk.
D) Not to me.
A) And you have a strange face.
0) And some people are still on all fours.
I) No two people walk the same.
3) I can't help it.

60. Why don't you straighten up and fly right?
(D) I do fine.
(I) The wind's too strong.
(S) Give me time.
(0) I don't see any good example around here.
(A) Why don't you wake up and die right.

88

Appendix 2. Six Faces used on Acquisition PrOgrams,

 

.Iwn-J

 

V. "9 I‘M.
thumbs.

   
 

 

..4‘ A —L .‘h“

Herr-Mr

 

 

 

....» ..N-rrv’WH1-"M- *' ‘

-..—.... -w‘

»- -..-Weh-

......

89

Appendix 3. A Note on Training the JUdges

The two Judges classifying the five alternative
replies into the five classes of response agreed on 92%
of the #50 replies classified. This rather high level of
agreement was reached with very little instruction to the
Judges. The ease in classifying the responses seemed to
be due to a verbal guide given to the Judges. They were
told to look for the "hooker", i.e. for a loophole or way
out that the reply leaves for the responder so that he
can say in one way or another (response class), "I am not
attacking you." Each class of response has a different
"hooker“.

To the original attack statement "You have a very
annoying laugh! the responder could defend as follows, his
reply to the general question; "Are you attacking me (by
the original attacker)?"

For Isolation, the responder could say "I am not

 

attacking you. All I said was laughter is the medicine of

 

the soul."

For Self‘attack, the responder could say "I am not

 

attacking you. I am attacking myself. I said I can't
change it.“

For ObJect‘displacement, the responder could say "I

 

am not attacking you, or I am not attacking you. I didn't
say anything about you. All I said was, with me it's only
my laugh that's annoying."

For Denial the responder could say "I am not attacking

90

you. all I said was it doesn't bother me.“

For Anger the responder has no way of saying "I am

 

not attacking you."

91

Appendix k Raw Data for Pre to Post Changes X Post
Testing Conditions X EXperimental and Control Groups

NSNG EXperimental

N H
!
mm mm 3’

\N
Om

c
10 17

I

mo

13

p:
8:
c:

o:

S O
p c
o 6
0 p
-18 “h
p c
'12 '1
° P
'5 u
o s
'14 '2
c s
10 -u
o c
-2 -t
s c
-6 ..3
p s
‘5 '3
s o
-9 -12
punished
safe
correct
modal

+4 r4
~40 one -40

\510

SNG EXperimental

A I S O D
c p s c o
1 Li -8 '2 2
p c o c s
2 '1 '3 '1 2 3
c fl 3 c o
3 8 - -6 23 '21
o c p s c
11,-12 9 -u '8 15
p c c s o
5 o -5 2 -6 9
c o s c p
6 lo -12 -6 2 6
C s o c p
7 8 h- '12 'h l;
c s c p o
8 13 '1 '2 l '11
c p o s c
9 2 -6 '5 -2 11
s c o p o
lo '3 '+ '7 '1 '7
S = Stress
NS = Nonstress
G = Generalization
NG : Nongeneralization

92

NSG EXperimental SG EXperimental
I S O D A I S O
0 0 P C c o s c p
-8 6 2 1+ l 15 -12 '5 10 -s
s o C P c s c p o
‘7 -12 13 0 2 10 -3 -2 -2 '3
8 C P 0 c o s c
-2 7 5 ‘H 3 10 -12 J: -u 10
P 0 8 c c s o c p
0 ’2 2 1 u -2 -6 12 -u o
p 0 ° 3 s o c p
’1 5 1 "i 5 -2 “11+ 10 -u l
P 0 8 C c o p s
-u '8 1+ O 6 1+ -114 ‘11 O 21
0 0 p 0 c p o s c
'10 ‘L' '9- 16 7 u '6 u -2 o
8 0 C P c o s c p
O “18 2 '2 8 lo ’18 l 7 0
p C 0 3 o c s c p
O 1 2 16 9 -6 o o 3 3
0 C C 8 p c o c s
'8 -2 0 12 10 o lo -8 -2 o
p ' punished S ' Stress
s = safe NS = Nonstress
c - correct G : Generalization

0
ll

modal NC = Nongeneralization

SNG Control

NSNG Control

c
12

o
‘10

p
“2

CH.

p7

08

3

s
"10

o c
-19 14

c
11

ii

0.4

892

c
10

o
’20

c
2

0
--19

s
O

08

pl

0
—10

s
O

88

'3

C6

po

57:

10

08

Stress

S:

punished

:3:

NS : Nonstress

safe

G : Generalization

correct

0:

NG = Nongeneralization

modal

O:

91+

SG Control

NSG Control

Dav

l

1.

C6

ck.

p8

po

06

p
'10

o
’5

86

86

Cl

8......

pg

80

o
’13 9

c
0

p0

Ch.

10

Stress

S:

punished

p:

Nonstress

NS

safe

8

G : Generalization

correct

0:

NG = Nongeneralization

o = modal